El Niño Southern Oscillation in a Changing Climate 9781119548126, 1119548128

Citation preview

Geophysical Monograph Series

Geophysical Monograph Series 200 Lagrangian Modeling of the Atmosphere John Lin (Ed.) 201 Modeling the Ionosphere‐Thermosphere Jospeh D. Huba, Robert W. Schunk, and George V. Khazanov (Eds.) 202 The Mediterranean Sea: Temporal Variability and Spatial Patterns Gian Luca Eusebi Borzelli, Miroslav Gacic, Piero Lionello, and Paola Malanotte-Rizzoli (Eds.) 203 Future Earth – Advancing Civic Understanding of the Anthropocene Diana Dalbotten, Gillian Roehrig, and Patrick Hamilton (Eds.) 204 The Galápagos: A Natural Laboratory for the Earth Sciences Karen S. Harpp, Eric Mittelstaedt, Noemi d’Ozouville, and David W. Graham (Eds.) 205 Modeling Atmospheric and Oceanic Flows: Insightsfrom Laboratory Experiments and Numerical Simulations Thomas von Larcher and Paul D. Williams (Eds.) 206 Remote Sensing of the Terrestrial Water Cycle Venkat Lakshmi (Ed.) 207 Magnetotails in the Solar System Andreas Keiling, Caitriona Jackman, and Peter Delamere (Eds.) 208 Hawaiian Volcanoes: From Source to Surface Rebecca Carey, Valerie Cayol, Michael Poland, and Dominique Weis (Eds.) 209 Sea Ice: Physics, Mechanics, and Remote Sensing Mohammed Shokr and Nirmal Sinha (Eds.) 210 Fluid Dynamics in Complex Fractured‐Porous Systems Boris Faybishenko, Sally M. Benson, and John E. Gale (Eds.) 211 Subduction Dynamics: From Mantle Flow to Mega Disasters Gabriele Morra, David A. Yuen, Scott King, Sang Mook Lee, and Seth Stein (Eds.) 212 The Early Earth: Accretion and Differentiation James Badro and Michael Walter (Eds.) 213 Global Vegetation Dynamics: Concepts and Applications in the MC1 Model Dominique Bachelet and David Turner (Eds.) 214 Extreme Events: Observations, Modeling and Economics Mario Chavez, Michael Ghil, and Jaime Urrutia-Fucugauchi (Eds.) 215 Auroral Dynamics and Space Weather Yongliang Zhang and Larry Paxton (Eds.) 216 Low‐Frequency Waves in Space Plasmas Andreas Keiling, Dong- Hun Lee, and Valery Nakariakov (Eds.) 217 Deep Earth: Physics and Chemistry of the Lower Mantle and Core Hidenori Terasaki and Rebecca A. Fischer (Eds.) 218 Integrated Imaging of the Earth: Theory and Applications Max Moorkamp, Peter G. Lelievre, Niklas Linde, and Amir Khan (Eds.) 219 Plate Boundaries and Natural Hazards Joao Duarte and Wouter Schellart (Eds.) 220 Ionospheric Space Weather: Longitude and Hemispheric Dependences and Lower Atmosphere Forcing Timothy Fuller‐ Rowell, Endawoke Yizengaw, Patricia H. Doherty, and Sunanda Basu (Eds.) 221 Terrestrial Water Cycle and Climate Change Natural and Human‐Induced Impacts Qiuhong Tang and Taikan Oki (Eds.) 222 Magnetosphere‐Ionosphere Coupling in the Solar System Charles R. Chappell, Robert W. Schunk, Peter M. Banks, James L. Burch, and Richard M. Thorne (Eds.) 223 Natural Hazard Uncertainty Assessment: Modeling and Decision Support Karin Riley, Peter Webley, and Matthew Thompson (Eds.) 224 Hydrodynamics of Time‐Periodic Groundwater Flow: Diffusion Waves in Porous Media Joe S. Depner and Todd C. Rasmussen (Auth.) 225 Active Global Seismology Ibrahim Cemen and Yucel Yilmaz (Eds.)

226 Climate Extremes Simon Wang (Ed.) 227 Fault Zone Dynamic Processes Marion Thomas (Ed.) 228 Flood Damage Survey and Assessment: New Insights from Research and Practice Daniela Molinari, Scira Menoni, and Francesco Ballio (Eds.) 229 Water-Energy‐Food Nexus – Principles and Practices P. Abdul Salam, Sangam Shrestha, Vishnu Prasad Pandey, and Anil K Anal (Eds.) 230 Dawn–Dusk Asymmetries in Planetary Plasma Environments Stein Haaland, Andrei Rounov, and Colin Forsyth (Eds.) 231 Bioenergy and Land Use Change Zhangcai Qin, Umakant Mishra, and Astley Hastings (Eds.) 232 Microstructural Geochronology: Planetary Records Down to Atom Scale Desmond Moser, Fernando Corfu, James Darling, Steven Reddy, and Kimberly Tait (Eds.) 233 Global Flood Hazard: Applications in Modeling, Mapping and Forecasting Guy Schumann, Paul D. Bates, Giuseppe T. Aronica, and Heiko Apel (Eds.) 234 Pre‐Earthquake Processes: A Multidisciplinary Approach to Earthquake Prediction Studies Dimitar Ouzounov, Sergey Pulinets, Katsumi Hattori, and Patrick Taylor (Eds.) 235 Electric Currents in Geospace and Beyond Andreas Keiling, Octav Marghitu, and Michael Wheatland (Eds.) 236 Quantifying Uncertainty in Subsurface Systems Celine Scheidt, Lewis Li, and Jef Caers (Eds.) 237 Petroleum Engineering Moshood Sanni (Ed.) 238 Geological Carbon Storage: Subsurface Seals and Caprock Integrity Stephanie Vialle, Jonathan Ajo-Franklin, and J. William Carey (Eds.) 239 Lithospheric Discontinuities Huaiyu Yuan and Barbara Romanowicz (Eds.) 240 Chemostratigraphy Across Major Chronological Eras Alcides N.Sial, Claudio Gaucher, Muthuvairavasamy Ramkumar, and Valderez Pinto Ferreira (Eds.) 241 Mathematical Geoenergy:Discovery, Depletion, and Renewal Paul Pukite, Dennis Coyne, and Daniel Challou (Eds.) 242 Ore Deposits: Origin, Exploration, and Exploitation Sophie Decree and Laurence Robb (Eds.) 243 Kuroshio Current: Physical, Biogeochemical and Ecosystem Dynamics Takeyoshi Nagai, Hiroaki Saito, Koji Suzuki, and Motomitsu Takahashi (Eds.) 244 Geomagnetically Induced Currents from the Sun to the Power Grid Jennifer L. Gannon, Andrei Swidinsky, and Zhonghua Xu (Eds.) 245 Shale: Subsurface Science and Engineering Thomas Dewers, Jason Heath, and Marcelo Sánchez (Eds.) 246 Submarine Landslides: Subaqueous Mass Transport Deposits From Outcrops to Seismic Profiles Kei Ogata, Andrea Festa, and Gian Andrea Pini (Eds.) 247 Iceland: Tectonics, Volcanics, and Glacial Features Tamie J. Jovanelly 248 Dayside Magnetosphere Interactions Qiugang Zong, Philippe Escoubet, David Sibeck, Guan Le, and Hui Zhang (Eds.) 249 Carbon in Earth’s Interior Craig E. Manning, Jung-Fu Lin, and Wendy L. Mao (Eds.) 250 Nitrogen Overload: Environmental Degradation, Ramifications, and Economic Costs Brian G. Katz 251 Biogeochemical Cycles: Ecological Drivers and Environmental Impact Katerina Dontsova, Zsuzsanna Balogh-Brunstad, and Gaël Le Roux (Eds.) 252 Seismoelectric Exploration: Theory, Experiments, and Applications Niels Grobbe, André Revil, Zhenya Zhu, and Evert Slob (Eds.)

Geophysical Monograph 253

El Niño Southern Oscillation in a Changing Climate Editors Michael J. McPhaden Agus Santoso Wenju Cai

This Work is a co‐publication of the American Geophysical Union and John Wiley and Sons, Inc.

This Work is a co‐publication between the American Geophysical Union and John Wiley & Sons, Inc. This edition first published 2021 by John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA and the American Geophysical Union, 2000 Florida Avenue, N.W., Washington, D.C. 20009 © 2021 American Geophysical Union All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions

Published under the aegis of the AGU Publications Committee Brooks Hanson, Executive Vice President, Science Carol Frost, Chair, Publications Committee For details about the American Geophysical Union visit us at www.agu.org. Wiley Global Headquarters 111 River Street, Hoboken, NJ 07030, USA For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com. Limit of Liability/Disclaimer of Warranty While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials, or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Library of Congress Cataloging‐in‐Publication Data Names: McPhaden, Michael J., editor. Title: El Niño southern oscillation in a changing climate / Michael J. McPhaden, Agus Santoso, Wenju Cai. Description: First edition. | Hoboken, NJ : Wiley-American Geophysical Union, 2020. | Series: Geophysical monograph series | Includes index. | Includes bibliographical references and index. Identifiers: LCCN 2020028098 (print) | LCCN 2020028099 (ebook) | ISBN 9781119548126 (cloth) | ISBN 9781119548119 (adobe pdf) | ISBN 9781119548157 (epub) Subjects: LCSH: El Niño Current. | Climatic changes. | Ocean-atmosphere interaction. Classification: LCC GC296.8.E4 E58 2020 (print) | LCC GC296.8.E4 (ebook) | DDC 551.5/24648–dc23 LC record available at https://lccn.loc.gov/2020028098 LC ebook record available at https://lccn.loc.gov/2020028099 Cover Design: Wiley Cover Image: © NOAA National Environmental Satellite, Data, and Information Service (NESDIS) Set in 10/12pt Times New Roman by SPi Global, Pondicherry, India 10 9 8 7 6 5 4 3 2 1

CONTENTS List of Contributors...............................................................................................................................................vii Acknowledgments................................................................................................................................................xiii Preface.................................................................................................................................................................. xv

Section I: Introduction 1. Introduction to El Niño Southern Oscillation in a Changing Climate Michael J. McPhaden, Agus Santoso, and Wenju Cai........................................................................................3 2. ENSO in the Global Climate System Kevin E. Trenberth..........................................................................................................................................21

Section II: Observations 3. ENSO Observations Michael J. McPhaden, Tong Lee, Severine Fournier, and Magdalena A. Balmaseda..........................................41 4. ENSO Diversity Antonietta Capotondi, Andrew T. Wittenberg, Jong-Seong Kug, Ken Takahashi, and Michael J. McPhaden.....................................................................................................................................65 5. Past ENSO Variability: Reconstructions, Models, and Implications Julien Emile-Geay, Kim M. Cobb, Julia E. Cole, Mary Elliot, and Feng Zhu.......................................................87

Section III: Theories and Dynamics 6. Simple ENSO Models Fei-Fei Jin, Han-Ching Chen, Sen Zhao, Michiya Hayashi, Christina Karamperidou, Malte F. Stuecker, Ruihuang Xie, and Licheng Geng......................................................................................121 7. ENSO Irregularity and Asymmetry Soon-Il An, Eli Tziperman, Yuko M. Okumura, and Tim Li..............................................................................153 8. ENSO Low‐frequency Modulation and Mean State Interactions Alexey V. Fedorov, Shineng Hu, Andrew T. Wittenberg, Aaron F. Z. Levine, and Clara Deser..........................173

Section IV: Modeling and Prediction 9. ENSO Modelling: History, Progress and Challenges Eric Guilyardi, Antonietta Capotondi, Matthieu Lengaigne, Sulian Thual, and Andrew T. Wittenberg..................................................................................................................................201 10. ENSO Prediction Michelle L. L’Heureux, Aaron F. Z. Levine, Matthew Newman, Catherine Ganter, Jing-Jia Luo, Michael K. Tippett, and Timothy N. Stockdale..............................................................................................227

v

vi Contents

Section V: Remote and External Forcing 11. ENSO Remote Forcing: Influence of Climate Variability Outside the Tropical Pacific Jong-Seong Kug, Jerome Vialard, Yoo-Geun Ham, Jin-Yi Yu, and Matthieu Lengaigne.....................................249 12. The Effect of Strong Volcanic Eruptions on ENSO Shayne McGregor, Myriam Khodri, Nicola Maher, Masamichi Ohba, Francesco S. R. Pausata, and Samantha Stevenson.............................................................................................................................267 13. ENSO Response to Greenhouse Forcing Wenju Cai, Agus Santoso, Guojian Wang, Lixin Wu, Mat Collins, Matthieu Lengaigne, Scott Power, and Axel Timmermann.............................................................................................................289

Section VI: Teleconnections and Impacts 14. ENSO Atmospheric Teleconnections Andréa S. Taschetto, Caroline C. Ummenhofer, Malte F. Stuecker, Dietmar Dommenget, Karumuri Ashok, Regina R. Rodrigues, and Sang-Wook Yeh..........................................................................311 15. ENSO Oceanic Teleconnections Janet Sprintall, Sophie Cravatte, Boris Dewitte, Yan Du, and Alexander Sen Gupta........................................337 16. Impact of El Niño on Weather and Climate Extremes Lisa Goddard and Alexander Gershunov .....................................................................................................361 17. ENSO and Tropical Cyclones I-I Lin, Suzana J. Camargo, Christina M. Patricola, Julien Boucharel, Savin Chand, Phil Klotzbach, Johnny C. L. Chan, Bin Wang, Ping Chang, Tim Li, and Fei-Fei Jin..........................................377 18. ENSO-Driven Ocean Extremes and Their Ecosystem Impacts Neil J. Holbrook, Danielle C. Claar, Alistair J. Hobday, Kathleen L. McInnes, Eric C. J. Oliver, Alex Sen Gupta, Matthew J. Widlansky, and Xuebin Zhang ..........................................................................409 19. ENSO Impact on Marine Fisheries and Ecosystems Patrick Lehodey, Arnaud Bertrand, Alistair J. Hobday, Hidetada Kiyofuji, Sam McClatchie, Christophe E. Menkès, Graham Pilling, Jeffrey Polovina, and Desiree Tommasi.............................................429 20. ENSO and the Carbon Cycle Richard A. Betts, Chantelle A. Burton, Richard A. Feely, Mat Collins, Chris D. Jones, and Andy J. Wiltshire....................................................................................................................................453

Section VII: Closing 21. ENSO in a Changing Climate: Challenges, Paleo‐Perspectives, and Outlook Christina Karamperidou, Malte F. Stuecker, Axel Timmermann, Kyung-Sook Yun, Sun-Seon Lee, Fei-Fei Jin, Agus Santoso, Michael J. McPhaden, and Wenju Cai....................................................................473 Glossary..............................................................................................................................................................485 Index...................................................................................................................................................................491

LIST OF CONTRIBUTORS Soon‐Il An Department of Atmospheric Sciences Yonsei University Seoul, Republic of Korea

Suzana J. Camargo Lamont‐Doherty Earth Observatory Columbia University Palisades, New York, USA

Karumuri Ashok Centre for Earth, Ocean and Atmospheric Sciences University of Hyderabad Hyderabad, India

Antonietta Capotondi University of Colorado, CIRES Boulder, Colorado, USA; and NOAA Physical Sciences Laboratory Boulder, Colorado, USA

Magdalena A. Balmaseda European Centre for Medium-Range Weather Forecasts Reading, UK

Johnny C. L. Chan Guy Carpenter Asia‐Pacific Climate Impact Centre School of Energy and Environment City University of Hong Kong Hong Kong, China

Arnaud Bertrand Institut de Recherche pour le Développement (IRD), MARBEC (Univ Montpellier, CNRS, Ifremer, IRD) Sète, France

Savin Chand Centre for Informatics and Applied Optimization Federation University Australia Ballarat, Victoria, Australia

Richard A. Betts Met Office Hadley Centre Exeter, UK; and Global Systems Institute University of Exeter Exeter, UK

Ping Chang Department of Oceanography and Department of Atmospheric Sciences Texas A&M University College Station, Texas, USA

Julien Boucharel LEGOS-CNRS, University of Toulouse Toulouse, France

Han‐Ching Chen Department of Atmospheric Sciences, SOEST University of Hawai’i at Mānoa Honolulu, Hawai’i, USA

Chantelle A. Burton Met Office Hadley Centre Exeter, UK

Danielle C. Claar School of Aquatic and Fisheries Sciences University of Washington Seattle, Washington, USA

Wenju Cai Centre for Southern Hemisphere Oceans Research (CSHOR) CSIRO Oceans and Atmosphere Hobart, Tasmania, Australia; and Key Laboratory of Physical Oceanography/Institute for Advanced Ocean Studies, Ocean University of China and Qingdao National Laboratory for Marine Science and Technology Qingdao, China

Kim M. Cobb Department of Earth Sciences & Technology Georgia Institute of Technology Atlanta, Georgia, USA Julia E. Cole Department of Earth and Environmental Sciences University of Michigan Ann Arbor, Michigan, USA

vii

viii  LIST OF CONTRIBUTORS

Mat Collins College of Engineering, Mathematics, and Physical Sciences University of Exeter Exeter, UK

Alexey V. Fedorov Earth and Planetary Sciences, Yale University New Haven, Connecticut, USA; and LOCEAN-IPSL, Sorbonne University Paris, France

Sophie Cravatte Laboratoire d’Etudes en Géophysique et Océanographie Spatiales (LEGOS), IRD, CNES, CNRS, UPS Toulouse, France

Richard A. Feely NOAA Pacific Marine Environmental Laboratory Seattle, Washington, USA

Clara Deser NCAR, Climate and Global Dynamics Division Boulder, Colorado, USA Boris Dewitte Centro de Estudios Avanzado en Zonas Áridas (CEAZA); Departamento de Biología, Facultad de Ciencias del Mar, Universidad Católica del Norte; Millennium Nucleus for Ecology and Sustainable Management of Oceanic Islands (ESMOI) Coquimbo, Chile; and Laboratoire d’Etudes en Géophysique et Océanographie Spatiales (LEGOS), IRD, CNES, CNRS, UPS Toulouse, France Dietmar Dommenget School of Earth, Atmosphere and Environment ARC Centre of Excellence for Climate Extremes Monash University Melbourne, Victoria, Australia Yan Du State Key Laboratory of Tropical Oceanography South China Sea Institute of Oceanology Chinese Academy of Sciences Guangzhou, Guangdong, China; University of Chinese Academy of Sciences Beijing, China; and Southern Marine Science and Engineering Guangdong Laboratory Guangzhou, Guangdong, China Mary Elliot Université de Nantes, LPG (UMR6112) Nantes, France Julien Emile‐Geay Department of Earth Sciences University of Southern California Los Angeles, California, USA

Severine Fournier NASA/Jet Propulsion Laboratory California Institute of Technology Pasadena, California, USA Catherine Ganter Australian Bureau of Meteorology Melbourne, Victoria, Australia Licheng Geng Department of Atmospheric Sciences, SOEST University of Hawai’i at Mānoa Honolulu, Hawai’i, USA Alexander Gershunov Scripps Institution of Oceanography University of California–San Diego La Jolla, California, USA Lisa Goddard International Research Institute for Climate and Society Columbia University Palisades, New York, USA Eric Guilyardi LOCEAN-IPSL, CNRS/Sorbonne University/IRD/ MNHN Paris, France; and NCAS‐Climate, University of Reading Reading, UK Yoo‐Geun Ham Department of Oceanography Chonnam National University Gwangju, Republic of Korea Michiya Hayashi Department of Atmospheric Sciences, SOEST University of Hawai’i at Mānoa Honolulu, Hawai’i, USA; Now at Center for Global Environmental Research National Institute for Environmental Studies Tsukuba, Ibaraki, Japan

LIST OF CONTRIBUTORS  ix

Alistair J. Hobday CSIRO Oceans and Atmosphere Hobart, Tasmania, Australia Neil J. Holbrook Institute for Marine and Antarctic Studies University of Tasmania Hobart, Tasmania, Australia; and Australian Research Council Centre of Excellence for Climate Extremes University of Tasmania Hobart, Tasmania, Australia Shineng Hu Lamont‐Doherty Earth Observatory Columbia University Palisades, New York, USA Fei‐Fei Jin Department of Atmospheric Sciences, SOEST University of Hawai’i at Mānoa Honolulu, Hawai’i, USA Chris D. Jones Met Office Hadley Centre Exeter, UK Christina Karamperidou Department of Atmospheric Sciences, SOEST University of Hawai’i at Mānoa Honolulu, Hawai’i, USA Myriam Khodri LOCEAN-IPSL, IRD/Sorbonne Université/CNRS/MNHN Paris, France Hidetada Kiyofuji National Research Institute of Far Seas Fisheries Japan Fisheries Research and Education Agency Shimizu, Shizuoka, Japan Phil Klotzbach Department of Atmospheric Science Colorado State University Fort Collins, Colorado, USA Jong‐Seong Kug Division of Environmental Science & Engineering Pohang University of Science and Technology (POSTECH) Pohang, Republic of Korea Michelle L. L’Heureux National Oceanic and Atmospheric Administration NWS/NCEP/Climate Prediction Center College Park, Maryland, USA

Tong Lee NASA/Jet Propulsion Laboratory California Institute of Technology Pasadena, California, USA Sun‐Seon Lee Center for Climate Physics, Institute for Basic Science Busan, Republic of Korea; and Pusan National University Busan, Republic of Korea Patrick Lehodey Collecte Localisation Satellite, 11 rue Hermès 31520 Ramonville St Agne, France Matthieu Lengaigne LOCEAN-IPSL, Sorbonne Universités/UPMCCNRS-IRD-MNHN Paris, France; and MARBEC, University of Montpellier, CNRS, IFREMER, IRD Sète, France Aaron F. Z. Levine Department of Atmospheric Sciences University of Washington Seattle, Washington, USA Tim Li Department of Atmospheric Sciences/IPRC University of Hawai’i at Mānoa Honolulu, Hawai’i, USA I‐I Lin Department of Atmospheric Sciences National Taiwan University Taipei, Taiwan Jing‐Jia Luo Institute for Climate and Application Research (ICAR)/ CICFEM/KLME/ILCEC Nanjing University of Information Science and Technology Nanjing, China Nicola Maher Max Planck Institute for Meteorology Hamburg, Germany Sam McClatchie 38 Upland Rd, Huia Auckland 0604, New Zealand

x  LIST OF CONTRIBUTORS

Shayne McGregor School of Earth, Atmosphere and Environment, and ARC Centre of Excellence for Climate Extremes, Monash University Melbourne, Victoria, Australia Kathleen L. McInnes Climate Science Centre, CSIRO Oceans and Atmosphere Aspendale, Victoria, Australia Michael J. McPhaden NOAA/Pacific Marine Environmental Laboratory Seattle, Washington, USA Christophe E. Menkès Institut de Recherche pour le Développement (IRD) ENTROPIE (IRD/CNRS/Univ. La Réunion) Nouméa, New Caledonia Matthew Newman Cooperative Institute for Research in the Environmental Sciences University of Colorado; and NOAA/ESRL Physical Sciences Division Boulder, Colorado, USA Masamichi Ohba Central Research Institute of Electric Power Industry Chiba, Japan Yuko M. Okumura Institute for Geophysics, Jackson School of Geosciences The University of Texas at Austin Austin, Texas, USA Eric C. J. Oliver Department of Oceanography Dalhousie University Halifax, Nova Scotia, Canada

Graham Pilling The Pacific Community (SPC), BP D5 Noumea, New Caledonia Jeffrey Polovina 196 Pauahilani Pl. Kailua, Hawai’i, USA Scott Power Australian Bureau of Meteorology Melbourne, Victoria, Australia; and School of Earth, Atmosphere and Environment, and ARC Centre of Excellence for Climate Extremes Monash University Melbourne, Victoria, Australia Regina R. Rodrigues Department of Oceanography Federal University of Santa Catarina Florianópolis, Santa Catarina, Brazil Agus Santoso ARC Centre of Excellence for Climate Extremes and Climate Change Research Centre University of New South Wales Sydney, New South Wales, Australia; and Centre for Southern Hemisphere Oceans Research (CSHOR), CSIRO Oceans and Atmosphere Hobart, Tasmania, Australia Alexander Sen Gupta Climate Change Research Centre and ARC Centre of Excellence for Climate Extremes University of New South Wales Sydney, New South Wales, Australia Janet Sprintall Scripps Institution of Oceanography University of California–San Diego La Jolla, California, USA

Christina M. Patricola Climate and Ecosystem Sciences Division Lawrence Berkeley National Laboratory Berkeley, California, USA; and Iowa State University Ames, Iowa, USA

Samantha Stevenson Bren School of Environmental Science & Management University of California Santa Barbara Santa Barbara, California, USA

Francesco S. R. Pausata Department of Earth and Atmospheric Sciences University of Quebec in Montreal Montreal, Quebec, Canada

Timothy N. Stockdale European Centre for Medium‐Range Weather Forecasts Reading, UK

LIST OF CONTRIBUTORS  xi

Malte F. Stuecker Center for Climate Physics, Institute for Basic Science Busan, Republic of Korea; and Pusan National University Busan, Republic of Korea Now at Department of Oceanography and International Pacific Research Center University of Hawai’i at Mānoa Honolulu, Hawai’i, USA Ken Takahashi Servicio Nacional de Meteorología e Hidrología del Perú—SENAMHI Lima, Peru Andrèa S. Taschetto Climate Change Research Centre and ARC Centre of Excellence for Climate Extremes University of New South Wales Sydney, New South Wales, Australia Sulian Thual Institute of Atmospheric Sciences/Department of Atmospheric and Oceanic Sciences Fudan University Shanghai, China Axel Timmermann Center for Climate Physics, Institute for Basic Science Busan, Republic of Korea; and Pusan National University Busan, Republic of Korea Michael K. Tippett Department of Applied Physics and Applied Mathematics Columbia University New York, USA Desiree Tommasi Institute of Marine Sciences University of California Santa Cruz Santa Cruz, California, USA; and NOAA Southwest Fisheries Science Center La Jolla, California, USA Kevin E. Trenberth National Center for Atmospheric Research Boulder, Colorado, USA Eli Tziperman Department of Earth and Planetary Sciences and School of Engineering and Applied Sciences Harvard University Cambridge, Massachusetts, USA

Caroline C. Ummenhofer Department of Physical Oceanography Woods Hole Oceanographic Institution Woods Hole, Massachusetts, USA; and ARC Centre of Excellence for Climate Extremes University of New South Wales Sydney, New South Wales, Australia Jerome Vialard LOCEAN-IPSL, CNRS/Sorbonne Université/IRD/MNHN Paris, France Bin Wang Department of Atmospheric Sciences University of Hawai’i Honolulu, Hawai’i, USA Guojian Wang Centre for Southern Hemisphere Oceans Research (CSHOR) CSIRO Oceans and Atmosphere Hobart, Tasmania, Australia; and Key Laboratory of Physical Oceanography/Institute for Advanced Ocean Studies Ocean University of China and Qingdao National Laboratory for Marine Science and Technology Qingdao, China Matthew J. Widlansky Joint Institute for Marine and Atmospheric Research School of Ocean and Earth Science and Technology University of Hawai’i at Mānoa Honolulu, Hawai’i, USA Andy J. Wiltshire Met Office Hadley Centre Exeter, UK; and Global Systems Institute, University of Exeter Exeter, UK Andrew T. Wittenberg NOAA Geophysical Fluid Dynamics Laboratory Princeton, New Jersey, USA Lixin Wu Key Laboratory of Physical Oceanography/Institute for Advanced Ocean Studies, Ocean University of China and Qingdao National Laboratory for Marine Science and Technology Qingdao, China

xii  LIST OF CONTRIBUTORS

Ruihuang Xie Key Laboratory of Ocean Circulation and Waves Institute of Oceanology Chinese Academy of Sciences Qingdao, China Sang‐Wook Yeh Department of Marine Science and Convergent Technology Hanyang University Ansan, Republic of Korea Jin‐Yi Yu Department of Earth System Science University of California–Irvine Irvine, California, USA Kyung‐Sook Yun Center for Climate Physics, Institute for Basic Science Busan, Republic of Korea; and Pusan National University Busan, Republic of Korea

Xuebin Zhang CSIRO Oceans and Atmosphere Hobart, Tasmania, Australia Sen Zhao Department of Atmospheric Sciences, SOEST University of Hawai’i at Mānoa Honolulu, Hawai’i, USA Feng Zhu Department of Earth Sciences University of Southern California Los Angeles, California, USA

ACKNOWLEDGMENTS Each of the 21 chapters in this book was peer reviewed by at least two experts in the field before acceptance for publication. We would like to recognize those individuals, listed below, who participated in this process. We are indebted to

them for their willingness to offer thoughtful and constructive critiques of submitted manuscripts. Their efforts have assured that material presented in this book is up to date, of the highest quality, and of the greatest relevance.

Soon‐Il An Magdalena Balmaseda Julien Boucharel Antonietta Capotondi Matthieu Carré Bo Christiansen Mat Collins Boris Dewitte Pedro DiNezio Dietmar Dommenget Alexey Fedorov Ming Feng Yoo‐Geun Ham Neil Holbrook Shineng Hu Sarah Ineson Nathaniel Johnson Karumuri Ashok Jin‐Soo Kim Andrew King Mojib Latif

Andrew Lenton Michelle L’Heureux Janice Lough Jing‐Jia Luo Shayne McGregor Christophe Menkès Kathy Pegion Scott Power Hamish Ramsay Harun Rashid Alex Sen Gupta Toshiaki Shinoda Georgiy Stenchikov Malte Stuecker Ken Takahashi Pascal Terray Caroline Ummenhofer Kevin Walsh Yan Xue Jin‐Yi Yu Dongliang Yuan

xiii

PREFACE El Niño (EN) and the Southern Oscillation (SO) constitute a richly textured scientific puzzle that has fascinated scientists for well over a century. They were originally thought to be unrelated; however, the Norwegian‐born meteorologist Jacob Bjerknes realized in the mid‐1960s that El Niño and the Southern Oscillation were oceanic and atmospheric manifestations of the same phenomenon, which we now refer to collectively as ENSO. ENSO is spawned in the tropical Pacific, but its reach is global. It alters the general circulation of the atmosphere from one season to the next through far‐field teleconnections, leading to droughts, floods, heat waves, and other extreme weather events across our planet. These natural hazards have widespread impacts on human and natural systems, including agriculture, public health, power generation and consumption, financial markets, transportation, tourism, freshwater resources, civil conflict, wildfires, fisheries, marine and terrestrial ecosystems, and biodiversity. The character of ENSO, the warm phase of which is referred to as El Niño and the cold phase La Niña, depends on the long‐term average background climatic conditions on which it develops. However, because of human activities since the start of the Industrial Revolution in the mid‐ 18th century, these background conditions have been changing at a pace that has greatly accelerated in recent decades. We have come to appreciate just how profound these changes are, thanks to five Intergovernmental Panel on Climate Change assessments since 1990. The most recent of these assessments in 2013-2014 leaves no doubt: “Warming of the climate system is unequivocal … human influence on the climate system is clear, and recent anthropogenic emissions of green‐house gases are the highest in history.” Greenhouse gas emissions moreover show no sign of abating in the near term. It is natural to ask then whether the character of ENSO has changed already, whether it will in the future, and if so, how. This book grew out of a realization that, while there are several excellent treatises available focusing on ENSO dynamics, its climate impacts, and its history, there has been no comprehensive examination of ENSO in a changing climate and what it means for society. This book is designed to fill that gap. Our purpose is to review the current state of knowledge regarding ENSO variability, predictability, and impacts, and how a changing climate may affect them. Emphasis is on developments over the past 20 years, since the last extensive review of

ENSO research was published as a collection of papers in the Journal of Geophysical Research in 1998. Those papers appeared in the aftermath of the 1997–1998 El Niño, the most extreme El Niño event in the instrumental record thus far. Since then, we have experienced another extreme El Niño in 2015–2016, different in character but equally consequential, providing further impetus to summarize recent advances in ENSO science. Given ENSO’s profound effects on society and the environment, our intent is to reach a broad audience of experts and nonspecialists alike. The goal is to provide authoritative information on a subject of great scientific interest and practical value, at the same stimulating further research on many important outstanding issues. The book is unique in scope, encompassing a wide range of topics related to ENSO that heretofore have not been covered extensively in a single volume. We begin with an introductory chapter on the ENSO cycle and its global impacts, the history of foundational ideas and watershed events that propelled ENSO to a position of prominence in the study of Earth system science, and why ENSO in a changing climate is such an urgent problem today. This is followed in chapter 2 by an overview of ENSO in the global climate system. We then describe instrumental observations of ENSO variability, event diversity, and paleo‐ reconstructions of ENSO in the distant past (chapters 3–5). Next, we discuss theories of ENSO dynamics and evolution (chapters 6 and 7) and the modulation of ENSO on decadal time scales (chapter  8). Computer modeling and prediction are covered in chapters 9 and 10. How climate variability outside the tropical Pacific, volcanic eruptions, and rising greenhouse gas concentrations in the atmosphere affect ENSO are covered chapters 11–13. Atmospheric and oceanic teleconnections are treated in chapters 14 and 15. The impacts of ENSO on weather and climate extremes, tropical cyclones, ocean extremes (e.g. marine heatwaves, coral bleaching, and sea level rise), fisheries and marine ecosystems, and the global carbon cycle are described in chapters 16–20. The book concludes with a supplemental analysis and interpretation of paleoreconstructions to highlight some key unresolved issues (chapter 21). A glossary is provided at the end to define many of the technical terms used in this book. The book focuses on fundamental concepts for which there is a broad consensus of expert opinion. However, the science of ENSO in a changing climate is far from

xv

xvi PREFACE

settled so that, as in any field where cutting‐edge ideas continue to emerge, differences in interpretation of the same evidence may be found on certain topics. Readers should recognize that these instances on the pages of this book are an indicator that the field is healthy and advancing through robust debate. This book would not have been possible without the effort of many individuals. First and foremost are the many outstanding scientists who contributed to writing the various chapters. It has been an enjoyable and rewarding experience working with them to bring this collective effort to fruition. We are also indebted to the reviewers who dedicated their time and expertise to provide constructive and critical reviews for chapters on which they were not an author. Their efforts contributed to the high quality of scholarship and balanced perspectives presented in this book. We have listed these reviewers on a separate page to recognize their critical involvement in the process. We would also like to acknowledge support from the Centre of Southern Hemisphere Ocean Research (CSHOR) for sponsoring a symposium and author workshop on ENSO under greenhouse warming in Hobart, Australia, in January–February 2019 (https://cshor.csiro. au/news/cshor‐enso‐science‐symposium/) to review the latest scientific advances and coordinate inputs to the various chapters. We thank Leonie Wyld, Benjamin Ng, and Guojian Wang of CSHOR for their assistance in organizing these events and in facilitating coordination among authors. We also thank Evelyn Ong for her help in putting together the list of 98 contributing authors to the book.

MJM would like to thank the U.S. National Oceanic and Atmospheric Organization (NOAA) for its support; AS and WC wish to acknowledge the support of CSHOR and the Earth System and Climate Change Hub of the Australian Government’s National Environment Science Program (NESP). Finally, we would like to express our sincere appreciation to Rituparna Bose, Emily Bae, Nithya Sechin, Bobby Kilshaw, Vimali Joseph and the copyediting team of Wiley for their help in publishing this book. Michael J. McPhaden NOAA/Pacific Marine Environmental Laboratory, Seattle, Washington, USA Agus Santoso ARC Centre of Excellence for Climate Extremes and Climate Change Research Centre, University of New South Wales, Sydney, Australia Centre for Southern Hemisphere Oceans Research (CSHOR), CSIRO Oceans and Atmosphere, Hobart, Tasmania, Australia Wenju Cai Centre for Southern Hemisphere Oceans Research (CSHOR), CSIRO Oceans and Atmosphere, Hobart, Tasmania, Australia Key Laboratory of Physical Oceanography/Institute for Advanced Ocean Studies, Ocean University of China and Qingdao National Laboratory for Marine Science and Technology, Qingdao, China

Section I Introduction

1 Introduction to El Niño Southern Oscillation in a Changing Climate Michael J. McPhaden1, Agus Santoso2,3, and Wenju Cai3,4

ABSTRACT El Niño and the Southern Oscillation (ENSO) is a naturally occurring fluctuation of the climate system that is generated in the tropical Pacific through interactions between the ocean and the atmosphere. It is the strongest year‐to‐year climate variation on the planet, with environmental and societal consequences felt worldwide. ENSO warm events (El Niño) and cold events (La Niña) are occurring in the context of a global climate system that is rapidly changing through human activities that have raised heat‐trapping greenhouse gas concentrations in the atmosphere to historically unprecedented levels. As a result, the planet has warmed, and it will continue to warm at a rate dependent on future greenhouse gas emissions. This raises questions about whether climate change has affected the ENSO cycle already, whether it will in the future, and if so, how. Here, we briefly describe ENSO and its impacts; highlight the history of ideas and events that have shaped our understanding of ENSO; discuss current challenges in ENSO research; and address why ENSO in a changing climate is such an urgent problem in Earth system science today.

1.1. INTRODUCTION

and patterns of weather variability worldwide (Figure 1.2; Yeh et al., 2018), with f­ ar‐reaching effects on human and natural systems (McPhaden et al., 2006). Floods, droughts, heat waves, and other extreme events associated with both warm and cold phases of ENSO have major impacts on agricultural production, food security, freshwater resources, public health, power generation, and economic vitality in many nations (Figure 1.3). ENSO can also disrupt the normal functioning of marine and terrestrial ecosystems, pelagic fisheries, and the global carbon cycle. The most recent major El Niño, the first of the 21st century and one of the strongest on record, occurred in 2015–2016 (Figure  1.4; L’Heureux et  al., 2017; Santoso et  al., 2017), with widespread impacts that affected millions of people around the globe. As our understanding and ability to predict ENSO has evolved over the past few decades, so too has the climate system itself. Human activities, through the combustion of fossil fuels and deforestation, have raised ­heat‐­trapping greenhouse gas (GHG) concentrations in the atmosphere to unprecedented levels since the start of the Industrial Revolution in the mid‐18th

El Niño Southern Oscillation (ENSO) is generated in the tropical Pacific through interactions between the atmosphere and ocean, mediated by surface wind and sea surface temperature (SST) feedbacks (Figure  1.1). It is the most energetic year‐to‐year variation of the climate system on Earth, with ENSO warm events (El Niño) and cold events (La Niña) occurring roughly every 2–7 years. ENSO events alter the global atmospheric circulation

1 NOAA Pacific Marine Environmental Laboratory, Seattle, WA, USA 2 ARC Centre of Excellence for Climate Extremes and Climate Change Research Centre, University of New South Wales, Sydney, NSW, Australia 3 Centre for Southern Hemisphere Oceans Research (CSHOR), CSIRO Oceans and Atmosphere, Hobart, Tasmania, Australia 4 Key Laboratory of Physical Oceanography/Institute for Advanced Ocean Studies, Ocean University of China and Qingdao National Laboratory for Marine Science and Tech­ nology, Qingdao, China

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 3

4  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE La Niña Conditions

Normal Conditions Walker Circulation

Equator

Equator

Thermocline 120°E

Thermocline 120°E

80°W

80°W

El Niño Conditions

Equator

Thermocline 120°E

80°W

Figure 1.1  Schematic of La Niña, normal, and El Niño conditions in the tropical Pacific. Arrows indicate the directional sense of circulation in the atmosphere and ocean. Focusing on normal conditions in the top right panel first, the atmospheric circulation cell characterized by rising air masses, deep atmospheric convection, and heavy rainfall over warm surface waters of the western Pacific and descent over cooler surface waters of the eastern Pacific is referred to as the Walker Circulation. Equatorial upwelling transports cold water upward from the ocean interior to create a “cold tongue” in sea surface temperature that extends all the way from the coast of South America to the international date line (green and blue shades). To the west of the cold tongue lies the western Pacific “warm pool” (orange and red shades), which are the warmest waters in the open ocean on Earth. The thermocline is a region of rapid vertical temperature change that separates the warm upper ocean from the cold deep interior; its tilt in the east‐west direction is related to the strength of the trade winds. When the trade winds weaken during El Niño (bottom panel), the warm pool shifts eastward, the thermocline flattens out, and upwelling is reduced in the cold tongue. The unusually warm surface waters then feed back to the atmosphere to cause further weakening of the trade winds. As the central and eastern Pacific warm up, the ascending air masses that lead to deep atmospheric convection and heavy rainfall in the western Pacific migrate eastward with the warm water. Air flow into the convective center from the west causes the trade winds to weaken further, which then leads to more surface warming. In this way, the atmosphere and the ocean become locked in a reinforcing positive feedback loop in which weakening winds and warming sea surface temperatures continue to amplify. The termination of El Niño is brought about by delayed negative feedbacks involving ocean dynamical processes that eventually return the system to normal or sometimes cause it to overshoot into cold La Niña conditions (top left panel). During La Niña, intensified trade winds create a steeper westward tilt to the thermocline, more intense upwelling in the cold tongue, and a warm pool displaced further to the west. The detailed processes that determine the fluctuations between normal, El Niño, and La Niña phases of the ENSO cycle are a major focus of this book.

century. These alterations in atmospheric chemistry have caused the planet to warm, as evident in the rise of global and regional surface air temperatures, melting glaciers, disappearing Arctic sea ice, ocean heat uptake,

sea level rise, more extreme weather events, and other indicators of systematic environmental change such as habitat loss and species extinctions (IPCC, 2013). We have entered the Anthropocene (Crutzen & Stoermer,

INTRODUCTION  5

Figure 1.2  Typical impacts of El Niño (top) and La Niña (bottom) on global weather patterns during the peak season of development in December–February (after Ropelewski & Halpert, 1987; Courtesy of NOAA/Climate Prediction Center).

2000), the term coined for the most recent geological era characterized by the imposing influence of human activity on climate and the environment. The far‐reaching consequences for society are evident now and likely will be even more so in the future. Against this backdrop, some of the most urgent questions we face today in Earth system science revolve around whether climate change has already affected the character of ENSO and

its impacts, or whether it will do so in the future. This book addresses these questions through a series of overview chapters on relevant topics authored by leading experts in the field. This introductory chapter describes the motivation for and purpose of the book, highlights the history of ideas and events that have shaped our understanding of ENSO, and emphasizes how the Earth’s changing climate

Figure 1.3  Images of typical ENSO impacts: Flooding in Peru (top) during the 1997–1998 El Niño (courtesy of the University of Piura); (middle) drought in New South Wales, Australia, early in the 2018–2019 El Niño (photo credit: Graham Jepson); and (bottom) the Thomas wildfire in Southern California in December 2017, during a La Niña winter (photo credit: Kari Greer).

INTRODUCTION  7 Global Sea Surface Temperature Anomalies 90°N

4

60°N

3 2

30°N

1

0°

0°C –1

30°S

–2 –3

60°S 90°S 0°E

December 2015 60°E

120°E

–4 180°

120°W

60°W

0°E

Figure 1.4  Global SST anomalies for December 2015, relative to a 30‐year (1981–2010) climatological average, based on the Reynolds et al. (2002) blended satellite and in situ data product.

adds challenging new dimensions to the study of ENSO and its impacts. 1.2. HISTORICAL BACKGROUND El Niño is Spanish for the Christ child. It derives from “Corriente del Niño,” the name local Peruvian fishermen gave to a warm southward‐flowing coastal ocean current that became noticeable every year around Christmas time (Carrillo, 1892). In some years, the current would be stronger, warmer, last longer, and extend over a broader region. The unusual warmings would disrupt regional fisheries and were accompanied by heavy rains in the coastal zone of western South America (Glantz, 2001). These periods of unusual warming are what we now refer to as “El Niño.” El Niño is coupled to the Southern Oscillation, an atmospheric phenomenon discovered by the Englishman Sir Gilbert Walker in the early 20th century. Walker was appointed Director General of Observatories in India in 1904 after a devastating drought and famine in ­1899–1900 claimed more than a million lives. Walker believed that the lack of summer monsoon rainfall which caused these droughts might be predictable. Using surface atmospheric pressure data gathered from all over the world, he published papers in the 1920s and 1930s, describing what he called the Southern Oscillation (Walker, 1924; Walker & Bliss, 1932). The pattern represented a year‐to‐year seesaw in surface atmospheric pressure between the Eastern and Western Hemispheres (Figure  1.5): when pressure was unusually low in the southeastern tropical Pacific, it was high over the western tropical Pacific and Indian Ocean, and vice versa. High pressure over India corresponded with deficient summer monsoon rainfall, whereas low pressure corresponded with abundant summer monsoon rainfall. Curiously, he also found that

drought in Southern Africa and warm winters in Canada were correlated with the pressure pattern related to the Southern Oscillation. Regrettably for Walker, the Southern Oscillation was not a useful monsoon predictor because it lagged rather than led Indian summer monsoon rainfall. Moreover, his contemporaries were skeptical of these statistical relationships because he could not explain physically why, for example, climatic variations in disparate regions of the globe should be related. Walker died in 1958, and his obituary the following year in the Quarterly Journal of the Royal Meteorological Society (1959, p. 185) summed it up: Walker’s hope was presumably not only to unearth relations useful for forecasting but to discover sufficient and sufficiently important relations to provide a productive starting point for a theory of world weather. It hardly seems to be working out like that.

Ironically, Walker died during the 1957–1958 International Geophysical Year (IGY). Had he lived just a little longer, he would have seen his reputation rehabilitated. New data collected during the IGY, the first major international scientific study of the global atmosphere, ocean and solid earth, would prove to be crucial. Enter Jacob Bjerknes, a Norwegian‐born meteorologist who provided the breakthrough conceptual framework for understanding ENSO in the 1960s (Bjerknes, 1966, 1969). Using data from the IGY, which just happened to coincide with a major El Niño in 1957–1958, Bjerknes identified the dynamic relationship between Walker’s Southern Oscillation and unusual El Niño warmings along the west coast of South America. Bjerknes also realized that it was not just the coastal zone off western South America that warmed during El Niño but the entire tropical Pacific basin, out to near the international date line (e.g. Figure  1.4). In addition, he recognized that El Niño could affect weather over North America and

8  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE 2

80°N

0 –2 2

0

60°

L

2

0

L 2

2

–5 L

0

H

L –2

–2

2

–2 H

2

6

L –8 4

0 2

20°W

20°E

–4 60°

100°

140°

0° –4

20° 40°

L

–2

60°S

0 180°

20°

–6

0

2

0 2

0

8

40°

140°

100°

60°W

Figure 1.5  The pattern of annual mean sea level pressure anomalies associated with the Southern Oscillation. Cross‐hatching indicates regions where sea level pressure varies in phase with Darwin, Northern Australia, and shading indicates regions where sea level pressure varies out of phase with Darwin. Units are in correlation coefficient (×10), with large absolute values indicating a more consistent relationship with Darwin. Dashed contours indicate regions of low data availability (after Trenberth & Shea, 1987).

elsewhere via far-field atmospheric teleconnections, which were anticipated in Walker’s earlier work on the Southern Oscillation. In honor of Sir Gilbert Walker, Bjerknes (1969) named the east‐west pattern of ascending and descending air masses in the tropics, connected by winds aloft and at the surface (Figures  1.1, 1.6), the “Walker Circulation.” Bjerknes correctly postulated that El Niño arises through reinforcing feedbacks between changes in surface winds and SSTs, as described in brief below (see also chapter  2). Normally, easterly trade winds drive ocean surface currents to the west, piling up water heated by the sun in a deep western Pacific warm pool (Figure 1.1). In the eastern Pacific, warm surface water that has been transported westward is replaced by cooler water from below through a process called equatorial upwelling. Upwelled water creates a “cold tongue” in SST that extends all the way from the coast of South America to the international date line (Figure  1.1). In the western Pacific, ascending air masses, deep atmospheric convection, and heavy rainfall are normally concentrated over the warm surface water that is found there. During El Niño, the trade winds systematically weaken, which allows warm water piled up in the west to migrate eastward. Upwelling is also reduced, causing SSTs in the eastern Pacific to rise. Warming east of the dateline causes the ascending air masses, deep convection, and

heavy rainfall to shift eastward as well. This in turn leads to a further relaxation of the trade winds to the west of the convective center, causing additional surface warming along the equator. This self‐amplifying sequence of events involving weakening zonal winds, warming sea surface temperatures, and reduction in equatorial upwelling has been widely referred to as the “Bjerknes feedback.” In contrast to Bjerknes, who emphasized the importance of equatorial Pacific SSTs on large‐scale atmospheric circulation, Jerome Namias proposed that SST anomalies in the North Pacific had a greater influence on North American seasonal climate (Namias, 1969). The North Pacific Experiment (NORPAX) was established in the late 1960s with support from the U.S. National Science Foundation and Office of Naval Research to test Namias’s and Bjerknes’s ideas (Anonymous, 1974). However, NORPAX found little convincing evidence from observations or atmospheric modeling studies that the North Pacific Ocean forced the overlying atmosphere to significantly affect seasonal climate over the U.S. (Chervin et  al., 1976; Davis, 1978). Hence, by the late‐ 1970s, NORPAX had concentrated its efforts on the equatorial ocean and El Niño (Wyrtki et al., 1981). According to conventional wisdom at the time, local weakening of alongshore winds off South America was assumed to cause the onset of El Niño warming in the

INTRODUCTION  9 Walker Circulation

ITCZ

20N

io

n

0

C

irc ul

at

10S

22

23

24

150E

25

160W

26

27

10W

60W

110W

28

H

100E

50E

ad

le y

20S

29

Figure 1.6 Schematic of the long‐term average large‐scale atmospheric Walker and Hadley circulations. The Walker Circulation (thick arrows) involves rising moist air over warm equatorial ocean waters, subsidence over cooler waters, and east‐west atmospheric flow at the surface and in the upper troposphere to close the loop. The tropical Pacific hosts the warmest region of the open ocean on Earth, namely the western Pacific warm pool, with SSTs (color shading) higher than 28°C. The trade winds flow westward toward the warm pool, causing upwelling of cold subsurface waters to form a narrow equatorial strip of relatively low SSTs in the eastern Pacific, often referred to as the cold tongue. Major rainfall bands, the Intertropical Convergence Zone and South Pacific Convergence Zone, form above warm SSTs north and south of the equator, respectively (see also Figure 2.2 in chapter 2). The large‐scale general atmospheric circulation in the north‐south direction, referred to as the Hadley circulation (dashed arrows), involves surface flow converging into regions of warm SST where air masses ascend and form precipitating clouds. Compensating poleward flow aloft then descends in subtropical high‐pressure zones in the Northern and Southern Hemispheres to close the Hadley cell in the meridional direction.

eastern basin. Wyrtki (1975), however, found no support for this idea in observations and so proposed a radically different hypothesis: the onset of El Niño was remotely forced by a strengthening of the trade winds in the central Pacific, followed by their sudden collapse. This sequence of wind variations would force a downwelling equatorial Kelvin wave that would propagate thousands of kilometers to the east across the basin over the course of a few months. Along its path, the Kelvin wave would depress the thermocline and, on reaching the eastern Pacific, elevate SSTs through a reduction in equatorial upwelling of cold thermocline water. Wyrtki’s hypothesis stimulated new theoretical and numerical modeling studies in which wind forced Kelvin waves played a prominent role in El Niño development (e.g. McCreary, 1976; Hurlburt et al., 1976). However, these waves had not yet been clearly detected in ocean observations. The 1970s also saw the establishment of NOAA’s Equatorial Pacific Ocean Climate Studies (EPOCS) program to study El Niño and its climatic impacts over North America. EPOCS supported an observational program in the eastern Pacific, a region where El Niño anomalies in SST, thermocline depth, and sea level were large (Halpern et  al., 1983). Using combined measurements from NORPAX and EPOCS moored buoy arrays in the central and eastern Pacific, Knox and Halpern (1982) were the first to unambiguously detect an eastward propagating Kelvin wave along the equator. This

milestone set theories of remote forcing in the equatorial Pacific on a firm observational foundation. EPOCS also supported a groundbreaking observational study by Rasmusson and Carpenter (1982) of how a typical El Niño event evolved from start to finish. Their analysis, based on averaging or “compositing” all available surface oceanic and atmospheric data in the tropical Pacific across six warm events from the 1950s to the 1970s, was widely viewed as the definitive statement about El Niño’s life cycle. The scientific community was therefore shocked when the 1982–1983 El Niño, at that time the strongest of the 20th century, did not develop as anticipated from the composite event. Even more startling was the confusion about whether an El Niño was even underway. The first signs of warming based on the composite were expected along the west coast of South America in March to May, which didn’t happen in 1982. In addition, the Mexican volcano El Chichón erupted in March–April 1982, ejecting a cloud of sulfuric aerosols high into the stratosphere that caused large undetected cold biases in satellite measurements of SST. The lack of expected precursory coastal warming and the failure of satellites to observe the onset of El Niño further to the west along the equator was compounded by the fact that there were almost no data available in real time from the ocean for routine day‐to‐day monitoring of evolving environmental conditions in the Pacific. It was against this backdrop of misconceptions and inadequate data that Klaus Wyrtki, who pioneered

10  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

ocean observations for El Niño (Chapter 3), proclaimed while attending a scientific conference at Princeton University in October 1982: “to call this El Niño would be a case of child abuse!” (Dennis Moore, personal communication). Later that month, a research cruise discovered the telltale signs of major El Niño near its peak in the eastern Pacific (Toole & Borges, 1984) and the scientific community realized how much it had been in the dark. The climatic and societal impacts of the stunning 1982– 1983 El Niño reverberated around the globe (Canby, 1984) and helped motivate the 10‐year (1985–1994) Tropical Ocean Global Atmosphere (TOGA) program (McPhaden et  al., 2010). TOGA, guided in part by the modeling and observational successes of NORPAX and EPOCS, was the first international research effort conducted under sponsorship of the newly established World Climate Research Program (WCRP). Its main goals were (1) to determine the predictability of the c­ oupled ocean‐ atmosphere system in the tropics on seasonal‐to‐interannual time scales, (2) to understand the mechanisms responsible for that predictability, and (3) to establish an observing system to support seasonal climate prediction. Although there were efforts undertaken to study all three tropical ocean basins during TOGA, ENSO in the Pacific was the main phenomenological target because of its strength and global impacts (McPhaden et al., 2010). TOGA forged the fields of theory, observation, and modeling into a coherent program that proved to be a spectacular success. It revolutionized our ability to model and predict ENSO (Stockdale et  al., 1998; Latif et  al., 1998), profoundly deepened our understanding of ENSO dynamics through the development of new theories for the ENSO cycle (Neelin et  al., 1998), gave us a greater appreciation for ENSO’s far-field impacts through atmospheric teleconnections (Trenberth et  al., 1998), and established a basin scale observing system to deliver data in real time to support both research and forecasting (McPhaden et  al., 1998). These advances were summarized in a series of review papers (some of which are cited above) in the Journal of Geophysical Research in 1998. It was also during TOGA that the cold phase of the ENSO cycle was christened “La Niña” (Philander, 1990), recognizing that the Bjerknes feedback could operate in reverse to produce unusually cold sea surface temperatures and strengthened trade winds in the tropical Pacific. 1.3. RECENT PROGRESS AND CURRENT CHALLENGES Since the end of TOGA, several new research themes have emerged that have deepened our understanding of ENSO and its impacts. This book is intended in part to review and elaborate on this rich history and to highlight areas where additional research is necessary. Below, we describe some recent advances and current challenges as

an introduction to a more complete and in‐depth review in subsequent chapters. ENSO variability arises through a complex mix of processes including seasonal time scale deterministic dynamics, nonlinearity, and noise forcing both in the atmosphere and ocean (Timmermann et  al., 2018, chapter  6). The most prominent form of noise or stochastic forcing is the atmospheric “westerly wind burst,” a sudden weakening of the trade winds along the equator lasting a few days to a few weeks that helps to trigger and amplify warm SST anomalies. Recently, “easterly wind surges,” the easterly counterpart to westerly wind bursts, have been identified to play a role in the ENSO cycle analogous to that of westerly wind bursts, particularly during the onset of La Niña events (e.g. Chiodi & Harrison, 2015). This weather noise forcing is not completely random but is “state dependent,” i.e. influenced by evolving ENSO conditions themselves (e.g. Eisenmann et al., 2005). In other words, westerly wind bursts tend to occur more frequently during El Niño and easterly wind surges during La Niña. The sources of this stochastic wind forcing and the physical mechanisms by which it influences ENSO evolution are an ongoing matter of debate. No two El Niño or La Niña events are exactly alike, and the differing anomaly spatial patterns exhibited among events is referred to as ENSO diversity (Figure 1.7; Capotondi et  al., 2015). While there is a continuum of longitudes where SST anomalies can peak, two commonly referred to end‐members are eastern Pacific (EP) events, i.e. those that have their largest anomalies in the eastern Pacific, and central Pacific (CP) events, those that have their largest anomalies in the central Pacific. (They are called by other names as well, as described in chapter 4). The 1997–1998 El Niño is an example of an EP El Niño, the 2009–2010 El Niño is an example of a CP El Niño, and the most recent major El Niño in 2015– 2016 is a hybrid between the two (Figure 1.7). El Niños exhibit greater diversity in spatial structure than La Niñas, and EP El Niños as a class are generally stronger than CP El Niños. These distinctions are important because ENSO impacts depend on where the SST anomalies are largest in the equatorial Pacific (Yeh et al., 2018; chapter 14). The detailed structure of ENSO SST anomalies determines the character of ENSO‐related shifts in tropical deep convection, rainfall, and atmospheric heating (Figure 1.7), and thus perturbations of the atmospheric general circulation and far‐field teleconnections to higher latitudes. Because climate impacts can vary between EP and CP events, predicting not only the sign of the SST anomaly but also its spatial structure is important (Santoso et al., 2019; chapter  10). So far, however, forecasting whether an El Niño event will be an EP or CP type more than a season in advance remains a challenge (e.g. Hendon et al., 2009).

INTRODUCTION  11 December Anomalies SST (°C), Winds (m s–1) and rainrate (mm hr–1) 20°N 10°N

4

(a) 2015

2 0

0°

–2

10°S

–4

20°S

5 m s–1

20°N 10°N

4

(b) 2009

2 0

0°

–2

10°S

–4

20°S 20°N 10°N

5 m s–1 4

(c) 1997

2 0

0°

–2

10°S

–4

20°S 20°N 10°N

5 m s–1 4

(d) 1998

2 0

0°

–2

10°S

–4

20°S 120°E

140°E

160°E

180°

160°W

140°W

120°W

100°W

80°W

60°W

Figure 1.7  Comparison of surface wind, SST, and precipitation anomalies for four Decembers during the boreal winter season when ENSO events typically reach peak development. Panels (a)–(c) show December 2015, 2009, and 1997 during the 2015–2016, 2009–2010 and 1997–1998 El Niños, respectively, and panel (d) shows December 1998 during the 1998–1999 La Niña. The 1997–1998 El Niño was an eastern Pacific (EP) event, while 2009–2010 was a central Pacific (CP) event. The 2015–2016 El Niño has characteristics of both types of El Niño, with large anomalies shifted to the west compared to an EP El Niño and larger anomalies in the eastern Pacific compared to a CP El Niño. Panel (d) illustrates how La Niña SST anomalies for strong cold events tend to be shifted westward compared to SST anomalies for strong warm events, as in 1997–1998 and 2015–2016. Differences in the structure of the precipitation anomalies account in part for the differences in far-field teleconnections between El Niño and La Niña events and between CP and EP El Niño events. Anomalies for all variables are computed relative to a mean seasonal cycle over the 20‐year period of 1998–2017, as determined by the length of the Multi‐Sensor Blended High‐Resolution Sea Surface Wind data set (Desbiolles et al., 2017). SST data are from Reynolds et al. (2007), rainrate data from the CPC Merged Analysis of Precipitation (Xie & Arkin, 1997).

This stems in part from the fact that state‐of‐the‐art climate models to varying degrees suffer from systematic errors (e.g. a mean equatorial cold tongue that is too cold and extends too far west) that limit their ability to accurately simulate ENSO diversity (chapter 9). El Niño and La Niña events are not simply mirror images of one another. El Niño events tend to be shorter lived (~9–12 months’ duration) and stronger than La Niña events, which sometimes occur over 2 successive years (see Figure 1.A2 in the appendix to this chapter). The spatial structure of the SST anomalies is also different, with cold

La Niña anomalies generally extending further west along the equator than warm El Niño anomalies (cf. Figure 1.7c, d). State‐dependent noise forcing, nonlinear ocean advection, nonlinear ocean‐atmosphere coupling, and other processes contribute to ENSO asymmetry and irregularity (chapter 7). Asymmetries in the evolution and structure of El Niño and La Niña events translate to differences in teleconnections and climate impacts (Hoerling et al., 1997; chapter 14), adding a layer of complexity for state‐of‐the‐art climate models in accurately simulating and predicting ENSO variability. Also, not well understood is what determines why

12  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

some rare El Niño and La Niña events grow to extreme amplitude, an aspect to ENSO irregularity that is of great societal consequence because the strongest events typically have the largest impacts. ENSO influences regions outside the Pacific via shifts in tropical deep convection and rainfall associated with the Hadley and Walker circulations (chapters 2 and 14), thermally forced atmospheric overturning cell in the north‐south and east‐west directions, respectively (Figure  1.6). Changes in the Pacific Hadley circulation affect the position and strength of the subtropical jet streams and associated storm tracks and are source of planetary scale atmospheric waves that drive far‐field teleconnections to higher latitudes in both the Northern and Southern Hemispheres (Trenberth et  al., 1998). Changes in the Pacific Walker Circulation on the other hand affect the tropical Atlantic and Indian Oceans via longitudinal shifts in positions of ascending and descending air masses. ENSO’s remote impacts also depend on event strength and type (e.g., EP or CP El Niño), as well as the background oceanic and atmospheric conditions in the tropics and elsewhere (Yeh et al., 2018). ENSO has consequences felt around the globe, but regions outside the tropical Pacific can in turn influence the evolution of ENSO events (chapter 11). Specifically, perturbations of the Walker Circulation originating in the tropical Atlantic and/or Indian Oceans can influence the tropical Pacific (Cai et  al., 2019), as can ocean‐ atmosphere interactions at higher latitudes (e.g. Di Lorenzo et  al., 2015; Liu & Di Lorenzo, 2018). These remote influences may contribute to ENSO diversity and the structural differences in development of individual events (e.g. Paek et al., 2017); they may therefore be a source of additional predictability for ENSO. However, assessing the relative importance of these various factors is challenging because of systematic errors in climate models and in some cases data records that are too short to identify statistically robust relationships between forcing outside the tropical Pacific and ENSO variability. ENSO variability is modulated on decadal timescales, with periods during which El Niños occur more frequently and with larger amplitude than La Niñas, or during which the reverse is true. Likewise, the relative frequency of EP and CP El Niños varies on decadal timescales (e.g. Yeh et  al., 2011; Newman et  al., 2011; McPhaden, 2012). These variations may simply occur by chance as the result of noise or chaotic nonlinearities in the climate system. Conversely, it is possible that decadal variations in background conditions associated with the Pacific Decadal Oscillation (Mantua et al., 1997; Newman et  al., 2016), or the very closely related Interdecadal Pacific Oscillation (Power et al., 1999), determine the variations in ENSO characteristics on decadal timescales

(Fedorov & Philander, 2000). However, Pacific decadal variability itself could be the residual of an asymmetric distribution of El Niños (CP or EP) and La Niñas in any given period of a decade or longer (Rodgers et al., 2004; McPhaden et al., 2011), rather than the result of intrinsically decadal timescale dynamical processes. The nature of this relationship between ENSO and decadal variations has major implications for our ability to make skillful climate predictions on interannual to decadal timescales in the tropical Pacific (­chapters 8 and  10). Thus, sorting out whether decadal variations in the tropical Pacific are the cause of, or the effect of, the decadal modulation of ENSO is an outstanding question of fundamental importance. Decadal variability in the Pacific affects the Earth’s global warming rate, as dramatically illustrated by the slowdown or “hiatus” in global surface warming during the first decade of the 21st century (Easterling & Wehner, 2009; Trenberth & Fasullo, 2013). This slowdown coincided with a decade‐long unprecedented cooling of the eastern tropical Pacific (Kosaka & Xie, 2013; England et al., 2014) that was clearly linked to decadal changes in the ENSO cycle (McPhaden, 2012). In particular, the period 2000–2012 was characterized by a series of weak CP El Niños and several strong and prolonged La Niñas. The hiatus ended with 3 successive warm years (2014– 2016) in the tropical Pacific, culminating in the major 2015–2016 El Niño (C. Zhang et al., 2019). SSTs in the central and western equatorial Pacific during the 2015–2016 El Niño were the highest on record, prompting the question of whether these record temperatures are a fingerprint of global warming on the ENSO cycle. While that is a possibility (Newman et al., 2018), the most likely explanation is that the western tropical Pacific has been steadily warming since the mid‐1950s, arguably as a response to anthropogenic GHG forcing (Cravatte et al., 2009; Seager et al., 2019). Warm anomalies associated with the major El Niño, superimposed on this decades‐long warming trend, are what pushed the SSTs in 2015–2016 into record territory (Santoso et al., 2017). These recent developments underscore the nece­ssity to advance our understanding of ENSO cycle variability and its impacts in the context of a global climate system that is undergoing profound change.

1.4. ENSO IN A CHANGING CLIMATE The near certainty of a human imprint on the Earth’s climate has emerged through the course of five assessments by the Intergovernmental Panel on Climate Change (IPCC) between 1990 and 2013-14. The planet is warming, and humans are responsible (IPCC, 2013). GHG

INTRODUCTION  13

concentrations in the atmosphere are higher now than at any time in the past 800,000 years based on ice core records—the result of fossil fuel combustion and deforestation in the 250 years since the beginning of the Industrial Revolution (Figure 1.8). The radiative effect of these gasses on the atmosphere‐ocean‐land system is to trap heat that would otherwise escape to space. Most of this heat imbalance is stored in the ocean, but some goes into raising global atmospheric temperatures. When a major El Niño occurs, there is an anomalous loss of heat from the ocean to the atmosphere, so that global mean surface air temperatures rise by 0.1–0.2 °C (Trenberth et al., 2002). Likewise, when a major La Niña occurs, the ocean takes up more heat from the atmosphere so that global mean surface air temperatures fall by a comparable amount. Thus, year‐to‐year variations in ENSO affect the global mean heat balance and are reflected in global mean surface temperature (GMST). As a result, 2016 was the warmest on record because of the combined effects of GHG forcing and El Niño (Figure 1.8). But can global warming, or more precisely anthropogenic GHG forcing, affect the ENSO cycle? If so, has it affected ENSO already? And how may ENSO change in a future world that is warmer because of climate change? With regard to the question of whether climate change has already affected ENSO, there are some suggestions, albeit inconclusive, from the instrumental record and

paleo‐climate data. The last 40 years have seen three extreme El Niños (1982–1983, 1997–1998, and 2015– 2016) unlike any comparable period in the nearly 150‐ year‐long instrumental record (Santoso et  al., 2017). However, even 150 years is too short to say that this behavior is evidence for an effect of climate change on ENSO, considering how much natural variability there is in the ENSO cycle (Wittenberg, 2009) and how poorly measured ENSO events were before 1950 (e.g. Giese et al., 2010). Paleo‐proxies (discussed in chapter  5) have the advantage of providing much longer records than the instrumental record and so can be valuable for inferring changes in ENSO properties over time, testing aspects of ENSO theory, and constraining model simulations of ENSO over the past 3 million years. However, high‐ quality paleo‐proxies relevant to ENSO studies are still relatively limited in their geographical distribution, and their interpretation is complicated by the convolution of biological, geochemical, and physical factors not related to climate. Thus, one cannot unambiguously conclude solely from analysis of paleo-reconstructions that anthropogenic GHG forcing has altered the ENSO cycle over the course of the Industrial Revolution. The global warming evident since preindustrial times to the present is expected to continue into the future, even if we were to stop emitting greenhouse gases into the atmosphere today. The ocean takes up more than 90% of

0.9

90 400 60 CO2

0.3

30 350

0.0

0

CO2 Pre-industrial temperature Pre-industrial CO2

–0.3

–0.6

Temperature CO2

Base period: 20th Century 1890

1920

1950

1980

320 ppmv

Temperature (°C)

0.6

–30

–60

280 Total ppmv

2010

Figure 1.8  Estimated changes in annual global mean surface temperatures (°C, bars) and CO2 concentrations (thick black line) over the past 149 years relative to 1901–2000 average values. The temperature scale (in °C) is shown on the left y‐axis and the CO2 concentration in parts per million by volume (ppmv) is shown on the right y‐axis. Carbon dioxide concentrations since 1957 are from direct measurements at Mauna Loa, Hawaii, while earlier estimates are derived from ice core records. CO2 concentrations are relative to the 20th century mean of 320 ppmv, while the temperature anomalies are relative to the 20th century mean of 13.9°C. Preindustrial estimated values for carbon dioxide of 280 ppmv are shown on the right. (Updated from Trenberth, https:// theconversation.com/the‐hottest‐year‐on‐record‐signals‐that‐global‐warming‐is‐alive‐and‐well‐53480)

14  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

the Earth’s energy imbalance (IPCC, 2013), and its vast thermal inertia implies a delayed equilibrium response to present‐day GHG forcing. Thus, there is more “warming in the pipeline” based on past emissions. Just how much warming, though, depends crucially on future emissions. The 2018 IPCC special report on pathways to 1.5°C GMST above preindustrial (IPCC, 2018) makes it clear that only by extraordinary coordinated international efforts undertaken now with a sense of urgency will we keep GMST below this level. For all future GHG emissions scenarios except those involving the most aggressive mitigation efforts, it is likely that by the end of the 21st century, GMST will rise 2–5  °C above the early 20th century (Knutti & Sedlácek, 2012; Figure 1.9). Thus, we should expect that climate change may affect ENSO in the future, but precisely how is uncertain. We have some clues, but to address this question we must rely on the IPCC class computer models, which are sensitive to spatial grid resolution and parameterized sub‐grid scale physical processes (chapter 9 provides an overview of ENSO modeling). The climate sensitivity of these models varies widely, and they are known to have in some cases severe systematic errors that limit our confidence in all but the most robust conclusions regarding ENSO in a warmer world. However, recent results suggest that under business‐as‐usual GHG emission scenarios (i.e. those scenarios that assume a high reliance on fossil fuels in the future with little or no effort to

limit CO2 emissions), EP ENSO variance may increase by the end of the 21st century by roughly 15% (Cai et  al., 2018), and extreme El Niño and La Niña events may double in frequency (Cai et al., 2014, 2015). It has been suggested that the frequency of CP El Niño events compared to EP El Niño events will increase under global warming (Yeh et al., 2009), though the robustness of this conclusion has been challenged (Yeh et al., 2011; Newman et al., 2011). Chapter 13 will address these issues in more depth. Regardless of whether GHG forcing has already affected ENSO, or whether it will in the future, there is evidence to suggest that the impacts of ENSO are being compounded by climate change even now simply because of the superposition of ENSO conditions on a warmer background state. This became most evident during the extreme ­2015–2016 El Niño with, for example, the record year for tropical Pacific cyclones in 2015 (W. Zhang et al., 2016), the unprecedented global coral bleaching event in ­2014–2016 (Hughes et al., 2017), and the extreme disruption of ecosystems and fisheries in the central Pacific in 2015–2016 linked to record high SSTs there (Brainard et  al., 2017). Extremes in Australian rainfall and global sea level likewise occurred during the strong 2010–2011 La Niña as a consequence of the combination of La Niña–induced warming in the western tropical Pacific plus regional SST trends due to GHG forcing (Fasullo et  al., 2013; Ummenhofer et  al., 2015). These factors combined to produce unusually high mois-

CMIP5 models, RCP scenarios 5

Global surface warming (°C)

4

Historical (42) RCP 2.6 (26) RCP 4.5 (32) RCP 6.0 (17) RCP 8.5 (30)

3 2 1 0 –1 1900

1950

2000

2050

2100

Year

Figure 1.9  Global mean surface temperature projections under different Representative Concentration Pathways (RCPs) based on Climate Model Intercomparison Project phase 5 (CMIP5) climate model simulations. Means are shown by solid lines and one standard deviation by shading. Numbers in parentheses indicate the number of CMIP5 models used to compute the means and standard deviations. Mean temperatures from model historical runs are shown in black (after Knutti & Sedlácek, 2012).

INTRODUCTION  15

ture transport by the trade winds in the western Pacific to feed exceptionally intense precipitation over Australia. The transfer of water mass from the ocean to land was extreme enough to lower global sea level by 5 mm (Boening et al., 2012) despite a background sea level rise of 3 mm yr‐1 over the past 25 years. As yet another example, the strong 2010– 2011 La Niña was also responsible for an unprecedented marine heatwave (the so‐called “Ningaloo Niño”) in a biodiversity hotspot off the coast of western Australia (Feng et al., 2013). Intensified Pacific trade winds drove an unusually strong flow of warm water from the Pacific to the Indian Ocean through the passages of the Indonesian seas during the La Niña. This warm water was transported southward along the coast of western Australia where, combined with multidecadal GHG‐forced ocean warming trends, SSTs in February–March 2011 rose 5°C above normal for 2 weeks. The heatwave led to major fish kills, severe coral bleaching, and significant disruption of the marine ecosystem in the region. As the climate system continues to warm, we can expect extremes related to ENSO like those described above to become more frequent (see chapters 16–20 for details). 1.5. CONCLUSION Bolstered by the extreme 2015–2016 El Niño, 2016 set a GMST record at 1.43°C above the 20th century average. The 5 warmest years on record have all occurred in the past 5 years (2014–2018). Also, in 2016 atmospheric CO2 concentrations exceeded 400 ppm all year long for the first time in the modern record, with an annual average of 402.9 ppm (Figure 1.8). Despite society’s best efforts to mitigate GHG emissions through international agreem­ ents such as the landmark 2015 Paris accord, GHG concentrations will continue to rise for the foreseeable future. We can be sure then that in this context of a changing global climate, nature will continue to deliver new surprises when it comes to ENSO. We have, for instance, witnessed three extreme El Niños in the past 40 years, namely, 1982–1983, 1997–1998, and 2015–2016 (Santoso et  al., 2017). The 1997–1998 El Niño (McPhaden, 1999), famously dubbed “the climate event of the 20th century” (Changnon, 1999), was the most extreme by many measures, but it may only be a matter of time before another surpasses it both in magnitude and impact. Given the tremendous consequences ENSO has on society and the environment and how it may change in response to anthropogenic GHG forcing, the urgency to understand ENSO in a changing climate is greater now than it has ever been. ENSO INDICES There are a variety of indices that are used to characterize the ENSO cycle, and here we describe those that are

most widely utilized. One set of indices relates to the ocean and is based on large areal averages of sea surface temperature anomalies, where anomaly is defined as the deviation from a 30‐year average climatological norm. The most commonly used regions to compute these anomalies are Niño‐1+2 (0°–10°S, 80°–90°W), Niño‐3 (5°N– 5°S, 90°–150°W), Niño‐3.4 (5°N–5°S, 120°–170°W), and Niño‐4 (5°N–5°S, 150°W–160°E) (Figure 1.A1). Niño‐1+2 captures variability close to the South American coast, Niño‐3 captures variability in the equatorial cold tongue of the eastern Pacific, and Niño‐4 captures variability further to the west in the warm pool. The Niño‐1+2, Niño‐3, and Niño‐4 index regions were designed by NOAA’s Climate Diagnostics Center (later called the Climate Prediction Center) in 1982 and have been in use for real‐time monitoring of evolving ENSO conditions ever since. The Niño‐3.4 region, which overlaps the Niño‐3 and Niño‐4 regions, was a later addition based on its high correlation with the Southern Oscillation Index (see below) and the strength of its correlation with remote ENSO‐related climate anomalies around the world (Barnston et  al., 1997). NOAA’s Oceanic Niño Index (ONI) uses a three‐month running average of Niño‐3.4 SSTs to track El Niño and La Niña events as they develop. The three‐month average is designed to filter out significant month‐to‐month variations that occur in the tropical Pacific in order to provide a clearer picture of evolving ENSO conditions. NOAA classifies an event as an El Niño when the ONI rises above 0.5°C for five consecutive months or more and as a La Niña when it falls below –0.5°C for five consecutive months or more. Another commonly used index designed to characterize the state of the atmosphere is the Southern Oscillation Index (SOI). This index is based on the anomalies from a 30‐year climatology of surface air pressure at Tahiti, French Polynesia (17° 39’S, 149° 28’W), minus that at Darwin, Northern Australia (12° 28’S, 130° 51’E) (Figures 1.A1 and 1.A2), after normalizing by the respective standard deviations at each station. Tahiti and Darwin are situated so as to capture the seesaw in atmospheric pressure between the Eastern and Western Hemispheres (Figure  1.5). The pressure difference between Tahiti and Darwin is a measure of trade wind strength since at low latitudes surface winds tend to flow down the pressure gradient. Thus, when the air pressure is high at Tahiti relative to Darwin (positive SOI), the trade winds are stronger than normal and when the air pressure at Tahiti is low relative to Darwin (negative SOI), the trade winds are weaker than normal. A measure of how strongly the ocean and atmosphere are coupled on ENSO timescales is the very clear anticorrelation of the SOI and the Niño‐3.4 SST (Figure 1.A2). These indices show that when the trade winds are anomalously weak (negative SOI), the central Pacific is anomalously

16  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

NINO3.4 NINO2

NINO4 Darwin

NINO3

NINO1

Tahiti

SOI

Figure 1.A1  Geographic distribution of ENSO index regions (boxes) and the location of the Tahiti and Darwin weather stations used to compute the Southern Oscillation Index from surface atmospheric pressure data. (Courtesy of the Australian Bureau of Meteorology) 3

Southern Oscillation Index

Standard Deviation

2 1 0 –1 –2 –3 3

Temperature (°C)

2

Nino 3.4 Sea Surface Temperature Anomalies (5°N–5°S, 120°W–170°W)

1 0 –1 –2 1950

1960

1970

1980

1990

2000

2010

Figure 1.A2  Niño‐3.4 SST anomalies and the Southern Oscillation Index, 1950–2019. Monthly values have been smoothed with a 5‐month running mean filter for clarity. Red peaks indicate El Niño events and blue troughs indicate La Niña events. Normal conditions (between ±0.5°C for Niño‐3.4 SST and between ±0.5 standard deviations for the Southern Oscillation Index) are unshaded.

warm (positive Niño‐3.4), conditions that define El Niño. Conversely, when the trade winds are anomalously strong (positive SOI), the central Pacific is anomalously cold (negative Niño‐3.4), conditions that define La Niña. The zero-lag

correlation between these two time series over the past 70 years (since 1950) is approximately –0.9, which is remarkable considering that the two indices, one oceanic and the other atmospheric, are derived completely independently.

INTRODUCTION  17

Other ENSO indices have been developed for specific purposes, such as the Cold Tongue Index based on SST anomalies in the region 6°N–6°S, 90°W–180° (Deser & Wallace, 1990) and the Multivariate ENSO Index, which uses a statistical combination of SST, air pressure and temperature, surface winds, and cloudiness (Wolter & Timlin, 1998). In addition, several indices have been developed to characterize the spatial diversity of ENSO SST patterns as described in chapter  4. These latter indices are often based on combinations of the more traditional Niño‐ indices, recognizing that no single index can describe the full range of variability that we observe in the ENSO cycle. ACKNOWLEDGMENTS Thanks to Rich Leach for Figure 1.2, Severine Fournier for help with the data in Figure 1.7, Kevin Trenberth for providing Figure  1.8, Catherine Ganter for providing Figure  1.A1, and Dai McClurg for general graphics support. Thanks also to Rodney Martinez for help with photographs of ENSO impacts in South America and to Rebecca Lindsey and Climate.gov for help with images of ENSO impacts in the United States. Two anonymous reviewers and Kevin Trenberth offered many valuable suggestions on how to improve earlier versions of this chapter. A. Santoso and W. Cai are supported by the Centre for Southern Hemisphere Oceans Research (CSHOR), a joint research center between QNLM and CSIRO, and the Earth Systems and Climate Change Hub of the Australian Government’s National Environmental Science Program. M. McPhaden is supported by NOAA. This is PMEL contribution no. 4949. REFERENCES Anonymous (1974). North Pacific research program. Bulletin of the American Meteorological Society, 55, 251–253. Barnston, A. G., Chelliah, M., & Goldenberg, S. B. (1997). Documentation of a highly related ENSO‐related SST region in the equatorial Pacific. Ocean‐Atmosphere, 35, 367–383. Bjerknes, J. (1966). A possible response of the atmospheric Hadley circulation to equatorial anomalies of ocean temperature. Tellus, 18, 820–829. doi:10.1111/j.2153‐3490.1966. tb00303.x Bjerknes, J. (1969). Atmospheric teleconnections from the equatorial Pacific. Monthly Weather Review, 97, 163–172. doi:10.1175/1520‐0493(1969)0972 .3.CO;2 Boening, C., Willis, J. K., Landerer, F. W., Nerem, R. S., & Fasullo, J. (2012). The 2011 La Niña: So strong, the oceans fell. Geophysical Research Letters, 39, L19602. doi:10.1029 /2012GL053055 Brainard, R. E., Oliver, T., McPhaden, M. J., Cohen, A., Veneers, R., Heenan, A., et al. (2017). Ecological impacts of the  2015‐2016 El Niño in the central equatorial Pacific.

Bulletin of the American Meteorological Society, 98, S21–S26. Cai, W., Borlace, S., Lengaigne, M., van Rensch, P., Collins, M., Vecchi, G., et al. (2014). Increasing frequency of extreme El Niño events due to greenhouse warming. Nature Climate Change, 4, 111–116. doi:10.1038 /nclimate2100 Cai, W., Wang, G., Santoso, A., McPhaden, M. J., Wu, L., Jin,   F.‐F., et  al. (2015). More frequent extreme La Niña events under greenhouse warming. Nature Climate Change, 5, 132–137. Cai., W., Wang, G., Dewitte, B., Wu, L., Santoso, A., et  al. (2018). Increased variability of eastern Pacific El Niño under greenhouse warming, Nature, 564, 201–206. doi: 10.1038/ s41586‐018‐0776‐9 Cai, W., Wu, L. Lengaigne, M., Li, T., McGregor, S., Kug, J.‐S., et al. (2019). Pan‐tropical climate interactions. Science, 363, eaav4236. doi: 10.1126/science.aav4236 Canby, T. Y. (1984). El Niño’s Ill Wind, National Geographic, 165(2), 144–183. Capotondi, A., et  al. (2015). Understanding ENSO diversity, Bulletin of the American Meteorological Society, 96, 921–938. Carrillo, C. N. (1892). Desertacion sobre las corrientes y estudios de la corriente Peruana de Humboldt. Bol. Soc. Geogr. Lima, 11, 72–110. Changnon, S.A. (1999). Impacts of 1997–98 El Niño generated weather in the United States. Bulletin of the American Meteorological Society, 80, 1819–1827. Chervin, R., Washington, W., & Schneider, S. (1976). Testing the statistical significance of the response of the NCAR general circulation model to North Pacific Ocean surface temperature anomalies. Journal of the Atmospheric Sciences, 33, 413–423. Chiodi, A. M., & Harrison, D. E. (2015). Equatorial Pacific easterly wind surges and the onset of La Niña events. Journal of Climate, 28, 776‐792. Cravatte, S. E., Delcroix, T., Zhang, D., McPhaden, M. J., & Leloup, J. (2009). Observed freshening and warming of the western Pacific warm pool. Climate Dynamics, 33, 565–589. Crutzen, P. J., & Stoermer, E. F. (2000). The Anthropocene. Global Change Newsletter, 41, 17–18. Davis, R. E. (1978). Predictability of sea level pressure anomalies over the North Pacific Ocean. Journal of Physical Oceanography, 8, 233–246. Desbiolles, F., Bentamy, A., Blanke, B., Roy, C., Mestas‐ Nunez,  A., Grodsky, S., et  al. (2017). Two decades (1992– 2012) of surface wind analyses based on satellite scatterometer observations. Journal of Marine Systems, 168, 38–56. http:// doi.org/10.1016/j.jmarsys.2017.01.003 Deser, C., & Wallace, J. M. (1990). Large‐scale atmospheric circulation features of warm and cold episodes in the tropical Pacific. Journal of Climate, 3, 1254–1281. Di Lorenzo, E., Liguori, G., Schneider, N., Furtado, J. C., Anderson, B. T., & Alexander, M. A. (2015). ENSO and meridional modes: A null hypothesis for Pacific climate variability. Geophysical Research Letters, 42, 9440–9448. doi: 10.1002/2015GL066281 Easterling, D. R., & Wehner, M. F. (2009). Is the climate warming or cooling? Geophysical Research Letters, 36, L08706. doi: 10.1029/2009GL037810

18  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Eisenman, I., Yu, L., & Tziperman, E. (2005). Westerly wind bursts: ENSO’s tail rather than the dog? Journal of Climate, 18, 5224–5238. England, M. H., McGregor, S., Spence, P., Meehl, G. A., Timmermann, A., Cai, W., et al. (2014). Recently intensified Pacific Ocean wind‐driven circulation and the ongoing warming hiatus. Nature Climate Change, 4, 222–227. doi: 10.1038/nclimate2106 Fasullo, J. T., Boening, C. Landerer, F. W., & Nerem, R. S. (2013). Australia’s unique influence on global sea level in 2010–2011. Geophysical Research Letters, 40, 4368–4373, doi: 10.1002/grl.50834 Fedorov, A. V., & Philander, S.G.H. (2000). Is El Niño changing? Science, 288, 1997–2002. Feng, M., McPhaden, M. J., Xie, S.‐P., & Hafner, J. (2013). La Niña forces unprecedented Leeuwin Current warming in 2011. Nature Scientific Reports, 3, 1277. doi: 10.1038/srep01277 Giese, B. S., Compo, G. P., Slowey, N. C., Sardeshmukh, P. D., Carton, J. A., Ray, S., & Whitaker, J. S. (2010). The 1918/19 El Niño. Bulletin of the American Meteorological Society, 91, 177–183, https://doi.org/10.1175/2009BAMS2903.1 Glantz, M. H. (2001). Currents of Change: Impacts of El Niño and La Niña on Climate and Society. Cambridge University Press, 252 pp. Halpern, D., Hayes, S. P., Leetmaa, A., Hansen, D. V., & Philander, S. G. H. (1983). Oceanographic observations of the 1982 warming of the tropical eastern Pacific. Science, 221, 1173–1175. Hendon, H. H., Lim, E., Wang, G., Alves, O. & Hudson, D. (2009). Prospects for predicting two flavors of El Niño. Geophysical Research Letters, 36, L19713, doi: 10.1029/ 2009GL040100. Hoerling, M. P., Kumar, A., and Zhong, M. (1997). El Niño, La Niña, and the nonlinearity of their teleconnections. Journal of Climate, 10, 1769–1786, https://doi.org/10.1175/1520‐0442 (1997)010 2.0.CO;2 Hughes, T. P., et al. (2017). Global warming and recurrent mass bleaching of corals. Nature, 543, 373–377. Hurlburt, H. E., Kindle, J. C., & O’Brien, J. J. (1976). A numerical simulation of the onset of El Niño. Journal of Physical Oceanography, 6, 621–631. IPCC (2013). Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker, T. F., Qin, D., Plattner, G.‐K., Tignor, M., Allen, S. K., Boschung, J., et al. (eds.)]. Cambridge University Press, Cambridge and New York, 1535 pp. IPCC (2018). Global Warming of 1.5°C. An IPCC Special Report on the impacts of global warming of 1.5°C above pre‐ industrial levels and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change, sustainable development, and efforts to eradicate poverty [Masson‐Delmotte, V., Zhai, P., Pörtner, H.‐O., Roberts, D., Skea, J., Shukla, P. R., et  al. (eds.)]. Available from https://www.ipcc.ch/sr15/ Knox, R., & D. Halpern (1982). Long range Kelvin wave propagation of transport variations in Pacific Ocean equatorial currents, Journal of Marine Research, 40, suppl., 329–339.

Knutti, R., & Sedlácek, J. (2012). Robustness and uncertainties in the new CMIP5 climate model projections. Nature Climate Change, 3, 369–373. Kosaka, Y., & Xie, S.‐P. (2013). Recent global‐warming hiatus tied to equatorial Pacific surface cooling. Nature, 501(7467), 403–407. Latif, M., Anderson, D., Barnett, T., Cane, M., Kleeman, R., Leetmaa, A., et al. (1998). A review of the predictability and prediction of ENSO. Journal of Geophysical Research, 103, 14,375–14,394. L’Heureux, M., et  al. (2017). Observing and predicting the 2015/16 El Niño. Bulletin of the American Meteorological Society, 98, 1363‐1382. Liu, Z., & Di Lorenzo, E. (2018). Mechanisms and predictability of Pacific decadal variability. Current Climate. Chane Reports, 4, 128–144. Mantua, N. J., Hare, S. J., Zhang, Y., Wallace, J. M., & Francis, R. C. (1997). A Pacific interdecadal oscillation with impacts on salmon production. Bulletin of the American Meteoro­ logical Society, 76, 1069–1079. McCreary Jr., J. P. (1976). Eastern tropical ocean response to changing wind systems, with application to El Niño, Journal of Physical Oceanography, 6, 632–645. McPhaden, M. J. (1999). Genesis and evolution of the 1997–98 El Niño. Science, 283, 950–954. McPhaden, M. J. (2012). A 21st century shift in the relationship between ENSO SST and warm water volume anomalies. Geophysical Research Letters, 39, L09706. doi: 10.1029/ 2012GL051826 McPhaden, M. J. (2015). Playing hide and seek with El Niño. Nature Climate Change, 5, 791–795. McPhaden, M. J., Busalacchi, A. J., Cheney, R., Donguy, J. R., Gage, K. S., Halpern, D., et  al. (1998). The tropical ocean‐ global atmosphere (TOGA) observing system: A decade of progress. Journal of Geophysical Research, 103, 14,169–14,240. McPhaden, M. J., Zebiak, S. E., & Glantz, M. H. (2006). ENSO as an integrating concept in Earth science. Science, 314, 1740–1745. doi: 10.1126/science.1132588 McPhaden, M. J., Busalacchi, A. J., & Anderson, D.L.T. (2010). A TOGA retrospective. Oceanography, 23, 86–103. McPhaden, M. J., Lee, T., & McClurg, D. (2011). El Niño and its relationship to changing background conditions in the tropical Pacific. Geophysical Research Letters, 38, L15709, doi: 10.1029/2011GL048275 Namias, J. (1969). Seasonal interactions between the North Pacific and the atmosphere during the 1960s. Monthly Weather Review, 97, 173–192. Neelin, J. D., Battisti, D. S., Hirst, A. C., Jin, F.‐F., Wakata, Y., Yamagata, T., & Zebiak, S. (1998). ENSO theory. Journal of Geophysical Research 103, 14,261–14,290. Newman, M., Shin, S.‐I., & Alexander, M. A. (2011). Natural variation in ENSO flavors. Geophysical Research Letters, 38, L14705, doi: 10.1029/ 2011GL047658 Newman, M., Alexander, M. A., Ault, T. R., Cobb, K. M, Deser, C., Di Lorenzo, E., et al. (2016). Pacific decadal oscillation revisited. Journal of Climate, 29, 4399–4427. Newman, M., Wittenberg, A. T., Cheng, L. Y., Compo, G. P., & Smith, C. A. (2018). The extreme 2015/16 El Niño, in the con-

INTRODUCTION  19 text of historical climate variability and change. Bulletin of the American Meteorological Society, 99(1), S16–S20. doi: 10.1175/Bams‐D‐17‐0116.1 Paek, H., Yu, J.‐Y., & Qian, C. (2017). Why were the 2015/2016 and 1997/1998 extreme El Niños different? Geophysical Research Letters, 44, 1848–1856. https://doi.org/10.1002/ 2016GL071515 Philander, S.G.H. (1990). El Niño, La Niña, and the Southern Oscillation. Academic Press, San Diego, CA, 293pp. Power, S., Casey, T., Folland, C., Colman, A, & Mehta, V. (1999). Inter‐decadal modulation of the impact of ENSO on Australia. Climate Dynamics, 15, 319–324. Rasmusson, E. M., & Carpenter, T. H. (1982). Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Monthly Weather Review 110, 354–384. Reynolds, R. W., Rayner, N. A., Smith, T. M., Stokes, D. C., & Wang W. (2002). An improved in situ and satellite SST analysis for climate. Journal Climate, 15, 1609–1625. Reynolds, R. W., Smith, T. M., Liu, C., Chelton, D. B., Casey,­K. S., & Schlax, M. G. (2007). Daily high‐resolution‐blended analyses for sea surface temperature. Journal Climate, 20, 5473–5496. https://doi.org/10.1175/2007JCLI1824.1 Rodgers, K. B., Friederichs, P. & Latif, M. (2004). Tropical Pacific decadal variability and its relation to decadal modulations of ENSO. Journal of Climate, 17, 3761–3774, doi:10.1175/1520-0442(2004)0172.0.CO;2. Ropelewski, C. F. & Halpert, M. (1987). Global and regional scale precipitation patterns associated with the El Niño/ Southern Oscillation, Monthly Weather Review, 115, 1606–1626. Santoso, A., McPhaden, M. J., & Cai, W. (2017). The defining characteristics of ENSO extremes and the strong 2015/16 El Niño. Reviews of Geophysics, 55, 1079–1129. Santoso, A., Hendon, H., Watkins, A., Power, S., Dommenget,  D., England, M. H., et  al. (2019). Dynamics and predictability of the El Niño‐Southern Oscillation: An Australian perspective on progress and challenges. Bulletin of the American Meteorological Society, 100, 403–420, https:// doi.org/10.1175/BAMS‐D‐18‐0057.1 Seager, R., Cane, M., Henderson, N., Lee, D.‐E., Abernathy, R., & Zhang, H. (2019). Strengthening tropical Pacific zonal sea surface temperature gradient consistent with rising greenhouse gases. Nature Climate Change, 9, 517–522. Stockdale, T. N., Busalacchi, A. J., Harrison, D. E., & Seager, R. (1998). Ocean modeling for ENSO. Journal of Geophysical Research, 103, 14,325–14,355. Timmermann, A., An, S.‐I., Kug, J.‐S., Jin, F.‐F., Cai, W., et al. (2018). El Niño Southern Oscillation Complexity, Nature, 559, 535–545. doi: 10.1038/s41586‐018‐0252‐6 Toole, J. M., & Borges, M. D. (1984). Observations of horizontal velocities and vertical displacements in the equatorial Pacific Ocean associated with the 1982/83 El Niño. Journal of Physical Oceanography, 14, 948–959. Trenberth, K. E., & Shea, D. J. (1987). On the Evolution of the Southern Oscillation, Monthly Weather Review, 115, 3078–3096.

Trenberth, K. E., Branstator, G. W., Karoly, D., Kumar, A., Lau, N. C., & Ropelewski, C. (1998). Progress during TOGA in understanding and modeling global teleconnections associated with tropical sea surface temperatures. Journal of Geophysical Research, 10, 14,291–14,324. Trenberth, K. E., Caron, J. M., Stepaniak, D. P., & Worley, S. (2002). El Niño–Southern Oscillation and global atmospheric temperatures. Journal of Geophysical Research, 107, 4065. doi: 10.1029/2000JD000298 Trenberth, K. E., & Fasullo, J. T. (2013). An apparent hiatus in global warming? Earth’s Future, 1(1), 19–32. https://doi. org/10.1002/2013EF000165 Ummenhofer, C. C., Sen Gupta, A., England, M. H., Taschetto,­A. S., Briggs, P. R., & Raupach, M. R. (2015). How did ocean warming affect Australian rainfall extremes during the 2010/2011 La Niña event? Geophysical Research Letters, 42, 9942–9951. Walker, G.T. (1924). Correlations in seasonal variations in weather. Part IX: A further study of world weather. Memoirs of the India Meteorological Department 24(4), 275–332. Walker, G. T., & Bliss, E. W. (1932). World weather. Part V. Memoirs of the Royal Meteorological Society 4(36), 53–84. Wittenberg, A. T. (2009). Are historical records sufficient to constrain ENSO simulations? Geophysical Research Letters, 36, L12702. doi: 10.1029/2009GL038710 Wolter, K., & Timlin, M. S. (1998). Measuring the strength of ENSO events: How does 1997/98 rank? Weather, 53, 315–324. Wyrtki, K. (1975). El Niño: The dynamic response of the equatorial Pacific Ocean to atmospheric forcing. Journal of Physical Oceanography, 5, 572–584. Wyrtki, K., Firing, E., Halpern, D., Knox, R., McNally, G. J., Patzert, W. C., et  al. (1981). The Hawaii‐Tahiti Shuttle Experiment. Science, 211, 22–28. Xie, P., & Arkin, P. A. (1997). Global precipitation: A 17‐year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bulletin of the American Meteorological Society, 78, 2539–2558. Yeh, S.‐W., Kug, J.‐S., Dewitte, B., Kwon, M.‐H., Kirtman, B. P., & Jin, F. F. (2009). El Niño in a changing climate. Nature, 461, 511–514. doi: 10.1038/nature08316 Yeh, S.‐W., Kirtman, B. P., Kug, J.‐S., Park, W., & Latif, M. (2011). Natural variability of the central Pacific El Niño event on multi‐centennial time‐scales. Geophysical Research Letters, 38, L02704. doi: 10.1029/2010GL045886 Yeh, S.‐W., Cai, W., Min, S.‐K., McPhaden, M. J., Dommenget, D., Dewitte, B., et al. (2018). ENSO atmospheric teleconnections and their response to greenhouse gas forcing. Reviews of Geophysics, 56, 185‐206. doi: 10.1002/2017RG000568 Zhang, C., Li, S., Luo, F., & Huang, Z. (2019). The global warming hiatus has faded away: An analysis of 2014–2016 global surface air temperatures. International Journal of Climatology. doi: 10.1002/joc.6114 Zhang, W., et  al. (2016). Influences of natural variability and anthropogenic forcing on the extreme 2015 accumulated cyclone energy in the western North Pacific. Bulletin of the American Meteorological Society, 97, S131–S135.

2 ENSO in the Global Climate System Kevin E. Trenberth

ABSTRACT In this chapter, the setting for ENSO is described, beginning with a brief introduction to the total climate system followed by a focus on the tropical Pacific and the mean annual cycle. These set the stage for ENSO to occur, and the main elements and components are introduced, highlighting why the tropical Pacific is so impor­ tant. ENSO is then described along with the main indices used to track it over time and the associated patterns in certain key fields, plus the diversity and complexity that can arise. Teleconnections and other modes of variability are then briefly introduced, including decadal variability in the tropical Pacific. Climate change ­influences are also presented, along with an introduction to the main impacts.

2.1. THE CLIMATE SYSTEM

mate and whether the main rain bands and storm tracks are shifted or not, especially in the tropics, where the rain patterns are especially distinctive. In turn, these displace­ ments lead to huge changes in heating patterns for the atmosphere through the latent heating associated with precipitation. This can drive large‐scale remote atmo­ spheric wave patterns, called teleconnections, with conse­ quences for the jet stream, storm tracks, and weather thousands of kilometers away. The following is adapted from Trenberth & Stepaniak (2004). The climate system consists internally of the atmos­ phere, ocean, cryosphere, and land, and externally of the sun and the orbit of the Earth around the sun, the distri­ bution of land and continents, the composition of the atmosphere and ocean, and volcanic eruptions that change the atmospheric composition episodically. On timescales of tens of thousands of years or longer, the continents drift, the mountains build and erode, the orbit of the Earth around the sun changes, the sun itself may change, and volcanoes may be more or less active. These give rise to paleo‐climate variations manifested as ice ages (glacials) and interglacials. Now human activities are changing the planet with magnitudes and speeds that are unprecedented and unequivocal (IPCC, 2013). The big­ gest global influence is the interference in natural flows of energy through the climate system due to changes in

There are huge variations in climate with the mean annual cycle as the sun moves from one hemisphere to the other. These are expected. The summer in one hemi­ sphere, with abundant sunshine and long days, forms the growing season while the winter in the other hemisphere can be a time of cold and snow, and short winter days. In the tropics, the seasons are less noticeable in terms of daylight hours and temperatures but are very perceptible in the wet and dry seasons. The summer monsoon rains develop with a risk of flooding and tropical storms, while the winter monsoon is dry and sunny. Because these are the normal sequence of events, we plan for them and can enjoy them. However, from time to time, the normal sequence breaks down owing to natural variability, the biggest of which is the El Niño pheno­ menon. It is often tempting to simply treat these pertur­ bations as linear additions or subtractions to the mean annual cycle, and this can work reasonably well for ­temperatures, but not so much for rainfall. The depar­ tures from normal depend enormously on the mean cli­

National Center for Atmospheric Research, Boulder, CO, USA

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 21

22  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

the composition of the atmosphere. The changes are caused mainly by fossil fuel burning, which gives rise to regional pollution plus global increases in carbon dioxide and other heat‐trapping greenhouse gases. Changes in land use and land cover are also widespread. Figure 1.8 of chapter 1 presents the changes in carbon dioxide and associated changes in global mean surface temperature (GMST) (after Trenberth, 2018), presented in a way to suggest a strong relationship between the overall changes over time. These links are well established through cli­ mate modeling. The main external influence on the climate is from the sun, and because of the sun‐Earth geometry most incoming solar (shortwave) radiation is in the tropics, while the outgoing infrared (longwave) radiation, which depends mostly on the fourth power of absolute temper­ ature (the Stefan‐Boltzmann Law), is more uniform. The incoming radiant energy is transformed into various forms (internal heat, potential energy, latent energy, and kinetic energy) moved around in various ways primarily by the atmosphere and oceans, stored and sequestered in the ocean, land, and ice components of the climate system, and ultimately radiated back to space as infrared radiation. The result is large temperature contrasts with latitude and a substantial transport of energy polewards by the atmosphere and ocean to offset the radiation imbalances. An equilibrium climate mandates a balance between the incoming and outgoing radiation and that the flows of energy are systematic. These drive the weather systems in the atmosphere, currents in the ocean, and fundamentally determine the climate. And they are perturbed due to climate change. The atmosphere is the most volatile component of the climate system. Winds in jet streams exceed 100 mph (160 km/h) or even 200 mph; winds move energy and water around. Of the thin envelope around the planet, 90% is within 15 km of Earth’s surface, 1/400 of the radius of Earth, so that clouds appear to hug the sur­ face from space. However, the atmosphere does not have much heat capacity. (Heat capacity depends on both the mass and the specific heat of a substance, which relates to how much heat it takes to raise the temperature by a given amount.) “Weather” occurs in the troposphere (lowest part), and weather systems (cyclones, anticy­ clones, cold and warm fronts, tropical storms/hurri­ canes) that move heat around mostly upwards and polewards. In mid‐latitudes, the warm and cold air flow side by side (southerlies are warm and northerlies are cold in the northern hemisphere). In the tropics, the main movement of energy is the overturning of circula­ tions as warm air rises and cold air sinks, giving rise to the Hadley Circulation (north‐south) or the Walker Circulation (west‐east). Major perturbations occur with El Niño.

Oceans cover 70.8% of the Earth’s surface. Water vapor from the surface ocean provides a source for rainfall and thus latent heat energy to the atmosphere. The heat capacity of the whole depth of the atmosphere is approx­ imately equivalent to that of 3.5 m of ocean (Trenberth & Stepaniak, 2004). The oceans slowly adjust to climate changes and can sequester heat for many years. The ocean is well mixed on average to a depth of about 20 m in summer and over 100 m in winter. An overall average of 90 m would delay any surface climate response to any perturbation by 6 years. The total ocean mean depth is about 3800 m. This would add a delay of 230 years if rap­ idly mixed. In reality, the response depends on the rate of ventilation of water through the thermocline (vertical mixing), and the estimate of delay overall is 10 to 100 years. Over 90% of the Earth’s energy imbalance (EEI) ends up in the ocean. (EEI is the net radiation into the Earth system given by the difference between the incoming solar radiation and the outgoing infrared radiation.) The ocean currents redistribute heat, fresh water, and dis­ solved chemicals around the globe through the currents that are often wind‐driven but that also have a density thermo‐haline component. These result in the large‐scale conveyer belt current systems that move heat and fresh water and salts around. El Niño produces huge changes throughout the tropics both in currents and heat (e.g. Cheng et al., 2019). For land, the heat penetration with the annual cycle is only about 2 m, and the heat capacity of land is much less than water: the specific heat of land is a factor of 4½ less than sea water, and for moist soil it may be a factor of 2. Hence, land plays a lesser role than oceans in storing heat. Consequently, surface air temperature changes over land are large and occur much faster than over the oceans. Land has an enormous variety of features: topography, soils, vegetation, slopes, and water capacity. Land sys­ tems are highly heterogeneous on small spatial scales. Changes in soil moisture affect the disposition of heat through changes in temperature versus evaporation. Changes in land surface and vegetation affect the climate through albedo, roughness, and evapotranspiration. Accumulation of EEI effects mainly occurs during dry spells, leading to wilting plants and increased risk of  wildfire, and these effects are more pervasive during El Niño. For the major ice sheets, e.g. Antarctica and Greenland, penetration of heat occurs primarily through conduction. Hence, the mass involved in changes from year‐to‐year is modest but important on multidecadal timescales. Unlike earth, land ice melts and contributes to changes in sea level. Ice volumes on Earth are about 28,000,000 km3 water in ice sheets, ice caps, and glaciers. Most ice is in the Antarctic ice sheet, which, if melted, would increase sea level by ~58 m, vs 7 m from Greenland and 0.41 m from

ENSO in the Global Climate System  23

the other glaciers and ice caps (Bahr et al., 2009; IPCC, 2013). In the Arctic the sea ice is up to 3–4 m thick, while around Antarctica, where it can spread out, it is up to 1–2 m thick. Ice is bright and reflects the solar radiation leading to an ice‐albedo feedback. When ice decreases in extent, less solar radiation is reflected, leading to warmer conditions and thus even less ice. Arctic sea ice has decreased by over 40% in September since the 1970s as one consequence of the EEI. The West Antarctic Ice Sheet (WAIS) is partly grounded below sea level. Warming could alter grounding of the ice sheet, making it float and become vulnerable to rapid (i.e. centuries) disintegration with a rise in sea level of 4–6 m. It may be irreversible if collapse begins. 2.2. THE TROPICAL PACIFIC AND MEAN ANNUAL CYCLE To set the stage for understanding the main variability in the climate system, Figure  2.1 presents the mean sea surface temperatures (SSTs) for January and July, along with the corresponding mean total column water vapor (TCWV) and precipitation (adapted from Trenberth, 2011). Figure  2.1 reveals the migration of the warmest SSTs greater than 28°C back and forth across the equator in the continental regions, but not over the Pacific or Atlantic Oceans. The TCWV in the atmosphere follows the SSTs very closely. Figure 2.2 introduces some key features in the tropics that are fully engaged in El Niño‐Southern Oscillation (ENSO) variability. These are the monsoons over Africa, southern Asia‐Australia, and the Americas, that experi­ ence huge annual cycles, while over the oceans the Intertropical Convergence Zone (ITCZ) and South Pacific Convergence Zone (SPCZ) have much more mod­ est changes. The ITCZs remain in the northern hemi­ sphere year‐round over the open ocean, away from land masses. In contrast, the warm pool of high SSTs in the western Pacific–Indonesian region migrates strongly across the equator accompanied by the highest values in TCWV (Figure 2.1). The peak rainfall follows the same track, but the contrasts are much greater spatially as huge ocean deserts are evident over the eastern Pacific and Atlantic in the downward branches of Walker Circulation and local Hadley cells (see Figure 1.6, chapter 1). Surface trade winds, the southeast trades over and south of the equator and the northeast trades north of the ITCZ, con­ verge moisture over the highest SSTs, leading to strong convection and upward motion overall, while the ­downward branch of the Walker Circulation suppresses rainfall. Figures  2.1 and 2.2 also introduce the cold tongue along the equator evident in the SST field, ­ magnified in the rainfall field by the deficit there (see also Figure 1.6 of chapter 1).

In the tropical ocean domains, the atmosphere and ocean are very strongly coupled. The surface winds drive surface ocean currents, which determine where the sur­ face waters flow and diverge, and thus where cooler nutrient‐rich waters upwell from below. Because of the Earth’s rotation, easterly winds along the equator deflect currents to the right in the Northern Hemisphere and to the left in the Southern Hemisphere and thus away from the equator, creating upwelling along the equator, as well as away from the coast of the Americas. Thus, the winds largely determine the SST distribution, the differential sea levels, and the heat content of the upper ocean (e.g. McPhaden, 2012; Cheng et  al., 2019). The presence of nutrients and sunlight in the cool surface waters along the equator and western coasts of the Americas favor development of tiny plant species (phytoplankton), which in turn are grazed on by microscopic sea animals (zoo­ plankton), which provide food for fish. In the equatorial Pacific Ocean, the result is a very deep upper ocean mixed layer, and sea level is high in the west­ ern tropical Pacific warm pool, as waters driven by the easterly trade winds pile up. Hence, the thermocline is about 150 to 200 m deep (Figure 2.3), while in the eastern Pacific upwelling cool waters result in a shallow thermo­ cline only 40 m deep on average, and sea level is relatively low. The Pacific sea surface slopes up by about 60 cm from east to west along the equator on average. Another outcome is that there is a flow of warm waters into the tropical Indian Ocean known as the Indonesian Throughflow (Sprintall et  al., 2014), and this is quite complex owing to the small islands and varying depths of channels. In Figure  2.3 the temperatures along the equator in the Pacific for December 1996, a fairly normal time, are shown with the 20°C isotherm highlighted as approximating the center of the thermocline: the strong temperature gradient in the vertical between the upper mixed layer and the deep cold abyss. A schematic of the tropical Pacific Ocean for the top 200 m (Figure  2.4) shows the relationships among the thermocline, cold tongue, and the northeast and the southeast trade winds. The South Equatorial Current extends well into the Southern Hemisphere and is strong near the equator, the North Equatorial Current is north of about 10°N, and the North Equatorial Countercurrent develops in between, north of the equator. The equatorial undercurrent is strong along the equator below the sur­ face, peaking at about 1 m s‐1 at 100 m depth from 120 to 140°W (Johnson et  al., 2002). The trade winds create divergence at the surface owing to the Earth’s rotation, as Ekman outflow, and this is compensated for by geo­ strophic inflow at greater depths. The surface winds drive the ocean currents, but the result depends mostly on the curl of the wind stress, as given by the Sverdrup equation (Sverdrup, 1947), and hence the countercurrent results

24  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE January

July

SST

18.2°C

1

5

9

13

17

21

25

TCWV

6

18.3°C

29

1

5

9

13

17

21

25.8 mm

23.1mm

12

18

24

30

36

42

48

54

6

12

18

24

30

36

42

48

2.7 mm/d

P

0.5

1

1.5

2

3

4

5

7

9

29

25

54 2.7 mm/d

0.5

1

1.5

2

3

4

5

7

9

Figure 2.1  Mean sea surface temperature (SST), total column water vapor (TCWV), and precipitation for January (left panels) and July (right panels). SSTs are from Smith et al. (2008) for 1979–2008; TCWV is from SSM/I over ocean and National Centers for Environmental Prediction reanalyses over land for 1987–2008; and precipitation is from the Global Precipitation Climatology Project Ver. 2.1 for 1979–2008 based on Huffman et al. (2009). Top right: global mean. Adapted from Trenberth (2011).

from the minimum in easterlies. Poleward of 25°N and 25°S, the prevailing westerly winds induce Ekman trans­ ports towards the equator, generating a convergence bet­ ween 15–30°N and 15–30°S, with high sea levels around 10–15°N and 10–15°S, respectively (D. Hu et al., 2015). Although January and July represent extremes in tem­ perature over land, following a few weeks behind the sol­ stice, the SSTs rise until the end of summer, March in the south and September in the north. The cold tongue is

strongest around September‐October of each year, when the ITCZ is at its farthest north, nearly 10°N, and the southeast trades along the equator are strong. Hence, this corresponds to the time when the northern oceans are warmest and the southern oceans are coldest. On the northern side of the cold tongue, there are frequently tropical instability waves that feed on the temperature gradients and act to transport ocean heat to reduce the cold tongue strength (e.g. Timmermann et  al., 2018).

ENSO in the Global Climate System  25 Monsoons

P

SPC

SPC

Z

Z

0.5

1

1.5

2

3

ITCZ

ITCZ

ITCZ

4

5

7

9

0.5

1

1.5

2

3

4

5

7

9

Figure 2.2 Schematic precipitation for January (left) and July (right) mean fields to illustrate the Intertropical Convergence Zone (ITCZ), South Pacific Convergence Zone (SPCZ), and monsoon rains, plus the cold tongue and associated dry zone along the equator.

In contrast, in March–April, the SSTs are warmest along the equator, the trade winds are weaker, and the cold tongue is less pronounced. Consequently, at this is the time of year, modest SST anomalies alter where the high­ est SST values occur, thereby shifting rainfall patterns in ways that can initiate a change in phase of ENSO. It is clear that the average and equilibrium state of the tropical Pacific is one that involves a fairly delicate balance between the trade winds piling up warm water in the west and the ocean currents that tend to flow in the other direction, but with considerable complexity related to the pattern of winds and their gradients. The tropical Pacific, therefore, is a region where the atmospheric winds are largely responsible for the tropical SST distribution which, in turn, is very much involved in determining the precipitation distribution and the tropical atmospheric circulation. However, there is really no such thing as

average conditions, and instead the tropical Pacific tends to sway back and forth between the El Niño and La Niña states. The stage is set for ENSO to occur. 2.3. EL NIÑO-SOUTHERN OSCILLATION El Niño events are not uncommon. Every three to seven years or so, a pronounced warming occurs of the surface waters of the tropical Pacific Ocean. The warm­ ings take place from the International Dateline to the west coast of South America and result in changes in the local and regional ecology, clearly linked with anomalous global climate patterns (see Trenberth, 2017). The warm­ ings have come to be known as “El Niño events.” Historically, “El Niño” referred to the appearance of unusually warm water off the coast of Peru, where it was readily observed as an enhancement of the normal

Nov-Dec-Jan 1996/97

2°S to 2°N

0

32

Depth m

24 20

200

20 16

300

12

Temperature °C

28 100

8

400

4 0

500 140 °E

160

180

160 °W 140

120

100

Figure 2.3 Cross section along the equator of temperatures within the ocean with depth from the Tropical Atmosphere‐Ocean (TAO) array (points given by dots) for November‐December‐January 1996–1997. The region from approximately 17 to 23°C, centered on the 20°C isotherm, is indicated as the main thermocline. Data downloaded 15 August 2018, courtesy GTMBA Project Office of NOAA/PMEL, https://www.pmel.noaa.gov/tao/ drupal/disdel/.

26  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Equ

Tropical Pacific Ocean

ato

r

Sou

Depth

th E

qua

tori a

No Courth Eq nte uato rCu r rren ial rren t t

l Cu

Nor

th E Cur quato ren t rial

Ekkm E maann Ouuttff O lloow w

Col

d To

Geo stro phi c In flow

Equ

ato

ngu

rial U

e

nde

rcu rren t

S

W

N E

Figure 2.4  Schematic of the tropical Pacific Ocean for the top 200 m. The equator is the dotted black line. Shown are the thermocline, cold tongue, and the northeast and the southeast trade winds (black arrows). The latter extend to and across the equator, and the trade winds come together around 5 to 10°N in the ITCZ (not shown). For ocean currents, the South Equatorial Current extends well into the southern hemisphere and is strong near the equator, the North Equatorial Current is north of about 10°N, and the North Equatorial Countercurrent develops in between, north of the equator. The equatorial undercurrent is strong along the equator below the surface, peaking at about 1 m s‐1 at 100 m depth from 120 to 140°W. The trade winds create divergence at the surface owing to the Earth’s rotation, given here as Ekman outflow (speckled blue), and compensated for by geostrophic inflow at greater depths (speckled brown). The surface winds drive the ocean currents, but the result depends mostly on the curl of the wind stress, as given by the Sverdrup equation, and hence the countercurrent results from the minimum in easterlies.

warming about Christmas (hence, El Niño, Spanish for “the Christ child”), and only more recently has the term come to be regarded as synonymous with the basinwide phenomenon (Bjerknes, 1969; Trenberth, 1997), see also chapter 1. The oceanic and atmospheric conditions in the tropical Pacific are seldom close to average, instead fluc­ tuating somewhat irregularly between the warm El Niño phase of ENSO and the cold phase of ENSO, consisting of basinwide cooling of the tropical Pacific, dubbed “La Niña events” (La Niña is “the girl” in Spanish). The most intense phase of each event typically lasts part of a year. The Southern Oscillation (SO) is principally a global‐ scale seesaw in atmospheric sea level pressure involving exchanges of air between eastern and western hemi­ spheres (see Figure  2.5a), centered in tropical and sub­ tropical latitudes with centers of action located near Indonesia and the tropical South Pacific Ocean (near Tahiti). Thus, there are the inverse variations in pressure anomalies (departures from average) at Darwin (12.4°S, 130.9°E) in northern Australia and Tahiti (17.5°S, 149.6°W) in the South Pacific Ocean, whose annual mean pressures are strongly and significantly oppositely corre­ lated (Trenberth, 1976; Trenberth & Caron, 2000). During an El Niño event, the sea level pressure tends to be higher

than usual at Darwin, northern Australia, and lower than usual at Tahiti in the southeast Pacific. While changes in near equatorial Pacific SSTs can occur without a swing in the SO, El Niño and the SO are linked so closely that the term ENSO is used to describe the atmosphere‐ocean interactions over the tropical Pacific. Warm ENSO events, therefore, are those in which both a negative SO extreme and an El Niño occur together (Figure 2.6). Consequently, the difference in pressure anomalies, Tahiti–Darwin, appropriately weighted, is often used as a Southern Oscillation Index (SOI) (Trenberth, 1984; Figure  2.6). The correlations with SST and surface air temperature (Figure 2.5b) with the SOI then show the El Niño patterns. To a first approximation, these capture much of what goes on, but the ENSO swings are not symmetric, as shown in Figure  2.6; El Niño departures from average can be much bigger than for La Niña, while La Niñas may last longer or come in double bursts. In fact, to describe the different flavors of El Niño, at least two patterns are required, the first related to mean anom­ alies in SST east of the Dateline, the so‐called Niño‐3.4 region SST anomalies from 5°N–5°S, 170–120°W, and the second involving SST gradients from west to east, summarized by the Trans Niño Index (Trenberth &

ENSO in the Global Climate System  27 Correlations SOI

SLP

N1+2 N3.4

Sfc T

Precip –0.9 –0.6 –0.3

0

0.3

0.6

0.9

Figure 2.5  Correlations with the Southern Oscillation Index (SOI), based on normalized Tahiti minus Darwin sea level pressures, for annual (May to April) means for sea level pressure (top) and surface temperature (center) for 1958–2004, and GPCP precipitation for 1979–2003 (bottom). On panel 2, the Niño 1+2 and 3.4 areas are identified. (Adapted from Trenberth et al., 2007).

Stepaniak, 2001). The latter involves contrasting normal­ ized values in the Niño 4 and Niño 1+2 regions. There are also intriguing lead‐lag relationships (Trenberth & Shea, 1987; Mamalakis et al., 2018; Cheng et al., 2019).

Higher than normal pressures are characteristic of more settled and fine weather, with less rainfall, whereas lower than normal pressures are identified with “bad” weather, more storminess and rainfall (Figure  2.5c). Thus, for El

28  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE 4 Southern Oscillation Index: Tahiti - Darwin

Standardized

2

0

–2

–4

1950

1960

1970

1980

1990

2000

2010

2020

97/98 82/83 El Niño Niño 3 2.5

Very strong

15/16

Strong

Niño 1+2 °C

2 1.5 1 0.5

Moderate Weak

0 –0.5 –1 Base 1950-79 –1.5 1950 1955 1960 1965 1970

La Niña 1975 1980 1985 1990 1995

3 2.5 2 ONI Index °C

1.5 1 0.5

Very strong

82/83

El Niño

2000 2005 2010

97/98

2015

15/16

Strong Moderate Weak

0 –0.5 –1 –1.5 –2

Weak Moderate Strong

–2.5 1950 1955 1960 1965 1970

La Niña 1975 1980 1985 1990 1995

2000 2005 2010

2015

Figure 2.6 Top: SOI based on Trenberth (1984). Anomalies of monthly sea level pressure at Tahiti normalized minus those at Darwin normalized, and then with the result normalized. An 11‐term filter has been applied to the monthly data along with a decadal filter. Red signifies El Niño conditions. Lower: Time series of SST anomalies from 1950 through 1917 relative to the means of 1950–1979 for (i) the traditional El Niño area Niño 1+2: 0°–10°S 90°W–80°W, and (ii) the region most involved in ENSO Niño‐3.4: 5°N–5°S, 170°E–120°W, labeled ONI (Oceanic Niño Index). Data from NOAA.

ENSO in the Global Climate System  29

Niño conditions, higher than normal pressures over Australia, Indonesia, southeast Asia, and the Philippines signal drier conditions or even droughts. Dry conditions also prevail at Hawaii, parts of Africa, and extend to the northeast part of Brazil and Colombia. On the other end of the seesaw, excessive rains prevail over the central and eastern Pacific, along the west coast of South America, parts of South America near Uruguay, and southern parts of the United States in winter, often leading to flooding. When the pressure pattern in Figure 2.5 reverses in sign, as for La Niña, the regions favored for drought in El Niño tend to become excessively wet, and vice versa. Naturally, the changes at the surface in the atmosphere have consequences for the atmospheric circulation. In particular, in association with El Niño, there are substantial changes in the west‐east overturning Walker Circulation near the equator (Figures 1.1 and 1.6, chapter 1). During the warm phase of ENSO, the warming of the waters in the central and eastern tropical Pacific shifts the location of the heaviest tropical rainfall eastward toward or beyond the Dateline from its climatological position centered over Indonesia and the far western Pacific, weakening the Walker Circulation. This shift in rainfall also alters the heating pat­ terns that force large‐scale waves in the atmosphere (sec­ tion 2.4). The waves in the airflow determine the preferred location of the extratropical storm tracks. Consequently, changes from one phase of ENSO to another have pro­ found impacts on regional temperatures (Figure 2.5). Although ENSO has a typical period of two to seven years, the strength of the oscillation has varied considerably (Figure 2.6). There were strong variations from the 1880s to the 1920s and after about 1950, but weaker variations in between (with the exception of the major 1939–1941 event; Trenberth, 1976). A remarkable feature of the SOI is the decadal and longer‐term variations that have become more pronounced since about 1970 (Figure 2.6; see Trenberth, 1990). As discussed below, several indices exist for ENSO events. However, most widely used is that El Niño and La Niña events are usually defined as when the Niño‐3.4 value exceeds a threshold (e.g. above and below 0.5°C), while in‐between values are considered as neutral (Trenberth, 1997). Hence, El Niño events clearly identifi­ able in Figure 2.6 since 1950 occurred in 1951, 1953, 1957– 1958, 1963, 1965, 1969, 1972–1973, 1976–1977, 1982–1983, 1986–1987, 1990–1995, 1997–1998, 2002–2003, 2004– 2005, 2006–2007, 2009–2010, and 2015–2016 (Trenberth, 1976, 1997; Santoso et  al., 2017). The 1990–1995 event might also be considered three modest events where the conditions in between failed to return to below normal so that they merged together and formed the longest El Niño on record, raising the specter of climate change (Trenberth & Hoar, 1996). Worldwide climate anomalies lasting sev­ eral seasons have been identified with all of these events. There have been three “super” El Niños, whereby the main index has made it into the “very strong” category:

1982–1983, 1997–1998 and 2015–2016 (Figure  2.6; see also Santoso et  al., 2017; L’Heureaux et  al., 2017; Newman et al., 2018). The 1997–1998 event was the larg­ est on record in terms of SST anomalies, and the GMST in 1998 was the highest on record last century, but for the SOI, the El Niño event of 1982–1983 still holds the record (Figure 2.6). There are no “very strong” La Niña events in Figure 2.6, which also highlights some aspects of the asymmetry between the two phases: La Niña events tend to last longer or be double‐phased more often. The effects of the 2015–2016 event were not as great in coastal South America. Nevertheless, a very unusual coastal El Niño occurred in the first few months of 2017 (Garreaud, 2018; Z. Z. Hu et al., 2018), causing devastating stormy weather over northern Chile, Peru, and Colombia. Under normal conditions, and even more so with La Niña, strong trade winds pile up warm waters in the western tropical Pacific (Figure  2.4), with profound effects on the thermocline. During El Niño, the trade winds weaken, which causes the thermocline to become shallower in the west and deeper in the eastern tropical Pacific (Figure 2.7), while sea level falls in the west and rises in the east by as much as 25 cm as warm waters surge eastward along the equator. Equatorial upwelling decreases or ceases and so the cold tongue weakens or disappears, and the nutrients for the food chain are substantially reduced. The resulting increase in sea temperatures warms and moistens the overlying air so that convection breaks out, and the convergence zones and associated rainfall move to a new location with a resulting change in the atmospheric circulation (sec­ tion  2.4). A further weakening of the surface trade winds completes the positive feedback cycle, leading to an El Niño event, known as the Bjerknes feedback (see chapter 6). Although ENSO has a clear preferred timescale, with a period of 2 to 7 years, every event is different. It has been said that El Niño comes in many different flavors (Trenberth, 1994). This has also been referred to as ENSO diversity (Capotondi et  al., 2015) or complexity (Timmermann et  al., 2018) (see also chapters 4 and 7). One form of variability arises in the tropical Pacific from just how the SSTs develop, where the warmest water lies, and what SST gradients emerge. The warmest water determines roughly where the main convection occurs, while the gradients determine the anomalous trade wind strength, and thus the strength of the anomalous convection. In 1997–1998 the negative SST anomalies in the tropical western Pacific were clear, but they were largely missing in 2015–2016, with the result that the anomalous teleconnections were not as strong in the later event in spite of similarly high SST anomalies (Figure 2.6). Much like the different flavors in the tropical Pacific, the teleconnections to higher latitudes occur through an ever‐changing chaotic atmosphere involving vigorous

30  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Temperature NDJ 5°N-5°S 30

28

22

20 18 16

150

Indonesia

Depth (m)

100

14

200

28

26

24

22 20 18 16

14

12

60E

90E

120E

150E

28

14

26 24 22

2010

300 0

30 16

12

250 12

2 182022 24 6

28 26

26 4 2

50

Americas

0

180

150W

120W

90W

60W

30W

0

18

30

28

20

150 200

26 24

Indonesia

18 16

14

16

26

22 20 18

22 1618 20 24

22

100 Depth (m)

28

Americas

26 24

50

14

14

12

16

10

14

°C

12

250

20

12 12

2015

300 60E

90E

120E

150E

180

150W

120W

90W

60W

30W

0

Figure 2.7  A cross section from 5°N to 5°S along the equator illustrates the temperature structure °C with depth for an El Niño in 2015 versus a La Niña in 2010 for NDJ. The thermocline is the temperature gradient centered on about the 20°C isotherm, highlighted by the dark black line. The white spaces denote the land masses. (Courtesy L. Cheng, http://159.226.119.60/cheng/).

baroclinic weather systems, jet streams, and frontal systems (section  2.4 and chapter  14). Even with identical SST anomalies, the response can be quite different, just by chance, over a particular season (Trenberth, 1994; Deser et al., 2017). Yet El Niño events do not last long enough to sample all weather variability and thus provide stable statistics for each event. Timmermann et  al. (2018) address some aspects of this complexity but do not address the importance of the changes in rainfall and thus latent heating that is so important for teleconnec­ tions (section 2.4; see also chapter 14). The traditional ENSO indices (Figure 2.6) identify either with the ocean (e.g. Niño‐3.4) or atmosphere (SOI), and accordingly these have many applications. In particular, SSTs and the ocean conditions can be specified to force the atmosphere in a model to adjudicate the response; or the atmospheric winds can be used to drive an ocean model to determine the response there. Wolter & Timlin (2011) dis­ cuss the desirability and formulation of a Multivariate ENSO Index, which initially included information from sea level pressure, zonal and meridional surface wind com­

ponents, SST, surface air temperature, and cloudiness, and varied with month of the year. However, the availability, quality, and homogeneity of the records of several of these other variables is questionable as one goes back in time, and a reduced MEI was formulated using sea level pressure and SST alone. L’Heureux et al. (2015) explored how well ENSO could be captured by simple indices and found that a combined index using SST and outgoing longwave radi­ ation was useful at certain times of the year. As the nonlin­ earity of precipitation associated with SSTs is not well captured by anomalies alone, a new index based on the longitude of tropical deep convection from 5°N to 5°S, the ENSO Longitude Index, has been proposed (Williams & Patricola, 2018). This index allows the diversity of ENSO to be captured with a single number that accounts for non­ linearity of convective response. The character of recent ENSO events seems to be differ­ ent than before 1976 (Trenberth, 1990; Trenberth & Hoar, 1997; Capotondi & Sardeshmukh, 2017; Newman et al., 2018). Several events have had a lot more activity in the central Pacific, in which the traditional South American

ENSO in the Global Climate System  31

coastal temperature anomaly is not as affected, but instead an anomaly arises in the central Pacific (Niño‐3.4), called El Niño Modoki (Ashok et al., 2007). Modoki is Japanese for "similar but different" and is related to the role of gra­ dients across the Pacific described by the Trans Niño Index, and the so‐called Central Pacific (CP) El Niño events as compared with the more traditional East Pacific (EP) events. However, this pattern is also part of the evo­ lution of ENSO, and it is unclear whether this is simply part of the continual variety (the different flavors of El Niño) (e.g. Johnson, 2013; Jadhav et al., 2015), or whether it might be part of a climate change signal (e.g. Yeh et al., 2009; McPhaden et al., 2011). Because of the enhanced activity in the Pacific and the changes in atmospheric circulation throughout the tropics, there is a decrease in the number of tropical storms and hurricanes in the tropical Atlantic during El Niño (e.g. Goldenberg et al., 2001; Bell & Chelliah, 2006). Good examples are 1997 and 2015, with 1997 being one of the quietest Atlantic hurricane seasons on record. In contrast, the El Niño events of 1990–1995, 1997–1998, and 2015–2016 terminated before the 1995, 1998, and 2017 hurricane seasons, which unleashed storms and placed those seasons among the most active on record in the Atlantic (Trenberth et  al., 2018). In 2015, super typhoon Pam ripped through Vanuatu in March, causing enormous damage, enabled by warm waters from El Niño. Less than a year later, the strongest hurricane on record in the Southern Hemisphere (Winston) severely damaged Fiji. The 2015 northern hurricane season fea­ tured by far the greatest number of category 4 and 5 hur­ ricanes/typhoons on record (25 vs previous record 18). Chapter 17 covers tropical cyclones in more detail. ENSO events involve large exchanges of heat between the ocean and atmosphere and affect GMST (Figure 1.8, chapter  1) (Trenberth et  al., 2002; Mayer et  al., 2014, 2018). Extremes of the hydrological cycle such as floods and droughts are common with ENSO (chapter 16) and are apt to be enhanced with global warming (Trenberth, 2011). For example, the modest 2002–2003 El Niño was associated with a drought in Australia, made much worse by record‐breaking heat. A strong La Niña event took place 2007–2008, contributing to 2008 being the coolest year since the turn of the 21st century. By far, the warmest year on record is 2016, followed by 2015, in part because of the El Niño event, and 2017 and 2018 follow in warmth (Figure 1.8, chapter 1). All of the impacts of El Niño are exacerbated by global warming (see section 2.5). 2.4. TELECONNECTIONS AND MODES OF VARIABILITY Teleconnections are linkages across great distances. Implied in the term teleconnection is that there is a

physical reason for the simultaneous variations, often of opposite sign, over distant parts of the globe, and the pri­ mary reason is the presence of Rossby waves in the atmosphere. In the tropics they may also originate from the monsoonal‐type overturning circulations that link the upward and downward branches. Following early work mainly for the Northern Hemisphere by Horel & Wallace (1981), a comprehensive review of teleconnections asso­ ciated with variability was given by Trenberth et  al. (1998), Alexander et  al. (2002), and Yeh et  al. (2018); much of the following derives from Trenberth and Hurrell (2019). Teleconnections can also occur through the ocean (cf. chapter  15) via exchanges between the Indian and Pacific oceans via the Indonesian Throughflow (Li et al., 2016). Effects from outside the tropics on ENSO are dis­ cussed in chapter 11, and chapter 14 goes into teleconnec­ tions in more detail. In the tropics, the temperature of the surface waters is readily conveyed to the overlying atmosphere, and because warm air is less dense it tends to rise, whereas cooler air sinks. As air rises into regions where the air is thinner, the air expands, causing cooling and therefore condensing moisture in the air, which produces rain. Low sea‐level pressures are set up over the warmer waters, while higher pressures occur over the cooler regions in the tropics and subtropics, and the moisture‐laden winds tend to blow toward low pressure so that the air converges, resulting in organized patterns of heavy rainfall. The rain comes from convective cloud systems, which preferentially occur in the convergence zones, the ITCZ and SPCZ, which are sepa­ rated by the equatorial dry zone. In the tropical atmosphere, anomalous SSTs force anom­ alies in convection and large‐scale overturning with subsi­ dence in the descending branch of the local Hadley Circulation, which is primarily in the winter hemisphere. In addition, the largest SST anomalies tend to occur in the eastern tropical Pacific from November to December (Trenberth, 1997). The resulting strong upper tropospheric divergence in the tropics and convergence in the subtropics act as a Rossby wave source (Figure 2.8). The climatolog­ ical stationary planetary waves and associated jet streams, especially in the Northern Hemisphere, can make the total Rossby wave sources somewhat insensitive to the position of the tropical heating that induces them and thus can cre­ ate preferred teleconnection response patterns, such as the Pacific–North American (PNA) pattern (Trenberth et al., 1998). In the subtropics, the changes in the jet stream are accompanied by changes in storm tracks in such a way that the transient disturbances (including storms) give feedback and influence the outcome through changes in heat, vor­ ticity, and momentum transports, as well as changes in dia­ batic heating through changes in rainfall‐induced latent heating. The large natural variability means that results are not deterministic but continually evolve, and this contributes

32  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

L

H

Rising air

Upper tropospheric outflow is bent to the right (NH) by Coriolis effect, creating an anticyclonic circulation

N High

SSTs Low pressure develops over high SSTs with convergence of winds and rain

Figure 2.8  Over the tropical Pacific Ocean where high SSTs exist, warm, moist air rises, creating low surface pressure and convergence of moisture. Widespread convection leads to strong latent heating of the atmosphere and divergent outflow in the upper troposphere and lower stratosphere. The Coriolis force acts on the outflow, creating anticyclonic conditions and thereby spinning up a Rossby wave downstream.

to the different flavors of El Niño effects, beyond the inevi­ table differences within the tropics. Many climate model– based experiments have clearly shown the dominant role of the tropical effects, however, while also indicating that large ensembles are essential to sort out signal from noise (Deser et al., 2014), possible only with models. Trenberth et  al. (2005) analyzed variability in global atmospheric mass distribution and found four key pat­ terns: the two annular modes (the Southern Annular Mode or SAM, and the Northern Annular Mode or NAM; Thompson & Wallace, 2000), a global ENSO‐ related pattern, and a fourth closely related to the North Pacific Index (NPI) and the Pacific Decadal Oscillation. Decadal to interdecadal variability in the atmospheric circulation is especially prominent in the North Pacific (e.g. Trenberth, 1990; Trenberth & Hurrell, 1994, 2019), where fluctuations in the strength of the wintertime Aleutian low pressure system, indicated by the NPI, co‐ vary with North Pacific SST in what has been termed the Pacific Decadal Oscillation (PDO; Trenberth, 2015), or its close cousin, the Interdecadal Pacific Oscillation (IPO). The PDO and associated NPI occur on longer timescales and larger spatial scales than ENSO.

Over the Atlantic sector, decadal variability has large amplitude relative to interannual variability, especially over the North Atlantic. The Atlantic decadal variability has been termed the Atlantic Multidecadal Oscillation (Enfield et  al., 2001; Trenberth & Shea, 2006). North Atlantic SSTs show a 65‐ to 75‐year variation (±0.2°C range), with a warm phase in 1930 to 1960 and after 1995, and cool phases during 1905 to 1925 and 1970 to 1995. Arguably the most prominent teleconnections over the Northern Hemisphere extratropics are the North Atlantic Oscillation (NAO), which is really a more regional ver­ sion of the NAM, and the PNA patterns. The NAO (Hurrell, 1995; Hurrell & Deser, 2009) refers to changes in the atmospheric sea level pressure difference between the Arctic and the subtropical Atlantic. Although NAO is the only teleconnection pattern evident throughout the year in the Northern Hemisphere, the climate anomalies associated with the NAO are largest during the northern winter months when the atmosphere is dynamically the most active. The NAO exerts a dominant influence on winter surface temperatures across much of the Northern Hemisphere, and on storminess and precipitation over Europe and North Africa (Hurrell et al., 2003). When the

ENSO in the Global Climate System  33

NAO index is positive, enhanced westerly flow across the North Atlantic in winter moves warm moist maritime air over much of Europe and far downstream, while stronger northerly winds over Greenland and northeastern Canada carry cold air southward and decrease land tem­ peratures and SST over the northwest Atlantic. Temperature variations over North Africa and the Middle East (cooling) and the southeastern United States (warming), associated with the stronger clockwise flow around the subtropical Atlantic high‐pressure center, are also notable. Positive NAO index winters are also associ­ ated with a northeastward shift in the Atlantic storm activity, with enhanced activity from Newfoundland into northern Europe and a modest decrease to the south. More precipitation than normal falls from Iceland through Scandinavia during high NAO index winters, while the reverse occurs over much of central and southern Europe, the Mediterranean, parts of the Middle East, the Canadian Arctic and much of Greenland. A positive PNA teleconnection pattern in the middle troposphere coincides with the warm‐phase ENSO pattern and is typically associated with higher‐than‐ normal pressure near Hawaii and over the northwestern United States and western Canada, while pressures are typically lower than normal over the central North Pacific and the southeastern United States (Alexander et  al., 2002). Variations in the PNA pattern represent changes in the north‐south migration of the large‐scale Pacific and North American air masses, storm tracks, and their associated weather, affecting precipitation in western North America and the frequency of Alaskan blocking events and associated cold air outbreaks over the western United States in winter. In the Southern Hemisphere wave structures do not emerge as readily, owing to the dominance of more zon­ ally symmetric variability (SAM) associated with synchronous pressure or height anomalies of opposite sign in middle and high latitudes, and therefore reflects changes in the main belt of subpolar westerly winds. Enhanced Southern Ocean westerlies occur in the positive phase of the SAM. SAM results mainly from the internal dynamics of the atmosphere and is an expression of storm track and jet stream variability. It contributes a significant proportion of southern midlatitude circulation variability on many time scales and is the leading mode in an analysis of monthly mean global atmospheric mass (Trenberth et  al., 2005), accounting for around 10% of total global variance. The SAM index (not shown) reveals a general long‐term increase beginning in the 1960s con­ sistent with a strengthening of the circumpolar vortex and intensification of the circumpolar westerlies that has been associated with the development of the ozone hole, especially in the southern summer (Swart et  al., 2015). Only in spring 2016 did the positive SAM turn abruptly

negative in association with large decreases in Antarctic sea ice in all sectors (Turner et al., 2017). The PDO/IPO have been described as a long‐lived El Niño–like pattern of Indo‐Pacific climate variability or as a low‐frequency residual of ENSO variability on mul­ tidecadal timescales (Newman et  al., 2016, and see chapter 8). Phase changes of the PDO/IPO are associated with pronounced changes in temperature and rainfall patterns across North and South America, Asia, and Australia. Furthermore, ENSO teleconnections on inter­ annual timescales around the Pacific basin are signifi­ cantly modified by the PDO/IPO. Low PDO goes with high NPI values, indicative of a weakened circulation over the North Pacific (1900–1924, 1945–1976, 1999– 2013), and predominantly high PDO values indicate a strengthened circulation (low NPI) (1925–1944, 1977– 1998, and since 2014) (Trenberth, 2015). The well‐known decrease in Aleutian low pressure from 1976 to 1977 (Trenberth, 1990) is analogous to transitions that occurred from 1946 to 1947 and from 1924 to 1925, and these earlier changes were also associated with SST fluc­ tuations in the tropical Indian and Pacific Oceans (e.g. Deser et al., 2004). The high PDO values relate to times of increases in the GMST (Figure 1.8, chapter  1), while the GMST no longer increases much for negative PDO values. From 1999 to 2013 this pause in the rise of GMST has also become known as a “hiatus” in warming (Trenberth & Fasullo, 2013; Trenberth et  al., 2014a; Trenberth, 2015; Fyfe et al., 2016). Although increases in GMST stall, the ocean heat content and sea level continue to rise (Cheng et al., 2017), showing that the heat from global warming is being redistributed within the ocean, both with depth and regionally in the West Pacific and Indian Oceans (England et al., 2014; Lee et al., 2015). The main pace­ maker of variability in rates of GMST increase appears to be the PDO, with aerosols likely playing a role in the earlier big hiatus from 1947 to 1976 (Trenberth, 2015; Kosaka & Xie, 2016). 2.5. CLIMATE CHANGE Earlier, in introducing the climate system, the role of external forcings versus internal variations was briefly dis­ cussed. Chapter  5 goes into detail about paleo‐climate aspects related to ENSO, chapter 11 delves into “external effects,” and chapter 13 considers projections into the next century related to climate change. On the timescale of decades, the main influences of concern are from human activities along with possible episodic volcanic eruptions and much smaller influences from the evolving sun. The sun has a distinct 11‐year sunspot cycle with changes in global irradiance of 0.5 to 1 W m‐2, equivalent to a radia­ tive forcing of less than 0.25 W m‐2. Volcanic eruptions

34  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

result in gases and particles in the stratosphere for up to about 2 years, and these can block the sun and/or absorb radiation, producing heating in the stratosphere but cooling the surface. However, these are not predictable. The main predictable change in external forcing is from the continual buildup of long‐lived greenhouse gases resulting from fossil fuel burning (carbon dioxide) and other human activities (Figure 1.8, chapter 1) (Trenberth, 2018). The total radiative forcing from increasing greenhouse gases is estimated to be about 3.0 W m‐2, but the net with aerosols also considered is about 2.3 W m‐2 (IPCC, 2013). The latter has an uncertainty of over 1 W m‐2. The result is global warming and an energy imbalance at the top‐of‐ atmosphere of order 0.9 W m‐2 (Trenberth et al., 2016). One consequence is an increase in GMST of 1°C, mainly since 1970. Half of the increase in carbon dioxide has occurred since 1985 (Figure 1.8, chapter 1). The energy imbalance is assessed by taking an inventory of the excess energy (Trenberth, 2009; Trenberth et  al., 2014b; von Schuckmann et al., 2016); over 90% ends up in the ocean, leading to increases in ocean heat content (Cheng et al., 2017), expansion of the ocean, and a rise in global sea level (Cazenave et al., 2018). Some of the energy imbalance goes into melting land ice (glaciers, Greenland, etc.) and sea ice, most notably in the Arctic. Some of the extra energy goes into warming the land and atmosphere and drying the land, thereby altering the hydrological cycle. Of particular note is that the atmosphere can hold more moisture at a rate of about 7% per degree Celsius (Clausius‐Clapeyron equation), and the main changes are more intense and faster‐developing droughts, heat waves, and wildfires, as well as heavier, more intense rains and snows (Trenberth, 2011). Because ENSO is the main source of both droughts and floods in different places around the world (Trenberth et al., 2014c), the biggest and most certain effect of climate change on ENSO is the risk of more heavy rains and dry spells, and thus flooding as well as water shortages, heat waves, and wildfires (see also Power et al., 2013). There are many theories about the ENSO phenomenon, discussed in chapters 6, 7, 8, and 13, for example, but one fundamental question is whether ENSO plays a vital role in the climate system beyond the obvious weather regime changes. Increasingly, it appears that the answer is yes. As noted above, heat builds up in the tropical western Pacific during normal and La Niña conditions, and during El Niño events, heat in the ocean first spreads across the Pacific, and then along the Americas outside of the tropics, all the while causing major changes in convection such that there is a major loss of ocean heat through evaporative cooling to the atmosphere. The increased atmospheric water vapor fuels storms and has a strong greenhouse effect. In turn this leads to heating of the atmosphere by latent heat in rainfall, and there is a mini– global warming. The latter stages of an El Niño event

produce the highest GMSTs (Trenberth et al., 2002). The result is a recharge and discharge of energy during the ENSO cycle, and ENSO effectively acts as a relief valve for the Pacific Ocean, keeping it cooler than it otherwise would be (Trenberth et  al., 2014b; Cheng et  al., 2019). Consequently, there has been some speculation about whether ENSO events could become more frequent and/ or bigger. Chapter  13 explores these aspects in more detail using models. 2.6. IMPACTS Changes associated with ENSO produce large variations in weather and climate around the world from year to year, and often these have a profound impact on humanity because of droughts, floods, heat waves, and other changes that can severely disrupt agriculture, fisheries, the environ­ ment, health, energy demand, and air quality, and also change the risks of fire. Chapters 16–20 explore these changes in much more detail, and this section is summa­ rized from Trenberth (2019). The normal upwelling of cold nutrient‐rich and CO2‐rich waters in the tropical Pacific is suppressed during El Niño. The presence of nutrients and sunlight fosters development of phytoplankton and zoo­ plankton, to the benefit of many fish species. Therefore, El Niño–induced changes in oceanic conditions can have disastrous consequences for fish and seabirds and thus for the fishing and guano industries, for example, along the South American coast. Other marine creatures may benefit so that unexpected harvests of shrimp or scallops occur in some places. Rainfall over Peru and Ecuador can trans­ form barren desert into lush growth and benefit some crops, but it can also be accompanied by swarms of grass­ hoppers and increases in the populations of toads and insects. Human health is affected by mosquito‐borne dis­ eases such as malaria, dengue, and viral encephalitis, and by water‐borne diseases such as cholera. In Africa, Rift Valley fever outbreaks may occur. Economic impacts can be large, with losses typically overshadowing gains. ENSO in both phases is the largest cause of drought around the world (Trenberth et al., 2014c), causing loss of agricultural production, widespread human suffering, and loss of life. El Niño also influences the incidence of fires, especially in Australia, Indonesia, and Brazil. With the fires come air quality and respiratory problems in adjacent areas up to 1000 km distant. Meanwhile, flooding occurs in Peru, Ecuador, and Chile, and coastal fisheries are disrupted. Very wet conditions can occur in California, but not always. REFERENCES Alexander, M. A., Bladé, I., Newman, M., Lanzante, J. R. Lau,  N.‐C., & Scott, J. D. (2002). The atmospheric bridge: The influence of ENSO teleconnections on air–sea interac­

ENSO in the Global Climate System  35 tion over the global oceans. Journal of Climate, 15, 2205– 2231, https://doi.org/10.1175/1520‐0442(2002)0152.0.CO;2 Ashok, K., Behera, S. K., Rao, S. A., Weng, H. Y., & Yamagata, T. (2007). El Niño Modoki and its possible tele­ connection. Journal of Geophysical Research Oceans, 112, C11007, https://doi.org/10.1029/2006JC003798 Bahr, D. B., Dyurgerov, M., & Meier M. F. (2009). Sea‐level rise from glaciers and ice caps: A lower bound. Geophysical Research Letters, 36, L03501, https://doi.org/10.1029/ 2008GL036309 Bell, G. D., & Chelliah, M. (2006). Leading tropical modes associated with interannual and multi‐decadal fluctuations in North Atlantic hurricane activity. Journal of Climate, 19, 590–612. Bjerknes, J. (1969). Atmospheric teleconnections from the equatorial Pacific. Monthly Weather Review, 97, 163–172. Capotondi, A., et  al. (2015). Understanding ENSO diversity. Bulletin American Meteorological Society, 96, 921–938. Capotondi, A., & Sardeshmukh, P. D. (2017). Is El Niño really changing? Geophys. Res. Lett., 44, 8548–8556. doi:10.1002/ 2017GL074515 Cazenave, A., et al. (2018). Global sea level budget 1993–pre­ sent, Earth System Science Data, 3, 10, 1551–1590. https:// doi.org/10.5194/essd‐10‐1551‐2018 Cheng L., Trenberth, K., Fasullo, J., Boyer, T., Abraham, J., & Zhu, J. (2017). Improved estimates of ocean heat content from 1960 to 2015, Science Advances, 3, e1601545. http:// dx.doi.org/10.1126/sciadv.1601545 Cheng, L., Trenberth, K. E., Fasullo, J., Mayer, M., Balmaseda, M., & Zhu, J. (2019). Evolution of ocean heat content related to ENSO. Journal of Climate, 32, 3529–3556. https://doi. org/10.1175/JCLI‐D‐18‐0607 Deser, C., Phillips, A. S., & Hurrell, J. W. (2004). Pacific inter­ decadal climate variability: Linkages between the tropics and the north Pacific during boreal winter since 1900. Journal of Climate, 17, 3109–3124. Deser, C., Phillips, A. S., Alexander, M. A., & Smoliak, B. V. (2014). Projecting North American climate over the next 50 years: Uncertainty due to internal variability. Journal of Climate, 27, 2271–2296. Deser, C., Simpson, I. R., McKinnon, K. A., & Phillips, A. S. (2017). The Northern Hemisphere extratropical atmospheric circulation response to ENSO: How well do we know it and how do we evaluate models accordingly? Journal of Climate, 30, 5059–5082. Enfield, D. B., Mestas‐Nunez, A. M., Trimble, P. J. (2001). The Atlantic Multidecadal Oscillation and its relationship to rain­ fall and river flows in the continental U.S. Geophysical Research Letters 28, 2077–2080. England, M. H., McGregor, S., Spence, P., Meehl, G. A., Timmermann, A., Cai, W., et al. (2014). Recently intensified Pacific Ocean wind–driven circulation and the ongoing warming hiatus. Nature Climate Change, 4, 222–227. https:// doi.org/10.1038/nclimate2106 Fyfe, J., Meehl, G., England, M. et al. (2016). Making sense of the early‐2000s warming slowdown. Nature Clim Change, 6, 224–228. https://doi.org/10.1038/nclimate2938

Garreaud, R. D. (2018). A plausible trigger for the 2017 coastal El Niño. International Journal of Climatology. https://doi. org/10.1002/joc.5426 Goldenberg, S. B., Landsea, C. W., Mestas‐Nuñez, A. M., & Gray, W. M. (2001). The recent increase in Atlantic hurricane activity: Causes and implications. Science, 293, 474–479. Horel, J. D., Wallace, J. M. (1981). Planetary‐scale atmospheric phenomena associated with the Southern Oscillation. Mon. Weather. Rev., 109, 813–829. Hu, D., Wu, L., Cai, W., Gupta, A.S., Ganachaud, A., Qiu, B., et  al. (2015). Pacific western boundary currents and their roles in climate. Nature, 522, 299–308. https://doi.org/10.1038/ nature14504 Hu, Z‐Z., Huang, B., Zhu, J., Kumar, A., & McPhaden, M. J. (2018). On the variety of coastal El Niño events. Climate Dynamics, 52, 7537–7552. https://doi.org.10.1007/s00382‐ 018‐4290‐4 Huffman, G. J., Adler, R. F., Bolvin, D. T., & Gu, G. (2009). Improving the global precipitation record. GPCP Version 2.1. Geophysical Research Letters, 36, L17808. https://doi. org/10.1029/2009GL040000 Hurrell, J. W. (1995). Decadal trends in the North Atlantic Oscillation and relationships to regional temperature and precipitation. Science, 269, 676–679. Hurrell, J. W., & Deser, C. (2009). Atlantic climate variability. Journal of Marine Systems, 78, 28–41. https://doi.org/10.1016/ j.jmarsys.2008.11.026 Hurrell, J. W., Kushnir, Y., Ottersen, G. & Visbeck, M. (2003). An overview of the North Atlantic Oscillation. In: Hurrell, J. W., Kushnir, Y., Ottersen, G., & Visbeck, M. (Eds.), The North Atlantic Oscillation: Climatic significance and environmental impact. Geophysical Monograph, 134, 1–35. American Geophysical Union, Washington, DC. IPCC (2013). Climate Change 2013: The Physical Science Basis. Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker, T.F., et  al. (Eds.)]. Cambridge University Press, Cambridge, UK, 1535 pp. Jadhav, J., Panickal, S., Marathe, S., Ashok, K. (2015). On the possible cause of distinct El Niño types in the recent decades. Scientific Reports, 5, 17009. https://doi.org/10.1038/srep17009 Johnson, G. C., Sloyan, B. M., Kessler, W. S., & McTaggart, K. E. (2002). Direct measurements of upper ocean currents and water properties across the tropical Pacific during the 1990s. Progress in Oceanography, 52, 31–36. Johnson, N. C. (2013). How many ENSO flavors can we distin­ guish? J Clim, 26, 4816–4827. doi:10.1175/JCLI‐D‐12‐00649.1 Kosaka, Y., & Xie, S‐P. (2016). The tropical Pacific as a key pacemaker of the variable rates of global warming. Nature Geoscience, 9, 669–673. https://doi.org/10.1038/NGEO2770 Lee, S.‐K., Park, W., Baringer, M. O., Gordon, A. L., Huber, B., & Liu, Y. (2015). Pacific origin of the abrupt increase in Indian Ocean heat content during the warming hiatus. Nature Geoscience, 8, 445–449, https://doi.org/10.1038/ngeo2438 L’Heureux, M. L., Tippett, M. K., & Barnston, A. G. (2015). Characterizing ENSO coupled variability and its impact on North American seasonal precipitation and temperature. Journal of Climate, 28, 4231–4245, doi:10.1175/JCLI‐D‐ 14‐00508.1.

36  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE L’Heureux, M. L., et al. (2017). Observing and predicting the 2015/16 El Niño. Bulletin American Meteorological Society, 98, 1363–1382. doi:10.1175/BAMS‐D‐16‐0009.1 Li, Y., Han, W., Hu, A., Meehl, G. A., & Wang, F. (2018). Multidecadal changes of the upper Indian Ocean heat content during 1965–2016. Journal of Climate, 31, 7863– 7884. https://doi.org/10.1175/JCLI‐D‐18‐0116.1 Mamalakis, A., Yu., J‐Y., Randerson, J. T., AghaKouchak, A., & Foufoula‐Georgiou, E. (2018). A new interhemispheric teleconnection increases predictability of winter precipitation in southwestern US. Nature Communications, 9, 2332. https:// doi.org/10.1038/s41467‐018‐04722‐7 Mayer, M., Haimberger, L., & Balmaseda, M. A. (2014). On the energy exchange between tropical ocean basins related to ENSO. Journal of Climate, 27, 6393–6403, https://doi. org/10.1175/JCLI‐D‐14‐00123.1 Mayer, M., Alonso Balmaseda, M., & Haimberger, L. (2018). Unprecedented 2015/2016 Indo‐Pacific heat transfer speeds up tropical Pacific heat recharge. Geophysical Research Letters, 45, 3274–3284. https://doi.org/10.1002/2018GL077106 McPhaden, M. J. (2012). A 21st century shift in the relationship between ENSO SST and warm water volume anomalies. Geophysical Research Letters, 39, 9706. https://doi. org/10.1029/2012GL051826 McPhaden, M. J., Lee, T., & McClurg, D. (2011). El Niño and its relationship to changing background conditions in the tropical Pacific Ocean, Geophysical Research Letters, 38, L15709, https://doi.org/10.1029/2011GL048275 Newman, M., et  al. (2016). The Pacific Decadal Oscillation, revisited. Journal of Climate, 29, 4399–4427. https://doi. org/10.1175/JCLI‐D‐15‐0508.1 Newman, M., Wittenberg, A. T., Cheng, L., Compo, G. P., & Smith, C. A. (2018). The Extreme 2015/16 El Niño, in the context of historical climate variability and change. Bulletin of the American Meteorological Society, 71, S16–S20. Power, S., Delage, F., Chang, C., Kociuba, G., & Keay, K. (2013): Robust twenty‐first‐century projections of El Niño and related precipitation variability. Nature, 502, 541–545, doi:10.1038/nature12580 Santoso, A., McPhaden, M. J., & Cai, W. (2017). The defining characteristics of ENSO extremes and the strong 2015/16 El Niño. Reviews of Geophysics, 55, 1079–1129. https://doi. org/10.1002/2017RG000560 Smith T. M., Reynolds, R. W., Peterson, T. C., & Lawrimore, J. (2008). Improvements to NOAA’s historical merged land– ocean surface temperature analysis (1880–2006). Journal of Climate, 21, 2283–2296. Sprintall, J., Gordon, A. L., Koch‐Larrouy, A., Lee, T., Potemra, J. T., Pujiana, K., & Wijffels, S. (2014). The Indonesian seas and their role in the coupled ocean‐clime system. Nature Geoscience, 7, 487–492. https://doi.org/10.1038/NGEO2188. Sverdrup, H. U. (1947). Wind‐driven currents in a baroclinic ocean; with application to the equatorial currents of the east­ ern Pacific. Proceedings National Academy of Science, 33, 318–326. https://doi.org/10.1073/pnas.33.11.318. Swart, N. C., Fyfe, J. C., Gillett N., & Marshall, G. J. (2015). Comparing trends in the Southern Annular Mode and sur­

face westerly jet. Journal of Climate, 28, 8840–8855. https:// doi.org/10.1175/JCLI‐D‐15‐0334.1 Thompson, D. W., & Wallace, J. M. (2000). Annular modes in the extratropical circulation. Pt I: Month‐to‐month vari­ ability. Journal of Climate, 13, 1000–1016. Timmermann, A., An, S., Kug, H‐S, Jin, F‐F., Cai, W., Capotondi, A., et  al. (2018). El Niño‐Southern Oscillation complexity. Nature, 559, 535–545. https://doi.org/10.1038/ s41586‐018‐0252‐6. Trenberth, K. E. (1976). Spatial and temporal variations of the Southern Oscillation. Quarterly Journal of the Royal Meteoro­ logical Society, 102, 639–653. Trenberth, K. E. (1984). Signal versus noise in the Southern Oscillation. Monthly Weather Review, 112, 326–332. Trenberth, K. E. (1990). Recent observed interdecadal climate changes in the Northern Hemisphere. Bulletin of the American Meteorological Society, 71, 988–993. Trenberth, K. E. (1994). The different flavors of El Niño. 18th Annual Climate Diagnostics Workshop, Boulder CO. 1–5 Nov. 1993; 50–53. Trenberth, K. E. (1997). The definition of El Niño. Bulletin of the American Meteorological Society, 78, 2771–2777. Trenberth, K. E. (2009). An imperative for climate change planning: Tracking Earth’s global energy. Current Opinion Environ. Sustainability, 1, 19–27, https://doi.org/10.1016/j. cosust.2009.06.001 Trenberth, K. E. (2011). Changes in precipitation with climate change. Climate Research, 47, 123–138. doi: 10.3354/cr00953 Trenberth, K. E. (2015). Has there been a hiatus? Science, 349, 691–692. https://doi.org/10.1126/science.aac9225 Trenberth, K. E. (2018). Climate change caused by human activities is happening and it already has major consequences. Journal of Energy & Water Resources Law, 36, 463–481. https://doi.org/10.1080/02646811.2018.1450895 Trenberth, K. E. (2019). El Niño Southern Oscillation. In: Cochran, J. K., Bokuniewicz, J. H., & Yager, L. P. (Eds.), Encyclopedia of Ocean Sciences (3rd ed., Vol. 6, pp. 420–432). Elsevier. https://doi.org/10.1016/B978‐0‐12‐409548‐9.04082‐3 Trenberth, K. E., & Caron, J. M. (2000). The Southern Oscillation revisited: Sea level pressures, surface temperatures and precipitation. Journal of Climate, 13, 4358–4365. Trenberth, K. E., & Hoar, T. J. (1996). The 1990–1995 El Niño– Southern Oscillation Event: Longest on record. Geophysical Research Letters, 23, 57–60. Trenberth , K. E., & Hoar, T. J. (1997). El Niño and climate change. Geophysical Research Letters, 24, 3057–3060. Trenberth, K. E., & Hurrell, J. W. (1994). Decadal atmosphere‐ ocean variations in the Pacific. Climate Dynamics, 9, 303–319. Trenberth K. E. & Hurrell, J. W. (2019). Climate change. In: P.  O. Dunn & A. P. Møller (Eds.), The Effects of Climate Change on Birds, 2nd ed., 5–25. Oxford University Press. doi: 10.1093/oso/9780198824268.003.0002Trenberth, K. E., & Shea, D. J. (1987). On the evolution of the Southern Oscillation. Monthly Weather Review, 115, 3078–3096. Trenberth, K. E., & Shea, D. J. (2006). Atlantic hurricanes and natural variability in 2005. Geophysical Research Letters, 33, L12704, https://doi.org/10.1029/2006GL026894

ENSO in the Global Climate System  37 Trenberth, K. E., & Stepaniak, D. P. (2001). Indices of El Niño evolution. Journal of Climate, 14, 1697–1701. Trenberth, K. E., & Stepaniak, D. P. (2004). The flow of energy through the Earth’s climate system. Quarterly Journal of Royal Meteorological Society, 130, 2677–2701. Trenberth, K. E., Branstator, G. W., Karoly, D., Kumar, A., Lau, N.‐C., & Ropelewski, C. (1998). Progress during TOGA in understanding and modeling global teleconnections asso­ ciated with tropical sea surface temperatures. Journal of Geophysical Research, 103, 14291–14324. Trenberth, K. E., Caron, J. M., Stepaniak, D. P. & Worley, S. (2002). Evolution of El Niño Southern Oscillation and global atmospheric surface temperatures. Journal of Geophysical Research, 107(D8), 4065. https://doi.org/10.1029/ 2000JD000298 Trenberth, K. E., Stepaniak, D. P., & Smith, L. (2005). Interannual variability of patterns of atmospheric mass dis­ tribution. J. Climate, 18, 2812–2825. Trenberth, K. E., et al. (2007). Observations: Surface and Atmos­ pheric Climate Change. In: Solomon, S., et  al. (Eds.), Climate Change 2007: The Physical Science Basis. Intergover­nmental Panel on Climate Change, 235–336. Cambridge University Press. Trenberth, K. E., Fasullo, J. T., Branstator, G. & Phillips, A. S. (2014a). Seasonal aspects of the recent pause in surface warming. Nature Climate Change, 4, 911–916. https://doi. org/10.1038/NCLIMATE2341. Trenberth, K. E., Fasullo, J. T., & Balmaseda, M. A. (2014b). Earth’s energy imbalance. Journal of Climate, 27, 3129–3144. https://doi.org/10.1175/JCLI‐D‐13‐00294 Trenberth, K. E., Dai, A., van der Schrier, G., Jones, P. D., Barichivich, J., Briffa, K. R., & Sheffield, J. (2014c). Global

warming and changes in drought. Nature Climate Change, 4, 17–22. https://doi.org/10.1038/NCLIMATE2067 Trenberth, K. E., Fasullo, J. T., von Schuckmann, K., & Cheng,  L. (2016). Insights into Earth’s energy imbalance from multiple sources. Journal of Climate, 29, 7495–7505. https://doi.org/10.1175/JCLI‐D‐16‐0339 Trenberth, K. E., Cheng, L., Jacobs, P., Zhang, Y., & Fasullo, J. (2018). Hurricane Harvey links to ocean heat content. Earth’s Future, 6, 730–744, https://doi.org/10.1002/2018EF000825 Turner, J., Phillips, T., Marshall, G. J., Hosking, J. S., Pope, J. O., Bracegirdle, T. J., & Deb, P. (2017). Unprecedented spring­ time retreat of Antarctic sea ice in 2016. Geophysical Research Letters, 44, 6868–6875, https://doi.org/10.1002/2017GL073656 von Schuckmann, K., et al. (2016). Earth’s energy imbalance: An imperative for monitoring. Nature Climate Change, 6, 138–144. https://doi.org/10.1038/NCLIM‐15030445C Williams, I. N., & Patricola, C. M. (2018). Diversity of ENSO events unified by convective threshold sea surface tempera­ ture: A nonlinear ENSO index. Geophysical Research Letters, 45. https://doi.org/10.1029/2018GL079203 Wolter, K., & Timlin, M. S. (2011). El Niño/Southern Oscillation behaviour since 1871 as diagnosed in an extended multivar­ iate ENSO index (MEI.ext). International Journal of Climatology, 31, 1074–1087. https://doi.org/10.1002/joc.2336 Yeh, S.‐W., Kug, J.‐S., Dewitte, B., Kwon, M.‐H., Kirtman, B. P., & Jin F.‐F. (2009). El Niño in a changing climate. Nature, 461, 511–514. https://doi.org/10.1038/nature08316 Yeh, S.‐W., Cai, W., Min, S.‐K., McPhaden, M. J., Dommenget, ­D., Dewitte, B. et al. (2018). ENSO atmospheric teleconnections and their response to greenhouse gas forcing. Reviews Geophysics, 56, 185–206. https://doi.org/10.1002/2017RG000568

Section II Observations

3 ENSO Observations Michael J. McPhaden1, Tong Lee2, Severine Fournier2, and Magdalena A. Balmaseda3

ABSTRACT Observations of the ocean‐atmosphere system sustained over decades are essential for describing, understanding, modeling, and predicting variations associated with El Niño and the Southern Oscillation (ENSO). Here we discuss the history of observing system development in the tropical Pacific and profile the mix of satellite and in situ components that currently make up the sustained observing system. To illustrate key physical processes that give rise to ENSO variations, we focus on the period 2014–2019, which encompassed the first extreme El Niño of the 21st century in 2015–2016. We also provide an overview of model‐based oceanic and atmospheric reanalysis products that are used for monitoring climate variability and initializing seasonal to decadal timescale climate forecasts. In addition, we highlight new developments in coupled ocean‐atmosphere data assimilation and century‐long reanalyses. We conclude by addressing some of the challenges in sustaining and evolving the observing system over time.

3.1. INTRODUCTION

on agricultural production, food security, freshwater resources, public health, power generation, and economic vitality around the globe. ENSO also affects marine and terrestrial ecosystems, pelagic fisheries, and the global carbon cycle (described in more detail in chapters 18–20). This chapter will describe observations needed for ENSO research and forecasting. We will highlight the requirements for these observations, the history of observing system development, and examples of the types of observations (in situ and satellite‐based) that have advanced our ability to describe, understand, and predict ENSO. Our focus will be on instrumental measurements rather than on paleo‐proxies, which will be covered separately in chapter  5. We will likewise focus on physical oceanographic and related surface marine meteorological measurements, since those for carbon cycle, biogeochemistry, and ecosystems will be discussed in later chapters of this volume. We will concentrate on sustained observations rather than on process studies, recognizing that the two types of measurement strategies are complementary. Sustained observations can provide long‐term, basin‐scale context for intensive space and time limited field programs

El Niño and the Southern Oscillation (ENSO) is a year‐ to‐year fluctuation of the climate system that arises through interactions between the ocean and the atmosphere in the tropical Pacific. It is the most energetic variation of the climate system on Earth (chapter 2) and the greatest source of climate predictability on seasonal to interannual timescales (chapter  10). ENSO affects the global atmospheric circulation and patterns of weather variability worldwide, with far‐reaching effects on human and natural systems (McPhaden et  al., 2006; chapters 14–17 of this volume). Floods, drought, wildfires, and extreme weather events associated with the ENSO cycle of warm (El Niño) and cold (La Niña) events have major impacts  NOAA/Pacific Marine Environmental Laboratory, Seattle, WA, USA 2   NASA/Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA 3   European Centre for Medium-Range Weather Forecasts, Reading, UK 1

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 41

42  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

to examine specific mechanisms (e.g. ocean mixing, atmospheric convection, air‐sea exchanges of heat). Process studies on the other hand can be used as pilot demonstration projects for the introduction of new technologies into the sustained observing system, for refining our understanding of relevant space and timescales that may need to be more regularly resolved, and for determining if new environmental parameters need to be routinely measured. A primary source of atmospheric data is the World Weather Watch (WWW) of the World Meteorological Organization (WMO). Established in the 1960s, the WWW combines observing systems, telecommunication facilities, and data‐processing centers operated by WMO member states to provide necessary meteorological and related data for weather and climate services. The global observing system for the WWW provides land, sea, aircraft, and earth observing satellite measurements transmitted in real time to operational forecasting centers around the world through the Global Telecommunications System (GTS). Given the maturity of the WWW and the wealth of meteorological measurements that are routinely available, our emphasis will be on the development and function of the ocean observation system as it relates to the study of ENSO. Oceanographic data in the tropical Pacific are collected within the framework of the Global Ocean Observing System (GOOS), which is an international effort to provide ocean information on a global scale that is relevant to societal needs. Within GOOS, the vast expanse of the tropical Pacific presents unique challenges to deliver in situ measurements with the necessary space/time resolution, accuracy, and continuity over many years. This chapter will describe these measurements, which fall into two broad categories: those made remotely from satellites and those from platforms deployed in the ocean, i.e. in situ measurements. These two types of measurements are complementary, and both are essential for documenting the evolution of ENSO events, understanding the key mechanisms at work, and providing the data needed to develop, initialize, and validate computer forecast models. In this chapter we will also describe model‐based ocean reanalyses that synthesize the information content of dynamical models and observations via data assimilation procedures. These products have limitations associated with model biases, space/time inhomogeneities in assimilated data sets, and assumptions related to the error covariances used for assimilating data. They have the advantage, however, of providing a full suite of gridded physical variables both regionally and globally, allowing for a comprehensive description of the observed climate. Model‐data products are also valuable for diagnosing processes that govern the evolution of the climate system, for forecast model development, and for the initialization and validation of climate forecasts. In addition, operational ocean reanalyses have been routinely used in mon-

itoring subsurface ocean temperature and heat content anomalies to assist ENSO prediction (Xue et al., 2017). 3.2. A BRIEF HISTORY The history of ocean observing system development for ENSO has been recounted in several publications (e.g. Halpern, 1996; McPhaden, 2006; McPhaden et al., 2010). Here we briefly review some of the highlights relevant to our discussion, starting with Jacob Bjerknes, a Norwegian‐ born meteorologist who provided the conceptual framework for understanding ENSO in the 1960s (Bjerknes, 1966, 1969). Bjerknes, as described in chapter  1, was the first to realize that El Niño arose through coupled interactions between the ocean and the atmosphere. He also realized that it was not just the coastal zone off western South America that warmed during El Niño, but the entire tropical Pacific basin out to near the international date line. 3.2.1. In Situ Observations Building on the pioneering work of Bjerknes, oceanographers began field programs in the tropical Pacific during the 1970s to understand the oceanic mechanisms that gave rise to El Niño and to observe its evolution on a basin scale. Klaus Wyrtki of the University of Hawaii led the way by establishing a network of island and coastal tide gauge stations in the tropical Pacific to monitor seasonal to interannual timescale variations in sea level and ocean surface geostrophic currents (Wyrtki, 1974a,b;1975a). Using data from this network and from the volunteer observing ship program that provided routine measurements of winds across the basin, Wyrtki devised his fundamental concept for the onset of El Niño (Wyrtki, 1975b). Conventional wisdom at the time posited that warming in the eastern basin associated with El Niño was caused by local weakening of alongshore winds off South America. Instead, Wyrtki argued that strengthening of the trade winds in the central Pacific, followed by their sudden collapse, forced a downwelling equatorial Kelvin wave that propagated eastward across the basin over the course of a few months. Along its path, this Kelvin wave would depress the thermocline, and when it reached the eastern Pacific, it would lead to abnormally high sea surface temperatures (SSTs) through reduction in equatorial upwelling of cold thermocline water into the surface layer. Wind‐forced Kelvin waves were prominent features in early numerical models of El Niño (e.g. McCreary, 1976; Hurlburt et  al., 1976) but were only detected for the first time in the early 1980s from moored time series data collected along the equator central and eastern Pacific (Knox & Halpern, 1982). The 1982–1983 El Niño, the strongest of the 20th century at that point in time (Cane, 1983), was a milestone in the annals of ENSO research. This El Niño was not predicted,

ENSO Observations  43

nor was it even detected until it had nearly reached its peak (McPhaden et al., 1998). There were many causes for this spectacular failure, but among them was the eruption of El Chichón in April–May 1982 that injected a cloud of sulfuric aerosols into the stratosphere, blinding newly launched satellites from detecting unusually warm SSTs that were developing. This problem was compounded by the lack of real‐time in situ data from the tropical Pacific to identify what was actually transpiring in the ocean. The 1982–1983 El Niño occurred when the scientific community was planning a major, 10‐year international program called the Tropical Ocean–Global Atmosphere, or TOGA, program to study Niño (McPhaden et al., 2010). The 1982–1983 experience emphasized the need for TOGA to develop reliable computer forecast models for El Niño and an ocean observing system that could deliver data in real time from across the basin to support seasonal prediction and climate research. Rising to this challenge, Stan Hayes of the U.S. National Oceanic and Atmospheric Administration (NOAA) developed the idea for a basin scale network of nearly 70 moorings that he called the TOGA–Tropical Atmosphere Ocean (TAO) array (Hayes et  al., 1991). The building block of this array was the Autonomous Temperature Line Acquisition System (ATLAS) mooring, a relatively low‐ cost surface moored designed to provide real‐time data via satellite relay for SST, upper ocean temperature, surface winds, and other meteorological parameters. The TAO array, completed in the last year of the TOGA program (McPhaden, 1995), became the centerpiece of the in situ tropical Pacific observing system that included Wyrtki’s tide gauge network, surface drifting buoys, and ship of opportunity expendable bathythermograph transects. Yearly cruises to service the moored buoy array also provided a valuable source of repeat shipboard measurements to characterize the circulation, hydrography, and biogeochemistry of the tropical Pacific (e.g. Chavez et al., 1999; Johnson et al., 2002).These in situ systems were complemented by a constellation of satellites for measuring surface winds, SST, and surface height from space. The TOGA observing system was in place to fully capture the evolution of the 1997–1998 El Niño (McPhaden, 1999; Picaut et  al., 2002), which was even stronger than the 1982–1983 El Niño. Moreover, in stark contrast to what happened in 1982–1983, real‐time data streams from this observing system made it possible to track the event as it unfolded. This observing system also provided data through the GTS to operational centers around the world for ingestion into seasonal forecast models developed during TOGA. Based on these forecasts, it was possible to ­reliably anticipate many of the climate impacts of this El Niño at six‐ to nine‐month lead times (Barnston et al., 1999). Beginning in 2000, the TAO array became known as TAO/TRITON, a partnership between NOAA and the

Japan Agency for Marine Earth Science and Technology (JAMSTEC). JAMSTEC designed its Triangle Transocean Buoy Network (TRITON) moorings, which were deployed west of the date line, to replicate ATLAS mooring capabilities so as to ensure a seamless data stream across the basin (Ando et  al., 2017). The ATLAS mooring design itself has been updated over time with new electronics, sensor technologies, and data transmission capabilities (Freitag et al., 2018). The early 2000s also saw the advent of the global Argo array of robotic profiling floats (Riser et  al., 2016). This array consists of approximately 3000 floats maintained worldwide, providing temperature and salinity profiles in the upper 2000 m every 10 days. Typical data coverage from in situ platforms in the Pacific for a recent 10‐day period (Figure 3.1) shows the variety of data that are currently available in real-time. This figure also emphasizes the remarkable growth in in situ ocean observations available for observing and forecasting El Niño since the time of the 1982–1983 El Niño. 3.2.2. Satellite Observations Satellite‐derived data offer a synoptic view of the ocean with near‐uniform spatial and temporal sampling, generally finer spatial resolution, and broader coverage than measurements from most in situ platforms. The satellite era for oceanography took a major leap forward in 1981 with the launch of NOAA’s operational advanced very high resolution radiometer (AVHRR) for SST measurements. Since the 2000s, AVHRR measurements have been augmented with measurements from passive microwave (PMW) radiometers (Reynolds et  al., 2007). Unlike AVHRR‐derived infrared SSTs, PMW SSTs are not obscured by clouds, but they have much lower spatial resolution. Sea surface height (SSH) measurements from satellite altimetry have been made operationally since 1992, starting from the TOPEX/Poseidon to more recently the Jason‐series missions (Fu & Cazenave, 2001). Ocean surface vector wind measurements have been made through a series of scatterometer missions such as the ERS‐1 and ‐2, QuikSCAT, and ASCAT (e.g., Liu, 2002). In addition, wind speed measurements have been made from space‐based PMW radiometers since the late 1970s. The spatiotemporal sampling of satellite products presents a major advantage in resolving spatial gradients such as SST fronts, wind stress curl, and wind divergence/convergence that are important to ocean dynamics and air‐ sea interaction (e.g. Chelton et al., 2001). In addition to SST, SSH, and winds, satellite measurements of precipitation, radiation, ocean color, and sea surface salinity (SSS) are now available. Likewise, ocean surface current velocities are now routinely estimated from satellite altimeter and scatterometer measurements (Bonjean & Lagerloef, 2002; Dohan, 2017).

44  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Observing platforms By Type, 4–13 Jan 1983 30°N

15°N

0°

15°S

30°S 120°W

150°W

180° Drifting Buoys

150°E Moored Buoys

120°E XBT

90°E

60°E

Tide Gauges

Observing platforms By Type, 4–13 Jan 2019 30°N

15°N

0°

15°S

30°S 120°W

150°W Drifting Buoys

180° 150°E Profiling Floats and Gliders

120°E Moored Buoys

XBT

60°E 90°E Tide Gauges

Figure 3.1  Distribution of in situ ocean measurements from various observational platforms in the Pacific Ocean for the 10‐day period of (top) 4–13 January 1983 and (bottom) 4–13 January 2019. Only the last position of each drifting buoy is shown for clarity. A major difference between these two time periods, besides the actual number and types of data available, is that most of the data in the early 1980s, such as those shown in the top panel, were available only months after they were collected, whereas today data as shown in the bottom panel are routinely available in real time, typically via satellite relay.

3.2.3. Complementarity of Satellite and In Situ Observations Satellite measurements provide broad (often global) spatial coverage of the ocean surface, but they provide no information about subsurface vertical structure that is critical to the understanding of ENSO development. For example, satellite SST measurements are typically used as a proxy for mixed‐layer temperature, but the knowledge of mixed‐layer depth relies on in situ measurements. Similarly, sea surface height measurements from space are often used as a proxy for the fluctuations of thermocline depth and upper‐ocean heat content in the tropical oceans. However, the vertical structure of temperature and salinity and the relative contributions of temperature and salinity to upper

ocean density ­ variability must come from in situ observations. In situ measurements are less uniformly and densely distributed across the ocean than satellite measurements. However, they are often more accurate than satellite retrievals that require complex algorithms to remove the effects of the intervening atmosphere and other geophysical influences. The higher accuracy of in situ measurements makes them valuable for the calibration and validation of satellite data. Conversely, a w ­ ell‐ calibrated satellite data product can help with in situ data quality control via cross‐validation consistency checks. Thus, satellite and in situ measurements complement one another and both are essential for describing and understanding ENSO, as we illustrate in the next section.

ENSO Observations  45

3.3. ENSO VARIABILITY In this section, we describe the evolution of the tropical Pacific over the period 2014–2019 using satellite data and in situ data from the TAO/TRITON array, Argo, and other data sources. This period encompasses one of the strongest El Niños on record, which occurred in 2015–2016 (L’Heureux et al., 2017; Santoso et al., 2017), a borderline El Niño in 2014 (McPhaden, 2015), and a very late‐starting weak warm event in 2018–2019. Our purpose is to highlight how the evolution of ENSO events is governed by multi‐timescale processes that involve seasonal timescale deterministic physical processes like those encapsulated in theories of ENSO (Suarez & Schopf, 1988; Battisti, 1988; Jin, 1997) and higher frequency (on time scales of days to weeks) atmospheric weather noise forcing (see also chapters 6 and 7 for more detailed discussion of ENSO theories). Deterministic processes involve, for example, coupling of the ocean and atmosphere via positive feedbacks mediated by surface zonal wind and SST variations, as well as wind‐forced changes in ocean circulation that produce large‐scale changes in upper ocean heat content. The most common manifestation of atmospheric weather noise forcing is referred to as a “westerly wind event” or “westerly wind burst,” which is an abrupt reduction in trade wind intensity that extends some tens of degrees of longitude along the equator and lasts for a few days to a few weeks (McPhaden, 1999; Vecchi & Harrison, 2000). Recently, “easterly wind surges” or “easterly wind events,” the easterly counterpart to westerly wind bursts, have been identified (Chiodi & Harrison, 2015; Puy et  al., 2019). They play a role in the ENSO cycle analogous to that of westerly wind bursts, particularly during the onset of La Niña events. These wind events are considered “noise” on the ENSO cycle because any individual weather event that gives rise to them cannot be predicted on seasonal timescales. They introduce irregularity in terms of the amplitude, timing, and duration of individual El Niño and La Niña events. 3.3.1. The 2015–2016 El Niño The 2015–2016 El Niño was the first extreme event of the 21st century and one of the strongest on record, with SST anomalies that rivaled those observed during the 1982–1983 and 1997–1998 El Niños (Figure  3.2a). Like most El Niños, this event developed early in the calendar year, reached its maximum development in the boreal winter, and ended in the following spring (Figure 3.2b). Peak conditions in December 2015 illustrated from TAO/ TRITON data (Figure 3.3) show the collapse of the surface trade winds in the central and western Pacific and their convergence over the western Pacific warm pool (surface water >28–29°C), which shifted eastward along

the equator in response to the weakened trades. The thermocline, which normally tilts down to the west in response to easterly trade wind forcing, flattened out, being deeper than normal in the eastern Pacific and shallower than normal in the west (Figure  3.3e). Cold upwelled water that normally forms the equatorial “cold tongue” along the equator in the eastern Pacific was only weakly evident at the height of the event in December 2015 (Figure 3.3b). The reason is that near‐normal trade winds in this region were upwelling water that was warmer than normal because of the unusually deep thermocline in the eastern basin. The net result of an eastward shifted warm pool and reduced upwelling in the cold tongue was the development of warm SST anomalies along the equator from the coast of South America to the international date line. This pattern of anomalies was similar in some respects to that of the 1997–1998 El Niño, which was an eastern Pacific or EP El Niño (see Figure 1.5 in chapter 1 and chapter 4 on ENSO diversity). However, compared to 1997–1998, SST anomalies were displaced westward along the equator in 2015–2016 and reached historical highs to the west of the Niño‐3.4 region (Xue & Kumar, 2017), where central Pacific or CP El Niños have their largest SST anomalies (e.g., Lee & McPhaden, 2010). The 2015–2016 El Niño thus appears to be a hybrid event with characteristics of both an EP and CP El Niño (L’Heureux et al., 2017; Paek et al., 2017). Satellite data and blended satellite–in situ data products provide a broader view of anomalous conditions across the Pacific basin at the peak of the El Niño in December 2015 (Figure  3.4). To the west of the warm SST anomalies, one can see anomalously cold SSTs (Figure 3.4a) that result from enhanced evaporative heat loss across the air‐sea interface in the far western Pacific (McPhaden, 2002). Deep atmospheric convection and precipitation (Figure  3.4b), which are the driving force for far‐field teleconnections, migrate eastward along the equator in response to this SST pattern. Thus, precipitation is enhanced in the central Pacific and reduced in the western Pacific. Weakened trade winds (Figure  3.4a) to the west of the largest SST and precipitation anomalies are indicative of the strong coupling between the ocean and the atmosphere through mutually reinforcing feedbacks. One also observes enhanced precipitation just north of the equator in the eastern Pacific and a deficit poleward of that, indicating an equatorward shift of rain bands in the Intertropical Convergence Zone. Likewise, there is an equatorward shift of the South Pacific Convergence Zone in the southern hemisphere as indicated by the rainfall deficit south of 10°S and the elevated rainfall values north of that, between 160°E and 160°W. Surface salinity is an important parameter because it constrains the upper ocean density field, which in turn

46  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a)

4

Nino 3.4 Sea Surface Temperature Anomalies (5°N–5°S, 120°W–170°W)

Temperature (°C)

3 2 1 0 –1 –2

1980 1984 1988 1992 1996 2000 2004 2008 2012 2016

(b) 4

Temperature (°C)

3

2018–19 vs. past El Niños 1982–83 1997–98 2015–16 2018–19 Average Std Dev

NINO 3.4

2 1 0 –1 –2 –3

Jul yr–1

Jan yr 0

Jul

Jan yr+1

Jul

Jan yr+2

Figure 3.2  (a) Five‐month running mean values of the Niño‐3.4 SST index since 1980 with El Niño events indicated by red peaks (above 0.5°C) and La Niña events indicated by blue troughs (below 0.5°C). (b) Amplitude of El Niños in the Niño‐3.4 region since the late 1950s based on 2.5‐year segments of monthly Niño‐3.4 values centered on the season of peak development in boreal winter. The thick black line in the right panel is the average of 15 El Niños between 1957–1958 and 2015–2016, and the grey shading is one standard deviation around the mean. The most recent warm event in 2018–2019 is shown as a thick red line. The dashed horizontal line at 0.5°C is a threshold above which SST anomalies must rise for five or more months to be considered an El Niño. Niño‐3.4 SST index is one of the most commonly used ENSO indices because of strong ocean‐atmosphere coupling on seasonal timescales in this region shown in Figure 3.3. The base period for computing anomalies in this plot is 1981–2010. The time series have not been detrended. (Data source http://www.cpc.ncep. noaa.gov)

affects the surface layer heat balance (Lukas & Lindstrom, 1991; Ando & McPhaden, 1997). One expects that surface salinity should show the imprint of anomalous pattern in rainfall, becoming fresher where it is raining more and saltier where it is raining less. That is indeed the case as one can see from comparing Figures  3.4b and 3.4c. The magnitude of surface salinity anomalies along the equator in the central Pacific is also significantly influenced by the eastward transport of warm fresh water from the western Pacific (Delcroix et al., 2000; Delcroix & Picaut, 1998: Picaut et  al., 2002), where the warmest

waters and heaviest rainfall are normally found in non–El Niño years (see Figure 1.1 in chapter 1). When the trade winds weaken during El Niño, the surface current along the equator rapidly accelerates eastward (Figure 3.4c). This eastward flow transports warm fresh water from the western Pacific into the central Pacific as noted above. In addition, these currents have an equatorward component immediately north and south of the equator west of 150°W (Figure 3.4c). This meridional convergence of surface flow onto the equator in response to anomalous westerly wind forcing (Figure  3.4a,b) is expected from Ekman theory (Ekman, 1905), which implies a general tendency for surface currents to flow to the right of the winds in the Northern Hemisphere and to the left of winds in the Southern Hemisphere. The resulting convergence of flow onto the equator produces downward velocities that tend to depress the thermocline and lead to surface warming via a process referred to as downwelling. Water expands when heated, so sea level is higher in regions where there is more warm water than cold water below the surface. Thus, thermocline depth and sea level are mirror images of one another in the tropical Pacific. Sea level along the equator is normally 50 cm higher and the thermocline 100 m deeper in the western Pacific than in the eastern Pacific. This situation arises because the westward blowing trade winds drive currents to the west, draining water heated by the sun from the eastern Pacific and piling up in the western Pacific. When the trade winds relax during El Niño, eastward accelerating currents shift this warm upper ocean water towards the east. Thus, the thermocline deepens and sea level rises in the east while the thermocline shoals and sea level falls in the  west (cf. Figure  3.3f and 3.4d). As we will show below, the way the ocean responds to the weakened trade winds to bring about this mass and thermal field adjustment is through the excitation of eastward propagating equatorial Kelvin waves and westward propagating Rossby waves. Monthly averages as in Figures  3.2–3.4 are useful for depicting the phases of the ENSO cycle, but to understand dynamical mechanisms, higher frequency information is necessary. El Niño is linked to a weakening of the trade winds, but that weakening is not a smooth process. Rather it is characterized by abrupt and episodic reductions in trade wind intensity in the form of westerly wind bursts (WWBs), referred to earlier in this section. WWBs and their role in the evolution of the 2015–2016 El Niño are clearly evident based on 5‐day analyses of TAO/TRITON and other data sets (Figure  3.5). The intense winds associated with these wind bursts (Figure  3.5a) have three important effects on SST (Figure 3.5b). First, they cool the western Pacific warm pool via enhanced evaporation across the air‐sea interface (McPhaden, 2002). Second, they accelerate surface

ENSO Observations  47 (a) Climatology

December SST and Winds

140°E

160°E

180°

160°W

140°W

120°W

100°W

10°N

31°C 29 27 25 23 21

5°N 0° 5°S 10°S 10. ms–1

(b) 2015 Mean 10°N

31°C 29 27 25 23 21

5°N 0° 5°S 10°S 10. ms–1

(c) 2015 Anomalies 10°N

4°C 2 0 –2 –4

5°N 0° 5°S 10°S 140°E

160°E

180°

160°W

140°W

120°W

100°W 10. ms–1

December Temperature 2°S to 2°N Average (d) Climatology

(e) 2015 Mean 180°

140°W

100°W 32°C 28 24 20 16 12 8 4 0

0 Depth (m)

100 200 300 400 500

140°E 0

180°

140°W

100°W 32°C 28 24 20 16 12 8 4 0

100 Depth (m)

140°E

200 300 400 500

(f) 2015 Anomalies 0 Depth (m)

100 200 300 400 500

140°E

180°

140°W

100°W

8°C 6 4 2 0 –2 –4 –6 –8

Figure 3.3 December surface winds and SSTs (a–c) and upper ocean temperatures (d–f) with climatological monthly means in panels (a) and (d), December 2015 monthly means at the height of the 2015–2016 El Niño in panels (b, e), and December 2015 anomalies in panels (c, f). December 2015 means and anomalies are based TAO/TRITON data. The Niño‐3.4 region is outlined in the SST anomaly plot (c). Cold (warm) temperature anomalies at depth indicate an unusually shallow (deep) thermocline. The “x” in (e) and (f) indicates depths and longitudes where temperature data are collected.

currents to the east (Figure 3.5d), producing very strong zonal velocity anomalies that transport warm water from the western Pacific into the central Pacific, evident

in the eastward shift of the 29°C isotherm that defines the eastern edge of the western Pacific warm pool (white lines in Figure  3.5a, d‐f). Third, WWBs generate

48  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

December 2015 Sea Surface Temperature and Surface Wind Anomalies 120°E 140°E 160°E 20°N

180°

160°W 140°W 120°W 100°W 80°W

60°W

(a)

10°N 0°

5 m s–1

10°S

4°C 2 0 –2

20°S

Rain Rate and Surface Wind Anomalies 20°N

(b)

10°N 0°

5 m s–1

10°S 20°S

30 mm day–1 20 10 0 –10 –20

Sea Surface Salinity and Surface Current Anomalies 20°N

(c)

10°N

1 PSS 0

0°

25 cm s–1

10°S

–1 –2

20°S

Sea Level and Surface Current Anomalies 20°N

(d)

10°N 0°

25 cm s–1

10°S 20°S 120°E 140°E 160°E

180°

160°W 140°W 120°W 100°W 80°W

0.3 m 0.2 0.1 0.0 –0.1 –0.2 –0.3

60°W

Figure 3.4  Charts for December 2015 anomalies in (a) surface winds and SST, (b) surface winds and rain rate, (c) surface currents and salinity, and (d) surface currents and SSH. Anomalies are computed relative to a mean seasonal cycle over the period of 1998–2017 for all variables except salinity for which anomalies are computed relative to a seasonal cycle over 2010–2017. The currents have been smoothed over 4° of latitude and 20° of longitude for clarity. SST data are from Reynolds et al. (2007), rain rate data from the Tropical Rainfall Measurement Mission (TRMM) (Huffman et al., 2010), surface wind data from the Multi‐Sensor Blended High‐Resolution Sea Surface Wind data set (Desbiolles et  al., 2017), salinity data from the Soil Moisture Ocean Salinity (SMOS) satellite (Boutin et  al., 2018), SSH from the Copernicus Marine Environment Marine Service data set (http:// marine.copernicus.eu), and currents from the Ocean Surface Current Analysis Real‐time (OSCAR) data product (Bonjean & Lagerloef, 2002).

downwelling equatorial Kelvin waves that cross the basin in about two months, leaving in their wake a thermocline depressed by up to 40 m (Figure  3.5c). These Kelvin waves are also seen in space‐based SSH measurements because of the inverse relationship between thermocline depth and SSH (Figure 3.5h). The effect of these Kelvin waves is to reduce the upwelling of cold water from the thermocline, so the eastern equatorial Pacific becomes anomalously warm. The net effect of these three processes is to shift the center of warmest surface water eastward and, in tandem, deep atmospheric convection and rainfall

(Figure  3.5e). Rainfall exhibits the same episodic character as the WWBs since it is disturbed weather that leads to the WWBs. Surface salinity in the central Pacific decreases (Figure 3.5f) in response to both the eastward shift in rainfall and the transport of fresh water from the western Pacific. This decrease in salinity enhances upper ocean density stratification and causes the mixed layer to thin, trapping heat over a shallower layer, which tends to elevate SST even more (Ando & McPhaden, 1997). The weakening trade winds, warming SSTs, and eastward shift in deep atmospheric convection and rainfall reinforce

(a)

Zonal wind (m s–1)

180°

2015 2016

–8 –4

2015

(d)

140°W 0

4

100°W

140°E

8

(e)

140°W

–2

0

2

100°W

20°C Depth (m)

140°E

180°

–80 –40

4

Rain Rate (mm day–1)

(f)

140°W 0

40

100°W 80

Surface salinity (PSS)

J F M A M J J A S O N D J F M A M J J A S O N D

J F M A M J J A S O N D J F M A M J J A S O N D

180°

140°W

100°W

140°E

–80 –40 0 40 80

(g)

2015

180°

–4

Zonal Current (cm s–1)

140°E

2016

(c)

SST (°C)

J F M A M J J A S O N D J F M A M J J A S O N D

140°E

2016

(b)

J F M A M J J A S O N D J F M A M J J A S O N D

180° –20

(h)

5°S SSH (m)

140°W 0

20

100°W

140°E

40

2°N– 2°S SSH (m)

180°

(i)

140°W 0

–1

5°N SSH (m) J F M A M J J A S O N D J F M A M J J A S O N D

J F M A M J J A S O N D J F M A M J J A S O N D

140°E

100°W

1

180° –0.2

140°W 0.0

0.2

100°W

140°E

180° –0.2

140°W 0.0

0.2

100°W

140°E

180° –0.2

140°W 0.0

100°W

0.2

Figure 3.5  Anomalies averaged between 2°N and 2°S for January 2015–December 2016 of (a) zonal wind, (b) SST, (c) 20°C isotherm depth, (d) zonal currents, (e) rain rate, (f) surface salinity, and (h) SSH. Panels (g) and (i) are for SSH at 5°S and 5°N, respectively. All panels show 5‐day analyses except for salinity, which is based on a 7‐day analysis. Panels (a)–(c) are from TAO/TRITON moored time series data, (d) is from OSCAR surface currents, (e) is from TRMM, (f) is from SMOS, and (g)–(i) are from satellite altimetry. The depth of the 20°C isotherm in (c) is a measure of the depth of the thermocline. The white lines in panels (a) and (d)–(f) show the longitudinal position of the 29°C SST isotherm, which is an indicator for the eastern edge of the western Pacific warm pool. Black squares on the upper and lower axes of panels (a)–(c) show longitudes where TAO/TRITON data are available at the start (top) and end (bottom) of the time series. White areas indicate no data. The currents in (d) have been smoothed over 4° of latitude and 20° of longitude for clarity.

50  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

one another to amplify El Niño conditions as the event develops. While the thermocline deepens and sea level rises in the east during El Niño, opposite tendencies are observed in the western Pacific, where the thermocline shoals and sea level falls (Figures 3.3e, 3.4d, 3.5c, g–i). These tendencies in the west are the result of westward‐propagating upwelling Rossby waves that, like the eastward propagating downwelling Kelvin waves, are generated by the weakening trade winds. The signature of these Rossby waves in SSH is largest at around 4–5°S (Figure  3.5g) and 4–5°N (Figure 3.5i). The fastest Rossby waves travel westward at one‐third the speed of the Kelvin waves and so take several months to reach the far western Pacific from where they are generated in the central Pacific. According to delayed oscillator theory (Suarez & Schopf, 1988; Battisti, 1988), at the western boundary these upwelling Rossby waves reflect into upwelling Kelvin waves that propagate eastward to elevate the thermocline and reinitiate upwelling in the cold tongue. The eastward progression of upwelling, some of which may also be directly forced by intensification of the trade winds at the height of the El Niño in the western Pacific (Boulanger et al., 2003, 2004; Harrison & Vecchi, 1997), is very evident from the shoaling thermocline in early 2016 (Figure 3.5c). These delayed negative feedbacks associated with the reflection of Rossby waves into Kelvin waves at the western boundary and trade wind intensification in the far west lead to the eventual demise of warm SST anomalies and the onset of cool La Niña conditions in the latter half of 2016. An interesting feature in the 5°N SSH data in the latter half of 2016 is the westward‐propagating sea level oscilla-

tion with a period of roughly 30 days east of the date line. These are tropical instability waves (TIWs) that are generated by shear instability of the swift zonal currents near the equator (see chapter 2 for a description of these currents). These zonal currents, and the shear instabilities that generate TIWs, are stronger north of the equator than south, stronger typically during June to February than March to May, and stronger during La Niña than El Niño (Cox, 1980; Philander et al., 1986). Thus, TIWs are very evident at 5°N during June–December 2016 after the onset of the La Niña, whereas they are virtually absent during the latter half of 2015 while the El Niño was underway. TIWs normally transport warm water equatorward, even more so during cold La Niña episodes, and thus act as a negative feedback on the development of ENSO SST anomalies (Wang & McPhaden, 1999, 2000, 2001; An, 2008). As a final observation, it is evident that the westerly wind bursts in 2015 were mostly confined to the western Pacific warm pool west of the 29°C isotherm and that they disappear as the western Pacific cools in 2016. This highlights the state‐dependent nature of stochastic forcing in the ENSO cycle (Eisenman et  al., 2005). WWBs lead to eastward expansion of the warm pool, thus expanding the area over which conditions are favorable for deep convection, disturbed weather, and more WWBs (Levine et  al., 2016). This nonlinearity in the ocean‐ atmosphere system helps to rectify higher frequency WWB forcing into lower frequency ENSO variability (Kessler et  al., 1995; Lengaigne et  al., 2004). The interplay between ENSO timescale variations and noise forcing on ENSO predictability is described next.

Depth Averaged Temperature Anomalies (0/300 m) and NINO 3.4 SST (5°N–5°S, 120°E–80°W) 1.0

3

3

2

2

1

1

0.5

0

–1

–1

–2

–2

–3

–3

0.0

–0.5

–1.0 1980

1984

1988

1992

1996

2000

2004

2008

2012

2016

–0.5

2014

2016

2018

Year

Figure 3.6  Time series of Niño‐3.4 SST (blue line) and anomalous ocean heat content along the equator (T300; red line). Monthly means between 1980 and 2019 smoothed with a five‐month running mean are shown in the left panel, while unsmoothed monthly means are shown in right panel for 2014–2019. The red dashed horizontal line is 0.25°C, a value of T300 above which the likelihood of El Niño developing greatly increases. T300 is computed as the average temperature anomaly in the upper 300 m between 5°N and 5°S from the coast of New Guinea to the coast of western South America using TAO/TRITON, Argo, and XBT data.

T300 (°C)

0

0.0

SST (°C)

T300 (°C)

0.5

ENSO Observations  51

3.3.2. The Role of Heat Content and Random Wind Forcing on ENSO Predictability According to recharge oscillator theory, a buildup of excess heat content along the equator in the tropical Pacific is a necessary precondition for the development of El Niño (Jin, 1997). This concept is illustrated by comparing Niño‐3.4 SST with ocean heat content as computed by the average temperature anomaly in the upper 300 m between 5°N and 5°S (Figure  3.6). Heat content leads Niño‐3.4 SST by typically six to nine months (Meinen & McPhaden, 2000) except for the first decade of the 21st century, when the lead shortens to about three months (McPhaden, 2012). The largest buildup of heat content since 1980 was observed in 1997 prior to the peak of the 1997–1998 El Niño, and there was nearly as large a buildup that occurred in 2015 prior to the full development of the 2015–2016 El Niño. Thus, upper ocean heat content (which can be defined in a variety of different ways, e.g., Meinen & McPhaden, 2000; Neske & McGregor, 2018; Planton et al., 2018) is a major source of ENSO predictability. It is interesting to note, however, that the buildup of heat content in 2018 was as large as in 1997 and 2015, and yet 2018–2019 was a very weak El Niño event with Niño‐3.4 SST values far below those in 1997–1998 or 2015–2016

(Figures  3.2 and 3.6). It is evident that while ocean heat content may precondition the system to favor the development of ENSO events, it is not sufficient to determine the full character of those events. So, what in particular accounts for the dramatic SST differences observed between 2015–2016 and 2018–2019 in view of the equally favorable heat content precursors in these years? The answer lies in part in the differences in the WWB forcing (Figure 3.7). Westerly wind bursts can occur randomly at any time. But when they happen with sufficient amplitude and frequency, they favor El Niño development by amplifying initial warm SST anomalies, which leads to a sustained collapse of the trade winds (Puy et al., 2019). It is noteworthy that the greatest number of wind bursts during 2014–2019 occurred in 2015. In contrast, while there were several westerly wind bursts in 2018, they were either of limited zonal fetch and confined to the far western Pacific or relatively weak until late September 2018. Thus, development of warm SSTs, even with a large buildup of heat content below the surface, was delayed until the latter part of 2018 when wind bursts of sufficient amplitude and fetch developed. Why the evolution of wind burst forcing varied in these 3 years is an open question, though random chance is a reasonable null hypothesis (McPhaden, 2015).

5–Day Zonal wind Anomalies 2°S to 2°N Average J

2014

2015

2018

J

F

F

M

M

A

A

M

M

J

J

J

J

A

A

S

S

O

O

N

N

D

D 140°E –8

180° –4

140°W 0

4

100°W 140°E 8 m s–1

–8

180° –4

140°W 0

4

100°W 140°E 8 m s–1

180° –4

0

140°W

100°W

4

m s–1

Figure 3.7  Time series of 5‐day zonal wind anomalies averaged between 2°N and 2°S from TAO/TRITON data for 2014 (left), 2015 (middle), and 2018 (right). Positive values are westerly. The white line indicates the 29°C isotherm used to define the eastern edge of the western Pacific warm pool. Black squares on the upper and lower axes show longitudes where TAO/TRITON wind data are available at the start (top) and end (bottom) of the time series. White areas indicate no data.

52  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

The arrested development of El Niño in 2014 illustrates another interesting aspect of the interplay between high‐ frequency WWB forcing and low‐frequency seasonal variations. As in 2015 and 2018, heat content throughout 2014 was well above a threshold of about 0.25°C that favors El Niño development, but at best only a borderline El Niño developed that year. Three strong wind bursts helped initiate warming along the equator in early 2014, but wind burst forcing did not continue into midyear (Menkes et al., 2014). There was even a temporary resurgence of the trade winds in June–July that effectively thwarted further SST warming (Hu & Fedorov, 2016; Levine & McPhaden, 2016). The normal life cycle for El Niño events involves a purging of excess heat from equatorial to higher latitudes as the event peaks and declines, setting the stage for a return to normal conditions or, if the heat content anomaly overshoots zero into substantial negative territory, the onset of La Niña (Figure  3.7; Jin, 1997, Meinen & McPhaden, 2000). However, the arrested development of El Niño in 2014 meant that elevated heat content lingered below the surface into early 2015. Levine and McPhaden (2016) and Hu and Fedorov (2017) argued from modeling studies that this heat content provided the fuel necessary to help jump start the extreme 2015–2016 El Niño when WWBs began to develop in early 2015. 3.4. DATA PRODUCTS TAO/TRITON, Argo, and other elements of the in situ observing system provide key inputs in the tropical Pacific for many oceanic and surface meteorological data products such as the International Comprehensive Ocean‐ Atmosphere Data Set for surface marine meteorological measurements (Freeman et al., 2019), Tropflux (Praveen Kumar et al., 2013) and OAFLUX (Yu & Weller, 2007) for surface fluxes, and the World Ocean Database for a wide variety of ocean data at and below the surface (Boyer et al., 2018). They are also critical for calibrating and validating satellite data because they are generally more accurate than remotely sensed data. Beyond that, they are essential for development of blended satellite in situ data products that take advantage of satellite data coverage and in situ data accuracy. There are a great many such blended products; below we describe some that are widely used for SST, surface winds, SSH, and SSS. Satellite infrared (IR) SST retrievals, despite their long history (since the early 1980s), are obscured by clouds. Measurements of SST from passive microwave (PMW) sensors do not have this limitation, but their records are relatively short. Moreover, IR SST sensors measure SST in the top 10 microns of the ocean surface, while PMW SST sensors measure SST in the top millimeters of the ocean. This difference can cause relative biases between IR and PMW SST, especially due to diurnal heating and

evaporative cooling. Furthermore, the “skin” SST measured by both IR and PMW sensors can be different from the “bulk” SST measured by in situ instruments at depths of 1 m or greater. While skin SST are important to estimating surface turbulent heat fluxes, it is the bulk SST that are typically used for ENSO research. To address these issues, blended satellite and in situ SST products (e.g. Reynolds et al., 2007) have been generated, typically using the optimal interpolation (OI) method (e.g. the OISST produced by Reynolds et  al., 2007) to capitalize on the respective strength of satellite and in situ SST measurements. Satellite SST offers superior spatiotemporal sampling and coverage, while in situ SST provides much more accurate measurements that are necessary to de‐bias satellite SST and to perform the skin‐to‐bulk SST conversion. Blended products such as the OISST have been used for numerous ENSO studies, not only on seasonal to interannual timescales, but also for studies of multidecadal variations in ENSO characteristics (discussed in more detail in chapter 8). Lee and McPhaden (2010), for example, used the OISST to reveal a doubling of the amplitude of SST anomalies associated with El Niño in the central equatorial Pacific from the 1980s to the 2010s. This change was attributed to the more frequent occurrence of CP El Niños as well as the larger amplitudes of CP El Niños since the turn of the 21st century. The increasing amplitude of El Niño in the central equatorial Pacific during the 1980s–2010s has also contributed to the observed multidecadal warming trends in SST in areas of the western Pacific warm pool (Lee & McPhaden, 2010). The cross‐calibrated multiplatform winds are a synthesis of ocean surface wind measurements from various satellite scatterometers and radiometers with in situ wind measurements, mostly from the global tropical moored array, using a variational analysis method (e.g., Atlas et al., 2011). The mooring wind measurements are particularly important for representing the diurnal and semidiurnal variability that are not well captured by the limited number of satellite scatterometers, and for de‐biasing the satellite winds obtained from different missions. Such blended wind products have facilitated the research on oceanic and atmospheric processes associated with ENSO as well as ENSO diversity (e.g. Hou et al., 2016). In addition to SST and winds, joint analyses of satellite SSH and sea level measurements from historical tide gauge data have also been carried out. For example, Hamlington et al. (2014) used cyclostationary empirical orthogonal functions to obtain spatial structures from the altimetry record, which are then projected to the historical tide‐gauge data to reconstruct sea level back to the 1950s. Such SSH reconstructions enable the investigation of SSH characteristics associated with ENSO events

ENSO Observations  53

prior to altimeter era (which began in 1992). More recently, blended satellite and in situ SSS products have also been developed (Xie et al., 2014) by taking advantage of the finer spatial resolution of satellite SSS and the much more accurate near‐surface salinity measurements from Argo and tropical moorings to de‐bias the satellite SSS. These blended SSS products have facilitated studies of SSS horizontal gradient structures, variations in the western Pacific warm fresh pool associated with El Niño, and the evaluation of ocean reanalysis (e.g. Xue & Kumar, 2017). 3.5. OCEANIC AND ATMOSPHERIC REANALYSES

thus resulting in inconsistent processing of observations over a period of time. Nowadays, reanalyses produce multidecadal, gridded data sets that estimate a large variety of atmospheric, ocean, sea‐ice, sea‐state, and land surface parameters, including many that are not directly observed. Such data sets have become fundamental to research and education in the Earth sciences and are the backbone of climate services. Here we provide a short overview of current ocean reanalyses, focusing on climate monitoring and initialization of seasonal forecasts. We conclude this ­section by describing recent developments expected to shape reanalyses activities in the coming decades: exten­ sion back in time a century or more and coupled data assimilation.

3.5.1. Background Consistent historical records of the Earth’s climate are required as a reference for monitoring the current state of the climate, and to initialize/validate forecasts at different ranges, e.g. from medium range (3–10 day) to monthly, seasonal, and decadal forecasts. Observations alone are insufficient to generate the required information. Model simulations with adequate external forcing can provide some insight on the variability, but they are affected by biases due to errors in model formulation and specifications of forcing. Improved estimations of the climate history however can be achieved by optimally merging the information content of model, forcings, and observations. The method to optimally merge model with observations is called data assimilation. The resulting estimations are called reanalyses. Model‐based reanalyses rely on models to interpret, relate, and combine many different observations from multiple sources through data assimilation. Data assimilation uses prior information about uncertainties in models and observations for quality checks, to derive bias adjustments, and to assign proportional weights to the data. The equations of motion and physical processes as represented in a forecast model are used to generate data products that are spatially complete and physically consistent. In essence, the aim in reanalysis is to derive a comprehensive description of the observed climate by using as much information as possible. The idea of a reanalysis can be traced back to the atmospheric weather prediction community. The first reanalysis efforts took place in operational weather centers, in an attempt to produce consistent records of atmospheric variability, which implied a consistent reprocessing of archived weather observations using a modern forecasting system. This was different from a collection of archived operational “analyses” used in weather prediction, which would have been generated with different versions of the forecast model and data assimilation systems,

3.5.2. Ocean Reanalyses for Climate Monitoring The advent of atmospheric reanalyses in the early 1990s made possible the development for ocean reanalyses: for the first time, estimations of daily surface variables needed to force ocean models were available. The next steps were to combine these model estimations with ocean observations via data assimilation. By this time, ocean data assimilation methods were reaching maturity, and sustained large‐scale sampling of the ocean was becoming a reality: the TOGA program was providing unprecedented sampling of the thermal structure of the upper Equatorial Pacific, and the upcoming satellite altimeter programs promised observations of sea level at a global scale. Ocean reanalyses were produced at this time for both state estimation (Stammer et al., 2002) and for initialization of seasonal forecasts (Ji et al., 1995, Stockdale et al., 1998). With expanding technical capabilities, the demand for more sophistication grew, which led to higher spatial resolution and longer estimation periods, but also more complex applications, including biogeochemical investigations. Existing applications differ in their assimilation methods, data used, formulation of constraints, model numerics and resolution, surface boundary conditions (forcing), specification of model, and background errors, among others. Stammer et al. (2016) provide a review of data assimilation methods used in the production of ocean reanalyses. The production of ocean reanalyses (ORAs hereafter) is now an established activity in several research and operational centers. ORAs are revisited every so often, and new “vintages” are produced at intervals of about 5 years, as improvements in ocean models, data assimilation methods, atmospheric surface forcing fluxes, or ocean observations become available. Compared to the costly atmospheric counterpart, ORAs are relatively affordable: the number of ocean

54  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

observations is substantially smaller than that in the atmosphere. Therefore, there are a larger number of ORAs available. Routine efforts for coordinated evaluation of ORAs are now established to (1) identify robust climate signals and associated uncertainty, (2) measure progress in the quality of the reanalyses, (3) produce indices for ocean monitoring with associated error estimates, and (4) identify main deficiencies in order to guide future developments. One recent evaluation is the so‐called Ocean Reanalyses Intercomparison Project, described in Balmaseda et al. (2015). The studies within attempted to measure the maturity of the climate information provided by the current‐generation ORAs in terms of signal‐to‐ noise ratios of selected variables, such as ocean heat content, steric sea level, mixed layer depth, ocean salinity, surface fluxes over the ocean, sea‐ice variables, and Atlantic meridional overturning circulation, among others. The intercomparison concluded that deficiencies in the observing system are major obstacles for the reliable reconstruction of the past ocean climate. Other problems, such as insufficient quality of surface fluxes, data assimilation methods, and ocean models, were also identified. For instance, many data assimilation methods struggle to represent equatorial dynamics. The quality of estimated ocean currents and meridional transports at the equator in fact can deteriorate as a result of data assimilation. Ensemble‐based analysis of signal‐to‐noise ratio allows the identification of ocean characteristics for which the estimation is robust (such as tropical mixed‐ layer‐depth, upper ocean heat content), and where large uncertainty exists (equatorial circulation, deep ocean, Southern Ocean, sea ice thickness, and salinity), providing guidance for future enhancement of the observing and data assimilation systems. Some of the ORAs are continuously updated in quasi– real time, with the model and data assimilation methodology kept fixed. This is the case for the ORAs produced in operational centers to initialize coupled forecasts. These real‐time ORAs have the additional advantage that they allow real‐time monitoring of relevant climate variables (Xue et  al., 2017). Monitoring of tropical Pacific conditions with a multiocean reanalysis system helps to build confidence in the definition of key oceanic features relevant to climate, such as subsurface temperature structure along the equator, which is an important indicator and predictor of ENSO development (e.g. Figure 3.8). 3.5.3. Ocean Reanalyses for Initialization of Forecasts ORAs are also being used for initialization and calibration of coupled models providing forecasts at a range of timescales, from medium range to decadal. Here we focus on the use of reanalyses for initialization of seasonal

forecasts. There is clear demand for reliable forecasts of climate at seasonal timescales for a variety of societal applications. Good‐quality seasonal forecasts with reliable uncertainty estimates are of great value to society, allowing institutions and governments to plan actions to minimize risks, manage resources, and increase prosperity and security. Human and economic losses caused by adverse climate events can be mitigated with early warning systems (e.g. famine, epidemics) and disaster preparedness. Equally, adequate planning can aid the exploitation of favorable climate conditions. Seasonal forecasting is currently a routine activity in several operational centers, with a growing number of economic and societal applications such as agriculture, health, and energy. Seasonal forecasts predict variations in the atmospheric circulation in response to anomalous boundary forcing, variations that significantly change the probability of occurrence of weather patterns. In order to extend the predictability horizon, these boundary conditions need to be either slowly varying or predictable given the initial conditions. Examples of boundary forcing are variations of SST, land conditions (snow depth, soil moisture), sea‐ ice, and radiative gases. Of special importance are ENSO SST variations, which have the potential to alter large‐ scale tropical convective cells, the Walker and Hadley circulations, and the global atmospheric circulation. Seasonal forecasting systems are based on coupled ocean‐atmosphere general circulation models that predict both the surface boundary forcing and their impact on the atmospheric circulation. The chaotic nature of the atmosphere is accounted for by issuing probabilistic forecasts, obtained by performing an ensemble of coupled integrations. Forecast skill is assessed by calibrating against a series of past seasonal hindcasts, which in turn requires initial conditions based on reanalyses for a historical period (typically 20–30 years). The reliability of interannual variability in the ocean reanalyses will determine the forecast quality. Operational seasonal forecasting also requires near‐real‐time knowledge of the state of the climate, placing strong demands on the ocean observing system, which should deliver timely high‐ quality observations. Seasonal forecasts can be initialized in a number of ways (Balmaseda, 2017). In all of them, there is an ocean reanalysis where observations and models have been combined via data assimilation. For seasonal forecasts, the initialization of the density and thermal structure of the tropical oceans is a high priority. Special attention is paid to initialization of the upper 300 m of the tropical Pacific, given its role in ENSO dynamics. Maintaining the dynamical balance between the winds and the upper ocean thermal structure is a major challenge for current ocean data assimilation systems. High‐quality surface‐winds

ENSO Observations  55 Anomalous Temperature (C) Averaged in 5S–5N: FEB 2019 JMA

NCEP 50

50

100

100

150

150

200

200

250

250

300 120E

150E

180

150W 120W

300 120E

90W

150E

180

150W 120W

90W

150E

180

150W 120W

90W

150E

180

150W 120W

90W

180

150W 120W

90W

GFDL

ECMWF 50

50 100

100

150

150

200

200

250

250

300 120E

150E

180

150W 120W

300 120E

90W

BOM

NASA 50

50 100

100

150

150

200

200

250

250

300 120E

150E

180

150W 120W

300 120E

90W

ENS. Mean

SN Ratio 50

50 100

100

150

150

200

200

250

250

300 120E

150E

180

150W 120W

–4 –3 –2 –1 –0.5 0 0.5 1

2

300 120E

90W 3

4

5

6

150E

2

4

6

8

12

Figure 3.8  Comparison of subsurface temperatures along the equator between 5°N and 5°S for February 2019 from six different operational ocean reanalysis products: National Centers for Environmental Prediction (NCEP), ECMWF, NASA, Japan Meteorological Agency (JMA), Geophysical Fluid Dynamics Laboratory (GFDL), and the Australian Bureau of Meteorology (BOM). The ensemble mean is shown in the lower left, and the signal to noise (SN) ratio, computed as the ensemble mean divided by the ensemble standard deviation, is shown in the lower right. (Courtesy of NOAA/Climate Prediction Center, http://www.cpc.ncep.noaa.gov/products/GODAS/multiora_ body.html)

and ocean models are required to make most use of the available information. In turn, observations of surface fluxes, as well as upper ocean currents, temperature, and salinity, are needed to improve the ocean models and atmospheric reanalyses. The value of ocean observations and data assimilation for improving the skill of seasonal forecast is illustrated in Figure  3.9, showing the progress in ENSO forecast

skill in the successive ECMWF seasonal forecasting systems, from the first system in 1997 to the most recent, which became operational in November 2017 (Johnson et al., 2018). The progress is measured by the forecast lead time (in months), with sustained correlation for Niño-3.4 SST anomalies above 0.9. The figure also shows how the skill in the most recent system (S5) would change if no subsurface ocean observations or altimeter data were

56  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

assimilated, so the initial conditions are produced only by an ocean model constrained by SST and atmospheric surface fluxes. The drop in skill is comparable to more than 15 years of progress in seasonal forecasting. 3.5.4. Ongoing Efforts: Coupled Reanalyses Making optimal use of available observations is key to any successful reanalysis. Coupled data assimilation (CDA) is a major development thrust needed to better address the requirements of climate monitoring and seamless prediction. There is a need for methods that are designed to ingest observations of the atmosphere and ocean in a coupled model in a consistent manner, in particular by enforcing the model‐implied constraint across the component interfaces during the assimilation process. To achieve this, the data assimilation system itself must be coupled, in the sense that any adjustment due to observations near the surface must affect both atmospheric and oceanic variables. Consistent use of the model constraint in the assimilation process should improve estimates of SST, wind, precipitation, and associated surface fluxes and potentially make better use of available observations related to those variables. CDA should also reduce initialization shock resulting from imbalanced initial oceanic and atmospheric conditions, which should translate into overall improvements in forecast skill. CDA addresses two overarching goals of the World Meteorological Organization (WMO) World Weather Research Programme strategic plan: (1) to

improve environmental prediction and (2) to develop a seamless predictive capability (Brunet et al., 2015). Pioneering CDA efforts, such as the weakly coupled three‐dimensional variational data assimilation (3DVar) used by the National Centers for Environmental Prediction (NCEP) Climate Forecast System to generate its reanalysis (Saha et  al., 2006, 2010) and the coupled ensemble Kalman filter developed by Zhang et al. (2007) at the Geophysical Fluid Dynamics Laboratory, have led to many agencies now attempting to build coupled Earth system models (e.g. including atmosphere, ocean, sea ice, wave, land, aerosols, ionosphere, and biogeochemistry). These new systems will help extend forecast skill and achieve a seamless prediction capability. CDA methods have been used at ECMWF in the production of coupled reanalyses CERA‐20C and CERA‐SAT (Buizza et  al., 2018). For a review on recent developments on CDA, see Penny and Hamill (2017). 3.5.5. Ongoing Efforts: Century‐Long Reanalyses Extending ocean and atmosphere reanalyses further back in time is a tremendous scientific challenge, as the observing system is very sparse before the availability of satellite data from the 1970s onward, and especially before the advent of radiosonde measurements in the 1930s. Information about the nature and quality of early instrumentation is often incomplete. Furthermore, locating and gaining access to early weather observations requires dedicated efforts in data rescue and digitization.

Forecast lead-month with correlation above 0.9 SST anomalies NINO3.4 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 S1

S2

S3

S4

S5

S5-NoOobs

Figure 3.9  Lead time in months at which NINO3.4 SST anomaly correlation drops below 0.9, as measured for the five operational ECMWF seasonal forecasting systems (S1-S5), for the common period 1987–2002. For reference, the dashed bar (S5‐NoOobs) shows the skill that S5 would have if no subsurface ocean observations or altimeter were assimilated, so the initial conditions are produced only by an ocean model constrained by SST and atmospheric surface fluxes. The various generations of the ECMWF forecast systems were implemented in 1997 (S1), 2002 (S2), 2006 (S3), 2011 (S4), and 2017 (S5).

ENSO Observations  57

The production of a model‐based reanalysis that extends as far back as the instrumental record allows was first pursued in the 20th‐Century Reanalysis Project (Compo et al., 2006). The project was based on the idea that a reanalysis assimilating only surface pressure observations is relatively straightforward to compute and avoids many of the problems associated with observing system changes. Surface weather observations are available in reasonably large numbers throughout the 20th century, initially concentrated in the Northern Hemisphere but with global coverage increasing over time. Modern data assimilation systems are able to reconstruct a large‐scale tropospheric circulation from surface pressure observations alone, although the quality of the reconstruction strongly depends on the quality of the assimilating model (including its boundary conditions), especially where observations are sparse. Compo et  al. (2011) produced a reanalysis for the 140‐year period 1871–2010 using NCEP’s Global Forecast System, and ECMWF has produced the atmospheric reanalysis ERA‐20C, which covers the period January 1900 to December 2010 (Poli et al., 2016). ERA‐20C assimilates only conventional observations of surface pressure and marine wind obtained from well‐established climate data collections. Information on surface boundary conditions (SST and sea ice) and radiative forcing is critical in the production of these extended reanalyses, which pave the way for the production of equivalent extended ocean reanalyses. An example is the ORA‐20C uncoupled ocean reanalysis, which reconstructs the ocean and sea‐ice state over the 20th century (de Boisseson et  al., 2017). Temperature and salinity profiles are assimilated into the ocean model, which is also constrained by fluxes from ERA‐20C and a relaxation to observed sea surface temperature. Observations visibly impact the fields from ORA‐20C during the full record but are only able to constrain large ocean climate signals, such as ocean heat content, during the second half of the century. The lack of data constraints obviously poses a challenge for extended reanalyses. As discussed above, coupled reanalyses may be able provide a more realistic ensemble of realizations, where ocean and atmosphere observations can be exploited in a more dynamically consistent manner. This is the motivation for CERA‐20C (Laloyaux et  al., 2018), a reanalysis of the 20th century using the coupled data assimilation. The system simultaneously ingests atmospheric and ocean observations in the coupled Earth system model used for ECMWF’s ensemble forecasts (Laloyaux et al., 2016). CERA‐20C is the first coupled reanalysis of the 20th century that aims to reconstruct the past weather and climate of the Earth system including the atmosphere, ocean, land, ocean waves, and sea ice; it provides a 10‐member ensemble to account for errors in the observational record as well as model error.

The ability of the CERA‐20C to capture surface signals associated with ENSO is illustrated in a comparison of the multivariate ENSO index, which represents covariability of SST and mean sea level pressure in observations (Wolter & Timlin, 2011) ERA‐20C and CERA‐20C (Figure  3.10a). Both ERA‐20C and CERA‐20C agree very well with the observation‐based index, with correlation of 0.88 and 0.89, respectively. Capturing the signature of the ENSO variability in the ocean subsurface is also important as predictability studies based on coupled reforecasts will use initial conditions from coupled reanalyses such as CERA‐20C. Figure  3.10b shows the upper ocean heat content (0–300 m) in the equatorial Pacific (5°S–5°N). CERA‐20C compares very well with ocean‐only reanalyses ORA‐20C (de Boisseson et  al., 2017) and ORAS4 (Balmaseda et al., 2013) in the well‐ observed period, capturing the sequence of warming/ cooling associated with the major El Niño events of 1972–1973, 1982–1983, and 1997–1998. The early decades do not show large interannual signals in the ocean subsurface, with the exception of the 1940–1942 El Niño. This result may be interpreted as weaker ENSO variability during the early to mid‐20th century, as described by Wolter and Timlin (2011), or it could be related to insufficient observational information to constrain the reanalysis. More developmental efforts are clearly needed to advance the state of the art in century‐long reanalyses, but they will be essential for helping us to better understand and predict climate variability and extremes, especially those related to ENSO, in the context of changing climate. 3.6. SUMMARY AND CONCLUSIONS In this chapter we have discussed the history of observing system development in the tropical Pacific and its basis in complementary satellite and in situ measurements. Using these data, we have illustrated key physical processes that give rise to ENSO variations, focusing on the period 2014–2019, which encompassed the first extreme El Niño of the 21st century in 2015–2016. We have also provided an overview of available data products and model‐based reanalyses used for monitoring climate variability and initializing seasonal to decadal timescale climate forecasts. These observations, data products, and reanalyses have enabled fundamental advances in our understanding and ability to predict ENSO timescale variations and their climatic impacts. There are challenges, however, going forward in maintaining and evolving the observing system. Some data records from sites of the original TAO array, for example, are now 30 or more years long, allowing for studies of not only seasonal and interannual timescale variability but decadal variability (Amaya et  al., 2015), the decadal

58  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a)

4 ERA20C

Multivariate ENSO Index

3

CERA20C MEI

2 1 0 –1 –2 –3 –4

(b)

1910

1930

1950

1970

1990

2010

20 CERA-20C ORA-20C

Equatorial Pacific Ocean Heat Content Anomaly (in 108 J m–2)

15

ORAS4

10 5 0 –5 –10 –15 –20

1910

1930

1950

1970

1990

2010

Figure 3.10  (a) Multivariate ENSO index computed as in Wolter and Timlin (2011) for SST and mean sea level pressure. Shown are ERA‐20C, CERA‐20C, and the index estimated from observational products (Wolter & Timlin, 2011). (b) Equatorial Pacific Ocean Heat Content (OHC) anomalies in the upper 300 m between 5°N and 5°S as represented by ORA‐20C and CERA‐20C. For comparison, OHC from ORAS4 (Balmaseda et al., 2013) is also shown. Anomalies are computed with respect to a mean seasonal cycle over 1958–2010. Thick lines correspond to the control reanalysis and thin lines represent four plausible alternative analyses based on sampling uncertainties. An OHC increase of 1 × 108 J m‐2 corresponds to a temperature increase of 0.08°C averaged over the top 300 m. (Adapted from Laloyaux et al., 2018)

modulation of ENSO (McPhaden, 2012), and ENSO diversity as manifest by the distinctions between CP and EP El Niños (Lee & McPhaden, 2010; Capotondi et al., 2015). In addition, the major 2015–2016 El Niño was the first where the combined impacts of global warming and El Niño began to emerge in the tropical Pacific (King et al., 2016; Zhang et al., 2016; Brainard et al., 2017). In the future, both the amplitude of the ENSO cycle and the frequency of extreme El Niño and La Niña events is likely to increase in response to the unabated rise of

greenhouse gas concentrations in the atmosphere (e.g., Cai et al., 2015; Cai et al, 2018). Thus, the value of these moored time series will only increase as the records become longer and climate system continues to change. That realization, however, has not prevented major problems from developing in efforts to sustain the array. Data flow from TAO nosedived in 2013–2014 for lack of regular buoy maintenance cruises (Tollefson, 2014) at the start of what proved to be three highly unusual years of warm El Niño conditions. JAMSTEC subsequently

ENSO Observations  59

decommissioned most of its TRITON mooring sites in the western Pacific because of financial constraints. Financial limitations have been a problem for the Argo program as well (Durack et  al., 2016). Faced with such challenges, an international committee has reviewed the design of the tropical Pacific Ocean Observing System for the purpose of recommending modifications and enhancements that build both on the availability of new technologies and new scientific understanding accumulated over the past 25 years since the TAO array was completed (Kessler et al., 2019). Continuity of satellite measurements for the tropical Pacific Ocean, along with that of the in situ measurements, is also fundamental to ENSO research and prediction. The continuity of satellite IR SST and altimetry missions is ensured for the foreseeable future. However, the continuity of PMW SST and SSS measurements is not assured because previous satellite missions for these measurements are research as opposed to operational missions. There are currently two operational series of satellite scatterometers for ocean surface wind measurements (the ASCAT series from the European Space Agency and the OceanSat series from Indian Space Research Organization). They do not provide sufficient sampling to resolve the diurnal cycle. As a number of additional satellite scatterometers have been launched recently and will be launched in the next few years, public accessibility of their data in near real time is important to improve blended wind products. Satellite mission concepts to measure ocean surface currents directly from space are also being developed (e.g., Ardhuin et al., 2019; Rodriquez et al., 2019). These technologies provide a new opportunity to improve the quality of blended wind products by synthesizing satellite winds that are relative to the moving ocean surface and in situ winds that are relative to the Earth’s stationary frame. Availability of continuous and sustained observations is an essential requirement for the production of Earth‐ system reanalyses, which are widely used for monitoring of the Earth’s climate and are also an integral part of forecasting systems. Without the required observations, forecasts cannot be calibrated, and their skill cannot be estimated. In the last few years there have been two important developments regarding reanalyses. One is the production of century‐long products that require extensive efforts in data rescue, digitization, and quality control, as well as further developments in data assimilation methodology. The second initiative is the impetus towards coupled ocean‐atmosphere data assimilation systems that optimize the use of observations in both media in a dynamically consistent way. A sustained and enhanced climate observing system, including a robust ocean component of that system, will be needed for these reanalysis efforts to succeed.

ACKNOWLEDGMENTS The authors wish to thank Abderrahim Bentamy of IFREMER for help with access to recent merged satellite surface wind analyses, Conel Soci of ECMWF for help with data used in Figure 3.1, and Dai McClurg of PMEL for help with graphics. We also acknowledge two anonymous reviewers who offered constructive comments on an earlier version of this manuscript. The research by TL and SF described in this paper was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with NASA. MJM is supported by NOAA. This is PMEL contribution no. 4939.

REFERENCES Amaya, D. J., Xie, S.‐P., Miller, A. J., & McPhaden, M. J. (2015). Seasonality of tropical Pacific decadal trends during the 21st century global warming hiatus. Journal of Geophysical Research, 120, 6782‐6798. doi:10.1002/2015JC010906 An, S.‐I. (2008). Interannual variations of the topical ocean instability wave and ENSO. Journal of Climate, 21, 3680–3686. Ando, K., & McPhaden, M. J. (1997). Variability of surface layer hydrography in the tropical Pacific Ocean. Journal of Geophysical Research, 102, 23,063–23,078. Ando, K., Kuroda, Y., Fujii, Y., Fukuda, T., Hasegawa, T., Horii, T., et al. (2017). Fifteen years progress of the TRITON array in the Western Pacific and Eastern Indian Oceans. Journal of Oceanography, 73, 403–426. Doi 10.1007/ s10872‐017‐0414‐4 Ardhuin, F., Brandt, P., Gautier, L., et al. (2019). SKIM, a candidate satellite mission exploring global ocean currents and waves. Frontiers in Marine Science, accepted. Atlas, R., Hoffman, R. N. Ardizzone, J., Leidner, S. M., Jusem, J. C., Smith, D. K., & Gombos, D. (2011). A cross‐calibrated, multiplatform ocean surface wind velocity product for meteorological and oceanographic applications, Bulletin of the American Meteorological Society, 92, 157–174, doi:10.1175/ 2010BAMS2946.1 Balmaseda, M.A. (2017). Data assimilation for initialization of seasonal forecasts. The Sea: The Science of Ocean Prediction. Journal of Marine Research 75, 331–359. Balmaseda, M. A., Mogensen, K., & Weaver, A. (2013). Evaluation of the ECMWF ocean reanalysis system ORAS4. Quarterly Journal of the Royal Meteorological Society, 139, 1132–1161. Balmaseda, M. A., Hernandez, F., Storto, A., Palmer, M., Alves, O., Shi, L., et al. (2015). The ocean reanalyses intercomparison project (ORA‐IP). Journal of Operational Oceanography, 8(Supp. 1), s80–s97. Barnston, A. G., Leetmaa, A., Kousky, V. E., Livezey, R. E., O’Lenic, E. A., Van den Dool, H., et al. (1999). NCEP forecasts of the El Niño of 1997–98 and its U.S. impacts. Bulletin of the American Meteorological Society, 80, 1829–1852.

60  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Battisti, D. S. (1988). Dynamics and thermodynamics of a warming event in a coupled atmosphere‐ocean model. Journal of the Atmospheric Sciences, 45, 2889–2919. Bjerknes, J. (1966). A possible response of the atmospheric Hadley circulation to equatorial anomalies of ocean temperature. Tellus, 18, 820–829. Bjerknes, J (1969). Atmospheric teleconnections from the equatorial Pacific. Monthly Weather Review, 97, 163–172. Bonjean, F., & Lagerloef, G.S.E. (2002). Diagnostic model and analysis of the surface currents in the tropical Pacific Ocean. Journal of Physical Oceanography, 32, 2938–2954. Boulanger, J.‐P., Cravatte, S., & Menkes, C. (2003). Reflected and locally wind‐forced interannual equatorial Kelvin waves in the western Pacific Ocean. Journal of Geophysical Research, 108, 3311, doi: 10.1029/2002JC001760 Boulanger, J.‐P., Menkes, C., & Lengaigne, M. (2004). Role of high‐ and low‐frequency winds and wave reflection in the onset, growth and termination of the 1997–1998 El Niño. Climate Dynamics, 2, 267–280. doi: 10.1007/s00382‐003‐ 0383‐8 Boutin, J., Vergely, J. L., Marchand, S., D’Amico, F., Hasson, A., Kolodziejczyk, N., et al. (2018). New SMOS sea surface salinity with reduced systematic errors and improved variability. Remote Sensing of the Environment, 214, 115–134. doi: 10.1016/j.rse.2018.05.022 Boyer, T. P., Baranova, O. K., Coleman, C., Garcia, H. E., Grodsky, A., Locarnini, R. A., et  al. (2018). World Ocean Database 2018. In: A. V. Mishonov, A. V. (Technical Editor), NOAA Atlas NESDIS 87. [Available online at https://www. nodc.noaa.gov/OC5/WOD/pr_wod.html] Brainard, R. E., Oliver, T., McPhaden, M. J., Cohen, A., Veneers, R., Heenan, A., et  al. (2017). Ecological impacts of the 2015‐2016 El Niño in the central equatorial Pacific. Bulletin of the American Meteorological Society, 98, S21–S26. Brunet, G., Jones, S., & Ruti, P. M. (Eds.) (2015). Seamless prediction of the Earth system: From minutes to months. WMO Rep. 1156, 483 pp. [Available online at http://library.wmo.int/ pmb_ged/wmo_1156_en.pdf] Buizza, R., Bronnimann, S., Haimberger, L., Laloyaux, P., Martin, M. J., Fuentes, M., et al. (2018). The EU‐FP7 ERA‐ CLIM2 project contribution to advancing science and production of earth‐system climate reanalyses. Bulletin of the American Meteorological Society, 99, 1003–1014. https://doi. org/10.1175/BAMS‐D‐17‐0199.1 Cai, W. Santoso, A., Wang, G., Yeh, S.‐W., An, S.‐I., Cobb, K., et al. (2015). ENSO and greenhouse warming. Nature Climate Change, 5, 849–859. Cai, W., Wang, G. Dewitte, B., Wu, L., Santoso, A., Takahashi, K., et al. A. (2018). Increased variability of eastern Pacific El Niño under greenhouse warming. Nature, 564, 201–206. https://doi.org/10.1038/s41586‐018‐0776‐9 Cane, M. E. (1983). Oceanographic events during El Niño. Science, 222, 1189–1195. Capotondi, A., et  al. (2015). Understanding ENSO diversity. Bulletin of the American Meteorological Society, 96, 921–938. https://doi.org/10.1175/BAMS‐D‐13‐00117.1 Chavez, F. P., Strutton, P. G., Friederich, G. E., Feely, R. A., Feldman, G. C., Foley, D. G., & McPhaden, M. J. (1999).

Biological and chemical response of the equatorial Pacific Ocean to the 1997–1998 El Niño. Science, 286, 2126–2131. Chelton, D. B., Esbensen, S. K., Schlax, M. G., Thum, N., Freilich, M. H., Wentz, F. J., et  al. (2001). Observations of coupling between surface wind stress and sea surface temperature in the eastern Tropical Pacific. Journal of Climate, 14, 1479–1498. Chiodi, A.M., & Harrison, D.E. (2015). Equatorial Pacific easterly wind surges and the onset of La Niña events. Journal of Climate, 28, 776–792. Compo, G. P., Whitaker, J. S., & Sardeshmukh, P. D. (2006). Feasibility of a 100‐year reanalysis using only surface pressure data. Bulletin of the American Meteorological Society, 87(2), 175–190. Compo, G. P., Whitaker, J. S., Sardeshmukh, P. D., Matsui, N., Allan, R. J., Yin, X., et  al. (2011). The twentieth century reanalysis project. Quarterly Journal of the Royal Meteorological Society, 137(654), 1–28. Cox, M. D. (1980). Generation and propagation of 30‐day waves in the Pacific. Journal of Physical Oceanography, 10, 1168–1186. de Boisseson, E., Balmaseda, M. A., & Mayer, M. (2017). Ocean heat content variability in an ensemble of twentieth century ocean reanalyses. Climate Dynamics, 50, 3783–3783. Delcroix, T., DeWitte, B., & duPenhaot, Y., et  al. (2000). Equatorial waves and warm pool displacements during the 1992–1998 El Niño Southern Oscillation events: Observation and modeling. Journal of Geophysical Research, 105, 26045–26062. Delcroix, T., & Picaut, J. (1998). Zonal displacement of the western equatorial Pacific “fresh pool.” Journal of Geophysical Research, 103, 1087–1098. doi:10.1029/97JC01912 Desbiolles, F., Bentamy, A., Blanke, B., Roy, C., Mestas‐Nunez, A., Grodsky, S. et al. (2017). Two decades (1992–2012) of surface wind analyses based on satellite scatterometer observations. Journal of Marine Systems, 168, 38–56. http://doi. org/10.1016/j.jmarsys.2017.01.003 Dohan, K. (2017). Ocean surface currents from satellite data. Journal of Geophysical Research, 122, 2647–2651. doi:10.1002/2017JC012961 Durack, J. P., Lee, T., Vinogradova, N. T., & Stammer, D. (2016). Keeping the lights on for global ocean salinity observation. Nature Climate Change, 6, 228–231. Eisenman, I., Yu, L., & Tziperman, E. (2005). Westerly wind bursts: ENSO’s tail rather than the dog? Journal of Climate, 18, 5224–5238. Ekman, V. (1905). On the influence of the Earth’s rotation on ocean currents. Arkiv för matematik, astronomi och fysik, 2, 1–53. Freeman, E., Kent, E. C., Brohan, P., Cram, T., Gates, L., Huang, B., et  al., (2019). The International Comprehensive Ocean‐Atmosphere Data Set—meeting users’ needs and future priorities. Frontiers of Marine Research. 6, 435. doi: 10.3389/fmars.2019.00435 Freitag, H. P., McPhaden, M. J., & Connell, K. J. (2018). Comparison of ATLAS and T‐Flex Mooring Data. NOAA Tech. Memo. OAR PMEL‐149, NOAA/Pacific Marine Environmental Laboratory, Seattle, WA., 60 pp.

ENSO Observations  61 Fu, L.‐L. & Cazenave, A., Eds. (2001). Satellite altimetry and Earth sciences: A handbook of techniques and applications. Academic Press, San Diego, 463 pp. Halpern, D. (1996). Visiting TOGA’s past. Bulletin of the American Meteorological Society, 77, 233–242. Hamlington, B. D., Leben, R. R., Strassburg, M. W., & Kim, K.‐Y., 2014: Cyclostationary Empirical orthogonal function sea level reconstruction, Geosciences Data Journal, 1, 13–19. Harrison, D. E., & Vecchi, G. A. (1997). Westerly wind events in the tropical Pacific, 1986–95. Journal of Climate, 10, 3131–3156. Hayes, S. P., Mangum, L. J., Picaut, P., Sumi, A. & Takeuchi, A. (1991). TOGA TAO: A moored array for real‐time measurements in the tropical Pacific Ocean. Bulletin of the American Meteorological Society, 72, 339–347. Hou, X., Long, D., Hong, Y., et al. (2016). Seasonal to interannual variability of satellite‐based precipitation estimates in the Pacific Ocean associated with ENSO from 1998 to 2014. Remote Sensing, 8(10), 833. doi: 10.3390/rs8100833 Hu, S., & Fedorov, A. V. (2016). Exceptionally strong easterly wind burst stalling El Niño of 2014. Proceeding of the National Academy of Sciences, 113(8), 2005–2010. Hu, S., & Fedorov, A. V. (2017). The extreme El Niño of 2015– 2016: The role of westerly and easterly wind bursts, and preconditioning by the failed 2014 event. Climate Dynamics, 1–19. Huffman, G. J., Adler, R. F., Bolvin, D. T., & Nelkin E. J. (2010). The TRMM Multi‐satellite Precipitation Analysis (TMPA). In Gebremichael, M., & Hossain, F. (Eds.), Satellite rainfall applications for surface hydrology, Springer, 327 pp. doi:10.1007/978‐90‐481‐2915‐7 Hurlburt, H. E., Kindle, J. C., & O’Brien, J. J. (1976). A numerical simulation of the onset of El Niño. Journal of Physical Oceanography, 6, 621–631. Ji, M., Leetmaa, A., & Derber, J. (1995). An ocean analysis system for climate studies. Monthly Weather Review, 123, 460–481. Jin, F. (1997). An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model. Journal of the Atmospheric Sciences, 54, 811–829. https://doi.org/10.1h175/1520‐0469(19 97)0542.0.CO;2.kj Johnson, G.C., Sloyan, B.M., Kessler, W.S. & McTaggart, K.E. (2002). Direct measurements of upper ocean currents and water properties across the tropical Pacific Ocean during the 1990s. Progress in Oceanography, 52, 31‐61. Johnson, S. J., Stockdale, T. N., Ferranti, L., Balmaseda, M. A., Molteni, F., Magnusson, L., et  al. (2018). SEAS5: The new ECMWF seasonal forecast system. Geoscience Model Development Discussion, 12, 1087–1117. https://doi.org/10.5194/ gmd‐2018‐228 Kessler, W. S., McPhaden, M. J., & Weickmann, K. M. (1995). Forcing of intraseasonal Kelvin waves in the equatorial Pacific. Journal of Geophysical Research, 100, 10,613–10,631. Kessler, W. S., et al. (2019). Second report of TPOS 2020. GOOS‐234, 265 pp. Available online at http://tpos2020.org/ project‐reports/second‐report and https://doi.org/10.13140/ RG.2.2.17161.29289 King, A. D., van Oldenborgh, G. J., & Karoly, D. J. (2016). Climate change and El Niño increase likelihood of Indonesian heat and drought. Bulletin of the American Meteorological Society, 97, S113–S117.

Knox, R., & Halpern, D. (1982). Long range Kelvin wave propagation of transport variations in Pacific Ocean equatorial currents. Journal of Marine Research., 40(Suppl.), 329–339. Laloyaux, P., Balmaseda, M. A., Dee, D., Mogensen, K., & Janssen, P. (2016). A coupled data assimilation system for climate reanalysis. Quarterly Journal of the Royal Meteorological Society, 142(694), 65–78. Laloyaux, P., de Boisseson, E., Balmaseda, M. A., Bidlot, J.‐R., Broennimann, S., Buizza, R., et  al. (2018). CERA‐20C: A coupled reanalysis of the twentieth century. Journal of Advances in Modeling Earth Systems, 10, 1172–1195. https:// doi.org/10.1029/2018MS001273 Lee, T., & McPhaden, M. J. (2010). Increasing intensity of El Niño in the central‐equatorial Pacific. Geophysical Research Letters, 37, L14603. doi:10.1029/2010GL044007. Lengaigne, M., Guilyardi, E., Boulanger, J. P., Menkes, C., Delecluse, P., Inness, P., et al. (2004). Triggering of El Niño by westerly wind events in a coupled general circulation model. Climate Dynamics, 23, 601–620. Levine, A.F.Z., & McPhaden, M. J. (2016). How the July 2014 easterly wind burst gave the 2015‐6 El Niño a head start. Geophysical Research Letters, 43, 6503–6510. doi: 10.1002/2016GL069204. Levine, A.F.Z., Jin, F. F., & McPhaden, M. J. (2016). Extreme noise‐extreme El Niño: How state‐dependent noise forcing creates El Niño‐La Niña asymmetry. Journal of Climate, 29, 5483–5499. L’Heureux, M. L., et al. (2017). Observing and predicting the 2015/16 El Niño. Bulletin of the American Meteorological Society, 98, 1363–1382. Liu, W.T. (2002). Progress in scatterometer application. Journal of Oceanography, 58, 121–136. Lukas, R., & E. Lindstrom, E. (1991). The mixed layer of the western equatorial Pacific Ocean, Journal of Geophysical Research, 96(Suppl.), 3343–3357. McCreary, J. P. (1976). Eastern tropical ocean response to changing wind systems, with application to El Niño. Journal of Physical Oceanography, 6, 632–645. McPhaden, M. J. (1995). The Tropical Atmosphere‐Ocean Array is completed. Bulletin of the American Meteorological Society, 76, 739–741. McPhaden, M. J. (1999). Genesis and evolution of the 1997–98 El Niño. Science, 283, 950–954. McPhaden, M. J. (2002). Mixed layer temperature balance on intraseasonal time scales in the equatorial Pacific Ocean. Journal of Climate, 15(18), 2632–2647. McPhaden, M. J. (2006). El Niño and ocean observations: A personal history. In: M. Jochum & R. Murtugudde (Eds.), Physical oceanography, developments since 1950. Springer, New York, 79–99. McPhaden, M. J. (2012). A 21st century shift in the relationship between ENSO SST and warm water volume anomalies. Geophysical Research Letters, 39, L09706. doi: 10.1029/­ 2012GL051826 McPhaden, M. J. (2015). Playing hide and seek with El Niño. Nature Climate Change, 5, 791–795. McPhaden, M.J., Busalacchi, A.J., Cheney, R., Donguy, J.R., Gage, K.S., Halpern, D., et  al., (1998). The Tropical

62  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Ocean‐Global Atmosphere (TOGA) observing system: A decade of progress. Journal of Geophysical Research, 103, 14,169‐14,240. McPhaden, M. J., Zebiak, S. E., & Glantz, M. H. (2006). ENSO as an integrating concept in Earth science. Science, 314, 1740–1745. McPhaden, M. J., Busalacchi, A. J., & Anderson, D.L.T. (2010). A TOGA retrospective. Oceanography, 23, 86–103. Meinen, C. S., & McPhaden, M. J. (2000). Observations of warm water volume changes in the equatorial Pacific and their relationship to El Niño and La Niña. Journal of Climate, 13, 3551–3559. Menkes, C. E., Lengaigne, M., Vialard, J., Puy, M., Marchesiello, P., Cravatte, S., & Cambon, G. (2014). About the role of westerly wind events in the possible development of an El Niño in 2014, Geophysical Research Letters, 41, 6476–6483. doi:10.1002/2014GL061186 Neske, S., & McGregor, S. (2018). Understanding the warm water volume precursor of ENSO events and its interdecadal variation. Geophysical Research Letters, 45, 1577–1585. https://doi.org/10.1002/2017GL076439 Paek, H., Yu, J.‐Y., & Qian, C. (2017). Why were the 2015/16 and 1997/98 Extreme El Niños different? Geophysical Research Letters, 44, 1848–1856. Penny, S. G., & Hamill, T. M. (2017). Coupled data assimilation for integrated Earth system analysis and prediction. Bulletin of the American Meteorological Society, 98, ES169–ES172. https://doi.org/10.1175/BAMS‐D‐17‐0036.1 Philander, S.G.H., Hurlin, W. J., & Pacanowski, R. C. (1986). Properties of equatorial long waves in models of the seasonal cycle in the tropical Atlantic and Pacific Oceans. Journal of Geophysical Research, 91, 14,207–14,211. Picaut, J., Hackert, E., Busalacchi, A. J., & Murtugudde, R. (2002). Mechanisms of the 1997‐1998 El Nino La Nina, as inferred from space‐based observations. Journal of Geophysical Research, 107, 3037. doi: 10.1029/2001JC000850 Planton, Y., Vialard, J., Guilyardi, E., Lengaigne, M., & Izumo, T. (2018). Western Pacific oceanic heat content: A better predictor of La Niña than of El Niño. Geophysical Research Letters, 45(18), 9824–9833. doi: 10.1029/2018GL0793 Poli, P., Hersbach, H., Dee, D. P., Berrisford, P., Simmons, A. J., Vitart, F., et al. (2016). ERA‐20C: An atmospheric reanalysis of the twentieth century. Journal of Climate, 29(11), 4083–4097. Praveen Kumar, B., Vialard, J., Lengaigne, M., Murty, V.S.N., McPhaden, M.J., Cronin, M.F., et al. (2013). TropFlux wind stresses over the tropical oceans: Evaluation and comparison with other products. Climate Dynamics, 40, 2049–2071. doi: 10.1007/s00382‐012‐1455‐4 Puy, M., Vialard, J., Lengaigne, M., Guilyardi, E., Voldoire, A., Balmaseda, M A., et al. (2019). Influence of Westerly Wind Events stochasticity on El Niño amplitude: The case of 2014 vs. 2015. Climate Dynamics, 52, 7435–7454. doi: 10.1007/ s00382‐017‐3938‐9 Reynolds, R. W., Smith, T.M., Liu, C., Chelton, D.B., Casey, K., & Schlax, M.G. (2007). Daily high‐resolution‐blended analyses for sea surface temperature. Journal of Climate, 20, 5473–5496.

Riser, S. C., et al. (2016). Fifteen years of ocean observations with the global Argo array. Nature Climate Change, 6, 145– 153. doi: 10.1038/NCLIMATE2872. Rodriguez, E., Bourassa, M.A., & Chelton, D.B. (2019). The Winds and Currents Mission concept. Frontiers in Marine. Science. https://doi.org/10.3389/fmars.2019.00438 Saha, S., Moorthi, S., Pan, H., Wu, X., Wang, J., Nadiga, S., et  al. (2010). The NCEP climate forecast system reanalysis. Bulletin of the American Meteorological Society, 91, 1015–1057. Saha, S., Nadiga, S., Thiaw, C., Wang, J., Wang, W., Zhang, Q., et al. (2006). The NCEP Climate Forecast System. Journal of Climate, 19, 3483–3517. Santoso, A., McPhaden, M. J., & Cai, W. (2017). The defining characteristics of ENSO extremes and the strong 2015/2016 El Niño. Reviews of Geophysics, 55(4), 1079–1129. Stammer, D., et al. (2002). The global ocean circulation during 1992–1997, estimated from ocean observations and a general circulation model. Journal of Geophysical Research, 107, 3118. doi: 10.1029/2001JC000888 Stammer, D., Balmaseda, M. A, Heimbach, P., Köhl, A. & Weaver, A. (2016). Ocean data assimilation in support of climate applications: Status and perspectives. Annual Review of Marine Science, 8, 491–518. Stockdale, T. N., Busalacchi, A. J., Harrison, D. E., & Seager, R. (1998). Ocean modeling for ENSO. Journal of Geophysical Research, 103, 14,325–14,355. Suarez, M. J., & Schopf, P. S., (1988). A delayed action oscillator for ENSO. Journal of the Atmospheric Sciences, 45, 3283–3287. https://doi.org/10.1175/1520‐0469(1988)0452.0.CO;2 Tollefson, J. (2014). El Niño in failure mode. Nature. doi:10.1038/ nature.2014.14582 Vecchi, G. A., & Harrison, D. E. (2000). Tropical Pacific sea surface temperature anomalies, El Niño, and equatorial westerly wind events. Journal of Climate, 13, 1814–1830. Wang, W., & McPhaden, M. J. (1999). The surface layer heat balance in the equatorial Pacific Ocean, Part I: Mean seasonal cycle. Journal of Physical Oceanography, 29, 1812–1831. Wang, W., and McPhaden, M. J. (2000). The surface layer heat balance in the equatorial Pacific Ocean. Part II: Interannual variability. Journal of Physical Oceanography, 30, 2989–3008. Wang, W. & McPhaden, M. J. (2001). Surface layer heat balance in the equatorial Pacific Ocean during the 1997–98 El Niño and the 1998–99 La Niña. Journal of Climate, 14, 3393–3407. Wolter, K., & Timlin, M. S. (2011). El Niño/Southern Oscillation behaviour since 1871 as diagnosed in an extended multivariate ENSO index. International Journal of Climatology, 31(7), 1074–1087. Wyrtki, K. (1974a). Equatorial currents in the Pacific 1950 to 1970 and their relations to the trade winds. Journal of Physical Oceanography, 4, 372–380. Wyrtki, K. (1974b). Sea level and the seasonal fluctuations of the equatorial currents in the western Pacific Ocean. Journal of Physical Oceanography, 4, 91–103.

ENSO Observations  63 Wyrtki, K. (1975a). Fluctuations of the dynamic topography in the Pacific Ocean. Journal of Physical Oceanography, 5, 450–459. Wyrtki, K. (1975b). El Niño: The dynamic response of the equatorial Pacific Ocean to atmospheric forcing. Journal of Physical Oceanography, 5, 572–584. Xie, P., Boyer, T., Bayler, E., Xue, Y., Byrne, D., Reagan, J., et al. (2014). An in situ–satellite blended analysis of global sea surface salinity. Journal of Geophysical Research, 119, 6140–6160. doi: 10.1002/2014JC010046 Xue, Y., & Kumar, A. (2017). Evolution of the 2015/16 El Niño and historical persp ective since 1979. Science China, 60, 1572–1588. Xue Y., Wen, C., Kumar, A., Balmaseda, M. A., Fujii, Y., Alves,  O., et  al. (2017). A real‐time ocean reanalyses intercomparison project in the context of tropical pacific

observing system and ENSO monitoring. Climate Dynamics, 49, 3647–3672. doi: 10.1007/s00382‐017‐3535‐y Yu, L., & Weller, R.A. (2007). Objectively Analyzed air‐sea heat Fluxes (OAFlux) for the global ice‐free oceans. Bulletin of the American Meteorological Society, 88, 527–539. Zhang, S., Harrison, M. J., Rosati, A., & Wittenberg, A. T. (2007). System design and evaluation of coupled ensemble data assimilation for global oceanic climate studies. Monthly Weather Review, 135, 3541–3564. doi:https://doi.org/10.1175/ MWR3466.1 Zhang, W., et  al. (2016). Influences of natural variability and anthropogenic forcing on the extreme 2015 accumulated ­cycl­one energy in the western north Pacific. Bulletin of the American Meteorological Society, 97, S131–S135.

4 ENSO Diversity Antonietta Capotondi1,2, Andrew T. Wittenberg3, Jong‐Seong Kug4, Ken Takahashi5, and Michael J. McPhaden6

ABSTRACT ENSO events display large interevent differences in amplitude, spatial pattern, and temporal evolution. The differ­ ences in spatial pattern, which have important consequences for ENSO teleconnections and societal impacts, have become known as “ENSO diversity.” In this chapter we review key aspects of ENSO diversity, including ENSO’s surface and subsurface characteristics, underlying dynamics, predictability, low‐frequency variations, and long‐ term evolution, as well as the representation of ENSO diversity in climate models. To better understand the origin of ENSO diversity and identify specific characteristics of different event types, many different classification schemes have been proposed. Here we describe these different approaches and the insights they may provide on the nature of event‐to‐event differences. The last two decades have seen a greater number of El Niño events with the largest sea surface temperature anomalies in the central Pacific. Current research seeks to determine whether such changes in ENSO characteristics were the result of anthropogenic greenhouse gas forcing or just a manifes­ tation of natural variability, and whether and how climate change may affect ENSO diversity in the future.

4.1. INTRODUCTION

equatorial Pacific (~170°W, Figure  4.1b). This event marked the start of a seemingly atypical ENSO evolution during the following decade, with relatively weaker and more frequent El Niño events, occurring approximately every two years, and whose largest SSTAs were in the central equatorial Pacific, albeit with differences in the detailed spatial pattern and evolution (e.g., the 2009–2010 event, Figure  4.1c). The 2000–2014 period culminated with the strong 2015–2016 El Niño event, which exhibited SSTAs comparable in magnitude to those of the 1997– 1998 event, but displaced further west than in 1997–1998 (Figure 4.1d). A commonly reliable predictor of El Niño events, the equatorial Pacific upper‐ocean warm water volume, became less useful as an ENSO precursor during 2000–2014, suggesting that some aspects of ENSO dynamics involving the evolution of the thermocline were different than in preceding decades (McPhaden, 2012; Luebbecke & McPhaden, 2014; Cai et al., 2018). In 2017, a “coastal El Niño” produced severe flooding in Peru, even though ENSO was neutral (Garreaud, 2018; Takahashi et al., 2018; Z.‐Z. Hu et al., 2019; Peng et al.,

The first El Niño of the 21st century in 2002–2003 was different than preceding El Niños, especially the 1997– 1998 event (McPhaden, 2004). While the 1997–1998 El Niño achieved extreme sea surface temperature anomaly (SSTA) values in the eastern equatorial Pacific (Figure  4.1a), the largest SSTAs in the winter of 2002– 2003 were weaker and primarily confined to the central  University of Colorado, CIRES, Boulder, CO, USA  NOAA Physical Sciences Laboratory, Boulder, CO, USA 3  NOAA Geophysical Fluid Dynamics Laboratory, Princeton, NJ, USA 4  Division of Environmental Science & Engineering, Pohang University of Science and Technology (POSTECH), Pohang, Republic of Korea 5  Servicio Nacional de Meteorología e Hidrología del Perú— SENAMHI, Lima, Peru 6  NOAA Pacific Marine Environmental Laboratory, Seattle, WA, USA 1 2

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 65

66  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a) DJF 1997/98

(b) DJF 2002/03

20N

20N

0

0

20S 120E

20S 150E

180

150W

120W

90W

120E

150E

(c) DJF 2009/10

(d) DJF 2015/16

20N

20N

0

0

20S

20S

120E

150E

180

150W –4

120W

120E

90W

–3.2 –2.4 –1.6 –0.8

0

0.8

150E 1.6

2.4

180

150W

120W

90W

180

150W

120W

90W

3.2

4

Figure 4.1 Interannual SST anomalies during December‐January‐February (DJF) for the El Niño events of (a) 1997–1998, (b) 2002–2003, (c) 2009–2010, and (d) 2015–2016. Monthly SST data were obtained from the NOAA Optimum Interpolation (OISST) data set (Reynolds et al., 2002). Anomalies are computed relative to the 1982–2017 climatology.

2019). This changing character of ENSO has stimulated a renewed interest in the nature of this powerful phenomenon, and in the fundamental causes of event‐to‐ event differences. Research on ENSO diversity has pri­ marily focused on the El Niño phase, since La Niña SSTAs tend to peak in the central Pacific with more limited longitudinal excursions (Kug & Ham, 2011), although subsurface characteristics can actually show greater diver­ sity for La Niña than El Niño (Ashok et al., 2017). While these event‐to‐event differences were initially described in terms of two different types, or “flavors,” of El Niño, commonly referred to as eastern Pacific (EP) and central Pacific (CP) following the definition intro­ duced by Kao and Yu (2009), it has become increasingly clear that El Niño events (and to a lesser degree, La Niña events) exhibit a wide spectrum of spatial structures. In particular, the location of maximum SSTAs along the equator spans a broad range of longitudes rather than clustering around only two locations (Giese & Ray, 2011), except perhaps for the extreme El Niño in the eastern Pacific (Takahashi et al., 2011, 2018). This wide range of spatial patterns, whose precise statistical distribution is somewhat clouded by observational uncertainties (Marathe et  al., 2015), is illustrated by the longitudinal profiles of SSTAs along the equator for the El Niño events that occurred during 1951–2017 (Figure  4.2). Although these profiles may be broadly classified in two different groups, a large diversity of longitudinal ­structures can be seen among events. Indeed, to account for events that share elements of both EP and CP types, Kug et al. (2009) introduced a “mixed” El Niño type, with largest SSTAs in the Niño‐3.4 region (5°S–5°N, 170°–120°W).

El Niño events with SST anomalies that extended all the way to the eastern Pacific during some periods of their evolution, like the 2009 and 2014 CP events, were described as “basin‐wide warming events” by Ashok et al. (2012) and Jadhav et al. (2015). It is evident from Figure 4.2 that CP events (blue‐dashed lines) tend to be generally weak to moderate in strength, while events with the largest anomalies in the eastern equatorial Pacific exhibit a broader range of amplitudes from relatively weak to extreme (red dashed lines in Figure  4.2). An exception to the above description is the recent 2015–2016 El Niño, which achieved SSTAs typical of extreme events in the Niño‐3.4 region (5°S–5°N, 120°–170°W) but exhib­ ited only half of the amplitude seen in the previous extreme events in the far eastern Pacific (Figures 4.1d and 4.2; L’Heureux et al., 2017; Newman et al., 2018), so that the peak SSTA was located closer to the central Pacific. Empirical dynamical models constructed from observed SSTs, thermocline depth, and zonal surface wind stress anomalies over the period 1959–2000 (Newman et al., 2011) indicate the presence of growing modes similar to EP and CP El Niño events. These results suggest that the broad range of observed El Niño types may arise from the superposition of these basic modes/ structures, which themselves result from different dynam­ ical balances, as mediated by the background conditions. Due to their differences in growth rate and dominant timescale, these two modes can give rise to complex ENSO behavior in both spatial and temporal domains (Timmerman et al., 2018). In this chapter, we provide a synthesis of our current understanding of ENSO diversity. Section 4.2 reviews the

ENSO Diversity  67 5.0 EP 4.0

CP

82/83

97/98

SST (°C)

3.0 15/16

2.0 1.0 0.0 –1.0 135E

180

135W

90W

Figure 4.2 Equatorial SST anomaly profiles averaged over 2°S–2°N during winter (DJF) for CP (blue) and EP (red) El Niño events over the period January 1951–December 2017. The EP and CP events have been identified using the Niño‐3 and Niño‐4 indices: The DJF Niño‐3 index is larger than 0.5°C and larger than the Niño‐4 index for EP events, while the Niño‐4 index is larger than 0.5°C and larger than the Niño‐3 index for CP events. Thin dashed lines show the profiles of individual events, while the thick red and blue solid lines indicate the composite profiles for EP and CP El Niño events, respectively. Monthly SST data were obtained from the NOAA Extended Reconstructed SST dataset version 5 (ERSSTv5; Huang et  al., 2017). Anomalies were computed relative to the 1951–2017 climatology, and linearly detrended prior to the profile calculation. The extreme events of 1982– 1983, 1997–1998, and 2015–2016 are labeled.

large variety of approaches and indices proposed to capture the differences in El Niño spatial patterns, while section  4.3 describes the leading dynamical processes underlying different El Niño types. The precursors of these different El Niño types, and their influence on the predictability of ENSO diversity, are discussed in sec­ tion  4.4; the low‐frequency modulation and long‐term trend of ENSO diversity are presented in section 4.5; and the ability of climate models to simulate ENSO diversity is discussed in section 4.6. Conclusions and future direc­ tions are presented in section 4.7. 4.2. CHARACTERISTICS OF ENSO DIVERSITY It has long been recognized that “no two El Niño events are quite alike” (Wyrtki, 1975), but only in the early 2000s were these differences more systematically classified through the introduction of specific indices to charac­ terize different types of El Niño events. The need for a more systematic classification has largely been motivated by the recognition that the details of ENSO spatial pat­ terns may play an important role in ENSO teleconnec­ tions and societal impacts. Larkin and Harrison (2005),

for instance, noticed that “conventional” El Niños (events with their largest SSTAs in the eastern Pacific) were asso­ ciated with surface temperature and precipitation anom­ alies over the US that differed in spatial pattern, and in some locations even in sign, from those associated with “dateline” El Niños (events with their largest SSTAs in the central Pacific near the dateline). Although the statistical significance of the results was not very high due to the small sample size of the two classes of events, the possi­ bility that El Niño spatial pattern could exert such an influence on quantities of large societal importance drew much attention to the diversity of ENSO events. Ashok et  al. (2007) further emphasized the importance of the location of equatorial Pacific warming for atmospheric teleconnections, in particular those associated with the summer Indian monsoon as well as precipitation over Korea and Japan. Ashok et al. (2007) identified a specific pattern of SSTAs, characterized by positive anomalies in the central equatorial Pacific and cold anomalies in the eastern and western parts of the basin, which appeared responsible for those teleconnections, and which they called “El Niño Modoki,” a Japanese word that means “similar but different” (see the appendix to this chapter for a definition of the El Niño Modoki index). Further discussion of the influence of ENSO diversity on telecon­ nections is provided in chapter 14. The papers by Larkin and Harrison (2005) and Ashok et al. (2007) stimulated intense research activity on ENSO diversity (Capotondi et al., 2015a). To better characterize this diversity, it is useful to classify El Niño events into different categories to more easily identify differences in the leading dynamical processes, precursors, and impacts. To that end, several indices have been introduced to prop­ erly capture aspects of these El Niño groups, and different names have been suggested to define them. These names include, among others, Dateline El Niño and El Niño Modoki as mentioned above, and also warm pool (vs. cold tongue El Niño) and central Pacific (vs. eastern Pacific) El Niño. For simplicity, here we will refer to events with the largest SSTA in the eastern Pacific as “EP” and those with the largest SSTA in the central Pacific as “CP” El Niño types. A list of the most common indices introduced to classify these types is provided in the appendix. Different definitions, and the exact details of their implementation, can lead to differences in the class­ ification of individual events. For the NOAA‐ERSSTv5 data set, Table 4.1 shows that some events can be either EP or CP depending on the indices used (e.g. 1965–1966 and 1991–1992), and the season chosen to define El Niño events can also play a role in the event classification. A dependency on the specific SST data set used can also be expected (Diamond & Bennartz, 2015). Despite these discrepancies, similar average spatial characteristics of EP and CP events emerge with most of

68  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

the proposed indices. Since the Niño‐3 region (5°S–5°N, 90°–150°W) is located in the eastern Pacific, and the Niño‐4 region (5°S–5°N, 160°E‐150°W) in the central Pacific, the Niño‐3 and Niño‐4 indices (SSTAs averaged over the Niño‐3 and Niño‐4 regions, respectively) have been widely used to identify EP and CP El Niño events (Kug et  al., 2009; Yeh et  al., 2009; Kug et  al., 2010; Capotondi, 2013; see the appendix for definitions). Composites of the EP and CP events obtained with this approach (Figure 4.3) show that EP events have SSTAs that peak further east, as expected, but are also, on average, stronger than CP events. Sea surface height (SSH) anomalies, which are dynamically linked to ther­

mocline depth, show a strong zonal dipole in the case of EP events (Figure  4.3, top left), indicative of a deeper thermocline in the eastern and shallower thermocline in the western equatorial Pacific. CP events (Figure 4.3, top right), on the other hand, are characterized by weaker positive thermocline depth anomalies that extend further westward than for EP events, and negative thermocline depth anomalies confined to the far western Pacific. Sea level pressure (SLP) anomalies (Figure 4.3, middle panels) in the tropical Pacific exhibit the zonal seesaw typical of the Southern Oscillation (SO), with higher than average pressure in the western part of the basin and lower than average pressure in the eastern tropical Pacific. These

Table 4.1  Classification of El Niño events based on some of the commonly used indices described in the appendix. These indices include the Niño‐3/Niño‐4 approach, using either 0.5°C or one standard deviation as thresholds for event identification (first two columns); the E and C indices of Takahashi et al. (2011, column 3); the Modoki index (EMI, Ashok et al., 2007) to identify CP‐Modoki events (column 4), and the EPnew and CPnew indices of Sullivan et al. (2016). Notice that for the EMI we do not apply the criterion that the anomalous warming in the central Pacific must persist from boreal summer through winter as in Ashok et al. (2007). When this criterion is applied, the 2009–2010 El Niño does not qualify as an El Niño Modoki. Monthly SSTs from the NOAA‐ERSSTv5 data set over the period 1951–2017 have been used to prepare the table. EP events (denoted with “E”) are highlighted in red, while CP events (denoted with “C”) are shown in green. Since different winter seasons have been used in the literature to select El Niño events, we have considered both the November‐December‐January (NDJ), and the December‐January‐February (DJF) seasons. In those cases when results differ in the two seasons, subscripts are used to indicate the season. Notice that in some cases events start as EP events in NDJ and evolve into CP events in DJF, or vice versa, as indicated by the arrow. The blank entries indicate cases in which the SSTA conditions were not detected as an El Niño with those indices. In the “Consensus” column, an El Niño is registered as a consensus event if it gets at least two votes from among the five different methods. It is then labeled as E or C if it has a margin of at least two votes in favor of either E or C, respectively, while a “no consensus” case is indicated by a question mark. Niño‐3/Niño‐4 > 0.5°C 1951–1952 1953–1954 1957–1958 1958–1959 1963–1964 1965–1966 1968–1969 1969–1970 1972–1973 1976–1977 1977–1978 1979–1980 1982–1983 1986–1987 1987–1988 1991–1992 1994–1995 1997–1998 2002–2003 2004–2005 2006–2007 2009–2010 2014–2015 2015–2016

E C E CNDJ E E C ENDJ → CDJF E E C ENDJ → CDJF E E ENDJ → CDJF E C E ENDJ → CDJF C ENDJ → CDJF C C E

Niño‐3/Niño‐4 > 1 std

E/C

Modoki

EPnew/CPnew

Consensus

ENDJ

E

C

E CNDJ

C C

E C

E C ? ? C

C

E E C

E E C

E E C E C E C C C C C E

E E C ? C E C C C C C E

ENDJ E

C

E C

CNDJ C C

E

E

E EDJF C E C E CNDJ C CNDJ C

E CNDJ → EDJF C C C E C C CNDJ C

E

ENDJ → CDJF

C C CDJF C C CDJF

ENSO Diversity  69

SLP anomalies are associated with westerly wind anom­ alies along the equator, which are part of the positive feedback known as the Bjerknes feedback, that promotes El Niño anomaly growth. While EP events (Figure 4.3c) exhibit strong westerly wind anomalies extending to ~120°W, with northerly wind anomalies further east, CP events (Figure  4.3d) have a weaker equatorial SLP gra­ dient that is located further to the west, so that the asso­ ciated westerly wind anomalies are weaker and confined westward of ~150°W, while southeasterly wind anomalies are seen in the eastern part of the basin (Harrison & Chiodi, 2009). Similarly, precipitation anomalies (Figure  4.3, bottom panels) extend all the way to the South American coast in the case of EP events, while they are limited to the western part of the basin during CP events. In the extratropical North Pacific, El Niño events are associated with a deepened and eastward‐extended Aleutian Low, and with a deeper thermocline and warmer conditions along the West Coast of North America. Positive anomalies of SST and SSH are also seen along the coast of South America. These El Niño influences, which are critical for marine ecosystem dynamics along the west coasts of the Americas, appear to be more pro­ nounced during EP events. Although useful for classifying ENSO events, the Niño‐3 and Niño‐4 indices are highly correlated with each other (the correlation coefficient for the indices used to produce Figure 4.3 is 0.84), so that they are not suit­ able indices for describing the distinctive evolution of EP and CP events. Other indices that are largely uncorrelated with each other have therefore been introduced. For example, the EPnew and CPnew indices (see the appendix for definition) proposed by Sullivan et al. (2016) have been used to highlight the differences in skewness and spectral characteristics of the EP and CP events, as illustrated in Figure 4.4 for the time period 1951–2017. While the EPnew index has a positive skewness (warm events tend to be larger than cold events; Figure  4.4a) and has spectral peaks at about 1.2 and 3–5 years (Figure 4.4b), the CPnew index displays a negative skewness (negative events tend to be larger than positive events, Figure  4.4c) with enhanced spectral power around 2–2.5 years and in the decadal range (Figure  4.4d). CP activity with a quasi‐ biennial timescale appears, indeed, to undergo a quasi‐ decadal modulation, with multiyear periods dominated by CP La Niñas followed by multiyear periods populated with CP El Niños, as highlighted by the seven‐year low‐ pass filtered time series in Figure 4.4c. The predominance of CP events in the early 21st century, as seen in Figure 4.4c, had an imprint on the equatorial sea level, which was above average in the central Pacific and below average near the eastern and western boundaries during 2000–2004 (Behera & Yamagata, 2010). Differences in skewness and spectral characteristics were also noticed by

Yu et al. (2011) using the subsurface EP and CP indices (see appendix), but no decadal variability was identified with those indices. As stressed by Trenberth and Stepaniak (2001), two indices are needed to characterize differences in spatial patterns and temporal evolution of El Niño events. Since ENSO‐related equatorial SST variability has low dimen­ sionality (Karamperidou et  al. 2014), the two leading empirical orthogonal functions (EOFs) of SSTA can jointly explain a large fraction of the variance of typical ENSO SST indices (Takahashi et al., 2011). The leading EOF of equatorial SSTAs resembles a canonical El Niño pattern, with the largest anomalies in the Niño‐3.4 region and the same sign in the central and eastern equatorial Pacific, while the second EOF exhibits differences in sign between the eastern and central Pacific. Thus, the linear combination of these two EOFs, as described in the appendix, produces patterns characterized by enhanced warming in the far eastern (E pattern) or central (C pattern) equatorial Pacific, as seen in Figures  4.6a and 4.6b, respectively. Similar patterns can be obtained by considering alternative quasi‐orthogonal indices, like the Niño‐3 index and the Trans‐Niño index (TNI; Trenberth  & Stepaniak, 2001; see appendix for defini­ tion) as descriptors of ENSO diversity. The spatial pattern associated with the TNI displays a zonal SST dipole in the equatorial region, similar to that of the sec­ ond EOF of SST, so that a linear combination of the Niño‐3 and TNI indices results in patterns that are very similar to the E and C patterns (Santoso et al., 2017). The relationship between various indices of ENSO diversity in the space of the two leading principal components (PCs, the projection of the SSTA on the two leading EOF patterns at each time step) is shown in Figure 4.5. Because most of the SSTA variance lies near the PC1/PC2 plane, the linear correlation between any two indices is approxi­ mately given by the cosine of the angle between their corresponding axes in Figure 4.5. Thus, orthogonal axes indicate maximum independence between two indices, as is the case of the E and C indices, which are derived from the first two EOF modes and represent SST variability that can be exclusively attributed to the eastern and central equatorial Pacific, respectively (Takahashi et al., 2011). The extreme El Niño events of 1877–1878, 1982– 1983, and 1997–1998 had much higher values of E (as well as Niño‐1+2 and EPnew) than any other year, and it has been suggested that these events belong to a distinct dynamical regime (Takahashi et  al., 2011; Takahashi & Dewitte, 2016). Thus, El Niño amplitude diversity (strong vs. moderate/weak) is an important aspect to consider. El Niño events also differ in their temporal evolution. Using lag‐correlation analysis, Kao and Yu (2009) showed that EP El Niño SSTAs tend to develop in the eastern equatorial Pacific and propagate westward, as in

EP

(a)

CP

(b)

60N

60N

30N

30N

0

0

30S

30S

60S

60S 120E

150E

180

150W

120W

90W

(d)

(c) 60N

60N

30N

30N

0

0

30S

30S

60S

60S 120E

150E

180

150W

120W

90W

120E

150E

180

150W

120W

90W

120E

150E

180

150W

120W

90W

120E

150E

180

150W

120W

90W

(f)

(e)

60N

60N

30N

30N

0

0

30S

30S

60S

60S 120E

150E

180

150W

–2

–1.6

120W

–1.2

90W

–0.8

–0.4

0

0.4

0.8

1.2

1.6

2

Figure 4.3  Composites of anomalous SST (shaded, °C, same in each row), SSH (top, contours), SLP/Winds (middle, contours/vectors), and Precipitation (bottom, contours) for EP (left), and CP (right) El Niño events during winter (DJF). Contour intervals are 2 cm for SSH, 0.7 millibars for SLP, and 0.7 mm/day for precipitation. Vector winds are in m/s. Dashed contours indicate negative values. The EP and CP events have been identified using the Niño‐3– Niño‐4 index approach (appendix). Events are considered EP when the DJF Niño‐3 index is larger than the DJF Niño‐4 index, and larger than 0.5°C, while CP events are characterized by the DJF Niño‐4 index being larger than the DJF Niño‐3 index, and larger than 0.5°C. The SST and SSH fields are obtained from the ECMWF ORAS4 ocean reanalysis (Balmaseda et al., 2013) over the period January 1958–December 2015, SLP and surface winds are from the NCEP‐NCAR reanalysis (Kalney et al., 1996), while precipitation is obtained from a reconstructed data set over the global land and ocean (Chen et al., 2002). Anomalies are relative to the 1958–2015 climatology. All fields have been linearly detrended prior to the composite calculation.

ENSO Diversity  71 (a)

(b) s = 1.22

4.0

8.0

EPnew

Power x Frequency

6.0 2.0

0.0

4.0

2.0 –2.0 0.0 1960

1970

1980

1990

2000

10–1

100 101 Period (year)

102

10–1

100

102

2010

(c)

(d) 8.0 s = –0.53

4.0

CPnew

Power x Frequency

6.0 2.0

0.0

4.0

2.0

–2.0 0.0 1960

1970

1980

1990

2000

2010

101

Period (year)

Figure 4.4  (a) and (c) show the time series of the EPnew and CPnew indices, as defined by Sullivan et al. (2016), while (b) and (d) show the corresponding spectra in variance preserving form. Both time series are normalized to unit standard deviation. Notice that both indices are above one standard deviation (dashed line) during the 2015–2016 El Niño event (black arrow in a and c), consistent with the classification of this event as a mixture of CP and EP types (Paek et al., 2017). The CPnew index has a spectral peak close to 10 years; this quasi‐decadal component is evident in the 7‐year low‐pass filtered time series (thick black line in c). The s‐values indicate skewness. Monthly SST data are from the ERSSTv5 dataset over the period 1951–2017.

the case of the “canonical El Niño” described by Rasmusson and Carpenter (1982), while CP events develop in the central Pacific near the dateline without a clear zonal propagation direction (Xiang et  al., 2013). These central equatorial Pacific anomalies are often the equatorial signature of SSTAs that extend from the U.S. West Coast near Baja California toward the equator, associated with the Pacific meridional mode (Chiang & Vimont, 2004), as further discussed in section  4.4. The statistical description of the EP event propagation out­ lined by Kao and Yu (2009) has some notable exceptions,

like the extreme El Niño events of 1982–1983 and 1997– 1998. SSTAs during 1982 events developed in the central Pacific and propagated eastward, while during 1997 SSTAs above 1°C appeared simultaneously in the western and eastern Pacific during February and merged in the central Pacific. In contrast to the east or west direction of SSTA propagation along the equator during El Niño events, SSTAs during La Niña events generally propagate only to the west (McPhaden & Zhang, 2009). Another approach to identify “flavors” of spatiotem­ poral diversity makes use of EOF analysis of the

72  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE 4

2

CPnew* (R2 = 88%)

C -in de x

EMI* (R2 = 87%)

4

–4

Nino4*(K) (R2 = 96%)

2

2 2010

2 1941

1

–2

2

3

1973 2 2016

PC2

0 2017 1

1925

Nino3* (K) (R2 = 96%)

3

1 2 1878

3 1983 2

2

–2 –2

1889 2

1

2

4

Nino1+2* (K) (R2 = 89%)

1998 3

–4

Nino3.4* (K) (R2 = 95%)

4

3

EPnew* (R2 = 87%)

E-

in

4

–4

TNI* (R2 = 91%) –2

0

de x

6 2

4

6

PC1

Figure 4.5  December–February (DJF) mean equatorial Pacific (10°S–10°N) SSTA PC1 and PC2 from ERSSTv5 (1870–2018, climatology based on 1981–2010) following Takahashi et al. (2011). Values of various ENSO indices estimated with multiple linear regression with these PCs can be obtained by reading the values from the corresponding axes (variance explained for 1950–2017 indicated as R2).

longitudinal and temporal evolution of the SSTA fields along the equator between 5°S and 5°N (see appendix; Lee et al., 2014; Dewitte & Takahashi, 2019). In addition to capturing the diversity of amplitude and spatial pattern, this method also captures the large interevent diversity of SSTA evolution following the event peak, for example, distinguishing El Niños that persist through boreal spring from more short‐lived El Niños, and distin­ guishing resurgent El Niños from those that transition into La Niñas. These temporal details of ENSO evolu­ tion are critical for ENSO’s remote impacts, since many of the teleconnections and local conditions are strongly affected by the seasonal cycle. For example, Lee et  al. (2016) found that El Niños that persist into boreal spring (such as the 2015–2016 event) are associated with reduced risk of tornado outbreaks over most of the U.S., while early‐terminating El Niños boost the likelihood of tor­ nado outbreaks in the upper Midwest by up to 50% in May. Similarly, strong La Niñas that persist through boreal spring (such as the 1974 and 2011 events) enhance

the likelihood of tornado outbreaks over the Ohio Valley, Southeast U.S., and upper Midwest in boreal spring, while La Niñas that transition to El Niños boost the likelihood of outbreaks in the southern U.S. In another study, Lee et al. (2018) showed that only strong El Niños that persisted into boreal spring (like the 1982 and 1997 events) were associated with increased rainfall over the entire state of California, while transitioning El Niños enhanced rainfall mainly over northern California, and weak El Niños showed little impact on California rainfall. A characterization of ENSO diversity from an ocean energetics perspective is provided by the perturbation available potential energy (APE; Goddard & Philander, 2000; Brown & Fedorov, 2010; Hu et al., 2014), a quantity that measures the energy potentially available to the system from a horizontal redistribution of mass, as a result of the work done by the winds on the ocean. A positive APA corresponds to a steeper than average ther­ mocline along the equator, and vice versa. A strong linear

ENSO Diversity  73 (a)

(b)

20N

20N

0

0

20S

20S

120E

150E

180

120W

150W –1

90W

–0.8

–0.6

–0.4

120E –0.2

(c)

0

0.2

150E 0.4

0.6

180 0.8

1

150W

120W

90W

150W

120W

90W

1990

2000

2010

°C

(d)

20N

20N

0

0

20S

20S

120E

150E

180

150W –0.5

120W

90W

–0.4

–0.2

–0.3

120E –0.1

(e) 4.0

0

0.1

150E 0.2

0.3

180 0.4

0.5

°C

(f) corr = 0.68, Lead = 5

4.0

2.0

2.0

0.0

0.0

–2.0

–2.0

–4.0

corr = 0.83, Lead = 5

–4.0 1960

1970

1980

1990

2000

2010

Years

1960

1970

1980

Years

Figure 4.6  Optimal two‐season precursors (c and d) of the (a) E and (b) C modes of Takahashi et al. (2011). See appendix for definition. The optimal precursors are computed as the tropical SST (shaded) and SSH (contoured) conditions that lead to the largest growth of the E and C indices six months later. If y is either the E or C index, and x the tropical state vector, as characterized by the 20 leading SST and 10 leading SSH EOFs, the optimal precursor xopt can be obtained as the leading right singular vector of the operator H, such that y(t + τ) = H x(t), where H is computed through multiple linear regressions. The time series of the optimal precursors for the E and C modes, computed as the projection of the SST and SSH fields onto the optimal structures at each time step (black lines), are compared in (e) and (f) with the E (red line) and C (blue line) indices, respectively. The largest correlations between the two sets of standardized indices (0.68 for the E index, and 0.83 for the C index) are obtained when the “optimal” indices lead the E and C indices by five to six months. SST and SSH data are from the ECMWF ORAS4 reanalysis during 1958–2015 (Balmaseda et al., 2013).

anticorrelation is found between APA and the Niño‐3/ Niño‐4 ratio for all the El Niño events, with both Niño‐3 and Niño‐4 positive and either index greater than 0.5, indicating a preference for EP events to occur when the zonal slope of the thermocline is reduced relative to CP events (Hu et al., 2014). Indices of ENSO diversity based on outgoing longwave radiation (OLR) have been introduced to more directly

relate ENSO diversity to remote impacts (Chiodi & Harrison, 2010; Johnson & Kosaka, 2016; Williams & Patricola, 2018). OLR anomalies are an indication of atmospheric deep convection, which is a source of atmo­ spheric teleconnections through the excitation of atmo­ spheric Rossby waves. As the atmospheric Rossby waves propagate from the tropics to high latitudes, they alter the extratropical atmospheric circulation and its influence on

74  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

surface air temperature and precipitation. A motivation for using OLR‐based ENSO diversity indices is that deep convection in the eastern tropical Pacific tends to be asso­ ciated with more robust impacts over the U.S. During EP El Niño events, in particular, the eastern Pacific SSTs may exceed the convective threshold (about 27.5°C; Graham & Barnett, 1987; Takahashi & Dewitte, 2016), so that convection can shift eastward and affect atmospheric teleconnections. Indeed, Figure 4.3 shows that precipita­ tion, which is an indicator of deep convection, generally extends to the eastern tropical Pacific during EP events. Because of the relatively short OLR record dating back to 1979, however, the teleconnections associated with the  eastern Pacific OLR signature are still relatively uncertain. The EP and CP events that are identified by the various indices listed in the appendix have SSTAs that extend over a large portion of the equatorial Pacific. A different class of El Niño events includes cases in which high positive SSTAs develop rapidly along the coast of South America in the far eastern equatorial Pacific in boreal winter and spring, while the rest of the equatorial Pacific remains cold or neutral. Such events, called “coastal El Niños” (Takahashi & Martinez, 2019) can have devastating impacts on flooding in western South American countries, and are associated with a strengthening of the Intertropical Convergence Zone (ITCZ) south of the equator, and northerly wind anomalies across the equator in the far eastern Pacific. Notable examples were observed in 1891, 1925 (Schott, 1931; Takahashi & Martinez, 2018), and more recently in the boreal spring of 2017, a period char­ acterized by torrential rains over the northern coastal areas of Peru, causing enormous losses of life and property (Fraser, 2017; Garreaud, 2018; Hu et al., 2019; Takahashi et al., 2018; Peng et al., 2019). 4.3. EQUATORIAL DYNAMICAL PROCESSES UNDERLYING ENSO DIVERSITY One important question concerning ENSO diversity is whether EP‐ and CP‐type events are governed by similar or different dynamical processes. Heat budget analysis has been used to identify the leading dynamical feed­ backs responsible for the growth and decay of EP and CP events. The upper‐ocean heat budget is computed at each grid point as the balance between the heat storage term, the oceanic advective terms (Qadv), and the surface heat flux terms (QF) in a layer of depth H, usually chosen as the mean mixed layer, so that its temperature is approxi­ mately vertically homogeneous:



cp H

T t

Qadv QF

R, (4.1)

where ρ is the density of seawater, cp is the oceanic heat capacity, T is the upper‐ocean temperature, t is time, and R is a residual term that accounts for omitted processes (e.g. vertical and lateral mixing, solar pene­ tration, sub‐monthly advection). Qadv includes zonal, meridional, and vertical advection. For example, the vertical (Qz)  and zonal (Qx) advection terms can be written as 0

Qz

cp

w H

T dz and Qx z

0

cp

u H

T , dz x

where w and u are the vertical and zonal velocities, respec­ tively, and z and x are the vertical and zonal coordinates. These terms can be further divided into linear and non­ linear components by separating each variable into its time mean and anomaly. Previous studies have recog­ nized the leading role of two linear feedback terms:





Thermocline feedback =

Zonal advective feedback =

T (4.2) z

w

u

T (4.3) x

where the primes indicate anomalies and overbars denote climatological values. The thermocline feedback is the advection due to mean upwelling acting on the anomalous vertical temperature gradient, while the zonal advective feedback is the advection due to anomalous zonal cur­ rents acting on the mean zonal temperature gradient. The relative importance of these feedbacks varies along the equator, because the oceanic and atmospheric background mean states are zonally asymmetric. In the eastern Pacific where the thermocline is shallow and upwelling is strong, anomalous thermocline variations play a large role in the SST tendency and lead to a strong thermocline feedback. In the central Pacific a deep mean thermocline suppresses the thermocline feedback, but the strong mean zonal temperature gra­ dient at the warm pool edge, and strong zonal velocity anomalies induced by El Niño‐related westerly wind anomalies, make SSTA growth sensitive to the zonal advective feedback. The longitudinal distribution of wind speed anomalies can also contribute to the spatial structure of CP events. CP El Niño westerly wind anomalies over the Niño‐4 region reduce the local wind speed, which amplifies warm Niño‐4 SSTAs due to reduced evaporation and vertical mixing; but increased easterlies in the eastern Pacific (Figure  4.3d) act to damp SSTAs in the Niño‐3 region and contribute to the confinement of the SSTAs to the central Pacific (Kug et al., 2009).

ENSO Diversity  75

Thus, the growth and decay of events centered at ­ ifferent longitudes can be expected to be controlled by d different feedbacks. This is confirmed by a heat budget analysis of events peaking in different regions along the equator (Capotondi, 2013), which showed how the relative importance of the different feedbacks for the growth and decay of the events gradually varies along the equator, with the thermocline feedback dominating in the east and the zonal advective feedback becoming more important in the central Pacific. For this reason, EP events have been associated with the thermocline feedback and CP events with the zonal advective feedback. However, each El Niño event has a zonally broad struc­ ture that may extend beyond the eastern or central Pacific regions and be influenced by other processes. For in­ stance, the warm anomalies of EP events that extend into the central Pacific (as seen for example during 1997–1998 in Figure  4.1a) can also see large contributions from zonal advection processes. Similarly, CP events with SSTAs in the eastern Pacific can see contributions from the thermocline feedback in that region. The unique evolution of events with extreme eastern Pacific warming, like the 1877–1878, 1982–1983, and 1997–1998 events, distinguishes those events from moderate and CP El Niños (Figure 4.5). This difference has been explained in terms of the existence of an SST threshold (at about 27.5°C in the present‐day climate) above which atmospheric deep convection can occur (Graham & Barnett, 1987). This threshold introduces a nonlinearity in the Bjerknes feedback that is particu­ larly relevant in the eastern Pacific, where it would only be activated during extreme El Niños (Takahashi & Dewitte, 2016). Adding such nonlinearity to the damped recharge‐discharge oscillator model, while keeping the system in a stable regime, is sufficient to generate bimod­ ality associated with strong and moderate El Niños, although the action of high‐frequency stochastic forc­ ing blurs the mode separation in this model (Takahashi et al., 2019). 4.4. PRECURSORS AND PREDICTABILITY OF ENSO DIVERSITY ENSO is often described as a low‐frequency tropical mode of coupled ocean‐atmosphere variability energized by stochastic wind forcing (Penland & Sardeshmukh,1995). The “quasi‐oscillatory” nature of ENSO, which alter­ nates between warm and cold events approximately every two to seven years as measured by the equatorial SSTA, is connected with the evolution of the upper‐ocean warm water volume (WWV) which undergoes meridional dis­ placements toward and away from the equator, as described by the recharge oscillator paradigm for ENSO (Jin, 1997). The WWV, usually diagnosed as the volume

of water above the thermocline between 5°S and 5°N and across the Pacific basin, has been a very useful precursor for ENSO events, with a “recharged” equatorial state (larger WWV) usually preceding the peak SSTAs in the Niño‐3.4 region by about two to three seasons (Meinen & McPhaden, 2000). However, the relationship between WWV and ENSO SSTAs changed in the first decade of the 21st century, when WWV anomalies weakened and typically led the ENSO SSTAs only by one season (McPhaden, 2012). This changed relationship is likely associated with the dominance of CP El Niño events dur­ ing that period. Indeed, the difference in the thermocline depth anomalies for EP and CP events, as seen in Figure  4.3 (top) is indicative of a different anomalous zonal thermocline tilt, and hence different anomalous meridional geostrophic flow, implying different recharge/ discharge processes during the two event types. The large changes in the zonal slope of the thermocline during EP events lead to a rapid discharge of warm water from the equatorial thermocline, and a robust transition to a La Niña event immediately after the event peak. On the other hand, the smaller thermocline depth anomalies during CP events are associated with a much weaker dis­ charge, a longer duration of the positive SSTAs, and a reduced likelihood of transitioning into a La Niña (Kug et al., 2009, Kug et al., 2010, Capotondi, 2013). In addition to the oceanic thermocline processes, fast variations of the surface wind stress in the western and central equatorial Pacific also provide an important forc­ ing mechanism for El Niño events. These fast wind stress variations, commonly referred to as westerly wind bursts (WWBs), excite downwelling oceanic Kelvin waves which can propagate all the way to the eastern part of the basin (McPhaden, 1999), where they deepen the thermocline and potentially initiate an El Niño event. The WWBs are often, but not always (Chiodi et  al., 2014), associated with the Madden Julian Oscillation (MJO; Puy et  al., 2016), as well as with tropical cyclones (Tian et al., 2018). The WWBs are considered to be a state‐dependent sto­ chastic wind forcing of ENSO as their frequency and intensity increase with warmer SST conditions (Lengaigne et al., 2004; Yu et al., 2003; Gebbie et al., 2007; Kug et al., 2008; Capotondi et al., 2018, and references therein). The interplay between the ocean subsurface conditions and WWB activity can contribute to diversity in both amplitude and spatial pattern, as shown by recent mod­ eling studies (Hu et al., 2014; Fedorov et al., 2015; Jadhav et  al., 2015; Levine et  al., 2016; Puy et  al., 2019). In particular, based on coupled model experiments, a sub­ surface “recharged” state would evolve into a moderate CP El Niño in the absence of WWBs but may develop into a strong EP El Niño in the presence of WWBs (Vecchi et al., 2006b). Similarly, a “discharged” state that would develop into a La Niña without WWB activity

76  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

may result in a CP El Niño if WWBs are present. Seasonally, WWBs and related intraseasonal wind forc­ ing in boreal spring are particularly effective at triggering El Niño events (McPhaden et  al., 2006; Hendon et  al., 2007; Lopez & Kirtman, 2014); in addition, the presence of westerly wind stress anomalies above a threshold in the central Pacific starting in August during the El Niño onset indicates an increased likelihood that the event will become an extreme El Niño (Takahashi & Dewitte, 2016). Influences from regions outside the tropical Pacific have also been proposed as possible triggers of ENSO events and contributors to ENSO diversity. Within the Pacific basin, extratropical precursors include the Pacific meridional mode (PMM; Chiang & Vimont, 2004) in the Northern Hemisphere, and the south Pacific meridional mode (SPMM; Zhang et  al., 2014) in the Southern Hemisphere. The equatorial SSTAs associated with the PMM occur in the central Pacific, so that the PMM has been viewed as a precursor for CP El Niño events (Yu and Kim, 2011; Vimont et  al., 2014). The SPMM, on the other hand, is a mode of variability characterized by SSTAs in the southeastern tropical Pacific, and has been considered a possible precursor for EP events (Zhang et al., 2014; Vimont et al., 2014). Modulation of the trade winds by the southern lobe of the North Pacific Oscillation (Rogers, 1981; Linkin & Nigam, 2008), the second leading mode of wintertime SLP variability over the north Pacific, produces subtropical SSTAs that can propagate southwestward via a wind‐evaporation‐SST feedback (Xie, 1999) and promote the development of an ENSO event after they reach the equator (Chang et al., 2007). Wind stress curl anomalies associated with the PMM can also force an equatorward meridional transport which alters the equatorial heat content and favors the development of El Niño events, a mechanism known as “trade wind charging” (TWC; Anderson & Perez, 2015). Apart from the SST precursors, the subsurface initial state of the ocean appears to be a critical discriminating factor in the development of an EP or CP event (Capotondi & Sardeshmukh, 2015). Here we apply the same methodology of Capotondi & Sardeshmukh (2015) to time series of tropical SST and SSH anomalies from the ECMWF ORAS4 ocean reanalysis (Balmaseda et al., 2013), to determine the optimal precursors for EP and CP events at a 6‐month lead time. The latter are identified using the E and C indices of Takahashi et al. (2011). The E and C spatial patterns, computed as the regression of SSTAs on the E and C indices (Takahashi et al., 2011), are shown in Figures  4.6 a and b, respectively. The 20 leading EOFs of SST and 10 leading EOFs of SSH anomalies between 25°S and 25°N are used to charac­ terize the state of the equatorial ocean. The optimal SST and SSH initial conditions for the EP and CP events are

shown in Figures  4.6 c and d, respectively. Both initial conditions exhibit similar positive SST structures that are reminiscent of the PMM, SPMM, and northwest Pacific precursor (cold SSTAs in the northwestern tropical Pacific; Wang et al., 2012), although with different relative strengths in the two cases. However, the SSH fields show positive anomalies (deeper thermocline) in the eastern equatorial Pacific, extending westward to the dateline for EP events, but negative anomalies (shallower thermo­ cline) in the eastern Pacific in the case of CP events, in agreement with the results of Capotondi & Sardeshmukh (2015). The indices associated with the two optimal initial conditions, obtained by projecting the SST and SSH fields at each time step on the optimal patterns, are largely correlated with the E and C indices at a six‐month lead time (0.71 for EP, and 0.84 for CP, Figures 4.6 e and f). Given the critical role played by the zonal thermocline slope on event selection, it is conceivable that La Niña– like background conditions (similar to those present dur­ ing the first decade of the 21st century) may be more conducive to the development of CP events. Influences from other ocean basins may also contribute to ENSO diversity. North Tropical Atlantic (NTA) SST variations in boreal spring appear to favor the development of ENSO events by creating strong air‐sea interactions along the Pacific ITCZ (Ham & Kug, 2013a). In particular, NTA cooling is more conducive to the occurrence of CP El Niño events. In contrast to the NTA SSTAs, the Atlantic Niño in boreal summer is more related to the development of EP El Niño events (Ham et al., 2013b). These studies thus suggest that these two tropical Atlantic precursors may contribute to ENSO diversity to some extent, so that prediction of ENSO diversity in dynamical models may potentially be improved if the models simulate tropical Atlantic vari­ ability realistically. How predictable are EP and CP El Niño events? While the SST and SSH precursors in Figure 4.6 indicate some degree of predictability for EP and CP events, the ability to predict different event types using state‐of‐the‐art fore­ casting systems is still under investigation. For example, the ability of the Australian Bureau of Meteorology cou­ pled ocean‐atmosphere seasonal forecast model to pre­ dict differences in the SSTA patterns of EP and CP events is limited to less than one season lead time (Hendon et al., 2009). Similarly, Ren et al. (2019) showed that in six oper­ ational models the differences in SSTA, precipitation, and teleconnections associated with the two ENSO types could be detected only up to one‐month lead time, and only in two or three models. The North American Multimodel Ensemble (Kirtman et al., 2014), which pro­ duces ensemble forecasts from a suite of different climate models, shows some skill in capturing the SST and pre­ cipitation contrast between central and eastern Pacific

ENSO Diversity  77

warming. However, the models tend to systematically produce more warming in the east, so that strong EP events tend to be better predicted than CP events. Thus, model biases may be responsible for the limited skill in predicting ENSO diversity. Model studies have also shown that ENSO’s seasonal predictability, and seasonal sensitivity to transient external forcings, can depend on the initial flavor of ENSO. In studies with the GFDL‐CM2.1 model, Karamperidou et  al. (2014) found that active, EP‐dominated epochs tended to show greater seasonal predictability than qui­ eter, CP‐dominated epochs. In other studies with the CM2.1 model, Predybaylo et  al. (2017) showed that the ENSO evolution was most sensitive to tropical explosive volcanic eruptions at the onset of a CP event, with less sensitivity during neutral conditions or at EP event onset, and almost no sensitivity at La Niña onsets. 4.5. LOW-FREQUENCY VARIATIONS OF ENSO DIVERSITY AND CLIMATE CHANGE The increased intensity and frequency of CP El Niño events since the late 1990s relative to previous decades (Lee & McPhaden, 2010), has suggested the possibility that such changes in ENSO character could be due to global warming. In particular, the 2015–2016 warming of the Niño‐4 region, which was extreme by historical stan­ dards (L’Heureux et al., 2017), may have included a west Pacific warming trend attributable to anthropogenic forc­ ing (Knutson et al., 2014). It is unclear from the limited observational record, however, whether or not the ENSO SST variability relative to this long‐term warming trend has changed (Newman et al., 2018). Seasonally resolved coral records spanning the last four centuries indicate that the increased ratio of CP vs. EP events since the late 20th century seemed unusual in the context of that multi­ century record, suggesting possible anthropogenic influ­ ences on the dominance of CP events in recent decades (Freund et al., 2019). To help address this question, it is helpful to turn to model simulations. An examination of the Climate Model Intercomparison Project phase 3 (CMIP3) multimodel ensemble showed that the CP vs. EP ratio of occurrence increased in global warming scenario simulations relative to historical simulations (Yeh et al., 2009). In the CMIP5 models, however, results have been more nuanced. Chen et  al. (2017) found little consensus among the CMIP5 models regarding the relative likelihood of CP vs. EP events: comparing RCP8.5 projections against preindus­ trial simulations, roughly as many models showed increases in the likelihood ratio as decreases, and nearly all of those changes were statistically insignificant relative to the unforced variability in the preindustrial control runs. However, like Santoso et  al. (2013), Chen et  al.

(2017) found a robust and statistically significant shift under RCP8.5 forcing toward more eastward propaga­ tion of equatorial SSTAs (i.e. a seasonal evolution of individual events from a CP‐ toward an EP‐type SSTA pattern), especially in the more realistic models that had less of a bias toward excessive eastward SSTA propaga­ tion in their historical simulations. Kim and Yu (2012) found that the intensity of both CP and EP events strengthened from preindustrial simula­ tions to historical simulations, whereas for RCP4.5 projections the CP events continued to strengthen while the EP events weakened; thus, in the RCP4.5 scenario the CP events became progressively stronger relative to the EP events. This increase in CP intensity relative to EP was attributed to the future changes in the upper ocean thermal stratification in the scenario simulations. Climate models project a weakening of the Walker circulation with global warming (Vecchi et al., 2006a) resulting in a weakened eastern Pacific cold tongue and reduced zonal thermocline slope. These mean state changes are expected to reduce upwelling, thus weakening the thermocline feedback in the eastern Pacific, while the increased strati­ fication of the sloping thermocline in the central Pacific can enhance both the zonal advective feedback (DiNezio et al., 2012) and the thermocline feedback (Dewitte et al., 2013), resulting in a preferred central Pacific warming. On the other hand, Cai et al. (2018) found increased var­ iance in the E index and more frequent strong eastern Pacific El Niño events with climate change in models that represent the nonlinearity in the Bjerknes feedback. This change is associated with the increased stratification of the equatorial Pacific, which enhances the projection of the anomalous wind forcing onto the dominant oceanic baroclinic modes, hence increasing the ocean‐atmosphere coupling. In contrast with the mechanisms proposed in some of the above studies, the CP‐dominated 2000–2014 period was characterized by a steeper zonal thermocline slope (McPhaden et al., 2011). This is in agreement with results from single‐forcing ensemble simulations of the last mil­ lennium, which showed that the relative incidence of CP vs. EP events during the 20th century (1850–2005) com­ pared to the preindustrial period (850–1849) significantly increased in the presence of ozone/aerosol forcing, which is conducive to a stronger zonal tilt of the thermocline. The CP/EP frequency showed no significant change when only greenhouse gas forcing, which produces a reduced zonal thermocline slope, was prescribed (Stevenson et al. 2019). These results not only highlight the complexity of the climate change–ENSO relationship but also support the link between a zonally steeper equatorial thermocline and a higher frequency of CP events. Although most cli­ mate models project a weakening of the Walker circulation and a warming of the eastern Pacific cold tongue, other

78  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

studies based on observations (Compo & Sardeshmukh, 2010; Solomon & Newman, 2012; L’Heureux et al., 2013, Li et al., 2017) suggest an intensification of the equatorial easterly winds and a strengthening of the cold tongue over the 20th century. In particular, Li and colleagues (2017) related the increased frequency of CP El Niño events in recent decades to what they call the cold tongue mode (CTM), a cooling trend of the equatorial cold tongue that emerges as the second EOF of SSTAs based on the HadISST1 dataset over the period 1871–2010, which they attributed to global warming. The colder con­ ditions in the eastern equatorial Pacific associated with the CTM would cause a westward displacement of the ENSO‐related air‐sea interactions and a weakening of the Bjerknes feedback, resulting in a preferred occurrence of CP‐type events (Xiang et al., 2013). Whether the cold tongue is warming or cooling with climate change, and whether these changes in the tropical Pacific climate will influence (or are already influencing) ENSO diversity, remain open questions. The relative frequency of CP and EP events also appears to undergo natural decadal variations. Newman et  al. (2016), for example, show that CP events are more common during the negative phases of the Pacific Decadal Oscillation (PDO), which are (again) associated with stronger trade winds in the tropics and cooler conditions in the eastern equatorial Pacific. EP events, on the other hand, tend to preferentially occur during positive PDO phases. For example, the decade of prevailing CP events at the beginning of the 21st century examined by McPhaden et al. (2011) coincides with a negative phase of the PDO. Influences from other ocean basins could also contribute to the decadal modulation of ENSO diversity. For example, Yu et al. (2015) attributed the intensification of the PMM and the increased frequency of CP El Niño events in recent decades to a phase change of the Atlantic Multidecadal Oscillation (AMO) after the 1990s. The influence of the AMO occurred through a strengthening of the North Pacific Subtropical High and an intensifica­ tion of the background trade winds, which led to a stronger wind‐evaporation‐SST feedback and stronger atmosphere‐ ocean coupling in the subtropical north Pacific. The warm conditions in the tropical North Atlantic associated with the warm phase of the AMO in recent decades have also been invoked as a possible contributor to stronger eastern Pacific cross‐equatorial southerly winds, which could have induced a La Niña–like background state in the equatorial Pacific that favored CP events (Hu & Fedorov, 2018). Multicentury preindustrial climate model simulations also show a low‐frequency modulation of the CP/EP frequency ratio, with CP‐dominated epochs characterized by steeper zonal gradients of SST and thermocline depth relative to EP‐dominated decadal epochs (Kug et  al., 2010; Choi et al., 2011; Choi et al., 2012; Ogata et al., 2013; Atwood

et al., 2017). Perfect‐model reforecast experiments with a coupled GCM suggest that this intrinsic component of the low‐frequency modulation of ENSO diversity may be fun­ damentally unpredictable on decadal scales (Wittenberg et  al., 2014). Other studies report statistically significant differences in the linear dynamics of observed decadal epochs with different ENSO characteristics (Capotondi & Sardeshmukh, 2017). Such changes in linear dynamics could be consistent with either internal stochastic modula­ tions of stationary but nonlinear ENSO dynamics, and/or with a role for external forcings (anthropogenic or natural radiative forcings, or decadal interactions of the tropical Pacific with the extratropics or other tropical basins) in modulating ENSO via changes in the background climate of the tropical Pacific. These results open the question of whether ENSO’s observed past behavior has responded in a deterministic fashion to changes in the background con­ ditions, or whether decadal changes in ENSO characteris­ tics have arisen mostly by chance, as seen in unforced climate model simulations (Wittenberg, 2009; Kug et al., 2010; Choi et al., 2011, 2012; Wittenberg et al., 2014). Changes in background conditions could alter ENSO dynamical feedbacks, and favor either EP or CP events (Luebbecke & McPhaden, 2014). At the same time, changes in ENSO characteristics could rectify into the background state through nonlinearities and temporal blurring of fluctuating climatological features like the ITCZ, cold tongue, and thermocline (Watanabe et  al., 2012; Watanabe & Wittenberg 2012; Ogata et  al., 2013; Atwood et al., 2017). For example, McPhaden et al. (2011) noticed that changes in composite El Niño SST and ther­ mocline patterns during 2000–2010 (when CP events pre­ vailed) relative to 1980–1999 (when EP events prevailed) resembled the changes in mean SST and thermocline con­ ditions over the two periods, suggesting that the latter may be a rectification of the former. Thus, whether decadal modulation of ENSO diversity is a consequence or a cause of mean state changes remains at this point an important open question that needs to be addressed. 4.6. ENSO DIVERSITY REPRESENTATION IN CLIMATE MODELS Although the simulation of ENSO in climate models has significantly improved in recent decades, as shown by the large body of literature documenting the CMIP3 and CMIP5 multimodel archives, several aspects of ENSO are still not satisfactorily represented in climate models (Bellenger et al., 2014; see chapter 9 for more details). In particular, many models have difficulty in simulating El Niño events with sufficient diversity in spatial patterns along the equator (Ham & Kug, 2012). This model defi­ ciency is illustrated in Figure  4.7, which compares the composite equatorial profiles of EP and CP El Niño events

ENSO Diversity  79

in observations and in 20 models from the CMIP5 archive. The Niño‐3 and Niño‐4 indices were used to classify the events. While some models (NCAR‐CCSM4, CNRM‐ CM5, GFDL‐CM3, GFDL‐ESM2M) show distinct zonal maxima for the two groups of events, somewhat similar to the observations, other models (e.g. HadGEM2‐CC, HadGEM2‐ES, INM_CM4, MIROC‐ESM, MRI‐ CGCM3) display longitudinal evolutions for the two groups that are strongly overlapping. Chen et  al. (2017) found that most of the CMIP5 models tended to produce excessive numbers of CP events relative to EP events. This limitation in the representation of ENSO diversity likely arises from model biases in the background mean state (Guilyardi et  al., 2012a, 2012b; Capotondi et  al., 2015b; Guilyardi et al., 2016). In particular, the intensity of the equatorial cold tongue, which helps set the strength of the zonal and meridional SST gradients near the equator, is key for determining how readily atmospheric deep convection spreads into the equatorial eastern Pacific during El Niño. Since anomalous convective activity depends on the total SST relative to the tropical mean SST (He et al., 2018), convective responses to diverse anoma­ lous SST patterns rely on the mean state. If the mean state of the equatorial eastern Pacific is too cold, the eastern Pacific warming will not support local convection, and the atmospheric response will be confined to the west, result­ ing in a limited range of precipitation and SSTA patterns (Ham & Kug, 2012, Kug et  al., 2012). The westward extension of the cold tongue is also important, since it determines the position of the maximum zonal SST gra­ dient. If the cold tongue extends too far west, the ENSO SSTA pattern can take on too much of a “double‐peaked” longitudinal structure, in which SSTAs driven by zonal advection in the west are separated from SSTAs driven by vertical advection in the east (Graham et al., 2017). 4.7. CONCLUSIONS In this chapter we have provided a synthesis of current understanding of ENSO diversity. Our focus has pri­ marily been on El Niño events, since they exhibit a broader range of spatial structures relative to La Niña events (Kug & Ham, 2011). El Niño events vary in amplitude, spatial pattern, and temporal evolution in ways that make each event unique, so that “no two El Niño events are quite alike” (Wyrtki, 1975). Recent studies, using a variety of approaches and criteria, have often partitioned El Niño events into Central Pacific (CP) and Eastern Pacific (EP) types, in order to better represent their dynamics, origin, and evolution. These various approaches have helped identify salient features of the two El Niño groups, including spectral characteris­ tics, temporal evolution, and their outgoing longwave radiation signature as a proxy for event impacts. The

decadal variation of CP events, an aspect that has been highlighted in conjunction with some recently proposed indices, is an intriguing phenomenon that needs to be better understood. Whether EP and CP events are distinct entities or the extreme expressions of a continuum of ENSO flavors remains an open question. On one hand, empirical studies show the emergence of two “modes” resembling the EP and CP cases (Newman et al., 2011), and some degree of bimodality is found when considering extreme El Niño events (Takahashi et  al., 2011; Takahashi & Dewitte, 2016; Cai et  al., 2018). On the other hand, the broad range of spatial patterns of El Niño events does not seem to support bimodality of equatorial SSTA distributions. Also, both EP and CP events are controlled by the same underlying dynamical processes, although their relative importance is longitudinally dependent. El Niño–related SSTAs in the eastern Pacific are largely controlled by the thermocline feedback, since upwelling is enhanced and vertical temperature gradients are stronger in the eastern Pacific. SSTAs in the central Pacific are more influenced by the zonal advective feedback, due to the larger mean zonal temperature gradients and zonal current variations in that region. Decadal variations in ocean background conditions can perhaps influence the relative frequency of EP and CP events in different decadal epochs, via the spatial modulation of the leading dynamical feedbacks. Whatever the nature of ENSO diversity, continuous or bimodal, the ability to predict the longitudinal location of the largest equatorial SSTAs is very important for atmo­ spheric teleconnections. Can different event types be skill­ fully predicted? Different precursors, both within the tropical Pacific as well as from regions outside the tropical Pacific, have been suggested as more conducive to EP or CP events. In particular, the SST and wind anomalies asso­ ciated with the North Pacific Meridional Mode have been considered as possible triggers of CP El Niño events, while the South Pacific Meridional Mode has been viewed as more conducive to EP El Niño events. However, the initial/ background zonal thermocline slope appears to be an important discriminating factor for evolving a developing El Niño into either an EP or CP event. Subsurface ocean conditions with a zonally flatter thermocline are more con­ ducive to the development of EP El Niños, while a zonally steeper thermocline favors CP El Niños. This result emerges from an empirical calculation of the optimal EP and CP precursors and is consistent with results from both obser­ vations and long climate model simulations. The processes by which a zonally steeper thermocline favors CP events are still unclear. More importantly, it is still unknown whether these background changes are a cause or a consequence of ENSO diversity, an issue that is in urgent need of clarification. Going forward, continuation of exist­ ing long‐term observational records in the tropical Pacific

(0) MME

(–1) OBS 3 EP EI Nino CP EI Nino 2

3

1

1

0

0

–1

120E

(1) ACCESS1.0

150E

180

2

150W 120W

–1

90W 120E

(2) BCC–CSM1.1

150E

180

150W 120W

90W

(3) CanESM2

(4) CCSM4

3

3

3

3

2

2

2

2

1

1

1

1

0

0

0

0

–1

120E

150E

180

150W 120W

–1

90W 120E

(5) CNRM–CM5

150E

180

150W 120W

–1

90W 120E

150E

180

150W 120W

–1

90W 120E

(7) GFDL–CM3

(6) CSIRO–Mk3.6.0

3

3

3

3

2

2

2

2

1

1

1

1

0

0

0

0

–1

–1

–1

–1

120E

150E

180

150W 120W

90W 120E

(9) GFDL–ESM2M

150E

180

150W 120W

90W 120E

(10) GISS–E2–R

150E

180

150W 120W

90W 120E

(11) HadGEM2–CC

3

3

3

2

2

2

2

1

1

1

1

0

0

0

0

–1

120E

150E

180

150W 120W

–1

90W 120E

(13) INMCM4

150E

180

150W 120W

150E

180

150W 120W

90W 120E

(15) IPSL–CM5A–MR

3

3

3

2

2

2

2

1

1

1

1

0

0

0

0

–1

120E

150E

180

150W 120W

–1

90W 120E

(17) MIROC–ESM

150E

180

150W 120W

150E

180

150W 120W

90W 120E

(19) MRI–CGCM3

3

3

3

2

2

2

2

1

1

1

1

0

0

0

0

120E

–1 150E

180

150W 120W

90W 120E

–1 150E

180

150W 120W

180

150W 120W

90W

150E

180

150W 120W

90W

90W 120E

150E

180

150W 120W

150E

180

150W 120W

90W

(20) NorESM1–M

3

–1

150E

–1

90W 120E

(18) MPI–ESM–LR

90W

(16) MIROC5

3

–1

150W 120W

–1

90W 120E

(14) IPSL–CM5A–LR

180

(12) HadGEM2–ES

3

–1

150E

(8) GFDL–ESM2G

–1 90W 120E

150E

180

150W 120W

Figure 4.7  Composites of equatorial SSTA profiles averaged in 5°S–5°N for EP (red line) and CP (blue line) events for observations (ERSSTv5; Huang et al., 2017; panel –1) during 1951–2017, the multimodel ensemble mean (panel 0) and 20 models from the CMIP5 archive (panels 1–20). EP and CP events are identified using the normalized Niño‐3 and Niño‐4 indices, respectively. EP events are characterized by a value of the Niño‐3 index greater than one standard deviation, and greater than the value of the Niño‐4 index, and vice versa for the CP events. Equatorial profiles are shown as a function of longitude. Vertical axis units are °C (adapted from Capotondi et al., 2015b).

90W

ENSO Diversity  81

will be needed to understand the full diversity of ENSO events in the real world, together with observational enhancements to underpin future improvements in model simulations (Kessler et al., 2019). How will ENSO diversity change in a changing cli­ mate? The answer to this question depends on the tropical Pacific mean state response to climate change, and on the relationship of ENSO diversity with the underlying background conditions. Whether the equatorial Pacific cold tongue weakens or strengthens with climate change, it is expected to have a large impact on the nature of ENSO and its diverse expressions. Such changes might not be reliably detectable for several decades, however, given the strong intrinsic modulation of ENSO diversity suggested by historical observations and model simula­ tions. Many models still have difficulty simulating ENSO diversity, due in part to the severity of their cold tongue bias. Thus, improved simulation of ENSO diversity by the majority of the next generation of climate models, as well as reliable projections of tropical Pacific mean state climate by those models, are needed to understand how ENSO flavors will change in the future. APPENDIX: INDICES OF EL NIÑO DIVERSITY El Niño Modoki index (EMI; Ashok et al., 2007). This index is calculated as EMI = SSTC – 0.5 * (SSTE + SSTW), where SSTC is the average SSTA over the central equatorial Pacific (10°S–10°N, 165°E‐140°W), and SSTE and SSTW are SSTA averaged over an eastern (15°S–5°N, 110°– 70°W) and a western (10°S–20°N, 125°–145°E) region, respectively. This index was constructed to capture the “Modoki” SSTA pattern, characterized by positive values in the central equatorial Pacific and negative anomalies on the eastern and western sides of the positive anom­ alies. The original definition by Ashok et al. (2007) also includes the criterion that the anomalous warming in the central Pacific must persist from boreal summer through winter, that is, for three seasons. Niño‐3–Niño‐4 approach (Kug et  al., 2009; Yeh et  al., 2009). This method uses the Niño‐3 (average SSTA in 5°S–5°N, 90°–150°W) and Niño‐4 (average SSTA in 5°S– 5°N, 160°E–150°W) indices to identify cold tongue and warm pool events. Cold tongue events are identified by the criterion that the boreal winter Niño‐3 index is larger than one standard deviation (or larger than 0.5°C, depending on the application) and larger than Niño‐4, while warm pool events are characterized by the boreal winter Niño‐4 index exceeding one standard deviation (or 0.5°C) and exceeding the Niño‐3 index. EP–CP index method (Kao & Yu, 2009; Yu et al., 2012). CP events are defined as the leading empirical orthogonal function (EOF) and associated principal component (PC) of the SSTA after the regression of the SSTA onto

the Niño‐1+2 index, which is associated with eastern Pacific warming, is removed from the total SSTA field. EP events are obtained as the leading EOF/PC of the SSTA after the regression of the SSTA onto the Niño‐4 index, associated with central Pacific warming, is removed from the total SSTA field. EP–CP subsurface index method (Yu et al., 2011). The EP and CP indices are obtained by averaging the upper 100 m ocean temperature anomalies over the eastern (80°W–90°W, 5°S–5°N) and central (160°E–150°W, 5°S– 5°N) equatorial Pacific, respectively, exploiting the fact that CP events have their largest subsurface anomalies in the central Pacific, where EP El Niño events have only weak subsurface anomalies. NCT–NWP indices (Ren & Jin, 2011). The NCT and NWP indices are obtained as a linear combination of the Niño‐3 and Niño‐4 indices as: NCT = Niño‐3  –  αNiño‐4, NWP = Niño‐4 – αNiño‐3, with α = 0.4, when Niño‐3 • Niño‐4 > 0, and zero otherwise. This approach was motivated by the need of indices for CP and EP El Niño types that were uncorrelated, unlike the Niño‐3 and Niño‐4 indices. EPnew–CPnew indices (Sullivan et al., 2016). These indices are similar to the NWP and NCT indices, respectively, but the Niño‐3 and Niño‐4 indices are normalized by their standard deviation, and α = 0.5. TNI index (Trenberth & Stepaniak, 2001). This index is a measure of the SSTA difference between the Niño‐1+2 and the Niño‐4 regions: TNI = N1+2 – N4, where N1+2 and N4 are the SSTAs averaged over the Niño‐1+2 and Niño‐4 regions, each normalized by its standard deviation. E and C indices (Takahashi et al., 2011). The definition of these indices is based on the two leading principal components (PC1 and PC2, respectively) of SSTAs in the 10°S–10°N tropical Pacific band. The C and E indices are defined as C = (PC1 + PC2)/√2, and E = (PC1 – PC2)/√2, where PC1 > 0 corresponds to positive SSTAs in the east­ ern equatorial Pacific, and PC2 > 0 corresponds to positive anomalies in the central Pacific and negative anomalies in the far eastern equatorial Pacific. They are independent by construction and identify moderately warm events, primarily in the central equatorial Pacific, and extreme events in the eastern Pacific, respectively. Sea Surface Salinity (SSS) indices (Singh et al., 2011; Qu & Yu, 2014). The spatial patterns of SSS during EP and CP events, are used to characterize the different El Niño and La Niña types. EP events are characterized by a larger eastward displacement of the eastern edge of the west Pacific fresh pool and of precipitation than are CP events, resulting in different SSS signatures in the two cases. Singh et al. (2011) has used agglomerative hierar­ chical clustering to determine salinity patterns associated with EP/CP El Niño and EP/CP La Niña events, while Qu and Yu (2014) have shown that SSS variations over a

82  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

southeastern Pacific region (0°–10°S, 150°–90°W) are well correlated with the El Niño Modoki index. Indices of spatial shifts in atmospheric convection (Chiodi & Harrison, 2010; Johnson & Kosaka, 2016; Williams & Patricola, 2018). These approaches aim at identifying El Niño events characterized by deep convection in the east­ ern equatorial Pacific and associated with robust climate impacts over the U.S. Chiodi and Harrison (2010) defined a top‐of‐atmosphere outgoing longwave radiation (OLR) El Niño index based on the average OLR anomalies over the central Pacific (5°S–5°N, 170°E–100°W). Johnson and Kosaka (2016) identified EPC and EPN (east Pacific con­ vective and nonconvective, respectively) El Niño events based on the value of the relative SST (RSST), defined as the local minus the tropically (20°S–20°N) averaged SST, in an eastern equatorial Pacific box (5°S–5°N, 150°– 90°W). RSST values exceeding 0.7 identify EPC events, and RSST values less than 0.7 identify EPN events. Williams and Patricola (2018) define the ENSO Longitude Index (ELI), which identifies the average longitude, within the 5°S–5°N tropical Pacific band, of the points where the local SST is above a convective threshold defined as the SST averaged over the global tropics. Spatiotemporal indices (Lee et al., 2014). This approach examines interevent variations in both longitude and time, to characterize the diversity among ENSO events of amplitude, spatial pattern, growth, propagation, persis­ tence, decay and transition, and seasonal timing. The first step is to identify a set of events: e.g. El Niño events for which the three‐month running mean SSTA averaged over the Niño‐3.4 region (5°S–5°N, 120°–170°W) exceeds 0.5 K for at least five consecutive months. For each such event, a longitude‐time Hovmöller map of equatorial Pacific SSTAs (averaged 5°S–5°N) is constructed, span­ ning 120°E–80°W and extending for two years (from January of the onset year to December of the decay year). A PC analysis is then performed on the set of event Hovmöllers. The resulting EOF patterns (which are themselves Hovmöllers) then express the main directions of interevent diversity in spatiotemporal evolution, and the associated PCs express the amount of each EOF pre­ sent in each particular event. ACKNOWLEDGMENTS The authors would like to thank Dr. Y.‐G. Ham for his help in preparing one of the figures, and two anonymous reviewers for their excellent and constructive comments. Suggestions received from the participants of an ENSO meeting held in Hobart, Tasmania, 1–2 February 2019 are also gratefully acknowledged. AC was supported by the NASA Physical Oceanography Program (Award NNX15AG46G). This is PMEL contribution no. 4905.

REFERENCES Anderson, B. T. & R. Perez (2015), ENSO and non‐ENSO induced charging and discharging of the equatorial Pacific. Climate Dynamics, 45, 2309‐2327. Ashok, K., S. K. Behera, S. A. Rao, H. Weng, & T. Yamagata (2007). El Niño Modoki and its possible teleconnections. J. Geophys. Res., 112, C11007. doi:10.1029/2006JC003798 Ashok, K., T. P. Sabin, P. Swapna, & R. G. Murtugudde (2012). Is a global warming signature emerging in the tropical Pacific? Geophys. Res. Lett., 39, L02701. doi:10.1029/ 2011GL050232. Ashok, K., M. Shamal, A. K. Sahai, & P. Swapna (2017). Nonlinearities in the evolutional distinction between El Niño and La Niña types. J. Geophys. Res. (Oceans), 122, 9649–9662. Atwood, A. R., D. S. Battisti, A. T. Wittenberg, W. G. H. Roberts, & D. J. Vimont (2017). Characterizing unforced multi‐decadal variability of ENSO: A case study with the GFDL CM2.1 coupled GCM. Climate Dyn., 49(7–8), 2845– 2862. doi: 10.1007/s00382‐016‐3477‐9 Balmaseda, M. A., K. Mogensen, & A. T. Weaver (2013). Evaluation of the ECMWF ocean reanalysis system ORAS4. Q. J. R. Meteorol. Soc., 139, 1132–1161. Behera, S., & T. Yamagata (2010). Imprint of the El Niño Modoki on decadal sea level changes. Geophys. Res. Lett., 37, L23702. doi:10.1029/2010GL045936 Bellenger, H., E. Guilyardi, J. Leloup, M. Lengaigne, & J.  Vialard (2014). ENSO representation in climate models: from CMIP3 to CMIP5, Clim. Dyn., 42, 1999–2018. Brown, J. N., & A.V. Fedorov (2010). How much energy is trans­ ferred from the winds to the thermocline on ENSO time­ scales? J. Climate, 23, 1563–1580. Cai., W., G. Wang, B. Dewitte, L. Wu, A. Santoso, et al. (2018). Increased variability of eastern Pacific El Niño under greenhouse warming. Nature. doi:10.1038/s41586‐018‐0776‐9. Capotondi, A. (2013). ENSO diversity in the NCAR CCSM4. J. Geophys. Res., 118, 4755–4770. Capotondi, A., & P. D. Sardeshmukh (2015). Optimal precur­ sors of different types of ENSO events. Geophys. Res. Lett., 42, 9952–9960. doi:10.1002/2015GL066171. Capotondi, A., & P. D. Sardeshmukh (2017). Is El Niño really changing?, Geophys. Res. Lett., 44. doi:10.1002/2017GL074515. Capotondi, A., et al. (2015a). Understanding ENSO diversity. Bull. Amer. Meteor. Soc., 96, 921–938. doi:10.1175/ BAMS‐D‐13‐00117.1 Capotondi, A., Y.‐G. Ham, A. T. Wittenberg, & J.‐S. Kug (2015b). Climate model biases and El Niño Southern Oscillation (ENSO) simulation. US CLIVAR Variations, 13(1), 21–25. Capotondi, A., P.D. Sardeshmukh, & L. Ricciardulli (2018). The nature of the stochastic wind forcing of ENSO. J. Climate, 31, 8081–8099. Chang, P., L. Zhang, R. Saravanan, D. J. Vimont, J. C. H. Chiang, L. Ji, H. Seidal, & M. K. Tippett (2007). Pacific Meridional mode and El Niño‐Southern Oscillation. Geophys. Res. Lett., 34, L16608. doi:10.1029/2007GL030302 Chen, C., M. A. Cane, A. T. Wittenberg, & D. Chen (2017). ENSO in the CMIP5 simulations: Life cycles, diversity, and

ENSO Diversity  83 responses to climate change. J. Climate, 30(2), 775–801. doi: 10.1175/JCLI‐D‐15‐0901.1 Chen, M., P. Xie, J. E. Janowiak, & P. A. Arkin (2002). Global land precipitation: A 50‐year analysis based on gauge obser­ vations. J. Hydrometeorology, 3, 249–266. Chiang, J. C. H., & D. J. Vimont (2004). Analogous Pacific and Atlantic meridional modes of tropical atmosphere ocean var­ iability. J. Climate, 17, 4143–4158. Chiodi, A. M., & D. E. Harrison (2010). Characterizing warm‐ ENSO variability in the equatorial Pacific: An OLR perspec­ tive. J. Climate, 23, 2428–2439. Chiodi, A. M., D. E. Harrison, & G. A. Vecchi (2014). Subseasonal atmospheric variability and El Niño waveguide warming: Observed effects of the Madden–Julian oscillation and westerly wind events. Journal of Climate, 27(10), 3619–3642. Choi, J., S.‐I. An, J.‐S. Kug, & S.‐W. Yeh (2011). The role of mean state on changes in El Niño flavor, Clim. Dyn., 37, 1205–1215. Choi, J., S.‐I. An, & S.‐W. Yeh (2012). Decadal amplitude mod­ ulation of two types of ENSO and its relationship with the mean state, Clim. Dyn., 28, 2631–2644. Compo, G., & P. D. Sardeshmukh (2010). Removing ENSO‐ related variations from the climate record, J. Climate, 23, 1957–1978. Dewitte, B., & K. Takahashi (2019). Diversity of moderate El Niño events evolution: Role of air‐sea interactions in the eastern tropical Pacific. Clim. Dyn., 52, 7455. doi: 10.1007/ s00382‐017‐4051‐9 Dewitte, B., S.‐W. Yeh, & S. Thual (2013). Reinterpreting the thermocline feedback in the western‐central equatorial Pacific and its relationship with the ENSO modulation. Clim. Dyn., 41, 819–830. Diamond, M. S., & R. Bennartz (2015). Occurrence and trends of eastern and central Pacific El Niño in different recon­ structed SST data sets. Geophys. Res. Lett., 42, 10,375–10,381. doi: 10.1002/2015GL066469. DiNezio, P. N., B. P. Kirtman, A. C. Clement, S.‐K. Lee, G. A. Vecchi, & A. Wittenberg (2012). Mean climate controls on the simulated response of ENSO to increasing greenhouse gases. J. Climate, 25, 7399–7420. doi: 10.1175/JCLI‐D‐ 11‐00494.1 Fedorov, A. V., S. Hu, M. Lengaigne, & E. Guilyardi (2015). The impact of westerly wind bursts and ocean initial state on the development, and diversity of El Niño events. Clim. Dyn., 44, 1381–1401. Fraser, B. (2017). Peru’s floods teach tough lessons, Nature, 544, 405–406. Freund, M. B., B. J. Henley, D. J. Karoly, HMcGregor, N.  J.  Abram, & D. Dommenget (2019). Nature Geoscience, 12, 450–455. Garreaud, R. D. (2018). A plausible atmospheric trigger for the 2017 coastal El Niño, International Journal of Climatology, 38, e1296–31302. doi: 10.1002/joc.5426 Gebbie, G., I. Eisenman, A. Wittenberg, & E. Tziperman (2007). Modulation of westerly wind bursts by sea surface tempera­ ture: A semistochastic feedback for ENSO. J. Atmos. Sci., 64, 3281–3295. doi: 10.1175/JAS4029.1

Giese, B. S., & S. Ray (2011). El Niño variability in simple ocean data assimilation (SODA) 1871–2008. J. Geophys. Res., 116, C02024. doi: 10.1029/2010JC006695 Goddard, L., & S. G. Philander (2000). The energetics of El Niño and La Niña. J. Climate, 13, 1496–1516. Graham, F. S., A. T. Wittenberg, J. N. Brown, S. J. Marsland, & N. J. Holbrook (2017). Understanding the double peaked El Niño in coupled GCMs, Climate Dyn., 48(5), 2045–2063. doi: 10.1007/s00382‐016‐3189‐1 Graham, N. E., & Barnett, T. P. (1987). Sea surface tempera­ ture, surface wind divergence, and convection over tropical oceans. Science, 238(4827), 657–659. Guilyardi, E., et al. (2012a). A first look at ENSO in CMIP5. CLIVAR Exchanges, 17, 29–32. Guilyardi, E., W. Cai, M. Collins, A. Fedorov, F.‐F. Jin, A. Kumar, D.‐Z. Sun, & A. Wittenberg (2012b). New strat­ egies for evaluating ENSO processes in climate models. Bull. Amer. Met. Soc., 93, 235–238. doi: 10.1175/BAM S‐D‐11‐00106.1 Guilyardi, E., A. Wittenberg, M. Balmaseda, W. Cai, M. Collins, M. J. McPhaden, M. Watanabe, & S.‐W. Yeh (2016). Fourth CLIVAR workshop on the evaluation of ENSO processes in climate models: ENSO in a changing climate. Bull. Amer. Meteor. Soc., 97(5), 817–820. doi: 10.1175/ BAMS‐D‐15‐00287.1 Ham, Y.‐G., & J.‐S. Kug (2012). How well do current climate models simulate two types of El Niño? Clim. Dyn., 39, 383–398. Ham, Y.‐G., J.‐S. Kug, & F.‐F. Jin (2013a). Sea surface temper­ ature in the north tropical Atlantic as a trigger for El Niño/ Southern Oscillation events. Nature Geoscience, 6, 112. Ham, Y.‐G., J.‐S. Kug, D. Kim, Y. H. Kim, & D.H. Kim (2013b). What controls phase‐locking of ENSO to boreal winter in coupled GCMs? Clim. Dyn., 40, 1551–1568. Harrison, D. E., & A. M. Chiodi (2009). Pre and post 1997/1998 westerly wind events and equatorial Pacific cold tongue warming. J. Climate, 22, 568–581. He, J., N. C. Johnson, G. A. Vecchi, B. Kirtman, A. T. Wittenberg, & S. Sturm (2018). Precipitation sensitivity to local variations in tropical sea surface temperature. J. Climate, 31(22), 9225– 9238. doi: 10.1175/JCLI‐D‐18‐0262.1 Hendon, H. H., M. C. Wheeler, & C. Zhang (2007). Seasonal dependence of the MJO‐ENSO relationship. Journal of Climate, 20(3), 531–543. Hendon, H. H., E. Lim, G. Wang, O. Alves, & D. Hudson (2009). Prospects for predicting two flavors of El Niño. Geophys. Res. Lett., 36, L19713. doi: 10.1029/2009G L040100. Hu, S., & A. V. Fedorov (2018). Cross‐equatorial winds control El Niño diversity and change. Nature Climate Change, 8, 798–802. Hu, S., A. V. Fedorov, M. Lengaigne, & E. Guilyardi (2014). The impact of westerly wind bursts on the diversity and pre­ dictability of El Niño events: An ocean energetics perspec­ tive. Geophys. Res. Lett., 41, 4654–4663. Hu, Z.‐Z., B. Huang, J. Zhu, A. Kumar, & M. McPhaden (2019). On the variety of coastal El Niño events. Clim. Dyn., 52, 7537–7552. doi: 10.1007/s00382‐018‐4290‐4

84  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Huang, B., P. W. Thorbe, V. F. Banzon, T. Boyer, G. Chepurin, et  al. (2017). Extended Reconstructed Sea Surface Temperature, version 5 (ERSSTv5): Upgrades, validations, and intercomparisons, J. Climate, 30, 8179–8205. Jadhav, J., S. Panickal, S. Marathe, & K. Ashok (2015). On the possible cause of distinct El Niño types in the recent decades. Sci. Reports, 5, 17009. doi: 10.1038/srep17009 Jin, F.‐F. (1997). An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model. J. Atmos. Sci., 54, 811–829. Johnson, N. C., & Y. Kosaka (2016). The impact of eastern equatorial Pacific convection on the diversity of boreal winter El Niño teleconnection patterns. Clim. Dyn., 47, 3737–3765. Kalney, E., et al. (1996). The NCEP/NCAR 40‐year reanalysis project. Bull. Amer. Meteor. Soc., 77, 437–471. Kao, H.‐Y. & J.‐Y. Yu (2009). Contrasting eastern‐Pacific and eentral‐Pacific types of ENSO. Journal of Climate, 22, 615– 632. doi: 10.1175/2008JCLI2309.1 Karamperidou, C., M. A. Cane, U. Lall, & A. T. Wittenberg (2014). Intrinsic modulation of ENSO predictability viewed through a local Lyapunov lens. Climate Dyn., 42, 253–270. doi: 10.1007/s00382‐013‐1759‐z Kessler, W. S., et  al. (2019). Second report of TPOS 2020. GOOS‐234, 265 pp. Available online at http://tpos2020.org/ project‐reports/second‐report and https://doi.org/10.13140/ RG.2.2.17161.29289 Kim, S. T., & J.‐Y. Yu (2012). The two types of ENSO in CMIP5 models. Geophys. Res. Lett., 39, L11704. doi: 10.1029/2012GL052006. Kirtman, B. P., et al. (2014). The North American multimodel ensemble. Bull. Amer. Meteor. Soc., 95, 585–601. doi: 10.1175/ BAMS‐D‐12‐00050.1 Knutson, T. R., F. Zeng, & A. T. Wittenberg (2014). Multimodel assessment of extreme annual‐mean warm anomalies during 2013 over regions of Australia and the western tropical Pacific. Section 8 of “Explaining extreme events of 2013 from a climate perspective.” Bull. Amer. Meteor. Soc., 95 (9), S26– S30. doi: 10.1175/1520‐0477‐95.9.S1.1 Kug, J.‐S., & Y.‐G. Ham (2011). Are there two types of La Niña? Geophys. Res. Lett., 38. doi:10.1029/2011GL048237. Kug, J.‐S., F.‐F. Jin, K.P. Sooraj, & I.‐S. Kang (2008). State‐ dependent atmospheric noise associated with ENSO. Geophys. Res. Lett., 35, L05701. https://doi.org/10.1029/ 2007GL032017 Kug, J.‐S., F.‐F. Jin, & S.‐I. An (2009). Two types of El Niño: Cold tongue El Niño and warm pool El Niño. J. Climate, 22, 1499–1515. Kug, J.‐S., et al. (2010). Warm pool and cold tongue El Niño events as simulated by the GFDL CM2.1 coupled GCM. J. Climate, 23, 1226–1239. doi: 10.1175/2009JCLI3293.1 Kug, J.‐S., Y.‐G. Ham, J.‐Y. Lee, & F.‐F. Jin (2012). Improved simulation of two types of El Niño in CMIP5 models. Environ. Res. Lett., 7. doi: 10.1088/1748‐9326/3/034002 Larkin, N. K., & D. E. Harrison (2005). On the definition of El Niño and associated seasonal average U.S. weather anom­ alies. Geophys. Res. Lett., 32 L13705. doi: 10.1029/2005 GL022738

Lee, S.‐K., P. Di Nezio, E.‐S. Chung, S.‐W. Yeh, A. T. Wittenberg, & C. Wang (2014). Spring persistence, transition, and resur­ gence of El Niño. Geophys. Res. Lett., 41, 8578–8585. doi:10.1002/2014GL062484 Lee, S.‐K., A. T. Wittenberg, D. B. Enfield, S. J. Weaver, C. Wang, & R. M. Atlas (2016). U.S. regional tornado out­ breaks and their links to spring ENSO phases and North Atlantic SST variability. Environ. Res. Lett., 11(4), 044008. doi: 10.1088/1748‐9326/11/4/044008 Lee, S.‐K., H. Lopez, E.‐S. Chung, P. DiNezio, S.‐W. Yeh, & A. T. Wittenberg (2018). On the fragile relationship between El Niño and California rainfall. Geophys. Res. Lett., 45(2), 907– 915. doi: 10.1002/2017GL076197 Lee, T., & M. J. McPhaden (2010). Increasing intensity of El Niño in the central‐equatorial Pacific. Geophys. Res. Lett., 37, L14603. doi: 10.1029/2010GL044007 Lengaigne, M., E. Guilyardi, J. P. Boulanger, C. Menkes, P. Delecluse, P. Inness, J. Cole, & J. Slingo (2004). Triggering of El Niño by westerly wind events in a coupled general circulation model. Clim. Dyn., 23, 601–620. L’Heureux, M., S. Lee, & B. Lyon (2013). Recent multidecadal strengthening of the Walker circulation across the tropical Pacific. Nat. Climate Change, 3, 571–576. L’Heureux, M. L., K. Takahashi, A. B. Watkins, et al. (2017) Observing and predicting the 2015/16 El Niño. Bull. Amer. Meteor. Soc., 98(7), 1363–1382. doi: 10.1175/BAMS‐D‐ 16‐0009.1 Levine, A. F. Z., F.‐F. Jin, & M. J. McPhaden (2016). Extreme noise–extreme El Niño: How state‐dependent noise forcing creates El Niño‐La Niña asymmetry. J. Climate, 29, 5483– 5499. doi: http://dx.doi.org/10.1175/JCLI‐D‐16‐0091.1. Li, Y., J. Li, W. Zhang, Q. Chen, J. Feng, F. Zheng, et al. (2017). Impacts of the tropical Pacific cold tongue mode on ENSO diversity under global warming. J. Geophys. Res. (Oceans), 122, 8524–8542. Linkin, M. E., & S. Nigam (2008). The North Pacific oscilla­ tion–West Pacific teleconnection pattern: Mature‐phase structure and winter impacts. J. Climate, 21, 1979–1997. Lopez, H., & Kirtman, B. P. (2014). WWBs, ENSO predict­ ability, the spring barrier and extreme events. J. Geophys. Res., 119, 10,114–10,138. doi: 10.1002/2014JD021908 Luebbecke, J. F., & M. J. McPhaden (2014). Assessing the 21st century shift in ENSO variability in terms of the Bjerknes stability index, J. Climate, 27, 2577–2587. http://dx.doi. org/10.1175/JCLI‐D‐13‐00438.1 Marathe, S., K. Ashok, P. Swapna, & T. P. Sabin (2015). Revisiting El Niño Modokis. Clim. Dyn., 45, 3527–3545. McPhaden, M. J. (1999). Genesis and evolution of the 1997‐98 El Niño. Science, 283, 950–954. McPhaden, M. J. (2004). Evolution of the 2002‐03 El Niño. Bull. Amer. Meteor. Soc., 85, 677–695. doi: 10.1175/BAMS‐85‐5‐677 McPhaden, M. J. (2012). A 21st century shift in the relationship between ENSO SST and warm water volume anomalies. Geophys. Res. Lett., 39, L09706. doi: 10.1029/2012GL051826 McPhaden, M. J., & X. Zhang (2009). Asymmetry in zonal phase propagation of ENSO sea surface temperature anom­ alies. Geophys. Res. Lett., 36, L13703. doi: 10.1029/ 2009GL038774

ENSO Diversity  85 McPhaden, M. J., X. Zhang, H. H. Hendon, & M. C. Wheeler (2006). Large scale dynamics and MJO forcing of ENSO var­ iability. Geophysical Research Letters, 33(16), L16702. doi: 10.1029/2006GL026786 McPhaden, M. J., T. Lee, & D. McClurg (2011). El Niño and its relationship to changing background conditions in the tropical Pacific. Geophys. Res. Lett., 38, L15709. doi: 10.1029/2011GL048275 Meinen, C. S., & M. J. McPhaden (2000). Observations of warm water volume changes in the equatorial Pacific and their rela­ tionship to El Niño and La Niña. Journal of Climate, 13, 3551–3559. Newman, M., S.‐I. Shin, & M. A. Alexander (2011). Natural variations in ENSO flavors. Geophys. Res. Lett., 38, L14705. https://doi.org/10.1029/2011GL047658 Newman, M., et al. (2016). The Pacific decadal oscillation, revis­ ited, J. Climate, 29, 4399–4427. doi: 10.1175/JCLI‐D‐15‐0508.1 Newman, M., A. T. Wittenberg, L. Cheng, G. P. Compo, & C. A. Smith (2018). The extreme 2015/16 El Niño, in the context of historical climate variability and change, Section  4 of “Explaining extreme events of 2016 from a climate perspec­ tive.” Bull. Amer. Meteor. Soc., 99(1), S16–S20. doi: 10.1175/ BAMS‐D‐17‐0116.1 Ogata, T., S.‐P. Xie, A. Wittenberg, & D.‐Z. Sun (2013). Interdecadal amplitude modulation of El Niño/Southern Oscillation and its impacts on tropical Pacific decadal variability. J. Climate, 26, 7280–7297. doi: 10.1175/JCLI‐D‐12‐00415.1 Paek, H., J.‐Y. Yu, & C. Qian (2017). Why were the 2015/2016 and 1997/1998 extreme El Niño different? Geophys. Res. Lett., 44, 1848–1856. Peng, Q., Xie, S. P., Wang, D., Zheng, X. T., & Zhang, H. (2019), Coupled ocean‐atmosphere dynamics of the 2017 extreme coastal El Niño. Nature Communications, 10(1), 1–10. doi: 10.1038/s41467‐018‐08258‐8 Penland, C., & P. D. Sardeshmukh (1995). The optimal growth of tropical sea surface temperature anomalies. J. Climate, 8, 1999–2024. Predybaylo, E., G. Stenchikov, A. T. Wittenberg, & F. Zeng (2017). Impacts of a Pinatubo‐scale volcanic eruption on ENSO. J. Geophys. Res. Atmos., 122, 925–947. doi: 10.1002/2016JD025796 Puy, M., J. Vialard, M. Lengaigne, & E. Guilyardi (2016). Modulation of equatorial Pacific westerly/easterly wind events by the Madden‐Julian oscillation and convectively‐ coupled Rossby waves. Clim. Dyn., 46(7), 2155–2178. doi:10.1077/s00382‐015‐2695‐x Puy, M., J. Vialard, M. Lengaigne, E. Guilyardi, P. Di Nezio, A.  Voldoire, et  al. (2019). Influence of westerly wind events stochasticity on El Niño amplitude: The case of 2014 vs. 2015. Clim. Dyn., 52, 7435–7454. doi: 10.1007/s00382‐017‐3938‐9 Qu, T., & J.‐Y. Yu (2014). ENSO indices from sea surface salinity observed by Aquarius and Argo. Journal of Oceanography, 70, 367–375. Rasmusson, E. M., & T. H. Carpenter (1982). Variations in tropical sea surface temperature and surface wind fields asso­ ciated with the Southern Oscillation/El Niño. Mon. Wea. Rev., 110, 354–384. Ren, H.‐L., & F.‐F. Jin (2011). Niño indices for two types of ENSO, Geophys. Res. Lett., 38. doi: 10.1029/2010GL046031.

Ren, H. L., A. A. Scaife, N. Dunstone, B. Tian, Y. Liu, S. Ineson, et al. (2019). Seasonal predictability of winter ENSO types in operational dynamical model predictions. Climate Dynamics, 52(7‐8), 3869–3890. doi: 10.1007/s00382‐018‐4366‐1 Reynolds, R. W., N. A. Rayner, T. M. Smith, D. C. Stokes, & W. Wang (2002). An improved in situ and satellite SST analysis for climate, J. Climate, 15, 1609–1625. Rogers, J. C. (1981). The North Pacific Oscillation, Int. J. Climatol., 1, 39–57. Santoso, A., S. McGregor, F.‐F. Jin, W. Cai, M. H. England, S.‐I. An, et  al. (2013). Late 20th century emergence of El Niño propagation asymmetry and future projections. Nature, 504, 126–130. Santoso, A., M. McPhaden, & W. Cai (2017). The defining characteristics of ENSO extremes and the strong 2015/2016 El Niño. Rev. Geophys., 55. doi: 10.1002/2017RG000560 Schott, G. (1931). Der Peru‐Strom und seine nördlichen Nachbargebiete in normaler und anormaler Ausbildung. Ann Hydrogr Mar Meteor, 59, 161–169, 200–213, 240–257. Singh, A., T. Delcroix, & S. Cravatte (2011). Contrasting the flavors of El Niño‐Southern Oscillation using sea surface salinity observations, J. Geophys. Res., 116. doi: 10.1029/ 2010JC006862 Solomon, A., & M. Newman (2012). Reconciling disparate twentieth‐century Indo‐Pacific ocean temperature trends in the instrumental record. Nat. Climate Change, 2, 691–699. Stevenson, S., A. Capotondi, J. Fasullo, B. Otto‐Bliesner (2019),2017). Forced changes to twentieth century ENSO diversity in a last millennium context, Clim. Dyn., 52, 7359– 7374. doi: 10.1007/s00382‐017‐3573‐5 Sullivan, A., J.‐J. Luo, A. C. Hirst, D. Bi, W. Cai, & J. He (2016). Robust contribution of decadal anomalies to the fre­ quency of central Pacific El Niño. Sci. Rep., 6, 38540. doi: 10.1038/srep38540 Takahashi, K., & B. Dewitte (2016). Strong and moderate non­ linear ENSO regimes. Climate Dynamics, 46, 1627–1645. Takahashi, K., & A. G. Martinez (2019). The very strong coastal El Niño in 1925 in the far‐eastern Pacific. Clim. Dyn., 52, 7389–7415. doi: 10.1007/s00382‐017‐3702‐1 Takahashi, K., A. Montecinos, K. Goubanova, & B. Dewitte (2011). ENSO regimes: Reinterpreting the canonical and Modoki El Niño. Takahashi, K., C. Karamperidou, & B. Dewitte (2019). A theo­ retical model of strong and moderate El Niño regimes. Climate Dynamics, 52, 7477–7493. doi: 10.1007/s00382‐018‐4100‐z. Takahashi, K., V. Aliaga‐Nestares, G. Avalos, M. Bouchon, A.  Castro, L. Cruzado, et  al. (2018). The 2017 coastal El Niño. In “State of the Climate in 2017.” Bulletin of the American Meteorological Society, 99(8), S210–S211. Timmermann, A., S.‐I. An, J.‐S. Kug, F.‐F. Jin, W. Cai, et  al. (2018). El Niño Southern Oscillation complexity. Nature, 559, 535–545. doi: 10.1038/s41586‐018‐0252‐6 Trenberth, K. E., & D. P. Stepaniak (2001). Indices of El Niño evolution. J. Climate, 14, 1697–1701. Vecchi, G. A., B. J. Soden, A. T. Wittenberg, I. M. Held, A. Leetma, & M. J. Harrison (2006a). Weakening of tropical Pacific atmospheric circulation due to anthropogenic forcing. Nature, 441, 73–76. doi: 10.1038/nature04744

86  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Vecchi, G. A., A. T. Wittenberg, & A. Rosati (2006b). Reassessing the role of stochastic forcing in the 1997–8 El Niño. Geophys. Res. Lett., 33, L01706. doi: 10.1029/2005GL024738 Vimont, D. J., M. A. Alexander, & M. Newman (2014). Optimal growth of central and east Pacific ENSO events, Geophys. Res. Lett., 41. doi: 10.1002/2014GL059997 Wang, S.‐Y., M. L’Heureux, & H.‐H. Chia (2012). ENSO pre­ diction one year in advance using western North Pacific sea surface temperatures. Geophys. Res. Lett., 39, L05702. doi: 10.1029/2012GL050909 Watanabe, M., & A. T. Wittenberg (2012). A method for disen­ tangling El Niño‐mean state interaction. Geophys. Res. Lett., 39, L14702. doi: 10.1029/2012GL052013 Watanabe, M., J.‐S. Kug, F.‐F. Jin, M. Collins, M. Ohba, & A. T. Wittenberg (2012) Uncertainty in the ENSO amplitude change from the past to the future. Geophys. Res. Lett., 39, L20703. doi: 10.1029/2012GL053305 Williams, I. N., & C. M. Patricola (2018). Diversity of ENSO events unified by convective threshold sea surface temperature: A nonlinear ENSO index, Geophys. Res. Lett., 45, 9236–9244. Wittenberg, A. T. (2009). Are historical records sufficient to con­ strain ENSO simulations? Geophys. Res. Lett., 36, L12702. doi: 10.1029/2009GL038710 Wittenberg, A. T., A. Rosati, T. L. Delworth, G. A. Vecchi, & F.  Zeng (2014). ENSO modulation: Is it decadally predict­ able? J. Climate, 27, 2667–2681. doi: 10.1175/ JCLI‐D‐13‐00577.1 Wyrtki, K. (1975). El Niño: The dynamic response of the equatorial Pacific Ocean to atmospheric forcing. J. Phys. Oceanogr., 5, 572–594.

Xiang, B., B. Wang, & T. Li (2013). A new paradigm for the pre­ dominance of standing central Pacific warming after the late 1990s. Clim. Dyn., 41, 327–340, doi: 10.1007/s00382‐012‐1427‐8 Xie, S.‐P. (1999). A dynamic ocean‐atmosphere model of the tropical Atlantic decadal variability. J. Climate, 12, 64–70. Yeh, S.‐W., J.‐S. Kug, B. Dewitte, M.‐H. Kwon, B. P. Kirtman, & F.‐F. Jin (2009). El Niño in a changing climate. Nature, 461. doi: 10.1038/nature08316. Yu, J.‐Y., & S. T. Kim (2011). Relationships between extra‐ tropical sea level pressure variations, and the central‐Pacific and eastern‐Pacific types of ENSO. J. Climate, 24, 708–720. Yu, J.‐Y., H.‐Y. Kao, T. Lee, & S.‐T. Kim (2011). Subsurface ocean temperature indices for central‐Pacific and eastern‐ Pacific types of El Niño and La Niña events. Theor. Appl. Climatol., 103, 337–344. Yu, J.‐Y., M.‐M. Lu, & S.‐T. Kim (2012). A change in the rela­ tionship between tropical central Pacific SST variability and the extratropical atmosphere around 1990. Env. Res. Lett., 7. doi: 10.1088/1748‐9326/7/3/034025 Yu, J.‐Y., P.‐K. Kao, H. Paek, H.‐H. Hsu, C.‐W. Hung, M.‐M.  Lu, & S.‐I. An (2015). Linking emergence of the central‐Pacific El Niño to the Atlantic multi‐decadal oscilla­ tion. J. Climate, 28, 651–662. Yu, L., R. A. Weller, & T. W. Liu (2003). Case analysis of a role of ENSO in regulating the generation of westerly wind bursts in the western equatorial Pacific. J. Geophys. Res, 108(C4), 3128. doi: 10.1029/2002JC001498 Zhang, H., A. Clement, & P. Di Nezio (2014). The South Pacific meridional mode: A mechanism for ENSO‐like variability. J. Climate, 27, 769–783.

5 Past ENSO Variability: Reconstructions, Models, and Implications Julien Emile‐Geay1, Kim M. Cobb2, Julia E. Cole3, Mary Elliot4, and Feng Zhu1

ABSTRACT This chapter investigates ENSO variability before the instrumental era. Though generally indirect, paleoclimate observations provide information that no other source can, sampling ENSO behavior across different base states, subject to many types and intensities of external forcing, and providing a much longer statistical sample than afforded by the instrumental record. After first reviewing the nature, strengths, and caveats of the paleoclimate observations most relevant to ENSO, we outline how these observations may be used to infer changes in ENSO properties over time. The chapter then synthesizes the most robust paleoclimate inferences about ENSO over various time intervals: the Pliocene, Quaternary Ice Ages, the Holocene, the last millennium, and the anthropogenic era. ENSO appears to have operated on Earth for at least 3 million years, and the existing observations support the view that variations in ENSO amplitude and frequency arise primarily from processes internal to the climate system. However, multiple lines of evidence support the notion that ENSO properties are sensitive to large changes in mean climate, such as those seen during the anthropogenic era. Throughout these examples, a case is made that paleoclimate observations are now mature enough to offer quantitative constraints on ENSO and its representation in climate models, offering a key out‐of‐sample test of model predictions across a variety of climate scenarios. The chapter closes with a roadmap for furthering the relevance of paleoclimate observations to the study of ENSO.

5.1. CLIMATIC CONTEXT FOR PALEO-ENSO RECONSTRUCTION

projections of ENSO properties remain highly uncertain, owing to the presence of structural biases in coupled ocean‐atmosphere models (Bellenger et  al., 2014), combined with uncertainties in the relative sensitivity and strength of competing feedbacks that might change under continued warming (Collins et  al., 2010). The relatively short instrumental record (chapter 3) provides a truncated view of ENSO dynamics as expressed over a handful of El Niño and La Niña events, hampering both short‐ and long‐term prediction of future ENSO properties. This is especially true given that multicentury realizations of ENSO variability from long, unforced model integrations (Vecchi & Wittenberg, 2010; Wittenberg, 2009, and others) indicate a rich spectrum of ENSO variability, largely consistent with paleo‐ENSO archives (Cobb et  al., 2003, 2013). As such, it would require several centuries of high‐ quality instrumental data to fully characterize ENSO properties and how they might be changing in response to

As the largest source of interannual climate variability on Earth, ENSO represents a critical target for projections of future climate change impacts. While the seasonal prediction of ENSO extremes has advanced considerably over the last several decades, continued progress relies in part on deciphering the relationship between ENSO properties and the background climate state. Moreover, decadal‐scale 1 Department of Earth Sciences, University of Southern California, Los Angeles, CA, USA 2 Department of Earth Sciences & Technology, Georgia Institute of Technology, Atlanta, GA, USA 3 Department of Earth and Environmental Sciences, University of Michigan, Ann Arbor, MI, USA 4 Université de Nantes, LPG, Nantes, France

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 87

88  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

greenhouse forcing (Stevenson et  al., 2010). Given that long (multidecadal to multicentury) records of ENSO properties are readily available from a wide variety of paleoclimate archives, there is great potential in combining such records with climate models to investigate fundamental questions, like the dependence of ENSO frequency and amplitude on the mean state, their relationship with other major modes (e.g. the Asian Monsoon or extratropical annular modes), and their sensitivity to a variety of climate forcings, including greenhouse gases. In the broadest sense, paleoclimate reconstructions of ENSO variability allow for the assessment of ENSO’s sensitivity to a wide variety of climate forcings and associated changes in mean climate. In particular, changes in external radiative forcing associated with elevated CO2 concentrations, changes in Earth’s orbit driving ice ages, variations in solar luminosity, and large volcanic eruptions provide the most relevant analogues for ongoing anthropogenic climate forcing. Each of these areas has received considerable attention in the literature, both with respect to reconstructions using a wide variety of paleoclimate archives as well as numerous studies with a hierarchy of numerical models. While early investigations hinted at significant ENSO responses to most classes of external climate forcing, expanded observations of past ENSO variability demonstrate a high degree of intrinsic variability, consistent with long, unforced model simulations. Therefore, the threshold for detecting forced changes in ENSO has increased significantly, requiring many centuries worth of ENSO observations (Russon et al., 2014, 2015; Stevenson et al., 2010). Indeed, at this point the only signal that has emerged from the background is associated with trends in ENSO properties over the last decades as compared to long records of pre‐industrial ENSO properties – a likely sign of anthropogenic influence on recent ENSO properties (e.g. Cai et al., 2018; chapter 13). Here we describe how paleoclimate records from the tropical Pacific and other ENSO‐affected regions broaden our observational perspective of ENSO variability and how such information can constrain ENSO’s evolution in a warming world. The chapter is structured as follows. First, we describe the paleoclimate observations most commonly used in paleo‐ENSO studies, detailing their strengths and weaknesses (section  5.2). We explore the diverse ways that such data have been synthesized to reconstruct ENSO indices and patterns (section 5.3). We then examine how paleodata have been used to address the critically important question of ENSO response to past and ongoing changes in external climate forcings (section  5.4). Past forced changes appear to be subtle compared to the large internal unforced variability, and we discuss how we might refine efforts to identify ENSO responses (section  5.6), with implications for detecting anthropogenic effects on ENSO. Finally, we highlight

ways forward to improve and refine the use of paleo‐ ENSO records to evaluate physical theories of ENSO and inform model evaluations. 5.2. OBSERVATIONAL CONSTRAINTS ON PALEO-ENSO BEHAVIOR Paleoclimate proxies are, by definition, indirect recorders of climate. As such, they all harbor strengths and weaknesses, which we synthesize here in the context of ENSO. Notably, not all proxies are created equal: some, like corals, can sample ENSO behavior at monthly scales from the heart of the tropical Pacific. Other proxy systems contribute critically complementary records, often with a longer and more continuous temporal coverage (Figure 5.1; Table A1 in the chapter appendix), but often with a noisier relationship to ENSO itself. 5.2.1. Marine Archives Marine paleorecords of ENSO can sample directly near ENSO’s centers of action, tracking the tropical Pacific ocean temperature that defines ENSO and underpins its physics. Corals and bivalves share the ability to resolve seasonal as well as interannual scales, with limited record lengths: at most a few centuries for corals, and decades for tropical bivalves. However, both can also be found in fossil form, enabling climate reconstruction over discontinuous “windows” of commensurate length, or longer ones when overlapping segments can be spliced (Cobb et al., 2003). Because these organisms live in shallow waters, their presence closely tracks sea level; modern sea levels stabilized about 7000 years ago. It is particularly challenging to obtain coral or bivalve records from glacial intervals, when global sea level was about 120 m lower than today (Lambeck & Chappell, 2001). Only where tectonic uplift is rapid, for example, in Papua New Guinea (Murray‐Wallace and Woodroffe, 2014) and Indonesia (Pedoja et al., 2018), are coral and bivalve samples preserved from lower sea level intervals. More recent fossil material is preserved in beach deposits (Cobb et al., 2003) and archaeological sites (Carré et al., 2014). Ocean drilling on submerged reefs may recover usable coral skeletons, but this is serendipitous (e.g. Felis et al., 2012). 5.2.1.1. Corals Corals offer many strengths in reconstructing ENSO variability (Lough, 2010). They originate from coastlines across the tropical Pacific and record surface ocean temperature and salinity anomalies associated with ENSO activity in their skeletal geochemistry. They deliver monthly‐scale reconstructions that resolve the growth, peak, and decay of ENSO extremes. The relatively abundant supply of modern corals supports rigorous calibra-

PAST ENSO VARIABILITY  89 ENSO proxy archives (after 1000 AD) 0.0 0.00.0

0.0

0.0

0.2

0.4

0.4

800

0.2

0.0

0.0

0.0

0.2

0.4 0.6

0.2

age [yrs] 1000

0.2 0.4

0.4

0.0

0.8 0.6 0.2

400

0.2 0.0 0.0

0.0

0.0

0.2

0.2

0.2

0.2 0.0

0.2 0.0

0.4

0.0

0.0

coral (n = 87)

0.0

0

bivalve (n = 3)

0.2 0.4

0.0

0.2

0.0

0.4 0.2

0.0

0.0

0.4

0.2 0.0

0.2

0.0

0.4 0.2 0.0

–0.2

0.2

0.0

100

0.0

0.8 0.6

0.0 0.2

0.0

1000

0.0

0.2

0.0

0.4

age [ky BP]

0.0

0.00.0

0.4 0.6

0.2

0.2

0.2

0.0

0.2

200

0.0

ENSO proxy archives (before 1000 AD) 0.0 0.00.0

600

0.0

0.2 0.0 0.2 0.0 0.0

10

0.0

coral (n = 36)

bivalve (n = 9)

glacier ice (n = 1)

speleothem (n = 1)

lake sediment (n = 9)

tree (n = 1)

1

marine sediment (n = 9)

Figure 5.1  Spatiotemporal coverage of the paleo‐ENSO records cited in this study. Colors denote the most ancient age captured by each record (BP: before 1950 CE), while shapes denote archive types. Note the linear color scale for the past millennium (top), and logarithmic color scale for previous time intervals (bottom). Contours represent SST regressed onto the Niño‐3.4 index in ERSST v5 (Huang et al., 2017).

tion with instrumental records and reproducibility studies to inform uncertainty estimates (Figure 5.2b, c). Lastly, sub‐fossil corals preserved in beach deposits and uplifted terraces enable ENSO tracking in earlier intervals. Taken together, coral archives extend our quantitative view of ENSO variability into the last centuries (Charles et  al., 2003; Cobb et al., 2001; Cole et al., 1993; DeLong et al., 2012; Gorman et  al., 2012; Hereid et  al., 2013; Linsley et al., 1994; Urban et al., 2000), millennia (Cobb et al., 2003; Cobb et al., 2013; Hughen et al., 1999; McGregor et  al., 2013a; Tudhope et  al., 2001; Woodroffe et  al., 2003), and up to 3 million years ago (Watanabe et  al., 2011). For these reasons, coral geochemical records represent the gold standard of paleo‐ENSO measurements, with multiple records across the global tropics (Corrège, 2006; Gagan et  al., 2000; Lough, 2010; Tierney et  al.,

2015; Table  A1, appendix). Nonetheless, these records have important limitations, discussed below, including length of record, potential for skeletal alteration, age model error, biological stress responses, and geographical constraints. Most coral‐based climate reconstructions use the oxygen isotopic (δ18O) composition of the coral’s aragonite skeleton as a proxy for temperature (Epstein et al., 1953) and/or seawater δ18O, which correlates to salinity (Conroy et al., 2014; Fairbanks et al., 1997). In the case of ENSO, coral δ18O typically captures a signal of warm‐ wet (negative δ18O) versus cool‐dry (positive δ18O), yielding high signal‐to‐noise ratios in many such isotope‐based reconstructions. The relative contributions of temperature and hydrology are quantifiable over recent decades, but likely varied in the past, adding uncertainty to coral

δ18O Tridacna (‰)

(a)

–2.0

–6.0

–1.8

–5.8

–1.6

–5.6

–1.4

–5.4

–1.2 –1.0

–5.2

–0.8 –0.6

–4.8

–5.0

–0.2 0.0 1985

(b)

–4.6 –4.4

MT03-7 H01-9 H95-64

–0.4

1990

1995

δ18O Porites (‰)

90  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

–4.2 –4.0 2005

2000

–5.4

δ18O (‰)

–5.2 –5.0 –4.8 –4.6 SB7 – 0.06‰ CH9 – 0.06‰ SB5 + 0.19‰ SB6 + 0.11‰ CH5 + 0.04‰

–4.4 –4.2 1320

1340

SB7, CH9 R = 0.68

1360

1380

1400

CH5, SB5 R = 0.71

CH5, SB5 R = 0.69

1420

1440

1460

Year AD

(c) –5.5

Evans99– 0.15 Nurhati09+0.03 X12-6+0.09 X12-3+0.03

X14-1+0.06 X14-14+0.12 X14-7-0.033

X14-9+0.00 X16-MC1+0.14 X16-MC2+0.04

32

28

–5

26 –4.5

–4 1980

SST (C)

δ18O (‰)

30

24 IGOSS SST 1985

1990

1995

2000

2005

2010

22

2015

Figure 5.2  Reproducibility of paleo‐ENSO reconstructions using bivalve and coral δ18O profiles. (A) Comparison of δ18O profiles for two coral profiles (H95‐64 and H01‐9) and one bivalve (Tridacna) profile (MT03‐7) for the period 1985–2003 (Welsh et al., 2011). (B) δ18O profiles measured on five coral samples of different overlapping ages from the 14th and 15th century. Independent U/Th dates are used to constrain age model (triangles) and location of splice (S). The key reports the offsets in mean coral δ18O that were applied prior to plotting. Correlation coefficients for sets of overlapping coral δ18O records are shown along the bottom (Cobb et al., 2003). (C) Ten monthly‐resolved modern coral δ18O records from Christmas Island plotted with IGOSS SST (Reynolds et  al., 2002) from Christmas Island (grey). Each coral record has been shifted by up to 0.15‰ to match the mean δ18O values in overlapping intervals of the records. (Corals from left to right: Evans et al. (1999) (navy); Nurhati et al. (2009) (crimson); Grothe et  al. (2019), X12‐6 (emerald) and X12‐3 (purple); T. Chen, X14‐1 (sienna), X14‐14 (orange), X14‐7 (blue) and X14‐9 (hot pink); G. O’Connor, X16‐MC1 (brown) and X16‐MC2 (teal)). From Grothe et al. (2019).

PAST ENSO VARIABILITY  91

δ18O‐based interpretation of SST (LeGrande & Schmidt, 2011; Russon et al., 2013). Additional geochemical data can address this issue; skeletal Sr/Ca reflects dominantly SST and allows separation of SST‐ and SSS‐related contributions to coral δ18O variability (Alibert & McCulloch, 1997; Beckmann & Doescher, 1997; Gagan et al., 1998). These “paired” coral δ18O and Sr/Ca records are still relatively rare (e.g. Nurhati et al., 2011). Chronological control in coral‐based climate reconstructions relies on counting annual cycles in skeletal density and/or geochemistry. For fossil records, radiometric dates (usually U/Th) anchor these seasonal chronologies to an absolute age (e.g. Cobb et  al., 2003). In rare cases during the last millennium, a sufficient number of U/Th dates can constrain the absolute age of a given fossil coral sequence to a single year (Dee et  al., 2020). Within a core, annual or semi‐annual extremes are identified as seasonal extremes in local climate, and data are linearly interpolated between these tie points. Because the timing of max/min temperature/salinity can vary by 1–2 months, and growth rates vary intra‐annually, subannual age assignments are only accurate to within 2–3 months. Further, when counting years from the drill date to time periods preceding the start of the instrumental record in multicentury coral records, the potential to double‐count or miss a year compounds as the record length grows. Robust estimates of counting errors are rare but center on 1–2 years per century (DeLong et al., 2013). In a probabilistic assessment of counting error, Comboul et  al. (2014) found that failing to account for these uncertainties in combining multiple coral timeseries into multiproxy temperature reconstructions produces artificial weakening of ENSO variance further back in time, owing to the inevitable misalignment of ENSO extremes back in time. This issue likely creates artificial variance trends in studies that use networks of band‐counted records (section 5.3.2). Nonclimatic factors can also influence coral geochemical records. For example, postdepositional skeletal alteration can occur through dissolution, secondary aragonite precipitation, or conversion to calcite (Enmar et al., 2000; Hendy et  al., 2007; McGregor & Gagan, 2003; Muller et al., 2001). Extensive screening is now standard practice (Sayani et al., 2011). Older corals are more susceptible to diagenetic alteration, but corals as young as several decades can also be affected (Enmar et al., 2000; Hendy et al., 2007; Nurhati et al., 2009). Trace elements such as Sr/Ca are more susceptible to alteration than oxygen isotopes (Hendy et al., 2007). An additional challenge comes from recent observations that exceptionally warm conditions can disrupt coral growth, creating artifacts in skeletal growth as well as associated geochemistry in the most extreme cases (Clarke et al., 2017; Damassa et al., 2006; D’Olivo and McCulloch, 2017; D’Olivo et al., 2019;

Oliver et  al., 2009). Careful inspection of coral skeletal morphology and replication of geochemical sampling transects across horizons corresponding to exceptionally large El Niño events is therefore warranted at sites where this is a documented feature of modern‐day coral cores. Such information can then inform the assessment of older coral material for the presence of such artifacts. 5.2.1.2. Bivalves Modern shell studies in the tropical Pacific demonstrate that δ18O from marine bivalves closely tracks observed SST and accurately captures ENSO variability (Carré et al., 2013; Driscoll et al., 2014; Welsh et al., 2011). Such geochemical records provide snapshots of past climate over generally shorter intervals (years to decades, depending on the species) corresponding to the lifespan of these mollusks. Fossil shells are typically dated with radiocarbon or are sampled from deposits of known radiometric age. Bivalve studies are not as numerous as coral studies, but work focused on different species in the eastern and western tropical Pacific shows great potential for reconstructing the seasonal‐interannual statistics of ENSO. In the eastern Pacific, δ18O profiles from Peruvian surf clams, Mesoderma donacium, record ENSO fluctuations; they experience mass mortality when SSTs warm dramatically. Despite a short (2–4 year) lifespan, the distribution of SST anomalies reconstructed from modern samples captures the positively skewed distribution of ENSO in the eastern Pacific. These shells are abundant in some archeological middens, and large numbers can be analyzed to reveal past ENSO statistics (Carré et al., 2014). In the western Pacific, long‐lived (up to 40 years) giant clam shells of Tridacna sp. provide information on mean climate (Aharon, 1991) and ENSO during the Holocene and the last Glacial (Driscoll et al., 2014). Geochemical profiles from co‐located modern coral and Tridacna gigas in Papua New Guinea (Figure 5.2a) reveal the same history of both cool and warm events (Welsh et al., 2011). Holocene records with this species have also been derived from Papua New Guinea, New Caledonia (Duprey et al., 2012), and the South China Sea (Yan et al., 2017). 5.2.1.3. Marine Sediments Despite their ubiquitous distribution, marine sediments are not ideally suited to reconstructing the interannual ENSO phenomenon. In settings with unusually high deposition rates, sediments may preserve quasi‐annual layers that allow for interannual reconstructions, for example in the Santa Barbara Basin (Beaufort & Grelaud, 2017) and along the coast of Baja California (Marchitto et al., 2010). Coastal sediments on the Peru margin reveal a history of flood events, arguably ENSO‐driven (Rein, 2007; Rein et al., 2004, 2005). More typically, open‐ocean sediment records provide information on large‐scale

92  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

background conditions relevant to ENSO, for example, the zonal SST gradient (Wara et  al., 2005). Barriers to ENSO reconstructions include slow sedimentation rates, sediment mixing, calcite dissolution, age model limitations, and variable habitat preferences of the organisms that carry a climate signal. Some of the above complications may be circumvented by analyses of single foraminiferal shells: because a single foraminifer lives approximately one month, its δ18O or Mg/Ca is equivalent to a monthly temperature ­observation. The small mass of calcite in a single foram (~8–10 μg) makes this analytically challenging. Sampling many individual foraminifera in a given sediment horizon (representing ~100 years) captures the distribution of hydrographic variations at the habitat depth of the target species, which may be tied to ENSO statistics. This approach has been applied to explore ENSO variance changes over recent centuries to millennia (Khider et al., 2011; Koutavas et  al., 2006; Leduc et  al., 2009; Rustic et al., 2015). However, a thorough analysis of the sources of variance in such studies, including ENSO, seasonality, analytical error, and salinity, suggests that the importance of non‐ENSO variance, particularly seasonality, is a critical factor (Thirumalai et al., 2013). Recent advances in this area include the analysis of distributions from geochemical measurements on several dozen individual foraminiferal shells to detect ENSO changes inferred from single‐foram analyses across zonal and depth gradients (Ford et al., 2015; White et al., 2018). 5.2.2. Terrestrial Archives 5.2.2.1. Lake Sediments Lake sediment records are widely used to assess changes in ENSO‐related variability, within and beyond the equatorial Pacific. Lake sediments preserve environmental indicators such as plant pollen, physical sedimentology, geochemical proxies, and aquatic microfossils. This diversity of sediment‐based sensors allows investigators to infer environmental conditions that relate to ENSO: rainfall amount and intensity, vegetation, lake level, hydrologic balance, and (indirectly) SST. These records may span thousands of years continuously; their resolution is typically decadal or longer, and age models are usually radiocarbon based. The most relevant inferences that emerge from lake‐based ENSO studies bear on long‐term (>decadal) mean conditions and on indicators of changes in variability and extremes. Tight calibration of lake proxies to the local environment and ENSO remains a challenge, for many reasons: lack of local instrumental climate data; age model uncertainties; a dearth of lake process and monitoring studies; and nonlinearities in the proxy‐climate relationship. For example, detrital proxies are commonly used to infer

rainfall intensity (Conroy et al., 2008; Moy et al., 2002; Rein et al., 2005), yet calibration of these records to rainfall amount remains mostly inferential due to lack of local data and long‐term lake monitoring (Schneider et  al., 2018). Moreover, clastic sediment deposition is a highly nonlinear recorder of precipitation, and an even more nonlinear recorder of ENSO. This nonlinearity distorts temporal trends and exaggerates change points (Emile‐Geay & Tingley, 2016). While simple precautions may alleviate these issues, few studies have done so (Rein, 2007). 5.2.2.2. Tree Rings  Tree‐ring reconstructions hold many strengths: they are dated to the year, strongly replicated within a chronology, and typically well calibrated to seasonal or site‐ specific signals. Record lengths vary from a few centuries to 1–2 millennia. These reconstructions are most robust over interannual to decadal timescales, due to unavoidable steps in data processing that remove lower frequency variability and differ among reconstructions (Ault et al., 2013a; Cook et al., 1995; Fritts, 1976). Most tree‐ring chronologies have been developed in the subtropics and midlatitudes, where seasonal growth rings provide precise chronologies. Like most lake records, tree‐ring reconstructions usually reflect ENSO’s precipitation teleconnections, which are more variable and subject to interferences by local factors (chapter 14). An early ENSO reconstruction combined Pacific coral records with longer tree‐ring reconstructions from the southwestern US and Indonesia to reconstruct the SOI (Stahle et al., 1998). Several subsequent studies built on this multiproxy approach to generate ENSO‐relevant reconstructions that rely heavily on tree‐ring reconstructions (section 5.3). One of the most significant tree‐ring contributions to ENSO reconstruction has been the North American Drought Atlas (NADA; Cook et al., 2004). In this spatially explicit reconstruction of North American drought over the past millennium, ENSO emerges the most important contributor to reconstructed drought variance. A parallel effort using Asian tree rings (the Monsoon Area Drought Atlas, or MADA; Cook et al., 2010) also identifies a tropical Pacific signature in patterns of reconstructed drought. Combining ENSO‐sensitive tree‐ring chronologies from NADA, MADA, and other circum‐ Pacific sites, Li et al. (2013) obtained a 700‐year ENSO reconstruction of comparable statistical quality to coral‐ derived reconstructions of ENSO from the central Pacific. Oxygen isotope measurements made on tropical trees compose a relatively new source of paleo‐ENSO information, based on capturing ENSO‐related hydrological signatures in cellulose (Brienen et al., 2012; Evans

PAST ENSO VARIABILITY  93

& Schrag, 2004). Such records reflect the isotopic content of the water reservoir in the tree, which is heavily influenced by the oxygen isotopic composition of rainfall at the site (Baker et al., 2015). Ongoing studies of rainfall isotopes in the tropics indicates that they integrate regional‐scale hydrological variability through space and time, making them well suited to ENSO reconstruction (Moerman et al., 2013). However, tropical trees generally lack strong annual growth rings, making the establishment of chronologies challenging. 5.2.2.3. Other Terrestrial Archives ENSO imprints on other terrestrial climate archives, but these are rarely used for quantitative reconstructions. One exception is the Quelccaya ice core δ18O record (Thompson et al., 2013), which correlates strongly with an El Niño– like pattern of Pacific SST, particularly in the Niño‐4 region. Thompson et al. attribute this link not to local climate but to the control of tropical vapor δ18O by tropical SST patterns. Such inferences should be tested with expanded isotope observations and isotope‐enabled model simulations. The authors present an 1800‐year reconstruction of Niño‐4 SST but find that the spatial pattern of correlation is unstable even in the 20th century; the reconstruction itself is offered at decadal resolution only. Another potentially useful ENSO proxy may come from stalagmite records (McDermott, 2004). Frappier et  al. (2002) showed that a 13C record from a central American stalagmite contained a clear ENSO signal, though the mechanism remains elusive. In Borneo, where precipitation δ18O reflects regional convective activity (thus ENSO; Moerman et al., 2013), a recent cave study provided subannual constraints on interannual variability in several windows across the Holocene (Chen et  al., 2016). Another ENSO‐sensitive cave site has been identified on Niue Island (Rasbury & Aharon, 2006). In most cave records, however, interannual variability is smoothed away as precipitation mixes in the subsurface to become cave dripwater (Dee et al., 2017). In summary, paleo‐observations of ENSO come from a wide array of sources; their major attributes, strengths, and weaknesses are synthesized in Table 5.1. We now turn to how these observations are converted into quantitative estimates of ENSO state. 5.3. QUANTITATIVE APPROACHES TO ENSO RECONSTRUCTION Translating paleoclimate measurements to quantitative constraints on ENSO behavior is a cutting‐edge research question. Historically, paleoclimatologists have focused on studies from a single site, while increasingly they are attempting to weave together records from multiple sites and/or proxy types. Methods have ranged from purely

statistical, inverse methods to forward approaches leveraging climate models. There are advantages and disadvantages to each, which we flag below. 5.3.1. Single Proxy Reconstructions Several studies have employed paleoclimate observations to estimate ENSO’s temporal evolution. A simple approach is to use a single record to track temporal variations, within the assumptions and caveats of this particular record. The advantage of such an approach is that, if the interpretation is sound and the series can be assumed stationary, a single record provides the most stable picture of ENSO evolution at a given site. Calibration (to local conditions, or to an index of ENSO) is generally straightforward, and strong proxy‐climate relationships offer confidence in extending the record to pre‐instrumental periods. ENSO reconstructions from individual corals have identified, for example, changes in the strength of interannual variance (Cole et  al., 1993; Tudhope et  al., 1995); changes in decadal variability (Cobb et al., 2001; Linsley et al., 2015; Urban et al., 2000) and the nature of long‐term trends. Single sites, however, can only inform on local conditions, and the diversity of ENSO’s spatial footprint (chapter 4) means that multiple records are critical for understanding its full range of variability. A single site can be strongly impacted by one flavor of ENSO, yet unaffected by another, and the extent of ENSO’s footprint may thus appear to change through time (Tangri et al., 2018). This problem is compounded for teleconnected records, because teleconnection patterns are even more variable (Trenberth et al., 1998). Over the Holocene, there is perhaps no more iconic emblem of the single‐site approach than the runoff proxy from Laguna Pallcacocha in the Ecuadorean Andes (Moy et al., 2002; Rodbell et al., 1999). Based on early work that matched sediment color variations to a few 20th‐century events, Rodbell et  al. (1999) interpreted the record as an intensification of El Niño‐related rainfall extremes starting ~7000 years ago and peaking ~1200 years ago. However, more recent records from the region (e.g. Conroy et al., 2008; Thompson et al., 2017; Zhang et al., 2014) show different patterns of temporal change. Attempts at local replication have not been successful (Schneider et al., 2018), and analysis of instrumental data and high‐resolution model simulations reveal that ENSO impacts on rainfall are inconsistent in this topographically complex area (Kiefer & Karamperidou, 2019). As with all detrital records, any link to ENSO would be extremely nonlinear (Emile‐ Geay & Tingley, 2016). This reappraisal reminds us that tight calibration, regional replication, and connection to a stable physical mechanism are key to producing durable inferences.

Table 5.1  Paleoclimate archives of ENSO. “Analytical intensity” refers to the amount of effort required to produce each data point in the laboratory.

Realm

Archive

Marine

Corals

Bivalves

Foraminifera

Length Resolution and chronological precision

Analytical Intensity

Measurements

Proxies for

Length of record: 100‐400 y (modern); 50‐100y (fossil) spicing is possible. Resolution: weekly‐monthly. Chronology: 14C and U/Th (for fossils)

stable isotopes : δ18O trace elements: Sr/Ca trace elements: Ba/Ca

SST, seawater δ18O (SSS) temperature

2

Seawater Ba/ Ca: upwelling or runoff (sediments) Coral growth

2

Length of record: Years to decades. Resolution: weekly‐monthly. Chronology: 14C dating and/or stratigraphic uncertainties.

stable isotopes : δ18O trace elements: Sr/Ca, Mg/Ca

SST, seawater δ18O (SSS) Strong biological imprint nutrients

2

Length of record: > 1ka and longer. Resolution: Variance averaged over ~10’s‐100’s y (function of sedimentation rates) Chronology: 14C dating and stratigraphic uncertainties.

stable isotopes : δ18O trace elements: Mg/Ca

Temperature, seawater δ18O temperature

2

Extension rate, calcification rate

trace elements: Ba/Ca

2

1

2 2

2

Strengths

Caveats

Located in core ENSO regions. Direct measure of ocean conditions. Monthly to seasonal resolution. Well calibrated and widely used. Separate Temperature/ salinity Annual banding typically present, aids chronology. Located in core ENSO regions. Direct measure of ocean conditions. Seasonal resolution. Well calibrated and widely used. Small vital effects on δ18O.

Relatively short snapshots (10’s‐100’s years). Potential for vital and diagenetic effects. Analytically intensive, limits replication Coastal sites may be impacted by local effects and runoff. Banding reflects a complex relationship with local climate Limited number of study sites (coastal areas, uplifted areas)

Length of record. Potential location across the tropical oceans.

Relatively short snapshots (years ‐ < 80 years). Analytically intensive, limits replication Coastal sites may be impacted by local effects and runoff. Mixed SST and SSS signa, no independent temperature proxy. Seasonal cycles can be truncated. Limited number of tudy sites (coastal areas, uplifted areas) Low resolution. No timeseries of ENSO per se. Statistical conflation of seasonal cycle and other non‐ENSO components (analytical error, salinity, vertical habitat changes). Age model uncertainty. Vital effects, vertical distribtion uncertainty.

Terrestrial

Lakes

Trees

Caves

Length of record: 1ka and longer. Resolution: annual‐century. Chronology: 14C and stratigraphic uncertainties.

Length of record: 100’s to > 1ka. Resolution: Annual. Chronology: 14C (fossil) and annually precise chronologies.

Sedimentology (e.g. grain size, magnetic susceptibility, color scans) Pollen Isotopes on aquatic organisms or authigenic sediments Isotopes on terrestrial orga-­ nics deposited into lake Ring width/ Density

Stable isotopes (δ18O)

Length of record: >1ka. Stable isotopes Resolution: subannual (δ18O) to decadal. Chronology: band‐ counting (rare); U/Th chronology

Erosion intensity inferred as intense rainfall Vegetation Hydrologic balance

Precipitation δD, thus amount, tra jec­tory, seaso­nality of precipitation. Locally calibrated to seasonal temperature and/or hydroclimate Precipitation δ18O, thus amount, trajectory, seasonality of precipitation

Precipitation δ18O, thus amount, trajectory, seasonality of precipitation

1 or 2

3 2 or 3

Length and geographic distribution of records. Enables to explore older periods (LGM Pleistocene) Direct measure of plant presence. Plant‐climate relationships have been well studied. Length and geographic distribution of records.

Links to core ENSO region may be distant. Age model uncertainties. Nonlinear relationship between sedimentation and climate/ hydrological forcings. Nonlinearities related to pollen production and transport. Time lags of vegetation response to climate. Multiple influences on climate‐precipitation δD relationship. Calibration is challenging.

Annually precise and accurate chronologies. Statistical rigor with massive replication. Underpinned by extensive research on tree growth‐climate relationships. Statistical calibrations to climate are strong on interannual‐decadal scales. Biases can be identified. Some studies can separate summer and winter variables. δ18O profiles capture seasonality and tend to prese­rve low frequency informa­tion better than ring widths. Length, resolution, and geographic distribution of records.

Links to core ENSO region may be distant (tropical trees rarely form clear annual rings that crossdate). Ring width records must be detrended to remove growth effects, which removes lower frequencies from reconstructions. Seasonal biases affect what parameters can be reconstructed. δ18O records analytically intensive, acting as barrier to replication.

3

1

3

2

Depending on location, links to core ENSO region may be distant. Age model uncertainties. Multiple influences on climate‐ precipitation δ18O relationship. Records are analytically intensive, making replication challenging. Calibration is challenging.

96  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

5.3.2. Multiproxy Reconstructions Given the limitations inherent to each paleoclimate proxy (Table 5.1), another approach is to combine information from multiple proxies so their common signal emerges (Mann, 2002). This line of research parallels reconstructions of other important paleoclimate indicators (e.g. Northern Hemisphere temperature) and may be classified according to three criteria: (1) the target of the reconstruction, whether an index (e.g. Niño‐3, SOI) or a field (e.g. tropical Pacific SST); (2) the network of paleoclimate timeseries used as input; and (3) the method used for the estimation. An implicit assumption is that proxy errors are independent, so linear combinations like averaging amplify the climate signal over the noise. This assumption makes the auxiliary hypothesis that records are synchronous. As shown by Comboul et al. (2014), this is unlikely to be the case unless records are cross‐dated, and may lead to apparent trends in ENSO statistics over time. Thus, multiproxy reconstructions must be interpreted with this caveat in mind. To sidestep this difficulty, several studies have focused on reconstructions of the variance itself (Li et  al., 2011, 2013; McGregor et  al., 2013b). The latter study showed, using simulated data, that reconstructing the variance on running windows prior to aggregating records effectively mitigated age uncertainties in high‐resolution records of the past millennium. However, uncertainties in variance estimates must still be dealt with to confidently refute the null hypothesis that changes in variance arise solely because of sampling (Russon et al., 2014, 2015). In principle, multiproxy reconstructions are possible for any time period. However, the paucity of early records has, so far, restricted such estimates to the last millennium. These attempts were reviewed in Emile‐Geay et al. (2013a). In addition, the authors reconstructed Niño‐3.4 variability over the past 850 years, using two complementary statistical methodologies and a focused set of ENSO records, buttressed by pseudoproxy experiments. They found high consistency on interannual estimates but large uncertainties in the estimation of lower frequency components, with large sensitivity to the choice of instrumental calibration dataset (Emile‐Geay et  al., 2013b). This highlights how uncertainties in long‐term instrumental data from the tropical Pacific (chapter 3) impose a fundamental “event horizon” on multiproxy reconstructions, in addition to uncertainties stemming from paleoclimate observations themselves. Perhaps surprisingly, then, improving instrumental SST estimates would also benefit paleoclimate inference about ENSO. A coral‐based reconstruction of tropical Pacific SST field (Emile‐Geay et  al., 2012, 2013c) highlights the promise and challenges of the multiproxy reconstruction of climate fields relevant to ENSO. The authors used a

network of coral records from the past millennium (Table  A1, appendix), together with a space‐preserving climate field reconstruction technique (GraphEM; Guillot et  al., 2015). Cross‐validation exercises demonstrated high reconstruction skill over the instrumental record, resolving over 50% of Niño‐3.4 variance with correct amplitude and mean Figure  5.3). Over space, Figure 5.4 (left) highlights three tropical years (Apr/Mar periods, straddling the peak ENSO time) reconstructed by this method, as well as the Last Millennium Reanalysis (section 5.3.3). It would be tempting to use such spatial reconstructions to investigate how ENSO “flavors” (chapter 4) have changed over time. However, Emile‐Geay et  al. (2013c) showed that this coral observing system (Figure  5.1) overemphasizes the western and central Pacific, resulting in a bias towards CP events, and underestimating the amplitude of EP events. Overall, the coral network captures La Niña events more faithfully than El Niño events, whose most extreme EP events (e.g. 1982–1983, 1997– 1998) are muted in the reconstruction. This results in a more symmetric distribution of anomalies than observed; that is, this particular set of observations underestimates SST skewness, a key measure of ENSO’s nonlinearity (Timmermann et  al., 2018). Emile‐Geay et  al. (2013c) also found that the effect of age uncertainties is nonnegligible and should be accounted for in any coral‐based approach to detect long‐term trends in the prevalence of ENSO flavors (e.g. Freund et al., 2019). In summary, while advances in multivariate statistics now enable the faithful retrieval of SST patterns from paleoclimate observations, the changing observing network through time and the unavoidable age uncertainties that come with even the most carefully generated coral records impose limitations on the information that may be inferred from multiproxy reconstructions. 5.3.3. Fusing Data and Models The reconstruction methods described thus far employ an inverse approach: they all express ENSO state as a linear combination of proxy observations. An alternative approach is to design estimation methods built around the recognition that climate influences proxies values, not the reverse. Bayesian methods (e.g. Tingley et al., 2012) generally accomplish this, though none has so far been applied to ENSO reconstruction. An emergent tool in paleoclimatology is data assimilation (DA), which optimally combines information from climate models and paleoclimate observations (Bhend et  al., 2012; Franke et al., 2017; Goosse et al., 2006, 2010; Matsikaris et al., 2015; Steiger et al., 2014; Widmann et al., 2010). The Last Millennium Reanalysis (LMR; Hakim et al., 2016; Tardif et  al., 2019) is one such example. LMR

PAST ENSO VARIABILITY  97 Annual Niño 3.4 index (corr = 0.82; CE = 0.64) 4

LMR (median)

GraphEM

Bunge & Clarke (2009)

Niño 3.4

2 0 –2 –4 1880

1990

1920

1940

1960

1980

2000

Year (AD) 4

LMR (median)

GraphEM

Niño 3.4

2 0 –2 –4 800

1000

1200

1400

1600

1800

2000

Year (AD)

Figure 5.3  The Niño‐3.4 index as reconstructed by GraphEM (Emile‐Geay et al., 2013c; Guillot et al., 2015) and the Last Millennium Reanalysis (LMR; Hakim et al., 2016; Tardif et al., 2019; this study). Top: comparison to an instrumental series (Bunge & Clarke, 2009). Bottom: comparison over 800–2000 CE. See text for details. GraphEM - 1642 (AD)

LMR - 1642 (AD)

[K] 1.0

0.5 GraphEM - 1799 (AD)

LMR - 1799 (AD)

0.0

GraphEM - 1806 (AD)

LMR - 1806 (AD)

–0.5

–1.0

Figure 5.4  Reconstructed surface temperature with two different methods/proxy networks: (left) GraphEM/corals and (right) LMR/PAGES 2k records. We display three individual years to highlight similarities and differences: 1642, 1799, and 1806 (year of the Tambora eruption).

98  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

proceeds from a prior of physically plausible climate states drawn from a long transient general circulation model (GCM) simulation, which is forward‐modeled into observation space, then compared to paleo observations via an ensemble Kalman filter. In the present case, we used the “all‐forcings” simulation of last millennium climate (Stevenson et  al., 2019) with the isotope‐enabled Community Earth System Model (iCESM; Brady et al., 2019) as the prior, though many other choices are possible. An essential feature of LMR is that this prior is used as a static ensemble of climate states, so all temporal variations arise solely from paleoclimate observations. Spatial relationships, however, are derived from the climate model, modulated by observations via the Kalman update equations (Hakim et al., 2016; Steiger et al., 2014). A major advantage is that relationships within and between climate fields are dynamically consistent, unlike those obtained by purely statistical means. A drawback is that the reconstructed patterns inherit model biases (e.g. the cold tongue bias, cf. chapter 9). To illustrate this approach, we apply the LMR framework to the PAGES 2k Consortium (2017) dataset, whose coral component (Tierney et  al., 2015) includes many of the same records as used above (Figure  5.1, Table A1). A key specification of the LMR framework concerns the modeling of paleoclimate observations. Here, all observations are modeled as linear functions of annual temperature, except for tree‐ring records (modeled as a linear combination of seasonal temperature and moisture signals; Tardif et  al., 2019) and coral records (using the process model of Thompson et  al., 2011). While the tree‐ring model requires calibration against instrumental observations of temperature and moisture, the coral model does not; its only coefficient (linking δ18O to SST) is determined from laboratory experiments (Epstein et  al., 1953), and the isotopic composition of seawater provided by iCESM, therefore eschewing calibrations to salinity. Figure  5.3 shows the Niño‐3.4 reconstruction, whose median is highly correlated with the instrumental estimate of Bunge and Clarke (2009) (r = 0.82, p < 10−3). An advantage of DA is its ensemble‐based uncertainty quantification, primarily achieved through randomly withholding parts of the prior and observations to generate a large collection of plausible realizations of paleoclimate histories (red envelopes on Figure  5.3, which encompass the instrumental data). Contrasting this LMR reconstruction of Niño‐3.4 with that obtained through GraphEM (section  5.3.2), Figure  5.3 reveals a notable consistency: despite diametrically opposite methodologies, different observational inputs and different instrumental targets (annual SST for LMR, Dec‐Jan‐Feb SST for GraphEM), the two reconstructions share a correlation of 0.63 after 1850 (p < 10−3 using an isospectral test; Ebisuzaki, 1997), decreasing to 0.25 (p < 10−3) over

1600–1850, a low but statistically significant correlation, reflecting the data sources common to both reconstructions. We see that the GraphEM reconstruction is mostly contained within the uncertainties of the LMR reconstruction. Figure 5.4 (right) shows SST snapshots for the same 3 years as before. The LMR spatial patterns are highly reminiscent of the prior (here, iCESM), which like most GCM simulations of ENSO tends to extend too far west, a direct consequence of the cold tongue bias. There is remarkable agreement on the spatial pattern, sign, and amplitude of reconstructed surface temperature for the year 1806, where a strong La Niña episode is inferred in both reconstructions. In 1642, both reconstructions infer a moderate El Niño episode, though its pattern evokes an EP event in the GraphEM case and a CP event in the LMR case. On the other hand, both reconstructions are at odds over all aspects of the 1799 event, even its sign (La Niña in GraphEM, El Niño in LMR). There is ongoing research to uncover the source of such discrepancies. A simple diagnostic is the correlation between the reconstructed (median) Niño‐3.4 and the coral observations of Cobb et  al. (2003), which are the longest and most proximate observations to constrain this index (Tierney et al., 2015). This correlation is −0.95, meaning that when Palmyra data are available, they drive 90% of the variance in Niño‐3.4; over the intervals where those observations are not available, other (usually teleconnected) sites are used for inference. In summary, paleoclimate DA offers a new way to weave together information from GCMs and paleo‐ observations in a systematic way. It cannot, however, perform miracles: the large uncertainties of Figure 5.3 are a reminder that many more annually resolved observations from the heart of the tropical Pacific (Comboul et  al., 2015), and to a lesser extent from teleconnected regions, will be needed to adequately constrain ENSO behavior over the past millennium. 5.4. PALEO-CONSTRAINTS ON ENSO DYNAMICS We now synthesize the current paleoclimate literature to identify key messages on how ENSO varied over various intervals, subjected to various forcings. 5.4.1. Pliocene Reconstruction of ENSO variability during periods significantly different to modern are key to understanding the relationship between the mean climatic state and ENSO. Evidence for ENSO variability during the Pliocene warm period (3 to 5 million years ago) has been derived from sediment cores and fossil Porites samples. These proxy‐based studies show conflicting results (Rickaby & Halloran, 2005; Wara et al., 2005; Watanabe

PAST ENSO VARIABILITY  99

et al., 2011), which stem in part from a confusion between ENSO and ENSO‐like variability. Rickaby and Halloran (2005) suggested that there is a shallower thermocline during this period, indicative of La Niña–like conditions. Wara et  al. (2005) challenged this conclusion using high‐resolution data from two sediment cores from the eastern and western Pacific ocean, suggesting a reduced zonal temperature gradient, dubbed “permanent El Niño–like conditions.” This “Pliocene paradox” was initially the topic of much research (Fedorov et  al., 2006, 2010; Haywood et  al., 2007), but the concept has been refuted by Bonham et  al. (2009), who argued that sparse Pliocene observations are likely to give the appearance of permanent El Niño–like conditions even when ENSO is present, or the mean zonal SST gradient is similar to modern. More recently, a monthly resolved coral record from the Philippines (Watanabe et al., 2011) showed a similar pattern of interannual variability when compared to a modern specimen from the same site. The authors interpret this as indicating that ENSO operated during this particular interval, contradicting the claim of a permanent El Niño state. Discrepancies could arise from a combination of factors: first, chronological uncertainties, as obtaining accurate age constraints during this period is difficult and the records described above may not be precisely from the same period; second, ENSO is known to exhibit large internal variability, necessitating long records to characterize. Finally, limitations come from the scarcity of well‐preserved fossil coral samples from this time period. The exceptional preservation of the samples presented by Watanabe et al. (2011) is noteworthy in this context, and, until contradicted by more direct evidence, supports the claim that ENSO operated similarly to present during at least some of the Pliocene, irrespective of the strength of the zonal SST gradient. 5.4.2. Glacial‐Interglacial Variability The last glacial maximum (LGM; 18–21 ky BP) was characterized by lower atmospheric CO2 (Petit et  al., 1999), much larger continental ice sheets (Peltier, 2004), and mean temperatures around 5°C colder than modern (MARGO Project Members, 2009). A key driver of glacial‐interglacial variability is insolation, particularly variations in precession that modulate the seasonal cycle, as well as fluctuations in atmospheric CO2 driving changes in global temperatures and ocean structure (DiNezio et al., 2011). The presence of large continental ice sheets also led to different latitudinal temperature gradients affecting teleconnections (Cook & Held, 1988) and freshwater fluxes affecting ocean structure and circulation (Luan et al., 2015). Forcing of ENSO may also arise from remote processes such as the Asian Monsoon, which varies significantly over these time periods (e.g. Zhisheng et al., 2011).

Tudhope et  al. (2001) analyzed a series of fossil corals from the Huon Peninsula in Papua New Guinea, demonstrating that ENSO had existed for the past 130,000 years, operating even during colder glacial times with varying amplitude. They suggested that the observed changes in amplitude may have resulted from the combined effects of a cooler mean state and precessional forcing. These observations were in line with initial results from studies using an intermediate complexity ENSO model (Zebiak & Cane, 1987), which showed precessionally controlled changes in ENSO variability through the last glacial cycle (Clement & Cane, 1999; Clement et al., 1999, 2000). Since then, ENSO behavior over glacial‐interglacial cycles has been the subject of many investigations with more comprehensive models (Lu et al., 2018). The LGM is a key target to evaluate ENSO behavior under glacial boundary conditions and has long been a focus of experiments by the Paleoclimate Modelling Intercomparison Project (PMIP; Kageyama et al., 2017). Such experiments, however, have been inconclusive, simulating both decreased and increased ENSO activity during the LGM (Brady et  al., 2012; Liu et  al., 2014; Masson‐Delmotte et al., 2013; Otto‐Bliesner et al., 2003, 2006; Zheng et al., 2008; Zhu et al., 2017). A significant limitation when studying this period is that lowered sea level (around 120 m below modern; Lambeck & Chappell, 2001) implies that the corals or bivalves that experienced ENSO conditions are now located around 120 m below modern sea level, making sampling exceedingly difficult. This explains why most of the LGM reconstructions of ENSO (Figure  5.1) are based on marine sediment records, particularly individual foraminiferal analyses (IFA, section  5.2.1.3). These studies show conflicting results. Records from the eastern equatorial Pacific show strengthened variability, e.g. Koutavas and Joanides (2012), who used surface‐dwelling foraminifera from a single site in the eastern Pacific cold tongue; or Sadekov et  al. (2013), who used subsurface species from both eastern and western Pacific sites. Other studies using thermocline‐dwelling species infer a weakened ENSO at the LGM (Ford et al., 2015; Leduc et al., 2009). Leaning on the interpretative tool of Thirumalai et al. (2013) and new single‐foramniferal measurements, Ford et al. (2015) reconciled previous results (Koutavas & Joanides, 2012; Koutavas et al., 2006; Leduc et al., 2009) and suggested that subsurface tracers are required to reconstruct past ENSO as they reflect changes in the thermocline depth, which more accurately reflects the Walker circulation and the state of ENSO. They argue for enhanced seasonality and reduced ENSO variability during the LGM, which they interpret as consistent with the frequency‐entrainment hypothesis (Chang et  al., 1994; Liu, 2002; Tziperman et al., 1994). Recent simulations with iCESM (Brady et  al., 2019; Nusbaumer et  al., 2017), which forward‐model the

100  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

distributions of individual foraminiferal analyses (Zhu et al., 2017), also conclude that ENSO activity was weaker at the LGM than the preindustrial baseline, but come to different conclusions regarding the role of the annual cycle. The primary mechanism appears to involve a shallower, sharper mean thermocline (Fedorov & Philander, 2001), mirroring the ENSO response to increased CO2 (Cai et  al., 2018; DiNezio et  al., 2009). The analysis of Zhu et al. (2017) also supports the contention that IFAs from the Eastern equatorial Pacific reflect an enhanced annual cycle, rather than increased ENSO variability, at the LGM (Ford et al., 2015; Thirumalai et al., 2013). More abundant fossil coral samples are available during previous interglacial periods, characterized by a sea level closer to modern. The last interglacial (aka Marine Isotope Stage 5e), dated around 125 ky BP (129–116 ky BP), has been of particular interest: fossil material is well preserved, chronologies based on U/Th are accurate, and sea level was on average several meters above modern (Dutton & Lambeck, 2012). Surprisingly, there are very few fossil coral records that provide information on ENSO for that time period. Tudhope et al. (2001) and Hughen et al. (1999) studied fossil Porites collected in Papua New Guinea and North Sulawesi, respectively. Both showed ENSO behavior not significantly different from modern, as expected from the similar boundary conditions. Recently, Thirumalai et al. (2019) used IFA to suggest the existence of an El Niño–like mode in the glacial Indian ocean, consistent with coupled GCM simulations. If supported by subsequent observations, this would have major implications for the dynamics of ENSO at the LGM, including its relationships to the Asian Monsoon and Indian ocean variability. 5.4.3. Holocene Variability The Holocene provides a unique laboratory in which to probe ENSO’s unforced behavior, as well as its response to insolation forcing. While ice volume and greenhouse gas concentrations were essentially similar to today’s, the latitudinal and seasonal distribution of incoming solar radiation (insolation) were markedly different as a result of orbital precession: during the classically defined “mid‐ Holocene” (6.5 ky BP), seasonal insolation contrast was stronger than today in the northern hemisphere and weaker in the southern hemisphere. This natural experiment provides an opportunity to explore the link between changes in the seasonal cycle, meridional asymmetry in the equatorial zone, and ENSO behavior (Luan et  al., 2012). The mid‐Holocene has been the target of concerted modeling efforts as part of PMIP (Braconnot et  al., 2012; Otto‐Bliesner et  al., 2017), and therefore offers a clean experimental framework to compare model simulations with paleo‐observations. Initially, various

proxies appeared to show reduced ENSO variance during the mid‐Holocene (Koutavas et al., 2006; McGregor and Gagan, 2004; Moy et  al., 2002; Rodbell et  al., 1999; Tudhope et  al., 2001; Woodroffe et  al., 2003), which inspired many theoretical explanations (An & Choi, 2014; Chiang et al., 2009; Clement et al., 2000; Liu et al., 2000, 2014; Saint‐Lu et  al., 2015; Zheng et  al., 2008), despite contradictory evidence (Corrège et al., 2000). In the past few years, an explosion in the high‐resolution data coverage of the Holocene from corals (Cobb et  al., 2013; Grothe et  al., 2019) and bivalves (Carré et  al., 2014; Driscoll et  al., 2014), together with a reexamination of teleconnected records (Kiefer & Karamperidou, 2019; Schneider et al., 2018), has provided a more nuanced view. In the central equatorial Pacific, Cobb et al. (2013) used 20 new coral sequences spanning short intervals of the past 7 ky to reveal highly irregular ENSO activity over the Holocene, contradicting the simplified picture of a gradual intensification of ENSO since the mid‐Holocene (Chiang, 2009). Notably, Cobb et  al. (2013) showed that ENSO activity in the late twentieth century is exceptionally high in the context of the Holocene, consistent with previous work (McGregor et al., 2013b; section 5.4.5). Using short‐lived clams from the Peruvian coast, Carré et al. (2014) showed that ENSO variance was close to the modern level in the early Holocene and severely damped ~ 4 to 5 ky ago. This finding is consistent with results from Grothe et al. (2019), who expand the Cobb et al. (2013) fossil coral archive and in doing so, document a statistically significant decrease in ENSO variability from 3–5 ky BP relative to the 5–7 ky and 1–3 ky adjacent intervals. Synthesizing these and other observations, as well as CMIP5‐PMIP3 model simulations, Emile‐Geay et al. (2016) argued that ENSO variability is only notably reduced across the Pacific during the interval 3–5 ky BP. This pattern bears no resemblance to any aspect of insolation forcing, making the latter an unlikely cause of such changes. A simpler explanation, consistent with theoretical models (Timmermann & Jin, 2002; Zebiak & Cane, 1987), is that such variations are endogenous to the climate system. Furthermore, the high density of seasonally resolved observations from corals allowed Emile‐Geay et al. (2016) to test the frequency entrainment hypothesis (Chang et al., 1994; Liu, 2002; Timmermann et al., 2007), which holds that ENSO amplitude is inversely related to the amplitude of the annual cycle, and has long been invoked to explain changes in ENSO variance (An & Choi, 2014; An et al., 2010; Ford et al., 2015; Liu et al., 2014; Lu and Liu, 2019; Timmermann et al., 2007). While the PMIP3 models do show this inverse relationship, the observations showed a neutral or positive relationship (Emile‐ Geay et al., 2016). One way to resolve this paradox comes from the work of Khon et  al. (2018), who used long transient integrations with a coupled GCM to show that

PAST ENSO VARIABILITY  101

it may display coeval intensification of both the seasonal and ENSO cycles, and that frequency entrainment only operates on short timescales. This analysis of seasonally resolved Holocene records supports the view that ENSO activity has no simple, if any, relation to insolation forcing (Emile‐Geay et  al., 2016). Investigating why state‐of‐the‐art models disagree on ENSO’s response to precession in PMIP3 models, An and Bong (2018) showed that it is the result of a delicate balance of opposing feedbacks, echoing similar conclusions regarding the annual cycle (Bellenger et  al., 2014; Lloyd et al., 2009, 2011, 2012). Complementing this picture from seasonally resolved marine archives, lake sediments from ENSO‐impacted regions (notably eastern/northern Australia and the western US) show clear increases in variability from the mid‐ Holocene to present, consistent with an intensification of ENSO’s impact on these areas. These include the expansion of drought‐adapted vegetation (McGlone et  al., 1992), intensified variability in reconstructed precipitation (Barr et al., 2019), and changes in sedimentology consistent with more variable hydroclimate. Initially interpreted as indicating changes in ENSO, these records are perhaps better interpreted as changes in the geographic reach of ENSO, as the modern climate regime becomes established following the retreat of ice sheets and insolation‐related changes in the monsoon and convective zones. In a promising development, researchers are leveraging lake sedimentary sequences from equatorial Pacific islands to explore the evolution of multidecadal to multicentury tropical Pacific climate variability (Conroy et  al., 2008, 2009; Sachs et  al., 2009; Thompson et  al., 2017; Zhang et al., 2014). A lake record from El Junco, Galápagos, supports the intensification of rainfall events starting around 4200 years BP, based on grain size variability (Conroy et al., 2008). Another Galápagos record from Bainbridge Crater, backed by lake monitoring, reconstructs low event frequency between 4000 and 6100 years BP and century‐scale variability in the past 2000 years (Thompson et al., 2017). Taken together, the lake‐based record of ENSO and its impacts support the idea of intensifying variance during the Holocene. Although most of these records are linked by remote teleconnections, the Galápagos records remind us that significant changes did occur in the equatorial Pacific itself, at least in the rainfall regime that feeds the lakes. The challenge remains to connect these hydroclimatic shifts with high‐resolution ocean temperature records that point to a different history of change. 5.4.4. Last Millennium Variability The last millennium is the most recent data‐rich period of past climate intervals, where ENSO’s relationship to

solar and volcanic forcing can be probed (Jungclaus et al., 2017; Schmidt et al., 2012). 5.4.4.1. Testing the Link to Solar Forcing Solar irradiance exhibits an 11‐year anharmonic cyclicity over the period of instrumental measurements (the Schwabe cycle, e.g. Foukal et al., 2006). Longer‐term variations in irradiance have been inferred from solar magnetic activity, taken as a proxy for irradiance: these comprise the Gleissberg (~88 y) and DeVries (~205 y) cycles (Peristykh & Damon, 2003; Wagner et al., 2001). Their irradiance scaling is extremely uncertain (Fröhlich and Lean, 2004; Gray et  al., 2010), reflected in widely varying estimates of solar forcing over the past millennium (Schmidt et al., 2012). Continuous reconstructions of ENSO indices offer a way to test whether a statistically significant association exists between proxies of solar irradiance and ENSO, a necessary but not sufficient condition to establish a causal link. Applying wavelet‐ transform coherency (Grinsted et al., 2004), Emile‐Geay et  al. (2013b) found evidence for a robust antiphasing between a variety of solar activity indices and their three reconstructions of  the Niño‐3.4 index. This is broadly consistent with the thermostat hypothesis (Clement et al., 1996), within the caveats of large reconstruction uncertainties for both the  forcing and Niño‐3.4. This provisional test should be periodically revisited in light of improved reconstructions of both aspects. 5.4.4.2. Testing the Link to Volcanic Forcing The potential effect of explosive volcanism on ENSO is reviewed in chapter  12, using both climate models and paleoclimate reconstructions. Here we therefore focus on aspects not explicitly discussed in that chapter. A common way to assess volcanic influence is superposed epoch analysis (Adams et al., 2003). This analysis, as well as others (D’Arrigo et al., 2009; Dätwyler et al., 2019; McGregor et al., 2010; Wahl et al., 2014) finds marginal support for El Niño occurrences anywhere from 1 to 4 years after an eruption. However, chapter  12 also finds ENSO occurrences prior to volcanic events, questioning causality. One possible reason is that such analyses require extremely accurate dating, whereas it is known that multiproxy reconstructions are vulnerable to dating uncertainties (Comboul et al., 2014). Indeed, Sigl et al. (2015) showed that even cross‐dated tree‐ring chronologies, often considered the gold standard of chronological accuracy in the paleo realm, were offset by 1–2 years in the case of some eruptions. Such offsets get compounded in a compositing methodology such as Superposed Epoch Analysis (SEA). A second reason is that SEA is rarely benchmarked against an appropriate null hypothesis (Haurwitz & Brier, 1981; Rao et  al., 2019), if at all. A third reason is that many of the reconstructions involve

102  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE NIN03.4 MTM spectra 1

500

200

100

50

20

10 8

6

4

2

Frequency × Spectral Density

10–1

10–2

10–3

10–4

CCSM4,PI CCSM4,LM GISS-E2-R, PI GISS-E2-R, LM IPSL-CM5A-LR, PI IPSL-CM5A-LR, LM MIROC-ESM, PI MIROC-ESM, LM MPI-ESM-P, PI MPI-ESM-P, LM

LIM EG12, ERSSTv3 EG12, HadSST2i EG12, kaplan

10–2

10–1

0.5

Frequency (cycles/year)

Figure 5.5  Spectral density of Niño‐3.4 in PMIP3 simulations and the reconstructions of Emile‐Geay et al. (2013a, 2013b). the logarithm of fS(f) is plotted against log(f) to preserve the area under the curves, allowing integration “by eye.”

extratropical tree rings, whose ENSO imprint is the result of hydrological perturbations (section  5.2.2.2). Using simulations with the CESM model, Stevenson et  al. (2016) showed that the hydrologic expression of volcanic eruptions is easily mistaken for ENSO teleconnections. Should this apply to the real world, it would call into question the use of extratropical tree rings (and the reconstructions based on them) to diagnose the link between ENSO and explosive volcanism. To sidestep these issues, Dee et  al. (2020) carried a detailed analysis on coral samples from Palmyra Island (Cobb et  al., 2003; Figure  5.1, Figure  5.2), including a new sequence comprising 319 years of continuous data at submonthly resolution. High‐precision U/Th dates yield age controls such that chronological errors are effectively reduced to zero. The record encompasses the largest eruptions of the past millennium, such as the 1257 (Samalas) and the 1452/3 (Kuwae) eruptions. Yet superposed epoch analysis reveals no consistent response of ENSO to tropical volcanic forcing in this record: while there is evidence for warming in the year following some large eruptions, the overall response cannot be distinguished from natural ENSO variability at the 95%

confidence level. The topic is likely to remain debated until more high‐resolution, well‐dated records of tropical Pacific climate become available. 5.4.4.3. Constraining Low-Frequency Variability The past millennium also offers spectral constraints on how climate models simulate fluctuations in tropical Pacific SST. Ault et  al. (2013b) compared simulated Niño‐3.4 indices in CMIP5‐era general circulation models to the multiproxy reconstructions of Emile‐Geay et al. (2013b), as well as a null hypothesis generated from “multivariate red noise”’ generated by a linear inverse model (LIM) fit to instrumental observations from 1960 to 2000 (Newman et al., 2011a, 2011b). The GCM simulations followed the CMIP5‐PMIP3 past1000 protocol (Schmidt et al., 2012). They found that on decadal to multidecadal timescales, variability in the reconstructed and simulated Niño‐3.4 is consistent with the null hypothesis that it arises from the stationary, damped tropical ocean‐atmosphere dynamics captured by the LIM. On centennial and longer timescales, both forced simulations and the paleoclimate reconstructions displayed significantly stronger variability than the LIM (Figure 5.5). However, the simula-

PAST ENSO VARIABILITY  103

tions and the reconstructions diverged: at decadal to centennial scales, cross‐wavelet analysis shows that in the reconstructions, Niño‐3.4 is out of phase with radiative forcing, consistent with the thermostat mechanism (Clement et al., 1996; Seager et al., 2019). On the other hand, PMIP3 models show Niño‐3.4 variability in phase with radiative forcing, suggestive of a direct, thermodynamic response (Held & Soden, 2006). As in the case of volcanism, the discrepancy lies partly in the considerable uncertainties in reconstructions of both ENSO and forcings. Nonetheless, the exercise illustrates how ENSO proxies of the past millennium may be used to constrain GCM behavior, generating new research hypotheses (Braconnot et al., 2012). Finally, another advantage of the past millennium over past intervals is the widespread existence of documentary evidence, which has been mined to reconstruct ENSO phases over recent centuries (Gergis & Fowler, 2005; Gergis et  al., 2006; Ortlieb & Macharé, 1993; Ortlieb et al., 2000; Quinn, 1992). A challenge of such documents that they only provide semiquantitative information (e.g. ENSO is “strong,” “very strong,” or “extreme”), and different sources can be difficult to reconcile with each other and with multiproxy reconstructions. 5.4.5. Anthropogenic Era Whether ENSO properties are affected by anthropogenic global warming remains a fundamental question in climate science (e.g. Cai et al., 2018; Collins, 2005; Collins et al., 2010; Timmermann et al., 1999; chapter 13). The application of paleo‐ENSO reconstructions to the detection of anthropogenic signals in ENSO properties has received considerable attention in recent years, with a number of studies lending support to a recent intensification of ENSO impacts. In attempting to address this with paleoclimate datasets, exceptionally long records that span many centuries of high‐resolution data are required, and such records are still relatively rare. While early efforts focused on reconstructing ENSO variance changes in multiproxy syntheses of ENSO variability (Mann et  al., 2000), more recent work stresses that relatively small errors (i.e. ± 1 y) in the absolute age models of individual proxy records can create a systemic bias towards low variance in time intervals prior to the 20th century (Comboul et al., 2014). In an effort to circumvent this effect, McGregor et al. (2013b) analyzed the evolution of ENSO variance across 13 different ENSO reconstructions covering the past few centuries, to argue that ENSO variance was significantly higher during the 1979– 2009 period versus the 1600–1900 period. In the same year, Li et al. (2013) used a 700‐year‐long reconstruction of ENSO variance based on over 2,000 individual ENSO‐ sensitive tree rings time series to argue much the same.

Long records of ENSO variability in the central tropical Pacific based on monthly‐resolved coral oxygen isotopes support these findings, detecting a statistically significant increase in the interannual variance of coral oxygen isotopes in the late 20th century versus the preindustrial era (Cobb et al., 2013; Grothe et al., 2019). 5.5. DISCUSSION We have shown how the paleoclimate record may be used to test theories of ENSO and constrain its representation in climate models. Paleo observations provide information that no other source can, sampling ENSO behavior across different base states, subjected to many types and intensities of external forcing, and providing a much longer statistical sample than afforded by the instrumental record. Increasingly, they are becoming an integral part of ENSO theory, offering a key out‐of‐sample test of model predictions (Schmidt, 2010). However, as this chapter made clear, the relative paucity of records, and the indirect nature of their relation to ENSO state, means that testing ENSO theories involves many auxiliary hypotheses, making such tests, at times, ambiguous. Indeed, studying ENSO via the prism of paleoclimate observations evokes the parable of the blind men and the elephant. Like the blind men, proxies only experience part of the whole and give seemingly conflicting accounts of this experience. Yet each experience is valid in its own way, and all experiences must be pieced together to understand the whole (here, ENSO). As seen throughout this review, different proxy types filter ENSO signals differently, or the same proxy system may simply record ENSO in different places (e.g. land vs. ocean; central vs. eastern Pacific). A key challenge is thus to best combine this information in a way that leverages what is already understood about ENSO. To achieve this goal, paleo observations must clear a higher bar. We see four major ways to achieve this: Ground‐truthing. Not all proxies are created equal. Many can be instrumentally calibrated, and in most cases, site monitoring may vastly improve a proxy’s quantitative understanding. Building upon the example of Laguna Pallcacocha and its recent reinterpretation, we recommend investments to better characterize iconic records. Furthermore, a majority of the proxies surveyed herein involve the oxygen isotope composition of biocarbonates. Improving their interpretation requires investments in both remotely sensed and in situ observing systems for oxygen isotopes in the tropical atmosphere and upper ocean. An additional motivation for such observations is that the δ18O of water vapor or seawater is a state variable (e.g. Risi et al., 2012), providing a unique picture of the tropical hydrological cycle (Conroy et al., 2014, 2017), particularly convective processes (Bony

104  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

et al., 2008; Hu et al., 2018; Risi et al., 2008). Investments in observing systems geared towards stable water isotopes would therefore have the double benefit of providing an independent constraint on tropical atmospheric convection, as well as reducing uncertainties in paleoclimate reconstructions relying on such isotopes. A better understanding of δ18O variations in seawater will thus have marked benefits for improving ENSO reconstructions based on coral δ18O measurements (Russon et  al., 2013; Stevenson et al., 2013, 2018). Process modeling. A complementary strategy is to develop models to understand proxy signals and biases; such is the object of proxy system modeling (Dee et al., 2015; Evans et  al., 2013). PSMs encode all known processes associated with a given proxy: a mathematically concise way of conveying what the paleoclimate community has come to understand about a proxy. In cases where competing models exist, this diversity may be used to address structural and parametric uncertainties. PSMs are becoming an integral part of data assimilation (Dee et  al., 2016), and they may also be used to devise optimal sampling strategies (Comboul et al., 2015). Replication is the basis of science. While there is often a drive to sample underexplored regions, repeated sampling using complementary proxies provides important cross‐checks on the validity of proxy interpertation. Pseudoproxy experiments (Smerdon, 2011) can help determine levels of replication needed to achieve the required signal/noise ratios for certain applications, including variance estimation (Russon et al., 2014, 2015) and sampling design (Comboul et al., 2015). Physical nulls. Beyond pseudoproxy experiments, much can be gleaned from physically motivated null hypotheses. These may originate from stationary models fit to modern observations (e.g. Newman et  al., 2011b) or from long, unforced simulations of climate models of varying degrees of complexity. The study of Wittenberg (2009), based on a 2,000‐year‐long simulation with the GFDL CM2.1 model, has durably shifted the baseline for detecting forced changes in ENSO. We recommend carrying out more such integrations, particularly with isotope‐ enabled GCMs (Brady et al., 2019). Similarly, Newman et al. (2011a) used a LIM to argue that recent changes in the frequency of CP events were consistent with stationary statistics. We recommend that more paleoclimate studies make use of such resources to generate physically plausible null hypotheses about ENSO’s nonstationarity. The emerging picture is that of a triad where theory, observations (both instrumental and paleoclimate), and models are inextricably linked. Climate models have helped, and will continue to help, refine the interpretation of proxies or understand the statistical consequences of nonstationary paleo‐observing networks, and provide physically based null hypotheses. Paleo observations, in

turn, provide key constraints on ENSO behavior and processes unavailable by other means. The scientific understanding of many proxies has advanced to the point that they can provide quantitative constraints on out‐of‐ sample model predictions and highlight areas of potential improvement. Nonetheless, the spatiotemporal sparsity and uncertainties of the paleo‐ENSO observational record prevents definitive conclusions in a majority of the cases reviewed here, highlighting the need for continued sampling and ground‐truthing efforts, as outlined above. With these caveats in mind, the existing observations are consistent with the view that variations in ENSO amplitude and frequency arise primarily from processes internal to the climate system; that is, one cannot presently refute the null hypothesis that ENSO is insensitive to natural forcings. Indeed, the view from the Holocene is one where ENSO activity displays marked millennial‐ scale changes (e.g. low activity 3–5 ky BP) without apparent relation to known forcings. This raises the bar against which to gauge other changes in the paleoclimate and modern record. An exception is changes in the mean state, forced primarily by atmospheric CO2: emerging evidence suggests that such changes during the anthropogenic era were large enough to generate changes in ENSO that stand out in the context of the Holocene and were only seen sporadically in the past 3 million years. The current observations from the last glacial maximum hint at a mirror image (subdued ENSO with lower atmospheric CO2), but this view may come to change as multiple independent lines of evidence (notably, seasonally resolved observations) are brought to bear on the question. ACKNOWLEDGMENTS The authors thank the editors, Michael McPhaden, Agus Santoso, and Wenju Cai, for their patience. JEG acknowledges NSF Grant AGS‐1003818; JEG and FZ acknowledge funding from NOAA grant NA18OAR4310426. KMC acknowledges support from NSF Awards 1836645, 1903640, and 2002458. ME acknowledges funding from JPI-Belmont Forum project PACMEDY (ANR-15JCLI-0003-01). JEC acknowledges support from NSF awards 1851587 and 1829613. REFERENCES Abram, N. J., M. K. Gagan, J. E. Cole, W. S. Hantoro, & M. Mudelsee (2008). Recent intensification of tropical climate variability in the Indian Ocean. Nature Geoscience, 1, 849–853. doi:10.1038/ngeo357 Adams, J., M. Mann, & C. Ammann (2003). Proxy evidence for an El Niño‐like response to volcanic forcing. Nature, 426, 274–278. doi:10.1038/nature02101

PAST ENSO VARIABILITY  105 Aharon, P. (1991). Recorders of reef environment histories: Stable isotopes in corals, giant clams, and calcareous algae. Coral Reefs, 10(2). 71–90. doi: 10.1007/BF00571826 Alibert, C., & L. Kinsley (2008). A 170‐year Sr/Ca and Ba/Ca coral record from the western Pacific warm pool: 2. A window into variability of the New Ireland Coastal Undercurrent. Journal of Geophysical Research, 113(C6). 1–10. doi: 10.1029/2007JC004263 Alibert, C., & M. T. McCulloch (1997). Strontium/calcium ratios in modern porites corals From the Great Barrier Reef as a proxy for sea surface temperature: Calibration of the thermometer and monitoring of ENSO. Paleoceanography, 12(3). 345–363. doi: 10.1029/97PA00318 An, S.‐I., & H. Bong (2018). Feedback process responsible for the suppression of ENSO activity during the mid‐Holocene. Theoretical and Applied Climatology, 132(3‐4). 779–790. doi: 10.1007/s00704‐017‐2117‐6 An, S.‐I., & J. Choi (2014). Mid‐Holocene tropical Pacific climate state, annual cycle, and ENSO in PMIP2 and PMIP3. Climate Dynamics, 43(3‐4). 957–970. doi: 10.1007/s00382‐013‐1880‐z An, S.‐I., Y.‐G. Ham, J.‐S. Kug, A. Timmermann, J. Choi, & I.‐S. Kang (2010). The inverse effect of annual‐mean state and annual‐cycle changes on ENSO. Journal of Climate, 23(5). 1095–1110. doi: 10.1175/2009JCLI2895.1 Ariztegui, D., P. Bösch, & E. Davaud (2007). Dominant ENSO frequencies during the Little Ice Age in Northern Patagonia: The varved record of proglacial Lago Frías, Argentina. Quaternary International, 161(1). 46–55. doi: 10.1016/ j.quaint.2006.10.022 Asami, R., T. Yamada, Y. Iryu, T. M. Quinn, C. P. Meyer, & G. Pauley (2005). Interannual and decadal variability of the western Pacific sea surface condition for the years 1787‐2000: Reconstruction based on stable isotope record from a Guam coral. Journal of Geophysical Research, 110. doi: 10.1029/ 2004JC002555 Ault, T. R., J. E. Cole, J. T. Overpeck, G. T. Pederson, S. St. George, B. Otto‐Bliesner, et  al. (2013a). The continuum of hydroclimate variability in western North America during the last millennium. Journal of Climate, 26(16). 5863–5878. doi: 10.1175/ JCLI‐D‐11‐00732.1 Ault, T. R., C. Deser, M. Newman, & J. Emile‐Geay (2013b). Characterizing decadal to centennial variability in the equatorial pacific during the last millennium. Geophysical Research Letters, 40, 3450–3456. doi: 10.1002/grl.50647 Bagnato, S., B. K. Linsley, S. S. Howe, G. M. Wellington, & J. Salinger (2004). Evaluating the use of the massive coral Diploastrea heliopora for paleoclimate reconstruction, Paleoceanography, 19. doi: 10.1029/ 2003PA000935 Bagnato, S., B. K. Linsley, S. S. Howe, G. M. Wellington, & J. Salinger (2005). Coral oxygen isotope records of interdecadal climate variations in the South Pacific Convergence Zone region. Geochemistry, Geophysics, Geosystems, 6. doi: 10.1029/2004GC000879 Baker, J. C. A., S. F. P. Hunt, S. J. Clerici, R. J. Newton, S. H. Bottrell, M. J. Leng, et  al. (2015). Oxygen isotopes in tree rings show good coherence between species and sites in Bolivia. Global and Planetary Change, 133, 298–308. doi: 10.1016/j.gloplacha.2015.09.008 Barr, C., J. Tibby, M. J. Leng, J. J. Tyler, A. C. G. Henderson, J. T. Overpeck, et al. (2019). Holocene el niño–southern oscilla-

tion variability reflected in subtropical australian precipitation. Scientific Reports, 9(1), 1627. doi: 10.1038/ s41598‐019‐38626‐3 Beaufort, L., & M. Grelaud (2017). A 2700‐year record of ENSO and PDO variability from the Californian margin based on coccolithophore assemblages and calcification. Progress in Earth and Planetary Science, 4(1), 5. doi: 10. 1186/ s40645‐017‐0123‐z Beckmann, A., & R. Doescher (1997). A method for improved representation of dense water spreading over topography in geopotential‐coordinate models. J. Phys. Oceanogr., 27(581–591). Bellenger, H., E. Guilyardi, J. Leloup, M. Lengaigne, & J. Vialard (2014). ENSO representation in climate models: from CMIP3 to CMIP5. Climate Dynamics, 42(7‐8), 1999–2018. doi: 10.1007/s00382‐013‐1783‐z Bhend, J., J. Franke, D. Folini, M. Wild, & S. Brönnimann (2012). An ensemble‐based approach to climate reconstructions. Climate of the Past, 8(3). 963–976. doi: 10.5194/cp‐8‐963‐2012 Boiseau, M., A. Juillet‐Leclerc, P. Yiou, B. Salvat, P. Isdale, & M. Guillaume (1998). Atmospheric and oceanic evidences of El Niño‐Southern Oscillation events in the south central Pacific Ocean from coral stable isotopic records over the last 137 years. Paleoceanography, 13, 671–685. doi: 10.1029/ 98PA02502 Bonham, S. G., A. M. Haywood, D. J. Lunt, M. Collins, & U. Salzmann (2009). El Niño‐Southern Oscillation, Pliocene climate and equifinality. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 367(1886). 127–156. doi: 10.1098/rsta.2008.0212 Bony, S., C. Risi, & F. Vimeux (2008). Influence of convective processes on the isotopic composition (δ18O and δD) of precipitation and water vapor in the tropics: 1. Radiative‐convective equilibrium and Tropical Ocean–Global Atmos phere–Coupled Ocean‐Atmosphere Response Experiment (TOGA‐COARE) simulations. Journal of Geophysical Research: Atmospheres, 113, D19,305. doi: 10.1029/2008JD009942 Braconnot, P., S. P. Harrison, M. Kageyama, P. J. Bartlein, V. Masson‐Delmotte, A. Abe‐Ouchi, et  al. (2012). Evaluation of climate models using palaeoclimatic data. Nature Clim. Change, 2(6). 417–424. doi: 10.1038/nclimate1456 Brady, E., S. Stevenson, D. Bailey, Z. Liu, D. Noone, J. Nusbaumer, et al. (2019). The connected isotopic water cycle in the Community Earth System Model version 1. Journal of Advances in Modeling Earth Systems, 11, 2547–2566. doi:10.1029/2019MS001663 Brady, E. C., B. L. Otto‐Bliesner, J. E. Kay, & N. Rosenbloom (2012). Sensitivity to glacial forcing in the CCSM4. Journal of Climate, 26(6). 1901–1925. doi: 10.1175/JCLI‐D‐11‐00416.1 Brienen, R. J. W., G. Helle, T. L. Pons, J.‐L. Guyot, & M. Gloor (2012). Oxygen isotopes in tree rings are a good proxy for Amazon precipitation and El Nino‐Southern Oscillation variability, Proceedings of the National Academy of Sciences of the United States of America, 109(42), 16,957–16,962. doi: 10.1073/pnas.1205977109 Bunge, L., & A. J. Clarke (2009). A Verified Estimation of the El Niño Index Nino‐3.4 since 1877. Journal of Climate, 22, 3979. doi: 10.1175/ 2009JCLI2724.1 Cai, W., G. Wang, B. Dewitte, L. Wu, A. Santoso, K. Takahashi, et al. (2018). Increased variability of Eastern Pacific El Niño

106  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE under greenhouse warming. Nature, 564(7735). 201–206. doi: 10.1038/s41586‐018‐0776‐9 Carré, M., J. P. Sachs, A. J. Schauer, W. E. Rodríguez, & F. C. Ramos (2013). Reconstructing El Niño‐Southern Oscillation activity and ocean temperature seasonality from short‐lived marine mollusk shells from Peru. Palaeogeography, Palaeoclimatology, Palaeoecology, 371(0). 45–53. doi: 10. 1016/j.palaeo.2012.12.014 Carré, M., J. P. Sachs, S. Purca, A. J. Schauer, P. Braconnot, R. A. Falcón, et al. (2014). Holocene history of ENSO variance and asymmetry in the eastern tropical Pacific. Science, 345(6200). 1045–1048. doi: 10.1126/science.1252220 Chakraborty, S., & R. Ramesh (1998). Stable isotope variations in a coral (Favia speciosa) from the Gulf of Kutch during 1948‐1989 AD: Environmental implications. Earth Planet. Sci., 107(4). 331–341. Chang, P., B. Wang, T. Li, & L. Ji (1994). Interactions between the seasonal cycle and the Southern Oscillation: Frequency entrainment and chaos in a coupled ocean‐atmosphere model. Geophys. Res. Lett., 21, 2817–2820. doi: 10.1029/94GL02759 Charles, C. D., D. E. Hunter, & R. G. Fairbanks (1997). Interaction between the ENSO and the Asian Monsoon in a coral record of tropical climate. Science, 277(5328). 925–928. doi: 10.1126/science.277.5328.925 Charles, C. D., K. Cobb, M. D. Moore, & R. G. Fairbanks (2003). Monsoon‐tropical ocean interaction in a network of coral records spanning the 20th century. Marine Geology, 201, 207–222. doi: 10.1016/S0025‐3227(03) 00217‐2 Chen, S., S. S. Hoffmann, D. C. Lund, K. M. Cobb, J. Emile‐ Geay, & J. F. Adkins (2016). A high‐resolution speleothem record of western equatorial Pacific rainfall: Implications for holocene ENSO evolution. Earth and Planetary Science Letters, 442, 61–71. doi: 10.1016/j.epsl.2016.02.050 Chiang, J. C. (2009). The tropics in paleoclimate. Annual Review of Earth and Planetary Sciences, 37(1). 263–297. doi: 10.1146/ annurev.earth.031208. 100217 Chiang, J. C. H., Y. Fang, & P. Chang (2009). Pacific climate change and ENSO activity in the mid‐Holocene. Journal of Climate, 22(4). 923–939. doi: 10.1175/2008JCLI2644.1 Clarke, H., J. P. D’Olivo, J. Falter, J. Zinke, R. Lowe, & M. McCulloch (2017). Differential response of corals to regional mass‐warming events as evident from skeletal Sr/Ca and Mg/ Ca ratios. Geochemistry Geophysics Geosystems, 18(5), 1794– 1809. doi: 10.1002/2016GC006788 Clement, A., & M. A. Cane (1999). A role for the tropical Pacific coupled ocean‐atmosphere system on Milankovitch and millenial timescales. Part I: A modeling study of tropical Pacific variability. In P. Clark & R. Webb (Eds.), Mechanisms of millennial‐scale global climate change (Vol. 29, pp. 363– 372), American Geophysical Union. Clement, A. C., R. Seager, M. A. Cane, & S. E. Zebiak (1996). An ocean dynamical thermostat, J. Clim., 9(9). 2190–2196. Clement, A. C., R. Seager, & M. A. Cane (1999). Orbital controls on the El Niño/Southern Oscillation and the tropical climate. Paleoceanography, 14(4), 441–456. Clement, A. C., R. Seager, & M. A. Cane (2000). Suppression of El Niño during the mid‐Holocene by changes in the Earth’s orbit. Paleoceanography, 15(6), 731–737.

Cobb, K. M., C. D. Charles, & D. E. Hunter (2001). A central tropical Pacific coral demonstrates Pacific, Indian, and Atlantic decadal climate connections. Geophys. Res. Lett., 28, 2209–2212. doi: 10.1029/2001GL012919 Cobb, K. M., C. D. Charles, H. Cheng, & R. L. Edwards (2003). El Niño/Southern Oscillation and tropical Pacific climate during the last millennium. Nature, 424, 271–276. Cobb, K. M., N. Westphal, H. R. Sayani, J. T. Watson, E. Di Lorenzo, H. Cheng, et al. (2013). Highly variable El Niño– Southern Oscillation throughout the Holocene. Science, 339(6115), 67. doi: 10.1126/science.1228246 Cole, J. E., R. G. Fairbanks, & G. T. Shen (1993). Recent variability in the Southern Oscillation: Isotopic results from a Tarawa Atoll coral. Science, 260, 1790–1793. doi: 10.1126/ science.260.5115.1790 Cole, J. E., R. B. Dunbar, T. R. McClanahan, & N. A. Muthiga (2000). Tropical Pacific forcing of decadal SST variability in the western Indian Ocean over the past two centuries, Science, 287, 617–620. doi: 10.1126/ science.287.5453.617 Collins, M. (2005). El Niño‐ or La Niña‐like climate change? Climate Dynamics, 24(1), 89–104. Collins, M., S.‐I. An, W. Cai, A. Ganachaud, E. Guilyardi, F.‐F. Jin, et  al. (2010). The impact of global warming on the tropical Pacific Ocean and El Niño. Nature Geosci, 3(6), 391– 397. doi: 10.1038/ngeo868 Comboul, M., J. Emile‐Geay, M. N. Evans, N. Mirnateghi, K. M. Cobb, & D. M. Thompson (2014). A probabilistic model of chronological errors in layer‐counted climate proxies: Applications to annually banded coral archives. Climate of the Past, 10(2), 825–841. doi: 10.5194/cp‐10‐825‐2014 Comboul, M., J. Emile‐Geay, G. J. Hakim, & M. N. Evans (2015). Paleoclimate sampling as a sensor placement problem. Journal of Climate, 28, 7717–7740. doi: 10.1175/JCLI‐D‐14‐00802.1 Conroy, J. L., J. T. Overpeck, J. E. Cole, T. M. Shanahan, & M. Steinitz‐Kannan (2008). Holocene changes in eastern tropical Pacific climate inferred from a Galápagos lake sediment record. Quaternary Science Reviews, 27(11–12). 1166 –1180. doi: 10.1016/j.quascirev.2008.02.015 Conroy, J. L., A. Restrepo, J. T. Overpeck, M. Steinitz‐Kannan, J. E. Cole, M. B. Bush, and P. A. Colinvaux (2009). Unprecedented recent warming of surface temperatures in the eastern tropical Pacific Ocean. Nature Geosci, 2(1). 46–50. doi: 10.1038/ngeo390 Conroy, J. L., K. M. Cobb, J. Lynch‐Stieglitz, & P. J. Polissar (2014). Constraints on the salinity–oxygen isotope relationship in the central tropical Pacific ocean. Marine Chemistry, 161(0). 26–33. doi: 10.1016/j.marchem.2014. 02.001 Conroy, J. L., D. M. Thompson, K. M. Cobb, D. Noone, S. Rea, & A. N. Legrande (2017). Spatiotemporal variability in the 18o‐salinity relationship of seawater across the tropical Pacific ocean. Paleoceanography, 32(5), 484–497. doi: 10.1002/2016PA003073 Cook, E., C. Woodhouse, C. Eakin, D. Meko, & D. Stahle (2004). Long‐term aridity changes in the western United States. Science, 306(5698), 1015–1018. Cook, E. R., K. R. Briffa, D. M. Meko, D. A. Graybill, & G. Funkhouser (1995). The “segment length curse” in long tree‐ ring chronology development for palaeoclimatic studies. The Holocene, 5(2), 229–237. doi: 10.1177/ 095968369500500211

PAST ENSO VARIABILITY  107 Cook, E. R., K. J. Anchukaitis, B. M. Buckley, R. D. D’Arrigo, G. C. Jacoby, & W. E. Wright (2010). Asian monsoon failure and megadrought during the last millennium. Science, 328(5977), 486–489. doi: 10.1126/science.1185188 Cook, K. H., & I. M. Held (1988). Stationary waves of the Ice Age climate. J. Climate, 1, 807–819. Corrège, T. (2006). Sea surface temperature and salinity reconstruction from coral geochemical tracers. Palaeogeography, Palaeoclimatology, Palaeoecology, 232(2–4), 408–428. doi: http://dx.doi.org/10.1016/j.palaeo.2005.10.014 Corrège, T., T. Delcroix, J. Récy, W. Beck, G. Cabioch, & F. Le Cornec (2000). Evidence for stronger El Niño‐Southern Oscillation (ENSO) events in a mid‐Holocene massive coral. Paleoceanography, 15, 465–470. doi: 10.1029/ 1999PA000409 Damassa, T. D., J. E. Cole, H. R. Barnett, T. R. Ault, & T. R. McClanahan (2006). Enhanced multidecadal climate variability in the seventeenth century from coral isotope records in the western Indian Ocean. Paleoceanography, 21, PA2016. doi: 10.1029/2005PA001217 D’Arrigo, R. D., R. Wilson, & A. Tudhope (2009). The impact of volcanic forcing on tropical temperatures during the past four centuries. Nature Geosci, 2(1). 51–56. Dassié, E. P., B. K. Linsley, T. Corrège, H. C. Wu, G. M. Lemley, S. Howe, & G. Cabioch (2014). A Fiji multi‐coral 18o composite approach to obtaining a more accurate reconstruction of the last two‐centuries of the ocean‐climate variability in the South Pacific convergence zone region. Paleoceanography, 29(12), 1196–1213. doi: 10.1002/2013PA002591 Dätwyler, C., N. J. Abram, M. Grosjean, E. R. Wahl, & R. Neukom (2019). El Niño–Southern Oscillation variability, teleconnection changes and responses to large volcanic eruptions since AD 1000. International Journal of Climatology, 39(5), 2711–2724. doi: 10.1002/joc.5983 Dee, S., J. Emile‐Geay, M. N. Evans, A. Allam, E. J. Steig, & D. M. Thompson (2015). PRYSM: An open‐source framework for PRoxY System Modeling, with applications to oxygen‐ isotope systems. Journal of Advances in Modeling Earth Systems, 7(3), 1220–1247. doi: 10.1002/2015MS000447 Dee, S. G., N. J. Steiger, J. Emile‐Geay, & G. J. Hakim (2016). On the utility of proxy system models for estimating climate states over the common era. Journal of Advances in Modeling Earth Systems, 8. doi: 10.1002/ 2016MS000677 Dee, S. G., L. A. Parsons, G. R. Loope, J. T. Overpeck, T. R. Ault, & J. Emile‐Geay (2017). Improved spectral comparisons of paleoclimate models and observations via proxy system modeling: Implications for multi‐decadal variability. Earth and Planetary Science Letters, 476(Supplement C), 34–46. doi: 10.1016/j.epsl.2017.07.036 Dee, S. G., K. M. Cobb, J. Emile‐Geay, T. R. Ault, R. L. Edwards, H. Cheng, & C. D. Charles (2020). No consistent ENSO response to volcanic forcing over the last millennium, Science, 367(6485), 1477, doi: 10.1126/science.aax2000. DeLong, K. L., T. M. Quinn, F. W. Taylor, K. Lin, & C.‐C. Shen (2012). Sea surface temperature variability in the southwest tropical Pacific since AD 1649. Nature Clim. Change, 2, 799– 804. doi: 10.1038/nclimate1583 DeLong, K. L., T. M. Quinn, F. W. Taylor, C.‐C. Shen, & K. Lin (2013). Improving coral‐base paleoclimate reconstructions by

replicating 350 years of coral Sr/Ca variations. Palaeogeography, Palaeoclimatology, Palaeoecology, 373, 6 – 24. doi: 10.1016/j.palaeo.2012.08.019 DiNezio, P. N., A. C. Clement, G. A. Vecchi, B. J. Soden, B. P. Kirtman, & S.‐K. Lee (2009). Climate response of the equatorial Pacific to global warming. J. Clim., 22(18), 4873– 4892. doi: 10.1175/2009JCLI2982.1 DiNezio, P. N., A. Clement, G. A. Vecchi, B. Soden, A. J. Broccoli, B. L. Otto‐Bliesner, & P. Braconnot (2011). The response of the walker circulation to last glacial maximum forcing: Implications for detection in proxies. Paleoceanography, 26(3). doi:10.1029/2010PA002083 D’Olivo, J. P., & M. T. McCulloch (2017). Response of coral calcification and calcifying fluid composition to thermally induced bleaching stress. Scientific Reports, 7. doi: 10.1038/ s41598‐017‐02306‐x D’Olivo, J. P., L. Georgiou, J. Falter, T. M. DeCarlo, X. Irigoien, C. R. Voolstra, et  al. (2019). Long‐term impacts of the 1997‐1998 bleaching event on the growth and resilience of massive porites corals from the central Red Sea. Geochemistry Geophysics Geosystems, 20(6). 2936–2954. doi: 10.1029/ 2019GC008312 Driscoll, R., M. Elliot, T. Russon, K. Welsh, Y. Yokoyama, & A. Tudhope (2014). ENSO reconstructions over the past 60 ka using giant clams (Tridacna sp.) from Papua New Guinea. Geophysical Research Letters, 41, 6819–6825. doi: 10.1002/2014GL061446 Druffel, E. R. M., & S. Griffin (1999). Variability of surface ocean radiocarbon and stable isotopes in the southwestern Pacific. J. Geophys. Res., 104, 23,607–23,614. doi: 10.1029/ 1999JC900212 Dunbar, R. B., G. M. Wellington, M. W. Colgan, & P. W. Glynn (1994). Eastern Pacific sea surface temperature since 1600 A.D.: The δ18O record or climate variability in Galápagos corals. Paleoceanography, 9, 291–316. doi: 10.1029/93PA03501 Duprey, N., C. E. Lazareth, T. Corrège, F. Le Cornec, C. Maes, N. Pujol, et  al. (2012). Early mid‐Holocene SST variability and surface‐ocean water balance in the southwest Pacific. Paleoceanography, 27(4). PA4207. doi: 10.1029/ 2012PA002350 Dutton, A., & K. Lambeck (2012). Ice volume and sea level during the last interglacial. Science, 337(6091). 216. doi: 10.1126/science.1205749 Ebisuzaki, W. (1997). A method to estimate the statistical significance of a correlation when the data are serially correlated. Journal of Climate, 10(9). 2147–2153. doi: 10.1175/ 1520‐0442(1997)010-2147:AMTETS-2.0.CO;2 Emile‐Geay, J., & M. Tingley (2016). Inferring climate variability from nonlinear proxies: Application to palaeo‐ENSO studies. Climate of the Past, 12(1), 31–50. doi: 10.5194/ cp‐12‐31‐2016 Emile‐Geay, J., D. Guillot, K. Cobb, J. Cole, T. Corrège, A. Tudhope, & B. Rajaratnam (2012). Four centuries of tropical Pacific sea‐surface temperature from coral archives (invited). In Fall Meeting, AGU, San Francisco, Calif., 3‐7 Dec, Vol. Abstract OS52B‐01. Emile‐Geay, J., K. Cobb, M. Mann, & A. T. Wittenberg (2013a). Estimating central equatorial Pacific SST variability over the

108  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE past millennium. Part 1: Methodology and validation. J. Clim., 26, 2302–2328. doi: 10.1175/ JCLI‐D‐11‐00510.1 Emile‐Geay, J., K. Cobb, M. Mann, & A. T. Wittenberg (2013b). Estimating central equatorial Pacific SST variability over the past millennium. Part 2: Reconstructions and implications. J. Clim., 26, 2329–2352. doi: 10.1175/ JCLI‐D‐11‐00511.1 Emile‐Geay, J., M. Comboul, D. Guillot, K. Cobb, M. Evans, T. Corrège, & B. Rajaratnam (2013c). Multiproxy reconstructions of ENSO: Progress and challenges, in Fall Meeting, AGU, San Francisco, Calif., 9‐13 Dec, Vol. Abstract PP23D‐08. Emile‐Geay, J., K. M. Cobb, M. Carre, P. Braconnot, J. Leloup, Y. Zhou, et al. (2016). Links between tropical Pacific seasonal, interannual and orbital variability during the Holocene. Nature Geosci, 9(2), 168–173. doi: 10.1038/ngeo2608 Enmar, R., M. Stein, M. Bar‐Matthews, E. Sass, A. Katz, & B. Lazar (2000). Diagenesis in live corals from the Gulf of Aqaba: I. The effect on paleo‐oceanography tracers. Geochimica et Cosmochimica Acta, 64(18), 3123–3132. doi: 10.1016/S0016‐7037(00)00417‐8 Epstein, S., R. Buchsbaum, H. A. Lowenstam, & H. C. Urey (1953). Revised carbonate‐water isotopic temperature scale. Geological Society of America Bulletin, 64(11). 1315–1326. doi: 10.1130/0016‐7606(1953)64(1315:RCITS)2. 0.CO;2 Evans, M., R. Fairbanks, & J. Rubenstone (1999). The thermal oceanographic signal of El Niño reconstructed from a Kiritimati Island coral. Journal of Geophysical Research (Oceans), 104, 13,409–13,422. doi: 10.1029/1999JC900001 Evans, M. N., & D. P. Schrag (2004). A stable isotope‐based approach to tropical dendroclimatology. Geochimica et Cosmochimica Acta, 68(16). 3295–3305. doi: https://doi. org/10.1016/j.gca. 2004.01.006 Evans, M. N., A. Kaplan, & M. A. Cane (1998). Optimal sites for coral‐based reconstruction of global sea surface temperature. Paleoceanography, 13, 502–516. doi: 10.1029/98PA02132 Evans, M. N., S. E. Tolwinski‐Ward, D. M. Thompson, & K. J. Anchukaitis (2013). Applications of proxy system modeling in high resolution paleoclimatology. Quaternary Science Reviews, 76(0), 16–28. doi: 10.1016/j. quascirev.2013.05.024 Fairbanks, R. G., M. N. Evans, J. L. Rubenstone, R. A. Mortlock, K. Broad, M. D. Moore, & C. D. Charles (1997). Evaluating climate indices and their geochemical proxies measured in corals. Coral Reefs, 16(1), S93–S100. doi: 10.1007/s003380050245 Fedorov, A. V., & S. G. Philander (2001). A stability analysis of tropical ocean‐atmosphere interactions: Bridging measurements and theory for El Niño. J. Climate, 14, 3086–3101. Fedorov, A. V., P. S. Dekens, M. McCarthy, A. C. Ravelo, P. B. deMenocal, M. Barreiro, et al. (2006). The Pliocene paradox (mechanisms for a permanent El Niño). Science, 312(5779), 1485. doi: 10.1126/science.1122666 Fedorov, A. V., C. M. Brierley, & K. Emanuel (2010). Tropical cyclones and permanent El Niño in the early Pliocene epoch. Nature, 463(7284), 1066–1070. doi: 10.1038/nature08831 Felis, T., J. Pätzold, Y. Loya, M. Fine, A. H. Nawar, & G. Wefer (2000). A coral oxygen isotope record from the northern Red Sea documenting NAO, ENSO, and North Pacific teleconnections on Middle East climate variability since the year

1750. Paleoceanography, 15, 679–694. doi: 10.1029/ 1999PA000477 Felis, T., U. Merkel, R. Asami, P. Deschamps, E. C. Hathorne, M. Kölling, et al. (2012). Pronounced interannual variability in tropical south Pacific temperatures during Heinrich stadial 1. Nature Communications, 3(1), 965. doi: 10.1038/ ncomms1973 Ford, H. L., A. C. Ravelo, & P. J. Polissar (2015). Reduced El Niño‐Southern Oscillation during the last glacial maximum. Science, 347(6219), 255. doi: 10.1126/science.1258437 Foukal, P., C. Frohlich, H. Spruit, & T. M. L. Wigley (2006). Variations in solar luminosity and their effect on the Earth’s climate. Nature, 443, 161–166. doi: 10.1038/nature05072 Franke, J., S. Brönnimann, J. Bhend, & Y. Brugnara (2017). A monthly global paleo‐reanalysis of the atmosphere from 1600 to 2005 for studying past climatic variations. Scientific Data, 4, 170,076. Frappier, A., D. Sahagian, L. A. González, & S. J. Carpenter (2002). El Niño events recorded by stalagmite carbon isotopes. Science, 298(5593), 565. doi: 10.1126/science.1076446 Freund, M. B., B. J. Henley, D. J. Karoly, H. V. McGregor, N. J. Abram, & D. Dommenget (2019). Higher frequency of central Pacific El Niño events in recent decades relative to past centuries. Nature Geoscience. doi: 10.1038/ s41561‐019‐0353‐3 Fritts, H. C. (1976). Tree rings and climate. Academic Press, New York. Fröhlich, C., & J. Lean (2004). Solar radiative output and its variability: Evidence and mechanisms. Astron. Astrophys. Rev., 12, 273–320. Gagan, M. K., L. K. Ayliffe, D. Hopley, J. A. Cali, G. E. Mortimer, J. Chappell, et  al. (1998). Temperature and surface‐ocean water balance of the mid‐holocene tropical western Pacific, Science, 279(5353), 1014–1018. doi: 10.1126/ science.279.5353.1014 Gagan, M. K., L. K. Ayliffe, J. W. Beck, J. E. Cole, E. R. M. Druffel, R. B. Dunbar, & D. P. Schrag (2000). New views of tropical paleoclimates from corals. Quaternary Science Reviews, 19(1–5), 45 – 64. doi: 10.1016/ S0277‐3791(99)00054‐2 Gergis, J., & A. Fowler (2005). Classification of synchronous oceanic and atmospheric El Niño ‐ Southern Oscillation (ENSO) events for palaeoclimate reconstruction. Int. J. Climatol., 25, 1541–1565. Gergis, J., K. Braganza, A. Fowler, S. Mooney, & J. Risbey (2006). Reconstructing El Niño‐Southern Oscillation (ENSO) from high‐resolution palaeoarchives. Journal of Quaternary Science, 21, 707–722. doi: 10.1002/jqs. 1070 Goosse, H., H. Renssen, A. Timmermann, R. Bradley, & M. Mann (2006). Using paleoclimate proxy‐data to select optimal realisations in an ensemble of simulations of the climate of the past millennium. Climate Dynamics, 27, 165–184. doi: 10.1007/s00382‐006‐0128‐6 Goosse, H., E. Crespin, A. de Montety, M. E. Mann, H. Renssen, & A. Timmermann (2010). Reconstructing surface temperature changes over the past 600 years using climate model simulations with data assimilation. Journal of Geophysical Research: Atmospheres, 115(D9). doi:10.1029/ 2009JD012737

PAST ENSO VARIABILITY  109 Gorman, M. K., T. M. Quinn, F. W. Taylor, J. W. Partin, G. Cabioch, J. A. Austin, et al. (2012). A coral‐based reconstruction of sea surface salinity at Sabine Bank, Vanuatu from 1842 to 2007 CE. Paleoceanography, 27(3), PA3226. doi: 10.1029/ 2012PA002302 Gray, L. J., J. Beer, M. Geller, J. D. Haigh, M. Lockwood, K. Matthes, et  al. (2010). Solar influences on climate. Rev. Geophys., 48(4). doi:10.1029/2009RG000282 Grinsted, A., J. C. Moore, & S. Jevrejeva (2004). Application of the cross wavelet transform and wavelet coherence to geophysical time series. Nonlinear Processes in Geophysics, 11, 561–566. Grothe, P. R., K. M. Cobb, G. Liguori, E. Di Lorenzo, A. Capotondi, Y. Lu, et  al. (2019). Enhanced El Niño‐ Southern Oscillation variability in recent decades. Geophysical Research Letters, 46.https://doi.org/10.1029/2019GL083906. Guilderson, T. P., & D. P. Schrag (1999). Reliability of coral isotope records from the western Pacific warm pool: A comparison using age‐optimized records. Paleoceanography, 14, 457–464. doi: 10.1029/1999PA900024 Guillot, D., B. Rajaratnam, & J. Emile‐Geay (2015). Statistical paleoclimate reconstructions via Markov random fields. Ann. Applied. Statist., 324–352. doi: 10.1214/14‐AOAS794 Hakim, G. J., J. Emile‐Geay, E. J. Steig, D. Noone, D. M. Anderson, R. Tardif, et  al. (2016). The last millennium climate reanalysis project: Framework and first results. Journal of Geophysical Research: Atmospheres, 121, 6745 – 6764. doi: 10.1002/2016JD024751 Haurwitz, M. W., & G. W. Brier (1981). A critique of the superposed epoch analysis method: Its application to solar– weather relations. Monthly Weather Review, 109(10), 2074–2079. doi: 10.1175/1520‐0493(1981)109〈2074: ACO TSE〉2.0.CO;2 Haywood, A. M., P. J. Valdes, & V. L. Peck (2007). A permanent El Niño–like state during the Pliocene? Paleoceanography, 22(1). doi:10.1029/ 2006PA001323 Heiss, G. (1994). Coral reefs in the Red Sea: Growth, production and stable isotopes. Tech. Rep., Leibniz Institute of Marine Sciences. Held, I. M., & B. J. Soden (2006). Robust responses of the hydrological cycle to global warming. J. Clim., 19, 5686–5699. doi: 10.1175/JCLI3990.1 Hendy, E. J., M. K. Gagan, J. M. Lough, M. McCulloch, & P. B. deMenocal (2007). Impact of skeletal dissolution and secondary aragonite on trace element and isotopic climate proxies in porites corals. Paleoceanography, 22(4). doi:10.1029/2007PA001462 Hereid, K. A., T. M. Quinn, F. W. Taylor, C.‐C. Shen, R. L. Edwards, & H. Cheng (2013). Coral record of reduced El Niño activity in the early 15th to middle 17th centuries. Geology, 41(1), 51–54. doi: 10.1130/G33510.1 Hu, J., J. Emile‐Geay, J. Nusbaumer, & D. Noone (2018). Impact of convective activity on precipitation δ18O in isotope‐enabled general circulation models. Journal of Geophysical Research: Atmospheres, 123(23), 13,595–13,610. doi: 10.1029/ 2018JD029187 Huang, B., P. W. Thorne, V. F. Banzon, T. Boyer, G. Chepurin, J. H. Lawrimore, et  al. (2017). Extended reconstructed sea

surface temperature, version 5 (ERSSTV5): Upgrades, validations, and intercomparisons. Journal of Climate, 30(20), 8179–8205. doi: 10.1175/JCLI‐D‐16‐0836.1 Hughen, K. A., D. P. Schrag, S. B. Jacobsen, & W. Hantoro (1999). El Niño during the last interglacial period recorded by a fossil coral from Indonesia. Geophysical Research Letters, 26(20), 3129–3132. doi: 10.1029/ 1999GL006062 Isdale, P. J., B. J. Stewart, K. S. Tickle, & J. M. Lough (1998). Palaeohydrological variation in a tropical river catchment: A reconstruction using fluorescent bands in corals of the Great Barrier Reef, Australia. The Holocene, 8(1). 1–8. doi: 10.1191/095968398670905088 Jimenez, G., J. E. Cole, D. M. Thompson, & A. W. Tudhope (2018). Northern Galápagos corals reveal twentieth century warming in the eastern tropical Pacific. Geophysical Research Letters, 45, 1981–1988. doi: 10.1002/ 2017GL075323 Jungclaus, J. H., E. Bard, M. Baroni, P. Braconnot, J. Cao, L. P. Chini, et al. (2017). The PMIP4 contribution to CMIP6: Part 3. The last millennium, scientific objective, and experimental design for the PMIP4 Past1000 simulations. Geoscientific Model Development, 10(11). 4005–4033. doi: 10.5194/ gmd‐10‐4005‐2017 Kageyama, M., S. Albani, P. Braconnot, S. P. Harrison, P. O. Hopcroft, R. F. Ivanovic, F. et al. (2017). The PMIP4 contribution to CMIP6: Part 4. Scientific objectives and experimental design of the PMIP4‐CMIP6 last glacial maximum experiments and PMIP4 sensitivity experiments. Geoscientific Model Development, 10(11), 4035–4055. doi: 10.5194/gmd‐10‐4035‐2017 Khider, D., L. D. Stott, J. Emile‐Geay, R. Thunell, & D. Hammond (2011). Assessing El Niño‐Southern Oscillation variability during the past millennium. Paleoceanography, 26(PA3222). doi:10.1029/2011PA002139 Khon, V. C., B. Schneider, M. Latif, W. Park, & C. Wengel (2018). Evolution of eastern equatorial Pacific seasonal and interannual variability in response to orbital forcing during the Holocene and Eemian from model simulations. Geophysical Research Letters, 45(18), 9843–9851. doi: 10.1029/ 2018GL079337 Kiefer, J., & C. Karamperidou (2019). High‐resolution modeling of ENSO‐induced precipitation in the tropical Andes: Implications for proxy interpretation. Paleoceanography and Paleoclimatology, 34(2). 217–236. doi: 10.1029/2018PA003423 Kilbourne, K. H., T. M. Quinn, F. W. Taylor, T. Delcroix, & Y. Gouriou (2004). El Niño‐Southern Oscillation–related salinity variations recorded in the skeletal geochemistry of a Porites coral from Espiritu Santo, Vanuatu. Paleoceanography, 19, PA4002. doi: 10.1029/2004PA001033 Koutavas, A., & S. Joanides (2012). El Niño‐Southern Oscillation extrema in the Holocene and Last Glacial Maximum. Paleoceanography, 27(4). PA4208. doi: 10.1029/2012PA002378 Koutavas, A., P. deMenocal, G. C. Olive, & J. Lynch‐Stieglitz (2006). Mid‐Holocene El Niño‐Southern Oscillation (ENSO) attenuation revealed by individual foraminifera in eastern tropical Pacific sediments. Geology, 34(12), 993–996. Kuhnert, H., J. Pätzold, B. Hatcher, K. H. Wyrwoll, A. Eisenhauer, L. B. Collins, et  al. (1999). A 200‐year coral stable oxygen isotope record from a high‐latitude reef off

110  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Western Australia. Coral Reefs, 18, 1–12. doi: 10.1007/ s003380050147 Kuhnert, H., J. Pätzold, K.‐H. Wyrwoll, & G. Wefer (2000). Monitoring climate variability over the past 116 years in coral oxygen isotopes from Ningaloo Reef, Western Australia. International Journal of Earth Sciences, 88, 725–732. doi: 10.1007/s005310050300 Lambeck, K., & J. Chappell (2001). Sea level change through the last glacial cycle. Science, 292(5517). 679. doi: 10.1126/ science.1059549 Leduc, G., L. Vidal, O. Cartapanis, & E. Bard (2009). Modes of eastern equatorial Pacific thermocline variability: Implications for ENSO dynamics over the last glacial period. Paleoceanography, 24(3). doi:10.1029/2008PA001701 LeGrande, A. N., & G. A. Schmidt (2011). Water isotopologues as a quantitative paleosalinity proxy. Paleoceanography, 26(3). PA3225. doi: 10. 1029/2010PA002043 Li, J., S.‐P. Xie, E. R. Cook, G. Huang, R. D’Arrigo, F. Liu, et al. (2011). Interdecadal modulation of El Niño amplitude during the past millennium. Nature Clim. Change, 1(2), 114– 118. doi: 10.1038/ nclimate1086 Li, J., S.‐P. Xie, E. R. Cook, M. S. Morales, D. A. Christie, N. C. Johnson, et  al. (2013). El Nino modulations over the past seven centuries. Nature Clim. Change, 3(9), 822–826. Linsley, B. K., R. B. Dunbar, G. M. Wellington, & D. A. Mucciarone (1994). A coral‐based reconstruction of intertropical convergence zone variability over Central America since 1707. J. Geophys. Res., 99, 9977–9994. doi: 10.1029/ 94JC00360 Linsley, B. K., L. Ren, R. B. Dunbar, & S. S. Howe (2000). El Niño Southern Oscillation (ENSO) and decadal‐scale climate variability at 10° in the eastern Pacific from 1893 to 1994: A coral‐based reconstruction from Clipperton Atoll. Paleoceanography, 15, 322–335. doi: 10.1029/1999PA000428 Linsley, B. K., A. Kaplan, Y. Gouriou, J. Salinger, P. B. deMenocal, G. M. Wellington, & S. S. Howe (2006). Tracking the extent of the South Pacific Convergence Zone since the early 1600s. Geochemistry, Geophysics, Geosystems, 7, 5003–+. doi: 10.1029/2005GC001115 Linsley, B. K., P. Zhang, A. Kaplan, S. S. Howe, & G. M. Wellington (2008). Interdecadal‐decadal climate variability from multicoral oxygen isotope records in the South Pacific Convergence Zone region since 1650 A.D. Paleoceanography, 23(26), A262,219+. doi: 10.1029/2007PA001539 Linsley, B. K., H. C. Wu, E. P. Dassié, & D. P. Schrag (2015). Decadal changes in South Pacific sea surface temperatures and the relationship to the Pacific decadal oscillation and upper ocean heat content. Geophysical Research Letters, 42(7), 2358–2366. doi: 10.1002/2015GL063045 Liu, Y., K. M. Cobb, H. Song, Q. Li, C.‐Y. Li, T. Nakatsuka, et al. (2017). Recent enhancement of central Pacific El Niño variability relative to last eight centuries. Nature Communications, 8, 15,386 EP –. doi: 10.1038/ncomms15386 Liu, Z. (2002). A simple model study of ENSO suppression by external periodic forcing. J. Clim., 15(9), 1088–1098. doi: 10. 1175/1520‐0442(2002)015〈1088: ASMSOE〉2.0.CO;2 Liu, Z., J. Kutzbach, & L. Wu (2000). Modeling climate shift of El Niño variability in the Holocene. Geophysical Research Letters, 27(15), 2265–2268. doi: 10.1029/2000GL011452

Liu, Z., Z. Lu, X. Wen, B. L. Otto‐Bliesner, A. Timmermann, & K. M. Cobb (2014). Evolution and forcing mechanisms of El Niño over the past 21,000 years. Nature, 515(7528), 550–553. doi: 10.1038/nature13963 Lloyd, J., E. Guilyardi, H. Weller, & J. Slingo (2009). The role of atmosphere feedbacks during ENSO in the CMIP3 models. Atmospheric Science Letters, 10(3), 170–176. doi: 10.1002/ asl.227 Lloyd, J., E. Guilyardi, & H. Weller (2011). The role of atmosphere feedbacks during ENSO in the CMIP3 models. Part II: Using AMIP runs to understand the heat flux feedback mechanisms. Clim. Dyn., 37(7‐8), 1271–1292. doi: 10. 1007/s00382‐010‐0895‐y Lloyd, J., E. Guilyardi, & H. Weller (2012). The role of atmosphere feedbacks during ENSO in the CMIP3 models. Part III: The shortwave flux feedback. Journal of Climate, 25(12), 4275–4293. doi: 10.1175/ JCLI‐D‐11‐00178.1 Lough, J. M. (2007). Tropical river flow and rainfall reconstructions from coral luminescence: Great Barrier Reef, Australia. Paleoceanography, 22(26), A262,218+. doi: 10.1029/ 2006PA001377 Lough, J. M. (2010). Climate records from corals. Wiley Interdisciplinary Reviews: Climate Change, 1(3), 318–331. doi: 10.1002/wcc.39 Lough, J. M. (2011). Great barrier reef coral luminescence reveals rainfall variability over northeastern Australia since the 17th century. Paleoceanography, 26(2), doi:10.1029/2010PA002050 Lu, Z., & Z. Liu (2019). Orbital modulation of ENSO seasonal phase locking. Climate Dynamics, 52(7), 4329–4350. doi: 10.1007/s00382‐018‐4382‐1 Lu, Z., Z. Liu, J. Zhu, & K. Cobb (2018). A review of paleo El Niño‐Southern Oscillation. Atmosphere, 9(4), 130. doi: 10.3390/atmos9040130 Luan, Y., P. Braconnot, Y. Yu, W. Zheng, & O. Marti (2012). Early and mid‐Holocene climate in the tropical Pacific: Seasonal cycle and interannual variability induced by insolation changes. Climate of the Past, 8(3), 1093–1108. doi: 10.5194/cp‐8‐1093‐2012 Luan, Y., P. Braconnot, Y. Yu, & W. Zheng (2015). Tropical Pacific mean state and ENSO changes: Sensitivity to freshwater flux and remnant ice sheets at 9.5 ka BP. Climate Dynamics, 44(3‐4). 661–678. doi: 10.1007/ s00382‐015‐2467‐7 Mann, M., E. Gilleb, R. Bradley, M. Hughes, J. Overpeck, F. Keimigc, & W. Gross (2000). Global temperature patterns in past centuries: An interactive presentation. Earth Interact, 4, 1–29. Mann, M. E. (2002). Climate reconstruction: The value of multiple proxies. Science, 297(5586), 1481–1482. doi: 10.1126/ science.1074318 Marchitto, T. M., R. Muscheler, J. D. Ortiz, J. D. Carriquiry, & A. van Geen (2010). Dynamical response of the tropical Pacific Ocean to solar forcing during the early Holocene. Science, 330(6009), 1378–1381. doi: 10. 1126/ science.1194887 MARGO Project Members (2009). Constraints on the magnitude and patterns of ocean cooling at the last glacial maximum. Nature Geosci, 2(2), 127–132. doi: 10.1038/ ngeo411 Masson‐Delmotte, V., M. Schulz, A. Abe‐Ouchi, J. Beer, A. Ganopolski, J. G. Rouco, et  al. (2013). Information from

PAST ENSO VARIABILITY  111 paleoclimate archives. In T. Stocker, D. Qin, G.‐K. Plattner, M. Tignor, S. Allen, J. Boschung, et al. (Eds.), Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (chap. 5), Cambridge University Press, Cambridge and New York. Matsikaris, A., M. Widmann, & J. Jungclaus (2015). On‐line and off‐line data assimilation in palaeoclimatology: A case study. Climate of the Past, 11(1), 81–93. doi: 10.5194/ cp‐11‐81‐2015 McDermott, F. (2004). Palaeo‐climate reconstruction from stable isotope variations in speleothems: A review. Quat. Sci. Rev., 23, 901–918. doi: 10.1016/ j.quascirev.2003.06.021 McGlone, M., A. Kershaw, & V. Markgraf (1992). El Niño/ Southern Oscillation and climatic variability in Australasian and South American paleoenvironmental records. In H. Diaz & V. Markgraf (Eds.), El Niño: Historical and paleoclimatic aspects of the Southern Oscillation (pp. 435–462), Cambridge University Press, Cambridge. McGregor, H. V., & M. K. Gagan (2003). Diagenesis and geochemistry of porites corals from Papua New Guinea: Implications for paleoclimate reconstruction. Geochimica et Cosmochimica Acta, 67(12), 2147–2156. doi: 10. 1016/ S0016‐7037(02)01050‐5 McGregor, H. V., & M. K. Gagan (2004). Western Pacific coral δ18O records of anomalous Holocene variability in the El Niño‐Southern Oscillation. Geophysical Research Letters, 31, L11,204. doi: 10.1029/2004GL019972 McGregor, H. V., M. J. Fischer, M. K. Gagan, D. Fink, S. J. Phipps, H. Wong, & C. D. Woodroffe (2013a). A weak El Niño‐Southern Oscillation with delayed seasonal growth around 4,300 years ago. Nature Geoscience, 6(11), 949–953. doi: 10.1038/ngeo1936 McGregor, S., A. Timmermann, & O. Timm (2010). A unified proxy for ENSO and PDO variability since 1650. Climate of the Past, 6(1), 1–17. doi: 10.5194/ cp‐6‐1‐2010 McGregor, S., A. Timmermann, M. H. England, O. Elison Timm, & A. T. Wittenberg (2013b). Inferred changes in El Niño‐Southern Oscillation variance over the past six centuries. Climate of the Past, 9(5), 2269–2284. doi: 10.5194/ cp‐9‐2269‐2013 Moerman, J. W., K. M. Cobb, J. F. Adkins, H. Sodemann, B. Clark, & A. A. Tuen (2013). Diurnal to interannual rainfall δ18O variations in northern borneo driven by regional hydrology. Earth and Planetary Science Letters, 369–370, 108–119. doi: 10.1016/j.epsl.2013.03.014 Moy, C., G. Seltzer, D. Rodbell, & D. Anderson (2002). Variability of El Niño/Southern Oscillation activity at millennial timescales during the Holocene epoch. Nature, 420(6912), 162–165. Muller, A., M. Gagan, & M. McCulloch (2001). Early marine diagenesis in corals and geochemical consequences for paleoceanographic reconstructions. Geophysical Research Letters, 28(23), 4471–4474. doi: 10.1029/ 2001GL013577 Murray‐Wallace, C. V., & C. D. Woodroffe (2014). Quaternary sea‐ level changes: A global perspective. Cambridge University Press. Newman, M., S.‐I. Shin, & M. A. Alexander (2011a). Natural variation in ENSO flavors. Geophys. Res. Lett., 38(14), L14,705. doi: 10.1029/ 2011GL047658

Newman, M., M. Alexander, & J. Scott (2011b). An empirical model of tropical ocean dynamics. Climate Dynamics, 37(9– 10), 1823–1841. doi: 10. 1007/s00382‐011‐1034‐0 Nurhati, I. S., K. M. Cobb, C. D. Charles, & R. B. Dunbar (2009). Late 20th century warming and freshening in the central tropical Pacific. Geophysical Research Letters, 36(21), doi:10.1029/2009GL040270 Nurhati, I. S., K. M. Cobb, & E. Di Lorenzo (2011). Decadal‐ scale SST and salinity variations in the central tropical Pacific: Signatures of natural and anthropogenic climate change. J. Clim., 24(13), 3294–3308. doi: 10.1175/ 2011JCLI3852.1 Nusbaumer, J., T. E. Wong, C. Bardeen, & D. Noone (2017). Evaluating hydrological processes in the community atmosphere model version 5 (CAM5) using stable isotope ratios of water. Journal of Advances in Modeling Earth Systems, 9(2), 949–977. doi: 10.1002/2016MS000839 Oliver, J. K., R. Berkelmans, & C. M. Eakin (2009). Coral bleaching in space and time (pp. 21–39), Springer Berlin Heidelberg. doi: 10.1007/978‐3‐540‐69775‐6_3 Ortlieb, L., & J. Macharé (1993). Former El Niño events: Records from western South America. Global and Planetary Change, 7(1–3), 181–202. doi: 10.1016/0921‐8181(93)90049‐T Ortlieb, L., A. M. Hocquenhem, & M. Soto (2000). The documented historical record of El Niño events in Peru: An update of the Quinn record (sixteenth through nineteenth centuries). In H. F. Diaz & V. Markgraf (Eds.), El Niño and the Southern Oscillation: Multiscale Variability and Global and Regional Impacts (chap. 7, pp. 207–296), Cambridge University Press, Cambridge. doi: 10.1017/CBO9780511573125.008 Otto‐Bliesner, B. L., E. C. Brady, S.‐I. Shin, Z. Liu, & C. Shields (2003). Modeling El Niño and its tropical teleconnections during the last glacial‐interglacial cycle. Geophys. Res. Lett., 30, 4–1. doi: 10.1029/ 2003GL018553 Otto‐Bliesner, B. L., E. C. Brady, G. Clauzet, R. Tomas, S. Levis, & Z. Kothavala (2006). Last Glacial Maximum and Holocene climate in CCSM3. J. Climate, 19, 2526–2544. doi: 10.1175/JCLI3748.1 Otto‐Bliesner, B. L., P. Braconnot, S. P. Harrison, D. J. Lunt, A. Abe‐Ouchi, S. Albani, et al. (2017). The PMIP4 contribution to CMIP6: Part 2. Two interglacials, scientific objective and experimental design for Holocene and last interglacial simulations. Geoscientific Model Development, 10(11), 3979–4003. doi: 10.5194/gmd‐10‐3979‐2017 PAGES 2k Consortium(2017). A global multiproxy database for temperature reconstructions of the Common Era. Scientific Data, 4, 170,088 EP. doi: 10. 1038/sdata.2017.88 Pedoja, K., L. Husson, A. Bezos, A.‐M. Pastier, A. M. Imran, C. Arias‐Ruiz, et al. (2018). On the long‐lasting sequences of coral reef terraces from SE Sulawesi (Indonesia): Distribution, formation, and global significance. Quaternary Science Reviews, 188, 37–57. doi: 10.1016/j.quascirev.2018.03. 033 Peltier, W. R. (2004). Global glacial isostasy and the surface of the Ice‐Age Earth: The ICE‐5G (VM2) model and GRACE. Ann. Rev. Earth Planet. Sci., 32, 111–149. Peristykh, A. N., & P. E. Damon (2003). Persistence of the Gleissberg 88‐year solar cycle over the last ~12,000 years: Evidence from cosmogenic isotopes. J. Geophys. Res. Space Phys., 108, 1–1. doi: 10.1029/2002JA009390

112  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Petit, J. R., J. Jouzel, D. Raynaud, N. I. Barkov, J. M. Barnola, I. Basile, et al. (1999). Climate and atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica. Nature, 399(6735). 429–436. doi: 10.1038/20859 Pfeiffer, M., O. Timm, W.‐C. Dullo, & S. Podlech (2004). Oceanic forcing of interannual and multidecadal climate variability in the southwestern Indian Ocean: Evidence from a 160 year coral isotopic record (La Réunion, 55°, 21°). Paleoceanography, 19, 4006–+. doi: 10.1029/2003PA000964 Quinn, T. M., T. J. Crowley, & F. W. Taylor (1996). New stable isotope results from a 173‐year coral from Espiritu Santo, Vanuatu. Geophys. Res. Lett., 23, 3413–3416. doi: 10.1029/96GL03169 Quinn, T. M., T. J. Crowley, F. W. Taylor, C. Henin, P. Joannot, & Y. Join (1998). A multicentury stable isotope record from a New Caledonia coral: Interannual and decadal sea surface temperature variability in the southwest Pacific since 1657 A.D. Paleoceanography, 13, 412–426. doi: 10. 1029/98PA00401 Quinn, T. M., F. W. Taylor, & T. J. Crowley (2006). Coral‐based climate variability in the Western Pacific Warm Pool since 1867. Journal of Geophysical Research (Oceans), 111, 11,006– +. doi: 10.1029/2005JC003243 Quinn, W. H. (1992). Large‐scale ENSO events, the El Niño and other important regional features. In L. Macharé, José Ortlieb, Registro del fenómeno El Niño y de eventos ENSO en América del Sur (Vol. 22, pp. 13–22), Institut Fran cais d’Etudes Andines, Lima. Rao, M. P., E. R. Cook, B. I. Cook, K. J. Anchukaitis, R. D. D’Arrigo, P. J. Krusic, & A. N. LeGrande (2019). A double bootstrap approach to superposed epoch analysis to evaluate response uncertainty. Dendrochronologia, 55, 119–124. doi: 10.1016/j.dendro.2019.05.001 Rasbury, M., & P. Aharon (2006). ENSO‐controlled rainfall variability records archived in tropical stalagmites from the mid‐ ocean island of Niue, South Pacific. Geochemistry, Geophysics, Geosystems, 7, 7010–+. doi: 10. 1029/2005GC001232 Rein, B. (2007). How do the 1982/83 and 1997/98 El Niño’s rank in a geological record from Peru? Quaternary International, 161(1), 56–66. doi: 10.1016/j.quaint.2006.10.023 Rein, B., A. Lückge, & F. Sirocko (2004). A major Holocene ENSO anomaly during the Medieval period. Geophys. Res. Lett., 31(17), L17,211. doi: 10. 1029/2004GL020161 Rein, B., A. Lückge, L. Reinhardt, F. Sirocko, A. Wolf, & W.‐C. Dullo (2005). El Niño variability off Peru during the last 20,000 years. Paleoceanography, 20(4). PA4003. doi: 10.1029/2004PA001099 Reynolds, R. W., N. A. Rayner, T. M. Smith, D. C. Stokes, & W. Wang (2002). An improved in situ and satellite SST analysis for climate. Journal of Climate, 15(13), 1609–1625. doi: 10.1175/1520‐0442(2002) 015〈1609:AIISAS〉2.0.CO;2 Rickaby, R. E. M., & P. Halloran (2005). Cool La Niña during the warmth of the Pliocene? Science, 307(5717),1948. doi: 10.1126/science.1104666 Riedinger, M. A., M. Steinitz‐Kannan, W. M. Last, & M. Brenner (2002). A 6~100 14C yr record of El Niño activity from the Galápagos Islands. Journal of Paleolimnology, 27(1), 1–7. doi: 10.1023/A:1013514408468 Risi, C., S. Bony, & F. Vimeux (2008). Influence of convective processes on the isotopic composition (δ18O and δD) of precipitation and water vapor in the tropics: 2. Physical interpre-

tation of the amount effect. Journal of Geophysical Research: Atmospheres, 113, D19,306. doi: 10.1029/2008JD009943 Risi, C., D. Noone, J. Worden, C. Frankenberg, G. Stiller, M. Kiefer, et  al. (2012). Process‐evaluation of tropospheric humidity simulated by general circulation models using water vapor isotopologues: 1. Comparison between models and observations. Journal of Geophysical Research: Atmospheres, 117, D05,303. doi: 10.1029/2011JD016621 Rodbell, D. T., G. O. Seltzer, D. M. Anderson, M. B. Abbott, D. B. Enfield, & J. H. Newman (1999). An 15,000‐year record of El Niño‐driven alluviation in southwestern Ecuador. Science, 283(5401), 516–520. doi: 10. 1126/science.283.5401.516 Russon, T., A. W. Tudhope, G. C. Hegerl, M. Collins, & J. Tindall (2013). Inter‐annual tropical Pacific climate variability in an isotope‐enabled CGCM: Implications for interpreting coral stable oxygen isotope records of ENSO. Climate of the Past, 9(4), 1543–1557. doi: 10.5194/cp‐9‐1543‐2013 Russon, T., A. W. Tudhope, G. C. Hegerl, A. Schurer, & M. Collins (2014). Assessing the significance of changes in ENSO amplitude using variance metrics. Journal of Climate, 27(13), 4911–4922. doi: 10.1175/ JCLI‐D‐13‐00077.1 Russon, T., A. W. Tudhope, M. Collins, & G. C. Hegerl (2015). Inferring changes in ENSO amplitude from the variance of proxy records. Geophysical Research Letters, (42), 1197–1204. doi: 10.1002/2014GL062331 Rustic, G. T., A. Koutavas, T. M. Marchitto, & B. K. Linsley (2015). Dynamical excitation of the tropical Pacific Ocean and ENSO variability by Little Ice Age cooling. Science, 350(6267), 1537. doi: 10.1126/science.aac9937 Sachs, J. P., D. Sachse, R. H. Smittenberg, Z. Zhang, D. S. Battisti, & S. Golubic (2009). Southward movement of the Pacific intertropical convergence zone AD 1400‐1850. Nature Geosci, 2(7), 519–525. doi: 10. 1038/ngeo554 Sadekov, A. Y., R. Ganeshram, L. Pichevin, R. Berdin, E. McClymont, H. Elderfield, & A. W. Tudhope (2013). Palaeoclimate reconstructions reveal a strong link between El Niño‐Southern Oscillation and tropical Pacific mean state. Nature Communications, 4(2692. Saint‐Lu, M., P. Braconnot, J. Leloup, M. Lengaigne, & O. Marti (2015). Changes in the ENSO/SPCZ relationship from past to future climates. Earth and Planetary Science Letters, 412(0), 18–24. doi: 10.1016/j.epsl.2014.12.033 Sayani, H. R., K. M. Cobb, A. L. Cohen, W. C. Elliott, I. S. Nurhati, R. B. Dunbar, et al. (2011). Effects of diagenesis on paleoclimate reconstructions from modern and young fossil corals. Geochimica et Cosmochimica Acta, 75(21), 6361–6373. doi: 10.1016/j.gca.2011.08.026 Schmidt, G. A. (2010). Enhancing the relevance of palaeoclimate model/data comparisons for assessments of future climate change. Journal of Quaternary Science, 25(1), 79–87. doi: 10.1002/jqs.1314 Schmidt, G. A., J. H. Jungclaus, C. M. Ammann, E. Bard, P. Braconnot, T. J. Crowley, et al. (2012). Climate forcing reconstructions for use in PMIP simulations of the last millennium (v1.1). Geoscientific Model Development, 5(1), 185–191. doi: 10.5194/gmd‐5‐185‐2012 Schneider, T., H. Hampel, P. V. Mosquera, W. Tylmann, & M. Grosjean (2018). Paleo‐ENSO revisited: Ecuadorian Lake Pallcacocha does not reveal a conclusive El Niño signal.

PAST ENSO VARIABILITY  113 Global and Planetary Change, 168, 54–66. doi: https://doi. org/10.1016/j.gloplacha.2018.06.004 Seager, R., M. Cane, N. Henderson, D.‐E. Lee, R. Abernathey, & H. Zhang (2019). Strengthening tropical Pacific zonal sea surface temperature gradient consistent with rising greenhouse gases. Nature Climate Change, 9(7), 517–522. doi: 10.1038/s41558‐019‐0505‐x Sigl, M., M. Winstrup, J. R. McConnell, K. C. Welten, G. Plunkett, F. Ludlow, et al. (2015). Timing and climate forcing of volcanic eruptions for the past 2,500 years. Nature, 523(7562), 543–549. Smerdon, J. E. (2011). Climate models as a test bed for climate reconstruction methods: Pseudoproxy experiments. WIREs Clim Change. doi: 10.1002/wcc. 149 Stahle, D. W., M. K. Cleaveland, M. D. Therrell, D. A. Gay, R. D. D’Arrigo, P. J. Krusic, et al. (1998). Experimental dendroclimatic reconstruction of the Southern Oscillation. Bulletin of the American Meteorological Society, 79, 2137–2152. doi: 10.1175/1520‐0477(1998)079 Steiger, N. J., G. J. Hakim, E. J. Steig, D. S. Battisti, & G. H. Roe (2014). Assimilation of time‐averaged pseudoproxies for climate reconstruction. Journal of Climate, 27(1), 426–441. doi: 10.1175/JCLI‐D‐12‐00693.1 Stevenson, S., B. Fox‐Kemper, M. Jochum, B. Rajagopalan, & S. G. Yeager (2010). ENSO model validation using wavelet probability analysis. Journal of Climate, 23(20), 5540–5547. doi: 10.1175/2010JCLI3609.1 Stevenson, S., H. V. McGregor, S. J. Phipps, & B. Fox‐Kemper (2013). Quantifying errors in coral‐based ENSO estimates: Toward improved forward modeling of δ18O. Paleoceanography, 28(4), 633–649. doi: 10.1002/palo.20059 Stevenson, S., B. Otto‐Bliesner, J. Fasullo, & E. Brady (2016). “El Niño‐like” hydroclimate responses to last millennium volcanic eruptions. Journal of Climate, 29(8), 2907–2921. doi: 10.1175/JCLI‐D‐15‐0239.1 Stevenson, S., B. Powell, K. Cobb, J. Nusbaumer, M. Merrifield, & D. Noone (2018). Twentieth century seawater δ18O dynamics and implications for coral‐based climate reconstruction. Paleoceanography and Paleoclimatology, 33(6), 606–625. doi: 10.1029/2017PA003304 Stevenson, S., B. L. Otto‐Bliesner, E. C. Brady, J. Nusbaumer, C. Tabor, R. Tomas, et al. (2019). Volcanic eruption signatures in the isotope‐enabled last millennium ensemble. Paleoceanography and Paleoclimatology, 34(8). doi: 10.1029/2019PA003625 Tangri, N., R. B. Dunbar, B. K. Linsley, & D. M. Mucciarone (2018). ENSO’s shrinking twentieth‐century footprint revealed in a half‐millennium coral core from the South Pacific convergence zone. Paleoceanography and Paleoclimatology, 33(11), 1136–1150. doi: 10.1029/2017PA003310 Tardif, R., G. J. Hakim, W. A. Perkins, K. A. Horlick, M. P. Erb, J. Emile‐Geay, et al. (2019). Last millennium reanalysis with an expanded proxy database and seasonal proxy modeling. Climate of the Past, 15(4), 1251–1273. doi: 10.5194/ cp‐15‐1251‐2019 Thirumalai, K., J. W. Partin, C. S. Jackson, & T. M. Quinn (2013). Statistical constraints on El Nino‐Southern Oscillation reconstructions using individual foraminifera: A sensitivity analysis. Paleoceanography, 28(3), 401–412. doi: 10.1002/palo.20037

Thirumalai, K., P. N. DiNezio, J. E. Tierney, M. Puy, & M. Mohtadi (2019). An El Niño mode in the glacial Indian Ocean? Paleoceanography and Paleoclimatology, 34. doi: 10.1029/2019PA003669 Thompson, D. M., T. R. Ault, M. N. Evans, J. E. Cole, & J. Emile‐Geay (2011). Comparison of observed and simulated tropical climate trends using a forward model of coral δ18O, Geophys. Res. Lett., 38, L14,706. doi: 10.1029/2011GL048224 Thompson, D. M., J. L. Conroy, A. Collins, S. R. Hlohowskyj, J. T. Overpeck, M. Riedinger‐Whitmore, et  al. (2017). Tropical Pacific climate variability over the last 6000 years as recorded in Bainbridge Crater Lake, Galápagos. Paleoceanography, 32(8), 903–922. doi: 10.1002/2017PA003089 Thompson, L. G., E. Mosley‐Thompson, M. E. Davis, V. S. Zagorodnov, I. M. Howat, V. N. Mikhalenko, & P.‐N. Lin (2013). Annually resolved ice core records of tropical climate variability over the past ~1800 years. Science, 340(6135), 945– 950. doi: 10.1126/science.1234210 Tierney, J. E., N. J. Abram, K. J. Anchukaitis, M. N. Evans, C. Giry, K. H. Kilbourne, et al. (2015). Tropical sea surface temperatures for the past four centuries reconstructed from coral archives. Paleoceanography, 30(3), 226–252. doi: 10.1002/ 2014PA002717 Timmermann, A., & F. Jin (2002). A nonlinear mechanism for decadal El Niño amplitude changes. Geophys. Res. Lett., 29, 1003–1006. doi: 10.1029/ 2001GL013369 Timmermann, A., J. Oberhuber, A. Bacher, M. Esch, M. Latif, & E. Roeckner (1999). Increased El Niño frequency in a climate model forced by future greenhouse warming. Nature, 398(6729), 694–697. doi: 10.1038/19505 Timmermann, A., S. J. Lorenz, S.‐I. An, A. Clement, & S.‐P. Xie (2007). The effect of orbital forcing on the mean climate and variability of the tropical Pacific. Journal of Climate, 20(16), 4147–4159. doi: 10.1175/JCLI4240.1 Timmermann, A., S.‐I. An, J.‐S. Kug, F.‐F. Jin, W. Cai, A. Capotondi, et al. (2018). El Niño–Southern Oscillation complexity. Nature, 559(7715), 535–545. doi: 10.1038/ s41586‐018‐0252‐6 Tingley, M. P., P. F. Craigmile, M. Haran, B. Li, E. Mannshardt, & B. Rajaratnam (2012). Piecing together the past: Statistical insights into paleoclimatic reconstructions. Quaternary Science Reviews, 35, 1–22. doi: 10.1016/j. quascirev.2012.01.012 Trenberth, K. E., G. W. Branstator, D. Karoly, A. Kumar, N.‐C. Lau, & C. Ropelewski (1998). Progress during TOGA in understanding and modeling global teleconnections associated with tropical sea surface temperatures. J. Geophys. Res., 103, 14,291–14,324. doi: 10.1029/97JC01444 Tudhope, A. W., G. B. Shimmield, C. P. Chilcott, M. Jebb, A. E. Fallick, & A. N. Dalgleish (1995). Recent changes in climate in the far western equatorial Pacific and their relationship to the Southern Oscillation: Oxygen isotope records from massive corals, Papua New Guinea. Earth and Planetary Science Letters, 136, 575–590. doi: 10.1016/0012‐ 821X(95)00156‐7 Tudhope, A. W., C. P. Chilcott, M. T. McCulloch, E. R. Cook, J. Chappell, R. M. Ellam, et al. (2001). Variability in the El Niño‐Southern Oscillation through a glacial‐interglacial cycle. Science, 291, 1511–1517. doi: 10.1126/science.1057969

114  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Turney, C. S. M., A. P. Kershaw, S. C. Clemens, N. Branch, P. T. Moss, & L. Keith Fifield (2004). Millennial and orbital variations of El Niño/Southern Oscillation and high‐latitude climate in the last glacial period. Nature, 428(6980), 306–310. doi: 10.1038/nature02386 Tziperman, E., L. Stone, M. A. Cane, & H. Jarosh (1994). El Niño chaos: Overlapping of resonances between the seasonal cycle and the Pacific ocean‐atmosphere oscillator. Science, 264(5155), 72–74. doi: 10.1126/science. 264.5155.72 Urban, F. E., J. E. Cole, & J. T. Overpeck (2000). Influence of mean climate change on climate variability from a 155‐year tropical Pacific coral record. Nature, 407, 989–993. Vásquez‐Bedoya, L. F., A. L. Cohen, D. W. Oppo, & P. Blanchon (2012). Corals record persistent multidecadal SST variability in the Atlantic warm pool since 1775 AD. Paleoceanography, 27(3), PA3231. doi: 10.1029/2012PA002313 Vecchi, G. A., & A. T. Wittenberg (2010). El Niño and our future climate: Where do we stand? Wiley Interdisciplinary Reviews: Climate Change, 1(2), 260–270. Wagner, G., J. Beer, J. Masarik, R. Muscheler, P. Kubik, W. Mende, et  al. (2001). Presence of the solar De Vries cycle (~  205 years) during the last Ice Age. Geophys. Res. Lett., 28(2), 303–306. Wahl, E. R., H. F. Diaz, J. E. Smerdon, & C. M. Ammann (2014). Late winter temperature response to large tropical volcanic eruptions in temperate western North America: Relationship to ENSO phases. Global and Planetary Change, 122, 238–250. doi: https://doi.org/10.1016/j.gloplacha.2014.08.005 Wara, M. W., A. C. Ravelo, & M. L. Delaney (2005). Permanent El Niño‐like conditions during the Pliocene warm period. Science, 309(5735), 758. doi: 10.1126/science.1112596 Watanabe, T., A. Suzuki, S. Minobe, T. Kawashima, K. Kameo, K. Minoshima, et al. (2011). Permanent El Niño during the Pliocene warm period not supported by coral evidence. Nature, 471, 209–211. doi: 10.1038/nature09777 Welsh, K., M. Elliot, A. Tudhope, B. Ayling, & J. Chappell (2011). Giant bivalves (Tridacna gigas) as recorders of ENSO variability. Earth and Planetary Science Letters, 307(3–4), 266–270. doi: 10.1016/j.epsl.2011.05.032 White, S. M., A. C. Ravelo, & P. J. Polissar (2018). Dampened El Niño in the early and mid‐Holocene due to insolation‐forced warming/deepening of the thermocline. Geophysical Research Letters, 45(1), 316–326. doi: 10. 1002/2017GL075433 Widmann, M., H. Goosse, G. Schrier, R. Schnur, & J. Barkmeijer (2010). Using data assimilation to study extratropical northern

hemisphere climate over the last millennium. Climate of the Past, 6(5), 627–644. Wittenberg, A. T. (2009). Are historical records sufficient to constrain ENSO simulations? Geophys. Res. Lett., 36, L12,702. doi: 10.1029/2009GL038710 Woodroffe, C. D., & M. K. Gagan (2000). Coral microatolls from the central Pacific record Late Holocene El Niño. Geophysical Research Letters, 27(10), 1511–1514. doi: 10.1029/2000GL011407 Woodroffe, C. D., M. R. Beech, & M. K. Gagan (2003). Mid‐ late Holocene El Niño variability in the equatorial Pacific from coral microatolls. Geophysical Research Letters, 30(7), 1358. doi: 10.1029/2002GL015868 Wu, H., B. Linsley, E. Dassié, B. Schiraldi, & P. deMenocal (2013). Oceanographic variability in the South Pacific convergence zone region over the last 210 years from multi‐site coral Sr/Ca records. Geochem. Geophys. Geosyst, 14, 1435– 1453. doi: 10.1029/2012GC004293 Yan, H., C. Liu, W. Zhang, M. Li, X. Zheng, G. Wei, et  al. (2017). ENSO variability around 2,000 years ago recorded by Tridacna gigas δ18O from the South China Sea. Quaternary International, 452, 148–154. doi: 10.1016/j.quaint.2016.05.011 Zebiak, S. E., & M. A. Cane (1987). A model El Niño‐Southern Oscillation. Mon. Weather Rev., 115(10), 2262–2278. Zhang, Z., G. Leduc, & J. P. Sachs (2014). El Niño evolution during the Holocene revealed by a biomarker rain gauge in the Galápagos Islands. Earth and Planetary Science Letters, 404, 420–434. doi: 10.1016/j.epsl.2014.07. 013 Zheng, W., P. Braconnot, E. Guilyardi, U. Merkel, & Y. Yu (2008). ENSO at 6ka and 21ka from ocean–atmosphere coupled model simulations. Clim. Dyn., 30(7‐8), 745–762. doi: 10.1007/s00382‐007‐0320‐3 Zhisheng, A., S. C. Clemens, J. Shen, X. Qiang, Z. Jin, Y. Sun, et al. (2011). Glacial‐interglacial Indian summer monsoon dynamics. Science, 333(6043), 719. doi: 10.1126/ science.1203752 Zhu, J., Z. Liu, E. Brady, B. Otto‐Bliesner, J. Zhang, D. Noone, et al. (2017). Reduced ENSO variability at the LGM revealed by an isotope‐enabled earth system model. Geophysical Research Letters, 44(13), 6984–6992. doi: 10.1002/2017GL073406 Zinke, J., W.‐C. Dullo, G. A. Heiss, & A. Eisenhauer (2004). ENSO and Indian Ocean subtropical dipole variability is recorded in a coral record off southwest Madagascar for the period 1659 to 1995. Earth and Planetary Science Letters, 228, 177–194. doi: 10.1016/j.epsl.2004.09.028DATA CITATIONS

PAST ENSO VARIABILITY  115

APPENDIX: DATA CITATIONS Table A1  Data sources for the records in Figure 5.1, together with essential metadata. Age is expressed as years before 1950 CE. Site Name

Archive

Latitude

Longitude

Age

Reference

Bayes Islet Huon Peninsula, PNG Huon Peninsula, PNG Malo Channel, Vanuatu Malo Channel, Vanuatu Peruvian Coast Fanning Is. Christmas Is. Muschu Is. Pirotan

coral mollusk coral coral

21°S 7°S 6°S 16°S

165°E 148°E 148°E 167°E

‐46 y ‐44 y ‐37 y ‐31 y

Emile‐Geay et al. (2016) Welsh et al. (2011) Tudhope et al. (2001) Kilbourne et al. (2004)

coral

16°S

167°E

‐31 y

Kilbourne et al. (2004)

mollusk coral coral coral coral

15°S 4°N 2°N 3°S 22.6°N

74°W 159°W 158°E 144°E 70°E

‐30 y ‐28 y ‐23 y ‐4 y 2y

Palmyra Is. Isla Wolf, Galapagos Savusavu, Fiji Laing, PNG Christmas Is. Madang, PNG Malo Channel Malo Channel Madang Rarotonga(99) Fiji Nauru Tarawa Clipperton Atoll Palmyra Palmyra Laing Madang Suwarrow

coral coral coral coral coral coral coral coral coral coral coral coral coral coral coral coral coral coral coral

6°N 1.5°N 16.8°S 4°S 2°N 5°S 15.7°S 15.7°S 5.2°S 21.2°S 16.9°S 0.5°S 1°N 10.3°N 5.9°N 5.9°N 4.2°S 5.2°S 13.2°S

162°W 91.5°W 179.2°E 145°E 157.3°W 146°E 167.2°E 167.2°E 145.8°E 159.8°E 177.6°E 166°E 172°E 109.2°W 162.1°W 162.1°W 144.9°E 145.8°E 163.1°W

8y 10 y 11 y 11 y 12 y 13 y 22 y 22 y 28 y 44 y 53 y 53 y 56 y 57 y 64 y 64 y 66 y 70 y 70 y

Ningaloo Reef Rarotonga(3R) Rabaul2006 Rabaul2006 Bunaken Mentawai Moorea Zanzibar

coral coral coral coral coral coral coral coral

21.9°S 21.2°S 4.2°S 4.2°S 1.5°N 0.1°S 17.5°S 6°S

114°E 159.8°E 152°E 152°E 124.8°E 98.5°E 149.8°W 39°E

72 y 76 y 83 y 83 y 90 y 92 y 98 y 99 y

Jarvis

coral

0°S

160°W

100 y

Tutia

coral

8°S

39°E

103 y

Mahe GBR COO01E Sabine Bank Maiana Atoll

coral coral coral coral

4.6°S 18°S 15.9°S 1°N

55°E 146.2°E 166°E 173°E

104 y 106 y 108 y 110 y

Carré et al. (2013) Cobb et al. (2013) McGregor et al. (2013a) McGregor & Gagan (2004) Chakraborty & Ramesh (1998) Cobb et al. (2003) Jimenez et al. (2018) Bagnato et al. (2004) Tudhope et al. (2001) Evans et al. (1998) Tudhope et al. (2001) Kilbourne et al. (2004) Kilbourne et al. (2004) Tudhope et al. (1995) Linsley et al. (2006) Dassié et al. (2014) Guilderson & Schrag (1999) Cole et al. (1993) Linsley et al. (2000) Cobb et al. (2001) Nurhati et al. (2011) Tudhope et al. (2001) Tudhope et al. (2001) Alexander Tudhope et al., personal communication (2012) Kuhnert et al. (2000) Linsley et al. (2006) Quinn et al. (2006) Quinn et al. (2006) Charles et al. (2003) Abram et al. (2008) Boiseau et al. (1998) Heidi Barnett et al., personal communication (2012) Alexander Tudhope et al., personal communication (2012) Heidi Barnett et al., personal communication (2012) Charles et al. (1997) Lough (2011) Gorman et al. (2012) Urban et al. (2000) (Continued)

116  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Table A1  (Continued) Site Name

Archive

Latitude

Longitude

Age

Reference

Palmerston

coral

19.2°S

169.3°W

117 y

La Reunion GBR LUP01C GBR CID01A Kavieng GBR SNH01A GBR MAG01D Espiritu Santo Malindi Abrolhos Tonga TH1 Ha’afera Is. Guam Aqaba 18 GBR BRO01A Lombok Strait Fiji 1F Savusavu Bramble Cay

coral coral coral coral coral coral coral coral coral coral coral coral coral coral coral coral coral coral

21°S 20.3°S 20.3°S 2.5°S 20.1°S 19.1°S 15°S 3°S 28.5°S 19.9°S 19.9°S 13°N 29.5°N 18.1°S 8.2°S 16.8°S 16.8°S 9.2°S

55.2°E 149.1°E 148.9°E 150.5°E 148.9°E 146.9°E 167°E 40°E 113.8°E 174.7°W 174.7°W 145°E 35.1°E 146.3°E 115.5°E 179.2°E 179.2°E 143.9°E

118 y 124 y 125 y 127 y 130 y 136 y 144 y 149 y 156 y 156 y 159 y 160 y 162 y 167 y 168 y 169 y 174 y 176 y

Puerto Morelos

coral

20.8°N

86.7°W

177 y

Ras Umm Sidd Rarotonga(2R) Rarotonga(2R) Rotuma

coral coral coral coral

27.8°N 21.2°S 21.2°S 12.5°S

34.3°E 159.8°E 159.8°E 177°E

199 y 224 y 224 y 230 y

Secas Ifaty GBR HMP01B Wallis

coral coral coral coral

8°N 23.1°S 23.2°S 12.3°S

82.1°W 43.6°E 151°E 176.1°W

243 y 262 y 265 y 270 y

Palmyra Is. GBR HKO01B Amedee Light. Ifaty Amedee Havannah Is. GBR HAV01A Abraham Reef Queensland River Mafia Fiji AB Urvina Bay Christmas Is. American Samoa Ile Des Pins

coral coral coral coral coral coral coral coral coral coral coral coral coral coral coral

6°N 20.1°S 22.5°S 23.1°S 22.5°S 18.9°S 18.9°S 22.1°S 20°S 8°S 16.8°S 0.4°S 2°N 14°S 22.5°S

162°W 148.9°E 166.4°E 43.6°E 166.5°E 146.6°E 146.6°E 153°E 148.5°E 39.5°E 179.2°E 91.2°W 157°W 170°W 167.4°E

281 y 289 y 290 y 291 y 302 y 306 y 311 y 312 y 319 y 328 y 333 y 343 y 411 y 430 y 475 y

Misima Misima Peruvian Coast Palmyra Is.

coral coral mollusk coral

10.6°S 10.6°S 15°S 6°N

152.8°E 152.8°E 74°W 162°W

537 y 537 y 537 y -1 ky

Alexander Tudhope et al., personal communication (2012) Pfeiffer et al. (2004) Lough (2011) Lough (2011) Alibert & Kinsley (2008) Lough (2011) Lough (2011) Quinn et al. (1996) Cole et al. (2000) Kuhnert et al. (1999) Linsley et al. (2008) Wu et al. (2013) Asami et al. (2005) Heiss (1994) Lough (2011) Charles et al. (2003) Linsley et al. (2008) Bagnato et al. (2005) Julia E. Cole, personal communication (2012) Vásquez‐Bedoya et al. (2012) Felis et al. (2000) Linsley et al. (2008) Linsley et al. (2006) Thierry Correge, personal communication (2014) Linsley et al. (1994) Zinke et al. (2004) Lough (2011) Thierry Correge, personal communication (2014) Cobb et al. (2003) Lough (2011) Quinn et al. (1998) Zinke et al. (2004) DeLong et al. (2012) Isdale et al. (1998) Lough (2011) Druffel & Griffin (1999) Lough (2007) Damassa et al. (2006) Linsley et al. (2008) Dunbar et al. (1994) Cobb et al. (2013) Tangri et al. 2018 Thierry Correge, personal communication (2014) Hereid et al. (2013) Hereid et al. (2013) Carré et al. (2014) Cobb et al. (2003) (Continued)

PAST ENSO VARIABILITY  117 Table A1  (Continued) Site Name

Archive

Latitude

Longitude

Age

Reference

E. Pacific near Galapagos Taiwan Quelccaya ice cap Christmas Is. Christmas Is. Christmas Is. Dongdao Is. Lago Frías Christmas Is. Madang, PNG Laing, PNG Muschu Is. Santa Barbara Basin Christmas Is. Peruvian Coast Fanning Is. Christmas Is. Christmas Is. Christmas Is. Christmas Is. Bayes Islet Christmas Is. Christmas Is. Peruvian Coast Christmas Is. Christmas Is. Christmas Is. Muschu Is. Muschu Is. Muschu Is. Fanning Is. Fanning Is. Muschu Is. Fanning Is. Fanning Is. Huon Peninsula, PNG Bainbridge Crater Lake Bainbridge Crater Lake Fanning Is. Ratua Is., Vanuatu Fanning Is. Muschu Is. Peruvian Coast Muschu Is. Huon Peninsula, PNG Koil Is. Swallow Lagoon Huon Peninsula, PNG Peruvian Coast Huon Peninsula, PNG Peruvian Coast

marine sed

1°S

90°W

1 ky

Rustic et al. (2015)

tree glacier ice coral coral coral mollusk lake sed coral coral coral coral marine sed coral mollusk coral coral coral coral coral coral coral coral mollusk coral coral coral coral coral coral coral coral coral coral coral coral lake sed

24°N 14°S 2°N 2°N 2°N 16.5°N 40°N 2°N 6°S 4°S 3°S 34.1°N 2°N 15°S 4°N 2°N 2°N 2°N 2°N 21°S 2°N 2°N 15°S 2°N 2°N 2°N 3°S 3°S 3°S 4°N 4°N 3°S 4°N 4°N 6°S 0.3°S

121°E 71°W 157°W 157°W 157°W 112.5°E 71°E 157°W 148°E 145°E 144°E 120°W 157°W 74°W 159°W 157°W 157°W 157°W 157°W 165°E 157°W 157°W 74°W 157°W 157°W 157°W 144°E 144°E 144°E 159°W 159°W 144°E 159°W 159°W 148°E 90.6°W

1 ky 1 ky 1 ky 2 ky 2 ky 2 ky 2 ky 2 ky 2 ky 3 ky 3 ky 3 ky 3 ky 3 ky 3 ky 3 ky 4 ky 4 ky 4 ky 4 ky 4 ky 4 ky 5 ky 5 ky 5 ky 5 ky 5 ky 6 ky 6 ky 6 ky 6 ky 6 ky 6 ky 6 ky 6 ky 6 ky

Liu et al. (2017) Thompson et al. (2013) Cobb et al. (2013) Cobb et al. (2013) Woodroffe & Gagan (2000) Yan et al. (2017) Ariztegui et al. (2007) Cobb et al. (2013) Tudhope et al. (2001) Tudhope et al. (2001) McGregor & Gagan (2004) Beaufort & Grelaud (2017) Woodroffe et al. (2003) Carré et al. (2014) Cobb et al. (2013) Cobb et al. (2013) Woodroffe et al. (2003) Cobb et al. (2013) Cobb et al. (2013) Emile‐Geay et al. (2016) Cobb et al. (2013) McGregor et al. (2013a) Carré et al. (2014) Cobb et al. (2013) Cobb et al. (2013) Cobb et al. (2013) McGregor & Gagan (2004) McGregor & Gagan (2004) McGregor & Gagan (2004) Cobb et al. (2013) Cobb et al. (2013) McGregor & Gagan (2004) Cobb et al. (2013) Cobb et al. (2013) Tudhope et al. (2001) Riedinger et al. (2002)

lake sed

0.3°S

90.6°W

6 ky

Thompson et al. (2017)

coral coral coral coral mollusk coral mollusk coral lake sed mollusk mollusk mollusk mollusk

4°N 16°S 4°N 3°S 15°S 3°S 7°S 3°S 27.5°S 7°S 15°S 7°S 15°S

159°W 167°E 159°W 144°E 74°W 144°E 148°E 144°E 153.5°E 148°E 74°W 148°E 74°W

7 ky 7 ky 7 ky 7 ky 7 ky 7 ky 7 ky 8 ky 8 ky 8 ky 9 ky 9 ky 9 ky

Cobb et al. (2013) Duprey et al. (2012) Cobb et al. (2013) McGregor & Gagan (2004) Carré et al. (2014) McGregor & Gagan (2004) Driscoll et al. (2014) McGregor & Gagan (2004) Barr et al., 2019 Driscoll et al. (2014) Carré et al. (2014) Driscoll et al. (2014) Carré et al. (2014) (Continued)

118  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Table A1  (Continued) Site Name

Archive

Latitude

Longitude

Age

Reference

Northern Borneo El Junco Crater Lake El Junco Crater Lake Laguna Pallcacocha Soledad Basin SO147‐106KL Site V21‐30 CD38‐17P Site 806 Site 849 Lynch’s Crater MD022529 Bunaken Philippines

speleothem lake sed lake sed lake sed marine sed marine sed marine sed marine sed marine sed marine sed lake sed marine sed coral coral

4°N 0.8°S 0.3°S 2.8°S 25.2°N 12°S 1°S 1.5°S 0.3°N 0.1°N 17.4°S 8°N 1.5°N 13°N

114°E 89.3°W 90.6°W 79.2°W 112.7°W 78°W 90°W 90°W 159.4°E 110.5°W 145.7°E 84°W 125°E 124°E

10 ky 10 ky 10 ky 12 ky 14 ky 21 ky 21 ky 21 ky 21 ky 21 ky 45 ky 50 ky 124 ky 3 my

Chen et al. (2016) Conroy et al. (2008) Zhang et al. (2014) Moy et al. (2002) Marchitto et al. (2010) Rein et al. (2005) Koutavas & Joanides (2012) Sadekov et al. (2013) Ford et al. (2015) Rustic et al. (2015) Turney et al. (2004) Leduc et al. (2009) Hughen et al. (1999) Watanabe et al. (2011)

Section III Theories and Dynamics

6 Simple ENSO Models Fei‐Fei Jin1, Han‐Ching Chen1, Sen Zhao1, Michiya Hayashi1,#, Christina Karamperidou1, Malte F. Stuecker2,3,4,5,*, Ruihuang Xie6, and Licheng Geng1

ABSTRACT The fundamental dynamical mechanisms of the El Niño‐Southern Oscillation (ENSO) phenomenon have been extensively studied since Bjerknes envisioned ocean‐atmosphere interaction in the equatorial Pacific as its main cause. This chapter provides a review of the recent progress in ENSO theory based on two classes of relatively simple models: (i) the Cane‐Zebiak (CZ) type models of intermediate complexity and (ii) conceptual ­low‐­­order models reducible from the CZ‐type models. The leading mode of ENSO variability, in reanalysis data, the ­CZ‐type models, and comprehensive climate models, can be reduced into the simplest possible coupled oscillator known as the recharge oscillator (RO). Incorporating seasonality, nonlinearity, and multiscale ­ processes into the RO framework, allows for basic understanding of how key physical processes determine ENSO’s properties, such as its amplitude, periodicity, phase‐locking, asymmetry, and nonlinear rectification onto the mean state. As these key physical processes can be easily quantified in both model and reanalysis data, the RO framework can be used to assess the simulations and projections of ENSO in climate models.

6.1. INTRODUCTION The early stages of understanding and modeling of the ENSO phenomenon were marked by the seminal papers of Bjerknes (1969), Wyrtki (1985), and Cane and Zebiak (1985). These three visionary studies advanced a clear ­hypothesis that casts ENSO as a phenomenon originating 1 Department of Atmospheric Sciences, SOEST, University of Hawai‘i at Mˉanoa, Honolulu, HI, USA 2 Department of Oceanography, SOEST, University of Hawai‘i at Mˉanoa, Honolulu, HI, USA 3 International Pacific Research Center, SOEST, University of Hawai‘i at Mˉanoa, Honolulu, HI, USA 4 Center for Climate Physics, Institute for Basic Science, Busan, Republic of Korea 5 Pusan National University, Busan, Republic of Korea 6 Key Laboratory of Ocean Circulation and Waves, Institute of Oceanology, Chinese Academy of Sciences, Qingdao, China *now at the University of Hawai‘i at Mˉanoa, HI, USA #  now at Center for Global Environmental Research, National Institute for Environmental Studies, Tsukuba, Ibaraki, Japan

from an oscillatory coupled ocean‐atmosphere instability. First, Bjerknes (1969) hypothesized that ENSO SST anomalies (SSTAs) can grow via reorganizing of the equatorial Pacific trade winds and altering of the ocean mixed‐layer heat budget through horizontal currents and upwelling. Then, Wyrtki (1985) noted that in order for a phase transition to occur and for ENSO to be cyclic, a redistribution of heat content driven by the trade winds is required, namely, a slow recharge of warm water in the western Pacific before the onset of El Niño and a discharge after its peak. Meanwhile, Cane and Zebiak (1985) built the first dynamical ENSO model to successfully test the above hypotheses. While these hypotheses for ENSO growth and its phase transitions were based on very limited data, and the Cane‐Zebiak model is of modest complexity, they set the stage for the advancement of ENSO theory, modeling, and prediction for years to come. Decades’ worth of research studies since then viewed ENSO as a coupled ocean‐atmosphere mode (i.e., a spatially and temporally coherent pattern of covariability in

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 121

122  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

oceanic and atmospheric variables that arises from their dynamical coupling) that can be described on the basis of linear instability theory using relatively simple coupled models that eventually boil down to the delayed and recharge oscillator paradigms (e.g., Cane & Zebiak, 1985; Gill, 1985; Hirst, 1986; Suarez & Schopf, 1988; Battisti & Hirst, 1989; Philander, 1990; Neelin, 1990; Jin & Neelin, 1993a, 1993b; Neelin & Jin, 1993; Neelin et al., 1994; Jin, 1997a, 1997b; Neelin et  al., 1998; Wang et  al., 1999). ENSO behaviors observed in nature or simulated in various coupled models were viewed qualitatively as the result of a sensitive dependence of the leading coupled modes to variations in the climate background state or in physical parameters related to the ocean‐atmosphere coupling (e.g., Philander et al., 1984; Jin & Neelin, 1993a; An & Jin, 2000; Fedorov & Philander, 2000, 2001; Bejarano, 2006; Bejarano & Jin, 2008; Xie & Jin, 2018; Timmermann et  al., 2018). The oscillatory nature of ENSO has been debated (Kessler, 2002; Philander & Fedorov, 2003; chapter  7 in this book), as it is being obscured by noise, nonlinearity, asymmetry, and abundant multiscale interactions. However, the largely linear view of ENSO described above has set the foundation for the understanding of fundamental ENSO dynamics. We will thus briefly review the progress that has been made on ENSO linear instability theory in section 6.2. Adopting the principles of linear instability theory, a class of highly simplified conceptual models was developed since the 1980s, including the delayed oscillator (Suarez & Schopf, 1988; Battisti & Hirst, 1989), wave oscillator (Cane et  al., 1990; Jin, 1997b), recharge oscillator (Jin, 1996, 1997a, 1997b), and advective‐reflective oscillator (Picaut et al., 1997). Despite their simplifications and their limitations in explaining the observed ENSO spatial pattern diversity and temporal complexity, most of these conceptual models are based on solid theoretical foundations and may be consolidated into a generalized recharge oscillator (RO) framework (Jin & An, 1999). This RO framework can be used to formulate a set of simple, approximate, but systematic measures for ENSO instability, namely, the Bjerknes and Wyrtki indices for ENSO linear growth rate and periodicity, respectively (Jin et  al., 2006; Lu et al., 2018). This quantitative assessment of the ENSO instability indices using observational data and outputs from comprehensive coupled general circulation models (CGCMs) make the RO conceptual model framework a useful tool, particularly for connecting ENSO theory to observed and modeled ENSO behaviors under various background climate conditions. In section 6.3, we will discuss these simple conceptual models, their consolidation and generalization, and the formulation of the ENSO instability index, as well as its applications and limitations. The role of seasonal modulation and stochastic and external forcing on ENSO dynamics and variability will

be discussed in section  6.4, using linear and nonlinear versions of the conceptual models and the ENSO instability index. This chapter will end with a brief discussion of its connections to previous and following chapters and an outlook on future research and the role of conceptual models in a hierarchical approach to improve our understanding of ENSO dynamics in past, present, and future climates. 6.2. COUPLED LINEAR INSTABILITY The observed variability of the coupled climate system and its underlying physical processes can be investigated via a study of the stability of a simplified linearized representation of the coupled climate system in the mathematical framework of dynamical systems. Similar to the role that baroclinic instability plays in the development of synoptic weather systems, coupled ocean‐atmosphere instability is important for generating interannual variability in the tropical climate system. The frontal theory for cyclones by Bjerknes and Solberg (1922) and the coupled ocean-atmosphere positive feedback hypothesis for ENSO by Bjerknes (1969) foresaw the role of these two fundamental instabilities for Earth’s dominant weather and climate systems. Advances in the understanding of tropical atmospheric waves and circulation (e.g., Matsuno, 1966; Webster, 1973; Gill, 1980; Zebiak, 1982; Lindzen & Nigam, 1987) and in parallel of tropical ocean dynamics (Moore, 1968; Cane & Sarachik, 1977, 1979, 1981; Cane, 1984) led to the ingenious formulation of the Cane‐Zebiak (CZ) model, which was the first dynamical model that realistically simulated and predicted ENSO (Cane & Zebiak, 1985; Cane et al., 1986; Zebiak & Cane, 1987). Since then, CZ‐type frameworks have become well‐utilized tools in the advancement of ocean‐atmosphere instability theory for ENSO (Gill, 1985; Hirst, 1986; Neelin, 1990; Jin et al., 1994; Jin, 1997a, 1997b; An & Jin, 2000; Fedorov & Philander, 2000, 2001; Bejarano & Jin, 2008; Xie & Jin, 2018; Timmermann et al., 2018). 6.2.1. Brief Description of the Cane‐Zebiak Model The CZ model is an anomaly model with a prescribed annually varying climate mean state (Figure  6.1a) and consists of simple atmospheric and oceanic components that can be derived from first principles with reasonable simplifications. As illustrated schematically in Figure 6.1b, it comprises a simple quasi‐linear Gill-Matsuno atmospheric component (Gill, 1980; Zebiak, 1982; Zebiak & Cane, 1987; also see the appendix to this chapter) that simulates the tropical wind response to ENSO‐associated SST anomalies. Its oceanic dynamical component is a 1.5‐layer linear reduced gravity model that describes the

Simple ENSO Models  123 (a)

(b)

La titu de

Climatological fields 20°N

28

24

10°N

10°N 10 m/s

Equator

u1

50

ws

28

100

Eq.

24

200 250 120°E

Nino4 Nino3

17 18 19 20 21 22 23 24 25 26 27 28 29

150°E

180°

mixed layer

16

Depth [m]

Depth [m]

50

150

ZC model schema Gill-type atmospheric 20°N model response

150°W Longitude

120°W

90°W 10

20

30

Temp. [°C]

100 subsurface layer

10 m/s

u1 u2

T ws

h~Tsub

h

5 3 1 −1

150 200

7

−3 −5

motionless layer

250 120°E

150°E

−7

180°

150°W

120°W

90°W

[°C]

Longitude

Figure 6.1 Schematic diagrams for the Cane‐Zebiak (CZ) model. (a) Climatological states prescribed for the model. Horizontal plane: mean SST (shadings) and surface winds (vector). Vertical plane: mean vertical ocean temperature profile along equator (shadings, left panel), and in the Niño‐4 (red line, right panel) and Niño‐3 (blue line, right panel) regions. Embedded black arrow indicates mean zonal current in the mixed layer, and orange arrow denotes mean upwelling. The 18°C isotherm is highlighted as thermocline. (b) Schematic for CZ model dynamics. Horizontal plane: Simulated Gill-Matsuno wind response (vectors) to observed SSTAs (shadings) averaged between October 1997 and February 1998 (example El Niño event). Vertical plane: observed ocean temperature anomalies (shadings) in the same period and simulated thermocline depth (green dashed line). The thick black arrows schematically present the simulated anomalous ocean current in the mixed layer 1 and subsurface layer 2. The change in the thermocline depth (h) is used to derive the ocean temperature anomalies in the subsurface (Tsub), which further influence SSTA (T) through Eq. A4a.

upper‐layer current and thermocline depth anomalies in response to wind anomalies. Within this upper layer ocean there is an embedded mixed layer with a fixed depth and an underlying subsurface layer with a prescribed mean depth. This approximation in representing the vertical structure of ocean dynamics allows the CZ oceanic model to simulate anomalous horizonal and vertical velocities in the mixed layer and at the mixed‐layer base, respectively. By introducing an assumption that subsurface ocean temperature is adiabatically redistributed following the thermocline depth variations, the CZ model captures to a large extent mixed‐layer SSTA dynamics governed by the heat budget equation (Zebiak & Cane, 1987; also see the appendix for details). The atmosphere and ocean are coupled in the CZ model by three main closure approximations: (i) an approximation for wind‐driven ocean current anomalies in the mixed layer that combines a vertically sheared current from an Ekman‐flow model with an upper‐layer current from the reduced‐gravity wave dynamics model, (ii) a nonlinear relation between the subsurface ocean temperature and the thermocline depth, partly motivated by the dependence of the vertical profiles on the thermocline depth (Figure  6.1a), and (iii) a parameterized atmospheric heating anomaly in response to SSTA and atmospheric moisture convergence anomalies in the Gill model framework (see Zebiak & Cane, 1987, and the appendix for details). Importantly, the CZ model focuses on the dynamical coupling processes: SSTA yields wind

stress anomalies, which produce horizontal current and upwelling anomalies, as well as subsurface temperature anomalies that are related to dynamic redistribution of ocean heat and thereby feed back onto SSTA. The complex thermodynamic surface heat flux feedbacks are greatly simplified by assuming a constant Newtonian SSTA damping. By focusing on the dynamical coupling in the tropical Pacific domain (29°S–29°N, 124°E–80°W), this model provides a simple coupled framework that fully embodies the key hypotheses for the coupled dynamical feedbacks contributing to ENSO growth as envisioned by Bjerknes (1969), and the ENSO turnabout mechanism through upper ocean heat content redistribution envisioned by Wyrtki (1985), that allows for ENSO to oscillate between El Niño and La Niña phases. 6.2.2. Linear ENSO Stability Analysis in the CZ Model The CZ model framework has been the foundation for the advancements of ENSO linear instability theory in the past 30 years. The concept of linear ENSO instability helps us to understand the fundamental ENSO dynamics by exploring the most unstable intrinsic mode in a dynamical system described by many simple‐to‐intermediate complexity models. Early work of ENSO instability theory (Gill, 1985; Hirst, 1986; Battisti & Hirst, 1989; Neelin, 1990; Jin & Neelin, 1993a, 1993b; Neelin & Jin, 1993 (together JN93 hereafter); Jin et al., 1994; Jin, 1997a, 1997b; An & Jin, 2000; Fedorov & Philander, 2000, 2001)

124  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

was largely based on simpler versions of the CZ model. For example, by considering the SSTA equation only within the equatorial strip instead of the entire tropical Pacific basin, JN93 developed a stripped‐down version of CZ model with its full ocean wave dynamics component, as described in the appendix. This slightly simplified CZ‐ type model allows for thorough analytical and numerical eigen‐analyses of the leading coupled mode in wide ranges of the parameter spaces and thus a basic understanding of how key coupled processes give rise to ENSO‐ like modes (JN93). By converting the parameter spaces to mean state spaces, An and Jin (2000) and Fedorov and Philander (2000, 2001) used the instability theory based on the JN93 stripped‐down model to investigate the diverse ENSO behaviors simulated in climate models under past, present, and future mean climate conditions. There is one ENSO‐like leading mode that stands out from the large continuum of eigen modes in the various simplified versions of the CZ model, such as the JN93 stripped‐down version and a two‐strip model which further reduced the ocean equatorial wave dynamics within the equatorial and off‐equatorial strips (Jin, 1997b). This type of ENSO eigen mode analysis was carried out in linearized versions of the full CZ model (Bejarano, 2006; Bejarano & Jin, 2008; Xie & Jin, 2018) with respect to a broad range of both parameter space and basic‐state space. In these studies, the leading ENSO mode is largely near critical (i.e., subcritical/supercritical when their growth rates are slightly negative/positive) within a realistic range of the parameter and basic state spaces. When subcritical, it can be easily excited by stochastic forcing, and when supercritical it can be self‐ sustained (linear growth will be constrained by nonlinear dynamic damping). Thus, the leading ENSO mode tends to dominate the internal variability in a manner similar to the observed ENSO behavior. Here we give a brief account of the behaviors of the leading mode in the CZ model based on a recalculation of eigen solutions of the CZ model because the earlier eigen analyses by Bejarano (2006), Bejarano and Jin (2008), as well as Xie and Jin (2018) shared a coding error that corrupted their eigen solutions (it was recently uncovered by Mr. Licheng Geng while working on his PhD thesis on the subject). This error results in an artificial split of the single leading oscillatory ENSO‐like mode into two coexisting leading modes that were referred in Xie and Jin (2018) as EP and CP ENSO modes, as their SST patterns happened to resemble EP and CP El Niño patterns. After the error was cleared, the CZ model in fact allows only a single leading ENSO‐like mode in broad basic state space (Figure  6.2a, b) and parameter space (Figure  6.2c, f), as in the earlier strip‐ down version models of JN93, An and Jin (2000), and Fedorov and Philander (2000). Consistent with the earlier studies, the growth rate and frequency of this leading

ENSO mode vary sensitively in the parameter and basic state spaces (Figure 6.2). Two solutions marked as A and B in Figure 6.2 and 6.2b under relatively stronger (i.e., a stronger mean wind stress and thus colder SST in the cold tongue) and weaker (i.e., a weaker mean wind stress and thus a warmer SST in the cold tongue) cold tongue conditions display a moderate zonal shift in the mature SSTA pattern location, albeit much less than seen in the observations for CP and EP events (Figure 6.3b, f). However, there is a remarkable large difference in the linear ­frequency. It shifts from quasi‐biannual (QB) to quasi‐­ quadrennial (QQ) periodicity ranges when the basic state is changed from a relatively stronger cold‐tongue state (A) to a weaker cold‐tongue state (B). This strong dependence of the ENSO mode on the strength of the cold tongue basic state is important for understanding ENSO complexity as it is partly responsible for the ENSO pattern diversity simulated in CZ models as noted in Bejarano and Jin (2008) and Xie and Jin (2018) in their nonlinear solutions. Further detailed discussion on this subject will be presented in forthcoming papers. 6.2.3. Dynamical Mechanisms Controlling Linear ENSO Mode To gain insight into the dynamics of the leading ENSO mode in the linearized CZ model, one may examine the relative importance of the different terms that contribute to the SSTA tendency of the mode under different basic state conditions (Figure  6.3). Under a relatively weak cold‐tongue basic state (point B in Figure 6.2a), the mode tends to have a QQ periodicity (Figure 6.2b). The dominant contribution for the SSTA growth rate is the thermocline (TH) feedback through upwelling of anomalous subsurface temperature, while the zonal advective (ZA) feedback via anomalous zonal current transport is of secondary importance (Figure 6.3d). In contrast, under a relative stronger cold‐tongue basic state (point A in Figure  6.2a), the mode tends to have a QB periodicity (Figure 6.2b) and the role of the ZA feedback becomes similarly important as the TH feedback (Figure  6.3h). These different roles of ZA and TH were previously noted based on observational data analysis (Kug et  al., 2009; Ren & Jin, 2013; Timmermann et al., 2018). In the CZ model, thermodynamics are implemented as a simple Newtonian SSTA damping, as indicated in Eq.  (A4a). This is a major oversimplification in representing coupled thermodynamics in terms of how ENSO SSTA alters surface thermal fluxes through atmospheric moisture dynamics, radiative processes, and evaporative and sensible heat sources. Extending the CZ model framework to  incorporate physically consistent thermodynamic ­coupling is needed to better understand how coupled thermodynamics may shape the patterns of this leading ENSO mode.

Growth rate

120

(b)

[year–1]

A

110

6

1 0.6 0.4 0.2 0 –0.2

90

[year]

A

5.6

0.8

100

Period

120

1.2 mean wind stress (S; %)

mean wind stress (S; %)

(a)

110

5.2 4.8 4.4 4

100

3.6 3.2 2.8

90

–0.4 B

2.4 B

–0.6

80 145 150 155 135 140 reference thermocline depth (H; m)

Growth rate at A

(c)

160

3 2.7 2.4 2.1 1.8 1.5 1.2 0.9 0.6 0.3 0 –0.3 –0.6

1.2

1.0

0.8

0.6 0.6

0.8

1.0

1.2

135 140 145 150 155 reference thermocline depth (H; m)

Period at A

(d)

[year–1]

1.4 thermodynamic damping (αs)

130

6 5.2

1.2

4.8 4.4 4

1.0

3.6 3.2

0.8

2.8 2.4

0.6

2 0.6

1.4

(f)

[year–1] 3 2.7 2.4 2.1 1.8 1.5 1.2 0.9 0.6 0.3 0 –0.3 –0.6

1.4

1.2

1.0

0.8

0.6 0.6

0.8

1.0

1.2

dynamic coupling efficiency (μ)

0.8

1.0

1.2

1.4

dynamic coupling efficiency (μ)

1.4

Period at B

[year] 6

1.4 thermodynamic damping (αs)

Growth rate at B

[year]

5.6

dynamic coupling efficiency (μ)

(e)

160

1.4 thermodynamic damping (αs)

130

thermodynamic damping (αs)

2

80

5.6 5.2

1.2

4.8 4.4 4

1.0

3.6 3.2

0.8

2.8 2.4

0.6

2 0.6

0.8

1.0

1.2

1.4

dynamic coupling efficiency (μ)

Figure 6.2  Instability of the leading ENSO mode as a function of the (a, b) basic states and (c–f) dynamic coupling efficiency (μ; x‐axis) and thermodynamic damping relative to the default value in Table A (αS; y‐axis). Here, the basic state is characterized by the reference thermocline depth (H) in the model and the percentage change of the mean equatorial wind stress relative to the CZ standard configuration (S). Generally, the cold tongue SST is low when H is small and/or S is large, and vice versa (Xie & Jin, 2018). See Bejarano (2006) for the detailed description of how to produce these basic states. Figures c–f are the instability of the leading ENSO mode when perturbing parameters μ and αS under the basic states at A and B. Values of μ = 1.0 and αS = 1.0 correspond to the default model settings as in Zebiak and Cane (1987). Values of other parameters are listed in Table A. The units of growth rate and period are year–1 and year, respectively. These results were calculated by an eigen analysis of the linearized CZ model.

(b)

10N

QQ regime - EI Ni˜no mature: Phase II 1 0.8 0.6 0.4 0.2 –0.2 –0.4 –0.6 –0.8 –1

10N

EQ

[°C

]

EQ 10S

0.04 N/m

10N

10N

EQ

120E

2 –2 –5 –10 –15 –20

EQ

150E

(c)

20

180

150W

120W

90W

Vocn 0.4 m/s

QQ regime: Ni˜no-3.4 vs hw

10

120E

(d)

150E

180

150W

120W

90W

QQ regime: SSTA, ZA and TH

π

–2 –1.5 –1 –0.5

π/2

IV I

0.5 1 1.5 2

0.5

Phase lag

hw [m]

10S

0

2.5

0

III –10

1.5

2

1 0.5

–π/2

II

0.5 1 1.5

–20 –1

–0.5

0

0.5

–π 120E

1

20 15 10

h

10S

5

ZA TH 1 0.8 0.6 0.4 0.2 –0.2 –0.4 –0.6 –0.8 –1

SSTA [°C]

Vatm

SS TA

10S

]

QQ regime - EI Ni˜no onset: Phase I

[m

(a)

–2

150E

180

150W

120W

90W

Ni˜no-3.4 [°C]

(f)

10N

1 0.8 0.6 0.4 0.2 –0.2 –0.4 –0.6 –0.8 –1

10N

[°C

]

EQ

Vatm 0.04 N/m

SS TA

10S

10N

10N

EQ

2 –2 –5 –10 –15 –20

EQ

150E

(g)

20

180

150W

120W

90W

10S

Vocn 0.4 m/s

QB regime: Ni˜no-3.4 vs hw

120E

(h)

π

150E

180

150W

120W

90W

QB regime: SSTA, ZA and TH ZA TH

–1

10

Phase lag

I hw [m]

0.5 –1

π/2

IV

0

0.5 1

0.5 1 0.5 1

–20 –1

–0.5

0

0.5

1

1

0

–π/2

II

–0.5

0.5 1 1.5

III –10

–π 120E

20 15 10

h

10S

5

150E

180

150W

0.5 –1

120W

–0.5

1 0.8 0.6 0.4 0.2 –0.2 –0.4 –0.6 –0.8 –1

SSTA [°C]

EQ 10S

120E

QB regime - EI Ni˜no mature: Phase II

]

QB regime - EI Ni˜no onset: Phase I

[m

(e)

90W

Ni˜no-3.4 [°C]

Figure 6.3  Schematic diagrams for the leading ENSO mode under weak (a–d) and strong (e–h) cold tongue basic states. In each phase, SST (shadings; °C) and wind stress (vectors; N m–2) anomalies are shown in the upper panel, and thermocline depth (shadings; m) and mixed layer currents (vectors; ms‐1) in the bottom panel. Phases are determined from the (c, g) trajectories of Niño‐3.4 index (°C; x‐axis) and thermocline depth anomalies in the western Pacific (hw; m; y‐axis). See text for the definition of hw. Onset phase I is a three‐month average centered at the time when Niño‐3.4 index ascends cross zero (gray dots), and mature phase II is centered at the time when Niño‐3.4 index reaches the positive peak (red dots). Phases III and IV are opposite to phases I and II and indicate the transition and La Niña phases, respectively. Also shown are the time series of Niño‐3.4 index and hw in the upper‐right corner. (d, h) Hovmöller diagrams of equatorial SSTA for 5°S–5°N (shadings) and the associated zonal advective (ZA; green contours) and thermocline (TH; purple contours) feedbacks for 2°S–2°N (10–1 °C month‐1) that contribute to the evolutions of SSTA. Phase –π to phase π consists of a complete ENSO cycle.

Simple ENSO Models  127

The ENSO eigen mode displays strong sensitivity to changes in key parameters. Two such parameters are chosen as they are known to exhibit substantial ranges in CGCMs and because they have strong impacts on the growth rates and frequencies of the leading ENSO mode. The first is the dynamic coupling efficiency (μ) which is a control parameter multiplied onto the CZ  model’s atmospheric wind stress response to SSTA. The second is the thermodynamic damping (αs) (see Eq. [A4a] and Table  A of the appendix for details). The ENSO mode becomes unstable when the dynamic coupling efficiency increases and/or the thermodynamic damping decreases (Figure  6.2c, e). Interestingly, the growth rate contours are nearly diagonal under both relative weak and strong cold‐tongue basic states, indicating a strong compensation effect on ENSO growth rate from dynamic coupling effi­ ciency and thermodynamic damping, which is consistent with findings of strong error compensation in dynamic coupling and thermodynamic damping for “right” ENSO amplitude simulation in CGCMs for wrong reasons (Bellenger et al., 2014; Karamperidou et al., 2017; Bayr et al., 2018). The periodicity of the mode under different cold tongue strength basic states displays some dependency on these parameter changes as well. The QB periodicity under a strong cold tongue basic state tends to stay largely in the QB range but decreases/increases somewhat when the dynamic coupling and thermodynamic damping increase/decrease (Figure  6.2d). The QQ periodicity under a weak cold tongue basic state increases to a 5‐ to 6‐year range as thermodynamic damping and dynamic coupling weakens (Figure 6.2f). Further work to carefully validate the CZ model against the observed basic state and to obtain best estimates for model parameters is needed to make direct comparisons with observed changes in ENSO behavior. Nevertheless, the results here indicate a significant sensitivity of the leading mode to dynamic coupling and thermodynamic damping. Thus, it is important that models simulate the individual feedbacks correctly in order to faithfully capture both ENSO variance and pattern diversity. The latter still remains a great challenge for state‐of‐the‐art CGCMs, which are known for a cancellation effect between the dynamic coupling and thermodynamic coupling parameters; indeed, many models may simulate an ENSO with realistic growth rate and periodicity, however, due to an incorrect balance of feedbacks (Bellenger et  al., 2014; Karamperidou et al., 2017). The sensitive dependence of the leading ENSO mode on the basic state is conducive for ENSO pattern diversity because relatively small modulations of the basic state by either natural variability or external forcing may give rise to significant changes in ENSO pattern, growth rate, and periodicity. An alternative hypothesis for this ENSO diversity has also been put forward, attributing it to a single oscillatory ENSO mode that undergoes large

nonlinear modifications by processes in the atmosphere and ocean (e.g., Choi et al., 2013; Chen & Majda, 2016, 2017; Takahashi et al., 2019). The relatively short length of the observational record may not be adequate to settle this debate at this point. Further examinations of the existence and behavior of ENSO regimes and their great sensitivity to the climate background state, as well as dynamic and thermodynamic coupling in both CZ‐type models and comprehensive climate models are needed to advance our understanding and simulation of ENSO complexity. 6.3. RECHARGE OSCILLATOR (RO) AND BJERKNES-WYRTKI-JIN (BWJ) INDEX Parallel to the advancements in ENSO instability theory, simple conceptual models have been developed that depict ENSO as an oscillatory phenomenon due to the coupled instability envisioned by Bjerknes (1969), Wyrtki (1985), and Cane and Zebiak (1985). The first conceptual model was the delayed oscillator framework (Suarez & Schopf, 1988; Battisti & Hirst, 1989). This delayed oscillator (DO) paradigm was derived based on numerical simulations from intermediate complexity coupled ocean‐atmosphere models. The essence of this DO model reflects two basic processes that cause ENSO SSTA to grow and to oscillate, respectively: (i) the coupled positive Bjerknes feedback provides the fundamental growth mechanism, while (ii) wind‐driven oceanic Rossby waves and their western boundary reflection into the equatorial Kevin wave give rise to a delayed negative feedback that causes the ENSO phase transition. By realizing that the recharge and discharge of heat content in the equatorial ocean is achieved collectively by the equatorial waves and thus may be better described as an adjustment process instead of an oceanic wave process with a single delay, Jin developed the simple RO paradigm for ENSO (Jin, 1996, 1997a, 1997b)1. The RO model captures the coupled oscillatory instability of ENSO explicitly and succinctly in the simplest possible manner with a two‐degree freedom dynamical system. The simplicity of the RO paradigm allows for explicit quantifications of both ENSO growth rate and periodicity, which are referred to as the Bjerknes and Wyrtki indices, respectively (Jin et al., 2006; Lu et al., 2018), or collectively as the Bjerknes‐Wyrtki‐Jin (BWJ) index. In the next subsection, we briefly discuss the key steps of deriving the RO, and the formulations of the BWJ index that encompass all the main processes important for ENSO linear growth rate and periodicity.

1  The term recharge oscillator was coined by Mark Cane, who suggested it in his review of the papers by Jin (1997a, 1997b).

128  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Table A  Definitions and values of model variables and parameters Variable/ Parameter εa ua, va β0y ca β

Definition/Default Values

T,c

Linear dissipation in the atmosphere, εa = (2 days)–1 Zonal and meridional winds in the lower atmosphere layer Coriolis parameter at distance y away from the equator Propagation speed of atmospheric Kelvin wave, ca = 60 m s–1 Atmospheric heating parameter, β=1.6×104 m2 s–2 Mean SST and atmospheric convergence

T, c u, v τx,τy h H S

SST and atmospheric convergence anomalies Zonal and meridional ocean currents in the upper ocean layers Zonal and meridional wind stresses Fluctuations of thermocline depth Reference thermocline depth Percentage of standard mean wind stress during 1980–2000 (see Figure 6.2)

g

Reduced gravity acceleration, g = g

∆ρ r rs

Density difference between the upper and motionless layers Ocean dynamical damping rate, (2.5 years)–1 Dynamical damping rate of the velocity shear, rs = (2 days)–1

ws

Ocean upwelling, w s

M(x) TZ

H1

Heaviside function, M(x) =

u1 x

v1 y

x, x 0, x

0 0

Te γ αs

Mean vertical temperature gradient in the ocean Entrainment temperature anomaly Entrainment efficiency: γ is set to 0.6 instead of 0.75. Thermodynamic damping rate, αs = (125 days)–1

h

Mean thermocline depth along the equator

T1, b1, T2, b2

Constants controlling the subsurface temperature anomalies, T1 = 28°C, T2 = –40°C, b1 = (80 m)–1, b2 = (33 m)–1

6.3.1. The Formulations of the RO Model and the BWJ Index Motivated by well‐defined spatial patterns from both numerical and analytic eigen solutions of ENSO in broad parameter spaces in JN93, Jin (1997a) further reduced these linear ENSO dynamics into two prognostic equations for eastern equatorial Pacific SSTA (TE) and warm pool thermocline depth anomaly (hw):

dTE dt

RTE

dhw dt

hw

F1hw , (6.1)

F2TE .



(6.2)

Jin et al. (2006) demonstrated using the CZ model that the growth rate of the most unstable ENSO mode in the linear CZ model can be estimated by deriving the approximate but analytical form of the processes that are encompassed in the parameters R and ε. The former includes the six dominant positive and negative feedbacks that determine the growth of eastern Pacific SSTAs. The latter denotes the dynamic adjustment rate of the equatorial warm pool thermocline depth. The linear frequency of the RO can be evaluated as well using all four parameters. Parameter F1 represents the ocean dynamic feedback from anomalous zonal advection and vertical heat advection associated with the discharge/recharge of equatorial warm pool heat content. Parameter F2 indicates the efficiency of the recharge/discharge driven by equatorial wind stress anomalies induced by the ENSO SSTA. The  complex eigenvalue of the RO can be written as follows:

Simple ENSO Models  129

mean upwelling (also referred to as dynamical damping, DD), the thermocline feedback (TH), the zonal advective 2 4 feedback (ZA), the meridional advective feedback (MA), where the real part of the index was referred to as the ENSO the vertical advective feedback (VA), and the thermodyBjerknes stability (BJ) index in Jin et  al. (2006), and the namic damping (TD), respectively. Lx and Ly are the imaginary part as the Wyrtki frequency (WF) index in Lu effective zonal and meridional scales that reflect box et al. (2018). Together we refer to the complex eigenvalue as averaged quantities. The factor y in the second term on the Bjerknes‐Wyrtki‐Jin (BWJ) index for the ENSO oscilla- the RHS denotes the meridional coordinate, which tory instability as it was first derived in its approximate and comes from the assumption that the SSTAs have a quasi‐analytical formulation in Jin (1997a, 1997b). It has Gaussian‐like meridional pattern with an e‐folding decay x 0 1, been demonstrated to be useful for understanding and eval- (Jin et  al., 2006), M x is a Heaviside uating ENSO’s growth rate and periodicity in both reanal0, otherwise, ysis data and CGCM simulations. step function that only considers regions with upward vertical motion, Hm is the effective depth for the vertical 6.3.2. The Derivation of the BWJ Index advection (set to 50 m), and Tsub is the subsurface temperature at 75 m depth. As for the CZ model (Zebiak & For simplicity, we use the Niño‐3 index (SSTAs aver- Cane, 1987), the parameter γ (set to 0.75) measures the aged over 5°S–5°N, 150°–90°W) to represent ENSO SST effectiveness of vertical entrainment using box‐averaged anomalies. The volume‐averaged mixed‐layer tempera- quantities to represent the average vertical advection ture anomaly tendency equation in the Niño‐3 region can from gridded data. The nonlinear and subgrid terms will be written as follows: be considered in a later subsection. Following Jin (1997b) and Jin et al. (2006), we give a T brief account of the key assumptions for deriving the RO u xT v yT w zT u xT v yT w zT    model and the BWJ index using SODA version 3.3.1 t LDH reanalysis data (Carton et al., 2018) for the period 1980– u xT v yT w zT Q / oC p H R ,        2015. The Niño‐3 region (5°S–5°N, 150°–90°W) is used to SG NDH TDH represent ENSO SST variability (Figure 6.4a). The zonal (6.4) wind response to the positive SSTA shows strong anomawhere u, v, and w are the zonal, meridional, and vertical lous westerlies in the western and central Pacific and weak velocities, T the potential temperature, Q the net surface anomalous easterlies in the eastern Pacific (Figure 6.4a). heat flux into the ocean, H the mixed‐layer depth (set to This Gill-Matsuno response (Matsuno, 1966; Gill, 1980) 50m), Cp the specific heat of seawater at constant pressure delineates the different zonal wind responses to the west (set to 3994 J kg‐1 K‐1), and ρ0 the reference density of sea- and east of the heating source. We use the Sverdrup water (set to 1025 kg m‐3). R denotes subgrid‐scale balance to relate the thermocline anomalies in the western processes and overbars denote climatological means. The and eastern equatorial box to zonal wind stress anomalies terms on the right‐hand side (RHS) of Eq. (6.4) represent in the central Pacific (Figure  6.4a). Following the CZ linear dynamic heating (LDH), nonlinear dynamical model framework, the anomalous upwelling in the eastern heating (NDH), thermodynamic heating (TDH), and Pacific consists of both Ekman upwelling and wave‐ subgrid‐scale contributions (SG) (e.g., oceanic turbulent induced upwelling; the former is related to the local zonal mixing, nonlinear heating due to tropical instability wave wind stress anomalies, whereas the latter is related to the nonlocal thermocline tendency, which may be further and eddy activity) (An & Jin, 2004). Following Jin et al. (2006) and also denoting 〈T  〉 as TE, related to warm pool heat content and ENSO SST. The mixed‐layer equatorial zonal and meridional currents the above equation can then be linearized as follows: and upwelling anomalies in the Niño‐3 box all have both Ekman and geostrophic components. The Ekman part is M w w M w w u 2 yv TE TE Tsub 2 determined by the local wind stress in the Niño‐3 box, t Lx Ly Hm Hm      whereas the geostrophic part can be approximately related TH DD to remote zonal wind stress in the central Pacific and the Q / 0C p H m . u xT v A yT w M w zT thermocline depth in the equatorial Pacific (Jin, 1997b;       A  TD VA ZA MA Jin & An, 1999). Subsurface temperature anomalies in  (6.5) the eastern Pacific are related to the thermocline anomaly The terms on the RHS of Eq. (6.5) represent advection locally (Figure  6.4a). Finally, the net thermodynamic due to mean zonal and meridional currents as well as heating from surface heat flux anomalies is linearized in BWJ

R

i F1F2

R

2

,

(6.3)

10°N

Tsub mean [degC] 28 28

EQ

2

28

20

5°N

30 10°N 28 26 5°N 24 EQ 22 20 18 5°S 16 14 10°S

20 18

5°S

28

28

10°S 10°N

D20 mean [m] 90

150

5°N

150

EQ

90

5°S u mean [10–1m s–1] 0

0 2

EQ

2

2 0

2

2

0

–2

0 2

2

0

2

0 2

2

5°S

4 2 0 –2 –4

2

2

2

0

10°S v mean [10–1m s–1]

5°N

0

EQ

0

0.4

0.4

5°S

4

0.4

0.

0

0.4 4

0.

10°S 10°N

w mean [10–5m s–1]

EQ 5°S 10°S taux mean [10–2N m–2]

5°N 4

2 4

5°S 10°S 10°N

Q mean [W m–2]

5°N

0

40

40

EQ

80

150°E

180°

150°W

0 120 80 40

80

120°W

40

2

2

D20 anom [m]

90°W

0

8 8

16 8 0 –8 –16

8 16

8 8

u anom [10–1m s–1] 0 0.4

0.4

5°S

0.4

0 0.4

0.4 0.4

0 0.4 0

0.4 0.4

0.8

0.4 0.4

10°S 10°N

3 2 1 0 –1 –2 –3 3 2 1 0 –1 –2 –3

0

v anom [10–1m s–1]

EQ 0 –0.4 5°S –0.8 10°S

0.1

w anom [10–5m s–1]

0.6 0.4 0.2 0 –0.2 –0.4 –0.6

taux anom [10–2N m–2] 0.4 0.8

1.2

1.8 1.2 0.6 0 –0.6 –1.2 –1.8

0

–0.4

1.2 0.8

Q anom [W m–2] 120 10°N 80 5°N 40 0 EQ –40 –80 5°S –120 10°S 120°E 150°E

1.6 1.2 0.8 0.4 0 –0.4 –0.8 –1.2 –1.6 0.3 0.2 0.1 0 –0.1 –0.2 –0.3

5°N

0

40

5°S 10°S 120°E

EQ

8 10°N 6 4 5°N 2 0 EQ –2 –4 5°S –6 –8 10°S

2

EQ

5°N

2.4 10°N 1.6 5°N 0.8 0 EQ –0.8 5°S –1.6 –2.4 10°S

5°N

10°N

0.8 0.4

10°N

Tsub anom [degC]

0.4

5°N

10°N

150

210

10°S 10°N

210 10°N 180 5°N 150 120 EQ 90 5°S 60 30 10°S

1

4

10°S

1

1 1

0

28

0

0

22

5°S

24

26

0.4

26

28

EQ

0.

30 10°N 28 26 5°N 24 EQ 22 20 18 5°S 16 14 10°S

0

5°N

SODA331 lag = 0

T anom [degC]

0.8

10°N

SODA331

T mean [degC]

–10

–10

–10 –20

–10

–20

–10

0

180°

150°W

120°W

90°W

Figure 6.4  Equatorial Pacific annual mean climatology and relevant anomalies during El Niño peak phase. The El Niño peak state is represented by the regressed anomalies onto the normalized volume averaged mixed layer (0–50 m) ocean temperature anomaly over the Niño‐3 region. Panels from top to bottom are horizontal distributions of mixed layer temperature, subsurface temperature at 75 m depth, thermocline depth (20°C isotherm depth), mixed layer zonal current, mixed layer meridional current, vertical motion at bottom of mixed layer (at 50 m), zonal wind stress, and net heat flux into the ocean. The Niño‐3 region (5°S–5°N, 150°–90°W), hw (5°S–5°N, 120°E–155°W), he (5°S–5°N, 155°–80°W), and [τx] (5°S–5°N, 150°E–130°W) regions are indicated by black boxes.

30 20 10 0 –10 –20 –30

Simple ENSO Models  131

terms of local SSTA (Figure 6.4a). These quasi‐balance approximation linear relationships can be derived from the CZ framework and be expressed as follows:

x



x

TE ,

a

Using the linear relationships in Eqs. (6.6)–(6.13), those six RHS terms can be expressed linearly in TE and hw, which yields formulations for R and F1 as follows:

(6.6)



R

* a E

T , (6.7)

M w w M w w 2 yv u a h ah Hm Lx L2y Hm      TH DD

* yT xT a vr a vl   A

a ur

he



w

wr

u v

ur

A

vr

x

x

h

Q /

x

wl

x

,

h ,

wh w

h ,

uh w

x

vl

(6.8)



x

ul

Tsub



hw

TE ,

(6.13)

where [τx] denotes central equatorial Pacific zonal wind stress anomalies (5°S–5°N,150°E–130°W), 〈τx〉 the averaged zonal wind stress anomalies in the Niño‐3 region, he the averaged thermocline depth anomalies over the eastern Pacific box (5°S–5°N,155°–80°W), and hw the averaged thermocline depth anomalies over the western Pacific box (5°S–5°N, 120°E–155°W). By combining Eqs. (6.6) and (6.8), we get the following linear relation:

he

hw

T

TE .

(6.15)

a wr a wl M w zT  ,   TD VA

F1

 

xT

uh

 ZA

yT A    vh

MA

M w w M w zT ah .    H m  VA wh

TH

h , (6.11)

ah he , (6.12)

C pH

MA

*

vh w

x

0

ZA

  

(6.9) (6.10)

*

a ul



By considering both Ekman and quasi‐balanced wave dynamics induced zonal and meridional current anomalies and upwelling anomalies in Eqs. (6.9)–(6.11), our new formulation established here is more complete in assessing the roles of Ekman feedback than those in Jin et al. (2006), Kim and Jin (2011), and Lu et al. (2018) as it now appears consistently in the zonal, meridional, and vertical advection terms. Following the equatorial strip approximation to the ocean dynamics equations (Eqs. [3.4]–[3.5] in Jin, 1997b), the slow adjustment of the warm pool ocean heat content described by linear reduced gravity wave ocean dynamics can be further systematically reduced into its simplest possible form as

(6.14)

The feedback coefficients are determined using linear regressions. Figure  6.5 shows time series and scatter plots of the linear relations in Eqs. (6.6)–(6.13) obtained from SODA reanalysis. The linear balance equation for [τx] holds very well with a correlation coefficient between [τx] and μaTE of 0.81 (Figure 6.5a). However, the linear relation for 〈τx〉 is very weak (Figure 6.5b), partly because of its sensitivity to the SSTA pattern according to the Gill-Matsuno response (Gill, 1980). The Sverdrup balance, subsurface temperature anomalies, and surface heat flux anomalies relations hold remarkably well with correlation coefficients of 0.83, 0.97, and 0.87, respectively (Figure 6.5c, g, h). The linear closure relationships for ocean zonal and meridional currents and upwelling are slightly weaker, with correlation coefficients of 0.70, 0.87, and 0.80, respectively (Figure  6.5e, f, d), because of significant uncertainties in ocean current data (Hayashi & Jin, 2017) and oceanic nonlinearities (Dijkstra, 2005). The estimated parameters are listed in Table 6.1.

 (6.16)



dhw dt

hw

hw

x

F2TE ,



(6.17)

where ε = (1 − rwre)C/2L, F2 = κμa, and κ = (θ − re)L/4ρHC2. Here, ε is the slow adjustment scale, which depends on the Kelvin wave speed (C), the Pacific basin width (L), and constant wave reflection efficiencies at the lateral boundaries (rw, re). Parameter κ is the warm pool ocean heat content recharge/discharge efficiency. It depends on the proportionality of wind stress curl off the equator to the wind stress on the equator (θ), and the eastern boundary reflection efficiency (Jin, 1997b). Equation (6.17) can be solved analytically by presenting heat content information as the integral effect of the finite memory of SSTA, i.e.,



hˆw

t

F2 e

s t

TE s ds.



(6.18)

Parameters ε and F2 depend on the ENSO wind stress pattern, including its meridional width and longitudinal

–0.03

μa = 7.53e–03 1995

2000

2005

2010

2015

–6

–3

Year

0.04

〈τx〉 μa*TE

0.02

0 3 TE (°C)

6

0.4

R = 0.07

0.02 0

0 –0.02 –0.04 1980

1985

1990

1995

2000

2005

2010

–6

2015

–3

Year he – hw βh [τx]

60

he – hw (m)

–0.02

μ*a = 3.65e–04

90

0 3 TE (°C)

6

60 30

0

0 βh = 1383.9

–30

〈w〉(10–5 m s–2)

(d)

1985

1990

1995

2000

2005

2010

2015

Year

0.5

〈w〉 Fit 〈w〉

–30 –60 –0.06 –0.03 0 0.03 0.06 –1 [τx](°C month ) 0.5 R = 0.80 0

0

–0.5 1980

〈u〉(m s–1)

(e)

βwl = 7.77, βwr = 4.58 βwh = –6.75e–03 1985

1990

1995

2000

2005

2010

2015

Year

0.5

–0.5 –0.5 0 0.5 –βwl〈τx〉 – βwr [τx] + βwh hw (m s–1) 0.5 〈u〉 R = 0.70 Fit 〈u〉 0

0

–0.5 1980

(f)

βul = 0.39, βur = 4.91 βuh = 4.98e–03 1985

1990

1995

〈vA〉(m s–1)

2000

2005

2010

2015

Year

0.05

〈vA〉 Fit 〈vA〉

–0.5 –0.5 0 0.5 βul 〈τx〉 + βur [τx] + βuh hw (m s–1) 0.05 R = 0.87 0

0

–0.05 1980

〈Tsub〉(°C)

(g)

(h)

βvl = –0.95, βvr = –0.34 βvh = –1.73e–04 1985

1990

1995

2000

2005

2010

2015

Year

10

〈Tsub〉 ahhe

5

–0.05 –0.5 0 0.5 βvl 〈τx〉 + βvr [τx] + βvh hw (m s–1) 10 R = 0.97 5 0

0 –5 –10 1980

–0.04 90

R = 0.83

30

–60 1980

–0.06

–5

ah = 0.138 1985

1990

1995

2000

2005

2010

–60 –30

2015

Year

1.5

〈Q〉 –αTE

0.5

–10

0 30 60 he (m)

1.5

R = 0.87

0.5 –0.5

–0.5 α = 0.188 –1.5 1980

1985

1990

1995

2000 Year

2005

2010

2015

〈τx〉(N m–2)

1990

he – hw (m)

1985

〈w〉(10–5 m s–2)

–0.03

[τx](N m–2)

0

〈u〉(m s–1)

〈τx〉(N m–2)

0.03

0

(b)

〈Q〉(°C month–1)

0.06

R = 0.81

–6

–3

0 3 –TE (°C)

〈vA〉(m s–1)

0.03

–0.06 1980

(c)

[τx] μaTE

〈Tsub〉(°C)

0.06

6

〈Q〉(°C month–1)

[τx](N m–2)

(a)

–1.5

Figure 6.5  Time series (left panels) and scatter plots (right panels) from the balanced relations for deriving and * determining the parameters of the recharge oscillator (RO) model. (a) [τx] = μaTE, (b) x aTE, (c) he − hw = βh[τx], (d) 〈w〉 =  − βwr[τx] − βwl〈τx〉 + βwhhw, (e) 〈u〉 = βur[τx] + βul〈τx〉 + βuhhw, (f) 〈v〉A = βvr[τx] + βvl〈τx〉 + βvhhw, (g) 〈Tsub〉 = ahhe, and (h) 〈Q〉/(ρ0CpH) =  − αTE. The black and red curves denote the observed monthly anomalies and fitted monthly anomalies, respectively. The feedback coefficients and correlation coefficients between observed and fitted anomalies are indicated.

Simple ENSO Models  133 Table 6.1  The annual mean background state parameters, feedback coefficients, and Bjerknes‐Wyrtki‐Jin index for Niño‐3 region estimated using the SODA‐3.3.1 monthly data. Symbol

Value

Unit

Note

H

50

m

Mixed layer depth

Lx

6.67×10

m

Effective zonal scale

Ly

1.11×10

m

Effective meridional scale

Hm

50

m

Effective vertical scale for vertical advection

γ

0.75

1

Mixing efficiency of 〈T〉 − 〈Tsub〉

‐1.31×10

s

Scaled mean zonal current

‐1.36×10‐8

s‐1

Scaled mean meridional current

7.62×10‐8

s‐1

Scaled mean upwelling

T

‐5.58×10‐7

°C m‐1

Zonal gradient of mean temperature

T

4.13×10‐6

°C m‐1

Antisymmetric competent of meridional gradient of mean temperature

4.84×10‐2

°C m‐1

Vertical gradient of mean temperature

μa

7.53×10‐3

N m‐2 °C‐1

Eq. (6.6)

μ

‐3.65×10‐4

N m‐2 °C‐1

Eq. (6.7)

βh

1383.9

m3 N‐1

Eq. (6.8)

βur

4.91

m3 s‐1 N‐1

Eq. (6.10)

βul

0.39

m3 s‐1 N‐1

Eq. (6.10)

βuh

4.98×10‐3

s‐1

Eq. (6.10)

βvr

‐0.34

m3 s‐1 N‐1

Eq. (6.11)

βvl

‐0.95

m3 s‐1 N‐1

Eq. (6.11)

βvh

‐1.73×10‐4

s‐1

Eq. (6.11)

βwr

4.58×10‐5

m3 s‐1 N‐1

Eq. (6.9)

βwl

7.77×10‐5

m3 s‐1 N‐1

Eq. (6.9)

βwh

‐6.75×10‐8

m3 s‐1 N‐1

Eq. (6.9)

ah

0.138

°C m‐1

Eq. (6.12)

α

0.188

month‐1

Eq. (6.13)

βT

14.7

m°C‐1

Eq. (6.14)

R

0.79

year‐1

Eq. (6.15)

F1

0.46

°C m‐1 year‐1

Eq. (6.16)

ε

1.40

year‐1

Eq. (6.18)

F2

18.2

m °C‐1 year‐1

Eq. (6.18)

BJ

‐0.30

year‐1

(R − ε)/2

WJ

2.34

year

4 / 4FF 1 2

WJBF

2.17

year

2 / FF 1 2

6 6

‐8

u / Lx 2yv / L2y M w w / Hm x

y A

M w

* a

T

z

‐1

location as formulated in Jin (1997b). Here, they are calculated by least square fitting of observed hw and TE. Using SODA reanalysis data, this fitting method

R

2

results in ε = 1.40 year‐1 and F2 = 18.2 m °C‐1 year–1, with a ­ correlation coefficient between hw and hˆw of ~0.89 (Figure 6.6a).

134  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

hw ˆhw

0

0 –30 1995

2000 Year

2005

2010

–60 –30

2015

0

–60 30 60

ˆhw (m)

90 30

90

R = 0.94

he – hw βTTE

60

60 30 0

0 βT = 14.7

–30 1985

1990

1995

2000 Year

2005

2010

–30 –6 –3

2015

60

0 3 TE (°C)

30

he (m)

0

0 he βTTE + ˆhw

–30 –60 1980

1985

1990

1995

2000 Year

2005

2010

2015

–30 –60 0 30 60 ˆ w (m) βTTE + h

–60 –30

30

hm (m)

15

0

0 hm ˆw βTTE /2 + h

–15 –30 1980

1985

1990

1995

2000 Year

2005

2010

30

–15 –30 –30 –15 0 15 30 βTTE /2 + ˆhw (m)

2015

hm ˆh m

15 hm (m)

30

R = 0.80

15

(e)

–60

60

R = 0.88

30

(d)

6

he – hw (m)

1990

30

R = 0.59

15 0

0 –15

–15

εm = 0.24, F2m = 4.5

–30 1980

1985

1990

1995

2000 Year

2005

2010

he (m)

he – hw (m)

1985

–60 1980

(c)

–30

ε = 1.40, F2 = 18.2

–60 1980

(b)

30

hm (m)

hw (m)

30

60

R = 0.89

hw (m)

60

–30 –15

2015

hm (m)

(a)

0

–30 15 30

ˆh (m) w

Figure 6.6  Time series (left panels) and scatter plots (right panels) from the thermocline dynamics of the RO model. (a) Western Pacific thermocline depth anomalies hw and RO hˆw using Eq. (6.18), (b) thermocline depth tilt h  − h and β T , (c) eastern Pacific thermocline depth anomalies h and T hˆ , (d) basin mean thermocline e

w

T E

T E

e

w

depth anomalies hm and TTE / 2 hˆw, (e) same as (d) but for RO fitted hˆm using Eq. (6.20). The feedback coefficients and correlation coefficients between observed and fitted anomalies are indicated.

Warm pool heat content discharge and equatorial wide heat content discharge (hm) are related (Wyrtki, 1985; Jin, 1997a, 1997b):

hm

he

hw / 2 t



F2 e T TE / 2

s t

TE s ds.

(6.19)

Simple ENSO Models  135

The above expression for basinwide heat content is similar to Eq. (8.7) of Fedorov (2010), who derived a slightly more complex formulation also using both Sverdrup balance and a slow oceanic adjustment process under a weak oceanic damping assumption. As shown in Figure 6.6a–d, the slow ocean heat content adjustment and equatorial Sverdrup balance capture the relation of ENSO SST with western, eastern, and zonal mean equatorial heat content remarkably well, with improved correlations compared to previous studies (Fedorov, 2010; Izumo et al., 2019). Both warm pool heat content and basinwide heat content have been used in the past to describe the recharge/discharge process (Jin, 1997a; Lu et  al., 2018). In order to test which one is more suitable, we use a similar form of Eq. (6.2) but with two different parameters for basinwide heat content:

hˆm

t

F2m

e

m (s

t)

TE ( s )ds

(6.20)

The best fit gives an adjustment time scale of 0.24 year–1, which is much longer than that for western equatorial heat content. This is consistent with the notion that equatorial heat content has a slow adjustment time scale (Burgers et al., 2005). However, the resulting hˆm only captures the very slow variation of hm with a correlation coefficient of about 0.59 (Figure 6.6e), which is a significantly less effective expression for equatorial heat content than the expression that combines the Sverdrup balance and the western Pacific warm‐pool heat content (Eq. 6.19). This result is consistent with recent studies (Neske & McGregor, 2018; Planton et al., 2018; Izumo et al., 2019), suggesting that western equatorial heat content is better suited to describe the recharge/discharge process. With these analytical formulations of ε and F2, we now derive all the parameters in the BWJ index for the RO under a number of reasonable quasi‐balance approximations. For the SODA reanalysis mean state, the BWJ index yields an ENSO growth rate of −0.30 year–1 and a period of 2.34 years (Table 6.1), which indicates that ENSO linear coupled dynamics are very close to criticality. This is significantly different from a much stronger damping rate estimated from linear inverse modeling. This difference largely comes from the fact that the nonlinear damping from both deterministic and stochastic nonlinear processes are already included in the estimated ENSO growth rate from linear inverse modeling by construction but not included in the BWJ index. Furthermore, the estimated linear period is substantially shorter than the average observed ENSO period. This periodicity is sensitive to the characterization of the basic state, the details of parameterizing the subsurface temperature anomalies, and the nonlinear corrections (see the appendix of Jin, 1997a). Nevertheless, this linear analysis of the BWJ index is roughly consistent with fitting

the linear RO model directly with observational data, although the latter method gives a more negative linear growth rate but longer period. The advantage of using the BWJ index comes from its decomposition of the contributions to ENSO growth rate and frequency from different coupled processes. As shown in Figure  6.7a, the TH feedback is the largest contributor to the instability growth, and the ZA feedback is the second largest. The TD damping is the largest damping term and the DD the second largest. These results are consistent with the results obtained from the SODA 2.0.2 reanalysis (Kim et al., 2014). 6.3.3. Seasonal Modulation of BWJ Index In this subsection, we consider seasonal modulations of the RO due to seasonality in the background state and feedback coefficients. As shown in Figure  6.7b, seasonality in the Sverdrup balance, i.e., the response of the ther­ mocline in the western equatorial Pacific to central Pacific wind stress anomalies (βh), has a positive TH feedback maximum during boreal winter and a minimum during late spring and summer. The seasonal response of thermodynamic heating to local SST (α) results in a negative TD feedback that has its maximum during late winter and spring and a minimum during fall. Seasonality in the strength of the mean ocean currents results in a negative MA feedback maximum during spring and a minimum during late summer and fall. The seasonal strength of zonal and meridional advection feedbacks results in a positive ZA feedback maximum during fall and a minimum during late winter and spring. The joint effect is a seasonal growth rate that is maximum during boreal late summer and minimum during spring, consistent with the results of Stein et al. (2014). The seasonal processes described have been noted to play an important role both in the seasonal synchronization of ENSO (Stein et  al., 2014) and the ENSO spring auto‐correlation barrier (Levine & McPhaden, 2016). There is also strong seasonality in the Wyrtki index, although the actual impact of seasonality on ENSO main periodicity should be properly assessed using the Floquet exponent analysis (Jin et al., 1996). Additional discussion of ENSO seasonality is given in section 6.4. 6.3.4. Nonlinear Dynamical Heating, Subgrid Processes, and Noise Forcing of ENSO Next, we decompose the SSTA equation as TE t

TE t

D

TE t

LD LN TD TN N

ND NN SGD SGN , (6.21)

136  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE 3

3

2

2.5

1

2 Period [year]

Growth Rate [year–1]

(a)

0 –1 –2 –3

DD

TD

ZA

MA

VA

TH

R

–ε

0

BJ

WJ

WJBF

WJ

WJBF

4

5 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

4 3 Growth Rate [year–1]

1 0.5

2 1

3 Period [year]

(b)

1.5

0

2

1

–1 –2 –3

DD

TD

ZA

MA

VA

TH

R

–ε

0

BJ

Figure 6.7  The Bjierknes‐Wyrtki‐Jin index and its individual components. (a) The estimates using the annual mean background state and feedback coefficients derived from monthly data. (b) The estimates using each month background state and feedback coefficients derived from data for each month. DD represents the dynamic damping, TD the thermodynamic damping, ZA the zonal advective feedback, MA the meridional advective feedback, VA the vertical advection feedback, TH the thermocline feedback, R = MA+TD+ZA+YA+TH+EK, ε represents dynamic adjustment rate of the equatorial warm‐pool thermocline depth, BJ is the Bjerknes stability index (R‐ε)/2, WJ represents Wyrtki period index, and WJBF is the period index ignoring contributions of R and ε, i.e., 2 / FF 1 2.

where

TE t

TE t

denote the total determin-



istic feedbacks and noise, LD and LN denote linear dynamic deterministic feedbacks and noise, TD and TN denote thermodynamic deterministic feedback and noise, ND and NN denote nonlinear dynamic deterministic terms and noise, and SGD and SGN denote subgrid dynamic deterministic terms and noise, respectively. Based on the RO framework, the linear deterministic terms can be written as



D

and

N





LD

R TD

TE

F1hw , (6.22)

TE . (6.23)

ND C N1TE2 C N 2TE hw C N 3TE3 C N 4TE2 hw C N 5TE hw2 , SGD CS1TE CS 2 hw CS3TE2 CS 4TE hw CS5TE3 CS6TE2 hw CS7TE hw2 .

(6.24)

(6.25)

Simple ENSO Models  137

The oceanic mixing and subgrid feedbacks included in the residual contribute to the linear feedbacks. This calls for a more thorough study of how these processes contribute to basic ENSO dynamics. Moreover, by fitting the NDH and SG with the form of Eqs. (6.24) and (6.25), we note that there are three cubic and two quadratic nonlinear terms that tend to dominate the ND and SGD terms. The coefficients are listed in Tables  6.2 and 6.3. This finding suggests that the nonlinear deterministic dynamics and nonlinearity from the subgrid processes may both contribute to the nonlinearity of ENSO. As shown in Figure 6.8, we find a correlation coefficient of R(LD, LDH) = 0.57. The thermodynamic term is captured remarkably well with a correlation coefficient of R(TD, TDH) = 0.87. By least square fitting NDH using Eq. (6.24), we find a correlation of R(ND, NDH) = 0.55, and by least square fitting SG using Eq. (6.25), we find a correlation of R(SGD, SG) = 0.49. The nonlinear terms, particularly the substantial cubic nonlinear terms, serve as nonlinear damping, whereas the relative weak quadratic nonlinear terms may contribute to ENSO asymmetry, all of which will be demonstrated in section 6.4. Similarly, we decompose the warm pool thermocline depth equation into the deterministic and noise terms as



hw t

hw t

hw t

D

, N



(6.26)

where the deterministic part can be written as



hw t

hw

F2TE .

D



(6.27)

In addition to the linear and nonlinear deterministic dynamics, we can also assess the noise forcing terms:

LN

LDH LD,

(6.28)

TN

TDH TD,

(6.29)

NN



NDH ND,

(6.30)

SGN SG SGD.

(6.31)

It is clear that the total noise forcing has considerable amplitude on time scales much faster than the ENSO time scale. However, this high‐frequency noise forcing estimated in this low‐order system is not at all effective in exciting low‐ frequency (i.e., interannual) ENSO variability. Only the low‐frequency part of the noise forcing can effectively drive ENSO and thereby affect ENSO amplitude. It is also worthy to mention that when the low‐frequency component of the noise is truly a part of near white noise, it has no predictability, Therefore, the predictability of ENSO largely resides in its linear and nonlinear deterministic dynamics. Figure  6.9 illustrates the linear, nonlinear, and noise forcing tendency vectors in RO phase space. The rotation and divergence of vectors demonstrate the oscillator and growth of ENSO. As shown in Figure 6.9, the linear deterministic vectors demonstrate an obvious clockwise rotation. In contrast, the rotation signal over the nonlinear deterministic and noise forcing tendency vectors are less clear; this indicates that ENSO oscillation is dominated by linear deterministic dynamics. 6.3.5. On the Relationship of the RO and DO Paradigms The DO and RO models for ENSO have been considered as two different but equivalent conceptual models for the basic dynamics of ENSO (Jin, 1997b; Fedorov, 2010). Both models are derived from the CZ framework. The only key difference between the two resides in the consideration of how the equatorial warm pool heat content or the thermocline respond to ENSO wind forcing. In the RO model, the warm pool heat content is governed by adjustment equation (6.13) with its adjustment time scale controlled by both eastern and western boundary reflections, wave propagation time scale, and the mass discharge/recharge due to the near‐equatorial

Table 6.2  The coefficients of nonlinear deterministic quadratic and cubic terms. Symbol

Value

Unit

CN1

3.68 + 21.61 cos ωat − 3.99 sin ωat

10 °C year

CN2

−1.48 + 0.83 cos ωat − 0.56 sin ωat

10‐2 m‐1 year‐1

TEhw

CN3

−27.67 − 2.93 cos ωat − 7.37 sin ωat

10 °C year

TE3

CN4

−5.01 − 0.69 cos ωat + 1.43 sin ωat

10‐2 °C‐1 m‐1 year‐1

TE2hw

CN5

−0.41 + 0.07 cos ωat + 0.33 sin ωat

10‐2 m‐2 year‐1

TEhw2

Note. ωa is the annual frequency.

‐2

‐2

Terms ‐1

‐2

‐1

‐1

TE2

138  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Table 6.3  The coefficients of subgrid deterministic linear, quadratic, and cubic terms. Symbol

Value

Unit

Terms

CS1

1.37 + 0.42 cos ωat + 0.31 sin ωat

year

CS2

−3.42 + 2.82 cos ωat − 1.15 sin ωat

10 °C m year

CS3

−26.69 − 12.41 cos ωat + 34.56 sin ωat

10 °C year

CS4

−14.79 + 14.99 cos ωat + 1.25 sin ωat

10‐2 m‐1 year‐1

TEhw

CS5

−0.14 + 0.69 cos ωat − 0.30 sin ωat

10 °C year

TE3

CS6

4.64 − 0.42 cos ωat + 6.75 sin ωat

10‐2 °C‐1 m‐1 year‐1

TE2hw

CS7

−0.19 + 0.57 cos ωat + 0.58 sin ωat

10‐2 m‐2 year‐1

TEhw2

TE

‐1

‐2

‐2

‐2

‐1

‐1

‐2

hw

‐1

TE2

‐1

‐1

Note. ωa is the annual frequency.

wind stress curl and wind stress longitudinal location (Jin, 1997b). In contrast, the DO as formulated in Battisti and Hirst (1989) assumed in an ad hoc way that ENSO‐ induced equatorial wind anomalies in the central Pacific drive an equatorial Rossby wave response that is reflected into an equatorial Kelvin wave and determines the western equatorial thermocline depth without any consideration of eastern boundary reflection, leading to the following simple equation:

hw

aw

C

t

.



(6.32)

There are two parameters in each model. The adjustment time scale and recharge/discharge efficiency in the RO model can both be systematically derived as shown in Jin (1997b). Both the adjustment dynamics formulation (RO) and the delayed response formulation (DO) are of comparable effectiveness in capturing the warm pool heat content evolution to a large extent, but adjustment dynamics not only consider the wind stress curl effect explicitly, but also filter out high frequency noise in the wind stress forcing effectively. In contrast, the delayed response dynamics are less clear on the role of wind stress curl and cannot filter out noise in the wind stress. Moreover, the DO formulation is in fact a more complex mathematical description than the RO formulation because the DO has an infinite number of degrees of freedom in terms of independent eigenmodes, whereas the RO has a minimal two degrees of freedom without redundancy. Nevertheless, conceptually the DO and RO are two succinct descriptions of ENSO basic dynamics, and both have played important roles in advancing ENSO theory. An obvious advantage that stems from the simplicity of the RO paradigm is an algebraic formulation for the BWJ index, which serves as a useful tool for dynamically assessing the various contributions of coupled processes to ENSO linear growth rate and frequency as envisioned by Bjerknes (1969), Wyrtki (1985), and Cane and Zebiak (1985). Further analyses of the phase

diagrams and vector fields associated with the simple RO paradigm as shown in Figure  6.9 (also Jin, 1997a; Takahashi et al., 2019) reveal its value in understanding the observed, simulated, and predicted ENSO behaviors. 6.4. FACTORS CONTROLLING ENSO AMPLITUDE, PERIODICITY, PHASE-LOCKING, ASYMMETRY, AND NONLINEAR RECTIFICATION Sections 6.2 and 6.3 provide a simple account of how the dynamics of the Pacific warm pool and cold tongue basic states allow perturbations to grow into ENSO cycles, and how it largely can be reduced into a minimal RO model for the simplest possible explicit depiction of the Bjerknes‐Wyrtki coupled instability for ENSO. In this section, we will use the RO model to systematically revisit basic questions of ENSO dynamics, i.e., what are the key factors that control ENSO amplitude, phase‐locking, El Niño/La Niña asymmetry, and its nonlinear rectification onto the mean state. 6.4.1. ENSO Amplitude To investigate these issues, we consider the following form of the RO model: dTE dt

RTE

dhw dt



hw

d dt

R

cTE3

1 BTE

h

0 w

m

R0 Ra sin

TE

0

(6.34)

(6.35)

t

(6.33)

 

w t

a

bTE 2

.

(6.36)

(°C month–1)

(a)

(°C month–1)

(b)

1.5 1 0.5 0 –0.5 –1 –1.5 1980

(°C month–1)

R = 0.57 1985

1990

1995

2010

2015

TDH TD TN

0.5 0 –0.5

R = 0.87 1985

1990

1995

2000

2005

2010

2015

Year 1

NDH ND NN

0.5 0 –0.5

R = 0.55 1985

1990

1995

2000

2005

2010

2015

Year

(d) (°C month–1)

2005

1

–1 1980 1.5 1 0.5 0 –0.5 –1 –1.5 1980

SG SGD SGN

R = 0.49 1985

1990

1995

2000

2005

2010

2015

Year

(e) (°C month–1)

2000 Year

–1 1980

(c)

LDH LD LN

1.5 1 0.5 0 –0.5 –1 –1.5 1980

∂TE /∂t [∂TE /∂t]D [∂TE /∂t]N

R = 0.55 1985

1990

1995

2000

2005

2010

2015

Year

(f) (m month–1)

10

∂hw/∂t [∂hw/∂t]D [∂hw/∂t]N

5 0 –5 –10 1980

R = 0.66 1985

1990

1995

2000

2005

2010

2015

Year

Figure 6.8  Time series of linear and nonlinear deterministic dynamics and noise forcing of the RO model. (a) Linear dynamic TE tendency and its deterministic part and noise (LDH, LD, and LN, respectively); (b) thermodynamic TE tendency and its deterministic part and noise (TDH, TD, and TN, respectively); (c) nonlinear dynamic TE tendency and its deterministic part and noise (NDH, ND and NN, respectively); (d) sub‐grid terms of TE tendency and its deterministic part and noise (SG, SGD, and SGN, respectively); (e) total TE tendency and its deterministic part and noise; (f) total hw tendency and its deterministic part and noise. Black curves denote original tendency of LDH, TDH, NDH, SG, ∂TE/∂t, and ∂hw/∂t. Red curves denote RO deterministic parts of LD, TD, ND, SGD, [∂TE/∂t]D, and [∂hw/∂t]D. Blue curves denote noise parts of LN, TN, NN, SGN, [∂TE/∂t]N, and [∂hw/∂t]N. The correlation coefficient between deterministic part and original term is indicated, respectively.

140  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a) RO linear deterministic tendency

(b) RO total deterministic tendency

(c) Noise tendency

20 10

hw (m)

0 –10 –20 –30 –40 –3

–2

–1

0

1

2

3

4

–3

–2

–1

0

TE(°C)

1

2

3

4

–3

–2

–1

0

TE(°C) 0.05 0.2

0.4

0.6

0.8

1

1.2

1

2

3

4

TE(°C) 1.4

1.6

1.8

2

2.2

2.4

2.6

Density (%)

Figure 6.9  Schematic representation of RO linear deterministic, total deterministic, and noise‐forcing dynamics. The vectors denote the tendency in phase space of the mixed layer volume averaged temperature anomalies over the Niño‐3 region and western Pacific 20°C isotherm depth anomalies (5°S–5°N, 120°E–155°W): (a) (LD+TD, [∂hw/∂t]D), (b) ([∂TE/∂t]D, [∂hw/∂t]D), and (c) ([∂TE/∂t]N, [∂hw/∂t]N), respectively. In the temperature equation, LD and LN denote linear dynamic deterministic feedbacks and noise, TD and TN thermodynamic deterministic feedback and noise, [∂TE/∂t]D and [∂TE/∂t]N total deterministic feedbacks and noise, respectively. In thermocline depth equation, [∂hw/∂t]D and [∂hw/∂t]N denote linear dynamic deterministic feedbacks and noise, respectively. The shading denotes a kernel density estimate of the joint probability distribution of the two time series, TE and hw. The vertical and horizontal dashed lines denote the peak of the probability distribution for TE and hw, respectively.

Here, a few additional parameters are introduced for the strength (Ra), frequency (ωa), and phase (φ) of a seasonal cycle modulation of the growth rate, for cubic (c) and quadratic (b) nonlinearities, and for additive (σ) and multiplicative (σB) noises. Westerly wind bursts (WWBs) affect ENSO, and importantly ENSO also modulates WWBs (e.g., Kessler et  al., 1995; Kleeman & Moore,1997; Vecchi & Harrison 2000; Yu et  al., 2003; Fedorov et al., 2003, Lengaigne et al., 2004; McPhaden, 2004; Zavala‐Garay et al., 2005; Eisenman et al., 2005). The state‐dependent stochastic forcing excitations included in the RO model here thus represent ENSO‐ modulated WWBs (Jin et al., 2007; Levine & Jin, 2010). The quadratic and cubic nonlinearities represent NDH from both deterministic nonlinear advection and upwelling as well as from the upscale effect of ENSO’s modulation of tropical instability waves (TIW) (An, 2008; Boucharel & Jin, 2020). One may also include a convective threshold that is nonlinear in the growth rate (Choi et al., 2013; Takahashi et al., 2019). These physically motivated nonlinearities are similar in terms of their qualitative impacts on the basic ENSO properties. As shown in section  6.3, there are a few more nonlinear

terms, but the details of how many of these terms are required to model ENSO realistically may vary depending on whether reanalysis data or CGCM simulations are investigated. Here, we will keep the minimum number of nonlinear terms in the nonlinear stochastically forced RO model that is adequate to qualitatively explore the aforementioned fundamental ENSO dynamics. For simplicity, we first consider a case in which the symmetry breaking processes are disabled (b = 0, B = 0). We also assume no seasonal cycle modulation (Ra = 0), resulting in a base case of the RO model under normal (cubic) nonlinearity and additive noise forcing. By considering a second order closure approximation (e.g., Launder et  al., 1975), the fourth moments of SSTA and of the warm pool thermocline anomaly covariability can be expressed in terms of their second moments:

TE 4

K 0 TE 2

2

, hTE

3

K1 hTE TE 2 . (6.37)

We can then derive an approximate ENSO SSTA variance equation as follows:

Simple ENSO Models  141

TE 2

R



2

Here

R

cK 0 TE 2

2

ˆ2

0.



(6.38)

cK1 TE 2

is a relatively small R cK1 TE 2 factor with weak dependence on the ENSO amplitude. This equation clearly indicates that ENSO amplitude depends on only three key factors: (i) the degree of near R criticality as measured by the linear growth rate , 2 (ii) the strength of cubic nonlinear damping (c), and (iii) the amplitude of stochastic excitation ( ˆ 2) as shown in Figure 6.10. How ENSO amplitude depends on the linear growth rate for different levels of noise forcing is shown in Figure  6.10b. When there is no stochastic excitation ( ˆ 2 0) (red curve), we have the standard form of ENSO amplitude dependence on criticality with zero self‐ sustained amplitude that is subcritical (R − ε  0). This result is similar to the solution obtained in the appendix of Jin (1997b). When ˆ 2 0, the stochastically forced solution transforms smoothly into a nonlinear dependence of ENSO amplitude on the criticality. Moreover, the increased level of noise indicates the distribution of ENSO’s probability density function becomes flatter (Figure  6.10a). ENSO amplitude decreases significantly when cubic nonlinear damping is increased (Figure  6.10c) in the super‐criticality regime (R − ε > 0), but less so in the stable (R − ε  0

0.5

BWJ = 0 BWJ < 0

–0.5

0.5

1

Ju n Ju l Au g Se p O ct N ov D ec Ja n Fe b M ar Ap r M ay

0

12

09

0.03 0.06 0.09 0.12 0.15 0.18

0.

06

0.

0.

0

03 0.

3

6

Histogram of EI Nino phase

c (°C–2month–1)

(f)

0.6 0.5

Model

Histogram of La Nina phase OBS Model

0.4 Probability

0.4 0.3 0.2

0.3 0.2

0.1

0.1

0

0

Ju n Ju l Au g Se p O ct N ov D ec Ja n Fe b M ar Ap r M ay

0

1

(R − ε)/2

0

(month–1)

OBS 0.5

0.5

1.5 Variance (°C2)

1.5

Ju n Ju l Au g Se p O ct N ov D ec Ja n Fe b M ar Ap r M ay

Seasonal variance/Growth rate OBS Model

Growth rate (yr–1) Probability

2

2

0

0

3

.0

2

–0

1

9

0

.0

–1

.0

–2

.1 2

–3

T (°C)

(d)

2.5

0.2

–0

0

Standard Deviation of T

3

2σ

1.8

0.04 0.02

(c)

Standard Deviation (°C)

0.5 σ

0.08

Standard Deviation of T

2

–0

0.1

OBS: SD = 0.87 S = 0.75 K = 3.96

(b)

0.5 σ: SD = 0.39 K = 2.61 σ: SD = 0.68 K = 2.41 2 σ: SD = 1.06 K = 2.18

–0

0.12

Probability

Probability distribution of T

0.14

Standard Deviation (°C)

(a)

Figure 6.10 (a) Probability distribution of temperature for the observations (gray bar) as well as the model with default, double, and half amplitude of stochastic forcing (black curves). The red curve indicates a Gaussian distribution with standard deviation of the default stochastic forcing amplitude. (b) ENSO amplitude dependence on linear growth for different levels of noise. Circles represent the corresponding results in (a). (c) ENSO amplitude dependence on strength of nonlinear damping for different linear growth rate regimes. (d) Seasonal ENSO variance (bar graph) and growth rate of ENSO (curves) for observations (OBS) and the RO model. Histogram of (e) El Niño and (f) La Niña peaking month according to calendar months for the observations and the RO model.

Simple ENSO Models  143

fundamental for ENSO phase locking (An & Jin, 2010; Stein et al., 2014). We use here the simple RO framework to discern the key factors that affect ENSO phase locking, which can be simply measured by the histogram of SSTA variance as a function of calendar month both for the observations and a typical solution from the RO model (Figure 6.10d). Another direct measure for capturing the ENSO peak phase is to examine the histogram of SSTA peak time as a function of calendar month (Figure 6.10e, f). These two methods both capture ENSO phase locking. For the observations, the peak times of both El Niño and La Niña tend to occur toward the end of the calendar year from November to January. The same ENSO phase locking is seen for the RO model simulation. In addition, how narrowly the histogram of SSTA peak time is distributed around the most preferred season is another important characteristic of ENSO phase locking. In both observations and model simulations, the distribution of the peak month is narrower for La Niña than El Niño, indicating that the sharpness of ENSO phase locking is asymmetric. In Chen and Jin (2020), the strength of ENSO phase locking is defined by the sharpness of the ENSO phase histogram. A sharp distribution of the phase histogram indicates a strong phase locking, whereas a wide distribution indicates weak phase locking. Chen and Jin (2020) demonstrated that the preferred month of ENSO peaking time depends on the phase and strength of the seasonal cycle modulation of ENSO growth rate, and the strength of ENSO phase locking mainly depends on the amplitude of the seasonal cycle of the growth rate, the linear growth rate regime, noise, and the linear frequency. Consequently, understanding how the seasonal cycles in the cold tongue and warm pool background states respectively control the phase and strength of seasonal cycle modulation of the ENSO growth rate may help quantify the main contributing sources to ENSO phase locking. 6.4.3. ENSO Periodicity and Frequency Locking ENSO periodicity appears to span a wide range, from 2 to 7 years in the modern record, with some evidence pointing to the preference of QQ and QB periodicities (Jiang et al., 1995; Ghil et al., 2002). The current generation of climate models still suffers from large biases in the simulation of ENSO periodicity (Lu et al., 2018). What controls ENSO periodicity is a fundamental question and important for understanding its basic dynamics. Linear instability theory, as briefly reviewed in section 6.2, suggests that the linear frequency of the leading ENSO mode varies sensitively from QQ to QB periodicity ranges, with modest changes from a relatively weaker to a stronger cold‐tongue basic state. Moreover, the linear eigen fre-

quency is not the sole factor that controls ENSO periodicity. Frequency locking induced by the seasonal cycle (Jin et al., 1994; Tziperman et al., 1994) and nonlinear corrections to ENSO periodicity (demonstrated analytically in the appendix of Jin, 1997a, and numerically in Eccles and Tziperman, 2004) can also affect the ENSO periodicity. We first examine the frequency‐locking phenomenon using the symmetric version of the RO model (b = B = 0) in the supercritical regime (R − ε > 0) without noise forcing. We vary the key parameters that control the linear period of RO ENSO

2 /

2 0

2

R 2

and the

strength of the annual modulation of the ENSO growth rate (Ra). The dominant frequency (main peak of the SSTA spectrum) is shown in Figure 6.11. The ENSO frequency is discretized into frequency‐locked steps known as a Devil's staircase (Bak, 1986), as noted in Jin et  al. (1994), and frequency‐locked regions that are called Arnold tongues (Arnol’d, 1961), as noted in Jin et  al. (1996) and Neelin et  al. (1998) in two‐parameter space. The dominant ENSO periodicity remains constant over an interval and then changes in discrete jumps to the next frequency‐locked solution (e.g., 2, 3, 4, 5, 6, 7, and 8‐year periodicities and a series of rational fraction frequencies corresponding to non‐integer‐year periodicity). When Ra is increased, the width of each frequency step is increased, which indicates stronger frequency locking. ENSO linear instability and the strong seasonal modulation of this linear growth rate tend to generate a preference for QQ or QB ENSO periodicities if the linear ENSO frequency is within the QQ or QB range. The linear ENSO frequency, as captured by the imaginary part of the BWJ index, is the main factor that controls ENSO periodicity. Seasonal modulations of ENSO, nonlinear processes, and symmetry breaking only have moderate effects on ENSO periodicity. Noise forcing not only can completely smooth out the the 4‐, 3‐ and 2‐year periodicity from nonlinear frequency locking but also shift ENSO’s main periodicity toward high frequencies due to nonlinear effects (Figure 6.11b). It is difficult to explain the observed broad range of ENSO periodicity with a single linear frequency in the RO model framework. The observed broad spectrum of ENSO likely involves sensitive modulation of ENSO linear periodicity by slow variations in the basic state; however, additional studies are needed to substantiate this conjecture. 6.4.4. ENSO Asymmetry The causes of the asymmetry between El Niño and La Niña in amplitude, duration, and spatial SST patterns are key questions in ENSO research. While the RO

144  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a)

Arnold tongues

0.6

(b)

0.4 0.3 0.2 2 3 4 5 6

0.1 0

σ = 0 (BWJ > 0) 0.5σ (BWJ > 0) 0.5σ (BWJ < 0)

5 ENSO Period (year)

Ra (month–1)

0.5

ENSO Periodicity

6

1

4

3

2

2

3

4

5

6

7

Linear Period (year)

1

1

2

3

4

5

6

Linear Period (year)

Figure 6.11  (a) Schematic representation of an Arnold tongue diagram for the model. The horizontal black line is related to the Devil’s staircase. (b) The Devil’s staircase, namely, a plot of the ENSO periodicity as a function of the linear period for different stochastic forcing in the unstable regime (blue and green dots) and stable regime (red dots). The dominant ENSO period is estimated from the maximum peak of the ensemble average of 100‐ member power spectral density calculated via Fast Fourier Transform (FFT). Each ensemble member is integrated for 200 years.

framework is not able to address the pattern asymmetry, the other asymmetries can be addressed qualitatively with the RO model (Levine & Jin, 2010), as shown in Figure 6.12a– c. A number of processes have been identified to explain ENSO amplitude asymmetry (Jin et al., 2003; An & Jin, 2004; Hayashi & Jin, 2017; Su et al., 2010; Vialard et al., 2001; Kang & Kug, 2002; Liang et al., 2017), and we refer the reader to chapter  7 for a detailed review. Here, we summarize all the processes that generate amplitude asymmetry into two types of symmetry‐breaking processes: (i) nonlinear dynamical heating and (ii) ENSO state‐dependent deterministic coupled feedback, such as a threshold nonlinearity (Choi et  al., 2013; Takahashi et al., 2019) or ENSO state‐dependent stochastic excitation. In our RO model framework, without loss of generality, they are related to the parameters b and B, respectively. The SSTA skewness has been the main measure for ENSO asymmetry. Using the statistical moment approach of Jin et  al. (2007) and Levine and Jin (2010), we may obtain the approximated solution for the third moment (skewness) as a nearly linear function of the ENSO symmetry breaking processes, as measured by parameters b and B (not shown). Both the deterministic and stochastic symmetry‐breaking processes can generate ENSO asymmetry (Figure  6.12d). Positive ENSO NDH and NDH from TIW can both increase the b parameter that in turn increases positive ENSO skewness. The activity of WWB/ MJO and TIW modulated by ENSO will increase the B

parameter, and thus also ENSO skewness. If b becomes negative, the effect on skewness also becomes negative. If La Niña can generate more easterly wind bursts, which may not be the case in nature (Hayashi & Watanabe, 2016; Puy et al., 2015), we will then find a linear relation between negative B and ENSO skewness as well. The ENSO linear growth rate regime can also affect ENSO skewness: ENSO skewness decreases when the linear growth rate changes from a stable to an unstable regime. Nevertheless, the key insight learned from the RO model is that the ENSO symmetry breaking processes may all additively contribute to ENSO asymmetry as measured by its SSTA skewness. 6.4.5. Rectification of ENSO onto the Climatological Mean State As synoptic weather systems transport heat, moisture, and momentum that affect the background mean state in which the weather systems develop, there is also a similar ENSO–mean state interaction through various nonlinear processes. Simple nonlinear conceptual models with nonlinear advection (Jin, 1996; Sun, 1997; Jin, 1998; Timmermann & Jin, 2002; Timmermann et al., 2003; Liang et al., 2012), more complex models (Rodgers et  al., 2004), and NDH diagnosed from reanalysis data (Jin et al., 2003; An & Jin, 2004; Hayashi & Jin, 2017) all point toward the importance of ENSO’s nonlinear rectification effect on the mean state.

Simple ENSO Models  145

Temperature (°C)

(b)

Observed Nino3 index (3-month running average)

Probability distribution of T

0.08 OBS (S = 0.7) Model (S = 0.0) Model (S = 0.7)

0.07 0.06 0.05

Simulated Temperature (3-month running average)

3 2 1 0 –1 –2 –3

0.04 0.03 0.02 0.01 0

5

0

(d)

(c)

3 2 1 0 –1 –2 –3 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Year

Probability

Nino3 (°C)

(a)

1

10

15

20

25

30

35 Year

40

45

50

Skewness of T

55

1.5

60

(e)

65

–3

–2

–1

0 T (°C)

1

2

Mean of T

1

3 (Unit:°C) 0.6

1

0.4

0.75

0.75

0

0.5

0.2

B

B

0.5

0

0.5

–0.2

–0.5 0.25

0.25 –0.4

–1

0 –0.2

–0.15 –0.10 –0.05

0

0.05

0.10

0.15

0.2

–1.5

0 –0.2

–0.6 –0.15 –0.10 –0.05

b (°C–1month–1)

0

0.05

0.10

0.15

0.2

b (°C–1month–1)

Figure 6.12  (a) Observed Niño‐3 index and (b) time series of the model simulated temperature using a three‐ month running average. (c) Probability distribution of temperature for the observations (gray bar) and RO model with zero (black curve) and normal (red curve) SST skewness values. (d) Skewness of simulated temperature related as a function of state‐dependent noise forcing amplitude and quadratic nonlinearity. (e) Mean of simulated temperature as a function of state‐dependent noise forcing amplitude and quadratic nonlinearity.

Whether it be the baroclinic instability for weather systems or the Bjerknes instability for ENSO, linear growth can only cause perturbations to grow within a finite timeframe as nonlinearities in the conservation laws of the governing equations (primarily through advective heat redistribution) will generate changes in the mean state to prevent unstable linear growth. A detailed discussion of the specific processes that are important for ENSO–mean state interaction can be found in chapter  8. We here will use the simple RO model to gain insight into the relationship between ENSO’s rectification effect onto the mean background state and ENSO skewness through symmetry‐breaking nonlinearities.

The time‐mean of the SSTA in the RO model can be written as 2

R

0

TE

bTE 2 cTE3

B TE

0.

(6.39)

In the absence of symmetry‐breaking processes (b = B = 0), the time‐mean solution is zero and there is no mean state change even though the system is nonlinear. This is because the cubic nonlinearity in SST and the additive noise are symmetry‐preserving nonlinearities. However, once the symmetry is broken by either state‐dependent noise or a quadratic nonlinearity, both of which can be easily related to ubiquitous advective processes, all the

146  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

nonlinear terms give rise to nonzero contributions to a time‐mean NDH. Considering the fact that the time‐ mean SSTAs are small, the time mean of nonlinear terms and multiplicative noise terms in Eq. (6.39) are domiB TE . Furthermore, we can nated by bTE 2 cTE 3 obtain a linear analytical solution to demonstrate that a positive ENSO‐like warming mean state is proportional to the skewness of ENSO TE S , as demonstrated in Figure  6.12d and 12e. The time‐mean temperature is linearly related to the ENSO skewness. This simple relation provides a tool for us to understand how ENSO may not only affect but also interact with the cold tongue mean state in the tropical Pacific and how biases in climate model simulations of ENSO may contribute to a climate mean state bias. 6.5. OUTLOOK Modern observational networks, deployed just in the past few decades, have already captured a rich variety of ENSO spatial patterns and temporal evolutions, as described in chapters 3 and 4, among others. These short records have most likely not yet encapsulated the full spectrum of possible ENSO behaviors. Nevertheless, the complexity we have seen has gone far beyond what was perceived by pioneers such as Bjerknes and Wyrtki, among others. Significant advances have indeed been made in the theory of ENSO complexity (e.g., JN93; Neelin et al., 1998; Fedorov & Philander, 2001; Jin et al., 2006; Levine & Jin, 2010; Stuecker et al., 2013; Takahashi et al., 2019; Xie & Jin, 2018), in comprehensive simulations of ENSO by climate models (e.g., Guilyardi et al., 2009; Wittenberg et al., 2014; Bayr et al., 2018), and in improving operational predictions of ENSO events and their climatic and societal impacts (e.g., Luo et al., 2008; McPhaden et al., 2010; Stockdale et al., 2011; Barnston et al., 2012; L’Heureux et al., 2017). In spite of the tremendous successes achieved, state‐of‐the‐art climate models are yet to converge in their simulations of basic ENSO characteristics such as its amplitude, period, phase locking, and warm‐cold phase asymmetry. Even more challenging is for models to simulate these characteristics as a result of capturing all key processes in the right balance, instead of as a result of error compensations (Bellenger et al., 2014; Karamperidou et al., 2017; Bayr et al., 2018; chapter 9 in this book). An additional challenge for models is the simulation of ENSO diversity, characterized by distinct spatial patterns, periodicities, duration, and asymmetries of the different ENSO “flavors.” These model biases have likely stalled the improvement of the skill and reliability of dynamical seasonal climate predictions that depend heavily on the representation of ENSO in climate models. Identifying

the sources and understanding the underlying mechanisms of stubborn climate model biases in the simulation of the mean state of the equatorial Pacific and the representation of ENSO thermodynamic and dynamic feedbacks will be essential in developing the next generation of models for seamless climate forecasts and projections. Future research on ENSO complexity needs to continue to undertake a hierarchical modeling approach that includes simple conceptual models, models of inter­ mediate complexity with consistent thermodynamic and dynamic coupled feedback processes and representation of multiscale interactions, and state‐of‐the‐art high‐resolution general circulation models (chapter 9 in this book). This long‐tested hierarchical approach that combines theoretical, observational, model, and diagnostic frameworks has the potential to lead to a new era of advances in ENSO research, advances that will not only enhance our understanding of the fundamental dynamics of eastern and central Pacific ENSO events, their interactions, and their relation to tropical Pacific decadal variability, but also decipher the role of multiscale processes involving WWBs, TIWs, and extratropical and pan‐basin precursors of ENSO spatiotemporal complexity. These are essential steps to ultimately improve predictive skill for the entire tropical‐extratropical climate system. ACKNOWLEDGMENTS The authors are grateful for valuable comments from Mark Cane, Alexey Fedorov, and an anonymous reviewer. F.F.J. was supported by US NSF grant AGS‐1813611 and Department of Energy grant DE‐ SC0005110. M.H. was supported by JSPS Overseas Research Fellowships 201860671. C.K. was supported by US NSF grant #AGS‐1602097. M.F.S. was supported by the Institute for Basic Science (project code IBS‐ R028‐D1). R.X. was supported by the National Natural Science Foundation of China (NSFC Grant 41506017). A BRIEF DESCRIPTION OF THE CZ MODEL The CZ model is an anomaly model with a prescribed annually varying climate mean state. It comprises a simple quasi‐linear Gill-Matsuno atmospheric component (Zebiak & Cane, 1987) that simulates the tropical wind response to ENSO‐associated SSTAs by considering condensational heating due to SSTA-induced moisture supply through evaporation and convective heating parameterized by moisture convergence by anomalous atmospheric flow:

u

a a

n 0

yva n

pn /

0 x

, (A1a)

Simple ENSO Models  147

a



n

v

0

a a

pn /

0

ca 2

ua n

Q s

yua n

pn /

va n

x

0 y

, (A1b)

Q s Q1n 1 ,

y

(A1c)

ws

Tsub

T Te H1 1

T , (A4a)

s

T , (A4b)

T1 tanh b1 h

h

tanh b1h , h

0

T2 tanh b2 h

h

tanh b2 h , h 0

Tsub h

Q1n

cn

M c

u t

0

yv

0 yu



h t



. (A1e)

M c

u x

H

ru (A2a)

H y

h y

g

x

h x

g

H v y

rv (A2b)

rh, (A2c)

where u = H–1(H1u1+H2u2). The subscripts 1 and 2 indicate that within the upper layer ocean, there is an embedded mixed layer with fixed depth H1 (50 m) and an underlying subsurface layer with fixed depth H2 (100 m). The equations governing the velocity shear (us) between layer 1 and 2 are x



rs us

0

yvs

rsvs

0

yus

H1 y



H1

, (A3a)

, (A3b)

where us = u1 − u2. The mixed‐layer SSTA is governed by the heat budget T t

Te

T exp T 30 / 16.7 , (A1d)

Definitions for all variables and parameters are listed in Table A. Here n denotes the sequence such that cumulus convection parameterized from moisture convergence can be solved iteratively. The oceanic dynamical component includes a 1.5‐layer linear reduced gravity model that describes the upper‐ layer current and thermocline depth anomalies in response to wind stress anomalies:



M ws

M ws

u1

T T ws

M ws

u1

T Tz





, (A4c)

where T , u1 , and ws are the mean SST, horizontal mixed‐layer ocean currents, and upwelling. Parameters and their values used in the model are listed in Table A. REFERENCES An, S.‐I. (2008). Interannual variations of the tropical ocean instability wave and ENSO. Journal of Climate, 21(15), 3680– 3686. https://doi.org/10.1175/2008JCLI1701.1 An, S.‐I., & Jin, F.‐F. (2000). An eigen analysis of the interdecadal changes in the structure and frequency of ENSO mode. Geophysical Research Letters, 27(16), 2573–2576. https://doi.org/10.1029/1999GL011090 An, S.‐I., & Jin, F.‐F. (2004). Nonlinearity and asymmetry of ENSO. Journal of Climate, 17(12), 2399–2412. https://doi.org /10.1175/1520‐0442(2004)017%3C2399:NAAOE%3E2.0 .CO;2 An, S.‐I., & Jin, F.‐F. (2010). Linear solutions for the frequency and amplitude modulation of ENSO by the annual cycle. Tellus A, 63(2), 238–243. https://doi.org/10.1111/j.1600‐ 0870.2010.00482.x Arnol’d, V. I. (1961). Small denominators: I. Mapping the circle onto itself. Izvestiya Rossiiskoi Akademii Nauk SSSR Seriya Matematicheskaya, 25(1), 21–86. Bak, P. (1986). Devil’s staircase. Physics Today, 39(12), 38. https://doi.org/10.1063/1.881047 Barnston, A. G., Tippett, M. K., L’Heureux, M. L., Li, S., & DeWitt, D. G. (2012). Skill of real‐time seasonal ENSO model predictions during 2002–11: Is our capability increasing? Bulletin of the American Meteorological Society, 93(5), 631–651. https://doi.org/10.1175/BAMS‐D‐11‐00111.1 Battisti, D. S., & Hirst, A. C. (1989). Interannual variability in a tropical atmosphere‐ocean model: Influence of the basic state, ocean geometry and nonlinearity. Journal of the Atmospheric Sciences, 46(12), 1687–1712. https://doi.org/10.1 175/1520‐0469(1989)046%3C1687:IVIATA%3E2.0.CO;2 Bayr, T., Latif, M., Dommenget, D., Wengel, C., Harlaß, J., & Park, W. (2018). Mean‐state dependence of ENSO atmospheric feedbacks in climate models. Climate Dynamics, 50(9–10), 3171–3194. https://doi.org/10.1007/s00382‐017 ‐3799‐2 Bejarano, L. (2006). Coexistence of leading equatorial coupled modes for ENSO (Ph.D. dissertation). Retrieved from FSU Digital Library (http://purl.flvc.org/fsu/fd/FSU_migr_etd‐1210). Tallahassee, FL: Florida State University.

148  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Bejarano, L., & Jin, F.‐F. (2008). Coexistence of equatorial coupled modes of ENSO. Journal of Climate, 21(12), 3051–3067. https://doi.org/10.1175/2007JCLI1679.1 Bellenger, H., Guilyardi, E., Leloup, J., Lengaigne, M., & Vialard, J. (2014). ENSO representation in climate models: From CMIP3 to CMIP5. Climate Dynamics, 42(7–8), 1999– 2018. https://doi.org/10.1007/s00382‐013‐1783‐z Bjerknes, J. (1969). Atmospheric teleconnections from the equatorial Pacific. Monthly Weather Review, 97(3), 163–172. https://doi.org/10.1175/1520‐0493(1969)097%3C0163:ATFT EP%3E2.3.CO;2 Bjerknes, J., & Solberg, H. (1922). Life cycle of cyclones and the polar front theory of atmospheric circulation. Geofysiske Publikationer, 3(1), 3–18. Boucharel, J., & Jin, F.‐F. (2020). A simple theory for the modulation of tropical instability waves by ENSO and the annual cycle. Tellus A, 72(1), 1–14. https://doi.org/10.1080/16000870 .2019.1700087 Burgers, G., Jin, F.‐F., & van Oldenborgh, G. J. (2005). The simplest ENSO recharge oscillator. Geophysical Research Letters, 32, L13706. https://doi.org/10.1029/2005GL022951 Cane, M. A. (1984). Modeling sea level during El Niño. Journal of Physical Oceanography, 14, 1864–1874. https://doi.org/10. 1175/1520‐0485(1984)014%3C1864:MSLDEN%3E2.0.CO;2 Cane, M. A., Münnich, M., & Zebiak, S. E. (1990). A study of self‐excited oscillations of the tropical ocean‐atmosphere system: I. Linear analysis. Journal of the Atmospheric Sciences, 47, 1562–1577. https://doi.org/10.1175/1520‐0469(1 990)047%3C1562:ASOSEO%3E2.0.CO;2 Cane, M. A., & Sarachik, E. S. (1977). Forced baroclinic ocean motions: II. The linear equatorial bounded case. Journal of Marine Research, 35, 395–432. Cane, M. A., & Sarachik, E. S. (1979). Forced baroclinic ocean motions: III. The linear equatorial basin case. Journal of Marine Research, 37, 355–398. Cane, M. A., & Sarachik, E. S. (1981). The response of a linear baroclinic equatorial ocean to periodic forcing. Journal of Marine Research, 39(4), 651–693. Cane, M. A., & Zebiak, S. E. (1985). A theory for El Niño and the Southern Oscillation. Science, 228(4703), 1085–1087. https://doi.org/10.1126/science.228.4703.1085 Cane, M. A., Zebiak, S. E., & Dolan, S. C. (1986). Experimental forecasts of El Niño. Nature, 321(6073), 827. https://doi. org/10.1038/321827a0 Carton, J. A., Chepurin, G. A., & Chen, L. (2018). SODA3: A new ocean climate reanalysis. Journal of Climate, 31(17), 6967–6983. https://doi.org/10.1175/JCLI‐D‐18‐0149.1 Chen, H.‐C. & Jin, F.‐F. (2020). Fundamental behavior of ENSO phase‐locking. Journal of Climate, 33(5), 1953–1968. https://doi.org/10.1175/JCLI‐D‐19‐0264.1 Chen, N., & Majda, A. J. (2016). Simple dynamical models capturing the key features of the central Pacific El Niño. Proceedings of the National Academy of Sciences, 113(42), 11732–11737. https://doi.org/10.1073/pnas.1614533113 Chen, N., & Majda, A. J. (2017). Simple stochastic dynamical models capturing the statistical diversity of El Niño Southern Oscillation. Proceedings of the National Academy of Sciences, 114(7), 1468–1473. https://doi.org/10.1073/ pnas.1620766114

Choi, K.‐Y., Vecchi, G. A., & Wittenberg, A. T. (2013). ENSO transition, duration, and amplitude asymmetries: Role of the nonlinear wind stress coupling in a conceptual model. Journal of Climate, 26(23), 9462–9476. https://doi.org/10.1175/ JCLI‐D‐13‐00045.1 Dijkstra, H. A. (2005). Nonlinear physical oceanography: A dynamical systems approach to the large scale ocean circulation and El Niño. Dordrecht, NY: Springer. Eccles, F., & Tziperman, E. (2004). Nonlinear effects on ENSO’s period. Journal of the Atmospheric Sciences, 61(4), 474–482. https://doi.org/10.1175/1520‐0469(2004)061%3C0474:NEOE P%3E2.0.CO;2 Eisenman, I., Yu, L., & Tziperman, E. (2005). Westerly wind bursts: ENSO’s tail rather than the dog? Journal of Climate, 18(24), 5224–5238. https://doi.org/10.1175/JCLI3588.1 Fedorov, A. V. (2010). Ocean response to wind variations, warm water volume, and simple models of ENSO in the low‐frequency approximation. Journal of Climate, 23, 3855–3873. https://doi.org/10.1175/2010JCLI3044.1 Fedorov, A. V., & Philander, S. G. (2000). Is El Niño changing? Science, 288(5473), 1997–2002. https://doi.org/10.1126/ science.288.5473.1997 Fedorov, A. V., & Philander, S. G. (2001). A stability analysis of tropical ocean‐atmosphere interactions: Bridging measurements and theory for El Niño. Journal of Climate, 14(14), 3086–3101. https://doi.org/10.1175/1520‐0442(2001)014%3C 3086:ASAOTO%3E2.0.CO;2 Fedorov, A. V., Harper, S. L., Winter, B. & Wittenberg, A. (2003). How predictable is El Niño? Bulletin of the American Meteorological Society, 84(7), 911–919. https://doi. org/10.1175/BAMS‐84‐7‐911 Ghil, M., Allen, M. R., Dettinger, M. D., Ide, K., Kondrashov, D., Mann, M. E., et al. (2002). Advanced spectral methods for climatic time series. Reviews of Geophysics, 40(1), 1003. https://doi.org/10.1029/2000RG000092 Gill, A. E. (1980). Some simple solutions for heat‐induced tropical circulation. Quarterly Journal of the Royal Meteorological Society, 106(449), 447–462. https://doi. org/10.1002/qj.49710644905 Gill, A. E. (1985). Elements of coupled ocean‐atmosphere models for the tropics. In J.C.J. Nihoul (Ed.), Coupled ocean‐ atmosphere models (Elsevier Oceanography Series, Vol. 40, pp. 303–327). Liège, Belgium: Elsevier. https://doi. org/10.1016/S0422‐9894(08)70718‐9 Guilyardi, E., Wittenberg, A., Fedorov, A., Collins, M., Wang, C., Capotondi, A., et  al. (2009). Understanding El Niño in ocean‐atmosphere general circulation models: Progress and challenges. Bulletin of the American Meteorological Society, 90(3), 325–340. https://doi.org/10.1175/2008BAMS2387.1 Hayashi, M., & Jin, F.‐F. (2017). Subsurface nonlinear dynamical heating and ENSO asymmetry. Geophysical Research Letters, 44(24), 12427–12435. https://doi.org/10.1002/ 2017GL075771 Hayashi, M., & Watanabe, M. (2016). Asymmetry of westerly and easterly wind events: Observational evidence. Scientific Online Letters on the Atmosphere, 12, 38–41. https://doi. org/10.2151/sola.2016‐009 Hirst, A. C. (1986). Unstable and damped equatorial modes in simple coupled ocean‐atmosphere models. Journal of the

Simple ENSO Models  149 Atmospheric Sciences, 43(6), 606–632. https://doi.org/10.1175 /1520‐0469(1986)043%3C0606:UADEMI%3E2.0.CO;2 Izumo, T., Lengaigne, M., Vialard, J., Suresh, I., & Planton, Y. (2019). On the physical interpretation of the lead relation between warm water volume and the El Niño Southern Oscillation. Climate Dynamics, 52(5–6), 2923–2942. https:// doi.org/10.1007/s00382‐018‐4313‐1 Jiang, N., Neelin, J. D., & Ghil, M. (1995). Quasi‐quadrennial and quasi‐biennial variability in the equatorial Pacific. Climate Dynamics, 12(2), 101–112. https://doi.org/10.1007/ BF00223723 Jin, F.‐F. (1996). Tropical ocean‐atmosphere interaction, the Pacific cold tongue, and the El Niño-Southern Oscillation. Science, 274(5284), 76–78. https://doi.org/10.1126/ science.274.5284.76 Jin, F.‐F. (1997a). An equatorial ocean recharge paradigm for ENSO: Part I. Conceptual model. Journal of the Atmospheric Sciences, 54(7), 811–829. https://doi.org/10.1175/1520‐0469(1 997)054%3C0811:AEORPF%3E2.0.CO;2 Jin, F.‐F. (1997b). An equatorial recharge paradigm for ENSO: II. A stripped‐down coupled model. Journal of the Atmospheric Sciences, 54(7), 830–847. https://doi.org/10.1175 /1520‐0469(1997)054%3C0830:AEORPF%3E2.0.CO;2 Jin, F.‐F. (1998). A simple model for the Pacific cold tongue and ENSO. Journal of the Atmospheric Sciences, 55(14), 2458– 2469. https://doi.org/10.1175/1520‐0469(1998)055%3C2458: ASMFTP%3E2.0.CO;2 Jin, F.‐F., & An, S.‐I. (1999). Thermocline and zonal advective feedbacks within the equatorial ocean recharge oscillator model for ENSO. Geophysical Research Letters, 26(19), 2989– 2992. https://doi.org/10.1029/1999GL002297 Jin, F.‐F., & Neelin, J. D. (1993a). Modes of interannual tropical ocean‐atmosphere interaction—A unified view. Part I: Numerical results. Journal of the Atmospheric Sciences, 50(21), 3477–3503. https://doi.org/10.1175/1520‐0469(1993)0 50%3C3477:MOITOI%3E2.0.CO;2 Jin, F.‐F., & Neelin, J. D. (1993b). Modes of interannual tropical ocean‐atmosphere interaction—A unified view. Part III: Analytical results in fully coupled cases. Journal of the Atmospheric Sciences, 50(21), 3523–3540. https://doi.org/10.1 175/1520‐0469(1993)050%3C3523:MOITOI%3E2.0.CO;2 Jin, F.‐F., Neelin, J. D., & Ghil, M. (1994). El Niño on the Devil’s staircase: Annual subharmonic steps to chaos. Science, 264(5155), 70–72. https://doi.org/10.1126/science.264.5155.70 Jin, F.‐F., Neelin, J. D., & Ghil, M. (1996). El Niño/Southern Oscillation and the annual cycle: Subharmonic frequency‐ locking and aperiodicity. Physica D: Nonlinear Phenomena, 98 (2–4), 442–465. https://doi.org/10.1016/0167‐2789(96)00111‐X Jin, F.‐F., An, S.‐I., Timmermann, A., & Zhao, J. (2003). Strong El Niño events and nonlinear dynamical heating. Geophysical Research Letters, 30(3), 1120. https://doi.org/10.1029/2002 GL016356 Jin, F.‐F., Kim, S. T., & Bejarano, L. (2006). A coupled‐stability index for ENSO. Geophysical Research Letters, 33(23), L23708. https://doi.org/10.1029/2006GL027221 Jin, F.‐F., Lin, L., Timmermann, A., & Zhao, J. (2007). Ensemble‐ mean dynamics of the ENSO recharge oscillator under state‐ dependent stochastic forcing. Geophysical Research Letters, 34(3), L03807. https://doi.org/10.1029/2006GL027372

Kang, I.‐S., & Kug, J.‐S. (2002). El Niño and La Niña sea surface temperature anomalies: Asymmetry characteristics associated with their wind stress anomalies. Journal of Geophysical Research,107(D19),4372.https://doi.org/10.1029/2001JD000393 Karamperidou, C., Jin, F.‐F. & Conroy, J. L. (2017). The importance of ENSO nonlinearities in tropical Pacific response to external forcing. Climate Dynamics, 49(7–8), 2695–2704. https://doi.org/10.1007/s00382‐016‐3475‐y Kessler, W. S. (2002). Is ENSO a cycle or a series of events? Geophysical Research Letters, 29(23), 2125. https://doi. org/10.1029/2002GL015924 Kessler, W. S., McPhaden, M. J., & Weickmann, K. M. (1995). Forcing of intraseasonal Kelvin waves in the equatorial Pacific. Journal of Geophysical Research, 100(C6), 10613– 10631. https://doi.org/10.1029/95JC00382 Kim, S. T., Cai, W., Jin, F.‐F., & Yu, J.‐Y. (2014). ENSO stability in coupled climate models and its association with mean state. Climate Dynamics, 42(11–12), 3313–3321. https://doi. org/10.1007/s00382‐013‐1833‐6 Kim, S. T., & Jin, F.‐F. (2011). An ENSO stability analysis. Part I: Results from a hybridcoupled model. Climate Dynamics, 36(7–8), 1593–1607. https://doi.org/10.1007/s00382‐010‐0796‐0 Kleeman, R., & Moore, A. M. (1997). A theory for the limitation of ENSO predictability due to stochastic atmospheric transients. Journal of the Atmospheric Sciences, 54(6), 753– 767. https://doi.org/10.1175/1520‐0469(1997)0542.0.CO;2 Kug, J.‐S., Jin, F.‐F., & An, S.‐I. (2009). Two types of El Niño events: Cold tongue El Niño and warm pool El Niño. Journal of Climate, 22(6), 1499–1515. https://doi.org/10.1175/2008J CLI2624.1 Launder, B. E., Reece, G. J., & Rodi, W. (1975). Progress in the development of a Reynolds‐stress turbulence closure. Journal of Fluid Mechanics, 68(3), 537–566. https://doi.org/10.1017/ S0022112075001814 Lengaigne, M., Guilyardi, E., Boulanger, J. P., Menkes, C., Delecluse, P., Inness, P., Cole, J., & Slingo, J. (2004). Triggering of El Niño by westerly wind events in a coupled general circulation model. Climate Dynamics, 23(6), 601–620. https:// doi.org/10.1007/s00382‐004‐0457‐2 Levine, A. F. Z., & Jin, F.‐F. (2010). Noise‐induced instability in the ENSO recharge oscillator. Journal of the Atmospheric Sciences, 67(2), 529–542. https://doi.org/10.1175/2009JAS3213.1 Levine, A. F. Z., & McPhaden, M. J. (2016). How the July 2014 easterly wind burst gave the 2015–2016 El Niño a head start. Geophysical Research Letters, 43(12), 6503–6510. https://doi. org/10.1002/2016GL069204 L’Heureux, M. L., Takahashi, K., Watkins, A. B., Barnston, A. G., Becker, E. J., Di LiBerto, T. E., et al. (2017). Observing and predicting the 2015/16 El Niño. Bulletin of the American Meteorological Society, 97(7), 1363–1382. https://doi. org/10.1175/BAMS‐D‐16‐0009.1 Liang, J., Yang, X.‐Q., & Sun, D.‐Z. (2012). The effect of ENSO events on the tropical Pacific mean climate: Insights from an analytical model. Journal of Climate, 25(21), 7590–7606. https://doi.org/10.1175/JCLI‐D‐11‐00490.1 Liang, J., Yang, X.‐Q., & Sun, D.‐Z. (2017). Factors determining the asymmetry of ENSO. Journal of Climate, 30(16), 6097–6106. https://doi.org/10.1175/JCLI‐D‐16‐0923.1

150  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Lindzen, R. S., & Nigam, S. (1987). On the role of sea surface temperature gradients in forcing low‐level winds and convergence in the tropics. Journal of the Atmospheric Sciences, 44(17), 2418–2436. https://doi.org/10.1175/1520‐0469(1987)0 44%3C2418:OTROSS%3E2.0.CO;2 Lu, B., Jin, F.‐F., & Ren, H.‐L. (2018). A coupled dynamic index for ENSO periodicity. Journal of Climate, 31(6), 2361– 2376. https://doi.org/10.1175/JCLI‐D‐17‐0466.1 Luo, J.‐J., Masson, S., Behera, S. K., & Yamagata, T. (2008). Extended ENSO predictions using a fully coupled ocean‐ atmosphere model. Journal of Climate, 21(1), 84–93. https:// doi.org/10.1175/2007JCLI1412.1 Matsuno, T. (1966). Quasi‐geostrophic motions in the equatorial area. Journal of the Meteorological Society of Japan. Ser. II, 44(1), 25–43. https://doi.org/10.2151/jmsj1965.44.1_25 McGregor, S., Timmermann, A., Schneider, N., Stuecker, M. F., & England, M. H. (2012). The effect of the South Pacific convergence zone on the termination of El Niño events and the meridional asymmetry of ENSO. Journal of Climate, 25(16), 5566–5586. https://doi.org/10.1175/JCLI‐D‐11‐00332.1 McPhaden, M. J. (2004). Evolution of the 2002/03 El Niño. Bulletin of the American Meteorological Society, 85(5), 677– 695. https://doi.org/10.1175/BAMS‐85‐5‐677 McPhaden, M. J., Busalacchi, A. J., & Anderson, D. L. T. (2010). A TOGA retrospective. Oceanography, 23(3), 86–103. https://doi.org/10.5670/oceanog.2010.26 Moore, D. W. (1968). Planetary‐gravity waves in an equatorial ocean (Ph.D. Dissertation). Retrieved from Harvard College Library. Cambridge, MA: Harvard University. Neelin, J. D. (1990). A hybrid coupled general circulation model for El Niño studies. Journal of the Atmospheric Sciences, 47(5), 674–693. https://doi.org/10.1175/1520‐0469(1990)047 %3C0674:AHCGCM%3E2.0.CO;2 Neelin, J. D., & Jin, F.‐F. (1993). Modes of interannual tropical ocean‐atmosphere interaction—A unified view. Part II: Analytical results in the weak‐coupling limit. Journal of the Atmospheric Sciences, 50(21), 3504–3522. https://doi.org/10.1175/1520‐0469(1993)050%3C3504:MOI TOI%3E2.0.CO;2 Neelin, J. D., Latif, M., & Jin, F.‐F. (1994). Dynamics of coupled ocean‐atmosphere models: The tropical problem. Annual Review of Fluid Mechanics, 26, 617–659. https://doi. org/10.1146/annurev.fl.26.010194.003153 Neelin, J. D., Battisti, D., Hirst, A., Jin, F.‐F., Wakata, Y., Yamagata, T., & Zebiak, S. E. (1998). ENSO theory. Journal of Geophysical Research, 103(C7), 14262–14290. https://doi. org/10.1029/97JC03424 Neelin, J. D., Jin, F.‐F., & Syu, H.‐H. (2000). Variations in ENSO phase locking. Journal of Climate, 13(14), 2570–2590. https://doi.org/10.1175/1520‐0442(2000)013%3C2570:VIEP L%3E2.0.CO;2 Neske, S., & McGregor, S. (2018). Understanding the warm water volume precursor of ENSO events and its interdecadal variation. Geophysical Research Letters, 45(3), 1577–1585. https://doi.org/10.1002/2017GL076439 Philander, S. G. (Ed.). (1990). El Niño, La Niña, and the Southern Oscillation. London, UK: Academic Press.

Philander, S. G., & Fedorov, A. (2003). Is El Niño sporadic or cyclic? Annual Review of Earth and Planetary Sciences, 31, 579– 594. https://doi.org/10.1146/annurev.earth.31.100901.141255 Philander, S. G. H., Yamagata, T., & Pacanowski, R. C. (1984). Unstable air‐sea interactions in the tropics. Journal of the Atmospheric Sciences, 41(4), 604–613. https://doi.org/10.1175 /1520‐0469(1984)0412.0.CO;2 Picaut, J., Masia, F., & du Penhoat, Y. (1997). An advective‐ reflective conceptual model for the oscillatory nature of the ENSO. Science, 277(5326), 633–666. https://doi.org/10.1126/ science.277.5326.663 Planton, Y., Vialard, J., Guilyardi, E., Lengaigne, M., & Izumo, T. (2018). Western Pacific oceanic heat content: A better predictor of La Niña than of El Niño. Geophysical Research Letters, 45(18), 9824–9833. https://doi.org/10.1029/2018GL079341 Puy, M., Vialard, J., Lengaigne, M., & Guilyardi, E. (2015). Modulation of equatorial Pacific westerly/easterly wind events by the Madden‐Julian oscillation and convectively‐ coupled Rossby waves. Climate Dynamics, 46(7–8), 2155– 2178. https://doi.org/10.1007/s00382‐015‐2695‐x Ren, H.‐L., & Jin, F.‐F. (2013). Recharge oscillator mechanisms in two types of ENSO. Journal of Climate, 26(17), 6506–6523. https://doi.org/10.1175/JCLI‐D‐12‐00601.1 Rodgers, K. B., Friederichs, P., & Latif, M. (2004). Tropical Pacific decadal variability and its relation to decadal modulations of ENSO. Journal of Climate, 17(19), 3761–3774. https://doi.org/10.1175/1520‐0442(2004)017%3C3761:TPDV AI%3E2.0.CO;2 Stein, K., Timmermann, A., Schneider, N., Jin, F.‐F., & Stuecker, M. F. (2014). ENSO seasonal synchronization theory. Journal of Climate, 27(14), 5285–5310. https://doi. org/10.1175/JCLI‐D‐13‐00525.1 Stockdale, T.N., Anderson, D.L.T., Balmaseda, M.A., Doblas‐ Reyes, F., Ferranti, L., Mogensen, K., et al. (2011). ECMWF seasonal forecast system 3 and its prediction of sea surface temperature. Climate Dynamics, 37(3–4), 455–471. https:// doi.org/10.1007/s00382‐010‐0947‐3 Stuecker, M. F., Timmermann, A., Jin, F.‐F., McGregor, S., & Ren, H.‐L. (2013). A combination mode of the annual cycle and the El Niño/Southern Oscillation. Nature Geoscience, 6(7), 540–544. https://doi.org/10.1038/ngeo1826 Stuecker, M. F., Jin, F.‐F., & Timmermann, A. (2015a). El Niño‐Southern Oscillation frequency cascade. Proceedings of the National Academy of Sciences, 112(44), 13490–13495. https://doi.org/10.1073/pnas.1508622112 Stuecker, M. F., Jin, F.‐F., Timmermann, A., & McGregor, S. (2015b). Combination mode dynamics of the anomalous northwest Pacific anticyclone. Journal of Climate, 28(3), 1093–1111. https://doi.org/10.1175/jcli‐d‐14‐00225.1 Su, J., Zhang, R., Li, T., Rong, X., Kug, J.‐S., & Hong, C.‐C. (2010). Causes of the El Niño and La Niña amplitude asymmetry in the equatorial eastern Pacific. Journal of Climate, 23(3), 605–617. https://doi.org/10.1175/2009JCLI2894.1 Suarez, M. J., & Schopf, P. S. (1988). A delayed action oscillator for ENSO. Journal of the Atmospheric Sciences, 45(21), 3283– 3287. https://doi.org/10.1175/1520‐0469(1988)045%3C3283: ADAOFE%3E2.0.CO;2

Simple ENSO Models  151 Sun, D.‐Z. (1997). El Niño: A coupled response to radiative heating? Geophysical Research Letters, 24(16), 2031–2034. https://doi.org/10.1029/97GL01960 Takahashi, K., Karamperidou, C., & Dewitte, B. (2019). A theoretical model of strong and moderate El Niño regimes. Climate Dynamics, 52(12), 7477–7493. https://doi. org/10.1007/s00382‐018‐4100‐z Timmermann, A., & Jin, F.‐F. (2002). A nonlinear mechanism for decadal El Niño amplitude changes. Geophysical Research Letters, 29(1), 1003. https://doi.org/10.1029/2001GL013369 Timmermann, A., Jin, F.‐F., & Abshagen, J. (2003). A nonlinear theory for El Niño bursting. Journal of the Atmospheric Sciences, 60(1), 152–165. https://doi.org/ 10.1175/1520‐0469 (2003)0602.0.CO;2 Timmermann, A., An, S.‐I., Kug, J.‐S., Jin, F.‐F., Cai, W., Capotondi, et al. (2018). El Niño‐Southern Oscillation complexity. Nature, 559(7715), 535–545. https://doi.org/10.1038/ s41586‐018‐0252‐6 Tziperman, E., Stone, L., Cane, M. A., & Jarosh, H. (1994). El  Niño chaos: Overlapping of resonances between the seasonal cycle and the Pacific ocean‐atmosphere oscillator. Science, 264(5155), 72–74. https://doi.org/10.1126/ science.264.5155.72 Tziperman, E., Zebiak, S. E., & Cane, M. A. (1997). Mechanisms of seasonal‐ENSO interaction. Journal of the Atmospheric Sciences, 54(1), 61–71. https://doi.org/10.1175/1520‐0469(199 7)054%3C0061:MOSEI%3E2.0.CO;2 Tziperman, E., Cane, M. A., Zebiak, S. E., Xue, Y., & Blumenthal, B. (1998). Locking of El Niño’s peak time to the end of the calendar year in the delayed oscillator picture of ENSO. Journal of Climate, 11(9), 2191–2199. https://doi.org/ 10.1175/1520‐0442(1998)0112.0.CO;2 Vecchi, G. A., & Harrison, D. E. (2000). Tropical Pacific sea surface temperature anomalies, El Niño and equatorial westerly events. Journal of Climate, 13(11), 1814–1830. https: //doi.org/10.1175/1520‐0442(2000)013 2.0.CO;2 Vecchi, G. A., & Harrison, D. E. (2006). The termination of the 1997–98 El Niño. Part I: Mechanisms of oceanic change. Journal of Climate, 19(12), 2633–2646. https://doi.org/10.1175/ JCLI3776.1

Vialard, J., Menkes, C., Boulanger, J.‐P., Delecluse, P., Guilyardi, E., McPhaden, M. J., & Madec, G. (2001). A model study of oceanic mechanisms affecting equatorial Pacific sea surface temperature during the 1997–98 El Niño. Journal of Physical Oceanography, 31(7), 1649–1675. https://doi.org/10.1175/152 0‐0485(2001)031%3C1649:AMSOOM%3E2.0.CO;2 Wang, B., Wu, R., & Lukas, R. (1999). Roles of the western North Pacific wind variation in thermocline adjustment and ENSO phase transition. Journal of the Meteorological Society of Japan. Ser. II, 77(1), 1–16. https://doi.org/10.2151/jmsj1965.77.1_1 Webster, P. J. (1973). Remote forcing of the time‐independent tropical atmosphere. Monthly Weather Review, 101(1), 58–68. https://doi.org/10.1175/1520‐0493(1973)101%3C0058:RFOT TT%3E2.3.CO;2 Wittenberg, A. T., Rosati, A., Delworth, T. L., Vecchi, G. A., & Zeng, F. (2014). ENSO modulation: Is it decadally predictable? Journal of Climate, 27(7), 2667–2681. https://doi. org/10.1175/JCLI‐D‐13‐00577.1 Wyrtki, K. (1985). Water displacements in the Pacific and the genesis of El Niño cycles. Journal of Geophysical Research: Oceans, 90(C4), 7129–7132. https://doi.org/10.1029/ JC090iC04p07129 Xie, R., & Jin, F.‐F. (2018). Two leading ENSO modes and El Niño types in the Zebiak‐Cane model. Journal of Climate, 31(5), 1943–1962. https://doi.org/10.1175/jcli‐d‐17‐0469.1 Yu, L., Weller, R. A., & Liu, T. W. (2003). Case analysis of a role of ENSO in regulating the generation of westerly wind bursts in the western equatorial Pacific. Journal of Geophysical Research, 108(C4), 3128. https://doi.org/10.1029/2002 JC001498 Zavala‐Garay, J., Zhang, C., Moore, A. M., & Kleeman, R. (2005). The linear response of ENSO to the Madden-Julian Oscillation. Journal of Climate, 18, 2441–2459. https://doi. org/10.1175/JCLI3408.1 Zebiak, S. E. (1982). A simple atmospheric model of relevance to El Niño. Journal of the Atmospheric Sciences, 39(9), 2017– 2027. https://doi.org/10.1175/1520‐0469(1982)0392.0.Co;2 Zebiak, S. E., & Cane, M. A. (1987). A model El Niño‐Southern Oscillation. Monthly Weather Review, 115(10), 2262–2278. https:// doi.org/10.1175/1520‐0493(1987)1152.0.Co;2

7 ENSO Irregularity and Asymmetry Soon‐Il An1, Eli Tziperman2, Yuko M. Okumura3, and Tim Li4

ABSTRACT The El Niño Southern Oscillation (ENSO) is characterized by being irregular or nonperiodic and asymmetric between El Niño and La Niña with respect to amplitude, pattern, and temporal evolution. These observed features suggest the importance of nonlinear dynamics and/or stochastic forcing. Both nonlinear deterministic chaos and linear dynamics subject to stochastic forcing and/or to non‐normal growth were introduced to explain the irregularity of ENSO, but no consensus has been reached to date given the short observational record. As a dominant source of stochastic forcing, westerly wind bursts play a role in triggering, amplifying, and determining the irregularity and asymmetry of ENSO, which are best treated as part of the deterministic dynamics or as a multiplicative noise forcing. Various nonlinear processes are responsible for the spatial and temporal asymmetry of El Niño and La Niña, which includes nonlinear ocean advection, nonlinear atmosphere‐ocean coupling, state‐dependent stochastic noise, tropical instability waves, and biophysical processes. In addition to the internal nonlinear processes, a capacitor effect of the Indian and Atlantic Oceans and atmospheric and oceanic teleconnections from extratropical Pacific could also contribute to the temporal and amplitude asymmetry of ENSO. Despite significant progress, most state‐of‐the‐art models are still lacking in simulation of the spatial and temporal asymmetry of ENSO.

7.1. INTRODUCTION The irregularity of the El Niño Southern Oscillation (ENSO), including its amplitude, interval between events, and spatial and temporal patterns, has been explained as being either a result of large‐scale deterministic nonlinear dynamics or the result of stochastic forcing, although these theories are not mutually exclusive. In the first case, the irregularity is due to jumping between nonlinear res Department of Atmospheric Sciences, Yonsei University, Seoul, Republic of Korea 2  Department of Earth and Planetary Sciences and School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA 3  Institute for Geophysics, Jackson School of Geosciences, The University of Texas at Austin, Austin, TX, USA 4  Department of Atmospheric Sciences/IPRC, University of Hawai’i at Mānoa, Honolulu, HI, USA 1

onances of the ENSO cycle and the seasonal cycle (Tziperman et  al., 1994, 1995; Jin et  al., 1994; Chang et al., 1994; An & Jin, 2011). This theory can also account for the tendency of El Niño to peak at the end of the calendar year, explaining it through phase locking to the annual cycle (above references). Yet Stein et  al. (2010, 2011) employed seasonally modulated linear dynamics under stochastic forcing and found phase locking as well, indicating that nonlinear dynamics may not be necessary for explaining ENSO’s seasonality. In the second case, stochastic variability, representing for example short‐ term weather events, leads to the irregularity of ENSO, potentially also amplified by non‐normal transient growth (e.g., Moore & Kleeman, 1999; Penland & Sardeshmukh, 1995). Since the observational record of tropical Pacific SST is still not long enough to distinguish between these two scenarios, no consensus on this matter has been reached (Kessler, 2002). Tropical climate state is a primary factor to

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 153

154  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

determine an atmosphere‐ocean coupled stability for ENSO system (T. Li, 1997b; An & Jin, 2000; Fedorov & Philander, 2000), and for example, depending on the coupling strength, ENSO system becomes a self‐sustained and possibly chaotic oscillator under a strong coupling and a damped oscillator under a weak coupling (An & Jin, 2001). It has been suggested that some decades may be characterized by a self‐sustained, possibly chaotic dynamics, while others show a damped ENSO cycle, excited by stochastic variability (Kirtman & Schopf, 1998). However, a bifurcation between stable and unstable regimes tends to be ambiguous in the presence of noise (e.g., Levine & Jin, 2010). Westerly wind bursts (WWBs) are episodic reversals of the equatorial trade winds with a strength of 5 to 7 ms–1, zonal extent of 20–40 degrees, duration of 5–30 days, and frequency of around 5 to 10 times per year (Harrison & Vecchi, 1997; L. Yu et al., 2003; Seiki & Takayabu, 2007a). These events, a dominant source of stochastic forcing, play a role in triggering, amplifying, and even determining the spatial pattern of ENSO events (Harrison & Vecchi, 1997; Eisenman et  al., 2005; Levine & Jin, 2010; Rong et al., 2011; D. Chen et al., 2015; Hayashi & Watanabe, 2017). WWBs were initially considered as additive stochastic forcing (e.g. Moore & Kleeman, 1999), yet it became clear that they depend on the background SST and tend to occur more frequently during a developing El Niño (Verbickas, 1998; L. Yu et al., 2003; Eisenman et al., 2005). These events are thus best treated as part of the  deterministic dynamics or as a state‐dependent multiplicative noise forcing, with important implications to amplitude and predictability of El Niño events. El Niño is not a simple mirror image of its opposite phase, La Niña. El Niño’s amplitude is on average greater than that of La Niña (Deser & Wallace, 1987; Burgers & tephenson, 1999; An & Jin, 2004). El Niño is often followed by a La Niña in the following year, but the opposite is much less common (Larkin & Harrison, 2002; M. Chen et al., 2016; An & Kim, 2017). After their mature phase, many La Niñas persist through the following year, but most of El Niños tend to decay rapidly by next summer (Ohba & Ueda, 2007; Okumura & Deser, 2010; Choi et al. 2013; DiNezio & Deser, 2014; An & Kim, 2018). Strong El Niños are mainly loaded over the eastern Pacific with focusing toward the equator, whereas strong La Niñas are mostly loaded over the central Pacific with a wider latitudinal extension (Hoerling et al., 1997; Kang & Kug, 2002; Takahashi et  al., 2011; Dommenget et  al., 2013). Such amplitude/duration/transition/pattern asymmetries between El Niño and La Niña may not be surprising given the nonlinear internal dynamics and/or selective external impacts (e.g., An & Kim, 2018). Asymmetrical internal nonlinear processes that are responsible for amplitude asymmetry include the vertical ocean temperature profile (Zebiak & Cane, 1986), ocean nonlinear advection (An &

Jin, 2004; Su et  al. 2010), asymmetric equatorial wind response to SST (Kang & Kug, 2002; Frauen &  Dommenget, 2010; Choi et  al., 2013), ocean wave response to the wind stress (An & Kim, 2017, 2018), outcropping thermocline nonlinearity (Battisti & Hirst, 1989; Galanti et al., 2002; An & Jin, 2004), state‐dependent stochastic forcing (Jin et  al., 2007; Kug et  al., 2008; Rong et  al., 2011; Levine et  al., 2016; Hayashi & Watanabe, 2017), tropical instability wave activity (J. Yu & Liu, 2003; An, 2008a, 2008b), biophysical feedback (Timmermann & Jin, 2002), shortwave feedback (Lloyd et al., 2012), etc. Transition/duration asymmetry has been attributed to a selective capacitor effect of the Indian and Atlantic oceans (Ohba & Ueda, 2007; Okumura & Deser, 2010; An & Kim, 2018), development of subtropical western Pacific atmospheric circulation during decaying phase of ENSO to boost ENSO transition (B. Wang et al., 1999; B. Wang et al., 2001; Y. Li et al., 2007; B. Wu et al., 2010a), and some of aforementioned internal nonlinear processes (Choi et al., 2013; Im et al., 2015; M. Chen et al., 2016; An & Kim, 2017, 2018; M. Chen & Li, 2018). This chapter focuses on the irregularity of ENSO and on its amplitude and evolution asymmetries. In section  7.2, the origin of irregularity will be addressed together with the role of westerly wind burst events. Mechanisms for amplitude asymmetry will be discussed in section 7.3. The cause of evolution asymmetry will be reviewed in section  7.4, and we include conclusion and discussion in section 7.5. 7.2. IRREGULARITY 7.2.1. Deterministic Chaos A dynamical system can display chaotic behavior without any external stochastic forcing, the most well‐ known example of this being the three nonlinear ordinary differential equations of the Lorenz system (Lorenz, 1963). It has been suggested that the irregularity of ENSO, including its amplitude, interval between events, and spatial and temporal patterns, may be a result of such deterministic large‐scale nonlinear dynamics (Tziperman et al., 1994; Jin et al., 1994). Chaotic dynamical systems are typically characterized via the “route to chaos” they undergo as a parameter is changed. There are three possibilities (Strogatz, 1994; Ott, 2002): the period doubling route, the intermittency route, and the quasi‐ periodicity route to chaos. This last route is typical of periodically forced nonlinear oscillators and is the relevant one in the case of ENSO, where the periodic forcing is provided by the seasonal cycle and the nonlinear oscillator is ENSO itself. When the periodic forcing (seasonal cycle in the case of ENSO) is weak, the nonlinear oscillator undergoes

ENSO Irregularity and Asymmetry  155

­ scillation at a frequency that is not simply related to the o forcing frequency. This is known as the quasi‐periodic regime. As the periodic forcing amplitude is increased, the forced nonlinear oscillator can enter a nonlinear resonance with the forcing when the ratio of its frequency and that of the forcing is that of two integers, p/q. Unlike a linear oscillator, a nonlinear one can adjust its period as function of its amplitude, and thus may be in more than a single nonlinear resonance with the given periodic forcing. When the nonlinearity is even stronger, these different nonlinear resonances may coexist for exactly the same parameters, and the resonances can then become unstable. That is, there would be a solution for ENSO that is perfectly periodic, with a period of, say 3 years (3/1 resonance), and another such solution with a period of 4 years (4/1 resonance), with the actual solution and thus the period determined by the initial conditions. In this strongly nonlinear/ strongly forced regime, these solutions are unstable, so that any slight deviation from the periodic solution (due to error in the initial conditions or due to finite accuracy in the calculation of the solution), would grow exponentially fast and the solution may then switch to another nonlinear resonance and thus to a different periodicity. This would lead to random jumping between these nonlinear resonances and thus to a chaotic solution with a limited predictability and irregular period and amplitude. This mechanism for ENSO’s irregularity has been demonstrated in the context of various simple toy models (Tziperman et  al., 1994; Jin et  al., 1994; Chang et  al., 1994; An & Jin, 2011) as well as in the Cane‐Zebiak Model (Zebiak & Cane, 1987) that was the first dynamical model to be used successfully for ENSO prediction (Cane et al., 1986). Figure 7.1 shows the three regimes for the CZ model: quasi‐periodic (left), phase‐locked nonlinear resonance (middle; note in lower panel that all events occur only in January and February in this case) and chaotic (right, where events happen throughout the year, yet preferentially at the end and beginning of the calendar year). Note that “phase locking” is strictly defined as the period of ENSO being related to that of the annual cycle as the ratio of two integers p/q. In the chaotic regime, though, we use the term more loosely to denote a preferential occurrence of warm events during a certain season. The quasi‐periodicity route to chaos can be seen in these models by varying the amplitude of the seasonal cycle or of the ocean‐atmosphere coupling, for example. As the seasonal cycle amplitude is increased, ENSO is seen to first be in a periodic solution that is not associated with the seasonal cycle, then for stronger seasonal forcing it enters a nonlinear resonance with the seasonal cycle such that its period is p/q times one year. Eventually, for even stronger forcing, the ENSO cycle becomes chaotic. An & Jin (2011) showed that the frequency modulation by the annual cycle can change

ENSO’s phases and frequency, while the amplitude modulation by the annual cycle intensifies the ENSO variability and also induces seasonal amplitude locking. When ENSO is in a nonlinear resonance with the seasonal cycle it is phased locked to this cycle. If the resonance were p/q = 4/1, for example, ENSO would always occur at the same season, every four years. In the chaotic regime, the different resonances would still tend to be phase locked and ENSO would still tend to preferentially occur at a given season. This is reminiscent of the known locking of El Niño events to the annual cycle, adding to the attractivity of the deterministic chaos explanation for ENSO’s irregularity and phase locking (above references, as well as Stein et al., 2010, 2011). This still leaves open the question of what precisely is the mechanism of the phase locking, and some attempts on that were made by noting the possible seasonal amplification of the different equatorial wave modes (Tziperman et al., 1997; Galanti & Tziperman, 2000) as well as via a cloud feedback (Dommenget & Yu, 2016). Similarly, ENSO termination time was suggested to be determined by the southward migration of westerly wind anomalies from the equator associated with climatological warm pool expansion (Vecchi, 2006; McGregor et al., 2012), or by the development of an anomalous western North Pacific anticyclone (B. Wang et  al., 2000; Stuecker et al., 2013). Future work will need to attempt to identify specific evidence for the nonlinear resonances involving the seasonal cycle. This may need to be done using very long integrations of state‐of‐the‐art general circulation models that allow separating the effects of small nonlinearities from weather noise. Simpler models that have been used to study these issues are useful mostly in suggesting hypotheses but not in testing their realism. 7.2.2. Stochastic Forcing Weather variability, although deterministic, has a much shorter timescale than that of ENSO, allowing us to treat it as noise, or stochastic forcing. The solution to a simple linear dynamical system dx/dt = Ax, where x(t) is a vector (say SST or thermocline depth at a set of grid points covering the tropical Pacific), and A a matrix, would ultimately decay if the eigenvalues of A all have negative real parts. However, if the matrix A is non‐normal, that is, if AAT ≠ ATA, its eigenvectors are not orthogonal and then x(t) may display potentially large growth before decaying (Farrell, 1988; Farrell & Ioannou, 1996). The initial conditions x(t = 0) of a unit norm, |x(t = 0)| = 1, leading to this non‐normal growth are known as optimal initial conditions. Such optimal initial conditions may be excited by noise and then amplified by non‐normal growth. It has been suggested that the warming during El Niño events is due to such non‐normal growth (e.g. Moore & Kleeman, 1999; Penland & Sardeshmukh, 1995).

156  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE 3.0

1.2

1.2

Nino3

0.6 0

0

–0.6

–0.6

–1.2

0

4

8 14 20 time (years)

26

–1.2

1.0 0 0

4

8

14

20

–1.0

26

time (years)

0

4

8

14

20

26

time (years)

200

20 # events

2.0

0.6

40

150

15 10

100

5

50 1 2 3 4 5 6 7 8 9 10 11 12

30 20 10 1 2 3 4 5 6 7 8 9 10 11 12

month

month

1 2 3 4 5 6 7 8 9 10 11 12 month

Figure 7.1  Transition to chaos of the CZ model as the seasonal forcing is increased. Upper panels: time series of the model Niño-3 index, lower panels: histogram of number of warm events per month. Left: quasi‐periodic regime obtained for perpetual July background state (no seasonal cycle), with a period of about 4.3 years that is not simply related to the annual cycle. Middle: phase‐locked regime, obtained for weak seasonality (background seasonality amplitude set to 0.2 times the standard), very nearly 4‐year period. Right: chaotic regime in the standard parameter with a full‐amplitude seasonal cycle.

Since the observational record of tropical Pacific SST is not sufficiently long to distinguish between the chaotic deterministic dynamics and stochastically driven random scenarios for ENSO’s irregularity, no consensus on this matter has been reached (Kessler, 2002). The tropical climate background state determines the atmosphere‐ocean coupled stability for ENSO (An & Jin, 2000; Fedorov & Philander, 2000), which can put ENSO in a self‐sustained and possibly chaotic regime under a strong coupling and a damped oscillator under a weak coupling. It has been suggested that some decades may be characterized by self‐ sustained, possibly chaotic dynamics, while others show a damped ENSO cycle, excited by stochastic variability (Kirtman & Schopf, 1998). This is further complicated by the presence of noise, as a bifurcation between stable and unstable regimes tends to be ambiguous then (e.g., Stone et al., 1998; Levine & Jin, 2010). Verification using fully‐ coupled general circulation models is complicated by the presence of weather noise in these models. In principle, it is possible to differentiate between irregularity due to weather noise and large‐scale nonlinear dynamics by examining the fractal dimension of the motion (Tziperman et al., 1994), yet this is complicated by the need to run the model for very long periods in order to reliably estimate the dimension, which is challenging for computationally expensive state‐of‐the‐art general circulation models. Finally, we note that non‐normal dynamics can play an important role in a purely deterministic and chaotic ENSO regime as well (Samelson & Tziperman, 2001).

7.2.3. Role and Dynamics of WWBs as State‐ Dependent ENSO Forcing WWBs trigger Kelvin waves that play a significant role in ENSO events (McPhaden et  al., 1992; Kessler et  al., 1995; Levine & Jin, 2010; D. Chen et al., 2015; Hayashi &Watanabe, 2017). Due to their short timescale, WWBs were initially considered as additive stochastic forcing (e.g. Moore & Kleeman, 1999). To analyze this, consider a stochastically driven set of linear ordinary differential equations, dx/dt = Ax + ξν(t), where ξ is a constant unit norm vector, |ξ| = 1, and ν(t) is a scalar stochastic forcing, say a gaussian white noise. One may now calculate the “stochastic optimals,” that is, the shape of ξ that leads to the maximum variance of x(t). It has been suggested by the above references that WWBs have a shape that is close to the stochastic optimals for ENSO, or that they excite anomalies (say in thermocline depth or SST) that are close to the optimal initial conditions that lead to strong El Niño growth. This would make WWBs an especially powerful stochastic forcing of ENSO. Yet it became clear that these events that strongly depend on the state of the SST tend to occur much more frequently during an already developing El Niño (Verbickas, 1998; L. Yu et al., 2003). That is, these events cannot be seen as a purely random wind forcing that is then amplified by the Bjerknes feedback. Instead, the occurrence of these events, as well as their location, scale, amplitude, and duration, while having a stochastic

ENSO Irregularity and Asymmetry  157

element, all strongly depend on the developing SST of warm events (Tziperman & Yu, 2007). These events are thus best treated as part of the deterministic dynamics. The SST dependence of WWBs means that rather than being an additive random forcing, they can be thought of as being “multiplicative random forcing” (e.g. Perez et al., 2005; Jin et al., 2007). The dependence of WWB probability of occurrence, amplitude, scale etc., on the SST has important implications to the amplitude and predictability of El Niño events (Eisenman et  al., 2005). The SST dependence of the WWB characteristics can be extracted using SVD analysis of the covariance matrix between these characteristics and the SST (Tziperman & Yu, 2007), and this allows to parameterize WWBs in ENSO models whose atmospheric component cannot produce these events realistically (Gebbie et  al., 2007; Gebbie & Tziperman, 2009; Lopez et  al., 2013). In addition to the fact that WWBs are best described as a state‐dependent multiplicative forcing, it has been shown that while WWBs have a near‐synoptic timescale, only the slow frequency component of these events is able to affect the ENSO cycle. This has been demonstrated for general noise forcing by Roulston & Neelin (2000), Levine & Jin (2010), and in the context of WWBs by Eisenman et al. (2005). The causes and dynamics of WWBs are still not well understood. WWBs are associated with a rapid intensification of atmospheric convection (Nitta & Motoki, 1987) and are more likely to occur in the boreal winter and less during cold ENSO conditions (Giese & Harrison, 1991; Harrison & Vecchi, 1997). WWBs have been associated with cold surges from midlatitudes (Chu, 1988), single and paired tropical cyclones (Keen, 1982; Nitta, 1989), Rossby waves (Kiladis & Wheeler, 1995; Puy et al., 2016), and inconclusively, to the Madden Julian Oscillation (MJO; C. Zhang, 1996; Seiki & Takayabu, 2007a, 2007b; Chiodi et  al., 2014; Slingo et  al., 1999). WWBs in the Community Climate System Model were found comparable to observations in many aspects (Lian et al., 2018). Yet the variability of these events (Fasullo &  Webster, 2000) makes it difficult to identify a unique mechanism (Lengaigne et  al., 2004). It seems that the MJO may modulate the frequency and characteristics of WWBs (Seiki & Takayabu, 2007a, 2007b; Chiodi et al., 2014; Puy et al., 2016), but that the WWB mechanism is independent of the MJO. A recent work (Fu & Tziperman, 2019) finds that convective heating plays a key role in the generation of model WWBs. Furthermore, wind‐induced surface heat exchange acts on a short time scale of about 2 days to dramatically amplify the model WWB winds near the peak of the event. On the other hand, it is found that radiation feedbacks (long wave and short wave) and sensible surface heat flux are not essential for the development of model WWBs.

7.3. ENSO AMPLITUDE ASYMMETRY The 1982–1983, 1997–1998 and 2015–2016 El Niños are usually called “extreme El Niños.” Although there is no consensus on the definition of an extreme El Niño, so far there has been no La Niña event comparable to such strong El Niño events. This amplitude asymmetry between El Niño and La Niña can be featured by a positively skewed probability distribution of ENSO index (Figure 7.5) (e.g. Burgers & Stephenson, 1999; Deser & Wallace, 1987) as well as a horizontal pattern of skewness of SST anomalies (Figure  7.5). Not only the strong positive skewness over the eastern Pacific but also weak negative skewness over the western Pacific were found mainly because of a pattern asymmetry between El Niño and La Niña (e.g. Burgers & Stephenson, 1999; Takahashi et al., 2011; Dommenget et al., 2013). Such higher order moment implies nonlinearity in a tropical atmosphere‐ ocean coupled system or asymmetric impact of external forcing. In this section, we introduce the current hypotheses on driving mechanisms of the amplitude asymmetry (section 7.3.1) and the extreme El Niño (section 7.3.2), as well as a conceptual model to explain the amplitude asymmetry (section  7.3.3). Finally, the amplitude asymmetry of ENSO appearing in climate system models is shown (section 7.3.4). 7.3.1. Cause of Amplitude Asymmetry Why is El Niño greater than La Niña? Since there is no strong evidence for a comparable asymmetry in an external forcing so far, it is highly likely that it is related to a nonlinear nature in tropical coupled ocean‐atmosphere system, particularly in its feedbacks. “Bjerknes feedback,” referring to a positive feedback between the equatorial surface winds and zonal SST contrast between equatorial western and eastern Pacific (Bjerknes, 1966, 1969) was known as a major growing mechanism of both El Niño and La Niña. Recent studies anatomized Bjerknes feedback to figure out detailed processes, and the positive feedback on ENSO system was revealed to be quite nonlinear, especially associated with nonlinear response of atmospheric pattern to SST anomalies (Kang & Kug, 2002; Im et al., 2015). These nonlinear Bjerknes feedback processes turn out to be responsible for amplitude asymmetry of ENSO (see Figure 7.2). One potentially important oceanic process in nonlinear Bjerknes feedback is a nonlinear dynamical heating (NDH), which indicates three‐dimensional adiabatic heat flux. The NDH produces positive SST tendency over the equatorial central‐to‐eastern Pacific by its vertical component (An & Jin, 2004) and far eastern Pacific by its zonal and meridional components (Su et  al., 2010), regardless of signs of SSTA (sea surface temperature

158  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

anomalies), and thus it enhances El Niño but suppresses La Niña (An, 2009). However, there is an unresolved issue on the actual role of each component of NDH. For example, earlier work by Zebiak and Cane (1986) and An and Jin (2004) showed that nonlinear vertical temperature advection results in the asymmetry between the amplitudes of El Niño and La Niña. Yet Su et al. (2010) found, using three different ocean reanalysis products, that nonlinear zonal and meridional advection plays a crucial role in leading to an El Niño amplitude that is larger than La Niña’s, while nonlinear vertical advection plays an opposite role, specifically over the far eastern Pacific. Moreover, the roles of nonlinear vertical temperature advection, especially during the development phase of El Niño, were inconsistent among ocean assimilation products (Su et al., 2010), and thus further analysis using an advanced assimilation data combined with higher quality observations in both space and time need to be pursued to resolve this issue. Another nonlinear oceanic process is related to the activity of tropical instability waves (TIWs), especially over off‐equatorial eastern Pacific. TIW is an intraseasonal oceanic phenomenon driven by barotropic and baroclinic instabilities, and thus relatively slowly varying El Niño and La Niña could not only modify their activity but also be influenced by them. During La Niña, the enhanced meridional SST gradient intensifies TIW activity, while during El Niño, TIW activity is suppressed (An, 2008a, 2008b; J. Yu & Liu, 2003). Therefore, strong

lateral mixing by TIW suppresses La Niña but does not influence El Niño much (Vialard et al., 2001). Large‐scale ocean waves driven by ENSO‐related wind stress are also responding asymmetrically. Oceanic wave response to wind stress depends on not only the wind stress itself but also thermocline depth. In particular, the equatorial Rossby wave response to the wind stress curl becomes more sensitive during El Niño compared with La Niña because of the shallow western Pacific thermocline depth during El Niño (An & Kim, 2017), which could cause amplitude asymmetry (Im et al., 2015). In addition to oceanic nonlinear process, the atmospheric response in ENSO timescale to SSTA between El Niño and La Niña have systematic differences. The equatorial zonal wind stress response to El Niño–induced SSTA is stronger than that to La Niña–induced SSTA (Choi et  al., 2013; Frauen & Dommenget, 2010; Kang &  Kug, 2002). This amplitude asymmetry in the wind response is also related to its pattern asymmetry (Kang & Kug, 2002). Usually the major surface wind patch of El Niño is located further east than that of La Niña. This eastward shift is related to the nonlinearity in the response of deep convection to SSTA (Ham & Kug, 2012; Hoerling et  al., 1997; Kang & Kug, 2002). Another nonlinear feedback process is a nonlinear shortwave‐cloud‐SST interaction (T. Li & Philander, 1996; T. Li, 1997a; Lloyd et  al., 2012). This shortwave feedback depends on the atmospheric stability. Over convectively unstable regions, the shortwave surface heat flux is reduced by more con-

Nonlinear shortwave feedback

Asymmertric wind response

State-dependent Stochastic forcing

w′ Rossby wave

Tropical Instability Waves

10°N

[τ] = μα E Equator

dT′ u′ dx

dT′ dz Nonlinear dynamical heating

Kelvin wave

4 3 2

Bio-Physical feedback

1 0 –1

Nonlinear oceanic wave response

–2 –3 –4

160E

160W

120W

80W

Figure 7.2  Schematic diagram for nonlinear processes responsible for asymmetric amplitude of ENSO (see text).

ENSO Irregularity and Asymmetry  159

vective clouds associated with the increase in SSTAs, and the opposite case occurs with the decrease in SSTA. While over stable regions, the destabilizing effect on the atmospheric boundary layer due to warmer SSTA destructs the stratiform layer clouds and leads to an increase in the shortwave surface heat flux. To colder SSTA, the situation becomes opposite. Therefore, the shortwave feedback becomes either positive or negative, depending on atmospheric stability condition. Furthermore, the increase (decrease) in convective clouds associated with positive (negative) SSTA enhances (reduces) the greenhouse effect, and thus the longwave surface flux feedback is  positive. Actually, the thermodynamical damping is stronger during El Niño compared to La Niña, which is mainly attributed to the difference in damping by the shortwave feedback between El Niño and La Niña (Im et al., 2015). Finally, more active westerly wind bursts during El Niño compared with La Niña was suggested for the amplitude asymmetry (Jin et  al., 2007) by enhancing a positive feedback of ENSO system during El Niño via a multiplicative noise effect (Levine & Jin, 2010). ENSO influences a supply of nutrients in ocean’s surface by changing upwelling, leading to change in phytoplankton concentration; the change in phytoplankton biomass in turn affects ocean mixed layer temperature by modifying the penetration of solar radiation. During La Niña especially, phytoplankton blooming due to the enhanced nutrient supply associated with strong upwelling leads to surface warming, thereby damping La Niña. As a result, the biophysical feedback leads to amplitude asymmetry, of which efficiency is further enhanced during La Niña because of shallower mixed depth (Marzeion et al., 2005; Timmermann & Jin, 2002). 7.3.2. Extreme El Niño Formation Mechanisms The strong El Niño events in 2015–2016, 1997–1998, and 1982–1983, referred to here as extreme El Niño, caused remarkably devastating weather and climate (floods, droughts, heat waves, and hurricanes) around the world. An extreme El Niño was anticipated in early 2014 (Tollefson, 2014), but it had failed to materialize by the end of 2014. While the scientific community was still puzzling about the cause of the aborted El Niño event in 2014, the remnants of the decaying warming in late 2014 unexpectedly reignited in February 2015 and grew into an extreme El Niño by the end of 2015. L. Chen et al. (2016) conducted an observational analysis to reveal statistically significant different precursor signals between an extreme and a regular El Niño group. The El Niño events during 1958–2008 were separated into two groups: an extreme El Niño group and a regular El Niño group. A composite analysis showed that a

significant SSTA tendency difference between the two groups occurs during the onset phase (April–May) when the SSTA is nearly zero for both the groups. A mixed‐ layer heat budget analysis indicates that the SSTA tendency difference between the two groups arises primarily from the difference in zonal current anomaly (u’) and associated zonal advection term. The major factors that causes the u’ difference is the thermocline depth anomaly (D’) in the off‐equatorial western Pacific prior to the onset phase. A further diagnosis showed that the D’ difference is caused by the difference in the local wind stress curl anomaly regulated by anomalous SST and precipitation over the Maritime Continent and equatorial western Pacific. It is interesting to note that precursory D’ signal in 2015’s extreme El Niño was very different from that of traditional extreme El Niños such as those in 1998 and 1982 (L. Chen et al., 2017). Figure 7.3a compares the evolutions of the Niño-3 SSTA for the 2015 extreme El Niño, the traditional extreme El Niño (defined as the composite of the 1982 and 1997 events), and the regular El Niño composite. Two marked differences are worth noting. First, in contrast to the traditional extreme El Niño that started from a cold episode in the preceding year, 2015 El Niño was preceded by a weak warming peak in November 2014 (Figure  7.3a). Second, a marked turnabout of the SSTA tendency (from negative to positive) happened around February 2015. The precondition of D’ and the associated SSTA evolution differed markedly between 2015’s El Niño and the traditional extreme El Niño during initial onset stage. The ocean‐atmosphere system prior to the traditional extreme El Niño exhibited a La Niña state, as seen in Figure  7.3a. Equatorial easterly anomalies associated with the precursory cold anomaly caused anticyclonic wind stress curl anomalies, which built up positive upper‐ ocean heat content anomalies off the equator. The positive off‐equatorial D’ signals propagated westward as Rossby waves and became downwelling equatorial Kelvin waves after being reflected in the western boundary. The positive D’ led to great thermocline and zonal advective feedbacks and thus a strong positive SSTA tendency during the initial developing stage of the traditional extreme El Niño. In contrast, the pre‐onset condition of the 2015 extreme El Niño was unfavorable for the occurrence of even a moderate El Niño event. During OND[‐1], the ocean‐ atmosphere system possessed a weak and decaying El Niño pattern. A negative D’ built up over off‐equatorial western Pacific during OND[‐1]. The negative D’ was supposed to move to the equator in the following months, reducing the remnants of preceding positive thermocline anomalies at the equator. However, a positive D’ signal unexpectedly intensified over central equatorial Pacific

160  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Initial developing stage

3.0

Late developing stage

Niño3 SSTA

2.0 2015/16 EN 1.0 0.0 regular EN –1.0

TR-super EN Jan Mar May Jul Sep Nov (–1) (–1) (–1) (–1) (–1) (–1)

0.2 Accumulated WWB-index

0.18 0.16 0.14

1982

JFM MJJ JFMAJJ

Jan (0)

Mar May (0) (0)

Jul (0)

Sep Nov Jan Mar (0) (0) (+1) (+1) 2015

1997

0.12 0.01 0.08 0.06 0.04 0.02 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

0

Figure 7.3  (a) Time evolution of Niño-3 SSTA. Purple line indicates the 2015–2016 El Niño, red line indicates the composite of traditional extreme EN events (i.e. 1982–1983 EN and 1997–1998 EN), and blue line indicates the composite of the regular EN events during 1980–2015 (including 1986–1987, 1987–1988, 1991–1992, 1994– 1995, 2002–2003, 2004–2005, 2006–2007, and 2009–2010 ENs). The light blue shading indicates the intercase spread, which is estimated with the intercase standard deviation of the regular EN events. (b) Time series of the accumulated WWB index, which is obtained through integrating the high‐frequency wind stress anomalies over (5°S–5°N, 120–180°E) for January–March (JFM; red curve), May–July (MJJ; blue curve) and January–July (JFMAJJ; green curve) of each year. (Figures from L. Chen et al., 2017)

(CEP) in FM[0] and expanded into the eastern equatorial Pacific in AM[0]. The sudden emergence of this positive D’ center in CEP is responsible for the turnabout of the SSTA tendency in February 2015, as clearly shown from an ocean mixed‐layer heat budget. The sudden increase of D’ over CEP in early 2015 was attributed to exceptional WWBs (Harrison & Vecchi, 1997; Lengaigne et  al., 2004). An accumulated WWB effect was introduced by L. Chen et  al. (2017) to quantitatively measure the strength of WWBs for each year. Their calculation showed that the intensity of the WWBs in early 2015 is the strongest during the past 40 years (Figure  7.3b). Oceanic general circulation model experiments further confirmed the role of the WWB in triggering the positive D’ in early 2015. In summary, the occurrence of a series of exceptionally strong WWBs in early 2015 was the major driver to flare

up a positive D’ center over CEP and cause the 2015 extreme El Niño formation. The unique developing characteristic breaks our traditional view of El Niño formation, which emphasized the off‐equatorial thermocline recharging process. The result suggests two routes for extreme El Niño formation (Figure  7.4). The first route is the occurrence of an exceptionally strong positive precursory D’ signal in off‐equatorial western Pacific. The 1997 and 1982 events are such examples. The second route is the occurrence of exceptionally strong WWBs. The formation of the 2015 extreme El Niño is such an example. While a precursory negative off‐equatorial D’ signal favored the occurrence of thermocline shoaling at the equator in subsequent months, such a discharging process was interrupted by the consecutive extremely strong WWBs. Thus, the 2015 episode is a shining example of the importance of WWBs. They can turn

ENSO Irregularity and Asymmetry  161

around slow coupled dynamics and cause the generation of an extreme El Niño. By analyzing CMIP3 and CMIP5 model outputs, Cai et  al. (2014) found that the increase of extreme ENSO frequency under global warming arises from projected surface warming over the eastern equatorial Pacific that appears greater than in the surrounding ocean waters. Such a warming pattern facilitates more frequent occurrences of atmospheric convection in the eastern equatorial Pacific region. Takahashi and Dewitte (2016) and Takahashi et al. (2019) showed a bimodal probability distribution of ENSO amplitude in a Geophysical Fluid Dynamics Laboratory coupled model and a simple theoretical model. They suggested the bimodality arose from the existence of a threshold of the SSTA above which zonal wind response is nonlinearly enhanced. Recently, Cai et  al. (2018) demonstrated based on CMIP5 model diagnosis that more frequent eastern Pacific–type El Niños would occur in the future warmer climate. 7.3.3. Conceptual Models to Explain Amplitude and Transition Asymmetries The delayed action oscillator (Suarez & Schopf, 1988; Battisti & Hirst, 1989) and later recharge oscillator (F.‐F. Jin, 1997a, 1997b; T. Li, 1997b) well explained an oscillatory nature of ENSO by adopting the tropical air‐sea coupled feedback (Bjerknes feedback) and the slow ocean adjustment (see chapter 6 of this book). However, because

they were built on a linear dynamic framework except for a cubic term that generates a symmetric nonlinearity in the delayed action oscillator model (see Eq. 7.1), they could not explain the asymmetric features of ENSO. To compensate for this shortcoming, recent studies attempted to expand either the delayed oscillator or the recharge oscillator model to a nonlinear model by adopting the afore­ mentioned nonlinear feedback processes (e.g. Frauen & Dommenget, 2010; Choi et al. 2013; Roberts et al., 2016; An & Kim, 2017; Timmermann et al., 2018). The delayed oscillator model (DOM) is given by T t



bT t

cT

eT 3 , (7.1)

where T represents the equatorial eastern Pacific SST anomaly; b and c indicate coefficients for the delayed negative feedback via equatorial ocean wave motions and a comprehensive simultaneous positive/negative feedback (hereafter just positive feedback because c > 0) via thermodynamical and dynamical air‐sea coupling processes, respectively; τ represents the delay time; and e a symmetric nonlinear damping to restrict an exponential growth. Basically, b and c are responsible for the transition and growth of ENSO, respectively. In the original DOM (e.g. Battisti & Hirst, 1989), both b and c were constants regardless of T. However, by modifying b and/or c based on nonlinear Bjerknes feedback, amplitude/transition asymmetry can be produced from DOM (e.g. Choi et al.,

1 2015

WWE-index (JFMAMJJ)

0.9 1997

0.8 0.7

2004

2014 1982

0.6 0.5

2002 1994

1986 1991

1987

0.4

2006

2009

0.3 –0.08

–0.06

–0.04

–0.02

0

0.02

0.04

0.06

0.08

0.1

D′ over 130E-180E, 10S-10N (ASON[–1])

Figure 7.4  Scatter diagram for each El Niño since 1979 as a function of precursory thermocline anomaly signal (horizontal axis) and an accumulated WWB index (vertical axis). Green circle indicates two distinctive regimes for extreme El Niño formation: exceptionally strong WWBs versus exceptional strong D’ signal. (Figure from L. Chen et al., 2017)

162  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

2013). A simple modification that T(t) converts to T(t)  +  r|T(t)| (Choi et  al., 2013), induces DOM to be a nonlinear DOM, where r is an asymmetric factor to enhance a feedback for the positive T(t) and to suppress a feedback for the negative T(t) (in case of r > 0). More specifically, the asymmetric factor, r, has a different value depending on a modified feedback (An & Kim, 2017), such that r for a delayed negative feedback and a positive feedback are represented by rb and rc, respectively. As a consequence of this modification, amplifying a delayed negative feedback by a factor of (1 + rb)/(1 − rb) (rb > 0) when transitioning from El Niño to La Niña relative to the other way around results in a quick termination of El Niño; and enhancing a positive feedback by a factor of (1 + rc)/(1 − rc) (rc > 0) for the growth of El Niño relative to that of La Niña produces a positive skewness. With such modification, DOM becomes T    t

bT t

cT

brb T t

crc T

eT 3 (7.2)

Depending on what nonlinear Bjerkness feedback ­ rocess is considered, rb and rc are determined. For a p positive r (both rb and rc) and the corresponding parameters, the modified DOM produced a positively skewed T and a relatively fast transition from El Niño to La Niña (e.g., An & Kim, 2017; Choi et  al., 2013). For example, Choi et al. (2013) focused on an asymmetrical response in the intensity of the equatorial central Pacific winds to SSTA. In addition to the wind‐SST asymmetrical relationship, DiNezio & Deser (2014) stressed the asymmetrical response of subsurface ocean temperature to thermocline depth anomalies. These two nonlinear processes require both rb and rc to be nonzero, and thus leading to change in both amplitude and transition asymmetries of ENSO. On the other hand, asymmetry associated with the thermo‐dynamical damping such as shortwave feedback requires to modify rc ≠ 0 and rb = 0, while the asymmetry in the reflected Kelvin wave response to the wind forcing does require to modify rb ≠ 0 and rc = 0. Therefore, the former mainly modifies ENSO amplitude asymmetry, and the latter mainly modifies ENSO transition asymmetry (An & Kim, 2017, 2018). Not only DOM but also a recharge oscillator model (ROM) has been modified to lead an amplitude and transition asymmetries of ENSO. Conceptual differences between DOM and ROM and their relative merits are discussed in chapter 6 of this book. Linear ROM is given by (e.g. Timmermann et al., 2018)



d TE

dt

I BJ TE

Fh

(7.3a)

dh



h

dt

TE , (7.3b)

where TE and h represent the equatorial eastern Pacific SSTA and zonal mean thermocline depth anomaly over an equatorial Pacific, respectively. IBJ indicates the Bjerknes stability index, i.e. a collective growth/damping rate of TE; ε is a damping rate of h and related to ocean adjustment time scale; and the frequency is determined by F , which is called the Wyrtki index (Lu et  al., 2018). Bjerknes stability is composed of “Thermal advective damping by mean currents,” “Thermo‐dynamical damping,” “Thermocline feedback,” “Zonal advection feedback,” and “Ekman feedback” (Jin et al., 2006). As in DOM, for example, a nonlinear process on asymmetric wind response to positive and negative SSTA (e.g., Choi et al., 2013; Kang & Kug, 2002) can be applied to ROM by using the absolute value function nonlinearity. In other words, TE is converted to TE  +  r|TE|. Furthermore, NDH, i.e. nonlinear oceanic thermal advection (e.g. An & Jin, 2004; Boucharel et al., 2015; Jin et  al., 2003; Su et  al., 2010), can be deformed by a combination of TE and h such as 1TE2 2TE h . Finally, the state‐dependent noise forcing is adopted as σ(1 + BTE)ξt, where σξ(t) is a stochastic noise with variance and B is a positive constant (e.g. Levine & Jin, 2010). Based on above modifications, the linear ROM becomes a nonlinear ROM as follows: d TE

dt

I BJ TE

rBJ TE

1 B TE

dh

dt

h

2 1 E

Fh rB TE TE

T

TE h

2

,

(7.4a)

r TE .

(7.4b)

t

As in DOM, Bjerknes stability (IBJ) increases by a factor of (1 + rBJ)/(1 − rBJ) (rBJ > 0) for El Niño growth relative to La Niña growth, producing a positive skewness; and Wyrtki index ( F ) also increases as (rα  >  0) for discharging phase F 1 r / 1 r relative to recharging phase, leading to a relatively fast transition from El Niño to La Niña compared to that from La Niña to El Niño. The quadratic term, 1TE2 (β1  >  0), always produces a positive SST tendency regardless of sign of TE, thus leading to a positive skewness. Another quadratic term, β2TEh, is related to a duration asymmetry of ENSO through a quarter‐cycle phase difference between TE and h. The state‐dependent noise forcing is also enhanced by a factor of (1 + rB)/(1 − rB) (rB > 0) for El Niño phase relative to La Niña phase, producing a positive skewness, where rB ≠ rBJ because rB is only related to a deterministic random noise. Therefore, the nonlinear ROM obviously produces amplitude and

ENSO Irregularity and Asymmetry  163

duration/transition asymmetries of ENSO. The aforementioned nonlinear formulas may produce somewhat similar behavior because of their mathematical similarity, but quantitative comparison of each nonlinear process has not been done yet. 7.3.4. Amplitude Asymmetry in Climate Models Most of the Earth system models are suffering to simulate the nonlinear properties of ENSO, even though the simulated ENSO amplitude is rather agreeing with the observation (An et  al., 2005a; T. Zhang & Sun, 2014). Figures  7.5a and b show the tropical Pacific skewness pattern of SSTA obtained from the observation and the multimodel ensemble (MME) of 36 CMIP5 models for a historical run, respectively. The observed SST skewness pattern features a cold tongue–like pattern of a positive skewness over eastern Pacific with its maximum at the west coast of South America; a horseshoe‐like pattern of negative skewness surrounding a positive skewness; and a weak positive skewness over the subtropical northern northwestern Pacific near 130°E. SST skewness from MME is very weak compared to the observed, although the spatial pattern is somewhat similar to its counterpart of observation. Smaller skewness is clearly demonstrated in a difference map between the observation and MME (Figure 7.5c), which is very similar to the observed pattern with opposite sign. T. Zhang & Sun (2014) argued that the underestimate of ENSO asymmetry in CMIP models is caused by the weaker precipitation anomalies over the eastern Pacific and westward shift of westerly wind anomalies during El Niño. It may be related to a common bias in mean states such as the stronger trade wind, smaller warm pool size, and far westward extension of cold tongue compared to the observation (e.g., Sun et al., 2013, 2016; Zheng et al., 2012). Figure 7.5d shows variance and skewness of the Niño‐3 index obtained from the historical runs of 36 CMIP5 models and observation. The observed variance and skewness are 0.8°C2and 0.54, respectively, for the period of 1901–2005. Skewness indicates the normalized third order moment (An & Jin, 2004). Scatter plot of both variance and skewness computed from each model simulation spread quite widely. The spread range of variance is about 0.2~2.3°C2, and the MME mean variance is 0.9, indicating that the MME mean variance is close to the observed variance. The spread range of skewness is about –0.3~1.2, and the MME mean skewness is 0.18. MME mean skewness is quite small compared to the observed skewness. Moreover, 9 out of 36 models produced negative skewness of the Niño‐3 index, and the observed skewness is out of range of one standard deviation of model’s skewness. In general, the CMIP5 model’s skewness is underestimated.

There is some inconclusive evidence of a weakening of the asymmetries in the amplitude (Kohyama et al., 2018) and transition (An & Kim, 2018) of ENSO due to global warming, based on future scenario experiments of Earth system models. Dynamically, it is feasible that changes associated with global warming to the ocean stratification or to a warm pool expansion may cause such changes, but further study is needed to examine these possibilities. 7.4. ENSO EVOLUTION ASYMMETRY El Niño and La Niña exhibit distinct asymmetry not only in the amplitude but also in the temporal evolution. Kessler (2002) noted the tendency for the equatorial Pacific Ocean to remain in a weak La Niña state for a few years and questioned the cyclic nature of ENSO. The systematic difference in the evolution of El Niño and La Niña cannot be explained by linear dynamics nor stochastic atmospheric forcing. In this section, we review the ENSO evolution asymmetry in observations and climate models (section  7.4.1) and the associated mechanisms, focusing on the nonlinearities in the tropical Pacific atmosphere and ocean (section 7.4.2) and the influences from remote tropical oceans (section 7.4.3). 7.4.1. ENSO Evolution Asymmetry in Observations and Climate Models Both observed El Niño and La Niña tend to develop in late boreal spring‐summer and peak toward the end of the calendar year. On average, El Niño terminates quickly after the mature phase and transitions into a cold phase by the following summer, whereas La Niña persists throughout the second year and reintensifies in winter (Larkin & Harrison, 2002; McPhaden & Zhang, 2009; Ohba & Ueda, 2009; Okumura & Deser, 2010; Figure  7.6). Approximately two‐thirds of observed El Niño events terminate after 1 year, while nearly half of La Niña events last 2 years or longer (X. Wu et  al., 2019). The asymmetric evolution of El Niño and La Niña is a robust feature of the observed ENSO throughout the past century (Okumura & Deser, 2010) and is particularly pronounced for strong ENSO events after the 1980s (McPhaden & Zhang, 2009). However, only a handful of the current and previous generations of climate models reproduce the observed ENSO evolution asymmetry (Ohba et  al., 2010; Deser et  al., 2012; Ohba & Watanabe, 2012; Choi et  al., 2013; DiNezio et al., 2017; An & Kim, 2018). Analysis of a long control simulation of one of these models suggests that the ENSO evolution asymmetry increases with the amplitude of ENSO (Okumura et al., 2017; see section 7.3.3 for a conceptual understanding).

164  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a)

observed skewness pattern

(d)

10N

OBS MME

0 10S

1.2 150E

180

–1 –0.8 –0.6 –0.4 –0.2

(b)

150W 0

0.2

0.4

120W 0.6

0.8

90W 1

(unitless)

MME skewness pattern

10N 0 10S 120E

150E

180

–1 –0.8 –0.6 –0.4 –0.2

(c)

150W 0

0.2

0.4

120W 0.6

0.8

90W 1

Skewness (unitless)

120E

0.8

0.4

0.0

(unitless)

Mean skewness bias

–0.4

10N 0 10S

–0.5

120E

150E

180

–1 –0.8 –0.6 –0.4 –0.2

150W 0

0.2

0.4

120W 0.6 0.8

90W 1

0.0

0.5

1.0

Variance

1.5

2.0

2.5

(°C2)

(unitless)

Figure 7.5  Horizontal distribution of the normalized SST skewness from (a) observation, (b) multimodel ensemble mean, and (c) their difference. (d) Scatter plot of variance (x‐axis; °C2) and skewness (y‐axis; normalized) of Niño‐3 index (monthly‐mean SST anomaly averaged over 5°S–5°N and 150–90°W) for Jan. 1901–Dec. 2005 obtained from the historical run of 36 CMIP5 models (multimodel ensemble: blue rectangular), and observation (ERSSTv5: red dot). Climatological mean and linear trend were removed to calculate anomaly. Error bar indicates the range of ±1 standard deviation.

7.4.2. Nonlinearities in the Tropical Pacific Atmosphere and Ocean In the equatorial Pacific, surface wind anomalies drive changes in the thermocline and upwelling, which in turn affect SSTs. Early studies thus explored the atmospheric origins for the asymmetric evolution of El Niño and La Niña. Indeed, the early termination of El Niño is preceded by a rapid decay in equatorial zonal wind anomalies that begins during the mature phase (Figure 7.7). In boreal winter, when the western Pacific warm pool migrates south of the equator and the South Pacific convergence zone intensifies, the center of precipitation and zonal wind anomalies associated with El Niño shift south of the equator, hastening the discharge of the equatorial oceanic heat content and hence the event termination (Harrison & Vecchi, 1999; Vecchi, 2006; McGregor et al., 2012). The southward shift of zonal wind anomalies is pronounced for strong El Niño but inconspicuous for La Niña (Ohba and Ueda, 2009; McGregor et  al., 2013). McGregor et  al. (2012, 2013) discuss that weak background winds south of the equator during El Niño promote the southward shift of wind anomalies by reducing surface momentum damping.

The equatorial precipitation and zonal wind anomalies are also shifted to the east during El Niño compared to La Niña, and wind anomalies reverse the direction in the far western equatorial Pacific after the mature phase of El Niño (Figure  7.7). The zonal displacement of atmospheric anomalies is caused by nonlinear dependence of the atmospheric deep convection on SSTs (e.g. Graham & Barnett, 1987; Kang & Kug, 2002): over the eastern equatorial cold tongue, large positive SST anomalies can induce atmospheric deep convection while negative anomalies have no further effect on the normally dry conditions (Hoerling et al., 1997). Okumura et al. (2011) suggest that the eastward displacement of atmospheric anomalies makes surface winds over the western equatorial Pacific more susceptible to the delayed negative feedback from the Indian Ocean during El Niño compared to La Niña (section 7.4.3). The Indian Ocean, as well as changes in local SSTs, force atmospheric circulation anomalies over the northwest tropical Pacific during the mature‐decay phase of ENSO, which act to reverse the equatorial wind anomalies (B. Wang et  al., 2000; Watanabe & Jin, 2002; B. Wu et al., 2010a). These off‐equatorial atmospheric circulation anomalies are also shifted eastward during El Niño compared to La Niña

ENSO Irregularity and Asymmetry  165 3.0

El Nino

2.0 1.0 0.0 –1.0 –2.0 –3.0 Jun(–1) Dec(–1) Jun(0)

Dec(0)

Jun(1)

3.0

Dec(1)

Jun(2)

La Nina

2.0 1.0 0.0 –1.0 –2.0 –3.0 Jun(–1) Dec(–1) Jun(0)

Dec(0)

Jun(1)

Dec(1)

Jun(2)

Figure 7.6 Time series of the Niño‐3.4 index overlaid from June(–1) to June(2) for strong (top) El Niño and (bottom) La Niña events during 1948–2018 based on the HadISST dataset (Rayner et al., 2003). Thick colored curves are composite time series. Strong El Niño (La Niña) events are defined when the Niño‐3.4 index is greater than 1 (less than –1) standard deviation in December(0). The Niño‐3.4 index is smoothed with a three‐month running mean filter and a linear trend is removed from the time series prior to the analysis.

(B. Wu et al., 2010b). There is a debate over whether or not the Indian Ocean SST can force a direct wind response in western equatorial Pacific (M. Chen et al., 2016). While a basinwide SST anomaly appears in the tropical Indian Ocean, rainfall anomaly exhibits a zonal dipole pattern. A westerly anomaly rather than an easterly anomaly was simulated by an atmospheric general circulation model forced by the observed dipole heating pattern. The result indicates that the Indian Ocean capacitor effect is season dependent (B. Wu et al., 2009), and it becomes effective

only during an El Niño decaying summer (see a thorough review on this subject by T. Li et al., 2017). The surface wind anomalies are asymmetric between El Niño and La Niña not only in the spatial pattern but also in the amplitude. The zonal wind response is considerably larger for positive than negative SST anomalies, and the larger wind anomalies during El Niño are suggested to result in stronger delayed negative oceanic feedback compared to La Niña (Choi et al., 2013; Dommenget et al., 2013; DiNezio & Deser, 2014). Atmospheric general circulation models forced with perfectly symmetric positive and negative SSTA patterns successfully simulate the asymmetric pattern and strength of atmospheric response that closely resemble observations, confirming the importance of atmospheric nonlinearities (Hoerling et  al., 1997; Kang & Kug, 2002; Ohba & Ueda, 2009; Frauen & Dommenget, 2010). The prominence of the atmospheric nonlinearity does not exclude the role of nonlinear processes in the ocean, which is more challenging to analyze due to the scarcity of long‐term in situ observations. A few recent studies explored the role of oceanic nonlinearities for the asymmetric evolution of El Niño and La Niña. DiNezio and Deser (2014) suggest that the delayed thermocline feedback is more effective at terminating El Niño than La Niña. During the decay phase of El Niño, the shoaling thermocline can induce large temperature anomalies at the base of the mixed layer, whereas the deepening thermocline during the decay phase of La Niña become decoupled from the mixed layer. An et al. (2005b) note that the equatorial oceanic heat content recovers more slowly during the decay phase of La Niña than El Niño. Furthermore, An and Kim (2017) discuss that dynamical response of the ocean to surface wind anomalies is also asymmetric between El Niño and La Niña. Surface wind anomalies induce larger oceanic wave response in the western equatorial Pacific during El Niño than La Niña because the atmospheric momentum is more efficiently trapped in the relatively shallow upper ocean layer. Besides the dynamical processes of the ocean and atmosphere, the asymmetry in thermodynamic air‐sea interactions may contribute to the asymmetric evolution of El Niño and La Niña. An oceanic mixed layer heat budget analysis was carried out by M. Chen et al. (2016), who showed that dynamic and thermodynamic air‐sea interaction processes are equally important in contributing to the El Niño and La Niña evolution asymmetry and that equatorial SST anomalies dampen more strongly during the decay phase of El Niño than La Niña due to larger negative cloud and evaporation feedbacks (Table 7.1). All these nonlinear processes in the atmosphere and ocean are likely to play important roles in the asymmetric evolution of El Niño and La Niña. The relative

166  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

importance of different processes may be sensitive to the choice of dataset and analysis region, and further assessment will require strategic model experiments.

El Nino Mar(2) Dec(1)

7.4.3. Influences from Remote Tropical Oceans

Sep(1) Jun(1) Mar(1) Dec(0) Sep(0) Jun(0) Mar(0)

1 150E

150W

180

120W

90W

La Nina Mar(2) Dec(1) Sep(1) Jun(1) Mar(1) Dec(0) Sep(0) Jun(0) Mar(0)

1 150E –2

–1.5

180 –1

–0.5

150W

120W

90W

0

1

2

0.5

1.5

Figure 7.7  Longitude–time sections of SST (°C, color shading), surface wind (m s–1, vectors), and precipitation (mm day–1; positive [negative] contours in green [brown] at ±1, 3, 5, …) anomalies along the equator (3°S–3°N) for strong (top) El Niño and (bottom) La Niña based on the OISST (Reynolds et  al., 2002), NCEP–DOE Reanalysis II (Kanamitsu et al., 2002), and CMAP (P. Xie & Arkin, 1996) datasets for 1982–2018. The anomalies are composited for strong El Niño (1982, 1986, 1991, 1994, 1997, 2002, 2009, and 2015) and La Niña (1984, 1988, 1998, 2007, and 2010) years. The time axis runs from January(0) to May(2).

ENSO exerts significant impacts on the tropical Indian and Atlantic Oceans through atmospheric teleconnections (T. Li et al., 2003; Xie & Carton, 2004; Chang et al., 2006; Schott et al., 2009). The resultant SST changes in these remote tropical oceans, in turn, affect the atmospheric circulation and feed back to the ENSO. This so‐ called “capacitor effect” (Xie et  al., 2009; B. Wu et  al., 2009) is particularly pronounced for the Indian Ocean: during the mature‐decay phase of El Niño, basinwide warming of the Indian Ocean forces an atmospheric Kelvin wave and induces easterly winds in the western equatorial Pacific, hastening the termination of El Niño (Annamalai et  al., 2005; Kug & Kang, 2006; Ohba & Ueda, 2007; Yoo et al., 2010). The basinwide cooling of the Indian Ocean during the mature‐decay phase of La Niña similarly forces westerly winds over the western equatorial Pacific. However, due to the westward displacement of the Pacific atmospheric anomalies during La Niña compared to El Niño, the negative feedback from the Indian Ocean is ineffective at reversing surface wind anomalies (Okumura et al., 2011). The Indian Ocean capacitor effect itself is not the cause of the asymmetric evolution of El Niño and La Niña: it is the nonlinearity in the tropical Pacific atmosphere that makes the impact of the Indian Ocean asymmetric. Nevertheless, the basinwide SST response of the Indian Ocean is larger for El Niño than La Niña when the ENSO events concur with the Indian Ocean dipole (Hong et al., 2010), which could result in stronger negative feedback during El Niño. The inclusion of the Indian Ocean capacitor effect significantly improves the forecasts of ENSO event evolution after the mature phase only for El Niño (Ohba & Watanabe, 2012). The delayed warming and cooling of the tropical Atlantic also act to terminate the ENSO events (Ham et  al., 2013; L. Wang et  al., 2017; T. Li et  al., 2017), although the role in the asymmetric evolution of El Niño and La Niña is not clear (An & Kim, 2018). The capacitor effect of the Atlantic Ocean is suggested to have increased since the early 1990s in association with an upward swing of Atlantic multidecadal variability (L. Wang et al., 2017). Given the important role of interbasin linkages in the ENSO evolution, the three tropical oceans should be viewed as a single system linked by means of the atmospheric circulation (Dommenget & Semenov, 2006; Jansen et  al., 2009; Dommenget & Yu, 2017; see chapter 10 for further discussion on this topic).

ENSO Irregularity and Asymmetry  167 Table 7.1  Composite mixed layer temperature anomaly budget analysis during the decaying phase of El Niño and La Niña averaged in the equatorial eastern Pacific region (5°N–5°S, 180°–80°W, unit, °C/month). “Adv” denotes advection terms, “Hflx” represents heat flux terms, “Sum” is the summation of “Adv” and “Hflx,” “sw” denotes the anomalous shortwave radiation term, and “lh” is the anomalous latent heat flux term. The values are based on the ensemble average of two ocean reanalysis datasets (GODAS and SODAv2.1.6) and two surface heat flux products (NCEPv2 and OAFlux). Adapted from M. Chen et al. (2016). El Niño La Niña

dT′/dt

Adv

Hflx

Sum

–0.28 0.13

–0.12 0.06

–0.20 0.11

–0.32 0.17

7.5. CONCLUSION AND DISCUSSION In this chapter, the observed characteristics of ENSO’s irregularity and asymmetry are described, and possible physical mechanisms are discussed. These ENSO characteristics have important implications for operational forecast, as ENSO’s remote impact on global climate depends on the structure, intensity, and temporal evolution of the anomalous heating source in the tropics associated with ENSO. In spite of progress in studies of the mechanisms behind the irregularity and asymmetry of El Niño and La Niña, significant problems are still unresolved, and further studies are needed. Many current state‐of‐the‐art coupled atmosphere‐ocean general circulation models fail to capture the observed amplitude and evolution asymmetry. In this regard, future change in ENSO asymmetries revealed by global warming scenario experiments of the coupled general circulation models cannot be conclusive so far. There are some unresolved issues regarding the asymmetry in El Niño and La Niña’s amplitude and evolution. For example, it is unclear if vertical NDH (Zebiak & Cane, 1986; An & Jin, 2004) or horizontal NDH (Su et al., 2010) plays a crucial role in leading to amplitude asymmetry. During the developing phase of El Niño, vertical NDH was especially inconsistent among ocean assimilation products (Su et  al., 2010). The role of the Indian Ocean capacitor effect on a quick El Niño’s termination by inducing anomalous easterlies was questioned by M. Chen et  al. (2016), who claimed that the Indian Ocean basin warming during mature El Niño wintertime events had little effect on the easterly anomalies in the equatorial western Pacific. Furthermore, the relative role and intensity of atmospheric nonlinearity (asymmetric wind response to warm and cold phase) and oceanic nonlinearities (NHD, thermocline outcropping, etc.) in producing amplitude and transition asymmetries of ENSO system have never been precisely compared. It must be very hard because as with a linear air‐sea coupling, nonlinear processes in the atmosphere and ocean are interacting. In addition to the asymmetry in their amplitude and evolution, El Niño and La Niña also exhibit a pattern asymmetry (see chapter  4 on ENSO diversity). For

u T/ x –0.19 0.10

sw′

lh′

–0.07 0.04

–0.14 0.06

example, El Niño events may be centered over either the central Pacific or the eastern Pacific, while La Niña’s cold SST pattern is typically in between these El Niño warm spots (e.g. Kug & Ham, 2011). It is not clear what mechanisms are responsible for this ENSO pattern asymmetry, although it is likely related to the amplitude and evolution asymmetry discussed above. While simple models such as those introduced in earlier sections are a useful tool in conceptually understanding ENSO’s complicated behavior, the cause of ENSO’s amplitude asymmetry may be beyond the scope of such models. For example, most prototype ENSO models represent explicitly only one or two spatial locations, such as the eastern and western Pacific, which does not allow for amplitude asymmetry caused by pattern asymmetry. An effort is required to reveal the cause of failure of current state‐of‐the‐art coupled general circulation models in capturing the observed amplitude, structure, and evolution asymmetry of ENSO. ACKNOWLEDGMENTS S.‐I. An was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF‐2017R1A2A2A05069383, NRF‐2018R1A 5A1024958) and appreciates J.‐W. Kim for drawing Figure 7.5. E. Tziperman was supported by the NSF climate dynamics program, grant AGS‐1826635, and by a Harvard‐UTEC collaborative grant, and would like to thank the Weizmann Institute for its hospitality during parts of this work. Y. Okumura was supported by the US National Oceanic Atmosphere Administration (NA17O­AR 4310149) and National Science Foundation (OCE1756883). T. Li was supported by NSFC grant 41630423, NSF grant AGS-2006553, and NOAA grant NA18OAR4310298.

REFERENCES Annamalai, H., Xie, S.‐P., McCreary, J. P., & Murtugudde, R. (2005). Impact of Indian Ocean sea surface temperature on a developing El Niño. Journal of Climate, 18, 302–319. https:// doi.org/10.1175/JCLI‐3268.1

168  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE An, S.‐I. (2008a). Interannual changes in the variability of tropical Pacific instability waves. Asia‐Pacific Journal of Atmospheric Sciences, 44, 249–258. An, S.‐I. (2008b). Interannual variations of the tropical ocean instability wave & ENSO. Journal of Climate, 21, 3680–3686. https://doi.org/10.1175/2008JCLI1701.1 An, S.‐I. (2009). A review of interdecadal changes in the nonlinearity of the El Ninño‐Southern Oscillation. Theoretical & Applied Climatology, 97, 29–40. https://doi.org/10.1007/ s00704‐008‐0071‐z An, S.‐I., Ham, Y.‐G., Jin, F.‐F., Kug, J.‐S, & Kang, I.‐S. (2005a). El Niño La Niña asymmetry in the Coupled Model Intercomparison Project simulations. Journal of Climate, 18, 2617–2627. https://doi.org/10.1175/JCLI3433.1 An, S.‐I., Hsieh, W.‐W., & Jin, F.‐F. (2005b). A nonlinear analysis of the ENSO cycle and its interdecadal changes. Journal of Climate, 18, 3229–3239. https://doi.org/10.1175/JCLI3466.1 An, S.‐I., & Jin, F.‐F. (2000). An eigen analysis of the interdecadal changes in the structure and frequency of ENSO mode. Geophysical Research Letters, 27, 1573–1576. An, S.‐I., & Jin, F.‐F. (2001). Collective role of thermocline and zonal advective feedbacks in the ENSO mode. Journal of Climate, 14, 3421–3432. An, S.‐I., & Jin, F.‐F. (2004). Nonlinearity and asymmetry of ENSO. Journal of Climate, 17, 2399–2412. https://doi.org/10. 1175/1520‐0442(2004)0172.0.CO;2 An, S.‐I., & Jin, F.‐F. (2011). Linear solutions for the frequency and amplitude modulation of ENSO by the annual cycle. Tellus, 63, 238–243. An, S.‐I., & Kim, J.‐W. (2017). Role of nonlinear ocean dynamic response to wind on the asymmetrical transition of El Niño and La Niña. Geophysical Research Letters, 44, 393–400. https://doi.org/10.1002/2016GL071971 An, S.‐I., & Kim, J.‐W. (2018). ENSO Transition asymmetry: Internal and external causes and intermodel diversity. Geophysical Research Letters, 45, 5095–5104. https://doi. org/10.1029/2018GL078476 Battisti, D. S., & Hirst, A. C. (1989). Interannual variability in a tropical atmosphere–ocean model: Influence of the basic state, ocean geometry and nonlinearity. Journal of the Atmospheric Sciences, 46, 1687–1712. Bjerknes, J. (1966). A possible response of the atmospheric Hadley circulation to equatorial anomalies of ocean temperature. Tellus, 18, 820–829. https://doi.org/10.3402/tellusa. v18i4.9712 Bjerknes, J. (1969). Atmosperic teleconection from the equatorial Pacific. Monthly Weather Review, 97, 163–172. https://doi.org /10.1175/1520‐0493(1969)0972.3.CO;2 Boucharel, J., Timmermann, A., Santoso, A., England, M.H., Jin, F.‐F., & Balmaseda, M. A. (2015). A surface layer variance heat budget for ENSO. Geophysical Research Letters, 42, 3529–3537. https://doi.org/10.1002/2015GL063843 Burgers, G., & Stephenson, D. B. (1999). The “normality” of El Niño. Geophysical Research Letters, 26, 1027–1030. Cai, W., et al. (2014). Increasing frequency of extreme El Niño events due to greenhouse warming. Nature Climate Change, 4, 111–116. Cai, W., Wang, G., Dewitte, B., Wu, L., Santoso, A., Takahashi, K., et  al. (2018). Increased variability of eastern Pacific El Niño under greenhouse warming. Nature, 564, 201–206.

Cane, M. A., Zebiak, S. E., & Dolan, S. C. (1986). Experimental forecasts of El Niño. Nature, 321, 827–832. Chang, P., Wang, B., Li, T., & Ji, L. (1994). Interactions between the seasonal cycle & the Southern Oscillation: Frequency entrainment & chaos in a coupled ocean‐atmosphere model. Geophysical Research Letters, 21, 2817–2820. Chang, P., Yamagata, T., Schopf, P., Behera, S. K., Carton, J., Kessler, W. S., et al. (2006). Climate fluctuations of tropical coupled systems: The role of ocean dynamics. Journal of Climate, 19, 5122–5174. https://doi.org/10.1175/JCLI3903.1 Chen, D. et al. (2015). Strong influence of westerly wind bursts on El Niño diversity. Nature Geoscience, 8, 339–345. Chen, L., Li, T., Behera, S. K., & Doi, T. (2016a). Distinctive precursory air–sea signals between regular and super El Niños. Advances in Atmospheric Sciences, 33, 996–1004. Chen, L., Li, T., Wang, B., & Wang, L. (2017). Formation mechanism for 2015/16 super El Niño. Scientific Reports, 7, 2975. doi:10.1038/s41598‐017‐02926‐3 Chen, M., Li, T., Shen, X., & Wu, B. (2016b). Relative roles of dynamic and thermodynamic processes in causing evolution asymmetry between El Niño and La Niña. Journal of Climate, 29, 2201–2220. https://doi.org/10.1175/JCLI‐D‐15‐0547.1 Chen, M., & Li, T. (2018). Why 1986 El Niño and 2005 La Niña evolved different from a typical El Niño and La Niña. Clim. Dyn., 51, 4309–4327. doi: 10.1007/s00382‐017‐3852‐1 Chiodi, A. M., Harrison, D. E., & Vecchi, G. A. (2014). Subseasonal atmospheric variability and El Niño waveguide warming: Observed effects of the Madden–Julian oscillation and westerly wind events. Journal of Climate, 27, 3619–3642. Choi, K. Y., Vecchi, G. A., & Wittenberg, A. T. (2013). ENSO transition, duration, and amplitude asymmetries: Role of the nonlinear wind stress coupling in a conceptual model. Journal of Climate, 26, 9462–9476. https://doi.org/10.1175/JCLI‐ D‐13‐00045.1 Chu, P.‐S. (1988) Extratropical forcing and the burst of equatorial westerlies in the western Pacific: A synoptic study. J. Meteorol. Soc. Jpn, 66, 549–563. Deser, C., Phillips, A. S., Tomas, R. A., Okumura, Y. M., Alexander, M. A., Capotondi, A., et  al. (2012). ENSO and Pacific decadal variability in the Community Climate System Model Version 4. Journal of Climate, 25, 2622–2651. https:// doi.org/10.1175/JCLI‐D‐11‐00301.1 Deser, C., & Wallace, J. M. (1987). El Niño events & their relation to the southern oscillation: 1925–1986. Journal of Geophysical Research: Oceans, 92, 14189–14196. DiNezio, P. N., & Deser, C. (2014). Nonlinear controls on the persistence of La Niña. Journal of Climate, 27, 7335–7355. https://doi.org/10.1175/JCLI‐D‐14‐00033.1 DiNezio, P. N., Deser, C., Okumura, Y., & Karspeck, A. (2017). Predictability of 2‐year La Niña events in a coupled general circulation model. Climate Dynamics, 49, 4237–4261. https:// doi.org/10.1007/s00382‐017‐3575‐3 Dommenget, D., Bayr, T., & Frauen, C. (2013). Analysis of the non‐linearity in the pattern and time evolution of El Niño southern oscillation. Climate Dynamics, 40, 2825–2847. https://doi.org/10.1007/s00382‐012‐1475‐0 Dommenget, D., & Semenov, V. (2006). Impacts of the tropical Indian and Atlantic Oceans on ENSO. Geophysical Research Letters, 33, L11701. https://doi.org/10.1029/2006GL025871

ENSO Irregularity and Asymmetry  169 Dommenget, D., & Yu, Y. (2016) The seasonally changing cloud feedbacks contribution to the ENSO seasonal phase‐locking. Climate Dynamics, 47, 3661–3672. Dommenget, D., &Yu, Y. (2017). The effects of remote SST forcings on ENSO dynamics, variability and diversity. Climate Dynamics, 49, 2605–2624. Eisenman, I., Yu, L., & Tziperman, E. (2005). Westerly wind bursts: ENSO’s tail rather than the dog? Journal of Climate, 18, 5224–5238. Farrell, B. (1988). Optimal excitation of neutral Rossby waves. Journal of the Atmospheric Sciences, 45, 163–172. Farrell, B. F., & Ioannou, P. J. (1996). Generalized stability theory. Part I: Autonomous operators. Journal of the Atmospheric Sciences, 53, 2025–2040. Fasullo, J., & Webster, P. J. (2000). Atmospheric and surface variations during westerly wind bursts in the tropical western Pacific. Quarterly Journal of the Royal Meteorological Society, 126, 899–924. Fedorov, A. V., & Philander, S. G. (2000). Is El Niño changing? Science, 288, 1997–2002. Frauen, C., & Dommenget, D. (2010). El Niño and la Niña amplitude asymmetry caused by atmospheric feedbacks. Geophysical Research Letters, 37, 1–6. https://doi. org/10.1029/2010GL044444 Fu, M., & Tziperman E. (2019). Essential ingredients to the dynamics of westerly wind bursts. Journal of Climate. doi: 10.1175/JCLI‐D‐18‐0584.1 Galanti, E., & Tziperman, E. (2000). ENSO’s phase locking to the seasonal cycle in the fast‐SST, fast‐wave, and mixed‐mode regimes. Journal of the Atmospheric Sciences, 57, 2936–2950. Galanti, E., Tziperman, E., Harrison, M., Rosati, A., Giering,  R., and Sirkes, Z. (2002). The equatorial thermocline outcropping: A seasonal control on the tropical Pacific ocean‐atmosphere instability strength. Journal of Climate, 15, 2721–2739. Gebbie, G., Eisenman, I., Wittenberg, A., & Tziperman, E. (2007). Modulation of westerly wind bursts by sea surface temperature: A semistochastic feedback for ENSO. Journal of the Atmospheric Sciences, 64, 3281–3295. doi:10.1175/ JAS4029.1 Gebbie, G., & Tziperman, E. (2009). Predictability of SST‐modulated westerly wind bursts. Journal of Climate, 22, 3894–3909. Giese, B. S., & Harrison, D. E. (1991). Eastern equatorial Pacific response to three composite westerly wind types. Journal of Geophysical Research: Oceans, 96, 3239–3248. Graham, N. E., & Barnett, T. P. (1987). Surface temperature, surface wind divergence, and convection over tropical oceans. Science, 238, 657–659. https://doi.org/10.1126/science.238.4827.657 Ham, Y.‐G., & Kug, J.‐S (2012). How well do current climate models simulate two types of El Niño? Climate Dynamics, 39, 383–398. https://doi.org/10.1007/s00382‐011‐1157‐3 Ham, Y.‐G., Kug, J.‐S., Park, J.‐Y., & Jin, F.‐F. (2013). Sea surface temperature in the north tropical Atlantic as a trigger for El Niño/Southern Oscillation events. Nature Geoscience, 6, 112–116. https://doi.org/10.1038/ngeo1686 Harrison, D. E., & Vecchi. G. A. (1997). Westerly wind events in the tropical Pacific, 1986–95. Journal of Climate, 10, 3131–3156.

Harrison, D. E., & Vecchi, G. A. (1999). On the termination of El Niño. Geophysical Research Letters, 26, 1593–1596. https:// doi.org/10.1029/1999GL900316 Hayashi, M., & Watanabe, M. (2017). ENSO complexity induced by state dependence of westerly wind events. Journal of Climate, 30, 3401–3420. Hoerling, M. P., Kumar, A., & Zhong, M. (1997). El Niño, La Niña, and the nonlinearity of their teleconnections. Journal of Climate, 10, 1769–1786. https://doi.org/10.1175/1520‐0442 (1997)0102.0.CO;2 Hong, C.‐C., Li, T., & Chen, Y.‐C. (2010). Asymmetry of the Indian Ocean basinwide SST anomalies: Roles of ENSO & IOD. Journal of Climate, 23, 3563–3576. https://doi. org/10.1175/2010JCLI3320.1 Im, S.‐H., An, S.‐I., Kim, S. T., & Jin, F.‐F. (2015). Feedback processes responsible for El Niño–La Niña amplitude asymmetry. Geophysical Research Letters, 42, 5556–5563. https:// doi.org/10.1002/2015GL064853 Jansen, M. F., Dommenget, D., & Keenlyside, N. (2009). Tropical atmosphere‐ocean interactions in a conceptual framework. Journal of Climate, 22, 550–567. https://doi. org/10.1175/2008JCLI2243.1 Jin, F.‐F. (1997a). An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model. Journal of the Atmospheric Sciences, 54, 811–829. https://doi.org/https://doi.org/10.1175/ 1520‐0469(1997)0542.0.CO;2 Jin, F.‐F. (1997b). An equatorial ocean recharge paradigm for ENSO. Part II: A stripped‐down coupled model. Journal of the Atmospheric Sciences, 54, 830–847. https://doi.org/10.1175/1 520‐0469(1997)0542.0.CO;2 Jin, F.‐F., Neelin, J. D., & Ghil, M. (1994). El Niño on the devil’s staircase: Annual and subharmonic steps to chaos. Science, 264, 70–72. doi: 10.1126/science.264.5155.70 Jin, F.‐F., An, S. I, Timmermann, A., & Zhao, J. (2003). Strong El Niño events and nonlinear dynamical heating. Geophysical Research Letters, 30, 1–4. https://doi.org/10.1029/2002 GL016356 Jin, F.‐F., Lin, L., Timmermann, A., & Zhao, J. (2007). Ensemble‐mean dynamics of the ENSO recharge oscillator under state‐dependent stochastic forcing. Geophysical Research Letters, 34, 1–5. https://doi.org/10.1029/2006 GL027372 Kanamitsu, M., Ebisuzaki, W., Woollen, J., Yang, S.‐K., Hnilo, J. J., Fiorino, M., & Potter, G. L. (2002) NCEP‐DOE AMIP‐ II Reanalysis (R‐2). Bulletin of the American Meteorological Society. 83(11), 1631–1644. Kang, I.‐S., & Kug, J.‐S (2002). El Niño and La Niña sea surface temperature anomalies: Asymmetry characteristics associated with their wind stress anomalies. Journal of Geophysical Research: Atmospheres, 107, 1–10. https://doi.org/10.1029/ 2001JD000393 Keen, R. A. (1982). The role of cross‐equatorial tropical cyclone pairs in the Southern Oscillation. Monthly Weather Review, 110, 1405–1416. Kessler, W. S. (2002). Is ENSO a cycle or a series of events? Geophysical Research Letters, 29, 2125. doi:10.1029/ 2002GL015924 Kessler, W. S., McPhaden, M. J., & Weickmann, K. M. (1995). Forcing of intraseasonal Kelvin waves in the equatorial

170  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Pacific. Journal of Geophysical Research: Oceans, 100, 10613–10631. Kiladis, G. N., & Wheeler, M. (1995). Horizontal and vertical structure of observed tropospheric equatorial Rossby waves. Journal of Geophysical Research: Atmospheres, 100, 22981–22997. Kirtman, B. P., & Schopf, P. S. (1998). Decadal variability in ENSO predictability and prediction. Journal of Climate, 11, 2804–2822. Kohyama, T., Hartmann, D. L., & Battisti, D. S. (2018). Weakening of nonlinear ENSO under global warming. Geophysical Research Letters, 45, 8557—8567. https://doi. org/10.1029/2018GL079085 Kug, J.‐S., & Ham, Y.‐G. (2011). Are there two types of La Niña? Geophysical Research Letters, 38, L16704. doi:10.1029/2011GL048237 Kug, J.‐S., & Kang, I.‐S. (2006). Interactive feedback between ENSO and the Indian Ocean. Journal of Climate, 19, 1784– 1801. https://doi.org/10.1175/JCLI3660.1 Kug, J.‐S., Jin, F.‐F., Sooraj, K. P., & Kang, I.‐S. (2008). State‐ dependent atmospheric noise associated with ENSO. Geophysical Research Letters, 35, L05701. doi:10.1029/ 2007GL032017 Larkin, N. K., & Harrison, D. E. (2002). ENSO warm (El Niño) and cold (La Niña) event life cycles: Ocean surface anomaly patterns, their symmetries, asymmetries, and implications. Journal of Climate, 15, 1118–1140. https://doi.org/10.1175/15 20‐0442(2002)0152.0.CO;2 Lengaigne, M., Boulanger, J. P., Menkes, C., Delecluse, P., & Slingo, J. (2004). Westerly wind events in the tropical Pacific and their influence on the coupled ocean‐atmosphere system. Earth Climate: The Ocean‐Atmosphere Interaction, Geophys­ ical Monograph, 147, 49–69. Levine, A. F. Z., & Jin, F.‐F. (2010). Noise‐induced instability in the ENSO recharge oscillator. Journal of the Atmospheric Sciences, 67, 529–542. https://doi.org/10.1175/2009JAS3213.1 Levine, A., Jin, F.‐F., & McPhaden, M. J. (2016). Extreme noise—extreme El Niño: How state‐dependent noise forcing creates El Niño–La Niña asymmetry. Journal of Climate, 29, 5483–5499. Li, T. (1997a). Air–sea interactions of relevance to the ITCZ: Analysis of coupled instabilities and experiments in a hybrid coupled GCM. Journal of the Atmospheric Sciences, 54, 134–147. Li, T., (1997b). Phase transition of the El Niño‐Southern Oscillation: A stationary SST mode. Journal of the Atmospheric Sciences, 54, 2872–2887. Li, T., & Philander, S.G.H., (1996). On the annual cycle of the equatorial eastern Pacific. J. Climate, 9, 2986–2998. Li, T., Wang, B., Chang, C.‐P., & Zhang, Y. (2003). A theory for the Indian Ocean dipole‐zonal mode. J. Atmos. Sci., 60, 2119–2135. Li, T., Wang, B., Wu, B., Zhou, T., Chang, C.‐P., & Zhang, R. (2017). Theories on formation of an anomalous anticyclone in western North Pacific during El Niño: A review. Journal of Meteorological Research, 31, 987–1006. doi:10.1007/ s13351‐017‐7147‐6 Li, Y., Lu, R., & Dong, B. (2007). The ENSO‐Asian monsoon interaction in a coupled ocean‐atmosphere GCM. Journal of Climate, 20, 5164–5177.

Lian, T., Tang, Y., Zhou, L., Islam, S. U., Zhang, C., Li, X., & Ling, Z. (2018). Westerly wind bursts simulated in CAM4 and CCSM4. Climate Dynamics, 50, 1353–1371. Lloyd, J., Guilyardi, E., & Weller, H. (2012a). The role of atmosphere feedbacks during ENSO in the CMIP3 models. Part III: The shortwave flux feedback. Journal of Climate, 25, 4275–4293. Lloyd, J., Guilyardi, E., & Weller, H. (2012b). The role of atmosphere feedbacks during ENSO in the CMIP3 models. Part III: The shortwave flux feedback. Journal of Climate, 25, 4275–4293. https://doi.org/10.1175/JCLI‐D‐11‐00178.1 Lopez, H., Kirtman, B. P., Tziperman, E., & Gebbie, J. (2013). Impact of interactive westerly wind bursts on CCSM3. Dynamics of Atmospheres and Oceans, 59, 24–51. Lorenz, E. N. (1963). Deterministic nonperiodic flow. Journal of the Atmospheric Sciences, 20, 130–141. Lu, B., Jin, F.‐F., and Ren, H.‐L. (2018) A coupled dynamic index for ENSO periodicity. Journal of Climate, 31, 2361–2376. Marzeion, B., Timmermann, A., Murtugudde, R., & Jin, F.‐F. (2005). Biophysical feedbacks in the tropical Pacific. Journal of Climate, 18, 58–70. https://doi.org/10.1175/JCLI3261.1 McGregor, S., Timmermann, A., Schneider, N., Stuecker, M. F., & England, M. H. (2012). The effect of the South Pacific convergence zone on the termination of El Niño events and the meridional asymmetry of ENSO. Journal of Climate, 25, 5566–5586. http://doi.org/10.1175/JCLI‐D‐11‐00332.1 McGregor, S., Ramesh, N., Spence, P., England, M. H., McPhaden, M. J., & Santoso, A. (2013). Meridional movement of wind anomalies during ENSO events and their role in event termination. Geophysical Research Letters, 40, 749–754. https://doi.org/10.1002/grl.50136 McPhaden, M. J., Bahr, F., Du Penhoat, Y., Firing, E., Hayes, S. P., Niiler, P. P., et  al. (1992). The response of the western equatorial Pacific Ocean to westerly wind bursts during November 1989 to January 1990. Journal of Geophysical Research: Oceans, 97, 14289–14303. McPhaden, M. J., & Zhang, X. (2009). Asymmetry in zonal phase propagation of ENSO sea surface temperature anomalies. Geophysical Research Letters, 36, L13703. https://doi. org/10.1029/2009GL038774 Moore, A. M., & Kleeman, R. (1999). Stochastic forcing of ENSO by the intraseasonal oscillation. Journal of Climate, 12, 1199–1220. Nitta, T. (1989). Development of a twin cyclone and westerly bursts during the initial phase of the 1986‐87 El Niño. Journal of the Meteorological Society of Japan. Ser. II, 67, 677–681. Nitta, T., & Motoki, T. (1987). Abrupt enhancement of convective activity and low‐level westerly burst during the onset phase of the 1986‐87 El Niño. Journal of the Meteorological Society of Japan. Ser. II, 65, 497–506. Ohba, M., & Ueda, H. (2007). An impact of SST anomalies in the Indian Ocean in acceleration of the El Niño to La Niña transition. Journal of the Meteorological Society of Japan, 85, 335–348. https://doi.org/10.2151/jmsj.85.335 Ohba, M., & Ueda, H. (2009). Role of nonlinear atmospheric response to SST on the asymmetric transition process of ENSO. Journal of Climate, 22, 177–192. https:/doi. org/10.1175/2008JCLI2334.1

ENSO Irregularity and Asymmetry  171 Ohba, M., & Watanabe, M. (2012). Role of the Indo‐Pacific interbasin coupling in predicting asymmetric ENSO transition and duration. Journal of Climate, 25, 3321–3335. https://doi.org/10.1175/JCLI‐D‐11‐00409.1 Ohba, M., Nohara, D., & Ueda, H. (2010). Simulation of asymmetric ENSO transition in the WCRP CMIP3 multimodel experiments. Journal of Climate, 23, 6051–6067. https://doi. org/10.1175/2010JCLI3608.1 Okumura, Y. M., & Deser, C. (2010). Asymmetry in the duration of El Niño and La Niña. Journal of Climate, 23, 5826– 5843. https://doi.org/10.1175/2010JCLI3592.1 Okumura, Y. M., Ohba, M., Deser, C., & Ueda, H. (2011). A proposed mechanism for the asymmetric duration of El Niño and La Niña. Journal of Climate, 24, 3822–3829. https://doi. org/10.1175/2011JCLI3999.1 Okumura, Y. M., Sun, T., & Wu, X. (2017). Asymmetric modulation of El Niño and La Niña and the linkage to tropical Pacific decadal variability. Journal of Climate, 30, 4705–4733. https://doi.org/10.1175/JCLI‐D‐16‐0680.1 Ott, E. (2002). Chaos in dynamical systems. Cambridge University Press. Penland, C., & Sardeshmukh, P. D. (1995). The optimal growth of tropical sea surface temperature anomalies. Journal of Climate, 8, 1999–2024. Perez, C. L., Moore, A. M., Zavala‐Garay, J., & Kleeman, R. (2005). A comparison of the influence of additive and multiplicative stochastic forcing on a coupled model of ENSO. Journal of Climate, 18, 5066–5085. Puy, M., Vialard, J., Lengaigne, M., & Guilyardi, E. (2016). Modulation of equatorial Pacific westerly/easterly wind events by the Madden–Julian oscillation and convectively‐ coupled Rossby waves. Climate Dynamics, 46, 2155–2178. Reynolds, R. W., Rayner, N. A., Smith, T. M., Stokes, D. C., & Wang, W. (2002). An improved in situ and satellite SST analysis for climate. Journal of Climate, 15, 1609–1625. Rong, X., Zhang, R., Li, T., & Su, J., (2011). Upscale feedback of high‐frequency winds to ENSO. Q. J. R. Meteorol. Soc., 137, 894–907. Roulston, M. S., & Neelin, J. D. (2000). The response of an ENSO model to climate noise, weather noise and intraseasonal forcing. Geophysical Research Letters, 27, 3723–3726. Samelson, R. M., & Tziperman, E. (2001). Instability of the chaotic ENSO: The growth‐phase predictability barrier. Journal of the Atmospheric Sciences, 58, 3613–3625. Schott, F. A., Xie, S.‐P., & McCreary, J. P. (2009). Indian Ocean circulation and climate variability. Reviews of Geophysics, 47, RG1002. https://doi.org/10.1029/2007RG000245 Seiki, A., & Takayabu, Y. N. (2007a). Westerly wind bursts and their relationship with intraseasonal variations and ENSO. Part II: Energetics over the western and central Pacific. Monthly Weather Review, 135, 3346–3361. Seiki, A., & Takayabu, Y. N. (2007b). Westerly wind bursts and their relationship with intraseasonal variations and ENSO. Part I: Statistics. Monthly Weather Review, 135, 3325–3345. Slingo, J. M., Rowell, D. P., Sperber, K. R., & Nortley, F. (1999) On the predictability of the interannual behavior of the Madden–Julian oscillation and its relationship with El Niño. Quart. J. Roy. Meteor. Soc., 125, 583–609.

Stein, K., Schneider, N., Timmermann, A., & Jin, F.‐F. (2010). Seasonal synchronization of ENSO events in a linear stochastic model. Journal of Climate, 23, 5629–5643. Stein, K., Timmermann, A., & Schneider, N. (2011). Phase synchronization of the El Niño‐Southern oscillation with the annual cycle. Physical Review Letters, 107, 128501. Stone, L., Saparin, P. I., Huppert, A., & Price, C. (1998). El Niño chaos: The role of noise and stochastic resonance on the ENSO cycle. Geophysical Research Letters, 25, 175–178. Strogatz, S. (1994). Nonlinear dynamics and chaos. Westview Press. Stuecker, M. F., Timmermann, A., Jin, F. F., McGregor, S. & Ren, H. L. (2013). A combination mode of the annual cycle and the El Niño/Southern Oscillation. Nat. Geosci., 6, 540–544. Su, J., Zhang, R., Li, T., Rong, X., Kug, J.‐S, & Hong, C. C. (2010). Causes of the El Niño & La Niña amplitude asymmetry in the equatorial Eastern Pacific. Journal of Climate, 23, 605–617. https://doi.org/10.1175/2009JCLI2894.1 Suarez, M. J., & Schopf, P. S. (1988). A delayed action oscillator for ENSO. Journal of the Atmospheric Sciences, 45, 3283–3287. Sun, Y., Sun, D. Z., Wu, L., & Wang, F. (2013). Western pacific warm pool and ENSO asymmetry in CMIP3 models. Advances in Atmospheric Sciences, 30, 940–953. https://doi. org/10.1007/s00376‐012‐2161‐1 Sun, Y., Wang, F., & Sun, D. Z. (2016). Weak ENSO asymmetry due to weak nonlinear air–sea interaction in CMIP5 climate models. Advances in Atmospheric Sciences, 33, 352–364. https://doi.org/10.1007/s00376‐015‐5018‐6 Takahashi, K., & Dewitte, B. (2016). Strong and moderate nonlinear El Nino regimes. Climate Dynamics, 46, 1627–1645. Takahashi, K., Karamperidou, C., & Dewitte, B. (2019) A theoretical model of strong and moderate El Nino regimes. Climate Dynamics, 52, 7477–7493. Takahashi, K., Montecinos, A., Goubanova, K., Dewitte, B. (2011). ENSO regimes: Reinterpreting the canonical and Modoki El Niño. Geophysical Research Letters, 38, L10704, https://doi.org/10.1029/2011GL047364 Timmermann, A., & Jin, F.‐F. (2002). Phytoplankton influences on tropical climate. Geophysical Research Letters, 29, 19‐1– 19‐4. https://doi.org/10.1029/2002GL015434 Timmermann, A., An, S.‐I., Kug, J.‐S., Jin, F.‐F., Cai, W., Capotondi, A., et  al. (2018). El Niño–Southern Oscillation complexity. Nature, 559, 535–545. https://doi.org/10.1038/ s41586‐018‐0252‐6 Tollefson, J. (2014). El Niño tests forecasters. Nature, 508, 20–21. Tziperman, E., & Yu, L. (2007). Quantifying the dependence of westerly wind bursts on the large‐scale tropical Pacific SST. Journal of Climate, 20, 2760–2768. Tziperman, E., Stone, L., Cane, M. A., & Jarosh, H. (1994). El Niño chaos: Overlapping of resonances between the seasonal cycle and the Pacific ocean‐atmosphere oscillator. Science, 264, 72–74. Tziperman, E., Cane, M. A., & Zebiak, S. E. (1995). Irregularity and locking to the seasonal cycle in an ENSO prediction model as explained by the quasi‐periodicity route to chaos. Journal of the Atmospheric Sciences, 52, 293–306.

172  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Tziperman, E., Zebiak, S. E., & Cane, M. A. (1997). Mechanisms of seasonal–ENSO interaction. Journal of the Atmospheric Sciences, 54, 61–71. Vecchi, G. A. (2006). The termination of the 1997/98 El Niño. Part II: Mechanisms of atmospheric change. Journal of Climate, 19, 2647– 2664. https://doi.org/10.1175/JCLI3780.1 Verbickas, S. (1998). Westerly wind bursts in the tropical Pacific. Weather, 53, 282–284. Vialard, J., Menkes, C., Boulanger, J.‐P., Delecluse, P., Guilyardi, E., McPhaden, M. J., & Madec, G. (2001). A model study of oceanic mechanisms affecting equatorial Pacific sea surface emperature during the 1997–98 El Niño. Journal of Physical Oceanography, 31, 1649–1675. https://doi.org/10.1175/1520‐0 485(2001)031 37 (37)

30 25

15

< 20 (3)

25

> 17 (13)

20 ...but the models that Strengthen, tend to do so more significantly.

15

5 26

10

20C.obs 20C.hist 21C.rcp85 ctrl.pi

10

1 24

More models weaken than

30 strengthen...

28

30

32

5

obs

20

Significant warming in the 21st century.

Most models are too cold.

obs

models sorted by 20th century values

mmm

1 34

SST (°C)

0

0.5

1

1.5

2

SSTA s.d. (°C)

Figure 9.4 Values simulated by the CMIP5 models for (a) climatological mean SST (°C) averaged over the Niño‐3.4 region (170°W–120°W, 5°S–5°N), and (b) standard deviation of SSTAs (°C) averaged over the Niño‐3.4 region. Simulations correspond to 20th‐century historical forcings (1900–1999, orange circles), the 21st‐century RCP8.5 scenario (2000–2099, blue triangles), and the central 95% range for 100‐year running statistics from the preindustrial control run (horizontal gray bars). Black vertical line indicates the observed value for 1900–1999 computed from the HadISST.v1.1 reconstruction. The multimodel mean (MMM) values for the orange and blue symbols are indicated with similarly colored vertical bars along the top abscissa. 21st‐century values that exceed the 20th century are shown with filled blue triangles, and their count is indicated by the > label (with the number of significant changes relative to the gray bars indicated in parentheses). Twenty‐first‐century values that fall below the 20th century are shown with hollow blue triangles, and their count is indicated by the < label (with the number of significant changes in parentheses). Brown vertical bars on the bottom abscissa indicate the (2.5, 50, 97.5) percentiles from a 4000‐year simulation of a statistical model fit to the HadISST.v1.1 observation. The models are sorted in the vertical according to the 20th‐century values (orange circles). Adapted from Figure 9 of C. Chen et al. (2017).

and the surface mixed layer and SST simulated by the ocean. The AMIP and OMIP frameworks remain very helpful tools, however, for isolating biases identified in coupled models. Another strategy is to only partially enable the cou­ pling, by nudging or overriding one of the variables involved in the exchange. For example, the simulated SST can be nudged (restored) with a short time scale (e.g. 5 days) toward observed time‐varying SSTs, to test the performance of the atmosphere and ocean simulations in the presence of both high‐frequency (diurnal) cou­ pling and close‐to‐observed SST variability (Kamenkovich & Sarachik, 2004; Zhu & Kumar, 2018; Vecchi et  al., 2019). Ocean biases that develop in the nudged context but not in the OMIP context can indi­ cate a problem, for example, with the simulation of the wind stress by the atmosphere component. The nudging term also provides a valuable diagnostic, indicating the degree of correction required to keep the SST close to those observed in different locations and dynamical con­ ditions. This approach can also be applied regionally, to further isolate the sources of model bias (Large & Danabasoglu, 2006; Small et  al., 2015; Song & Zhang, 2016; McGregor et al., 2018).

Another valuable technique is to compute a clima­ tology of the nudging term and wind stress biases from the nudged run described above. A new coupled run can then be performed in which the nudging is turned off, but now the seasonally varying (but interannually constant) nudging climatology is prescribed. This technique and its variants, known as “flux adjustment” or “flux correction,” help to maintain the climatology of the coupled model close to observations, without restricting the development of anomalies relative to that climatology. This framework can help diagnose how biases in simulated ENSO dynamics are related to a model’s climatological biases (Magnusson et  al., 2013a; Ray et  al., 2018b). Several studies have shown that flux adjustment can improve a model’s simulated tropical Pacific climatology, seasonal cycle, ENSO, and seasonal forecast skill (Manganello & Huang, 2009; Kröger & Kucharski, 2011; Magnusson et al., 2013a, 2013b), although results especially for fore­ cast skill appear to be model dependent (Spencer et al., 2007; Pan et al., 2011). The ability of a model to reconstruct and forecast either real‐world ENSO events or its own events can be used to assess the degree of ENSO’s long‐term memory and predictability (Wittenberg et al., 2014; Karamperidou

212  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

et  al., 2014; Ding et  al., 2018, 2019). Ding et  al. (2018) found that some models are now realistic enough that their unforced control runs contain close “model‐ana­ logs” of real‐world conditions, and the evolution of those analog trajectories can actually provide useful ensemble forecasts of the real world, with skill approaching that of state‐of‐the‐art assimilation and forecast systems. The ability of a model’s analogs to reconstruct and forecast real‐world ENSO conditions is thus a potentially pow­ erful diagnostic of model performance, assessing both the shape of the model’s attractor and the evolution of its trajectories on that attractor. Data assimilation can assist in diagnosing model biases, through the analysis of the increments needed to adjust the model solution toward observations. A systematic increment structure can help to localize persistent dynam­ ical biases in the model. The performance of model fore­ casts of the real world, initialized via data assimilation, can also provide valuable information about model biases. If the assimilated state lies far from the model attractor, this can induce an “initialization shock,” which rings through the model’s climate system and reduces forecast skill. Once the model is released from data assim­ ilation and run in forecast mode, the location and time scale of bias development can help to isolate biases in particular processes (Vannière et al., 2014). For example, a bias that appears within a single day would have to involve local biases in clouds, surface heat fluxes, or mixed layer dynamics but would rule out remotely forced internal waves in the ocean. Using this approach, it has been shown that a single bias (e.g. the ECT error described below) can have model‐dependent sources (Vannière et al., 2013). 9.4.2. Key Model Biases A number of common ENSO biases have been identi­ fied in CGCM simulations, some persisting since earlier model intercomparisons (Guilyardi, 2006, Guilyardi et  al., 2009, 2012a, 2012b; Bellenger et  al., 2014, Capotondi et  al., 2015a; Santoso et  al., 2019). Here we discuss those biases and how they are related to biases in the simulated background climatology. 9.4.2.1. Biases in the Background Climatology Common biases in the CGCM‐simulated background climatology include 1. An overly strong ECT that extends too far west, associated with a cold SST bias in the equatorial Pacific. 2. A warm SST bias near the coast of South America. 3. An equatorial Pacific dry bias, associated with the excessive ECT. 4. An excessive “double” ITCZ south of the equator in the east Pacific (J. Lin, 2007).

5. A southern Pacific convergence zone (SPCZ) that is too zonally oriented. 6. An overly intense hydrologic cycle over the tropical Pacific, with excessive evaporation and rainfall (de Szoeke & Xie, 2008; Wittenberg et al., 2018). 7. Biases in the cloud regimes over the eastern and central Pacific (Lloyd et  al., 2009, Sun et  al., 2009, Bellenger et al., 2014). 8. Equatorial τx that is too strong or too weak, which then affects equatorial upwelling and the zonal tilt of the equatorial thermocline. 9. Overly cyclonic wind stress off‐equator, associated with a too‐rapid poleward intensification of τx. This can lead to excessive Ekman suction and poleward Sverdrup transport off‐equator, shoaling the equatorial thermo­ cline and contributing to an ECT cold bias (Wittenberg et al., 2018). 10. Biases in the equatorial thermocline depth, inten­ sity, sharpness, and zonal slope. These biases can be masked by other deficiencies in the model via error compensation (Guilyardi et al., 2004, 2009; Wittenberg et al., 2018; Ray et al., 2018a, 2018b; Vijayeta & Dommenget, 2018). For example, many processes can influence the structure and maintenance of the ECT, including vertical mixing in the upper ocean off‐equator (Anderson et al., 2009), subtropical cloud albedos (Burls et  al., 2014), excessive equatorial zonal wind stress (Vannière et al., 2013), temperature biases subducted in the subtropics (Thomas & Fedorov, 2017), errors in the sub­ tropical wind stress that controls the strength and structure of the shallow meridional overturning subtropical‐tropical cells and equatorial upwelling (McPhaden & Zhang, 2002; Capotondi et  al., 2005), and SST biases in the tropical Indian and Atlantic oceans (Kajtar et al., 2017). This diver­ sity of possible mechanisms is a key reason that it remains a challenge to correctly simulate the ECT in GCMs. Climatological biases can affect ENSO feedbacks and sensitivities by displacing climatological features, and their associated ENSO variability, away from their observed locations. Background biases also affect the intensities and spatiotemporal phases of the leading terms in the mixed layer heat budget, namely, the heat flux damping, thermocline feedback, zonal advective feedback, and Ekman feedback, which then induce biases in ENSO properties (An & Wang 2000; Wittenberg 2002; Kim & Jin, 2011; Graham et al., 2017). 9.4.2.2. Biases in ENSO As described in recent reviews (Guilyardi, 2006; Guilyardi et  al., 2009, 2012a, 2012b; Bellenger et  al., 2014; Capotondi et  al., 2015a; Santoso et  al., 2019), common ENSO biases in CGCMs include 1. Amplitude errors, which can also affect the skew­ ness, diversity, and interdecadal modulation of ENSO

HISTORY AND PROGRESS OF ENSO MODELING  213

and the ability of ENSO to affect the multidecadal‐mean climate. 2. Errors in spectrum, including the dominant ENSO period and irregularity. In many models the ENSO spec­ trum is too sharply peaked, and the ENSO period is too regular and biennial. 3. Too little synchronization of ENSO to the annual cycle, or a synchronization of ENSO to the wrong season. This can result from problems simulating the climatolog­ ical seasonal cycle of the ITCZs (Wittenberg et al., 2006; Abellán et al., 2017), the SST‐cloud and thermocline‐SST feedbacks of ENSO (Rashid & Hirst, 2016), and the sea­ sonally mediated impacts of ENSO on remote regions (Lee et al., 2016, 2018; W. Zhang et al., 2016). 4. Errors in the level of interdecadal modulation of ENSO behavior. 5. SSTA patterns that are displaced too far west, dis­ connected from the South American coast, too symmetric about the equator, and show too little interevent diversity in peak longitude (Capotondi et al., 2015a; chapter 4, this volume). 6. Atmospheric response patterns of rainfall and tele­ connections that are displaced too far west. 7. Too little skewness of ECT SSTAs toward warm events (An et al., 2005; Dommenget et al., 2013; T. Zhang & Sun 2014; C. Chen et al., 2017), and too little skewness of central equatorial Pacific τx toward westerly anomalies (K.‐Y. Choi et al., 2013). 8. Equatorial τx anomalies that are too weak, too far west, and too narrow in the meridional direction (Guilyardi, 2006). This can affect ENSO amplitude via a reduced zonal wind feedback (Figure 9.1) and can accel­ erate the oceanic adjustment to wind anomalies and shorten the ENSO period (Capotondi et al., 2006) 9. Too little damping of SSTAs by surface heat fluxes, often due to a weak cloud shading response associated with biases in cloud regimes, and other mean state biases (Lloyd et al., 2012, Bellenger et al., 2014, Dommenget & Yu, 2016). 10. Insufficient cross‐timescale linkage between ENSO, its intraseasonal precursors, and Pacific decadal modes (Di Lorenzo et al., 2015; Newman et al., 2016; Wang & Miao, 2018; Liguori & Di Lorenzo, 2018; R. Lin et al., 2018). These errors appear to be linked in part to the magnitude of the climatological ECT cold SST bias (Lyu et al., 2015). Many of these errors in ENSO simulations can be linked to biases in the background climatology. Atmospheric convection and rainfall patterns over the tropical Pacific are sensitive to the relative temperature difference between the local SST and the tropical‐mean SST, especially over the warm pool (He et  al., 2018). Thus, models with excessively cold ECTs and a westward‐ displaced warm pool tend to have westward‐displaced

ENSO patterns (e.g. Wittenberg et al., 2006; Ham & Kug, 2015). This is often associated with a modified distribu­ tion of convective and subsidence regimes, leading to deficient atmosphere feedbacks in the east and central Pacific (Lloyd et  al., 2012, Bellenger et  al., 2014) and overamplified or overdamped ENSO events. Such biases in the background state can even lead to unrealistic dou­ ble‐peaked El Niños (Graham et  al., 2017), due to the zonal‐advective feedbacks at the warm pool’s eastern edge being displaced too far west of the thermocline feed­ backs occurring further east. Westward displacement of ENSO’s SSTAs can also weaken the interevent diversity of SSTA patterns, as it becomes more difficult to generate SSTAs in the eastern ECT that induce a τx response with sufficient zonal fetch to produce strong thermocline feed­ backs (Ham & Kug, 2012; Kug et  al., 2012). Improper balances between zonal and thermocline feedbacks can also affect the zonal propagation direction of SSTAs (Ham & Kug, 2015; C. Chen et al., 2017). A model with a poleward‐displaced ITCZ and SPCZ can also show reduced nonlinearity of the equatorial τx response to SSTAs, which then affects ENSO’s warm‐cold asymme­ tries of amplitude, duration, and transition (K.‐Y. Choi et al., 2013, 2015). As described above for biases in the background state, biases in ENSO can mask each other. For example, a model with both a weak equatorial τx response (which tends to weaken ENSO) and weak surface heat flux damp­ ing (which tends to strengthen ENSO) can exhibit a rea­ sonable ENSO amplitude for the wrong reasons (Guilyardi et  al., 2009; Vijayeta & Dommenget, 2018). Similarly, a model with a climatological cold SST bias in the ECT may still be able to produce realistic atmospheric responses during the El Niño phase, if it also has excessive warm SSTAs to flatten the zonal and meridional SST gradients across the tropical Pacific, thereby favoring atmospheric convection in the central and eastern equatorial Pacific. Errors in ENSO amplitude can also affect the multi­ decadal‐mean climate, via nonlinearity and temporal blur­ ring of the seasonal‐to‐interannual motions of features like the ITCZs, warm pool edge, and thermocline (Watanabe & Wittenberg, 2012; Watanabe et  al., 2012; Ogata et al., 2013; J. Choi et al., 2013; Atwood et al., 2017). 9.4.2.3. ENSO Response to External Forcings As described in chapter 13, GCMs produce diverse and nonmonotonic projected responses of ENSO to future changes in radiative forcings (Vecchi & Wittenberg, 2010; Collins et  al., 2010; C. Chen et  al., 2017, Rashid et  al., 2016). Depending on the model, the future amplitude of ENSO SSTAs can increase, decrease, or show no significant change, with no clear link to the magnitude of the time‐mean ECT SST bias in the historical simulation (C. Chen et al., 2017).

214  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Uncertainties in ENSO’s future SSTA amplitude arise from competing changes in ENSO’s intense air‐sea feed­ backs, which amplify the sensitivity of ENSO (and its future response) to biases in the models (Collins et  al., 2010; DiNezio et al., 2012). Studies have shown that the sensitivity of the background state and ENSO SSTAs to future anthropogenic forcings depends strongly on the present‐day simulated spatial structure of key climate regimes, including the regions of intense convection, cloudiness, and evaporation (Xie et al., 2010), the inten­ sity of the ECT (Huang & Ying, 2015), the strength of convective cloud shading response to SSTAs (Ying & Huang, 2016a), the intensity of equatorial Pacific upwelling (Ying & Huang, 2016b), and the future warming of the tropical Indian and Atlantic oceans relative to the tropical Pacific (Luo et al., 2012; Wieners et al., 2017). Model biases in these climatological features can therefore lead to biases in the sensitivity of ENSO to future change, reducing confidence in future projections. There is also substantial intrinsic modulation of ENSO that is unrelated to external forcings (Wittenberg, 2009, 2015; Stevenson et  al., 2010), which can mask forcing‐ induced changes in short observational records. For example, Newman et al. (2018) examined 30‐member his­ torical ensembles from the NCAR‐CESM‐LE and GFDL‐FLOR‐FA CGCMs, and found that while both ensembles simulated a slight trend toward stronger ENSO SSTAs over the past century, those trends were of the same magnitude as the intrinsic modulation of ENSO among realizations with the length of the observational record. This suggests that the forced component of change, even if it were as large in reality as in the models, would be difficult to detect in the single century–long realization yet available from the historical observations. Consistent with this, Newman et al. (2018) found that the observed trend in ENSO SSTA amplitude (the strength of which differed depending on the historical SST recon­ struction used) did not exceed the range expected for a null hypothesis of stationary ENSO dynamics, estimated using a stochastically forced linear inverse model tuned to the historical observations. It has also been suggested that models may systemati­ cally underestimate intrinsic decadal variations in the tropical Pacific (Kociuba & Power, 2015), for instance, due to reduced SSTA persistence and/or reduced inter­ basin connections associated with model SST biases in the tropical Atlantic (McGregor et  al., 2018). Reduced decadal variability might then affect the models’ ability to capture a realistic range of intrinsic ENSO modulation. Despite the uncertainties mentioned above, there are aspects of future ENSO change that are relatively robust among the model projections (chapter 13), including a tendency for ENSO rainfall variations over the tropical Pacific to increase and shift eastward and equatorward

as the tropical atmosphere moistens, especially during El Niño (Power et al., 2013, 2017; Cai et al., 2014; Huang & Chen, 2017), and a tendency for more eastward pro­ pagation of ENSO equatorial SSTAs in the future (C.  Chen et  al., 2017). However, intermodel consensus does not necessarily mean that the projections are correct, as many models have similar climate biases that could affect their sensitivities to future change. In particular, there is concern that the inability of most models to capture the recent observed decadal strengthening of the Walker circulation (Power et al., 2017) may indicate that they either underestimate the intrinsic decadal variability or overestimate the externally forced warming of the eastern equatorial Pacific, thereby reducing the east‐west SST gradient. 9.4.3. Emergent Constraints for Future Changes in ENSO A potentially powerful approach to deal with the above uncertainties is to examine how model projections depend on model biases. If robust relationships can be found, then it may be possible to extrapolate the expected sensi­ tivities of the real world from the diverse model results, providing “emergent constraints” for the real‐world response (e.g. Heinze et al., 2019). An example is shown in Figure 9.5 adapted from Ham & Kug (2015), which indicates that among the CMIP5 models there is a relationship between the ECT intensity and the longitudinal location of the ENSO anomaly pat­ terns. Models with weaker/better ECTs tend to show an ENSO response closer to observed, with patterns of rain­ fall, wind stress, and SSTAs that shift farther eastward and equatorward during El Niño. By extending such relation­ ships to include the observations, estimates can be made regarding the real‐world sensitivities, essentially leveraging the intermodel diversity to provide emergent constraints for the sensitivities based on a large set of models. A second example is that among the CMIP3 models that produced a reasonably realistic ENSO, models with meridionally broader τx responses during ENSO also tended to exhibit more eastward SSTA propagation and a stronger projected amplification of ENSO SSTAs in the future (Merryfield, 2006). Nearly all models underesti­ mate the meridional width of the τx response (Capotondi et al., 2006), which suggests that the future amplification of ENSO in the real world might be even stronger than suggested by the models (Vecchi & Wittenberg, 2010). A third example of an emergent constraint relates to future changes in the climatological equatorial Pacific zonal and meridional SST gradients, which as described above are key controls on the atmospheric response to ECT SSTAs during ENSO events. Huang & Ying (2015) found that CMIP5 models with weaker and more realistic

HISTORY AND PROGRESS OF ENSO MODELING  215 (a)

Rainfall response to ENSO

(d)

20N 10N

–1

10S

10N

–11

2

EQ

EQ

0

10S

1

20S 90E

(b)

120E

150E

180

150W

120W

90W

850 mb zonal wind response to ENSO

20N

2

10S

–0.5

90E

120E

(c)

150E

180

1.5

0.5

150W

120W

20S

90W

0.6

EQ

90W

150W 0.2

0.3

90E

120E

150E

180

150W

120W

90W

120W

90W

Correlation with mean SST

10N

0.0 1.2

EQ 10S

20S

0.1

120W

20N

0.6

10S

–0.6 –0.5 –0.4 –0.3 –0.2 –0.1

150W

Correlation with mean 850 mb zonal wind

(f)

10N

180

180

10S

0.3

150E

150E

EQ 1

SST response to ENSO

120E

120E

10N

20N

90E

90E

20N

0.5 1.5

EQ

20S

(e) 0.5

10N

20S

Correlation with mean rainfall

20N

120W 0.4

0.5

20S

90W 0.6

>

90E

120E

150E

–0.7 –0.6 –0.5 –0.4 –0.3 –0.2

180 0.2

150W 0.3

0.4

0.5

0.6

0.7

>

Figure 9.5  Links between ENSO anomaly patterns and the background climatology among the CMIP5 models. For each model, various anomaly fields are regressed onto SSTAs averaged over the Niño‐3.4 region (170°W–120°W, 5°S–5°N). Contours in the left column indicate these multimodel mean anomaly regressions (i.e. ENSO responses) for (a) rainfall (mm day−1 K−1), (b) 850 hPa zonal wind (m s−1 K−1), and (c) SST (K K−1). The leading intermodel principal component of ENSO precipitation anomalies, P1, is then used to stratify the models. Shading in the left column shows the intermodel regressions of ENSO anomaly patterns onto P1 for each variable. Right column shows intermodel correlations of climatological mean fields with P1 for (d) rainfall, (e) 850 hPa zonal wind, and (f) SST. Thus, a CMIP5 model with a more eastward‐shifted equatorial Pacific rainfall response during El Niño (east‐west dipole pattern in (a)) also tends to have a more eastward‐shifted (b) zonal wind response and (c) SST response, along with (d) more eastward‐ and equatorward‐shifted tropical Pacific mean rainfall, (e) weaker mean easterly trade winds along the equator, and (f) warmer SST in the equatorial Pacific cold tongue. Adapted from Figures 2, 4, and 5 of Ham and Kug (2015).

climatological ECTs also tended to project more future weakening of the climatological SST gradients, suggest­ ing that the future climatology might be more El Niño– like than most models suggest, and more conducive to enhanced eastward and equatorward shifts of atmo­ spheric convection during future El Niño events. Ying & Huang (2016a, 2016b) found that this relationship stemmed from both atmospheric and oceanic sources: (i)  models with stronger and more realistic damping of SSTAs (due to cloud shading) tended to inhibit future SST warming in the western equatorial Pacific; and (ii)  models with weaker and more realistic upwelling in the eastern equatorial Pacific tended to enhance SST warming in the east, as this reduced the inhibition of warming associated with anthropogenically enhanced thermal stratification of the tropical Pacific upper ocean. Further candidates for emergent constraints appear in the CMIP5 analysis of C. Chen et al. (2017), who found robust relationships between the El Niño/La Niña SSTA

asymmetries in historical simulations and the projected future changes in those asymmetries. It remains an open question whether emergent con­ straints diagnosed from a limited set of biased climate models offer a reliable way to infer future sensitivities of the real world. Key questions are whether the existing models are a sufficiently diverse, independent, and repre­ sentative sample of the relationships between biases and sensitivities, or whether deficiencies that are common to all of the models (e.g. limited resolution in the ocean and atmosphere) may reduce the utility of these emergent constraints. 9.4.4. Prospects for Improving ENSO Simulations ENSO simulations are affected by an inability of CGCMs to fully capture several important processes involving small scales that are difficult to represent ade­ quately with present atmosphere/ocean resolutions.

216  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

WWBs. As described in more detail in this chapter and others, westerly wind bursts (WWBs, also known as west­ erly wind events or WWEs) are a key stochastic forcing for ENSO. At the onset of El Niño, WWBs tend to strengthen and expand eastward as the warm pool expands eastward, thus representing a broadband multiplicative stochastic forcing for ENSO (Vecchi et al., 2006; Gebbie et  al., 2007; Zavala‐Garay et  al., 2008; Levine et al., 2016; Puy et al., 2016; Levine & Jin, 2017; Thual et al., 2016; Capotondi et al., 2018). WWBs con­ tribute to ENSO’s seasonality, its asymmetry between El Niño and La Niña, and its diversity from event to event (Lengaigne et  al., 2004; Levine et  al., 2016; Hayashi & Watanabe, 2017). Unfortunately, many CGCMs have dif­ ficulty reproducing the observed properties of the WWBs as well as the MJO (Ahn et al., 2017; Feng & Lian, 2018), which provides the large‐scale context for WWBs to occur (Puy et al., 2017). Barrier layers. In the warm pool region, heavy rain can cap the sea surface with a shallow lens of fresh water, leading to increased density stratification and formation of a “barrier layer” beneath the mixed layer that inhibits deep mixing. As momentum from the wind stress is deposited into a thinner surface layer, the barrier layer can enhance the response of the surface currents during WWBs, leading to greater eastward acceleration of the surface currents and stronger zonal advective–induced warming of the central Pacific at the onset of El Niño (Maes et  al., 2005; Maes & Belamari, 2011; Zhu et  al., 2014). Unfortunately, most AGCMs lack sufficient horizontal resolution to simulate the strong WWBs and strong rain events, and most OGCMs lack the upper‐ ocean vertical resolution to properly represent barrier layers and their associated strong surface currents. Thus, CGCMs might not fully capture the impacts of barrier layers on the WWB‐induced multiplicative stochastic forcing of ENSO. TIWs. In the real world, vigorous tropical instability waves (TIWs) stir water across the northern and southern flanks of the ECT. This stirs warm water equatorward at the surface, enhances ocean heat uptake from the atmosphere off‐equator, and induces strong transient shears that enhance mixing at the base of the mixed layer within the ECT (Jochum et  al., 2005; Jochum & Murtugudde, 2006; Menkes et  al., 2006; Holmes et  al., 2014; Holmes & Thomas, 2015). Together, these effects can act to thermally stratify the ECT and reduce vertical mixing near the surface (Ray et al., 2018a, 2018b), poten­ tially affecting ENSO indirectly by altering the Ekman and thermocline feedbacks. TIWs can also influence ENSO more directly. TIWs affect ENSO asymmetries, since they are more active during La Niña than El Niño (Nagura et al., 2008; Imada & Kimoto, 2012; R.‐H. Zhang, 2016), and TIWs also cause fluctuations in surface wind

stress over the ECT, which may constitute an additional stochastic forcing for ENSO (Jochum et al., 2007). Most CGCMs fail to fully represent TIWs and their effects on the ECT heat budget, due to coarse OGCM resolution and poor simulation of wind stress curls and associated upper‐ocean zonal jets for the tropical Pacific, which weaken the subsurface shears and TIW generation (Marchesiello et  al., 2011; Graham, 2014; Wittenberg et al., 2018). Numerous studies have demonstrated improved simu­ lations of the tropical Pacific climatology and ENSO as a result of improved model resolution in the atmosphere and ocean components (Roberts et al., 2009). Wittenberg et  al. (2018) found that refining the atmospheric horizontal grid from 200 km to 25 km improved the simulated tropical Pacific climatological upper‐ocean currents and temperatures, due to reduced biases in simulated rainfall and wind stress cyclonicity off‐equator. This greatly improved the simulation of ENSO and its impacts (Delworth et  al., 2012; Vecchi et  al., 2013; Jia et al., 2015; Krishnamurthy et al., 2015, 2016; W. Zhang et  al., 2016; Murakami et  al., 2015; Yang et  al., 2015). Refining the oceanic horizontal grid from 100 km to 10 km also leads to greatly improved simulation of the Pacific TIWs and their equatorward heat transport (Marchesiello et al., 2011; Griffies et al., 2015). There is also hope that refined vertical grids could help, e.g. by improving the atmospheric representation of tropical Pacific boundary layer moisture and cloudiness (espe­ cially near the coast of South America), and by improving the oceanic representation of upper‐ocean barrier layers, shears, and vertical mixing. Sufficient temporal resolu­ tion is important as well: studies have shown that resolving the diurnal cycle of solar radiation is important for the simulated time‐mean ocean mixed layer and sur­ face fluxes (Stockdale et  al., 1998; Bernie et  al., 2008; Weihs & Bourassa, 2014). As enhanced atmospheric resolutions gradually reduce the need for convective parameterization, many model biases (e.g. the double ITCZ) that are sensitive to those parameterizations may diminish. At present though, the resources required for seasonal forecasts and centennial projections currently limit atmospheric grids to about 50  km, well short of the 2–3 km needed to explicitly resolve deep convection. Atmospheric and coupled model simulations of the tropical Pacific continue to show strong sensitivity to parameterizations of atmospheric convection and clouds. This frequently leads to compromises during model development, since the combination of parameters that produces a realistic climatology or realistic ENSO may be very different from the combination that yields realistic MJO variability and tropical cyclone statistics. A key aspect of atmospheric convection that has been shown to affect ENSO simulations is the representation

HISTORY AND PROGRESS OF ENSO MODELING  217

of convective momentum transport (CMT), namely, the vertical transport of horizontal momentum that results from subgrid scale vertical motions advecting on the strong vertical shears associated with the Pacific Walker circulation. During El Niño, an eastward and equator­ ward shift of tropical Pacific deep convection leads to increased CMT over the equatorial central Pacific, bring­ ing upper‐level westerly momentum down toward the surface boundary layer where it can amplify and meridi­ onally broaden the existing westerly wind anomalies. This amplifies the Bjerknes feedback and slows the poleward discharge of equatorial ocean heat content, which in turn amplifies ENSO and lengthens its period (Wittenberg, 2002; Wittenberg et  al., 2006; Capotondi et  al., 2006; Kim et al., 2008; Neale et al., 2008). In ocean models, several parameterizations affect the sim­ ulation of tropical Pacific climate and ENSO. Parameterized vertical mixing affects the intensity and depth of the equatorial thermocline and undercurrent, as well as ENSO’s subsurface feedbacks (Meehl et  al., 2001; Wilson, 2000; Canuto et  al., 2004; Noh et  al., 2005; Ray et  al., 2018a, 2018b). The parameterization of lateral viscosity affects the intensity of the equatorial undercurrent and its associated shears, which in turn affect the intensity of Pacific TIWs and their equatorward heat transport (Stockdale et  al., 1998; Griffies et  al., 2005; Wittenberg et  al., 2018). The parameterization of solar penetration through the water column, which depends on both the ocean turbidity and the optical model used for radiative transfer through the ocean, can also strongly affect the ECT and ENSO, by modifying the structure of the equatorial thermocline, via both local and nonlocal effects (Murtugudde et  al., 2002; Anderson et al., 2007, 2009; Lengaigne et al., 2007). 9.5. CHALLENGES AND OPPORTUNITIES The ability of CGCMs to simulate ENSO continues to improve, offering exciting opportunities for research, forecasting, understanding past variations, and project­ ing the future behavior of ENSO and its global impacts. Many GCMs have also now evolved into more compre­ hensive Earth system models that simulate atmospheric chemistry, ocean biogeochemistry, land vegetation, dust, fire, and the carbon cycle, enabling pioneering new research and applications related to ENSO’s impacts on air quality, ecosystems, agriculture, and fisheries. Other GCMs have pushed toward higher resolution in the atmosphere and ocean, enabling seamless simulations of weather and climate with applications to ENSO’s impacts on tropical cyclones, severe weather, coastal communities, and regional extremes. The research and modeling communities also continue to become better organized. CMIPs are enabling ground­ breaking research by the academic community, by

coordinating multimodel experiments that support periodic assessments by the Intergovernmental Panel on Climate Change (IPCC). The quality and diversity of observational and reanalysis constraints for models are improving and are now being provided in formats that are convenient for modelers to use, e.g. via the Obs4MIPs project (Ferraro et al., 2015). Freely available diagnostic frameworks are enabling more rapid assessment of simu­ lations, and efforts are well underway to support those activities and the broader climate community by providing comprehensive sets of ENSO diagnostics and metrics for models (Guilyardi et al., 2016). There are many ways that models could be better lever­ aged to yield insight into ENSO’s mechanisms, sensitiv­ ities, and predictability. Key foci are (i) the atmosphere response to SSTAs, (ii) surface wind stress and heat flux feedbacks over the tropical Pacific, (iii) the upwelling and vertical mixing near the equator, and (iv) the upper‐ocean heat budgets for both the background climatology and ENSO. Long control runs and large ensembles can be used to illuminate ENSO’s diversity and interdecadal modulation, and perturbed‐physics ensembles can be used to systematically probe the sensitivity of ENSO to model parameters. EMICs and statistical emulators can be used as diagnostics of models, enabling more robust intercom­ parisons and evaluations against short observational records. Seasonal forecasts could be used to better under­ stand the seeds and amplifiers of model biases and initial­ ization shocks. Model analogs could be employed to assess the fundamental predictability of ENSO to provide a baseline of skill against which to evaluate initialized forecasts. Emergent constraints can provide insight into the relationship between model biases and model sensitiv­ ities and possibly leverage the intermodel diversity to yield more reliable projections of future ENSO behavior. Further improvement of ENSO simulations will rely on several factors. First, modeling advances must be sup­ ported by improved observational constraints. These should include more reliable, representative, and diverse observations from moorings, satellites, ships, and drifters, as well as maintenance of the long‐term climate records needed to assess simulated decadal variability and sensi­ tivities to external forcings. Improving observational con­ straints for ENSO simulations and forecasts is a major thrust of the Tropical Pacific Observing System 2020 (TPOS2020) project (Cravatte et al., 2016; Kessler et al., 2019). Also essential are statistical and GCM‐based reanalyses that reconcile the diverse observations, fill gaps between them, and impute variables that are not directly observed; such reanalyses are essential for evalu­ ating the detailed processes (e.g. heat budgets) and multi­ decadal behavior of ENSO in the models. Efforts to rescue and digitize historical observations, such as the Global Oceanographic Data Archaeology and Rescue

218  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Project (GODAR), are also essential to support these reanalyses. Paleo constraints should not be overlooked, as they provide a unique perspective on the longer‐term variability of ENSO. As proxy records become ever more numerous, consistent, and better understood, they can support multiproxy reanalyses extending deep into the past and may even be able to improve understanding of the early part of the instrumental era via merged reanaly­ ses involving both proxy and instrumental observations (e.g., Emile‐Geay et al., 2013, Freund et al., 2019). Second, it will be essential to maintain and advance a robust hierarchy of models, including simple conceptual models, EMICs, ocean‐only and atmosphere‐only ­simulations, statistical and empirical models, and com­ prehensive CGCMs (free, nudged, flux‐adjusted, or partially‐coupled). Quality documentation and software archiving (e.g. via github) are also essential to make EMICs and simple models easily available to the community. Investments in intellectual, software infra­ structure, and computing resources are needed to support improved model resolution, more realistic processes and parameterizations, larger ensembles to characterize extremes, and more diverse and creative sets of experi­ ments with the GCMs, as well as improved foundations in ENSO theory for the simpler models. Based on past experience, the return on these invest­ ments, from improved predictions and projections that benefit global economies and societies to better fundamental understanding of Earth’s climate variations, is likely to be exceptional. ACKNOWLEDGMENTS We thank the editors and three anonymous reviewers for their careful reading of the manuscript, and their help in improving it. EG is funded by the Centre National de la Recherche Scientifique (CNRS), ML is funded by the Institut de Recherche pour le Développement (IRD), and they both acknowledge the support of the Belmont Forum project GOTHAM, under grant ANR‐15‐ JCLI‐0004‐01 and the Agence Nationale de la Recherche project ARISE, under grant ANR‐18‐CE01‐0012. AC was supported by the NASA Physical Oceanography Program (Award NNX15AG46G). REFERENCES Abellán, E., S. McGregor, & M. H. England (2017). Analysis of the southward wind shift of ENSO in CMIP5 models. J. Climate, 30, 2415–2435, https://doi.org/10.1175/JCLI‐D‐16‐0326.1 AchutaRao, K., & K. Sperber (2006). ENSO simulations in coupled ocean‐atmosphere models: Are the current models better? Climate Dyn., 27, 1–16.

An, S.‐I., & Jin, F.‐F. (2001). Collective role of thermocline and zonal advective feedbacks in the ENSO mode. J. Climate, 14(16), 3421–3432. An, S.‐I., & F.‐F. Jin (2004). Nonlinearity and asymmetry of ENSO. J. Climate, 17, 2399–2412. An, S.‐I., & B. Wang (2000). Interdecadal change of the struc­ ture of ENSO mode and its impact on the ENSO frequency. J. Climate, 13, 2044‐2055. doi:10.1175/1520‐0442%282000% 290132.0.CO;2 An, S., Y. Ham, J. Kug, F. Jin, & I. Kang (2005). El Niño–La Niña asymmetry in the coupled model intercomparison project simulations. J. Climate, 18, 2617–2627. doi:10.1175/ JCLI3433.1 Ahn, M. S., D. Kim, K. R. Sperber, I.‐S. Kang, E. Maloney, D. Waliser, & H. Hendon (2017). MJO simulation in CMIP5 climate models: MJO skill metrics and process‐oriented diag­ nosis. Climate Dyn., 49, 4023–4045. doi:10.1007/s00382‐017‐ 3558‐4 Anderson, W. G, A. Gnanadesikan, R. Hallberg, J. Dunne, & B. L. Samuels (2007). Impact of ocean color on the mainte­ nance of t D. J. Vimont (2017). Characterizing unforced multi‐decadal variability of ENSO: A case study with the GFDL CM2.1 coupled GCM. Climate Dyn., 49(7–8), 2845– 2862. doi: 10.1007/s00382‐016‐3477‐9 Barnett, T. P., Graham, N., Pazan, S., White, W., Latif, M., & Flügel, M. (1993). ENSO and ENSO‐related predictability. Part I: Prediction of equatorial Pacific sea surface tempera­ ture with a hybrid coupled ocean–atmosphere model. Journal of Climate, 6(8), 1545–1566. Barnston, A.G., M. K. Tippett, M. L. L’Heureux, S. Li, & D. G. DeWitt (2012). Skill of real‐time seasonal ENSO model pre­ dictions during 2002–11: Is Our capability increasing? Bull. Amer. Meteor. Soc., 93, 631–651. https://doi.org/10.1175/ BAMS‐D‐11‐00111.1 Battisti, D. S., & A. C. Hirst, 1989). Interannual variability in the tropical atmosphere–ocean system: Influence of the basic state and ocean geometry. J. Atmos. Sci., 46, 1678–1712. Bellenger, H., E. Guilyardi, J. Leloup, J. Lengaigne, & J. Vialard (2014). ENSO representation in climate models: from CMIP3 to CMIP5. Climate Dyn., 42, 1999–2018. doi:10.1007/ s00382‐013‐1783‐z Bernie, D. J., E. Guilyardi, G. Madec, J. M. Slingo, S. J. Woolnough, & J. Cole (2008). Impact of resolving the durnal cycle in an ocean‐atmosphere GCM. Part 2: A diurnally cou­ pled CGCM. Climate Dyn., 31, 909–925. doi: 10.1007/ s00382‐008‐0429‐z Bianucci, M., A. Capotondi, S. Merlino, & R. Mannella (2018). Estimate of the average timing for strong El Niño events using the recharge oscillator model with a multiplicative per­ turbation. Chaos, 28, 103118. Bony S., K. Lau, & Y. C. Sud (1997). Sea surface temperature and large‐scale circulation influences on tropical greenhouse effect and cloud radiative forcing. J. Climate, 10, 2055–2077. Brown J. N., & A. V. Fedorov (2010). How much energy is trans­ ferred from the winds to the thermocline on ENSO time­ scales? J. Clim., 23, 1563–1580. Brown J., A. Fedorov, & E. Guilyardi (2011). How well do cou­ pled models replicate the energetics of ENSO? Clim. Dyn., 36, 2147–2158.

HISTORY AND PROGRESS OF ENSO MODELING  219 Burgers G., F.‐F. Jin, & G. J. van Oldenborgh (2005). The sim­ plest ENSO recharge oscillator. Geophys. Res. Lett., 32, L13706. doi:10.1029/2005GL022951 Burls, N. J., & A. V. Fedorov (2014). What controls the mean east–west sea surface temperature gradient in the equatorial Pacific: The role of cloud albedo. J. Climate, 27, 2757–2778. doi:10.1175/JCLI‐D‐13‐00255.1 Cai, W., et al. (2014). Increasing frequency of extreme El Niño events due to greenhouse warming. Nat. Clim. Change, 4, 111–116. doi:10.1038/NCLIMATE2100 Cai, W., G. Wang, B. Dewitte, L. Wu, A. Santoso, K. Takahashi, et al. (2018). Increased variability of eastern Pacific El Niño under greenhouse warming. Nature, 564, 201–206. Canuto, V. M., A. Howard, Y. Cheng, & R. L. Miller (2004). Latitude‐dependent vertical mixing and the tropical thermo­ cline in a global OGCM. Geophys. Res. Lett., 31, L16305. doi:10.1029/2004GL019891 Capotondi, A. (2013). ENSO diversity in the NCAR CCSM4. J. Geophys. Res., 118, 4755–4770. Capotondi, A., & P. D. Sardeshmukh (2017). Is El Niño really changing? Geophys. Res. Lett., 44, 8548–8556. doi:10.1002/ 2017GL074515 Capotondi, A., M. A. Alexander, C. Deser, & M. McPhaden (2005). Anatomy and decadal evolution of the Pacific sub­ tropical‐tropical cells (STCs). J. Climate, 18, 3739–3758. Capotondi, A., A. Wittenberg, & S. Masina (2006). Spatial and temporal structure of tropical Pacific interannual variability in 20th century coupled simulations. Ocean Modelling, 15, 274–298. doi: 10.1016/j.ocemod.2006.02.004 Capotondi, A., Y.‐G. Ham, A. T. Wittenberg, & J.‐S. Kug (2015a). Climate model biases and El Niño Southern Oscillation (ENSO) simulation. U.S. CLIVAR Variations, 13(1), 21–25. Capotondi, A., A. T. Wittenberg, et al. (2015b). Understanding ENSO diversity. Bull. Amer. Meteor. Soc., 96, 921–938. doi: 10.1175/BAMS‐D‐13‐00117.1 Capotondi, A., P. D. Sardeshmukh, & L. Ricciardulli (2018). The nature of the stochastic wind forcing of ENSO. J. Climate, 31, 8081–8099. https://doi.org/10.1175/JCLI‐D‐17‐0842.1 Chen, D., S. E. Zebiak, A. J. Busalacchi, & M. A. Cane (1995). An improved procedure for EI Nino forecasting: Implications for predictability. Science, 269(5231), 1699–1702. Chen, D., M. A. Cane, A. Kaplan, S. E. Zebiak, & D. Huang (2004). Predictability of El Niño over the past 148 years. Nature, 428(6984), 733. Chen, C., M. A. Cane, A. T. Wittenberg, & D. Chen (2017). ENSO in the CMIP5 simulations: Life cycles, diversity, and responses to climate change. J. Climate, 30(2), 775–801. doi: 10.1175/JCLI‐D‐15‐0901.1 Chiodi, A. M., & D.E. Harrison (2017). Observed El Niño SSTA development and the effects of easterly and westerly wind events in 2014/15. J. Climate, 30, 1505–1519, https://doi. org/10.1175/JCLI‐D‐16‐0385.1 Choi, J., S.‐I. An, J.‐S. Kug, & S.‐W. Yeh (2011). The role of mean state on changes in El Niño flavor. Clim. Dyn., 37, 1205–1215. Choi, J., S.‐I. An, & S.‐W. Yeh (2012). Decadal amplitude mod­ ulation of two types of ENSO and its relationship with the mean state. Clim. Dyn., 28, 2631–2644.

Choi, J., S. An, S. Yeh, & J. Yu (2013a). ENSO‐like and ENSO‐ induced tropical Pacific decadal variability in CGCMs. J. Climate, 26, 1485–1501. doi:10.1175/JCLI‐D‐12‐00118.1 Choi, K.‐Y., G. A. Vecchi, & A. T. Wittenberg (2013b). ENSO transition, duration and amplitude asymmetries: Role of the nonlinear wind stress coupling in a conceptual model. J. Climate, 26, 9462–9476. doi: 10.1175/JCLI‐D‐13‐00045.1 Choi, K.‐Y., G. A. Vecchi, & A. T. Wittenberg (2015). Nonlinear zonal wind response to ENSO in the CMIP5 models: Roles of the zonal and meridional shift of the ITCZ/SPCZ and the simulated climatological precipitation. J. Climate, 28, 8556– 8573. doi: 10.1175/JCLI‐D‐15‐0211.1 Clarke, A. J., S. Van Gorder, & G. Colantuono (2007). Wind stress curl and ENSO discharge/recharge in the equatorial Pacific. J. Phys. Oceanogr., 37(4), 1077–1091. Claussen, M., L. Mysak, A. Weaver, et  al. (2002). Climate Dynamics, 18, 579. https://doi.org/10.1007/s00382‐001‐0200‐1 Collins, M., S.‐I. An, W. Cai, A. Ganachaud, E. Guilyardi, F.‐F. Jin, et  al. (2010). The impact of global warming on the tropical Pacific and El Niño. Nature Geoscience, 3, 391–397. doi: 10.1038/ngeo868 Collins M., M. Sutherland, L. Bouwer, S.‐M. Cheong, T. Frölicher, H. Jacot Des Combes, et  al. (2019). Extremes, abrupt changes and managing risk. In H.‐O. Pörtner, D.C. Roberts, V. Masson‐Delmotte, P. Zhai, M. Tignor, E. Poloczanska, et al. (Eds.), IPCC special report on the ocean and cryosphere in a changing climate. IPCC. Cravatte, S., B. Kessler, N. Smith, S. Wijffels, L. Yu, K. Ando, et al. (2016). First report of TPOS 2020. GOOS‐215 (200 pp.) [Available online at http://tpos2020.org/first‐report.] doi: 10.13140/RG.2.2.28674.07363 Delecluse, P., M. Davey, Y. Kitamura, S. G. H. Philander, M. Suarez, & L. Bengtsson, 1998). TOGA review paper: Coupled general circulation modeling of the tropical Pacific. J. Geophys. Res., 103, 14,357–14,373. Delworth, T. L., A. Rosati, W. Anderson, A. J. Adcroft, V. Balaji, R. Benson, et  al. (2012). Simulated climate and cli­ mate change in the GFDL CM2.5 high‐resolution coupled climate model. J. Climate, 25, 2755–2781. doi: 10.1175/ JCLI‐D‐11‐00316.1 de Szoeke, S.P., & S. Xie (2008). The tropical eastern Pacific seasonal cycle: Assessment of errors and mechanisms in IPCC AR4 coupled ocean–atmosphere general circulation models. J. Climate, 21, 2573–2590. doi:10.1175/2007 JCLI1975.1 Di Lorenzo, E., G. Liguori, N. Schneider, J. C. Furtado, B. T. Anderson, & M. A. Alexander (2015). ENSO and meridional modes: A null hypothesis for Pacific climate variability. Geophys. Res. Lett., 42, 9440–9448. doi:10.1002/2015GL066281 DiNezio, P. N., B. P. Kirtman, A. C. Clement, S.‐K. Lee, G. A. Vecchi, & A. Wittenberg (2012). Mean climate controls on the simulated response of ENSO to increasing greenhouse gases. J. Climate, 25, 7399–7420. doi: 10.1175/JCLI‐D‐11‐00494.1 Ding, H., M. Newman, M. A. Alexander, & A. T. Wittenberg (2018). Skillful climate forecasts of the tropical Indo‐Pacific Ocean using model‐analogs. J. Climate, 31(14), 5437–5459. doi:10.1175/JCLI‐D‐17‐0661.1 Ding, H., M. Newman, M. A. Alexander, & A. T. Wittenberg (2019). Diagnosing secular variations in retrospective ENSO

220  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE seasonal forecast skill using CMIP5 model‐analogs. Geophys. Res. Lett., 46(3), 1721–1730. doi:10.1029/2018GL080598 Ding, H., M. Newman, M. A. Alexander, & A. T. Wittenberg (2020). Relating CMIP5 model biases to seasonal forecast skill in the tropical Pacific. Geophys. Res. Lett., 47(5), e2019GL086765. doi:10.1029/2019GL086765 Dommenget, D., & Y. Yu (2016). The seasonally changing cloud feedbacks contribution to the ENSO seasonal phase‐locking. Climate Dynamics, 47(12), 3661–3672. Dommenget, D., T. Bayr, & C. Frauen (2013). Analysis of the non‐linearity in the pattern and time evolution of El Niño Southern Oscillation. Climate Dyn., 40, 2825–2847. doi:10.1007/s00382‐012‐1475‐0 Eisenman, I., L. Yu, & E. Tziperman (2005). Westerly wind bursts: ENSO’s tail rather than the dog? Journal of Climate, 18(24), 5224–5238. Emile‐Geay, J., K. M. Cobb, M. E. Mann, & A. T. Wittenberg (2013). Estimating central equatorial Pacific SST variability over the past millennium. Part II: Reconstructions and implications. J. Climate 26, 2329–2352. Eyring, V., S. Bony, G. A. Meehl, C. A. Senior, B. Stevens, R. J. Stouffer, & K. E. Taylor (2016). Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experi­ mental design and organization. Geosci. Model Dev., 9, 1937– 1958, https://doi.org/10.5194/gmd‐9‐1937‐2016 Fedorov, A. V., & S. G. Philander (2000). Is El Niño changing? Science, 288, 1997–2002. Fedorov, A. V., & S. G. Philander (2001). A stability analysis of tropical ocean–atmosphere interactions: Bridging measure­ ments and theory for El Niño. J. Climate, 14, 3086–3101. Fedorov, A.V., S. L. Harper, S. G. Philander, B. Winter, & A. Wittenberg (2003). How predictable is El Niño? Bull. Amer. Meteor. Soc., 84, 911–920. https://doi.org/10.1175/ BAMS‐84‐7‐911 Feng, J., & T. Lian (2018). Assessing the relationship between MJO and equatorial Pacific WWBs in observations and CMIP5 models. J. Climate, 31, 6393–6410. https://doi. org/10.1175/JCLI‐D‐17‐0526.1 Ferraro, R., D. E. Waliser, P. Gleckler, K. E. Taylor, & V. Eyring (2015). Evolving Obs4MIPs to support Phase 6 of the Coupled Model Intercomparison Project (CMIP6). Bull. Amer. Meteor. Soc., 96, ES131–ES133. https://doi. org/10.1175/BAMS‐D‐14‐00216.1 Flato, G., J. Marotzke, B. Abiodun, P. Braconnot, S. C. Chou, W. Collins, et  al. (2013). Evaluation of climate models. In Stocker, T. F., D. Qin, G.‐K. Plattner, M. Tignor, S. K. Allen, J. Boschung, et al. (Eds.), Climate change 2013: The physical science basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC AR5). Cambridge University Press. Freund, M. B., Henley, B. J., Karoly, D. J., McGregor, H. V., Abram, N. J. & Dommenget, D. (2019). Higher frequency of central Pacific El Niño events in recent decades relative to past centuries. Nature Geoscience, 6, 450–455. doi:10.1038/ s41561‐019‐0353‐3 Gebbie, G., I. Eisenman, A. Wittenberg, & E. Tziperman (2007). Modulation of westerly wind bursts by sea surface tempera­ ture: A semistochastic feedback for ENSO. J. Atmos. Sci., 64, 3281–295. doi: 10.1175/JAS4029.1

Gill, A. E. (1980) Some simple solutions for heat‐induced tropical circulation. Q J R Meteor Soc, 106, 447–462. https:// doi.org/10.1002/qj.49710644905 Gillett, N. P., H. Shiogama, B. Funke, G. Hegerl, R. Knutti, K. Matthes, et al. (2016). The Detection and Attribution Model Intercomparison Project (DAMIP v1.0) contribution to CMIP6. Geosci. Model Dev., 9, 3685–3697. Graham, T. (2014). The importance of eddy permitting model resolution for simulation of the heat budget of tropical insta­ bility waves. Ocean Modelling, 79, 21–32. doi:10.1016/j. ocemod.2014.04.005 Graham, F. S., J. N. Brown, C. Langlais, S. J. Marsland, A. T. Wittenberg, & N. J. Holbrook (2014). Effectiveness of the Bjerknes stability index in representing ocean dynamics. Climate Dyn., 43, 2399–2414. doi: 10.1007/s00382‐ 014‐2062‐3 Graham, F. S., J. N. Brown, A. T. Wittenberg, & N. J. Holbrook (2015). Reassessing conceptual models of ENSO. J. Climate, 28, 9121–9142. doi: 10.1175/JCLI‐D‐14‐00812.1 Graham, F. S., A. T. Wittenberg, J. N. Brown, S. J. Marsland, & N. J. Holbrook (2017). Understanding the double peaked El Niño in coupled GCMs. Climate Dyn., 48(5), 2045–2063. doi: 10.1007/s00382‐016‐3189‐1 Griffies, S. M., A. Gnanadesikan, K. W. Dixon, J. P. Dunne, et al. (2005). Formulation of an ocean model for global climate sim­ ulations. Ocean Sci., 1, 45‐79. doi:10.5194/os‐1‐45‐2005 Griffies, S. M., M. Winton, W. G. Anderson, R. Benson, T. L. Delworth, C. O. Dufour, et al. (2015). Impacts on ocean heat from transient mesoscale eddies in a hierarchy of climate models. J. Climate, 28, 952–977. doi: 10.1175/ JCLI‐D‐14‐00353.1 Guckenheimer J., A. Timmermann, H. Dijkstra, & A. Roberts (2017). (Un)predictability of strong El Niño events. Dynamics and Statistics of the Climate System, 2(1). dzx004, https://doi. org/10.1093/climsys/dzx004 Guilyardi, E. (2006). El Niño–mean state–seasonal cycle inter­ actions in a multi‐model ensemble. Clim. Dyn., 26, 329–348, doi: 10.1007/s00382‐005‐0084‐6 Guilyardi E., S. Gualdi, J. M. Slingo, A. Navarra, P. Delecluse, J. Cole, et al. (2004). Representing El Niño in coupled ocean‐ atmosphere GCMs: The dominant role of the atmospheric component. J. Climate, 17, 4623–4629. Guilyardi, E., A. Wittenberg, A. Fedorov, M. Collins, C. Wang, A. Capotondi, et  al. (2009). Understanding El Niño in ocean‐atmosphere general circulation models: Progress and ­challenges. Bull. Amer. Meteor. Soc., 90, 325–340. doi: 10.1175/2008BAMS2387.1 Guilyardi, E., H. Bellenger, M. Collins, S. Ferrett, W. Cai, & A. Wittenberg (2012a). A first look at ENSO in CMIP5. CLIVAR Exchanges, 17, 29–32. ISSN: 1026‐0471. Guilyardi, E., W. Cai, M. Collins, A. Fedorov, F.‐F. Jin, A. Kumar, et  al. (2012b). New strategies for evaluating ENSO processes in climate models. Bull. Amer. Met. Soc., 93, 235– 238. doi: 10.1175/BAM S‐D‐11‐00106.1 Guilyardi, E., A. Wittenberg, M. Balmaseda, W. Cai, M. Collins, M. J. McPhaden, et  al. (2016). Fourth CLIVAR workshop on the evaluation of ENSO processes in climate models: ENSO in a changing climate. Bull. Amer. Meteor. Soc., 97(5), 817–820. doi: 10.1175/BAMS‐D‐15‐00287.1

HISTORY AND PROGRESS OF ENSO MODELING  221 Haarsma, R. J., M. J. Roberts, P. L. Vidale, C. A. Senior, A. Bellucci, Q. Bao, et  al. (2016). High Resolution Model Intercomparison Project (HighResMIP v1.0) for CMIP6. Geoscientific Model Development, 9, 11, 4185–4208. Ham, Y.‐G., & J.‐S. Kug (2012). How well do current climate models simulate two types of El Nino? Climate Dyn., 39, 383–398. doi:10.1007/s00382‐011‐1157‐3 Ham, Y.‐G., & J.‐S. Kug (2015). Improvement of ENSO simula­ tion based on intermodel diversity. J. Climate, 28, 998–1015. doi:10.1175/JCLI‐D‐14‐00376.1 Hayashi, M., & M. Watanabe (2017). ENSO complexity induced by state dependence of westerly wind events. J. Climate, 30, 3401–3420. doi:10.1175/JCLI‐D‐16‐0406.1 He, J., N. C. Johnson, G. A. Vecchi, B. Kirtman, A. T. Wittenberg, & S. Sturm (2018). Precipitation sensitivity to local variations in tropical sea surface temperature. J. Climate, 31(22), 9225– 9238. doi:10.1175/JCLI‐D‐18‐0262.1 Heinze, C., Eyring, V., Friedlingstein, P., Jones, C., Balkanski, Y., Collins, et al. (2019). ESD Reviews: Climate feedbacks in the Earth system and prospects for their evaluation. Earth Syst. Dynam., 10, 379‐452. https://doi.org/10.5194/esd‐10‐379‐ 2019 (2019. Held, I. M. (2005). The gap between simulation and under­ standing in climate modeling. Bull. Amer. Meteor. Soc., 86, 1609–1614. Holmes, R. M., & L. N. Thomas (2015). The modulation of equatorial turbulence by tropical instability waves in a regional ocean model. J. Phys. Oceanogr., 45, 1155–1173. doi:10.1175/JPO‐D‐14‐0209.1 Holmes, R., L. Thomas, L. Thompson, & D. Darr (2014). Potential vorticity dynamics of tropical instability vortices. J. Phys. Oceanogr., 44, 995–1011. doi:10.1175/JPO‐D‐13‐0157.1 Hu, S., & Fedorov, A. V. (2016). Exceptionally strong easterly wind burst stalling El Niño of 2014. Pnas, 113(8), 2005–2010. http://doi.org/10.1007/s00704‐008‐0069‐6 Huang, P., & D. Chen (2017). Enlarged asymmetry of tropical Pacific rainfall anomalies induced by El Niño and La Niña under global warming. J. Climate, 30, 1327–1343. doi:10.1175/ JCLI‐D‐16‐0427.1 Huang, P., & J. Ying (2015). A multimodel ensemble pattern regression method to correct the tropical Pacific SST change patterns under global warming. J. Climate, 28, 4706–4723. doi:10.1175/JCLI‐D‐14‐00833.1 Hung, M., J. Lin, W. Wang, D. Kim, T. Shinoda, & S. J. Weaver (2013). MJO and convectively coupled equatorial waves simulated by CMIP5 Climate Models. J. Climate, 26, 6185– 6214, https://doi.org/10.1175/JCLI‐D‐12‐00541.1 Imada, Y., & M. Kimoto (2012). Parameterization of tropical instability waves and examination of their impact on ENSO Characteristics. J. Climate, 25, 4568–4581, doi:10.1175/ JCLI‐D‐11‐00233.1 Izumo, T., Lengaigne, M., Vialard, J., Suresh, I., & Planton, Y. (2018). On the physical interpretation of the lead relation bet­ ween the warm water volume and the El Niño Southern Oscillation. Climate Dynamics, 52, 2923–2942. https://doi. org/10.1007/s00382‐018‐4313‐1 Jia, L., X. Yang, G. A. Vecchi, R. G. Gudgel, T. L. Delworth, A. Rosati, et al. (2015). Improved seasonal prediction of temper­ ature and precipitation over land in a high‐resolution GFDL

climate model. J. Climate, 28, 2044–2062. doi: 10.1175/ JCLI‐D‐14‐00112.1 Jin, F. (1997). An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model. J. Atmos. Sci., 54, 811–829. https://doi. org/10.1175/1520‐0469(1997)0542.0.CO;2 Jin, F.‐F. (2001). Low‐frequency modes of tropical ocean dynamics. J. Climate, 14(18), 3874–3881. Jin, F. F., & J. D. Neelin (1993). Modes of interannual tropical ocean–atmosphere interaction: A unified view. Part I: Numerical results. J. Atmos. Sci, 50, 3477–3503. Jin F‐F, S. T. Kim, & L. Bejarano (2006). A coupled‐stability index for ENSO. Geophys. Res. Lett., 33, L23708, doi:10.1029/ 2006GL027221 Jin, F.‐F., L. Lin, A. Timmermann, & J. Zhao (2007). Ensemble‐ mean dynamics of the ENSO recharge oscillator under state‐ dependent stochastic forcing. Geophys.Res.Lett., 34. doi:10.1029/2006GL027372 Jochum, M. & R. Murtugudde (2006). Temperature advection by tropical instability waves. J. Phys. Oceanogr., 36, 592–605. doi:10.1175/JPO2870.1 Jochum, M., et al. (2005). The impact of horizontal resolution on the tropical heat budget in an Atlantic Ocean Model. J. Climate, 18, 841–851. doi:10.1175/JCLI‐3288.1 Jochum, M., et  al. (2007). Tropical atmospheric variability forced by oceanic internal variability. J. Climate, 20, 765–771. doi:10.1175/JCLI4044.1 Kageyama, M., P. Braconnot, S. P. Harrison, A. M. Haywood, J. Jungclaus, B. L. Otto‐Bliesner, et  al. (2018). PMIP4‐ CMIP6: The contribution of the Paleoclimate Modelling Intercomparison Project to CMIP6. Geosci. Model Dev., 11, 1033–1057. https://doi.org/10.5194/gmd‐11‐1033‐2018 Kajtar, J. B., A. Santoso, M. H. England, & W. Cai (2017). Tropical climate variability: Interactions across the Pacific, Indian, and Atlantic Oceans. Climate Dyn., 48, 2173–2190. doi:10.1007/s00382‐016‐3199‐z Kamenkovich, I. V., & E. S. Sarachik (2004). Reducing errors in temperature and salinity in an ocean model forced by restoring boundary conditions. J. Phys. Oceanogr., 34, 1856–1869. doi:10.1175/1520‐0485(2004)0342.0.CO;2 Karamperidou, C., M. A. Cane, U. Lall, & A. T. Wittenberg (2014). Intrinsic modulation of ENSO predictability viewed through a local Lyapunov lens. Climate Dyn., 42, 253–270. doi: 10.1007/s00382‐013‐1759‐z Kessler, W. S., S. E. Wijffels, S. Cravatte, N. Smith, A. Kumar, Y. Fujii, et al. (2019). Second Report of TPOS 2020. GOOS‐234, 265 pp. Available online at http://tpos2020.org/project‐ reports/second‐report. doi: 10.13140/RG.2.2.17161.29289 Khodri M., T. Izumo, J. Vialard, S. Janicot, C. Cassou, M. Lengaigne, et  al. (2017). Tropical explosive volcanic erup­ tions can trigger El Niño by cooling tropical Africa. Nature Communications, 8, 778. doi:10.1038/s41467‐017‐00755‐6 Kim, D., J.‐S. Kug, I.‐S. Kang, F.‐F. Jin, & A. T. Wittenberg (2008). Tropical Pacific impacts of convective momentum transport in the SNU coupled GCM. Climate Dyn., 31, 213– 226. doi: 10.1007/s00382‐007‐0348‐4 Kim, S. T. & F.‐F. Jin (2011). An ENSO stability analysis. Part II: Results from the twentieth and twenty‐first century simu­ lations of the CMIP3 models. Climate Dynamics, 36, 1609–1627.

222  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Kirtman, B. P., & Zebiak, S. E. (1997). ENSO simulation and prediction with a hybrid coupled model. Monthly Weather Review, 125(10), 2620–2641. Kleeman, R. (1993). On the dependence of hindcast skill on ocean thermodynamics in a coupled ocean–atmosphere model. J. Climate, 6, 2012–2033. Kleeman, R., & S. B. Power (1994). Limits to predictability in a coupled ocean‐atmosphere model due to atmospheric noise. Tellus, 46A, 529–540. Kociuba, G., & S. B. Power (2015). Inability of CMIP5 models to simulate recent strengthening of the Walker Circulation: Implications for projections. J. Climate, 28, 20–35. doi:10.1175/JCLI‐D‐13‐00752.1 Krishnamurthy, L., G. Vecchi, R. Msadek, A. Wittenberg, T. Delworth, & F. Zeng (2015). The seasonality of the Great Plains Low‐Level Jet and ENSO relationship. J. Climate, 28, 4525–4544. doi: 10.1175/JCLI‐D‐14‐00590.1 Krishnamurthy, L., G. A. Vecchi, R. Msadek, H. Murakami, A. Wittenberg, & F. Zeng (2016). Impact of strong ENSO on regional tropical cyclone activity in a high‐resolution climate model in the North Pacific and North Atlantic. J. Climate, 29, 2375–2394. doi: 10.1175/JCLI‐D‐0468.1 Kröger, J., & F. Kucharski (2011). Sensitivity of ENSO charac­ teristics to a new interactive flux correction scheme in a cou­ pled GCM. Climate Dyn., 36, 119–137. doi:10.1007/ s00382‐010‐0759‐5 Kug, J.‐S., J. Choi, S.‐I. An, F.‐F. Jin, & A. T. Wittenberg (2010). Warm pool and cold tongue El Niño events as simulated by the GFDL CM2.1 coupled GCM. J. Climate, 23, 1226–1239. doi: 10.1175/2009JCLI3293.1 Kug, J.‐S., Y.‐G. Ham, J.‐Y. Lee, & F.‐F. Jin (2012). Improved simulation of two types of El Nino in CMIP5 models. Environ. Res. Lett., 7, 034002. doi:10.1088/1748‐9326/7/3/034002 Large, W. G., & G. Danabasoglu (2006). Attribution and impacts of upper‐ocean biases in CCSM3. J. Climate, 19, 2325–2346. doi:10.1175/JCLI3740.1 Larson, S. M., & Kirtman, B. P. (2015). Revisiting ENSO Coupled Instability Theory and SST error growth in a fully coupled model. Journal of Climate, 28(12), 4724–4742. http:// doi.org/10.1175/JCLI‐D‐14‐00731.1 Latif M., & A. Villwock (1990). Interannual variability in the tropical Pacific as simulated in coupled ocean‐atmosphere models. Journal of Marine Systems, 1, 51–60, https://doi.org/ 10.1016/0924‐7963(90)90119‐U Lee, S.‐K., A. T. Wittenberg, D. B. Enfield, S. J. Weaver, C. Wang, & R. M. Atlas (2016). U.S. regional tornado out­ breaks and their links to spring ENSO phases and North Atlantic SST variability. Environ. Res. Lett., 11(4), 044008. doi: 10.1088/1748‐9326/11/4/044008 Lee, S.‐K., H. Lopez, E.‐S. Chung, P. DiNezio, S.‐W. Yeh, & A. T. Wittenberg (2018). On the fragile relationship between El Niño and California rainfall. Geophys. Res. Lett., 45(2), 907– 915. doi:10.1002/2017GL076197 Lengaigne M., J. P. Boulanger, C. Menkes, P. Delecluse, & J. Slingo (2004). Westerly wind events in the tropical Pacific and their influence on the coupled ocean–atmosphere system: A review. In C. Wang, S.‐P. Xie, & J. A. Carton (Eds.), Earth’s climate: The ocean–atmosphere interaction (pp. 49–69). Amer. Geophys. Union.

Lengaigne M., C. Menkes, O. Aumont, T. Gorgues, L. Bopp, J.-M. André, & G. Madec (2007). Influence of the oceanic biology on the tropical Pacific climate in a Coupled General Circulation Model. Climate Dynamics, 28: 503–516, doi:10.1007/s00382-006-0200-2 Levine, A., & F. F. Jin (2010). Noise‐induced instability in the ENSO recharge oscillator. J. Atm. Sci., 67(2):529–542. Levine, A. F. Z., & F. F. Jin (2017). A simple approach to quan­ tifying the noise–ENSO interaction. Part I: Deducing the state‐dependency of the windstress forcing using monthly mean data. Climate Dyn., 48, 1–18. doi:10.1007/s00382‐015‐2748‐1 Levine, A. F. Z., F. F. Jin, & M. J. McPhaden (2016). Extreme noise–Extreme El Niño: How state‐dependent forcing creates El Niño–La Niña Asymmetry. J. Climate. 29, 5483‐5499. doi: 10.1175/JCLI‐D‐16‐0091.1 Liguori, G., & E. Di Lorenzo (2018). Meridional modes and increasing Pacific decadal variability under anthropogenic forcing. Geophys. Res. Lett., 45, 983–991. doi:10.1002/ 2017GL076548 Lin, J. (2007). The double‐ITCZ problem in IPCC AR4 coupled GCMs: Ocean–atmosphere feedback analysis. J. Climate, 20, 4497–4525. https://doi.org/10.1175/JCLI4272.1 Lin, R., F. Zheng, & X. Dong (2018). ENSO frequency asym­ metry and the Pacific decadal oscillation in observations and 19 CMIP5 models. Adv. Atmos. Sci., 35, 495–506. doi:10.1007/ s00376‐017‐7133‐z Liu, Z., & E. Di Lorenzo (2018). Mechanisms and predictability of Pacific decadal variability. Current Climate Change Reports. https://doi.org/10.1007/s40641‐018‐0090‐5 Lloyd J., E. Guilyardi, H. Weller, & J. Slingo (2009). The role of atmosphere feedbacks during ENSO in the CMIP3 models. Atmos. Sci. Let., 10, 170–176. Lloyd J., E. Guilyardi, H. Weller, (2012). The role of atmosphere feedbacks during ENSO in the CMIP3 models. Part III: The shortwave flux feedback. J. Clim., 25, 4275–4293. Luo J.‐J., et al. (2012). Indian Ocean warming modulates Pacific climate change. Proc. Natl. Acad. Sci., 109, 18701–18706. doi:10.1073/pnas.1210239109 Lutsko, N. J., & K. Takahashi (2018). What can the internal variability of CMIP5 models tell us about their climate sensi­ tivity? J. Climate, 31, 5051–5069. https://doi.org/10.1175/ JCLI‐D‐17‐0736.1 Lyu, K., X. Zhang, J. A. Church, & J. Hu (2015). Quantifying internally generated and externally forced climate signals at regional scales in CMIP5 models. Geophys. Res. Lett., 42, 9394–9403. doi: 10.1002/2015GL065508 Maes, C., & S. Belamari (2011). On the impact of salinity barrier layer in the Pacific ocean mean state and ENSO. Sci. Online Lett. Atmos., 7, 97‐100. doi:10.2151/sola.2011‐025 Maes, C., J. Picaut, & S. Belamari (2005). Importance of the salinity barrier layer for the buildup of El Niño. J. Climate, 18, 104–118. doi:10.1175/JCLI‐3214.1 Magnan, A. K., M. Garschagen, J‐P. Gattuso, J. E. Hay, N. Hilmi, E. Holland, et  al. (2019). Integrative cross‐chapter box on low‐lying islands and coasts. In H.‐O. Pörtner, D. C. Roberts, V. Masson‐Delmotte, P. Zhai, M. Tignor, E. Poloczanska, et  al., IPCC special report on the ocean and cryosphere in a changing climate. IPCC.Magnusson, L., M. Alonso‐Balmaseda, & F. Molteni (2013a). On the dependence

HISTORY AND PROGRESS OF ENSO MODELING  223 of ENSO simulation on the couped model mean state. Climate Dyn., 41, 1509–1525. doi:10.1007/s00382‐012‐1574‐y Magnusson, L., M. Alonso‐Balmaseda, S. Corti, F. Molteni, & T. Stockdale (2013). Evaluation of forecast strategies for seasonal and decadal forecasts in presence of systematic model errors. Climate Dyn., 41, 2393–2409. doi:10.1007/ s00382‐012‐1599‐2 Manganello, J. V., & B. Huang (2009). The influence of systematic errors in the Southeast Pacific on ENSO vari­ ability and prediction in a coupled GCM. Climate Dyn., 32, 1015–1034. doi:10.1007/s00382‐008‐0407‐5 Marchesiello, P., X. Capet, C. Menkes, & S. C. Kennan (2011). Submesoscale dynamics in tropical instability waves. Ocean Modelling, 39, 31–46. doi:10.1016/j.ocemod.2011.04.011 Marzeion, B., Timmermann, A., Murtugudde, R., & Jin, F. F. (2005). Biophysical feedbacks in the tropical Pacific. Journal of Climate, 18(1), 58–70. McGregor, S., M. F. Stuecker, J. B. Kajtar, M. H. England, & M. Collins (2018). Model tropical Atlantic biases underpin diminished Pacific decadal variability. Nat. Clim. Change, 8, 493–498. doi:10.1038/s41558‐018‐0163‐4 McPhaden, M. J., & D. Zhang (2002). Slowdown of the merid­ ional overturning circulation in upper Pacific Ocean. Nature, 415, 603. Meehl, G. A., P. R. Gent, J. M. Arblaster, B. L. Otto‐Bliesner, E. C. Brady, & A. Craig (2001). Factors that affect the amplitude of El Nino in global coupled climate models. Climate Dyn., 17, 515–526. doi:10.1007/PL00007929 Menkes, C., et  al. (2006). A modeling study of the impact of tropical instability waves on the heat budget of the eastern equatorial Pacific. J. Phys. Oceanogr., 36, 847–865. doi:10.1175/JPO2904.1 Merryfield, W. J. (2006). Changes to ENSO under CO2 dou­ bling in a multimodel ensemble. J. Climate, 19, 4009–4027, https://doi.org/10.1175/JCLI3834.1 Moore, A. M., & R. Kleeman (1999). Stochastic forcing of ENSO by the intraseasonal oscillation. Journal of Climate, 12(5), 1199–1220. Murakami, H., G. A. Vecchi, S. Underwood, T. L. Delworth, A. T. Wittenberg, W. G. Anderson, et al. (2015). Simulation and prediction of category 4 and 5 hurricanes in the high‐resolu­ tion GFDL HiFLOR coupled climate model. J. Climate, 28, 9058–9079. doi: 10.1175/JCLI‐D‐15‐0216.1 Murtugudde, R., J. Beauchamp, C. R. McClain, M. Lewis, & A. J. Busalacchi (2002). Effects of penetrative radiation on the upper tropical ocean circulation. J. Climate, 15, 470–486. doi:10.1175/1520‐0442(2002)0152.0.CO;2 Nagura, M., K. Ando, & K. Mizuno (2008). Pausing of the ENSO cycle: A case study from 1998 to 2002. J. Climate, 21, 342–363. doi:10.1175/2007JCLI1765.1 Neale, R. B., et al. (2008). The impact of convection on ENSO: From a delayed oscillator to a series of events. J. Climate, 21, 5904–5924. doi:10.1175/2008JCLI2244.1 Neelin, J. D. (1990). A hybrid coupled general circulation model for El Niño studies. Journal of the Atmospheric Sciences, 47(5), 674–693. Neelin, J. D., & F. Jin (1993). Modes of interannual tropical ocean–atmosphere interaction—A unified view. Part II: Analytical results in the weak‐coupling limit. J. Atmos. Sci., 50,

3504–3522, https://doi.org/10.1175/1520‐0469(1993)0502.0.CO;2 Neelin, J. D., D. S. Battisti, A. C. Hirst, F.‐F. Jin, Y. Wakata, T. Yamagata, & S. E. Zebiak (1998). ENSO theory. J. Geophys. Res., 103(C7), 14261–14290. doi:10.1029/97JC03424 Neske, S., & S. McGregor (2018). Understanding the warm water volume precursor of ENSO events and its interdecadal variation. Geophysical Research Letters, 45, 1577–1585. https://doi.org/10.1002/2017GL076439 Newman, M., et al. (2011). Natural variation in ENSO flavors. Geophys. Res. Lett., 38, L14705. doi: 10.1029/2011GL047658 Newman, M., M. A. Alexander, T. R. Ault, K. M. Cobb, C. Deser, E. Di Lorenzo, et al. (2016). The Pacific decadal oscil­ lation, revisited. J. Climate, 29, 4399–4427. doi:10.1175/ JCLI‐D‐15‐0508.1 Newman, M., A. T. Wittenberg, L. Cheng, G. P. Compo, & C. A. Smith (2018). The extreme 2015/16 El Niño, in the context of historical climate variability and change. Section  4 of "Explaining extreme events of 2016 from a climate perspec­ tive." Bull. Amer. Meteor. Soc., 99(1), S16–S20. doi: 10.1175/ BAMS‐D‐17‐0116.1 Noh, Y., Y. J. Kang, T. Matsuura, & S. Iizuka (2005). Effect of the Prandtl number in the parameterization of vertical mix­ ing in an OGCM of the tropical Pacific. Geophys. Res. Lett., 32, L23609. doi:10.1029/2005GL024540. Ogata, T., S.‐P. Xie, A. Wittenberg, & D.‐Z. Sun (2013). Interdecadal amplitude modulation of El Niño/Southern Oscillation and its impacts on tropical Pacific decadal variability. J. Climate, 26, 7280–7297. doi: 10.1175/JCLI‐D‐12‐00415.1 O’Neill, B. C., C. Tebaldi, D. van Vuuren, V. Eyring, P. Fridelingstein, G. Hurtt, et  al. (2016). The Scenario Model Intercomparison Project (ScenarioMIP) for CMIP6. Geosci. Model Dev., 9, 3461–3482. Pan, X., B. Huang, & J. Shukla (2011). Sensitivity of the tropical Pacific seasonal cycle and ENSO to changes in mean state induced by a surface heat flux adjustment in CCSM3. Climate Dyn., 37, 325–341. doi:10.1007/s00382‐010‐0923‐y Penland, C., & P. D. Sardeshmukh (1995). The optimal growth of tropical sea surface temperature anomalies. J. Climate, 8, 1999–2024. Philander, S. G., T. Yamagata, & R.C. Pacanowski (1984). Unstable air‐sea interactions in the tropics. J. Atmos. Sci., 41, 604–613, https://doi.org/10.1175/1520‐0469(1984)0412.0.CO;2 Picaut, J., F. Masia, and Y. du Penhoat (1997). An advective‐ reflective conceptual model for the oscillatory nature of the ENSO. Science, 277(5326), 663–666. Planton Y., J. Vialard, E. Guilyardi, M. Lengaigne, & T. Izumo (2018). Western Pacific oceanic heat content: A better ­predictor of La Niña than of El Niño. Geophys. Res. Lett. https://doi.org/10.1029/2018gl079341 Power, S. B., et  al. (2013). Robust twenty‐first‐century projections of El Nino and related precipitation variability. Nature, 502, 541–545. doi:10.1038/nature12580 Power, S. B., F. Delage, G. Wang, I. Smith, & G. Kociuba (2017). Apparent limitations in the ability of CMIP5 climate models to simulate recent multi‐decadal change in surface tempera­ ture: Implications for global temperature projections. Climate Dyn., 49, 53–69. doi:10.1007/s00382‐016‐3326‐x

224  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Predybaylo, E., G. Stenchikov, A. T. Wittenberg, & F. Zeng (2017). Impacts of a Pinatubo‐scale volcanic eruption on ENSO. J. Geophys. Res. Atmos., 122, 925–947. doi: 10.1002/2016JD025796 Puy, M., J. Vialard, M. Lengaigne, & E. Guilyardi (2016). Modulation of equatorial Pacific westerly/easterly wind events by the Madden–Julian oscillation and convectively‐ coupled Rossby waves. Climate Dyn., 46, 2155–2178. doi:10.1007/s00382‐015‐2695‐x Puy M., J. Vialard, M. Lengaigne, E. Guilyardi, P. N. Di Nezio, A. Voldoire, et al. (2017). Influence of westerly wind events stochasticity on El Niño amplitude: The case of 2014 vs. 2015. Clim. Dyn., doi:10.1007/s00382‐017‐3938‐9 Rashid, H. A., & A. C. Hirst (2016). Investigating the mecha­ nisms of seasonal ENSO phase locking bias in the ACCESS coupled model. Clim Dyn, 46, 1075–1090. https://doi. org/10.1007/s00382‐015‐2633‐y Rashid, H. A., A. C. Hirst, & S. J. Marsland (2016). An atmo­ spheric mechanism for ENSO amplitude changes under an abrupt quadrupling of CO2 concentration in CMIP5 models. Geophys. Res. Lett., 43, 1687–1694. doi:10.1002/2015GL066768 Ray, S., A. T. Wittenberg, S. M. Griffies, & F. Zeng (2018a). Understanding the equatorial Pacific cold tongue time‐mean heat budget, Part I: Diagnostic framework. J. Climate, 31(24), 9965–9985. doi:10.1175/JCLI‐D‐18‐0152.1 Ray, S., A. T. Wittenberg, S. M. Griffies, & F. Zeng (2018b). Understanding the equatorial Pacific cold tongue time‐mean heat budget, Part II: Evaluation of the GFDL‐FLOR cou­ pled GCM. J. Climate, 31(24), 9987–10011. doi:10.1175/ JCLI‐D‐18‐0153.1 Ren, H. L., J. Zuo, F. F. Jin, & M. Stuecker (2016). ENSO and annual cycle interaction: The combination mode representa­ tion in CMIP5 models. Clim Dyn, 46, 3753. https://doi. org/10.1007/s00382‐015‐2802‐z Roberts, M. J., A. Clayton, M. Demory, J. Donners, P. L. Vidale, W. Norton, et al. (2009). Impact of resolution on the tropical Pacific circulation in a matrix of coupled models. J. Climate, 22, 2541–2556. https://doi.org/10.1175/2008JCLI2537.1 Santoso, A., H. Hendon, A. Watkins, S. Power, et  al. (2019). Dynamics and predictability of the El Niño‐Southern Oscillation: An Australian perspective on progress and chal­ lenges. Bull. Amer. Meteor. Soc., 100, 403–420. doi:10.1175/ BAMS‐D‐18‐0057.1 Schopf, P. S., & M. J. Suarez (1988). Vacillations in a coupled ocean–atmosphere system. J. Atmos. Sci., 45, 3283–3287. Schopf, P. S., & M. J. Suarez (1990). Ocean wave dynamics and the time scale of ENSO. J. Phys. Oceanogr., 20, 629–645. https://doi.org/10.1175/1520‐0485(1990)0202.0.CO;2 Small, R. J., E. Curchitser, K. Hedstrom, B. Kauffman, & W. G. Large (2015). The Benguela upwelling system: Quantifying the sensitivity to resolution and coastal wind representation in a global climate model. J. Climate, 28, 9409–9432. doi:10.1175/JCLI‐D‐15‐0192.1 Song, F., & G. J. Zhang (2016). Effects of southeastern Pacific sea surface temperature on the double‐ITCZ bias in NCAR CESM1. J. Climate, 29, 7417–7433. doi:10.1175/ JCLI‐D‐15‐0852.1 Spencer, H., R. Sutton, & J. M. Slingo (2007). El Niño in a cou­ pled climate model: Sensitivity to changes in mean state

induced by heat flux and wind stress corrections. J. Climate, 20, 2273–2298. doi:10.1175/JCLI4111.1 Stein, K., A. Timmerman, N. Schneider, F.‐F. Jin, & M. Stuecker (2014). ENSO seasonal synchronization theory. J. Climate, 27, 5285–5310. doi:10.1175/JCLI‐D‐13‐00525.1 Stevenson, S., B. Fox‐Kemper, M. Jochum, B. Rajagopalan, & S.G. Yeager (2010). ENSO model validation using wavelet probability analysis. J. Climate, 23, 5540–5547. doi:10.1175/2010JCLI3609.1 Stockdale, T. N., A. J. Busalacchi, D. E. Harrison, & R. Seager (1998). Ocean modeling for ENSO. J. Geophys. Res., 103, 14325–14355. doi:10.1029/97JC02440 Stuecker, M. F., A. Timmermann, F.‐F. Jin, S. McGregor, & H.‐L. Re (2013). A combination mode of the annual cycle and the El Niño/Southern Oscillation. Nat Geosci 6, 540–544. Stuecker, M. F., A. Timmermann, F.‐F. Jin, Y. Chikamoto, W. Zhang, A. T. Wittenberg, et  al. (2017). Revisiting ENSO/ Indian Ocean dipole phase relationships. Geophys. Res. Lett., 44(5), 2481–2492. doi:10.1002/2016GL072308 Suarez, M. J., & P. S. Schopf (1988). A delayed action oscillator for ENSO. J. Atm. Sci., 45(21), 3283–3287. Sun D. Z., Y. Yu, & T. Zhang (2009). Tropical water vapor and cloud feedbacks in climate models: A further assessment using coupled simulations. J. Climate, 22, 1287–1304. Syu, H. H., J. D. Neelin, & D. Gutzler (1995). Seasonal and interannual variability in a hybrid coupled GCM. Journal of Climate, 8(9), 2121–2143. Takahashi, K., & B. Dewitte (2016). Strong and moderate non­ linear El Niño regimes. Clim Dyn, 46, 1627. https://doi. org/10.1007/s00382‐015‐2665‐3 Takahashi, K., C. Karamperidou, and B. Dewitte (2018). A the­ oretical model of strong and moderate el Niño regimes. Clim. Dyn. doi:10.1007/s00382‐018‐4100‐z Taschetto, A. S., A. S. Gupta, N. C. Jourdain, A. Santoso, C. C. Ummenhofer, & M. H. England (2014). Cold tongue and warm pool ENSO events in CMIP5: Mean state and future projections. J. Climate, 27, 2861–2885. https://doi. org/10.1175/JCLI‐D‐13‐00437.1 Terray P., S. Masson, C. Prodhomme, M. Koll Roxy & K. P. Sooraj (2015). Impacts of Indian and Atlantic oceans on ENSO in a comprehensive modeling framework. Climate Dynamics, 46(7–8), 2507–2533. doi: 10.1007/s00382‐015‐2715‐x Thomas, M. D., & A. V. Fedorov (2017). The eastern subtrop­ ical Pacific origin of the equatorial cold bias in climate models: A Lagrangian perspective. J. Climate, 30, 5885–5900. doi:10.1175/JCLI‐D‐16‐0819.1 Thual, S., A. J. Majda, N. Chen, & S. N. Stechmann (2016). Simple stochastic model for El Niño with westerly wind bursts. Proc. Nat. Acad. Sci, 113, 10245–10250. doi:10.1073/ pnas.1612002113 Timmerman, A., & F. Jin. (2003). A nonlinear theory for El Nino bursting. J. Atmos. Sci., 60, 152–165. Timmermann, A., S.‐I. An, J.‐S. Kug, F.‐F. Jin, W. Cai, A. Capotondi, et al. (2018). El Niño‐Southern Oscillation com­ plexity. Nature, 559 (7715), 535–545. doi:10.1038/ s41586‐018‐0252‐6 Tziperman, E., L. Stone, M. A. Cane, & H. Jarosh (1994). El Niño chaos: Overlapping of resonances between the seasonal

HISTORY AND PROGRESS OF ENSO MODELING  225 cycle and the Pacific ocean–atmosphere oscillator. Science, 264, 72– 74. Vannière B., E. Guilyardi, G. Madec, F. J. Doblas‐Reyes, & S. Woolnough (2013). Using seasonal hindcasts to understand the origin of the equatorial cold tongue bias in CGCMs and its impact on ENSO. Clim. Dyn., 40, 963–981, doi: 10.1007/ s00382‐012‐1429‐6 Vannière, B., E. Guilyardi, T. Toniazzo, G. Madec, & S. Woolnough (2014). A systematic approach to identify sources of tropical SST errors in coupled models using the adjust­ ment of initialised experiments. Clim. Dyn., 43, 2261–2282. doi: 10.1007/s00382‐014‐2051‐6 van Oldenborgh, G.J. (2000). What Caused the Onset of the 1997–98 El Niño?. Mon. Wea. Rev., 128, 2601–2607, doi:10.1175/1520‐0493(2000)1282.0.CO;2 van Oldenborgh, G. J., S. Philip, & M. Collins (2005). El Niño in a changing climate: A multi‐model study. Ocean Sci., 1, 81–95. Vecchi, G. A., & A. T. Wittenberg (2010). El Niño and our future climate: Where do we stand? Wiley Interdisciplinary Reviews: Climate Change, 1, 260–270. doi: 10.1002/wcc.33 Vecchi, G. A., A. T. Wittenberg, & A. Rosati (2006). Reassessing the role of stochastic forcing in the 1997‐8 El Niño. Geophys. Res. Lett., 33, L01706. doi: 10.1029/2005GL024738 Vecchi, G. A., R. Msadek, W. Anderson, Y.‐S. Chang, T. Delworth, K. Dixon, et al. (2013). Multiyear predictions of North Atlantic hurricane frequency: Promise and limitations. J. Climate, 26, 5337–5357. doi: 10.1175/JCLI‐D‐12‐00464.1 Vecchi, G. A., T. Delworth, H. Murakami, S. D. Underwood, A. T. Wittenberg, F. Zeng, et al. (2019). Tropical cyclone sensi­ tivities to CO2 doubling: Roles of atmospheric resolution, synoptic variability and background climate changes. Climate Dyn., 53, 5999–6033. Vialard J., C. Menkes, J. P. Boulanger, E. Guilyardi, P. Delecluse, M. McPhaden & G. Madec (2001). Oceanic mechanisms driving the SST during the 1997‐98 El Niño. J. Phys. Oceanog., 31, 1649–1675. Vijayeta, A., & D. Dommenget (2018). An evaluation of ENSO dynamics in CMIP simulations in the framework of the recharge oscillator model. Clim Dyn, 51, 1753. https://doi. org/10.1007/s00382‐017‐3981‐6 Wang, C., & J. Picaut (2004). Understanding ENSO physics: A review. In C. Wang, S.‐P. Xie, and J. A. Carton (Eds.), Earth’s climate: The ocean‐atmosphere interaction (Geophysical Mono­ graph Series, Vol. 147, pp. 21–48). Washington, DC: AGU. Wang, T., & J.‐P. Miao (2018). Twentieth‐century Pacific Decadal Oscillation simulated by CMIP5 coupled models. Atmos. Ocean. Sci. Lett., 11, 94–101. doi:10.1080/16742834.2 017.1381548 Watanabe, M., & A. T. Wittenberg (2012). A method for disen­ tangling El Niño‐mean state interaction. Geophys. Res. Lett., 39, L14702. doi: 10.1029/2012GL052013 Watanabe, M., J.‐S. Kug, F.‐F. Jin, M. Collins, M. Ohba, & A. T. Wittenberg (2012). Uncertainty in the ENSO amplitude change from the past to the future. Geophys. Res. Lett., 39, L20703. doi: 10.1029/2012GL053305 Weihs, R. R., & M. A. Bourassa (2014). Modeled diurnally varying sea surface temperatures and their influence on sur­ face heat fluxes. J. Geophys. Res. Oceans, 119, 4101–4123, doi:10.1002/2013JC009489

Wieners, C. E., et al. (2017). The influence of the Indian Ocean on ENSO stability and flavor. J. Climate, 30, 2601–2620. doi:10.1175/JCLI‐D‐16‐0516.1 Wilson, S. G. (2000). How ocean vertical mixing and accumu­ lation of warm surface water influence the "sharpness" of the equatorial thermocline. J. Climate, 13, 3638–3656. doi:10.1175/1520‐0442(2000)0132.0 .CO;2 Wittenberg, A. T. (2002). ENSO response to altered climates. Ph.D. thesis, Princeton University. 475 pp. doi: 10.13140/ RG.2.1.1777.8403/1 Wittenberg, A. T. (2009). Are historical records sufficient to constrain ENSO simulations? Geophys. Res. Lett., 36, L12702. doi: 10.1029/2009GL038710 Wittenberg, A. T. (2015). Low‐frequency variations of ENSO. U.S. CLIVAR Variations, 13(1), 26–31. Wittenberg, A. T., A. Rosati, N.‐C. Lau, & J. J. Ploshay (2006). GFDL’s CM2 global coupled climate models, Part III: Tropical Pacific climate and ENSO. J. Climate, 19, 698–722. doi: 10.1175/JCLI3631.1 Wittenberg, A. T., A. Rosati, T. L. Delworth, G. A. Vecchi, & F. Zeng (2014). ENSO modulation: Is it decadally ­predictable? J. Climate, 27, 2667–2681. doi: 10.1175/ JCLI‐D‐13‐00577.1 Wittenberg, A. T., G. A. Vecchi, T. L. Delworth, A. Rosati, W. Anderson, W. F. Cooke, et al. (2018). Improved simulations of tropical Pacific annual‐mean climate in the GFDL FLOR and HiFLOR coupled GCMs. J. Adv. Model. Earth Syst.,10, 3176–3220. doi:10.1029/2018MS001372 Xie, S.‐P., C. Deser, G. A. Vecchi, J. Ma, H. Teng, & A. T. Wittenberg (2010). Global warming pattern formation: Sea surface temperature and rainfall. J. Climate, 23, 966–986. doi: 10.1175/2009JCLI3329.1 Yang, X., G. A. Vecchi, R. G. Gudgel, T. L. Delworth, S. Zhang, A. Rosati, et al. (2015). Seasonal predictability of extratrop­ ical storm tracks in GFDL’s high‐resolution climate predic­ tion model. J. Climate, 28, 3592–3611. doi: 10.1175/ JCLI‐D‐14‐00517.1 Yeh, S.‐W., J.‐S. Kug, B. Dewitte, M.‐H. Kwon, B. P. Kirtman, & F.‐F. Jin (2009). El Niño in a changing climate. Nature, 461. doi:10.1038/nature08316 Ying, J., & P. Huang (2016a). Cloud‐radiation feedback as a leading source of uncertainty in the tropical Pacific SST warming pattern in CMIP5 models. J. Climate, 29, 3867– 3881. doi:10.1175/JCLI‐D‐15‐0796.1 Ying, J., & P. Huang (2016b). The large‐scale ocean dynamical effect on uncertainty in the tropical Pacific SST warming pattern in CMIP5 models. J. Climate, 29, 8051–8065. doi:10.1175/JCLI‐D‐16‐0318.1 Yu, Y., D. Dommenget, C. Frauen, G. Wang, & S. Wales (2015). ENSO dynamics and diversity resulting from the recharge oscillator interacting with the slab ocean. Clim Dyn. doi: 10.1007/s00382‐015‐2667‐1 Zavala‐Garay, J., C. Zhang, A. M. Moore, A. T. Wittenberg, M. J. Harrison, A. Rosati, et  al. (2008). Sensitivity of hybrid ENSO models to unresolved atmospheric variability. J. Climate, 21, 3704‐3721. doi: 10.1175/2007JCLI1188.1 Zebiak, S. E., & M. A. Cane (1987). A model El Niño–Southern Oscillation. Mon. Wea. Rev., 115, 2262–2278.

226  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Zhang, R.‐H. (2016). A modulating effect of Tropical Instability Wave (TIW)‐induced surface wind feedback in a hybrid cou­ pled model of the tropical Pacific. J. Geophys. Res. Oceans, 121, 7326–7353. doi: 10.1002/2015JC011567 Zhang, R.‐H., Zebiak, S. E., Kleeman, R., & Keenlyside, N. (2003). A new intermediate coupled model for El Nino simu­ lation and prediction. Geophysical Research Letter, 30(19), 2012. https://doi.org/10.1029/2003GL018010 Zhang, R., F. Tian, A. J. Busalacchi, & X. Wang (2019). Freshwater flux and ocean chlorophyll produce nonlinear feedbacks in the tropical Pacific. J. Climate, 32, 2037–2055, https://doi.org/10.1175/JCLI‐D‐18‐0430.1 Zhang, T., & D. Sun (2014). ENSO asymmetry in CMIP5 models. J. Climate, 27, 4070–4093. doi:10.1175/JCLI‐D‐13‐00454.1

Zhang, W., F.‐F. Jin, M. F. Stuecker, A. T. Wittenberg, A. Timmermann, H.‐L. Ren, et al. (2016). Unraveling El Niño’s impact on the East Asian monsoon and Yangtze River summer flooding. Geophys. Res. Lett., 43 (21), 11375‐11382. doi: 10.1002/2016GL071190 Zheng W., P. Braconnot, E. Guilyardi, U. Merkel & Y. Yu (2008). ENSO at 6ka and 21ka from ocean‐atmosphere cou­ pled model simulations. Clim. Dyn., 30, 745–762. Zhu, J., & A. Kumar (2018). Influence of surface nudging on cli­ matological mean and ENSO feedbacks in a coupled model. Climate Dyn, 50, 571–586. doi:10.1007/s00382‐017‐3627‐8 Zhu, J., B. Huang, R.‐H. Zhang, Z.‐Z. Hu, A. Kumar, M. A. Balmaseda, et  al. (2014). Salinity anomaly as a trigger for ENSO events. Nature Sci. Rep., 4, 6821. doi: 10.1038/srep06821

10 ENSO Prediction Michelle L. L’Heureux1, Aaron F. Z. Levine2, Matthew Newman3, Catherine Ganter4, Jing‐Jia Luo5, Michael K. Tippett6, and Timothy N. Stockdale7

ABSTRACT The El Niño‐Southern Oscillation (ENSO) is a coupled ocean‐atmosphere phenomenon of variability that is a leading source of seasonal climate prediction skill across the globe. The first ENSO prediction was made in the mid‐1970s, but it was another 10–15 years before operational centers, using simple, coupled climate models, began to make routine ENSO predictions. These early forecast models were succeeded in the 1990s by more sophisticated dynamical and statistical models, which created the basis for real‐time seasonal outlooks over the globe. These models, and more recent multimodel ensembles, also inform our understanding and estimates of the predictability and prediction skill of ENSO, which varies seasonally and from decade to decade. ENSO predictability largely stems from slowly evolving oceanic conditions, with short‐term atmospheric fluctuations often limiting predictability on seasonal timescales. Despite improved models and better initializations, prediction skill remains low for forecasts passing through the boreal spring, the so‐called spring prediction barrier. Furthermore, prediction skill and predictability have varied significantly over the past couple decades. Higher skill and predictability are evident during periods of larger amplitude ENSO events (e.g., Eastern Pacific El Niño), whereas lower skill/predictability is associated with lower amplitude events (e.g., Central Pacific El Niño). These natural variations in our ability to predict ENSO, together with challenges during 2014–2016, motivate the search for understanding of how anthropogenic warming will influence seasonal ENSO prediction.

10.1. HISTORY OF ENSO FORECASTING William Quinn of Oregon State University issued the first known El Niño prediction in 1974 (McPhaden et al.  National Oceanic and Atmospheric Administration, NWS/ NCEP/Climate Prediction Center, College Park, MD, USA 2   Department of Atmospheric Sciences, University of Washington, Seattle, WA, USA 3   University of Colorado/CIRES, NOAA/ESRL Physical Sciences Division, Boulder, CO, USA 4  Australian Bureau of Meteorology, Melbourne, VIC, Australia 5  Institute for Climate and Application Research (ICAR)/ CICFEM/KLME/ILCEC, Nanjing University of Information Science and Technology, Nanjing, China 6  Department of Applied Physics and Applied Mathematics, Columbia University, NY, USA 7  European Centre for Medium-Range Weather Forecasts, Reading, UK 1

2015; Figure  10.1). Based on the persistence of the Southern Oscillation Index, he forecasted that a weak El Niño would emerge in early 1975 and cause below‐normal precipitation over Indonesia through mid‐to‐late 1975 (Quinn, 1974a, 1974b). The predicted El Niño did not develop in 1975, and a retrospective analysis offered by McPhaden et  al. (2015) indicates that an insufficient buildup of oceanic heat content prevented its development. Regardless, Quinn’s prediction provided a start for ENSO forecasting. Furthermore, this first ENSO prediction illustrates the close relation between seasonal climate prediction and ENSO, which predated a compre­ hensive understanding of ENSO itself. Sir Gilbert Thomas Walker uncovered the Southern Oscillation in an effort to predict Indian monsoon rainfall (Katz, 2002), and ancient Andean farmers used the visibility of stars in the Pleiades, which is influenced by ENSO, to determine when to plant their crops (Orlove et  al., 2000). Even

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 227

228  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE 2009-2015 1974 W.Quinn El Niño Prediction

1986

1992

2002

First ENSO Diagnostic Advisory at NOAA CPC

Experimental Long Lead Forecast Bulletin initiated

First issuance of the IRI Niño-3.4 plume of monthly updated model forecasts

ENSO Alert Systems are implemented at BOM, CPC, and ENFEN/Peru

1985

1989

1997

2005

2012

Start of Tropical Ocean-Global Atmosphere (TOGA) program

Seasonal rainfall outlooks issued at BOM based on SOI correlations

Routine ENSO outlooks issued from NOAA CPC, BOM, the WMO. ECMWF model becomes operational.

EUROSIP multimodel system becomes operational

NOAA CPC ENSO outlooks are provided probabilistically out to 9 seasons

Figure 10.1  A timeline of events in ENSO forecasting.

today, the predictability of ENSO offers a scientific basis for seasonal climate outlooks around the globe. Currently, ENSO forecasts are produced by many national meteorological and climate forecast centers around the world. We focus here on the development of operational forecasts, or the regular, timely dissemination of forecasts and data. The operational setting provides a uniquely thorough test of prediction methods, in large part because the future state of ENSO being predicted is truly unknown. The operational scenario is quite differ­ ent from evaluating model forecasts initialized on histor­ ical data when what happened is known. Real‐time forecasts test model assumptions on truly independent data and carry the risk of being exposed as too reliant on past data or based on insufficient or inaccurate theories. Operational time constraints mean that observational data have to be rapidly and accurately collected, assem­ bled, and ingested to provide the starting conditions for forecast models. Users require regular and routine dis­ semination of forecasts for decisions and planning at all times, so operational forecasts must be issued even during periods or situations when forecasts might perform poorly. Some of the earliest historical operational devel­ opments for ENSO forecasting were made in the United States (National Oceanic and Atmospheric Adminis­ tration’s Climate Prediction Center, CPC) and Australia (Bureau of Meteorology, BOM). The strong 1982–1983 El Niño and its global impacts were well underway by the time scientists noticed it. Its unexpectedly strong development led to accelerated worldwide efforts in climate observations and forecasting. Until this event, ENSO research was not advanced enough to be applied to operational climate outlooks, in part because of the nascent state of climate forecasting. CPC, created just 3 years earlier (initially as the Climate Analysis Center) to serve as NOAA’s new climate forecast agency, did not issue its first bulletin advertising El Niño until November 1982. This delay was mostly due to inad­

equate monitoring infrastructure. Satellite monitoring of sea surface temperatures was in place, but contamination by volcanic aerosols from the eruption of El Chichón resulted in sea surface temperature estimates being too cool. Ship data, on the other hand, showed strong warming in the equatorial Pacific Ocean, but these were discounted as erroneous because they were too warm compared to what was expected (Reeves & Gemmill, 2004). These data gaps motivated the establishment of the 10‐year Tropical Ocean–Global Atmosphere (TOGA) project to improve tropical Pacific Ocean monitoring and ENSO prediction (McPhaden, et al., 2010; chapter 1). In the aftermath of the 1982–1983 El Niño, which brought widespread drought to agricultural regions of Australia, Australian farmers were desperate for long‐term rainfall forecasts. This in part led to the establishment of the National Climate Centre at BOM and the dedication of resources to improve long‐term forecast skill beyond numerical weather timescales. In the wider scientific community, the dramatic nature of the 1982–1983 event spurred interest in understanding, modeling and, in time, predicting ENSO variability. At CPC, advances in tropical Pacific monitoring led to the release of the first ENSO Diagnostic Advisory in January 1986 (Figure  10.2; Reeves & Gemmill, 2004). However, this provided only diagnostic information about current ENSO conditions; it was still a research and monitoring effort, and forecasts of ENSO were still not routinely issued. Barnett et al. (1988) were among the first to show that forecasts of the 1986–1987 El Niño had skill comparable to operational 30–90 day climate out­ looks issued at that time. As a result, in June 1989 predic­ tions of ENSO began to be routinely produced at CPC as part of monthly updates of the “Climate Diagnostics Bulletin.” Initially, the forecast forum of the bulletin fea­ tured forecasts of the Niño‐3 index from a simplified cou­ pled model of Cane et  al. (1986) and a hybrid statistical‐dynamical model based on Inoue and O’Brien

Figure 10.2  The first ENSO Diagnostics Advisory issued at NOAA Climate Prediction Center in March 1986, which assessed an emerging warm episode. Courtesy of Robert Reeves.

230  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

(1984). With the advent of the TOGA program on pre­ diction (Cane & Sarachik, 1991; Battisti & Sarachick, 1995), the Experimental Long‐Lead Forecast Bulletin began in the fall of 1992 to publish forecasts and advances in seasonal prediction from authors throughout the research and operational climate communities. Many other statistical and dynamical models came on board during the 1990s, some of which became the basis for real‐time ENSO forecasts and their verifications (Barnston et  al., 1994). Ultimately, these advances in modeling driven by ENSO paved the way for operational seasonal outlooks for rainfall and temperature. At BOM, seasonal rainfall outlooks began to be rou­ tinely disseminated in 1989 for the “zero‐lead,” the period that begins in the month of issuance (for example, a July to September forecast issued in July). These three‐cate­ gory (above, near, or below‐average) outlooks for Australian precipitation were based on the ENSO state as measured by the Southern Oscillation Index (SOI). In 1996, the precipitation outlooks were joined by 6‐monthly ENSO outlooks based on an intermediate coupled model. CPC’s seasonal outlooks were also only for zero‐lead until 1995, when increased confidence in ENSO predic­ tion skill led to the extension of United States tempera­ ture and precipitation outlooks out to a year (Barnston et al., 1994). However, not until the major 1997–1998 El Niño did CPC and BOM produce stand‐alone, opera­ tional ENSO discussions issued regularly, even during neutral ENSO conditions. Prior to 1997–1998, discussion related to ENSO was embedded within the seasonal cli­ mate discussion, and advisories or media releases were only issued when conditions appeared ripe for an event. In Europe, the UK Met Office worked on ENSO mod­ eling with coupled GCMs in the 1980s (e.g. Gordon, 1989), but this did not immediately lead to a dynamical ENSO forecasting system: instead, operational seasonal outlooks were first generated by a two‐tier system using persisted SST anomalies (Graham et al., 2000). ECMWF started the development of a seasonal forecast system in 1994, and because the 1997–1998 El Niño was captured well in their experimental system (Stockdale et al., 1998), they began issuing Niño SST and seasonal forecasts on an operational schedule from December 1997. European groups subsequently collaborated on the development of multimodel GCM seasonal forecast systems with two research projects (DEMETER, ENSEMBLES), which demonstrated the power of a multimodel approach (Palmer et al., 2004; Weisheimer et al., 2009). This led to the development of the EUROSIP multimodel system, which has produced ENSO and seasonal forecasts opera­ tionally since 2005 (Stockdale et al., 2009). A coordinated multimodel GCM approach was also picked up by APCC (the APEC Climate Centre, founded in 2005) in South Korea (Min et  al., 2009) and the NMME (North

American Multi‐Model Ensemble) in North America, starting in 2011 (Kirtman et  al., 2014). More recently, EUROSIP has been superseded by the Copernicus Climate Change Service (C3S) multimodel seasonal fore­ cast system. In recent years, there has been an explosion in the number of seasonal climate outlooks produced by cou­ pled dynamical models, all relying upon an expansion of observing networks, improvements in data assimilation, and a tremendous escalation of computing power. While these models make global predictions, ENSO remains a core component of any credible seasonal forecast system. One indicator of its importance is the number of readily available forecasts of ENSO posted online each month. The International Research Institute for Climate and Society (IRI) has the longest‐running archive of ENSO model forecasts, but other national and international efforts have resulted in various model displays that are posted in real or near real‐time. ENSO outlooks often showcase forecasts of SST anomalies in the Niño index regions, spanning the west­ ernmost Niño‐4 region to the easternmost Niño‐1+2 region. Of these regions, the Niño‐3.4 index is often fea­ tured because this region has strong correlations with other atmospheric measures of ENSO and with remote climate impacts (Barnston et al., 1997). An expert team was commissioned in the mid‐2000s to catalogue El Niño and La Niña definitions used operationally by member countries of the World Meteorological Organization (Horsfall et  al., 2006). While a wide variety of indices were used, one of the most common indices was based on 3‐month average values in the Niño‐3.4 index. Thus, this index is often relied upon for designations of El Niño and La Niña, though the threshold or cutoff can vary by agency. The method or format by which national and interna­ tional organizations disseminate their ENSO outlooks varies considerably. Some national meteorological ser­ vices do not routinely write forecast discussions but instead post forecast model output onto their websites, and on occasion issue notices of particularly noteworthy ENSO conditions. Other national meteorological services dedicate resources to ENSO forecast discussions that provide qualitative and quantitative descriptions of current conditions in the tropical Pacific, its likely upcoming state (e.g. likelihood of El Niño or La Niña), and potential influences of ENSO on the regional climate (e.g. rainfall and temperature). Many of these discussions are archived online, so users can retrospectively examine previous forecast discussions to see what forecasters were considering. Since 1997, the World Meteorological Organization has also put together a discussion when ENSO conditions appear possible or have arrived. This discussion is updated roughly four times per year and

ENSO Prediction  231

represents the international consensus of the member countries that provide input to the ENSO discussion. A more recent addition to operational ENSO outlooks is the implementation of ENSO alert systems beginning between 2009 and 2015 (see table in L’Heureux et  al., 2017) for an outline of systems at CPC, BOM, and in Peru). Increasingly, forecasts quantify the probability of El Niño, ENSO‐neutral, and La Niña for upcoming sea­ sons (this began in 2012 at CPC, in conjunction with the IRI; L’Heureux et  al., 2019). And with the advent of social media, meteorological centers are increasingly leaning on these tools to alert users. The use of online videos and blogs has also increasingly gained traction as a way to expand communication platforms to those who would not otherwise read a technical ENSO discussion. 10.2. ENSO PREDICTABILITY What are the physical and dynamical mechanisms that allow ENSO to be predicted? Conversely, which factors fundamentally limit ENSO forecast accuracy and skill? These are questions about the predictability of ENSO. Their answers depend on the ENSO aspect of interest, when the forecast is made, and the forecast lead time. Generally, ocean‐based ENSO indices are more predict­ able than atmospheric ones. ENSO predictability decreases with forecast lead and is often lower for fore­ casts made during the boreal spring compared with the rest of year, a characteristic known as the spring predict­ ability barrier. Understanding ENSO predictability is crucial for evaluating whether ENSO forecasts have the right level of uncertainty. As forecast lead times increase, the range of possible outcomes becomes larger. If the models that we use to predict ENSO have too large or too small of a range, our forecasts can be underconfident or overconfident, respectively. Broadly speaking, two factors limit our ability to fore­ cast ENSO. First, small errors in our estimates of the current state of the ocean and the atmosphere (initial conditions) may grow and lead to increasingly large fore­ cast errors as we forecast further ahead. In ENSO fore­ casting, the growth of initial condition errors is amplified by the air‐sea interactions due to the strong coupling in the tropics (Zebiak & Cane, 1987). When a forecast begins, the initial conditions are our best guess of the current state of the atmosphere and ocean given the avail­ able observations. However, observations themselves have measurement errors, not every quantity can be observed, and the methods by which we include observa­ tions that have different spatial or temporal resolution than the model (data assimilation) may introduce errors. Thus, the initial conditions are imperfect. One way of accounting for the uncertainty in initial conditions is to make multiple forecasts, or an ensemble, each starting

from a slightly different initial condition (Stockdale et al., 1998). The small differences between these initial condi­ tions grow, and the differences in the final forecasts pro­ vide an indication of the uncertainty due to initial condition errors. These ensemble forecasts also contain representations of weather, which is chaotic and unpre­ dictable after a few weeks. Since these ensemble forecasts present a range of possible outcomes, each of which is consistent with our best information, it is natural to express ENSO forecasts using probabilities. The challenge for models is to produce the correct range of possible outcomes given the initial conditions and unpredict­ ability of the weather. Second, all forecast models have important physical processes that are either missing or represented imper­ fectly, often because they occur at smaller spatial and/or temporal scales than the models are capable of resolving with the available computing capacity. The resulting “model error” restricts how well models capture poten­ tially realistic outcomes and therefore how well they pre­ dict the likelihood of an ENSO event occurring. Current research aims to better represent these processes by using higher resolution simulations, increased observations, and improved parameterizations. Forecast uncertainty due to model error is harder to quantify than that due to initial condition error since it pertains to “unknown unknowns,” but past forecast performance can provide indications of model deficiencies and their impacts on forecasts. To the extent that model error can be viewed as another source of uncertainty, it can be accounted for by combining several different model ensembles into one larger “multimodel” ensemble (discussed further in the next section). The presence or absence of various ocean conditions can be important for the extended predictability of ENSO. From recharge oscillator theory, it is understood that equatorial warm water volume (WWV; typically measured as the volume of water above 20°C between 5°S and 5°N) anomaly is an important predictor of future ENSO SSTs (Jin, 1997; Meinen & McPhaden, 2000). The WWV anomaly peaked 6–9 months before the SST peak during the period of 1980–1999 and 4–6 months from 2000–2010 (McPhaden, 2012). Therefore, WWV can act as an early precursor of ENSO anomalies and, in particular, can provide some predictability beyond the spring predictability barrier (Balmaseda et  al., 1995; McPhaden, 2003; Newman et al., 2011). However, more recently it has been shown that WWV is a better predictor of La Niña events than of El Niño events (Luo et  al., 2008; Levine and McPhaden, 2016; Di Nezio et al., 2017; Santoso et al., 2017; Planton et al., 2018). Extratropical SST patterns can also impact the generation of ENSO events. The North Pacific Meridional Mode (NPMM) has been shown to lead ENSO events, (see chapter 11)

232  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

particularly during the northern hemisphere winter and spring (Vimont et al., 2001; Vimont et al., 2003). Studies have suggested that the NPMM can either be forced from extratropical winter storms and propagated equatorward via the wind‐evaporation‐SST feedback (Vimont, 2010) or by the trade wind charging process (Anderson & Perez, 2015). Similarly, equatorward propagation of anomalies in the Southern Hemisphere can also impact El Niño. The South Pacific Meridional Mode (SPMM) can signif­ icantly impact the development of El Niño events during the boreal summer and fall. Some studies have suggested that the SPMM can influence the type of El Niño event that occurs during summer and fall months, the tradi­ tional growth period of El Niño (You & Furtado, 2017), although the exact role of the SPMM still remains unclear (Larson et al., 2018). Other tropical regions also impact ENSO predictability (Cai et  al., 2019). The tropical Indian Ocean and the Indian Ocean Dipole interacts with ENSO and may con­ tribute positively to ENSO predictability at long lead times (Luo et al., 2010; Izumo et al., 2010). The Indian Ocean has been suggested as a potential cause for the recent much anticipated non–El Niño event in 2014 and subsequent extreme event in 2015 (Dong & McPhaden, 2018). The tropical Atlantic, including the “Atlantic Niño” and tropical North Atlantic SST anomaly may provide some additional predictability for ENSO (Ham et al., 2013; Keenlyside et al., 2013; Luo et al., 2017), but its role is modulated on multidecadal timescales (Martin‐ Rey et al., 2014). Atmospheric forcing in the tropical Pacific on weather timescales can have a significant influence on ENSO development and evolution (Karspeck et  al., 2006; Menkes et al., 2014). Westerly wind bursts in the western and central Pacific act to force El Niño in two different important ways. First, the westerly wind burst acts to push the warmer surface water in the western Pacific towards the central Pacific (Anderson and McCreary, 1985; Puy et al., 2016; Levine et al., 2017). This eastward extension of the warm pool appears important to the formation of weaker, Central Pacific El Niño events (Ren & Jin 2013; Chen et al., 2015). Second, the westerly wind bursts act on the oceanic subsurface, displacing the ther­ mocline and increasing the WWV (McPhaden et  al., 1998; Fedorov, 2002; Fedorov et al., 2015). This thermo­ cline displacement propagates eastward as a Kelvin wave, deepening the thermocline across the entire Pacific over the course of 2 to 3 months and helping to create the incipient conditions for stronger, eastern Pacific El Niño events. Since westerly wind bursts are weather time‐scale events, they are inherently unpredictable on the time scale of seasonal forecasts. However, given the nature of the interactions between westerly wind bursts and El Niño, predicting the exact timing of the westerly wind burst is

less essential than predicting that they will occur over the course of the season (Roulston & Neelin, 2000; Levine & Jin, 2010). Westerly wind bursts are modulated by the conditions in the tropical Pacific, for example, occurring more frequently when the warm pool is extended east­ ward than when the warm pool is in its normal position or has retreated westward (Lengaigne et  al., 2004; Eisenman et al., 2005). Although these aspects of west­ erly wind bursts reduce the deterministic predictability of ENSO, it is important that they are correctly represented in the model in order to obtain reliable probabilistic pre­ dictions of ENSO (Fedorov et  al., 2003; Levine et  al., 2016). ENSO predictability also exhibits a pronounced sea­ sonality, primarily due to two related factors. First, over the last 50 years it appears that with one exception, all ENSO events peaked during the boreal winter. Second, ENSO forecasts in both dynamical and statistical models are generally much more skillful when they are started after boreal spring than before or during it (Jin et  al., 2008). A pronounced annual cycle in the eastern tropical Pacific, including the development of the equatorial cold tongue from boreal summer to fall, drives seasonal varia­ tions in the processes responsible for equatorial Pacific SST anomaly growth (Stein et  al., 2010, 2015). SST anomalies are most strongly damped during the boreal spring and most strongly amplified during the boreal fall (Liu et al., 2019). Under these conditions, predictable sig­ nals grow weakly during the spring and can be over­ whelmed by unpredictable factors. Also, all of the previous factors that reduce the predictability of ENSO, challenges in observing the initial conditions or the impact from the forcing, are potentially important for creating the spring predictability barrier because the source of the SST anomaly is unimportant for its growth (Yu et al., 2012; Levine & McPhaden, 2015). There has also been some suggestion that the seasonal cycle of the westerly wind bursts impacts the seasonality of ENSO predictability, with the Northern Hemisphere spring hav­ ing the most westerly wind bursts, and the Northern Hemisphere summer and fall having the least (Zheng & Zhu, 2010; Lopez & Kirtman, 2014). So, our lack of ability to predict ENSO during the boreal spring is due to the time it takes the SST anomalies to grow and the likelihood of the stochastic forcing occurring during specific seasons (Thomas et al., 2018). 10.3. ENSO PREDICTION SKILL The more closely a prediction corresponds to observa­ tions, the more skillful it is. Forecast verification, the act of assessing skill, usually employs multiple skill mea­ sures, or metrics. Choosing metrics is nontrivial and depends on user interests, including what aspects of the

ENSO Prediction  233

forecast they want to evaluate. Skill can be evaluated in space or in time, it can assess a single category or multiple categories, and it can be deterministic or probabilistic. Rigorous verification also requires considering the fol­ lowing: (i) Start time: when is your forecast initiated? and (ii) Target time: what period are you forecasting? The forecast “lead time” is then the time interval between the start and target times. ENSO prediction skill is com­ monly displayed as a function of some combination of start, target, or lead times. ENSO skill is often verified using Niño indices, which measure SST anomalies averaged within different regions along or near the equatorial Pacific Ocean. Probably the most common SST index to measure skill is the Niño‐3.4 index, which is based on SST anomalies in the east‐central equatorial Pacific and is highly correlated with other var­ iables and indicators of ENSO, such as sea level pressure and rainfall over the tropical Pacific (Barnston et  al., 1997; Trenberth, 1997), and with the extratropical response to ENSO (Kumar et al. 1996; L’Heureux et al., 2015). Furthermore, unlike other variables within the tropical Pacific, a substantial investment has been made to develop several homogenous, observationally based SST climate records, which can be used for skill valida­ tion over a long time period. So, while the Niño‐3.4 index is not the only indicator of ENSO, it is commonly used to assess the quality of ENSO predictions, with many model providers distributing real‐time forecasts of Niño‐3.4 (Barnston et al., 2012). That the Niño‐3.4 index is a single time series and has a near‐Gaussian distribution also makes it relatively easy to analyze. Also, tropical Pacific SST model forecasts are typically most skillful in the Niño‐3.4 region (e.g., Jin et  al., 2008; Newman & Sardeshmukh, 2017). On the other hand, not all ENSO events have the same SST pattern, so other Niño regions and indices also need to be considered to entirely capture ENSO (Capotondi et al. 2015). Figure 10.3 shows an evaluation of Niño‐3.4 skill from the North American Multi‐Model Ensemble (based on the 1982–2018 record). Target times across the calendar year are on the x‐axis, and the lead‐time on the y‐axis. The figure illustrates how far in advance the Niño‐3.4 index can be predicted with skill above specified thresh­ olds. The different colored bars correspond to different correlation thresholds. So, if one desires at least 50% of Niño‐3.4 index variance to be predicted (r = 0.7), then ENSO can be predicted 5 months (for August–September) to 10 months (for April–June) ahead, whereas Niño‐3.4 skill in excess of r = 0.9 (81% of variance explained) is only possible at 0‐month lead for June and July targets and up to 5 months in advance for January targets. Generally, the most trustworthy assessment of skill comes from assessing the skill of predictions that are made in “real time,” or on a regular, ongoing basis, as in an oper­

ational framework. Barnston et  al. (1994) noted that “performance in a real‐time setting is the ultimate test of the utility of a long‐lead forecast.” One major advantage of verifying real‐time predictions is that forecast providers cannot pick only specific, possibly advantageous, forecast times that could otherwise result in a biased verification. To truly assess the suitability of any model for decision making and planning, it must consistently be initialized on the latest (previously unknown) observations, run forward, and evaluated for its ability to capture ENSO evolution over many forecasts. Because models are constructed with past observations already known and based on previously observed physics or statistical relationships, the unrealized future is the only truly independent scenario to objectively determine the skill of various models. On the other hand, estimating skill from real‐time forecasts can be challenging because of the relatively short record of real‐time fore­ casts, as well as changing forecast methodologies. In addition to real‐time evaluation, another approach to assess ENSO skill is through hindcasts, or reforecasts, when a model is run in forecast mode on known, past data. Akin to making a forecast run using the most recent observational data gathered within the past day or so, the model is also run on the observed conditions of each day (or certain days) across many years, going back several decades (usually to ~1980, when denser, satellite‐based observations that can initialize models first became avail­ able). In this fashion, forecast skill over a longer period can be assessed across a variety of ENSO events, reducing (but not eliminating) dependence upon sample size. However, hindcast skill may differ from real‐time pre­ dictions because of practical issues involved with making real‐time forecasts, such as missing or faulty data ingests and other computing bugs (e.g. National Weather Service, 2016). Also, hindcast skill reflects the skill of state‐of the‐ art current generation forecast models, not the skill of previous generation models, which are eventually phased out in real‐time forecasting. This abandonment of older forecast systems complicates determining whether opera­ tional ENSO skill has improved over time due to advance­ ments in forecast models and their assimilation systems. A long record of real‐time predictions, using the model systems in place at the time the prediction was made, would be preferable to answer the question of whether model skill has improved over time. However, the longest known real‐time record, the IRI/CPC set of Niño‐3.4 outlooks going back to February 2002 (International Research Institute for Climate and Society, 2002), indi­ cates that natural, decadal variations in skill cloak any apparent change due to forecast technology (Barnston et  al., 2012). For instance, despite the advancements in models through the 2000s, this decade was accompanied by overall lower prediction skill compared to the 1990s (Hu, 2000; Xue et al., 2013).

234  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Forecast Lead Time When Correlation Reaches Threshold

12

r = 0.9

11

r = 0.8 r = 0.7

10 9

Lead Time

8 7 6 5 4 3 2 1 0 Jan

Feb

Mar

Apr

Jun

May

Jul

Aug

Sep

Ocp

Nov

Dec

Target Month

Figure 10.3  The forecast lead‐time (y‐axis) when Niño‐3.4 index skill exceeds certain correlation thresholds. On the x‐axis, the target month of the forecast is presented. Model data is based on the ensemble average of ~100 members from the North American Multi‐Model Ensemble from 1982 to 2018. Departures in the Niño‐3.4 index are based on 1982–2010 monthly averages.

While the first seasonal ENSO predictions were made in the early 1980s, prediction skill was not evaluated and compared across models until the 1990s (Barnston et al., 1994). Three broad types of models have been used to predict ENSO: (i) statistical, (ii) hybrid, and (iii) dynam­ ical models. Statistical models are based on historical relationships in the observational record and use analysis techniques such as linear regression, neural networks, or other machine learning methods. Dynamical models are based on physical equations and parameterizations of the coupled ocean‐atmosphere climate system and typically require supercomputing resources. Hybrid models use a dynamical ocean model and infer the atmosphere from statistical relationships. Hybrid models are not presently used in ENSO forecasting, though they played a role in the past (Latif et al., 1998). Within the operational cen­ ters, dynamical model development has been emphasized since the 1990s, and little statistical model development has taken place since the 2000s. The benefits of this focus may have been realized with Tippett et  al. (2012), Barnston et  al. (2012), and van Oldenborgh (2005a, 2005b) showing that dynamical models have very slightly outperformed their statistical counterparts for the Niño indices during the previous decade, although the differ­ ences for Niño‐3.4 are small and its statistical significance is debatable (Newman & Sardeshmukh, 2017). One of the most notable features in ENSO skill, which continues to bedevil forecasters, is a relatively lower skill level for forecasts made through the boreal spring. The previous section discusses ideas for the springtime predic­

tion or predictability barrier. Despite years of model improvement, its endurance suggests real, fundamental intrinsic limits on skill. This minimum in forecast skill, which impacts target months during the summer and early fall, is also seen in Figure 10.3. Forecasts made in the spring suffer a subsequent reduction in skill that is most noticeable up to 5 months later (e.g. a model initial­ ized in March or April is associated with a visible reduction of skill for August–September targets). Despite the decrease in skill across the spring, state‐of‐ the‐art dynamical models tend to outperform many statistical models owing to their higher skill for spring and summer targets (van Oldenborgh, 2005a; Barnston et al. 2012; Barnston et al., 2019). Among other possibil­ ities, this performance may be due to the advanced assim­ ilation systems of dynamical models, which more rapidly ingest the latest observations, such as a recent westerly wind burst, which can be a precursor for El Niño. Most statistical models tend to be coarser, using monthly or seasonal averaged data, so they do not resolve these short‐term developments (Barnston et  al., 1999). The relatively short and uncertain observational record may also limit statistical models. When trained on multi­ centennial output from climate simulations, some statistical models have skill comparable to dynamical model skill (e.g., Chen et  al. 2016; Ding et  al. 2018). Regardless, statistical models play a valuable role in providing benchmarks for comparison, and their lower dimensionality can help provide insight on important mechanisms and sources of predictability. Further,

ENSO Prediction  235

statistical methods are often applied to dynamical model output to calibrate and combine models (e.g. Stephenson et al., 2005). A hindcast record of sufficient length allows for the identification of robust physical model errors in both fore­ cast mean and uncertainty. This allows the skill of dynam­ ical models to be considerably enhanced by bias correction, and a forecast anomaly is developed by removing the lead‐ time dependent model climatology from the total fields (Stockdale, 1997; van Oldenborgh, 2005b). Not only does this process remove mean biases, but also it removes any systematic errors in the seasonal cycle, as well as lead‐time dependent drift of the model climatology away from the observed one. In this sense, all dynamical model forecasts are “statistically corrected.” Statistical methods are occa­ sionally also used to estimate flux correction terms within the model integration itself in order to reduce certain cli­ mate model errors, such as cold tongue and double ITCZ biases (Magnusson et al., 2013). Increased skill from dynamical models is partly due to the introduction of multiple member ensembles, which capture the range of uncertainty emanating from imperfect initial conditions and the unpredictable evolution of the atmosphere. More widespread availability of supercom­ puting resources has also allowed for combinations of ensembles associated with multiple models that are each run (and bias corrected) in a similar configuration, such as the North American Multi‐Model Ensemble (NMME; Kirtman et al., 2014), European Multi‐Model Seasonal‐ to‐Interannual Prediction project (EUROSIP; Palmer et  al., 2004), and Asia‐Pacific Economic Cooperation (APEC) Multi‐Model forecasts (Min et al., 2009). The use of multiple model ensembles generally improves skill over a single model (Kirtman et  al., 2001; Jin et  al., 2008; Weisheimer et al. 2009; DelSole et al., 2014) and, perhaps more importantly, estimates of forecast uncertainty. Multimodel systems, where each model is accompanied by a long reforecast, improve the statistical correction of forecasts not only about the mean, but also in their vari­ ance. Hence, their benefit is realized through improved reliability scores and other types of probabilistic verifica­ tion metrics (Tippett & Barnston, 2008; Barnston et al., 2015; Tippett et  al., 2019). However, ENSO variability errors common to all models (e.g., a westward displace­ ment of the SST anomaly; Newman & Sardeshmukh 2017; see also chapter  9) are not as easily corrected (Ding et al. 2018). Recent efforts have been made to increase the ocean resolution and better parameterize tropical convection, which appears to reduce bias and improve variability of various aspects of the coupled ocean‐ atmosphere system (MacLachlan et  al. 2015; Stockdale et al., 2018; Johnson et al., 2019). In general, the prediction skill for other SST regions of the tropical Pacific is lower than that for Niño‐3.4 and the

east‐central Pacific (Luo et  al., 2008; Magnusson et  al., 2013; Newman & Sardeshmukh, 2017), which has impli­ cations for forecasting the continuum of SST types, or ENSO flavors. Central Pacific events tend to have SST anomalies that are relatively larger within the Niño‐4 region, while eastern Pacific events have the largest amplitude closer to South America, within the Niño‐3 and Niño‐1+2 regions. Weaker El Niños often coincide with central Pacific events (Figure 3 in Capotondi et al., 2015; Timmerman et al., 2018), which can be trickier to predict because of their lower amplitudes and generally later onset times (Imada et  al., 2015). For example, Figure 10.4 shows the February 2019 prediction from the NMME, where the ensemble mean lies near El Niño thresholds for all lead times, but the spread of ensemble members encompasses conditions ranging from La Niña to El Niño. Thus, in an operational setting, it is easy to be more confident of an El Niño forecast when most ensem­ bles are well above required event thresholds (+0.5°C or +0.8°C) than of a weaker event when the forecast uncer­ tainty significantly spans other outcomes. While fewer in number, predictions related to the stron­ gest El Niños, or eastern Pacific events, tend to underesti­ mate peak intensity for most lead times beyond a couple months. This was an issue for real‐time predictions of both the 1997–1998 El Niño (Barnston et al., 1999) and the 2015–2016 El Niño (L’Heureux et al., 2017). However, these stronger events were also more skillfully predicted further in advance (e.g. 6–9 months) of their boreal winter peaks, in part due to persistent and strong signals that grew through the preceding year (Luo et  al., 2016; Zheng & Yu, 2017). In recent years, there have been growing efforts to better resolve the different spatial fla­ vors of ENSO in prediction models because of its poten­ tial importance on regional climate variability (Hendon et  al., 2009). For example, the new operational forecast model from BOM, the ACCESS‐s1, appears to better dis­ tinguish different flavors compared to its predecessor model (Hudson et al., 2017). 10.4. DECADAL VARIATION IN ENSO AND ITS SKILL Both ENSO forecast skill and predictability have varied substantially over the past few decades (e.g., Balmeseda et  al., 1995; Barnston et  al., 2012; Newman & Sardeshmukh, 2017; Huang et  al., 2017; Ding et  al., 2019). This is illustrated in Figure 10.5, which uses a mea­ sure called pattern correlation to show how well the ensemble‐mean SST forecast for a 6‐month lead time matches observations within the ENSO region spanning the central and eastern tropical Pacific. The resulting forecast skill, smoothed to emphasize the year‐to‐year variations, is shown for an empirical‐dynamical model

236  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE NMME initialized in February 2019 3

Obs (OSTIA) Ensemble Mean

Nino3.4 index (degC)

2

1

0

–1

–2

2

3

4

5

6

7

8

9

10

11

12

1

Month (2019–20) Figure 10.4  Predictions of the Niño‐3.4 index from the North American Multi‐Model Ensemble initialized at the beginning of February 2019. Predictions are shown out to 12 months lead time or through January 2020. The black solid line is the observations based on daily OSTIA, and the purple dashed line is the ensemble average. The red dashed line is the threshold for El Niño (+0.5°C) and the blue dashed line is the threshold for La Niña (–0.5°C). Remaining lines show the individual members. Departures in the Niño‐3.4 index are based on 1982–2010 monthly averages.

called the linear inverse model (LIM; Newman & Sardeshmukh, 2017), three physical‐dynamical ensem­ bles (the NMME operational multimodel ensemble, the ECMWF SEAS5 25‐member single model ensemble, and the SINTEX‐F 9‐member single model ensemble; Luo et al., 2008), and a statistical approach (model analogs; Ding et  al., 2018, 2019) that mimics ENSO forecasts made by traditional multimodel ensembles by finding matches to observed oceanic surface anomalies within two different sets of long climate model simulations from either NMME or CMIP5 models. The LIM hindcasts are determined using a tenfold cross‐validation, and the NMME, SEAS5, and SINTEX‐F hindcasts have been bias‐corrected separately by model; model analogs are bias‐corrected by construction. Also shown is the ENSO predictability, or average expected skill, estimated from the LIM as a function of the “signal‐to‐noise” ratio bet­ ween its predicted ENSO amplitude and its forecast uncertainty (or ensemble spread), assumed to depend only upon forecast lead time. Two things are immediately clear from this figure. First, despite using fairly different modeling techniques, all three approaches (models, LIM, analogs) often have sim­ ilar year‐to‐year skill variations. Note that the LIM expected skill has similar variations, showing how pre­ dictability estimates can be used to diagnose the potential for actual forecast skill. For this definition of skill, it

might be anticipated that the model forecasts have greater expected and actual skill at times of larger forecast ampli­ tudes, and this can be seen to be true for forecast skill values during years of moderate to strong ENSO events (indicated by shading in the figure). Additionally, since the expected skill can be determined at the time of fore­ cast issuance, it could be used to indicate when confidence in the spatial structure of the forecast anomaly is expected to be relatively high. Second, while no long‐term trend in ENSO skill is apparent since 1961 (Ding et  al., 2019), there are some extended periods when forecasts appear generally more skillful than during other periods. Recently, the relatively low skill that existed for much of the first decade of the 2000s led some researchers to suggest that ENSO predict­ ability had been reduced, possibly because of some fundamental change to average tropical conditions (e.g., Barnston et al., 2012; Zhao et al., 2016). Moreover, some studies have posited that the nature of ENSO itself has changed over the last few decades, with an increased prevalence of central Pacific events possibly due to changes in the average ocean, or base state, conditions (Lee & McPhaden, 2010), including those potentially driven by climate change (Yeh et al., 2009). For example, it has been suggested that WWV and related thermocline feedbacks had weaker impacts on ENSO SST anomaly development during the 2000s, causing more central

ENSO Prediction  237 ENSO Region SST Month 6 skill

1

0.8

0.6

0.4

0.2

LIM NMME hindcasts SEAS5 hindcasts SINTEX-F hindcasts NMME-model-analog CMIP5-model-analog Expected LIM skill

0

–0.2

1965

1970

1975

1980

1985

1990

1995

2000

2005

2010

2015

Verification time

Figure 10.5 Year‐to‐year variations of model hindcast/forecast skill in the tropical Pacific ENSO region for monthly forecasts initialized during 1961–2015, for 6‐month forecast lead time. The measure of skill used is the pattern correlation between ensemble mean forecast and corresponding observed monthly SST anomalies within the region bounded by 20°S–20°N, 170°E–70°W, smoothed with a 13‐month running mean filter to emphasize annual variations. All anomalies are determined relative to a 1961–2015 monthly climatology. The light red (blue) shaded epochs indicate periods of >1 sigma warm (< −1 sigma cold) Niño‐3.4 seasonal SST anomalies. Skill is shown for the linear inverse model (LIM), the North American Multi‐Model Ensemble (NMME) operational ensemble, the ECMWF SEAS5 ensemble, the JAMSTEC SINTEX‐F ensemble, and model analogs employed on either a 4‐model subset of the NMME models or a 10‐model subset of the CMIP5 models. The mean expected skill (a predictability estimate) from the LIM is also shown. The LIM is trained from tropical sea surface analyses over the 1958–2011 period, with hindcasts determined by a tenfold cross‐validation technique. The NMME and SEAS5 hindcasts are bias corrected over the 1982–2009 period. Therefore, all forecasts after 2011 for the LIM, NMME, SEAS5, and SINTEX‐F models are independent of observations during that period. There is no training period for the model‐analog forecasts, since they are based on analogs to tropical IndoPacific sea surface observations drawn from independent multicentennial uninitialized model simulations.

Pacific events with less predictability (McPhaden, 2012; Neske & McGregor, 2018). Such a base state change on long timescales, if predictable, could lead to long‐term ENSO forecasts of decadal ENSO variability. On the other hand, some researchers have suggested that variations in both the amplitude and type of ENSO events are largely driven by noise, that is, by random and essentially unpredictable year‐to‐year weather variations in both the tropics and extratropics (e.g., Newman et al., 2011; Thomas et al., 2018). In this case, year‐to‐year var­ iations in ENSO forecast skill might also occur simply due to differing noise events, and apparent base state changes might be the residual of random ENSO events rather than a forcing of them (Flügel et al., 2004; Kumar et al., 2015; Lee et al., 2016; Newman & Sardeshmukh, 2017; Wittenberg et al., 2014). Then, ENSO might not be predictable for forecast leads extending much beyond its life cycle. For example, since the LIM is constructed so

that its predictable dynamics are unchanging over the 1961–2010 period (apart from cross‐validation, which has minor impact), its expected decadal variations of ENSO activity are solely a consequence of unpredictable decadal variations in weather noise. Likewise, the LIM expects seasonal forecast skill to increase whenever, by chance, a substantial ENSO precursor is initiated. These expectations (dotted line in Figure  10.5) were in fact largely realized by all the hindcasts in Figure 10.5, with periods of enhanced skill in the mid 1970s and late 1990s coinciding with enhanced ENSO activity, and reduced skill in the late 1970s to early 1980s and early 2000s coin­ ciding with reduced ENSO activity. Of course, decadal variations in ENSO events and pre­ dictability could result from both weather noise and base state oceanic changes (e.g., Aiken et al., 2013; Capotondi & Sardeshmukh, 2017), as well as the convolution ­between them (e.g. Levine et al., 2016), with consequent

238  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

impacts on forecast skill. Still, if changes to the base state and its related ENSO variability are predictable on decadal times cales, current physical models seem unable to capture them. At present, on average these models have little SST skill within the ENSO region for forecast leads greater than about 2 years (Newman, 2013). Some recent studies, however, have suggested that longer lead forecast skill may exist for a few isolated events (Gonzalez & Goddard, 2015), particularly some La Niñas that either occur as a result of a transition from a strong El Niño (e.g., Thomas et al., 2018) or follow a previous La Niña as part of a 2‐year La Niña event (DiNezio et al., 2017). 10.5. RECENT ENSO PREDICTION CHALLENGES Some current challenges in ENSO prediction were highlighted and illustrated by the surprising evolution of ENSO during 2014, when early indications that a strong El Niño could be brewing (Carrington et al., 2014) turned out to be a false alarm. Despite a strong downwelling oceanic Kelvin wave in the boreal spring of 2014 and cli­ mate model forecasts for El Niño, only marginally warm tropical Pacific conditions with little atmospheric cou­ pling were observed during the boreal winter 2014–2015 (McPhaden, 2015). Better understanding of the reasons for such problematic ENSO forecasts can identify obsta­ cles to improved ENSO prediction. Many obstacles are related to the physics of the climate system, but better quantifying and communicating uncertainty is also a challenge, since some forecasting systems, such as the 100‐member North American multimodel ensemble, sug­ gested the observed evolution was unlikely but still pos­ sible (Figure 10.6, top), a view supported by the EUROSIP calibrated multimodel forecast (Figure 10.6, bottom). Another challenge is forecasting surface winds over the equatorial Pacific Ocean. These winds can have a consid­ erable impact on the evolution and strength of ENSO (Menkes et  al., 2014; Hu & Fedorov, 2016; Levine & McPhaden, 2016; Takahashi & Dewitte, 2016; Wang et  al., 2011), and the unpredicted and sudden onset of easterly wind anomalies during July 2014 was one of the reasons for reduced oceanic warmth during the rest of the year. Unfortunately, prediction errors in tropical winds arise quickly and are large by day 5 (Martin et al., 2010). However, some aspects of these winds may arise from low‐ frequency SST variability and may be more predictable (Eisenman et al., 2005; Tziperman & Yu, 2007; Levine & Jin, 2017; Capotondi et al., 2018; Ineson et al., 2018). The extent to which ENSO forecast uncertainty is due to limi­ tations in forecasting the surface winds over the equatorial Pacific Ocean remains an open question. The extent of the role of other ocean basins and the off‐ equatorial Pacific Ocean also needs to be clarified. For in­ stance, conditions in 2014 across the Indian Ocean (Dong

& McPhaden, 2018) and North and South Pacific Oceans (Larson & Kirtman, 2015; Min et  al., 2015; Zhu et  al., 2016; Wang & Hendon, 2017) have been labeled possible culprits for the aborted El Niño. Nonequatorial Pacific anomalies can be quite critical for the development of ENSO in general (Vimont et al., 2001; Chang et al., 2007; Di Lorenzo et al., 2010; Wang et al., 2012; Wen et al., 2014; Boschat et al., 2013; Keenlyside et al., 2013; Zhang et al., 2014; Luo et  al., 2017; Pegion & Selman, 2017; You & Furtado, 2017), but it is another task to translate this knowledge into improved predictions (Larson & Kirtman, 2014; Larson et al., 2018). So the question remains whether current climate models appropriately capture and predict these extratropical‐tropical linkages. While 2014–2015 was a recent example of a challenging ENSO forecast, prediction and characterization of marginal El Niño events are always particularly difficult. Since ENSO is a complex, coupled ocean‐atmosphere system that is measured using multiple regions and vari­ ables, those indicators are not necessarily all in alignment during borderline events. In 2014 some national meteoro­ logical services issued watches or alerts, indicating that conditions were favorable for the onset of El Niño later in the year. From the standpoint of SSTs in the Niño‐3.4 region, conditions later in the year were consistent with El Niño, but several forecast centers (e.g., NOAA, BOM) never declared the onset of El Niño due to the lack of significant atmospheric coupling over the equatorial Pacific (L’Heureux, 2015; Santoso et al., 2019). Therefore, while a major emphasis of real‐time model forecasts is on SSTs, there is no guarantee of a concomitant response in other indicators that may be important for impacts. Currently, operational predictions of atmospheric‐based ENSO indicators, such as the Southern Oscillation Index (SOI) and Outgoing Longwave Radiation (OLR), are not commonplace and are likely of lower skill than SST indices. Figure 10.7 shows log skill scores from the Climate Forecast System (version 2) for three different atmosphere‐ based ENSO indices, in addition to the Niño‐3.4 SST index. In this particular model, predictions of the Niño‐3.4 SST index are clearly more skillful than the atmospheric indices, though the Equatorial SOI appears to provide skill beyond the two OLR‐based indices. Finally, with the planet warming in association with anthropogenic climate change, SST values have begun to include warming trends not necessarily related to ENSO (Solomon & Newman 2012). For example, the 2015– 2016 El Niño appears to be record breaking based on Niño‐3.4 SST anomalies, but if detrended then it was not as warm and suggests nonlinear feedbacks were not trig­ gered in the eastern Pacific to the same extent as previous El Niño events (Santoso et  al. 2017; L’Heureux et  al., 2017). Newman et al. (2018) pointed out that SST anom­ alies in the westernmost Niño‐4 region were unprece­

ENSO Prediction  239 NMME initialized in April 2014

3

Obs (OISST)

Nino3.4 index (degC)

Ensemble Mean 2

1

0 4

5

6

7

8

9

10

11

12

1

2

3

Month (2014–15) NINO3.4 SST calibrated pdf EUROSIP multi-model forecast from 1 Apr 2014

Anomaly (deg C)

ECMWF, Met Office, Météo-France, NCEP Percentiles at 2% 10%, 25%, 50%, 75% 90% and 98%

2

2

1

1

0

0

–1

–1

Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Seg Oct Nov Dec 2014 2013

Figure 10.6  (top panel) Predictions of the Niño‐3.4 index from the North American Multi‐Model Ensemble initialized at the beginning of April 2014. Predictions are shown out to 12 months lead time or through March 2015. The black solid line is the observations based on daily OISST, and the purple dashed line is the ensemble average. Remaining lines show the individual members. Departures in the Niño‐3.4 index are based on 1982–2010 monthly averages. (bottom panel) Predictions of the Niño‐3.4 index from the EUROSIP multimodel system initialized on 1 April 2019, according to the operational calibrated product. Various percentiles of the distribution are shown, based on a multimodel combination of variance‐corrected ensemble forecasts, with an additional calibration of the forecast uncertainty to match historic forecast error.

dented in the observational record and were unlikely to have occurred without anthropogenic warming. How­ ever, it is not clear if or how this warming is associated with a change in ENSO variability itself, which is an important consideration given continued uncertainty in

the projections of future El Niño variability in the east­ ern Pacific Ocean (e.g., Stevenson et al. 2017; Cai et al., 2018). Thus, an emerging challenge in real‐time fore­ casting is to untangle El Niño dynamics from long‐term climate trends.

240  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Lead

8 6

0.5

4

0

2

–0.5

0 J

F M A M J

J A S O N D

CFSv2-C EPOLR

1

8

Lead

CFSv2-C Nino3.4

6

0.5

4

0

2

–0.5

0

–1

J

6

0.5

4

0

2

–0.5

0 J

F M A M J

J A S O N D

J A S O N D

CFSv2-C CPOLR

1

–1

Target month

8

Lead

Lead

8

F M A M J

–1

Target month

Target month CFSv2-C EQSOI

1

1

6

0.5

4

0

2

–0.5

0 J

F M A M J

J A S O N D

–1

Target month

Figure 10.7  Log skill score (LSS) from hindcasts (1981–2010) of the NCEP Climate Forecast System (CFS) version 2. Skill is presented as a function of target month (x‐axis) and lead time (y‐axis). The top left panel shows the skill of the Niño‐3.4 SST index region (5°S–5°N, 170°W–120°W), the bottom left panel shows the skill of the Equatorial Southern Oscillation Index (EQSOI; surface pressure differences based 5°S–5°N, 80°W–130°W minus 5°S–5°N, 90°E–140°E), the top right panel shows the skill of the eastern Pacific outgoing longwave radiation (EPOLR) index (5°S–5°N, 160°W–‐110°W; Chiodi & Harrison, 2013), and the bottom right panel shows the skill of the central Pacific OLR index (5°S–5°N, 170°E–140°W; L’Heureux et al., 2015). Positive values of LSS show forecasts that are better than a climatological forecast, and negative values of LSS show where forecasts are worse than climatology.

10.6. CONCLUDING REMARKS Our understanding of ENSO has matured, and with it forecast systems have improved in their representation of nature and handling of observations. Now skillful predic­ tions of ENSO are routinely made by operational centers around the world. The scientific and institutional invest­ ments that have improved ENSO forecasting have also enabled seasonal predictions of other climate anomalies, which provide advance guidance to a wide range of decision makers and users. While the infrastructure to predict ENSO and communicate its impacts has signifi­ cantly advanced since the 1980s, the community now has greater appreciation for variation in skill over the span of decades and tremendous diversity of ENSO types, which are not equally well predicted.

This chapter has focused on the history of ENSO pre­ diction, both its skill and its potential predictability, and highlighted some continuing prediction challenges that were illustrated by the 2014–2016 evolution of ENSO. It is clear that the upcoming decades will challenge the pre­ diction communities as anthropogenic climate change becomes more prominent, even while its impact on ENSO remains debated. Significant changes in the nature of ENSO, such as in its amplitude or frequency, would almost certainly impact our ability to predict it, and in ways we may not expect. Climate change could also have a substantial impact on the tools that forecasters rely upon, which critically assume a long (~30 year) stationary record, whether as a training period for statistical models or to bias correct dynamical models. Significant changes in the statistics of ENSO, not accounted for in these past

ENSO Prediction  241

records, would greatly increase the uncertainty of ENSO predictions. Fortunately, or unfortunately, forecasters and users making decisions off these forecasts will lead the pack in experiencing the consequences of ENSO in a warming climate. ACKNOWLEDGMENTS We especially thank these individuals for their time in helping us with the History of ENSO section: Tony Barnston (NOAA/IRI, retired), Grant Beard (BOM, retired), Mike Halpert (NOAA), Vernon Kousky (NOAA, retired), Neville Nicholls (formerly BOM, currently Monash University), Robert Reeves (NOAA, retired), and Chet Ropelewski (NOAA/IRI, retired). REFERENCES Aiken, C. M., Santoso, A., McGregor, S., & England, M. H. (2013). The 1970’s shift in ENSO dynamics: A linear inverse model perspective. Geophys. Res. Lett., 40(8), 1612–1617. doi: 10.1002/grl.50264 Anderson, D. L. T., & McCreary, J. P. (1985). Slowly Propagating Disturbances in a Coupled Ocean‐Atmosphere Model. J.  Atmos. Sci., 42(6), 615‐630. doi: 10 .1175/1520‐0469 (1985)042h0615:SPDIACi2.0.CO;2 Balmaseda, M. A., Davey, M. K., & Anderson, D. L. T. (1995). Decadal and Seasonal Dependence of ENSO Prediction Skill. J. Climate, 8(11), 2705‐2715. doi: 10.1175/1520‐0442 (1995)008h2705:DASDOEi2.0.CO;2 Barnett, T., Graham, N., Cane, M., Zebiak, S., Dolan, S., O’Brien, J., & Legler, D. (1988). On the Prediction of the El Niño of 1986‐1987. Science, 241(4862), 192–196. doi: 10.1126/science.241.4862.192 Barnston, A. G., Chelliah, M., & Goldenberg, S. B. (1997). Documentation of a highly ENSO‐related SST region in the equatorial pacific: Research note. Atmos.‐Ocean, 35(3), 367‐383. doi: 10.1080/07055900.1997.9649597 Barnston, A. G., Glantz, M. H., & He, Y. (1999). Predictive skill of statistical and dynamical climate models in sst forecasts during the 1997‐98 el niño episode and the 1998 la niña onset. Bull. Amer. Meteor. Soc., 80(2), 217‐244. doi: 10.1175/ 1520‐0477(1999)080h0217:PSOSADi2.0.CO;2 Barnston, A. G., Tippett, M. K., L’Heureux, M. L., Li, S., & DeWitt, D. G. (2012). Skill of real‐time seasonal ENSO model predictions during 2002‐11: Is our capability increasing? Bull. Amer. Meteor. Soc., 93(5), 631‐651. doi: 10.1175/BAMS‐D‐ 11‐00111.1 Barnston, A. G., Tippett, M. K., Ranganathan, M., & L’Heureux, M. L. (2019). Deterministic skill of enso predic­ tions from the north american multimodel ensemble. Clim Dyn., 53, 7215–7234. doi: 10.1007/s00382‐017‐3603‐3 Barnston, A. G., Tippett, M. K., van den Dool, H. M., & Unger, D. A. (2015). Toward an Improved Multimodel ENSO Prediction. J. Appl. Meteor. Climatol., 54(7), 1579‐1595. doi: 10.1175/JAMC‐D‐14‐0188.1

Barnston, A. G., van den Dool, H. M., Zebiak, S. E., Barnett, T. P., Ji, M., Rodenhuis, D. R., … Livezey, R. E. (1994). Long‐Lead Seasonal ForecastsWhere Do We Stand? Bull. Amer. Meteor. Soc., 75(11), 2097‐2114. doi: 10.1175/1520‐ 0477(1994)075h2097:LLSFDWi2.0.CO;2 Battisti, D. S., & Sarachik, E. S. (1995). Understanding and pre­ dicting ENSO. Rev. Geophys., 33(S2), 1367‐1376. doi: 10.1029/95RG00933 Boschat, G., Terray, P., & Masson, S. (2013). Extratropical forc­ ing of ENSO. Geophys. Res. Lett., 40(8), 1605‐1611. doi: 10.1002/grl.50229 Cai, W., Wang, G., Dewitte, B., Wu, L., Santoso, A., Takahashi, K., … McPhaden, M. J. (2018). Increased variability of east­ ern pacific el niño under greenhouse warming. Nature, 564(7735), 201–206. doi: 10.1038/s41586‐018‐0776‐9 Cai, W., Wu, L., Lengaigne, M., Li, T., McGregor, S., Kug, J.‐S., … Chang, P. (2019). Pantropical climate interactions. Science, 363(6430). doi: 10.1126/science.aav4236 Cane, M., & Sarachik, E. (1991). Prospectus: A TOGA Program on Seasonal to Interannual Prediction. UCAR Productions, NOAA Climate and Global Change Program Special Report, no. 4, 1‐46. Cane, M. A., Zebiak, S. E., & Dolan, S. C. (1986). Experimental forecasts of el niño. Nature, 321, 827 EP ‐. Retrieved from https://doi.org/10.1038/321827a0 Capotondi, A., & Sardeshmukh, P. D. (2017). Is El Niño really changing? Geophys. Res. Lett., 44(16), 8548–8556. doi: 10.1002/2017GL074515 Capotondi, A., Sardeshmukh, P. D., & Ricciardulli, L. (2018). The Nature of the Stochastic Wind Forcing of ENSO. J. Climate, 31(19), 8081‐8099. doi: 10.1175/JCLI‐D‐17‐0842.1 Capotondi, A., Wittenberg, A. T., Newman, M., Lorenzo, E. D., Yu, J.‐Y., Braconnot, P., … Yeh, S.‐W. (2015). Understanding ENSO diversity. Bull. Amer. Meteor. Soc., 96(6), 921‐938. doi: 10.1175/BAMS‐D‐13‐00117.1 Carrington, D., Godenberg, S., & Redfearn, G. (2014). How El Niño will change the world’s weather in 2014. The Guardian. Chang, P., Zhang, L., Saravanan, R., Vimont, D. J., Chiang, J. C. H., Ji, L., … Tippett, M. K. (2007). Pacific meridional mode and El Niño‐Southern Oscillation. Geophys. Res. Lett., 34(16). doi: 10.1029/2007GL030302 Chen, C., Cane, M. A., Henderson, N., Lee, D. E., Chapman, D., Kondrashov, D., & Chekroun, M. D. (2016). Diversity, Nonlinearity, Seasonality, and Memory Effect in ENSO Simulation and Prediction Using Empirical Model Reduction. J. Climate,29(5), 1809‐1830. doi: 10.1175/ JCLI‐D‐15‐0372.1 Chen, D., Lian, T., Fu, C., Cane, M. A., Tang, Y., Murtugudde, R., … Zhou, L. (2015). Strong influence of westerly wind bursts on El Niño diversity. Nature Geosci, 8, 339–345, doi: 10.1038/ngeo2399 Chiodi, A. M. and D. E. Harrison, (2013) El Niño Impacts on Seasonal U.S. Atmospheric Circulation, Temperature, and Pre­cipita­tion Anomalies: The OLR-Event Perspective. J. Climate, 26, 822–837, https://doi.org/10.1175/JCLI-D12-00097.1 DelSole, T., Nattala, J., & Tippett, M. K. (2014). Skill improvement from increased ensemble size and model diversity. Geophys. Res. Lett., 41(20), 7331‐7342. doi: 10.1002/2014GL060133

242  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Di Lorenzo, E., Cobb, K. M., Furtado, J. C., Schneider, N., Anderson, B. T., Bracco, A., … Vimont, D. J. (2010). Central Pacific El Niño and decadal climate change in the North Pacific Ocean. Nature Geosci, 3, 762–765, doi: 10.1038/ngeo984 DiNezio, P. N., Deser, C., Okumura, Y., & Karspeck, A. (2017). Predictability of 2‐year La Niña events in a coupled general circulation model. Climate Dyn., 49(11), 4237–4261. doi: 10.1007/s00382‐017‐3575‐3 Ding, H., Newman, M., Alexander, M. A., & Wittenberg, A. T. (2018). Skillful Climate Forecasts of the Tropical Indo‐Pacific Ocean Using Model‐Analogs. J. Climate, 31(14), 5437‐5459. doi: 10.1175/JCLI‐D‐17‐0661.1 Dong, L., & McPhaden, M. J. (2018). Unusually warm Indian Ocean sea surface temperatures help to arrest development of El Niño in 2014. Sci. Rep., 8(1), 2249. doi: 10.1038/ s41598‐018‐20294‐4 Eisenman, I., Yu, L., & Tziperman, E. (2005). Westerly Wind Bursts: ENSO’s Tail Rather than the Dog? J. Climate, 18(24), 5224‐5238. doi: 10.1175/JCLI3588.1 Fedorov, A. V. (2002). The response of the coupled tropical ocean?atmosphere to westerly wind bursts. Quart. J. Roy. Meteor. Soc., 128(579), 1‐23. doi: 10.1002/qj.200212857901 Fedorov, A. V., Harper, S. L., Philander, S. G., Winter, B., & Wittenberg, A. (2003). How Predictable is El Niño? Bull. Amer. Meteor. Soc., 84(7), 911920. doi: 10.1175/ BAMS‐84‐7‐911 Fedorov, A. V., Hu, S., Lengaigne, M., & Guilyardi, E. (2015). The impact of westerly wind bursts and ocean initial state on the development, and diversity of El Niño events. Climate Dyn., 44(5), 1381–1401. doi: 10.1007/s00382‐014‐2126‐4 Flügel, M., Chang, P., & Penland, C. (2004). The Role of Stochastic Forcing in Modulating ENSO Predictability. J. Climate, 17(16), 3125‐3140. doi: 10.1175/ 1520‐0442(2004)017h3125:TROSFIi2.0.CO;2 Gonzalez, P. L. M., & Goddard, L. (2016). Long‐lead ENSO predictability from CMIP5 decadal hindcasts. Climate Dyn., 46(9), 3127–3147. doi: 10.1007/ s00382‐015‐2757‐0 Gordon, C., Neelin, J. D., Charnock, H., & Philander, S. G. H. (1989). Tropicalocean‐atmosphere interactions in a coupled model. Philos. Trans. Royal Soc. A, 329(1604), 207‐223. doi: 10.1098/rsta.1989.0071 Graham, R. J., Evans, A. D. L., Mylne, K. R., Harrison, M. S. J., & Robertson, K. B. (2000). An assessment of seasonal pre­ dictability using atmospheric gen‐eral circulation models. Quart. J. Roy. Meteor. Soc., 126(567), 2211‐2240. doi: 10.1002/qj.49712656712 Ham, Y.‐G., Kug, J.‐S., & Park, J.‐Y. (2013). Two distinct roles of Atlantic SSTs in ENSO variability: North Tropical Atlantic SST and Atlantic Niño. Geophys. Res. Lett., 40(15), 4012‐4017. doi: 10.1002/grl.50729 Hendon, H. H., Lim, E., Wang, G., Alves, O., & Hudson, D. (2018). Prospects for predicting two flavors of el niño. Geophys. Res. Lett., 36(19). doi: 10.1029/ 2009GL040100 Horsfall, F. (2006). Catalogue of Indices and Definitions of El Niño and La Niña in Operational Use by WMO Members. World Meteorological Organization Commission for Climatology. Hu, S., & Fedorov, A. V. (2016). Exceptionally strong easterly wind burst stalling El Niño of 2014. Proceedings of the

National Academy of Sciences, 113(8), 2005–2010. doi: 10.1073/pnas.1514182113 Hu, Z., A. Kumar, B. Huang, J. Zhu, M. L’Heureux, M.J. McPhaden, and J. Yu, (2020). The Interdecadal Shift of ENSO Properties in 1999/2000: A Review. J. Climate, 33, 4441–4462. https://doi.org/10.1175/JCLI-D-19-0316.1. Huang, B., Shin, C.‐S., Shukla, J., Marx, L., Balmaseda, M. A., Halder, S., … Kinter, J. L. (2017). Reforecasting the ENSO Events in the Past 57 Years (1958–2014). J. Climate, 30(19), 7669–7693. doi: 10.1175/JCLI-D-16-0642.1 Hudson, D. (2017). ACCESS‐S1: The new Bureau of Meteorology multi‐week to seasonal prediction system. J So Hemisph Earth, 67(3), 132‐159. Imada, Y., Tatebe, H., Ishii, M., Chikamoto, Y., Mori, M., Arai, M., … Kimoto, M. (2015). Predictability of Two Types of El Nio Assessed Using an Extended Seasonal Prediction System by MIROC. Mon. Wea. Rev., 143(11), 4597‐4617. doi: 10.1175/MWR‐D‐15‐0007.1 Ineson, S., Balmaseda, M. A., Davey, M. K., Decremer, D., Dunstone, N. J., Gordon, M., … Weisheimer, A. (2018). Predicting El Niño in 2014 and 2015. Scientific Reports, 8(1), 10733. doi: 10.1038/s41598‐018‐29130‐1 Inoue, M., & O’Brien, J. J. (1984). A Forecasting Model for the Onset of a Major El Niño. Mon. Wea. Rev., 112(11), 2326‐2337. doi: 10.1175/1520‐0493(1984) 112h2326:AFM FTOi2.0.CO;2 International Research Institute for Climate and Society. (2002). IRI/CPC Plume Forecasts. Retrieved 2019‐03‐25, from https://iri.columbia.edu/~forecast/ensofcst/Data/enso forecast data.txt Izumo, T., Vialard, J., Lengaigne, M., de Boyer Montegut, C., Behera, S. K., Luo, J.‐J., … Yamagata, T. (2010). Influence of the state of the Indian Ocean Dipole on the following year’s El Niño. Nature Geosci, 3, 168–172, doi: 10.1038/ ngeo760 Jin, E. K., Kinter, J. L., Wang, B., Park, C.‐K., Kang, I.‐S., Kirtman, B. P., … Yamagata, T. (2008). Current status of ENSO prediction skill in coupled ocean–atmosphere models. Climate Dyn., 31(6), 647–664. doi: 10.1007/s00382‐008‐ 0397‐3 Jin, F.‐F. (1997). An Equatorial Ocean Recharge Paradigm for ENSO. Part I: Conceptual Model. J. Atmos. Sci., 54(7), 811– 829. doi: 10.1175/1520 ‐0469(1997)054h0811:AEORPFi2.0 .CO;2 Karspeck, A. R., Kaplan, A., & Cane, M. A. (2006). Predictability Loss in an Intermediate ENSO Model due to Initial Error and Atmospheric Noise. J. Climate, 19(15), 3572–3588. doi: 10.1175/JCLI3818.1 Katz, R. W. (2002). Sir Gilbert Walker and a Connection bet­ ween El Niño and Statistics. Stat. Sci., 17(1), 97–112. doi: 10.1214/ss/1023799000 Keenlyside, N. S., Ding, H., & Latif, M. (2013). Potential of equatorial Atlantic variability to enhance El Niño prediction. Geophys. Res. Lett., 40(10), 2278‐2283. doi: 10.1002/grl. 50362 Kirtman, B. P., Min, D., Infanti, J. M., James L. Kinter, I., Paolino, D. A., Zhang, Q., … Wood, E. F. (2014). The North American Multimodel Ensemble: Phase‐1 Seasonal‐to‐ Interannual Prediction; Phase‐2 toward Developing

ENSO Prediction  243 Intraseasonal Prediction. Bull. Amer. Meteor. Soc., 95(4), 585‐601. doi: 10.1175/BAMS‐D‐12‐00050.1 Kirtman, B. P., Shukla, J., Balmaseda, M., Graham, N., Penland, C., Xue, Y., & Zebiak, S. (2001). Current Status of ENSO Forecast Skill. International CLIVAR Publication Series(56), 1‐26. Kumar, A., Chen, M., Xue, Y., & Behringer, D. (2015). An Analysis of the Temporal Evolution of ENSO Prediction Skill in the Context of the Equatorial Pacific Ocean Observing System. Mon. Wea. Rev., 143(8), 3204–3213. doi: 10.1175/ MWR‐D‐15‐0035.1 Kumar, A., Hoerling, M., Ji, M., Leetmaa, A., & Sardeshmukh, P. (1996). Assessing a GCM’s Suitability for Making Seasonal Predictions. J. Climate, 9(1), 115129. doi: 10.1175/1520‐ 0442(1996)009h0115:AAGSFMi2.0.CO;2 Larson, S. M., & Kirtman, B. P. (2014). The Pacific Meridional Mode as an ENSO Precursor and Predictor in the North American Multimodel Ensemble. J. Climate, 27(18), 7018‐7032. doi: 10.1175/JCLI‐D‐14‐00055.1 Larson, S. M., & Kirtman, B. P. (2015). An alternate approach to ensemble ENSO forecast spread: Application to the 2014 forecast. Geophys. Res. Lett., 42(21), 9411‐9415. doi: 10.1002/ 2015GL066173 Larson, S. M., Pegion, K. V., & Kirtman, B. P. (2018). The South Pacific Meridional Mode as a Thermally Driven Source of ENSO Amplitude Modulation and Uncertainty. J. Climate, 31(13), 5127‐5145. doi: 10.1175/JCLI‐D‐17‐0722.1 Latif, M., Anderson, D., Barnett, T., Cane, M., Kleeman, R., Leetmaa, A., … Schneider, E. (1998). A review of the predict­ ability and prediction of ENSO. J. Geophys. Res., 103(C7), 14375‐14393. doi: 10.1029/97JC03413 Lee, J.‐W., Yeh, S.‐W., & Jo, H.‐S. (2016). Weather noise leading to El Niño diversity in an ocean general circulation model. Climate Dyn. doi: 10.1007/s00382 ‐016‐3438‐3 Lee, T., & McPhaden, M. J. (2010). Increasing intensity of El Niño in the central‐equatorial Pacific. Geophys. Res. Lett., 37(14). doi: 10.1029/ 2010GL044007 Lengaigne, M., Guilyardi, E., Boulanger, J.‐P., Menkes, C., Delecluse, P., Inness, P., … Slingo, J. (2004). Triggering of El Niño by westerly wind events in a coupled general circulation model. Climate Dyn., 23(6), 601–620. doi: 10.1007/ s00382‐004‐0457‐2 Levine, A., Jin, F. F., & McPhaden, M. J. (2016). Extreme Noise‐Extreme El Niño: How State‐Dependent Noise Forcing Creates El Niño‐La Niña Asymmetry. J. Climate, 29(15), 5483‐5499. doi: 10.1175/JCLI‐D‐16‐0091.1 Levine, A. F. Z., & Jin, F.‐F. (2010). Noise‐Induced Instability in the ENSO Recharge Oscillator. J. Atmos. Sci., 67(2), 529– 542. doi: 10.1175/ 2009JAS3213.1 Levine, A. F. Z., & Jin, F. F. (2017). A simple approach to quanti­ fying the noise‐ENSO interaction. Part I: deducing the state‐ dependency of the windstress forcing using monthly mean data. Climate Dyn., 48(1), 1–18. doi: 10.1007/s00382‐015‐2748‐1 Levine, A. F. Z., & McPhaden, M. J. (2015). The annual cycle in ENSO growth rate as a cause of the spring predictability barrier. Geophys. Res. Lett., 42(12), 5034–5041. doi: 10.1002/ 2015GL064309 Levine, A. F. Z., & McPhaden, M. J. (2016). How the July 2014 easterly wind burst gave the 2015‐2016 El Niño a head start.

Geophys. Res. Lett., 43(12), 6503‐6510. doi: 10.1002/2016 GL069204 L’Heureux, M. (2015). An overview of the El Niño‐Southern Oscillation (ENSO) since 2014. National Weather Service Climate Prediction S&T Digest. doi: 10.7289/V5RN35VW L’Heureux, M. L., Takahashi, K., Watkins, A. B., Barnston, A. G., Becker, E. J., Di Liberto, T. E., … Wittenberg, A. T. (2017). Observing and Predicting the 2015/16 El Niño. Bull. Amer. Meteor. Soc., 98(7), 1363‐1382. doi: 10.1175/BAMS‐ D‐16‐0009.1 L’Heureux, M. L., Tippett, M. K., & Barnston, A. G. (2015). Characterizing ENSO coupled variability and its impact on North American seasonal precipitation and temperature. J. Climate, 28(10), 4231‐4245. doi: 10.1175/JCLI‐D‐14‐ 00508.1 L’Heureux, M. L., Tippett, M. K., Takahashi, K., Barnston, A. G., Becker, E. J., Bell, G. D., … Wang, W. (2019). Strength Outlooks for the El Niño‐Southern Oscillation. Wea. Forecasting, 34(1), 165‐175. doi: 10.1175/WAF‐D‐ 18‐0126.1 Liu, Z., Jin, Y., & Rong, X. (2019). A Theory for the Seasonal Predictability Barrier: Threshold, Timing, and Intensity. J. Climate, 32(2), 423‐443. doi: 10 .1175/JCLI‐D‐18‐0383.1 Lopez, H., & Kirtman, B. P. (2014). WWBs, ENSO  predict­ ability, the spring barrier and extreme events. J.  Geophys. Res., 119(17), 10,114–10,138. Retrieved from https://doi. org/10.1002/2014JD021908 doi: 10.1002/2014JD021908 Luo, J.‐J., Liu, G., Hendon, H., Alves, O., & Yamagata, T. (2017). Inter‐basin sources for two‐year predictability of the multi‐year La Niña event in 2010–2012. Sci. Rep., 7(1), 2276. doi: 10.1038/s41598‐017‐01479‐9 Luo, J.‐J., Masson, S., Behera, S. K., & Yamagata, T. (2008). Extended ENSO Predictions Using a Fully Coupled Ocean‐ Atmosphere Model. J. Climate, 21(1), 84‐93. doi: 10.1175/2007JCLI1412.1 Luo, J.‐J., Yuan, C., Sasaki, W., Behera, S. K., Masumoto, Y., Yamagata, T., … Masson, S. (2016). Current status of intrase­ asonal-seasonal-to interannual prediction of the Indo-Pacific climate. In Indo‐Pacific Climate Variability and Predictability (p. 63‐107). doi: 10.1142/9789814696623 0003 Luo, J.‐J., Zhang, R., Behera, S. K., Masumoto, Y., Jin, F.‐F., Lukas, R., & Yamagata, T. (2010). Interaction between El Niño and Extreme Indian Ocean Dipole. J. Climate, 23(3), 726–742. doi: 10.1175/2009JCLI3104.1 MacLachlan, C., Arribas, A., Peterson, K. A., Maidens, A., Fereday, D., Scaife, A. A., … Madec, G. (2015). Global Seasonal forecast system version 5 (GloSea5): a high‐resolu­ tion seasonal forecast system. Quart. J. Roy. Meteor. Soc., 141(689), 1072‐1084. doi: 10.1002/qj.2396 Magnusson, L., Alonso‐Balmaseda, M., Corti, S., Molteni, F., & Stockdale, T. (2013, Nov 01). Evaluation of forecast strat­ egies for seasonal and decadal forecasts in presence of systematic model errors. Climate Dyn., 41(9), 2393–2409. doi: 10 .1007/s00382‐012‐1599‐2 Martin, G. M., Milton, S. F., Senior, C. A., Brooks, M. E., Ineson, S., Reichler, T., & Kim, J. (2010). Analysis and Reduction of Systematic Errors through a Seamless Approach to Modeling Weather and Climate. J. Climate, 23(22), 5933‐5957. doi: 10.1175/2010JCLI3541.1

244  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Mart´ın‐Rey, M., Rodr´ıguez Fonseca, B., Polo, I., & Kucharski, F. (2014). On the Atlantic–Pacific Niños connection: a multi­ decadal modulated mode. Climate Dyn., 43(11), 3163–3178. doi: 10.1007/s00382‐014‐2305‐3 McPhaden, M. J. (2003). Tropical Pacific Ocean heat content variations and ENSO persistence barriers. Geophys. Res. Lett., 30(9). doi: 10.1029/ 2003GL016872 McPhaden, M. J. (2012). A 21st century shift in the relationship between enso sst and warm water volume anomalies. Geophys. Res. Lett., 39(9). doi: 10.1029/2012GL051826 McPhaden, M. J. (2015). Playing hide and seek with El Niñ ño. Nature Clim. Change, 5(9), 791–795. doi: 10.1038/ nclimate2775 McPhaden, M. J., Busalacchi, A. J., & Anderson, D. L. (2010). A TOGA Retrospective. Oceanography, 23. Retrieved from https://doi.org/10.5670/oceanog.2010.26 McPhaden, M. J., Busalacchi, A. J., Cheney, R., Donguy, J.‐R., Gage, K. S., Halpern, D., … Takeuchi, K. (1998). The Tropical Ocean‐Global Atmosphere observing system: A decade of progress. J. Geophys. Res., 103(C7), 14169‐14240. doi: 10.1029/97JC02906 McPhaden, M. J., Timmermann, A., Widlansky, M. J., Balmaseda, M. A., & Stockdale, T. N. (2015). The Curious Case of the El Niño That Never Happened: A Perspective from 40 Years of Progress in Climate Research and Forecasting. Bull. Amer. Meteor. Soc., 96(10), 1647‐1665. doi: 10.1175/BAMS‐D‐14‐00089.1 Meinen, C. S., & McPhaden, M. J. (2000). Observations of Warm Water Volume Changes in the Equatorial Pacific and Their Relationship to El Niño and La Niña. J. Climate, 13(20), 3551–3559. doi: 10.1175/1520‐0442(2000)013h3551: OOWWVCi2.0.CO;2 Menkes, C. E., Lengaigne, M., Vialard, J., Puy, M., Marchesiello, P., Cravatte, S., & Cambon, G. (2014). About the role of Westerly Wind Events in the possible development of an El Niño in 2014. Geophys. Res. Lett., 41(18), 6476‐6483. doi: 10.1002/2014GL061186 Min, Q., Su, J., Zhang, R., & Rong, X. (2015). What hindered the El Nio pattern in 2014? Geophys. Res. Lett., 42(16), 6762‐6770. doi: 10.1002/2015GL064899 Min, Y.‐M., Kryjov, V. N., & Park, C.‐K. (2009). A Probabilistic Multimodel Ensemble Approach to Seasonal Prediction. Wea. Forecasting, 24(3), 812‐828. doi: 10.1175/2008WAF 2222140.1 National Weather Service. (2016). Technical Implementation Notice 16‐09. Retrieved 2019‐03‐25, from http://www.nws. noaa.gov/os/notification/ tin16‐09cfs.htm Neske, S., & McGregor, S. (2018). Understanding the Warm Water Volume Precursor of ENSO Events and its Interdecadal Variation. Geophys. Res. Lett., 45(3), 1577–1585. doi: 10.1002/2017GL076439 Newman, M. (2013). An Empirical Benchmark for Decadal Forecasts of Global Surface Temperature Anomalies. J. Climate, 26(14), 5260–5269. doi: 10.1175/ JCLI‐D‐12‐00590.1 Newman, M., & Sardeshmukh, P. D. (2017). Are we near the predictability limit of tropical Indo‐Pacific sea surface tem­

peratures? Geophys. Res. Lett., 44(16), 8520‐8529. doi: 10.1002/2017GL074088 Newman, M., Shin, S.‐I., & Alexander, M. A. (2011). Natural variation in ENSO flavors. Geophys. Res. Lett., 38(14). doi: 10.1029/2011GL047658 Newman, M., Wittenberg, A. T., Cheng, L., Compo, G. P., & Smith, C. A. (2018). The Extreme 2015/16 El Niño, in the Context of Historical Climate Variability and Change. Bull. Amer. Meteor. Soc., 99(1), S16‐S20. doi: 10.1175/ BAMS‐D‐17‐0116.1 Orlove, B. S., Chiang, J. C. H., & Cane, M. A. (2000). Forecasting Andean rainfall and crop yield from the influence of El Niño on Pleiades visibility. Nature, 403(6765), 68–71. doi: 10.1038/47456 Palmer, T. N., Doblas‐Reyes, F. J., Hagedorn, R., Alessandri, A., Gualdi, S., Andersen, U., … Thomson, M. C. (2004). Development of a European multimodel ensemble system for seasonal‐to‐interannual prediction (DEMETER). Bull. Amer. Meteor. Soc., 85(6), 853‐872. doi: 10.1175/ BAMS‐85‐6‐853 Pegion, K. V., & Selman, C. (2017). Extratropical precursors of the el niño‐southern oscillation. In Climate extremes (p. 299‐314). American Geophysical Union (AGU). doi: 10.1002/9781119068020.ch18 Planton, Y., Vialard, J., Guilyardi, E., Lengaigne, M., & Izumo, T. (2018). Western Pacific Oceanic Heat Content: A Better Predictor of La Niña Than of El Niño. Geophys. Res. Lett., 45(18), 9824‐9833. doi: 10.1029/2018GL079341 Puy, M., Vialard, J., Lengaigne, M., & Guilyardi, E. (2016). Modulation of equatorial Pacific westerly/easterly wind events by the Madden–Julian oscillation and convectively‐ coupled Rossby waves. Climate Dyn., 46(7), 2155–2178. doi: 10.1007/s00382‐015‐2695‐x Quinn, W. H. (1974a). Monitoring and Predicting El Niño Invasions. J. Appl. Meteor. Climatol., 13(7), 825‐830. doi: 10. 1175/1520‐0450(1974)013h0825: MAPENIi2.0.CO;2 Quinn, W. H. (1974b). Outlook for El Niño‐like conditions in 1975. NORPAX Highlights, 2(6), 2‐3. Reeves, R. W., & Gemmill, D. J. (2004). Reflections on 25 Years of Analysis, Diagnosis, and Prediction, 1979‐2004. U.S. Government Printing Office. Ren, H.‐L., & Jin, F.‐F. (2013). Recharge Oscillator Mechanisms in Two Types of ENSO. J. Climate, 26(17), 6506–6523. doi: 10.1175/JCLI‐D‐12 ‐00601.1 Roulston, M. S., & Neelin, J. D. (2000). The response of an ENSO Model to climate noise, weather noise and intrasea­ sonal forcing. Geophys. Res. Lett., 27(22), 3723–3726. doi: 10.1029/2000GL011941 Santoso, A., et al, (2019). Dynamics and Predictability of El Niño–Southern Oscillation: An Australian Perspective on Progress and Challenges. Bull. Amer. Meteor. Soc., 100, 403–420, https://doi.org/10.1175/BAMS-D-18-0057.1 Santoso, A., Mcphaden, M. J., & Cai, W. (2017). The Defining Characteristics of ENSO Extremes and the Strong 2015/2016 El Niño. Rev. Geophys., 55(4), 1079‐1129. doi: 10.1002/2017RG000560 Stein, K., Schneider, N., Timmermann, A., & Jin, F.‐F. (2010). Seasonal Synchronization of ENSO Events in a Linear

ENSO Prediction  245 Stochastic Model. J. Climate, 23(21), 5629–5643. doi: 10.1175/2010JCLI3292.1 Stein, K., Timmermann, A., Schneider, N., Jin, F.‐F., & Stuecker, M. F. (2014). ENSO Seasonal Synchronization Theory. J. Climate, 27(14), 5285–5310. doi: 10.1175/ JCLI‐D‐13‐00525.1 Stephenson, D., Coelho, C., Doblas‐Reyes, F., & Balmaseda, M. (2005). Forecast assimilation: a unified framework for the combination of multi‐model weather and climate predic­ tions. TELLUS A, 57(3), 253‐264. doi: 10.3402/tellusa. v57i3.14664 Stevenson, S., Capotondi, A., Fasullo, J., & Otto‐Bliesner, B. (2017). Forced changes to twentieth century ENSO diversity in a last Millennium context. Climate Dyn.. doi: 10.1007/ s00382‐017‐3573‐5 Stockdale, T., Doblas‐Reyes, F., & Ferranti, L. (2009). EUROSIP: multi‐model seasonal forecasting [Meteorology]. ECMWF Newsletter, 10‐16. Retrieved from https://www. ecmwf.int/node/17491 doi: 10.21957/7wc0nybvir Stockdale, T. N., Anderson, D. L. T., Alves, J. O. S., & Balmaseda, M. A. (1998). Global seasonal rainfall forecasts using a coupled ocean–atmosphere model. Nature, 392(6674), 370–373. Retrieved from https://doi.org/10.1038/ 32861 doi: 10.1038/32861 Takahashi, K., & Dewitte, B. (2016). Strong and moderate non­ linear El Niño regimes. Climate Dyn., 46(5), 1627–1645. doi: 10.1007/s00382‐015‐2665 ‐3 Thomas, E. E., Vimont, D. J., Newman, M., Penland, C., & Mart´ınez‐Villalobos, C. (2018). The Role of Stochastic Forcing in Generating ENSO Diversity. J. Climate, 31(22), 9125–9150. doi: 10.1175/JCLI‐D‐17‐0582.1 Timmermann, A., An, S.‐I., Kug, J.‐S., Jin, F.‐F., Cai, W., Capotondi, A., … Zhang, X. (2018). El niño–southern oscil­ lation complexity. Nature, 559(7715), 535–545. doi: 10.1038/ s41586‐018‐0252‐6 Tippett, M. K., & Barnston, A. G. (2008). Skill of Multimodel ENSO Probability Forecasts. Mon. Wea. Rev., 136(10), 3933‐3946. doi: 10.1175/2008MWR2431.1 Tippett, M. K., Barnston, A. G., & Li, S. (2012). Performance of Recent Multimodel ENSO Forecasts. J. Appl. Meteor. Climatol., 51(3), 637‐654. doi: 10.1175/JAMC‐D‐11‐093.1 Tippett, M. K., Ranganathan, M., L’Heureux, M., Barnston, A. G., & DelSole, T. (2019). Assessing probabilistic predic­ tions of ENSO phase and intensity from the North American Multimodel Ensemble. Climate Dyn., 53, 7497–7518. doi: 10.1007/s00382‐017‐3721‐y Trenberth, K. E. (1997). The Definition of El Niño. Bull. Amer. Meteor. Soc., 78(12), 2771‐2778. doi: 10.1175/1520‐0477 (1997)078h2771:TDOENOi2.0.CO; 2 Tziperman, E., & Yu, L. (2007). Quantifying the Dependence of Westerly Wind Bursts on the Large‐Scale Tropical Pacific SST. J. Climate, 20(12), 2760‐2768. doi: 10.1175/JCLI4138a.1 van Oldenborgh, G. J., Balmaseda, M. A., Ferranti, L., Stockdale, T. N., & Anderson, D. L. T. (2005a). Did the ECMWF Seasonal Forecast Model Outperform Statistical ENSO Forecast Models over the Last 15 Years? J. Climate, 18(16), 3240‐3249. doi: 10.1175/JCLI3420.1

van Oldenborgh, G. J., Balmaseda, M. A., Ferranti, L., Stockdale, T. N., & Anderson, D. L. T. (2005b). Evaluation of Atmospheric Fields from the ECMWF Seasonal Forecasts over a 15‐Year Period. Journal of Climate, 18(16), 32503269. doi: 10.1175/JCLI3421.1 Vimont, D. J. (2010). Transient Growth of Thermodynamically Coupled Variations in the Tropics under an Equatorially Symmetric Mean State.J. Climate, 23(21), 5771–5789. doi: 10.1175/2010JCLI3532.1 Vimont, D. J., Battisti, D. S., & Hirst, A. C. (2001). Footprinting: A seasonal connection between the tropics and mid‐latitudes. Geophys. Res. Lett., 28(20), 3923‐3926. doi: 10.1029/­ 2001GL013435 Vimont, D. J., Battisti, D. S., & Hirst, A. C. (2003). The Seasonal Footprinting Mechanism in the CSIRO General Circulation Models.J. Climate, 16(16), 2653–2667. doi: 10.1175/1520‐044 2(2003)016h2653: TSFMITi2.0.CO;2 Vimont, D. J., Wallace, J. M., & Battisti, D. S. (2003). The Seasonal Footprinting Mechanism in the Pacific: Implications for ENSO. J. Climate, 16(16), 2668–2675. doi: 10.1175/ 1520‐0442(2003)016h2668:TSFMITi2.0.CO;2 Wang, G., & Hendon, H. H. (2017). Why 2015 was a strong El Niño and 2014 was not. Geophys. Res. Lett., 44(16), 8567‐8575. doi: 10.1002/2017GL074244 Wang, S.‐Y., L’Heureux, M., & Chia, H.‐H. (2012). ENSO pre­ diction one year in advance using western North Pacific sea surface temperatures. Geophys. Res. Lett., 39(5). doi: 10.1029/2012GL050909 Wang, W., Chen, M., Kumar, A., & Xue, Y. (2011). How impor­ tant is intraseasonal surface wind variability to real‐time ENSO prediction? Geophys. Res. Lett., 38 L13705. doi:10.1029/2011GL047684. Weisheimer, A., Doblas‐Reyes, F. J., Palmer, T. N., Alessandri, A., Arribas, A., Dqu, M., … Rogel, P. (2009). ENSEMBLES: A new multi‐model ensemble for seasonal‐to‐annual predic­ tions‐ Skill and progress beyond DEMETER in forecasting tropical Pacific SSTs. Geophys. Res. Lett., 36(21). doi: 10.1029/2009GL040896 Wen, C., A., Kumar, Y., Xue, & M.J., McPhaden. (2014). Changes in Tropical Pacific Thermocline Depth and Their Relationship to ENSO after 1999. J. Climate., 27, 7230–7249, https://doi.org/10.1175/JCLI-D-13-00518.1 Wittenberg, A. T., Rosati, A., Delworth, T. L., Vecchi, G. A., & Zeng, F. (2014, 2018/11/14). ENSO Modulation: Is It Decadally Predictable? J. Climate, 27(7), 2667–2681. doi: 10.1175/JCLI‐D‐13‐00577.1 Xue, Y., Chen, M., Kumar, A., Hu, Z.‐Z., & Wang, W. (2013). Prediction Skill and Bias of Tropical Pacific Sea Surface Temperatures in the NCEP Climate Forecast System Version 2. J. Climate, 26(15), 5358‐5378. doi: 10.1175/JCLI‐D‐12‐00600.1 Yeh, S.‐W., Kug, J.‐S., Dewitte, B., Kwon, M.‐H., Kirtman, B. P., & Jin, F.‐F. (2009, 09 24). El Niño in a changing climate. Nature, 461, 511–514, doi: 10.1038/nature08316 You, Y., & Furtado, J. C. (2017). The role of South Pacific atmospheric variability in the development of different types of ENSO. Geophys. Res. Lett., 44(14), 7438‐7446. doi: 10.1002/2017GL073475

246  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Yu, Y., Mu, M., & Duan, W. (2012). Does model parameter error cause a significant ?spring predictability barrier? for el nio events in the zebiak?cane model? J. Climate, 25(4), 1263‐1277. doi: 10.1175/2011JCLI4022.1 Zebiak, S. E., & Cane, M. A. (1987). A Model El Niño?Southern Oscillation. Mon. Wea. Rev., 115(10), 2262‐2278. doi: 10.117 5/1520‐0493(1987)115h2262: AMENOi2.0.CO;2 Zhang, H., Clement, A., & Di Nezio, P. (2014). The South Pacific Meridional Mode: A Mechanism for ENSO‐like Variability. J. Climate, 27(2), 769‐783. doi: 10 .1175/JCLI‐D‐ 13‐00082.1 Zhao, M., Hendon, H. H., Alves, O., Liu, G., & Wang, G. (2016). Weakened Eastern Pacific El Niño Predictability in the Early

Twenty‐First Century. J. Climate, 29(18), 6805–6822. doi: 10.1175/JCLI‐D‐15‐0876.1 Zheng, F., & Yu, J.‐Y. (2017). Contrasting the skills and biases of deterministic predictions for the two types of El Niño. Advances in Atmospheric Sciences, 34(12), 1395–1403. doi: 10.1007/s00376‐017‐6324‐y Zheng, F., & Zhu, J. (2010). Coupled assimilation for an inter­ mediated coupled ENSO prediction model. Ocean Dynam., 60(5), 1061–1073. doi: 10.1007/ s10236‐010‐0307‐1 Zhu, J., Kumar, A., Huang, B., Balmaseda, M. A., Hu, Z.‐Z., Marx, L., & Kinter III, J. L. (2016). The role of off‐equatorial surface temperature anomalies in the 2014 El Niño predic­ tion. Sci. Rep., 6, doi: 10.1038/srep19677

Section V Remote and External Forcing

11 ENSO Remote Forcing: Influence of Climate Variability Outside the Tropical Pacific Jong‐Seong Kug1, Jerome Vialard2, Yoo‐Geun Ham3, Jin‐Yi Yu4, and Matthieu Lengaigne2,5

ABSTRACT Climate variabilities in the Pacific, Indian and Atlantic Oceans are tightly connected. The influence of El Niño– Southern Oscillation (ENSO) on the Atlantic and Indian oceans has been documented for long. There are recent lines of evidence that regions outside the tropical Pacific feed back onto ENSO characteristics, such as its amplitude, periodicity, and time‐sequence and spatial patterns, suggesting that basin interactions play a significant role for ENSO diversity and complexity. The climate variability that may influence ENSO includes the Pacific Meridional Mode, the Indian Ocean basin mode, the Indian Ocean Dipole, the Atlantic Niño and surface ­temperature variations in the North Tropical Atlantic, and the western hemisphere warm pool. The tendency of these climate modes to lead ENSO variability by several seasons could in particular provide an opportunity for improved long‐lead predictions of ENSO. This chapter will provide a comprehensive review of our current understanding of the influence of climate variability outside the tropical Pacific on ENSO.

11.1. INTRODUCTION The El Niño–Southern Oscillation (ENSO) is the dom­ inant mode of Earth’s climate variability on interannual timescales (chapters 1 and 2). ENSO is a climate mode that emerges from internal dynamics of the ocean‐ atmosphere coupled system in the tropical Pacific (chap­ ters 6–8). Positive El Niño sea surface temperature anomalies (SSTA) usually appear in spring and amplify through the Bjerknes feedback (Bjerknes, 1969), which is  Division of Environmental Science & Engineering, Pohang University of Science and Technology (POSTECH), Pohang, Republic of Korea 2  LOCEAN, IPSL, Sorbonne Université, CNRS-IRD-MNHN, Paris, France 3   Department of Oceanography, Chonnam National University, Gwangju, Republic of Korea 4   Department of Earth System Science, University of California at Irvine, Irvine, CA, USA 5  MARBEC, University of Montpellier, CNRS, IFREMER, IRD, Sete, France 1

a positive air‐sea feedback loop in the tropical Pacific. This positive feedback mechanism is offset by several negative feedbacks, including the instantaneous negative feedback from air to sea fluxes (e.g. Lloyd et al., 2009), nonlinear interactions of convective anomalies with the seasonal cycle (e.g. Lengaigne et  al., 2006) and delayed negative feedbacks from oceanic dynamics (e.g. Suarez & Schopf, 1988; Jin, 1997). This eventually leads to an El Niño event that peaks towards the end of the calendar year and then decays rapidly. While rooted in the tropical Pacific, ENSO influences climate and weather phenomena worldwide through atmospheric and oceanic teleconnections (chapters 14 and 15). Within the tropics, ENSO teleconnections can be conceptualized as zonal shifts of the Walker Circulation. For example, the eastward shift of the Walker Circulation during an El Niño induces a warming over the entire Indian Ocean (e.g. Klein et al., 1999; Xie et al., 2009) and the North Tropical Atlantic (e.g. Enfield & Mayer, 1997; Huang, 2004) in response to atmospheric subsidence. The wind speed and latent heat flux anomalies associated with

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 249

250  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

the circulation responses to convective anomalies also contribute to generating remote SSTA. In addition to those tropical teleconnections, El Niño–induced convec­ tive anomalies in the central Pacific also induce a stationary Rossby wave response that extends into the subtropics and midlatitudes (Hoskins & Karoly, 1981). Such a response alters the occurrence probability of extra­ tropical weather patterns such as the Pacific North American and North Atlantic Oscillation (e.g. Trenberth & Hurrell, 1994; Alexander et al., 2002), with impacts on the underlying SST through changes in air‐sea fluxes and Ekman currents (e.g. Alexander & Scott, 2008; Deser et al., 2010). In addition to the above studies focusing on the influence of ENSO on the other basins, it was suggested early that the Indian Ocean and Pacific basin interannual variability are interactive and better understood as an integrated climate mode designated as the Tropospheric Biennial Oscillation (Barnett, 1983; Meehl, 1987; Meehl et al., 2003). Some studies also discovered precursor sig­ nals to ENSO in other basins, such as the Indian Ocean (e.g. Clarke & Van Gorder, 2003; Kug et al., 2005) or in the North Pacific (Vimont et  al., 2001, 2003a, 2003b). There has been recently a growing number of studies sup­ porting potential influences of other oceanic basins on ENSO. The regions that may influence ENSO through teleconnections include the north (e.g. Vimont et  al., 2001, 2003a, 2003b, 2009) and south (Zhang et al., 2014) extratropical Pacific, the Southern Ocean (White et  al., 2002; White & Annis, 2004; Terray, 2011; Boschat et al., 2013), the equatorial (e.g. Rodríguez‐Fonseca et al., 2009; Martin‐Rey et al., 2012; Ding et al., 2012) and northern subtropical (e.g. Ham et  al., 2013a) Atlantic, and the tropical Indian Ocean (e.g. Kug & Kang, 2006; Ohba & Ueda, 2007; Luo et al., 2010; Izumo et al., 2010). These studies suggest that SSTA in a given region outside the tropical Pacific can induce wind changes over the equatorial Pacific. These wind changes induce an equatorial Pacific SSTA response, which can further be amplified by the Bjerknes feedback and interfere with the ENSO cycle. In the tropics, where high ambient SSTs favor an impact of SSTA on deep atmospheric convection (e.g. Gadgil et al., 1984), these wind changes in the Pacific are usually explained by a steady atmospheric response, i.e. Gill‐type response, to anomalous convective forcing in response to SST anomalies. Known modes of interannual SST vari­ ability in the other tropical oceans are thus potentially able to influence the El Niño evolution through this mechanism, which leads to wind changes over the tropical Pacific. The modes of tropical SST variability that have been most studied as precursors of ENSO (and will be briefly described later in the chapter) more specifically include the North Atlantic warming/cooling (Ham et al.,

2013a), Atlantic Niño (Zebiak, 1993), Indian Ocean basinwide warming/cooling, and Indian Ocean Dipole (Saji et al., 1999). Unlike in the tropics, midlatitudes SSTAs hardly trigger deep atmospheric convective anomalies, so that different mechanisms must operate to trigger remote wind anom­ alies in the equatorial Pacific. For instance, Vimont et al. (2001, 2003a) emphasized the potential influence of the North Pacific Oscillation, one of the dominant internal atmospheric modes in the North Pacific, on ENSO through the so‐called “footprinting mechanism.” In this mechanism, midlatitude Pacific stochastic atmospheric fluctuations drive SSTAs in winter through latent heat fluxes (see e.g. Chiang & Vimont, 2004). Off‐equatorial air‐sea interactions favor the propagation of these SSTAs into the equatorial Pacific by the following boreal spring and summer (Vimont et  al., 2003a). Once reaching the tropical region, those SSTAs can trigger convective anom­ alies and equatorial zonal wind anomalies, which influence ENSO development (Alexander et  al., 2010). A similar influence of the Southern Hemisphere subtropical Pacific has also been proposed (e.g. Zhang et al., 2014). In addition to acting as an ENSO precursor, SSTA var­ iations in other basins or in the extratropical Pacific may alter ENSO characteristics such as its magnitude, period­ icity, diversity, and predictability (Timmermann et  al., 2018; Cai et al., 2019). For instance, climate model exper­ iments where the Atlantic or the Indian Ocean are decou­ pled from the Pacific suggest that the interannual variability in these two basins damps ENSO and shortens its periodicity (e.g. Dommenget  al., 2006; Terray et  al., 2015). While many studies have suggested that SSTA pat­ terns in various regions may influence ENSO, the relative importance of each region, the detailed mechanisms through which this influence operates, and their conse­ quences for ENSO predictability are still unclear (e.g. Dayan et  al., 2014, 2015). There have been significant scientific advances in our understanding of the two‐way interactions between the tropical Pacific and other basins in recent years (Cai et  al., 2019), emphasizing that the influence of regions outside the tropical Pacific on the Pacific ENSO system is more vigorous than previously thought. The potential consequences of the influence of other basins on the El Niño phase transition (e.g. Kug & Kang 2006; Ohba & Ueda, 2007), diversity (Ham et al., 2013b; Capotondi et al., 2015; Dommenget & Yu, 2017; Timmermann et al., 2018), and for predicting El Niño and La Niña events at long leads (Park et al., 2018) thus call for a better understanding of these remote influences. The purpose of this chapter is to provide a comprehensive review of our current understanding of the influence of var­ ious regions on ENSO. In particular, we review the influence of the Indian Ocean in section 11.2, the Atlantic Ocean in section 11.3, and the extratropical Pacific in section 11.4. In

ENSO REMOTE FORCING  251

s­ection 11.5, we will summarize and compare the influence of various regions, and a further research direction will be discussed. 11.2. INDIAN OCEAN Despite the evident geographical separation between the Indian and the Pacific Oceans by the maritime conti­ nent, they share the Indo‐Pacific warm pool, the biggest span of SST in excess of 27.5°C, a necessary condition for deep atmospheric convection to develop (Gadgil et al., 1984; Graham & Barnett, 1987). This Indo‐Pacific warm pool hence maintains an intense deep atmospheric convective activity whose midtropospheric heating drives the ascending branch of the Walker Circulation and bridges these two oceans together. In addition to this, the western Pacific and the eastern equatorial Indian Ocean are connected through a gateway of narrow deep sills in the Indonesian Archipelago called the Indonesian Throughflow. The Indian and Pacific oceans mutually interact through these atmospheric and oceanic bridges, with potential influences on ENSO in the Pacific. The tropical Indian Ocean has two dominant modes of variability at the interannual timescale, both influenced by ENSO but also thought to influence ENSO: the Indian Ocean Basin Mode and the Indian Ocean Dipole. The Indian Ocean Basin Mode (IOBM) is the leading mode of Indian Ocean interannual variability (~40% of the total variance of interannual SST anomalies) and is asso­ ciated with a uniform warming of the entire Indian Ocean (Figure  11.1a). The IOBM is mostly explained by a delayed response to the zonal shifts in the Walker Circulation associated with the ENSO cycle (lag correla­ tion with ENSO > 0.8 up to ~7 months after the ENSO peak; Figure 11.1c; Klein et al., 1999; Lau & Nath, 2003; chapter 14). During El Niños, the eastward shift of the Walker Cell induces subsidence over the Indian Ocean, reducing cloudiness and inducing anticyclonic anomalies, which contribute to increasing SST through both enhanced downward solar and reduced latent upward heat fluxes (Klein et  al., 1999; Lau & Nath, 2003; Tokinaga & Tanimoto, 2004). The anticyclonic wind anomalies also force downwelling oceanic waves in the southern Indian Ocean, which also contribute to the sea surface warming in the southwestern Indian Ocean, the “thermocline ridge” region (Xie et  al., 2002; Huang & Kinter, 2002; Vialard et al., 2009). The IOBM exhibits a clear amplitude asymmetry, with a larger basinwide warming than the corresponding cooling (Hong et  al., 2010). It peaks in boreal spring, one season after the ENSO peak (Figure 11.1c), because local air‐sea interac­ tions maintain Indian Ocean SST anomalies beyond the end of the El Niño event, through boreal spring and summer. The southwestern Indian Ocean warming

indeed forces ­antisymmetric wind anomalies that weaken the summer monsoon flow, hence reducing latent heat losses and maintaining the warming through summer, in particular in the northern Indian Ocean (Wu et al., 2008; Du et al., 2009; Xie et al., 2009). Several studies have proposed that the IOBM affects the Pacific ENSO by modulating western Pacific wind anomalies (Kug & Kang 2006; Kug et al., 2006; Ohba & Ueda 2007, 2009, Okumura et  al., 2011). This western Pacific wind response to the IOBM can either be isolated statistically from observations (Figure 11.2c and e.g. Kug & Kang, 2006; Dayan et al., 2015; Izumo et al., 2016) or from atmospheric model simulations forced by anoma­ lous warming in the Indian Ocean (Figure 11.2d and e.g. Annamalai et  al., 2005; Kug & Kang, 2006; Ohba & Ueda, 2007; Dayan et  al., 2015). The Indian Ocean warming leads to enhanced convection, which influences the western North Pacific anticyclone anomaly (Watanabe & Jin, 2002; Kug & Kang, 2006; Xie et al., 2009, 2016) through atmospheric Kelvin waves, yielding easterly wind anomalies in the equatorial western Pacific in boreal winter (Figure  11.2c,d). Even though a controversial argument still exists on the seasonal dependency of the IOBM’s effect on the western Pacific easterlies (Chen et al., 2016), these boreal winter western Pacific easterly anomalies last until the following summer, favoring a fast transition from El Niño to La Niña via the Bjerknes feedback in the Pacific basin (Kug & Kang, 2006; Kug et al., 2006; Ohba & Ueda, 2007, 2009, Okumura et al., 2011). Both observations (Kug & Kang, 2006) and model simulations (Okumura et  al., 2011; Ohba & Watanabe, 2012) suggested that this IOBM feedback is larger for the warm phase than the cold phase; this asymmetry hence potentially contributes to the more systematic phase transition from El Niño to La Niña and a shorter dura­ tion of El Niños (e.g. Ohba & Ueda 2009; Okumura et al., 2011; Ohba & Watanabe, 2012). This feature could be attributable to the asymmetric IOBM amplitude and the zonal extension of the IOBM‐induced wind anom­ alies in the western Pacific (Ohba & Watanabe, 2012). The Indian Ocean also hosts a second prominent mode of interannual variability: the Indian Ocean Dipole (IOD; Reverdin et  al., 1986; Saji et  al., 1999; Webster et al., 1999; Murtugudde et al., 2000). Positive IOD events are characterized by cold sea surface anomalies near Java  and Sumatra and weaker and broader warm sur­ face  anomalies in the western tropical Indian Ocean (Figure 11.1b). IOD events tend to peak in boreal fall and to decay rapidly during the following winter (e.g. Saji et  al., 1999; Figure  11.1c). Similar to the Bjerknes feedback, positive air‐sea interaction also operates dur­ ing IOD events, giving rise to wind anomalies in the central equatorial Indian Ocean (Figure 11.1b), which in turn enhance the SST anomalies.

252  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a) EOF#1 (37%): IOBM

0.5°C

–0.5°C

10°N 0° 10°S

2 m/s (b) EOF#2 (15%): IOD

10°N 0° 10°S

45°E

75°E

105°E

(c) Lead/lag correlation with ENSO peak amplitude in NDJ(0) 0.80

ENSO IOD IOBM

0.40 0.00 –0.40 Aug(–1) Dec(–1) Apr(0)

Aug(0) Dec(0) Apr(1)

IOD/IOBM lead

Aug(1) Dec(1)

ENSO leads

Figure 11.1  (a) First and (b) second EOF of detrended SST anomalies in the tropical Indian Ocean, an associated wind signal obtained through regression of the corresponding normalized PC. (c) Lead‐lag correlation of the Niño‐3 (a proxy of ENSO, black), average Indian Ocean (a proxy of the IOB, red) and dipole mode index (DMI; a proxy of the IOD, green) SST anomalies to the NDJ Niño‐3 index.

While some IOD events can occur independently from ENSO (e.g. Yamagata et  al., 2004; Fischer et  al; 2005; Crétat et al., 2017; H. Wang et al., 2016), the anticyclonic wind anomalies in the southeastern Indian Ocean during El Niños tend to induce cold anomalies along Java and Sumatra that can grow into a positive IOD through the Bjerknes feedback (e.g. Xie et al., 2002; Annamalai et al., 2003). As a result, IOD events tend to peak in the boreal fall before the ENSO peak (r ~ 0.6; Figure 11.1c), to dis­ sipate quickly during winter while the IOBM settles through winter to summer (r ~ 0.8; Figure 11.1c). This tendency of the IOD to phase‐lock to ENSO results in a r ~ 0.6 synchronous correlation between the two climate

modes. Recent studies suggest that positive IOD events that co‐occur with El Niño tend to strengthen the ongoing El Niño (Luo et al., 2010) and foster the occurrence of extreme El Niño events (Saji et al., 2018). Those studies suggest that the eastern Indian Ocean cold SST anom­ alies during a positive IOD supress convective activity, inducing westerly anomalies over the equatorial western Pacific, that strengthen the El Niño development. Observations, however, also indicate a tendency for the IOD to lead ENSO events by ~14 months (r ~ –0.4, Figure  11.1c). Some argue that this lead correlation is simply a consequence of the IOD being a purely passive response to ENSO and its biennial tendency (Stuecker

ENSO REMOTE FORCING  253

SON (IOD)

(a)

Rainfall & wind stress response to IOD/IOB SST patterns Modelled Observed (b)

(d)

DJF (IOBM)

(c)

0.04 N/m2 –5

–4

–3

–2

–1

0

1

2

3

mm/day

Figure 11.2  Adapted from Dayan et al. (2015). Rainfall (colors) and wind stress (vectors) anomalies driven by the anomalous (a, b) SON IOD and (c, d) DJF IOBM SST anomaly patterns deduced from observations (as a regression to the IOD/IOBM indices after having linearly removed the ENSO signal) and ECHAM‐5 simulations forced by the IOD/IOBM patterns.

et al., 2017), while others interpret this as an IOD influence on the following year’s ENSO (e.g. Clarke & van Gorder, 2003; Izumo et  al., 2010; Izumo et  al., 2014; Jourdain et  al., 2016). The effect of the IOD on following year’s ENSO also relies on the Indian Ocean–induced western Pacific wind variability. Those studies hypothesize that the western Pacific westerly anomalies induced by a cold pole of the positive IOD in the eastern Indian Ocean suddenly disappear in boreal winter in relation with the fast IOD decay at that season. This sudden release of the wind forces upwelling Kelvin waves (Izumo et al., 2016), which lead to central and eastern Pacific cooling and transition to La Niña through the Bjerknes feedback (Izumo et al., 2010). Jourdain et al. (2016) suggest that the tendency of positive IOD events to lead La Niña events by ~14 months tends to be more robust than the opposite relation in observations and CMIP models. The studies discussed above suggest that both IOBM and IOD could play a role in leading rapid phase transition of ENSO. Basically, both phenomena affect ENSO by modulating western Pacific wind variability. Annamalai et  al. (2005), Ohba and Ueda (2007), and Dayan et  al. (2015) argued that the wind anomalies over the western Pacific remotely induced by the positive and negative

poles of the IOD tend to cancel each other (Figure 11.2a, b), so that the IOBM is more efficient at inducing anomalies over the western Pacific (Figure 11.2c, d). To the contrary, other studies argue that higher ambient SST in the eastern Indian Ocean favors a larger convective response to the IOD eastern pole, which dominates the wind response in the western Pacific (e.g. Izumo et  al., 2010; Saji et  al., 2018). Irrespective of which study is correct, the sudden demise of the IOD eastern pole in boreal winter will induce a fast wind change over the Pacific that is more efficient to force an oceanic response and trigger an ENSO (Izumo et  al., 2015). This is corroborated by Ha et  al. (2017), who also showed that co‐occurring IOBM and IOD leads to a more efficient ENSO phase transition in CMIP5 simulation. Independently or together, the IOD and IOBM thus favor ENSO phase transitions, hence strengthening ENSO’s biennial tendency, i.e. shortening its period (Kug & Kang, 2006; Izumo et al., 2010, 2014). Modeling studies that artificially constrain the Indian Ocean to a climato­ logical state support a strong influence of Indian Ocean variability on ENSO (Yu et  al., 2002; Wu & Kirtman 2004; Behera et al., 2006; Dommenget et al., 2006; Obha & Watanabe 2012; Frauen & Dommenget 2012; Santoso

254  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

et al., 2012; Terray et al., 2015; Dommenget & Yu, 2017; Kajtar et al., 2017). These model studies indicate that an interactive Indian Ocean consistently shortens ENSO’s dominant period. However, results are more scattered for ENSO amplitude, with studies suggesting that Indian Ocean variability either increases (Yu et al., 2002; Wu & Kirtman, 2004) or decreases (Dommenget et  al., 2006; Frauen & Dommenget, 2012; Santoso et al., 2012; Terray et al., 2015; Dommenget & Yu 2017; Kajtar et al., 2017) ENSO variance. One can, however, note that the most recent studies, which establish their results on longer sim­ ulations, mostly suggest that Indian Ocean variability decreases ENSO variance, implying a damping influence on ENSO. The IOBM and IOD influences on ENSO phase transition respectively have important consequences for ENSO modeling and predictability. The IOBM co‐occurs almost systematically with ENSO (r ~ 0.8), so that it can be viewed as an integral part of the ENSO cycle, favoring its turnabout, and hence necessary to be well represented in models in order to capture the ENSO phase transi­ tions. In contrast, some IOD events occur independently from ENSO (r ~ 0.6), and hence yield independent information that provides a potential additional source of predictability. Clarke and Van Gorder (2003) for in­ stance demonstrated that including Indian Ocean information considerably improved ENSO forecasts at 10–15 month lead times. More specifically, using the IOD index as a precursor yields a large improvement of ENSO peak intensity hindcasts at 14 months’ lead time in obser­ vations and CMIP5 models (Izumo et  al., 2010; Dayan et  al., 2014; Jourdain et  al., 2016). Izumo et  al. (2014) further found that this improvement is superior than that obtained using an IOBM index. While most studies have so far focused on the potential influence of the tropical Indian Ocean on ENSO, some studies also suggested that SST anomalies in the subtrop­ ical Indian Ocean can also act as an ENSO precursor (e.g. Terray, 2011; Boschat et  al., 2013). Other studies have emphasized the Indonesian throughflow oceanic channel rather than the atmospheric bridge for the Indian Ocean influence on the Pacific (Wajsowicz & Schneider, 2001; Yuan et al., 2011). There is a strong volume, heat, and freshwater transport from the Pacific to the Indian Ocean through the throughflow (e.g. Gordon et al., 2010). Coupled climate model experiments blocking the throughflow hence produce a Pacific mean state change that in turns generally reduces ENSO variance and shifts its centers of action eastward (Wajsowicz & Schneider, 2001; Santoso et  al., 2011; Kajtar et  al., 2015). Some studies argued that the IOD could also contribute to ENSO onset through coastal Kelvin waves propagating through the Indonesian seas and into the Pacific as equatorial Kelvin waves (Yuan et  al., 2011), but this

pathway seems much less efficient than that associated with the atmospheric bridge (Izumo et al., 2016). 11.3. ATLANTIC OCEAN The Atlantic Ocean hosts three main regions with prominent interannual SST variations: the North Tropical Atlantic (NTA), the equatorial Atlantic, and the southern subtropical Atlantic. The NTA warming/cooling is maximum during boreal spring and is caused either by ENSO teleconnections (Chiang & Sobel, 2002; Lee et al., 2008) or by the North Atlantic Oscillation with a few months’ delay (Czaja et al., 2002). The typical pattern of interannual SSTA over the equatorial Atlantic is often referred to as the Atlantic Niño due to its similarity with the Pacific El Niño. The Atlantic Niño displays positive SSTA in the central and eastern equatorial Atlantic and usually peaks in boreal summer (Keenlyside & Latif, 2007). There is, however, no robust influence of ENSO on the equatorial Atlantic, with only a weak concurrent cor­ relation between ENSO and the equatorial Atlantic SST (Enfield & Mayer, 1997; Keenlyside & Latif, 2007). The last prominent mode of interannual SST variability in the Atlantic is the South Atlantic subtropical dipole mode (Kayano et al., 2013). It is characterized by a northeast– southwest oriented dipole‐like pattern of SSTAs peaking in boreal winter. This mode may be influenced by central Pacific (CP) El Niños through the Pacific–South American wave train (Rodrigues et  al., 2015). While a number of studies document the ENSO influence on the Atlantic Ocean, with local air‐sea coupled processes either maintaining or amplifying those SST anomalies, less attention has been paid to the influence of the Atlantic on ENSO (Melice & Servain, 2003; Latif & Grötzner, 2000; Münnich & Neelin, 2005). ­ echanisms, Improved understanding of the dynamical m supported by targeted modeling experiments, has recently helped to reach some consensus on the influence of tropical Atlantic SST anomalies on ENSO (Dommenget et al., 2006; Rodríguez‐Fonseca et al., 2009; Jansen et al., 2009; Ding et  al., 2012; Ham et  al., 2013a, 2013b; Polo et al., 2015), challenging the conventional view of the one‐ way influence of the Pacific on the Atlantic (Latif & Grotzner, 2000; Enfield & Mayer, 1997; Saravanan & Chang 2000; Chiang & Sobel, 2002; Chang et al., 2006). An NTA cooling has, for instance, been found to lead El Niño (Ham et al., 2013a; L. Wang et al., 2017). The mechanism that may explain this lead‐lag relationship is illustrated in Figure  11.3a. An anomalously cold NTA SST during boreal spring suppresses the local convective activity. This gives rise to a low‐level anticyclonic flow over the subtropical far‐eastern Pacific as the Gill‐type response, which induces anomalous southerlies over the subtropical northeastern Pacific (yellow vector in

ENSO REMOTE FORCING  255

Warm/Moist Adv.

(a) North Tropical Atlantic

C CP-type El Niño

Warm/Moist Adv.

(b) Western Hemisphere Warm Pool

PMM C

Southward Migration of Westerly El Niño El Niño

Modulation of Zonal Walker circulation

(c) Atlantic Niño

Low Level conv. EP-type El Niño Cold SST

JJA (–1)

MAM

JJA

SON/DJF

Positive Prcp.

Negative Prcp.

Figure 11.3  Schematic diagram of the influence of the (a) North Tropical Atlantic SST, (b) Western Hemispheric Warm Pool (WHWP), and (c) Atlantic Niño variabilities on the ENSO. The colours denote the season when the signals are robust. The circles and crosses denote the location of positive and negative precipitation anomalies, respectively.

Figure  11.3a). These southerlies lead to a surface warming there through reduced evaporative cooling due to weaker wind speed and warm/wet air advection from further south. This warming in turn induces positive pre­ cipitation anomalies under the eastern Pacific ITCZ (orange circle on Figure  11.3a), which plays a critical role in conveying the Atlantic signals to the Pacific due to its strong convective instability and meridional gra­ dient of moist static energy. These precipitation anom­ alies induce an anomalous low‐level cyclonic flow over the subtropical central Pacific, which progressively strengthens and extends to the western Pacific through the wind‐evaporation‐SST feedback (Xie & Philander, 1994) (red vectors in Figure 11.3a). This results in a west­ erly wind anomaly over the equatorial western Pacific during boreal summer and fall, favoring an El Niño development. In addition, the NTA could alternatively influence ENSO by inducing a remote westerly anomaly in the equatorial Pacific through atmospheric Kelvin wave response and the Indian Ocean relaying effect (Ham et al., 2013a; Yu et al., 2016).

Recently, Park et al. (2018) further argued that when the sea surface cooling is confined to the Western Hemisphere Warm Pool (WHWP) region, this lead time could be extended up to 17 months. In this frame­ work, SST anomalies over the WHWP in late boreal summer contribute to the emergence of the Pacific meridional mode (PMM) during subsequent boreal spring (yellow in Figure  11.3b), which can further trigger ENSO during the subsequent winter through induced near‐equatorial surface wind anomalies (­section  11.4). This physical mechanism shares some similarities with that of Ham et al. (2013a), involving initially an atmospheric teleconnection to the subtrop­ ical Northern Pacific, and subsequently local air‐sea coupling processes that maintain the anomaly and favor its propagation. Although slightly weaker than that of the NTA, an influence of the equatorial Atlantic on ENSO since the 1970s has been reported. An equatorial Atlantic Niña, characterized by cold conditions in the equatorial Atlantic in boreal summer, tends to be followed by a

256  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Pacific La Niña development two seasons later. While the NTA or WHWP SSTA remotely influence the Pacific through the Pacific ITCZ, the influence of the Atlantic Niño on ENSO is mediated via shifts in the zonal Walker Circulation (Rodríguez‐Fonseca et al., 2009; Martin‐Rey et al., 2012; Polo et al., 2015). An Atlantic Niña induces anomalous subsidence over the Atlantic and anomalous ascending motion over the central Pacific, which leads to enhanced convection there. This positive convection anomaly results in surface westerly wind anomalies over the equatorial central Pacific (orange vectors in Figure 11.3c), which excite eastward propagating down­ welling Kelvin waves (red arrows on Figure  11.3c) and enhance the development of an El Niño event. Atlantic Niño conditions are also statistically linked to the South Atlantic subtropical dipole mode, implying this subtrop­ ical mode is a precursor of ENSO with a 1‐year lead time (Terray, 2011; Boschat et al., 2013). Pacemaker climate model experiments, in which observed historical SST is prescribed over the tropical Atlantic, further confirmed the key role played by the tropical Atlantic on ENSO variability. These experiments indicate that the tropical Atlantic contributes to one‐ fourth of Indo‐Pacific SST variance (Rodríguez‐Fonseca et  al., 2009; Ding et  al., 2012; Polo et  al., 2015). In addition, decoupling the Atlantic Ocean in coupled models generally strengthens the ocean‐atmosphere cou­ pling in the equatorial Pacific and shifts ENSO variations to lower frequencies and stronger ENSO amplitude (e.g. L. Wang et  al., 2017; Dommenget et  al., 2006, 2017; Frauen & Dommenget, 2012; Kajtar et al., 2017). The relationship between Atlantic Niño and NTA SST variability is weak, so these modes can be treated as independent precursors of ENSO. In addition, the Atlantic Niño and NTA SSTA preferentially excite dis­ tinct ENSO flavors (Ashok et al., 2007; Kug et al., 2009; Kao & Yu, 2009; Yeh et al., 2009). NTA cooling preferen­ tially triggers CP El Niño (Ham et al., 2013b), since the Atlantic‐induced anticyclonic flow over the subtropical far‐eastern Pacific yields equatorial easterlies, hence sup­ pressing warming in the eastern Pacific. On the other hand, two of the strongest recent El Niño events (i.e. 1982–1983, 1997–1998) were preceded by Atlantic Niña events, suggesting Atlantic Niñas tend to favor eastern Pacific (EP) El Niños (Martin‐Rey et al., 2015). This sug­ gests that the NTA and Atlantic Niño variabilities induce different flavors of the El Niño events and hence play rather independent roles (Ham et al., 2013b). The tropical Atlantic SST variability also appears to have a greater influence on ENSO during recent decades (Cai et al., 2019). The Atlantic Niño in boreal summer is indeed significantly correlated to the following‐winter ENSO over 1979–2001, but this relation is much weaker before (Rodríguez‐Fonseca et al., 2009). The interdecadal

modulation of the Atlantic Niño–ENSO relationship may be partly linked to the phase of the Atlantic Multidecadal Oscillation (AMO) (Martin‐Rey et  al., 2018). Negative AMO phases are associated with a stronger subtropical high, which leads to stronger east­ erlies and a shallower thermocline over the equatorial eastern Atlantic. This shoaling enhances the Atlantic Niño variability through a stronger thermocline feedback, and the Atlantic Niño–related SST pattern extended westward (Martin‐Rey et al., 2018), which may enhance the Atlantic Niño forcing on ENSO (Losada & Rodríguez‐ Fonseca, 2016). Similarly, the negative NTA‐ENSO relationship has recently strengthened. L. Wang et al. (2017) showed that the correlation between the boreal spring the NTA SST and following winter Niño‐3.4 progressively increases from 1948 to 2016, coincident with AMO phase changes: a positive AMO phase, such as the one observed during 1992–2012, provides a warmer background SST, increasing the local atmospheric response to the NTA SST anomaly and strengthening its impact on ENSO (Ham et al., 2018). Similarly, the influence of the WHWP on ENSO is also stronger after 1985, in relation with a warmer background SST in this region compared to previous decades (Park et  al., 2018). However, those decadal variations can be statistical artifacts due to the small number of degrees of freedom. Therefore, how changes in climate background state modulate the influence of the Atlantic on ENSO should be further investigated. The lagged relationships between the Atlantic SST variations and ENSO indices could potentially increase ENSO prediction skills by using Atlantic precursors. Dayan et al. (2014) argued that using the NTA SST and the index for the Atlantic Niño (i.e. Atl3 index, area‐ averaged SST over 20–0°W, 3°S–3°N) in addition to the Pacific predictors (i.e. Pacific warm water volume and Niño‐3.4 index) significantly increase the Niño‐3.4 hindcast skill. The WHWP SST can also significantly increase the statistical ENSO forecast skill up to 17 months’ lead (Park et al., 2018). Partially coupled exper­ iments prescribing the observed Atlantic SST indicate an active role of the Atlantic SST not only on ENSO evolution but also on its prediction (Ding et al., 2012). A sensitivity test performed with a dynamical forecast model (Luo et al., 2017) showed that a successful 2‐year forecast of the prolonged 2010–2012 La Niña can be performed if warm SSTA in the Atlantic and Indian Ocean are imposed. However, the dynamical forecast systems using state‐ of‐the‐art atmosphere‐ocean coupled models do not real­ istically simulate the Atlantic SST variability (Stockdale et  al., 2006; Richter et  al., 2014, 2017) or the Atlantic‐ Pacific connection strength (Ham & Kug, 2015).

ENSO REMOTE FORCING  257

Alleviating these biases may improve ENSO forecast by better accounting for Atlantic SST variations. This inability of most current climate models to properly sim­ ulate the Atlantic SST variability is attributable to a mean‐state warm bias in the eastern equatorial and southeastern tropical Atlantic, a cold bias in the western equatorial and northern tropical Atlantic, and a large error in the equatorial thermocline slope (Richter & Xie, 2008). More work is hence required to improve the repre­ sentation and prediction of Atlantic climate variability in climate models, its influence on ENSO, and ultimately ENSO prediction skills. 11.4. EXTRATROPICAL PACIFIC Atmospheric variability outside the tropics has also been suggested to influence ENSO evolution. For in­ stance, the North Pacific Oscillation (NPO; Rogers, 1981; Linkin & Nigam, 2008), the second leading mode of the extratropical atmospheric low‐frequency variability over the North Pacific, has been identified as a precursor of ENSO events one year ahead. This lead relation is explained by the seasonal footprinting mechanism (SFM) proposed by Vimont et al. (2001, 2003a, 2003b). In this hypothesis, anomalous winds associated with the southern pole of the NPO induce SSTA in the subtropical North Pacific by altering the heat fluxes, in particular its latent component. The NPO‐induced subtropical SSTA signals resemble those of the Pacific Meridional Mode (PMM; Chiang & Vimont, 2004), which has long been recognized as an important player in connecting the extratropical Pacific to ENSO, particularly in triggering ENSO events (e.g. Anderson, 2003; Chiang & Vimont, 2004; Chang et al., 2007; Alexander et al., 2010). The PMM is characterized by covarying SSTA and sur­ face wind anomalies extending southwestward from near Baja California toward the tropical central Pacific (Figure 11.4). The PMM in boreal spring is tightly related to ENSO in the following winter. Chang et al. (2007) sug­ gested that 70% of the El Niño events between 1958 and 2000 were preceded by SSTA and surface wind anomalies similar to the PMM. For instance, positive SST anom­ alies are evident off Baja California several months before the onset of the 1986, 1994, 1997, and 2015 El Niño events. These SST anomalies persist and progressively extend southwestward over the following months to reach the western equatorial Pacific (Figure 11.4c), where they can trigger an El Niño event through the Bjerknes feedback (Figure 11.4d). The PMM itself results from the coupling between the extratropical Pacific Ocean and the overlying atmosphere. An initial warming off Baja California, presumably forced by atmospheric fluctuations via surface heat fluxes, initially enhances convection on the northern edge of the

ITCZ. This induces wind anomalies further southwest (Xie & Philander, 1994), where new SSTA can develop, since the wind anomalies are opposed to the climatolog­ ical northeasterlies and thus reduce the evaporative cooling. This wind‐evaporation‐SST (WES) feedback (Xie & Philander, 1994) allows SSTA initially induced by the extratropical atmospheric fluctuations to progres­ sively extend southwestward toward the tropical central Pacific and form the spatial pattern associated with the PMM (Figure  11.4). This ocean‐atmosphere coupling through the WES feedback also sustains the PMM from boreal winter, when the extratropical atmospheric vari­ ability is most active, into the following spring or summer to excite El Niño events. The SSTA and wind anomalies associated with the PMM resemble the optimal structures of ENSO development identified by linear inverse models (Penland & Sardeshmukh, 1995; Xue et  al., 1997). Larson and Kirtman (2014) also reported some skill using the PMM to forecast ENSO events in North American Multi‐Model Ensemble (NMME) experiments. Different mechanisms through which the PMM anomalies could trigger ENSO events have been proposed. First, the PMM‐related sur­ face wind anomalies can excite downwelling Kelvin waves in response to the equatorial westerlies and to the reflec­ tion of off‐equatorial Rossby waves, that propagate east­ ward to trigger El Niño events (e.g. Alexander et al., 2010). Alternatively, wind anomalies during the PMM positive phase may also directly recharge the ocean heat content in the equatorial Pacific via a modulation of the trade winds intensity, favoring an El Niño onset (Anderson, 2004; Anderson & Maloney, 2006; Anderson et al., 2013). The PMM has been further recognized in recent years as a major contributor to ENSO diversity (e.g. Yu et al., 2010, 2017; Capotondi et al., 2015; Yang et al., 2018; Yu & Fang, 2018). The PMM could contribute to at least two aspects of ENSO diversity: its spatial pattern (or flavor) and its evolution. The SFM mechanism is arguably more efficient at exciting CP rather EP El Niño events (Yu & Kao, 2007; Kao & Yu, 2009). This argument is supported by the similarity between CP El Niños and PMM spatial patterns, with SSTA confined to the central Pacific and extending into the northeastern Pacific (Kao & Yu, 2009). In addition, as compared to EP El Niños, CP El Niños are not accompanied by significant subsurface ocean heat content variations across the Pacific basin (Kao & Yu, 2009), suggesting that CP El Niños’ underlying dynamics is less dependent on the equatorial Pacific thermocline variations and more related to external forcings. Consistent with this argument, Kim et al. (2012) showed that the performance of NCEP’s Climate Forecast System model in simulating CP El Niños was related to its ability to simulate the PMM. These studies point towards a close relationship between extratropical Pacific processes and

258  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

30°N 20°N 10°N 0° 10°S 20°S 30°S

PMM

(a) JFM

150°E 30°N 20°N 10°N 0° 10°S 20°S 30°S

30°N 20°N 10°N 0° 10°S 20°S 30°S

150°W

120°W

90°W

180°

150°W

120°W

90°W

180°

150°W

120°W

90°W

180°

150°W

120°W

90°W

0.4

1 m/s

(b) MAM

150°E 30°N 20°N 10°N 0° 10°S 20°S 30°S

180°

(c) MJJ

150°E

(d) NDJ

150°E

–0.4

–0.2

0.0

0.2

Figure 11.4  SST (contours) and surface wind (vectors) anomalies regressed onto the MAM PMM index at various leads and lags (in months) using NCEP‐NCAR reanalysis data during the period 1958–2014. The PMM index is from http://www.aos.wisc.edu/~dvimont/MModes/Home.html

CP El Niños. However, some CP events, such as in 2004– 2005 and 2009–2010, were not preceded by a PMM pre­ cursor, indicating that other physical processes can also yield CP events. The PMM and SFM also contribute to diversity in ENSO evolution. Yu and Fang (2018) suggested that the SFM is a key source of complexity in ENSO transitions: while the recharge oscillator mechanism mostly produces a cyclic pattern of transition (i.e., El Niño to La Niña or La Niña to El Niño), the SFM mechanism produces three types of ENSO transition patterns: a cyclic pattern, an episodic pattern (i.e., El Niño or La Niña preceded by a ENSO‐neutral state), and a multiyear ENSO pattern. They also indicated that the SFM can favor multiyear La

Niña events but not El Niño events, though other studies suggest strong discharge during strong El Niño and Indian Ocean SSTA can be also responsible for the multi­ year La Niña (Luo et al., 2017). Their study suggests that forcing from extratropical Pacific may be one of the rea­ sons why multiyear La Niña events occur more often than multiyear El Niño events (Ohba & Ueda, 2009; Hu et al., 2014). In addition to the PMM in the extratropical North­ eastern Pacific, other regions of the extratropical Pacific have also been suggested to influence ENSO, including the southeastern Pacific (Zhang et  al., 2014) and the northwestern Pacific (S.‐Y. Wang et  al., 2012). Zhang et al. (2014) identified a Southern‐Hemispheric analogue

ENSO REMOTE FORCING  259

to the SFM in the southeastern Pacific, which they termed the southern PMM. The southern PMM is also characterized by covarying SSTA and trade wind anom­ alies, which extend from the Peruvian coast toward the equatorial central Pacific. The southern PMM is capable of influencing the deep tropics through its connection with cold tongue ocean dynamics (e.g. mean advection) and impacting the development of the EP El Niños (Zhang et  al., 2014; You & Furtado, 2017). The north­ western Pacific also hosts covarying SST and wind patterns similar to the PMM, which may induce oceanic Kelvin wave activity in the western tropical Pacific and later lead to ENSO events (S.‐Y. Wang et al., 2012). This mode was suggested to be also related to NPO. 11.5. DISCUSSION In this chapter, we reviewed possible remote influences of SST anomalies in various regions outside the tropical Pacific on ENSO evolution. Various climate modes in the Indian Ocean (IOD, IOBM), the Atlantic (NTA, Atlantic Niño, WHWP) or the extratropical Pacific (NPO, PMM) can trigger or alter ENSO events, generally by altering the western Pacific wind variability (e.g. Kug et al., 2005). Every ENSO event is, however, not preceded by those precursors, as an ENSO event can occur spontaneously through coupled dynamics internal to the tropical Pacific. On the other hand, some ENSO events are preceded (and potentially influenced) by a combination of these precur­ sors. Tables  11.1 and 11.2 summarize which precursors preceded each ENSO event over the 1980–2017 period.

Based on a 0.5 standard deviation criterion, each of these precursors preceded 4 to 7 El Niño events out of 12, and 4 to 7 La Niña events out of 12. Interestingly, stronger ENSO events tend to be preceded by more active precur­ sors, the very strong 1997–1998 El Niño being preceded by 6 precursors out of 7 and the weak 2004–2005 El Niño event being preceded by none. Although the historical dataset is really too short to conclude, it is conceivable that the combination of several remote forcings can enhance the ENSO amplitude. The lagged correlations between Niño‐3.4 SST and pre­ cursor indices, introduced in this chapter, are significant as shown in Table 11.3. However, these correlations may be partly the result of ENSO influencing many regions and having a biennial tendency (Jourdain et al., 2016; Stuecker et  al., 2017). To exclude this possibility, we recomputed these correlations after linearly removing the simultaneous Niño‐3.4 SST. This generally increases the correlations for most indices, suggesting that these precursors indepen­ dently influence the Pacific and are not solely a result of ENSO teleconnections. That is, the internal variabilities over the Indian, Atlantic, and extratropical oceans, might be more important for affecting ENSO characteristics than the ENSO-induced variability over each basin. Among 24 El Niño and La Niña events, the PMM fre­ quently co‐occurred with ENSO events (13 events), and the partial correlation is high (Table  11.3), suggesting that the PMM index is an important precursor of ENSO as discussed in section  11.4. On the other hand, four positive NPO events preceded El Niño events, consistent with the SFM argument, but three negative NPO events

Table 11.1  Niño‐3.4 index and various precursor indices for individual El Niño events. Shading indicates the case that the index is greater (or less) than 0.5 std, 1 std, and 1.5 std (−0.5 std, −1 std, and −1.5 std) and the sign is consistent with the relation on ENSO discussed in the text. Each index is averaged value from two SST datasets of ERSST and HADISST. Year

Niño‐3.4 ND(0)J(1)

IOBMD(‐1) IODSON(‐1) JF(0) NTAMAM(0) ANiñoJJA(0)

NPOD(‐1) WHWPJAS(‐1) JF(0)

PMMFMAM(0)

82/83

2.18

−0.76

−0.52

−0.20

−1.87

1.45

−0.53

0.69

86/87

1.04

0.33

−0.81

−1.33

−0.15

−0.50

1.09

1.37

87/88

0.95

0.79

0.24

0.87

1.57

−0.75

−0.25

−0.57

91/92

1.48

−0.79

0.76

−0.93

0.74

0.10

−1.05

0.57

94/95

1.11

−0.19

−0.86

−1.07

−1.17

−0.60

−0.01

1.02

97/98

2.36

−1.79

−1.10

−0.33

−1.33

−1.70

1.94

0.67

02/03

1.19

−0.74

0.57

−0.37

0.24

−0.25

−0.44

−0.21

04/05

0.64

−0.24

0.24

0.30

−0.48

−0.30

0.39

0.33

06/07

0.91

−1.17

−1.14

0.23

0.20

1.40

−1.11

−0.34

09/10

1.57

0.07

−0.33

−1.50

−0.67

−0.95

0.09

−0.72

14/15

0.71

−0.33

−0.33

−1.30

−0.52

−1.10

0.54

0.93

15/16

2.62

−0.17

−0.10

−0.83

−0.78

−0.40

1.34

1.85

260  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Table 11.2  Niño‐3.4 index and various precursor indices for individual La Niña events. Shading indicates the case that the index is greater (or less) than 0.5 std, 1 std, and 1.5 std (−0.5 std, −1 std, and −1.5 std) and the sign is consistent with the relation on ENSO discussed in the text. Each index is averaged value from two SST datasets of ERSST and HADISST. YEAR

Niño‐3.4 ND(0)J(1)

IOBWD(‐1) IODSON(‐1) JF(0)

NTAMAM(0) ANiñoJJA(0)

WHWPJAS(‐1)

NPOD(‐1) JF(0) PMMFMAM(0)

83/84

−1.03

1.60

1.52

1.23

−1.30

0.25

1.09

−1.75

84/85

−1.26

0.33

−0.86

−0.13

1.07

1.10

0.74

−0.43

88/89

−2.06

0.88

2.00

1.13

1.96

2.10

−0.89

0.44

95/96

−0.87

1.98

−0.38

0.30

1.15

−1.30

0.61

0.97

98/99

−1.54

2.74

2.57

1.80

1.13

0.30

−1.04

−2.10

99/00

−1.66

−1.36

−1.00

−0.57

1.46

1.65

−1.66

−2.00

00/01

−0.85

−0.12

−0.62

0.00

−0.83

0.00

−0.67

−1.56

05/06

−0.78

0.10

0.14

1.77

−1.57

−0.25

0.48

0.52

07/08

−1.64

1.74

0.43

0.27

0.52

0.05

−0.18

−0.40

08/09

−0.75

0.33

−1.10

0.07

1.24

−0.25

−1.20

−1.80

10/11

−1.62

−0.19

1.43

2.77

0.96

0.00

0.93

−0.28

11/12

−1.03

−1.67

−1.38

0.10

−0.72

0.95

−0.58

−1.01

Table 11.3  Correlation and partial correlation with Niño‐3.4 SST at ND(0)J(1). The partial correlation is calculated after removing the effect of the simultaneous Niño‐3.4 SST.

Correlation Partial correlation

IOD SON(‐1)

IOBW D(‐1)JF(0)

NTA MAM(0)

Anino JJA(0)

WHWP JAS(‐1)

NPO D(‐1)JF(0)

NPMM FMAM(0)

−0.40** −0.44**

−0.27** −0.32**

−0.48** −0.62**

−0.45** −0.09

−0.33** −0.32**

−0.21* −0.25**

−0.50** −0.50**

 90% significant  95% significant

*

**

also preceded El Niño events, indicating a false alarm. Likewise, six negative NPO events are related to La Niña events, but still four positive NPO events are found for La Niña cases. This suggests that ENSO response to the NPO‐related forcing are not as systematic as for the PMM forcing. Interestingly, the PMM index is signifi­ cantly correlated to NPO index (cor = 0.45), but most NPO false alarm events did not co‐occur with coincident PMM events, except for the 2008–2009 La Niña event. The partial correlation with Niño‐3.4 SSTA is highest for the NTA index, suggesting that it is a good indicator for ENSO development. The Atlantic Niño events frequently co‐occurred with ENSO events (13 events), but the partial correlation is quite weak (–0.09). This weak relationship suggests that Atlantic Niño hardly affects ENSO phase but may possibly modulate ENSO magnitude. Cai et al. (2019) showed that using Indian and Atlantic ocean precursors improves ENSO prediction skill and the skill improvement is particularly more distinctive in the

recent decades. However, Tables 11.1 and 11.2 show that the Indian Ocean precursors are closely related to ENSO events during 1980–1999, but this relationship weakened recently to some extent. This contradictory result might be related to the decadal changes in ENSO stability. In the past decades (1980–1999), ENSO amplitude is strong and the phase transition is clear so that internal Pacific precursors such as heat content can be a dominant factor in driving ENSO evo­ lution, and the external remote forcings mostly play a role in enhancing ENSO variability. In recent decades (2000–2018), however, ENSO stability is weak so that external factors are more prominent in generating ENSO events. It might be also linked to why El Niño diversity is evident in recent decades. This speculation deserves further investigation. Our understanding of ENSO’s remote forcings is still in its infancy, with large uncertainty largely related to the short observational record. In particular, the observed relationship between ENSO and remote precursors has changed on the interdecadal timescales (Melice & Servain,

ENSO REMOTE FORCING  261

2003; Münnich & Neelin, 2005; Park et  al., 2018; Cai et  al., 2019). Therefore, it is difficult to quantitatively measure how much each precursor contributes to the evo­ lution of ENSO events. Current ENSO predictive skill is limited particularly for a long lead time, possibly due to our immature understanding of relative contributions of ENSO’s remote forcings to ENSO development. To overcome these observational issues, many studies have used climate models to support the observational arguments and quantitatively estimate the relative contri­ butions to ENSO evolution. Current climate models sim­ ulate to some extent the effects of the remote forcings on ENSO from the Indian Ocean (Kug et al., 2012; Jourdain et  al., 2016; Ha et  al., 2017), the Atlantic Ocean (Keenlyside et al., 2013; Ham & Kug, 2015; Park et al., 2018), and the extratropics (Vimont et al., 2003b). Various decoupled experiments, by switching off the feedbacks from the Indian or the Atlantic Oceans, confirmed that interbasin interactions play a significant role in ENSO variability (Yu et al., 2002; Wu & Kirtman, 2004; Obha & Watanabe 2012; Terray et  al., 2015; Dommenget & Yu, 2017; Kajtar et  al., 2017). The most recent studies e­xamining the role of each remote region within the same modeling framework further indicate that interactions with the Indian and Atlantic Oceans provide a delayed negative feedback to ENSO but also increase ENSO fre­ quency (Terray et  al., 2015; Kajtar et  al., 2017; Dommenget & Yu, 2017). However, current climate models tend to underestimate the effect of remote forcing on ENSO compared to the observational estimates, with a weaker impact of the NTA (Ham & Kug, 2015), the Atlantic Niño (Kucharski et al., 2015), and the Indian Ocean basin (Kug et al., 2012) on the Pacific basin. The underestimation and misrepresenta­ tion of remote impacts on ENSO in current climate models are related to model’s systematic biases (Richter et al., 2014; McGregor et al., 2018; Luo et al., 2018; Kajtar et  al., 2018). For example, the underestimation of the NTA SST effect is possibly related to the dry bias over the Atlantic warm pool area (Ham & Kug, 2015). The weaker equatorial Atlantic SST gradient may also lead to a weak convective response to Atlantic SSTA and a weaker shift of the Walker Circulation, resulting in an unrealistic impact of Atlantic SSTA on the Pacific wind variability. Biases in the distribution of climatological precipitation over the Indo‐Pacific warm pool region may be an impor­ tant factor in the underestimated strength of the Indian Ocean feedback (Kug & Ham, 2012). All these studies suggest that a realistic representation of the model mean state can substantially improve the model’s ability to sim­ ulate the influence of remote regions on ENSO evolution, which will eventually lead to improved ENSO predictive skill in dynamical forecast models.

ACKNOWLEDGMENTS We thank Dr. Takeshi Izumo for generating Figure 11.1. JSK and YGH are supported by the Korea Meteorological Administration Research and Development Program under Grant KMIPA 2018‐03212. JV’s and ML’s contri­ bution to this work benefited from the ARISE ANR project (ANR‐18‐CE01‐0012) funding. JYY was sup­ ported by the National Science Foundation under grants NSF AGS1505145 and AGS1833075. REFERENCES Alexander, M. A., & Scott, J. D. (2008). The role of Ekman ocean heat transport in the Northern Hemisphere response to ENSO. Journal of Climate, 21(21), 5688‐5707. Alexander M. A., I. Bladé, M. Newman, J. R. Lanzante, N.‐C. Lau, & J. D. Scott, 2002: The atmospheric bridge: The influence of ENSO teleconnections on air–sea interaction over the global oceans. J. Climate, 15, 2205–2231. Alexander M. A., Vimont, D. J., Chang, P., Scott, J. D. (2010). The impact of extratropical atmospheric variability on ENSO: Testing the seasonal footprinting mechanism using coupled model experiments. J. Climate, 23, 2885‐2901. Anderson, B. T. (2003). Tropical Pacific sea‐surface tempera­ tures and preceding sea level pressure anomalies in the sub­ tropical North Pacific. J. Geophys. Res., 108. doi: 10.1029/2003JD003805. Anderson, B. T. (2004). Investigation of a large‐scale mode of ocean atmosphere variability and its relation to tropical Pacific sea surface temperature anomalies. J. Climate, 17, 1089–4098. doi: 10.1175/1520‐0442(2004)0172.0.CO;2 Anderson, B. T., & Maloney, E. (2006). Interannual tropical Pacific sea surface temperatures and their relation to pre­ ceding sea level pressures in the NCAR CCSM2. J. Climate, 19, 998–1012. doi:10.1175/JCLI3674.1 Anderson, B. T., Perez, R. C., & Karspeck, A. (2013). Triggering of El Niño onset through trade wind–induced charging of the equatorial Pacific. Geophy. Res. Letts., 40, 1212–1216, doi:10.1002/grl.50200 Annamalai, H., Murtugudde, R., Potemra, J., Xie, S. P., Liu, P., & Wang, B. (2003) Coupled dynamics over the Indian Ocean: Spring initiation of the Zonal Mode. Deep Sea Res. 50, 2305–2330. Annamalai, H., Xie, S. P., & McCreary, J. P. (2005). Impact of Indian Ocean sea surface temperature on developing El Niño. J Climate, 18, 302–319. Ashok, K., Behera, S. K., Rao, S. A., Weng, H., & Yamagata, T. (2007). El Niño Modoki and its possible teleconnection. Journal of Geophysical Research: Oceans, 112(C11). Barnett, T. P. (1983). Interaction of the monsoon and Pacific trade wind system at interannual time scales Part I: The equatorial zone. Monthly Weather Review, 111, 756–773. Bjerknes, J. (1969). Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev. Weather Review, 97, 163–172.

262  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Boschat, G., Terray, P., & Masson, S. (2013). Extratropical forc­ ing of ENSO. Geophysical Research Letters, 40(8), 1605–1611. Cai, W., et al. (2019). Pantropical climate interactions, Science, 363, doi: 10.1126/science.aav4236 Capotondi, A., Wittenberg, A. T., Newman, M., Di Lorenzo, E., Yu, J.‐Y., Braconnot, P., et al. (2015). Understanding ENSO diversity. Bulletin of the American Meteorological Society, 96, 921–938. doi:10.1175/BAMS‐D‐13‐00117.1 Chang, P., Fang, Y., Saravanan, R., Ji, L., & Seidel, H. (2006). The cause of the fragile relationship between the Pacific El Niño and the Atlantic Niño. Nature, 443(7109), 324. Chang, P., Zhang, L., Saravanan, R., Vimont, D. J., Chiang, J.C.H., Ji, L., et al. (2007). Pacific meridional mode and El Niño‐Southern Oscillation. Geophys. Res. Lett. 34. doi: 10.1029/2007GL030302 Chen, M.‐C., Li, T., Shen, X.‐Y., & Wu, B. (2016). Relative roles of dynamic and thermodynamic processes in causing evolu­ tion asymmetry between El Niño and La Niña. J. Climate, 29, 2201–2220. Chiang. J. C., & Vimont, D. (2004). Analogous Pacific and Atlantic meridional modes of tropical atmosphere‐ocean variability. J. Climate, 17, 4143–4158. Chiang, J. C., & Sobel, A. H. (2002). Tropical tropospheric tem­ perature variations caused by ENSO and their influence on the remote tropical climate. Journal of Climate, 15(18), 2616–2631. Clarke, A. J., & Van Gorder, S. (2003). Improving El Niño pre­ diction using a space‐time integration of Indo‐Pacific winds and equatorial Pacific upper ocean heat content. Geophys. Res. Lett., 30, 1399. doi:10.1029/2002GL016673. Crétat, J., Terray, P., Masson, S., Sooraj, K. P., & Roxy, M. K. (2017). Indian Ocean and Indian summer monsoon: Relationships without ENSO in ocean–atmosphere coupled simulations. Climate Dynamics, 49, 1429–1448. Czaja, A., Van der Vaart, P., & Marshall, J. (2002). A diagnostic study of the role of remote forcing in tropical Atlantic vari­ ability. Journal of Climate, 15(22), 3280–3290. Dayan, H., Vialard, J., Izumo, T., & Lengaigne, M. (2014) Does sea surface temperature outside the tropical Pacific con­ tribute to enhanced ENSO predictability? Clim. Dyn., 43, 1311–1325. Dayan, H., Izumo, T., Vialard, J., Lengaigne, M., & Masson, S. (2015). Do regions outside the tropical Pacific influence ENSO through atmospheric teleconnections? Clim. Dyn., 45, 583–601. Deser, C., Alexander, M. A., Xie, S.‐P., & Phillips, A. S. (2010). Sea surface temperature variability: Patterns and mecha­ nisms. Annu. Rev. Mar. Sci., 2, 115–143. Ding, H., Keenlyside, N. S., & Latif, M. (2012). Impact of the equatorial Atlantic on the El Niño southern oscillation. Climate dynamics, 38(9–10), 1965–1972. Dommenget, D., & Yu, Y. (2017). The effects of remote SST forcings on ENSO dynamics, variability and diversity. Climate Dynamics, 49(7–8), 2605–2624. https://doi. org/10.1007/s00382‐016‐3472‐1 Dommenget, D., Semenov, V., & Latif, M. (2006). Impacts of the tropical Indian and Atlantic Oceans on ENSO, Geophys. Res. Lett., 33, L11701, doi:10.1029/2006GL025871

Du, Y., Xie, S.‐P., Huang, G., & Hu, K.‐M. (2009). Role of air‐ sea interaction in the long persistence of El Niño–induced North Indian Ocean warming. J. Climate, 22, 2023–2038. Enfield, D. B., & Mayer, D. A. (1997). Tropical Atlantic sea sur­ face temperature variability and its relation to El Niño‐ Southern Oscillation. Journal of Geophysical Research: Oceans, 102(C1), 929–945. Fischer, A. S., Terray, P., Guilyardi, E., Gualdi, S., & Delecluse, P., (2005). Two independent triggers for the Indian Ocean dipole/zonal mode in a coupled GCM. Journal of climate, 18, 3428–3449. Frauen, C., & Dommenget, D. (2012). Influences of the tropical Indian and Atlantic Oceans on the predictability of ENSO, Geophys. Res. Lett., 39, L02706. doi:10.1029/2011GL050520 Gadgil, S., Joseph, P. V., & Joshi, N. V. (1984). Ocean–atmosphere coupling over monsoon regions. Nature, 312, 141–143. Gordon, A. L., Sprintall, J., Van Aken, H. M., Susanto, D., Wijffels, S., Molcard, R., et  al. (2010). The Indonesian throughflow during 2004–2006 as observed by the INSTANT program. Dynamics of Atmospheres and Oceans, 50, 115–128. Graham, N. E., & Barnett, T. P. (1987). Sea surface tempera­ ture, surface wind divergence, and convection over the tropical oceans. Science, 238, 657–659. Ha, K.‐J., Chu, J.‐E., Lee, J.‐Y., & Yun, K.‐S. (2017). Interbasin coupling between the tropical Indian and Pacific Ocean on interannual timescale: Observation and CMIP5 reproduction. Clim Dyn., 48, 459–475. Ham, Y. G., & Kug, J. S. (2015). Role of north tropical Atlantic SST on the ENSO simulated using CMIP3 and CMIP5 models. Climate dynamics, 45(11–12), 3103–3117. Ham, Y. G., Kug, J. S., Park, J. Y., & Jin, F. F. (2013a). Sea sur­ face temperature in the north tropical Atlantic as a trigger for El Niño/Southern Oscillation events. Nature Geoscience, 6(2), 112. Ham, Y. G., Kug, J. S., & Park, J. Y. (2013b). Two distinct roles of Atlantic SSTs in ENSO variability: North Tropical Atlantic SST and Atlantic Niño. Geophysical Research Letters, 40(15), 4012–4017. Ham, Y. G., Kug, J. S., Yang, W. H., & Cai, W. (2018). Future changes in Extreme El Niño events modulated by North Tropical Atlantic variability. Geophysical Research Letters. Hong, C. C., Li, T., LinHo, & Chen, Y.‐C. (2010). Asymmetry of the Indian Ocean basinwide SST anomalies: Roles of ENSO and IOD. J. Climate, 23, 3563–3576. Hoskins, B., & Karoly, D. J. (1981). The steady linear response of a spherical atmosphere to thermal and orographic forcing. J. Atmos. Sci., 38, 1179–1196. Hu, Z. Z., Kumar, A., Xue, Y., & Jha, B. (2014). Why were some La Niñas followed by another La Niña? Climate Dynamics, 42, 1029–1042. Huang, B. (2004). Remotely forced variability in the tropical Atlantic Ocean. Climate Dyn., 23, 133–152. Huang, B., & Kinter III, J. L. (2002). Interannual variability in the tropical Indian Ocean. J. Geophys. Res., 107, 3199. doi:10.1029/2001JC001278 Izumo, T., Vialard, J., Lengaigne, M., de Boyer Montégut, C., Behera, S. K., Luo, J.‐J., et al. (2010). Influence of the state of the Indian Ocean Dipole on the following year’s El Niño. Nature Geoscience, 3, 168–172.

ENSO REMOTE FORCING  263 Izumo, T., Lengaigne, M., Vialard, J., Luo, J‐J., Yamagata, T., & Madec, G. (2014). Influence of Indian Ocean Dipole and Pacific recharge on following year’s El Niño: Interdecadal robustness. Clim Dyn., 42, 291–310. Izumo, T., Vialard, J., Dayan, H., Lengaigne, M., & Suresh, I. (2015). A simple estimation of equatorial Pacific response from windstress to untangle Indian Ocean Dipole and Basin influences on El Niño. Clim Dyn., 46, 2247–2268. Izumo, T., Vialard, J., Dayan, H., Lengaigne, M., Suresh, I. (2016). A simple estimation of equatorial 565 Pacific response from windstress to untangle Indian Ocean Dipole and Basin influences on 566 El Niño. Clim Dyn., 46, 2247–2268. Jansen, M. F., Dommenget, D., & Keenlyside, N. (2009). Tropical atmosphere–ocean interactions in a conceptual framework. Journal of Climate, 22(3), 550–567. Jin, F. F. (1997). An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model. J. Atmos. Sci., 54, 811–829. Jourdain, N.C., Lengaigne, M., Vialard, J., Izumo, T., & Sen Gupta, A. (2016). Further insights on the influence of the Indian Ocean Dipole on the following year’s ENSO from observations and CMIP5 Models. J. Climate, 29, 637–658. Kajtar, J. B., Santoso, A., England, M. H., & Cai, W. (2015). Indo‐Pacific climate interactions in the absence of an Indonesian Throughflow. Journal of Climate, 28, 5017–5029. Kajtar, J. B., Santoso, A., England, M. H., & Cai, W. (2017). Tropical climate variability: Interactions across the Pacific, Indian, and Atlantic Oceans. Climate Dynamics, 48, 2173–2190. Kajtar, J. B., Santoso, A., McGregor, S., England, M. H., & Baillie, Z. (2018). Model under‐representation of decadal Pacific trade wind trends and its link to tropical Atlantic bias. Clim. Dyn., 50, 1471–1484. Kao, H. Y., Yu, J. Y. (2009). Contrasting eastern‐Pacific and central‐Pacific types of El Niño. J. Clim., 22, 615–632. Kayano, M. T., Andreoli, R. V., & de Souza, R. A. F. (2013). Relations between ENSO and the South Atlantic SST modes and their effects on the South American rainfall. International Journal of Climatology, 33(8), 2008–2023. Keenlyside, N. S., & Latif, M. (2007). Understanding equatorial Atlantic interannual variability. Journal of Climate, 20(1), 131–142. Keenlyside, N. S., Ding, H. Latif, M. (2013). Potential of equatorial Atlantic variability to enhance 690 El Niño predic­ tion. Geophys. Res. Lett., 40, 2278–2283. Kim, S. T., Yu, J. Y., Kumar, A., & Wang, H. (2012). Examination of the two types of ENSO in the NCEP CFS model and its extratropical associations. Mon. Wea. Rev., 140, 1908–1923. Klein, S. A., Soden, B. J., & Lau, N.‐C. (1999). Remote sea sur­ face temperature variations during ENSO: Evidence for a tropical atmospheric bridge. J. Climate, 12, 917–932. Kucharski, F., Syed, F. S. Burhan, A. Farah, I. Gohar, A. (2015). Tropical Atlantic influence on 509 Pacific variability and mean state in the twentieth century in observations and CMIP5. Clim. Dyn., 44, 881–896. Kug, J.‐S., & Ham, Y.‐G., (2012). Indian Ocean feedback to the ENSO transition in a multimodel ensemble. J. Climate, 25, 6942–6957. Kug, J.‐S., & Kang, I.‐S. (2006). Interactive feedback between ENSO and the Indian Ocean. J. Clim., 19, 1784–1801.

Kug, J.‐S., An, S.‐I., Jin, F.‐F., & Kang, I.‐S. (2005). Preconditions for El Niño and La Niña onsets and their rela­ tion to the Indian Ocean. Geophy. Res. Lett., 32, L05706. doi: 10.1029/2004GL021674 Kug, J.‐S., Li, T., An, S.‐I., Kang, I.‐S., Luo, J.‐J., Masson, S., & Yamagata T. (2006). Role of the ENSO–Indian Ocean cou­ pling on ENSO variability in a coupled GCM. Geophys. Res. Lett., 33, L09710. doi:10.1029/2005GL024916 Kug, J. S., Jin, F. F., & An, S. I. (2009). Two types of El Niño events: cold tongue El Niño and warm pool El Niño. Journal of Climate, 22(6), 1499–1515. Larson, S. M., & Kirtman, B. P. (2014). The Pacific Meridional Mode as an ENSO precursor and predictor in the North American Multimodel Ensemble. J. Climate, 27, 7018–7032. doi: http://dx.doi.org/10.1175/JCLI‐D‐14‐00055.1 Latif, M., & Grötzner, A. (2000). The equatorial Atlantic oscillation and its response to ENSO. Climate Dynamics, 16(2–3), 213–218. Lau, N.‐C., & Nath, M. J. (2003). Atmosphere–ocean variations in the Indo‐Pacific sector during ENSO episodes. J. Climate, 16, 3–20. Lee, S. K., Enfield, D. B., & Wang, C. (2008). Why do some El Niños have no impact on tropical North Atlantic SST? Geophysical Research Letters, 35(16). Lengaigne, M., Boulanger, J. P., Menkes, C., & Spencer, H. (2006). Influence of the seasonal cycle on the termination of El Niño events in a coupled general circulation model. J. Climate, 19, 1850–1868. doi:10.1175/jcli3706.1 Linkin, M. E., & Nigam, S. (2008). The North Pacific Oscillation‐West Pacific teleconnection pattern: Mature‐ phase structure and winter impacts. J. Climate, 21, 1979– 1997, doi:10.1175/2007JCLI2048.1 Lloyd, J., Guilyardi, E., Weller, H., & Slingo, J. (2009). The role of atmosphere feedbacks during ENSO in the CMIP3 models. Atmos. Sci. Lett., 10, 170–176. Losada, T., & Rodríguez‐Fonseca, B. (2016). Tropical atmo­ spheric response to decadal changes in the Atlantic Equatorial Mode. Climate Dynamics, 47(3–4), 1211–1224. Luo, J. J., Zhang, R., Behera, S. K., Masumoto, Y., Jin, F. F., Lukas, R., & Yamagata, T. (2010). Interaction between El Niño and extreme Indian ocean dipole. Journal of Climate, 23, 726–742. Luo, J.‐J., Liu, G., Hendon, H., Alves, O., & Yamagata, T. (2017). Inter‐basin sources for two‐year predictability of the multi‐year La Niña event in 2010–2012. Scientific Reports, 7(1), 2276. doi: 10.1038/s41598‐017‐01479‐9 Luo, J.‐J., Wang, G., & Dommenget, D. (2018). May common model biases reduce CMIP5’s ability to simulate the recent Pacific La Niña‐like cooling? Clim. Dyn., 50, 1335–1351. Martín‐Rey, M., Polo, I., Rodríguez‐Fonseca, B., & Kucharski, F. (2012). Changes in the interannual variability of the tropical Pacific as a response to an equatorial Atlantic forc­ ing. Scientia Marina, 76(S1), 105–116. Martín‐Rey, M., Rodríguez‐Fonseca, B., & Polo, I. (2015). Atlantic opportunities for ENSO prediction. Geophysical Research Letters, 42(16), 6802–6810. Martín‐Rey, M., Polo, I., Rodríguez‐Fonseca, B., Losada, T., & Lazar, A. (2018). Is there evidence of changes in tropical Atlantic variability modes under AMO phases in the obser­ vational record? Journal of Climate, 31(2), 515–536.

264  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE McGregor, S., Stuecker, M. F. Kajtar, J. B. England, M. H. Collins, M. (2018). Model tropical Atlantic biases underpin diminished Pacific decadal variability. Nat. Clim. Change, 8, 493–499. Meehl, G. A. (1987). The annual cycle and interannual vari­ ability in the tropical Pacific and Indian Ocean regions. Monthly Weather Review, 115, 27–50. Meehl, G. A., Arblaster, J. M., & Loschnigg, J. (2003). Coupled ocean‐atmosphere dynamical processes in the tropical Indian and Pacific Oceans and the TBO. J. Climate, 16, 2138–2158. Melice, J. L., & Servain, J. (2003). The tropical Atlantic meridional SST gradient index and its relationships with the SOI, NAO and Southern Ocean. Climate Dynamics, 20(5), 447–464. Münnich, M., & Neelin, J. D. (2005). Seasonal influence of ENSO on the Atlantic ITCZ and equatorial South America. Geophysical Research Letters, 32(21). Murtugudde, R., McCreary, J. P., & Busalacchi, A. J. (2000). Oceanic processes associated with anomalous events in the Indian Ocean with relevance to 1997–1998. J. Geophys. Res., 105, 3295–3306. Ohba, M., & Ueda, H. (2007). An impact of SST anomalies in the Indian Ocean in acceleration of the El Niño to La Niña transition. JMSJ., 85, 335–348. Ohba, M., & Ueda, H. (2009). Role of nonlinear atmospheric response to SST on the asymmetric transition process of ENSO. J. Climate, 22, 177–192. Ohba, M., & Watanabe, M. (2012). Role of the Indo‐Pacific interbasin coupling in predicting asymmetric ENSO transition and duration. J. Climate, 25, 3321–3335. Okumura, Y. M., Ohba, M., Deser, C., & Ueda, H. (2011). A proposed mechanism for the asymmetric duration of El Niño and La Niña. J. Climate. 24, 3822–3829. Park, J. H., Kug, J. S., Li, T., & Behera, S. K. (2018). Predicting El Niño beyond 1‐year lead: Effect of the Western Hemisphere warm pool. Scientific Reports, 8(1), 14957. Penland, C., & Sardeshmukh, P. D. (1995). The optimal growth of tropical sea surface temperature anomalies. J. Climate, 8, 1999–2024. Polo, I., Martin‐Rey, M., Rodriguez‐Fonseca, B., Kucharski, F., & Mechoso, C. R. (2015). Processes in the Pacific La Niña onset triggered by the Atlantic Niño. Climate Dynamics, 44(1– 2), 115–131. Reverdin, G., Cadet, D. L., & Gutzler, D. (1986). Interannual displacements of convection and surface circulation over the Indian Ocean, Q. Jour. Roy. Met. Soc., 112, 43–67. Richter, I., & Xie, S. P. (2008). On the origin of equatorial Atlantic biases in coupled general circulation models. Climate Dynamics, 31(5), 587–598. Richter, I., Xie, S. P., Behera, S. K., Doi, T., & Masumoto, Y. (2014). Equatorial Atlantic variability and its relation to mean state biases in CMIP5. Climate Dynamics, 42(1–2), 171–188. Richter, I., Xie, S. P., Morioka, Y., Doi, T., Taguchi, B., & Behera, S. (2017). Phase locking of equatorial Atlantic vari­ ability through the seasonal migration of the ITCZ. Climate Dynamics, 48(11–12), 3615–3629. Rodrigues, R. R., Campos, E. J., & Haarsma, R. (2015). The impact of ENSO on the South Atlantic subtropical dipole mode. Journal of Climate, 28(7), 2691–2705.

Rodríguez‐Fonseca, B., Polo, I., García‐Serrano, J., Losada, T., Mohino, E., Mechoso, C. R., & Kucharski, F. (2009). Are Atlantic Niños enhancing Pacific ENSO events in recent decades? Geophysical Research Letters, 36(20). Rogers, J. C. (1981). The North Pacific Oscillation. Int. J. Climatol., 1, 39–57. doi: 10.1002/joc.3370010106 Saji, N. H., Goswami, B. N., Vinayachandran, P. N., & Yamagata, T. (1999). A dipole mode in the tropical Indian Ocean. Nature, 401, 360–363. Saji, H. N., Jin, D., & Thilakan, V. (2018). A model for super El Niños. Nat Comms, 9, 401. doi: 10.1038/s41467‐018‐04803‐7 Santoso, A., Cai, W., England, M. H., & Phipps, S. J. (2011). The role of the Indonesian throughflow on ENSO dynamics in a coupled climate model. J. Climate, 24, 585–601. Santoso, A., England, M. H., & Cai, W. (2012). Impact of Indo‐ Pacific feedback interactions on ENSO dynamics diagnosed using ensemble climate simulations. J. Climate, 25, 7743–7763. Saravanan, R., & Chang, P. (2000). Interaction between tropical Atlantic variability and El Niño–Southern oscillation. Journal of Climate, 13(13), 2177–2194. Stockdale, T. N., Balmaseda, M. A., & Vidard, A. (2006). Tropical Atlantic SST prediction with coupled ocean– atmosphere GCMs. Journal of Climate, 19(23), 6047–6061. Stuecker, M. F., Timmermann, A., Jin, F.‐F., Chikamoto, Y., Zhang, W., Wittenberg, A. T., et al. (2017). Revisiting ENSO/ Indian Ocean Dipole phase relationships. Geophys. Res. Lett., 44, 2481–2492. doi:10.1002/2016GL072308 Suarez, M. J., & Schopf, P. S. (1988). A delayed action oscillator for ENSO. J. Atmos. Sci., 45, 3283–3287. Terray, P. (2011). Southern Hemisphere extra‐tropical forcing: A new paradigm for El Niño‐Southern Oscillation. Climate Dynamics, 36(11–12), 2171–2199. Terray, P., Masson, S., Prodhomme, C., Roxy, M. K., & Sooraj, K. P. (2015). Impacts of Indian and Atlantic oceans on ENSO in a comprehensive modeling framework. Clim Dyn., 46, 2507–2533. Timmermann, A., An, S.‐I., Kug, J.‐S., Jin, F.‐F., Cai, W., Capotondi, A., et  al., (2018). El Niño‐Southern Oscillation complexity. Nature, 559, 535–545. https://doi.org/10.1038/ s41586‐018‐0252‐6 Tokinaga, H., & Tanimoto, Y. (2004). Seasonal transition of SST anomalies in the tropical Indian Ocean during El Niño and Indian Ocean dipole years. J. Meteor. Soc. Japan, 82, 1007–1018. Trenberth, K., & Hurrell, J. (1994). Decadal atmosphere–ocean variations in the Pacific. Clim. Dyn., 9, 303–319. Vialard, J., Duvel, J.‐P., McPhaden, M., Bouruet‐Aubertot, P., Ward, B., Key, E., et al. (2009a). Cirene: Air sea interactions in the Seychelles‐Chagos thermocline ridge region. Bull. Am. Met. Soc., 90, 45–61. Vimont, D. J., Battisti, D. S., & Hirst, A. C. (2001). Footprinting: A seasonal connection between the Tropics and mid‐lati­ tudes. Geophys. Res. Lett., 28, 3923–3926. Vimont, D. J., Wallace, J. M., & Battisti, D. S. (2003a). The seasonal footprinting mechanism in the Pacific: Implications for ENSO. J. Climate, 16, 2668–2675. Vimont, D. J., Battisti, D. S., & Hirst, A. C. (2003b). The seasonal footprinting mechanism in the CSIRO general circulation models. J Clim, 16, 2653–2667.

ENSO REMOTE FORCING  265 Vimont, D. J., Alexander, M., & Fontaine, A. (2009). Midlatitude excitation of tropical variability in the Pacific: The role of ther­ modynamic coupling and seasonality. J. Climate, 22, 518–534. Wajsowicz, R. C., & Schneider, E. K. (2001). The Indonesian throughflow’s effect on global climate determined from the COLA coupled climate system. Journal of Climate, 14, 3029–3042. Wang, H., Murtugudde, R., & Kumar, A. (2016). Evolution of Indian Ocean dipole and its forcing mechanisms in the absence of ENSO. Climate Dynamics, 47, 2481–2500. Wang, L., Yu, J. Y., & Paek, H. (2017). Enhanced biennial vari­ ability in the Pacific due to Atlantic capacitor effect. Nature Communications, 8, 14887. Wang, S.‐Y., L’Heureux, M. & Chia, H.‐H. (2012). ENSO pre­ diction one year in advance using western North Pacific sea surface temperatures. Geophys. Res. Lett., 39, L05702. doi:10.1029/2012GL050909 Watanabe, M., & Jin, F.‐F. (2002). Role of Indian Ocean warming in the development of Philippine Sea anticyclone during ENSO. Geophys. Res. Lett., 29, 116‐1–116‐4. Webster, P. J., Moore, A., Loschnigg, J., & Leban, M. (1999). Coupled ocean‐atmosphere dynamics in the Indian Ocean during 1997–98. Nature, 401, 23 September 1999, 356–360. White, W. B, & Annis, J. (2004). Influence of the Antarctic cir­ cumpolar wave on El Niño and its multidecadal changes from 1950–2001. J Geophys Res, 109(C0):6019. doi: 10.1029/2002JC001666 White, W. B., Chen, S.‐C., Allan. R. J., & Stone, R. C. (2002). Positive feedbacks between the Antarctic circumpolar wave and the global El Niño‐Southern Oscillation wave. J Geophys Res, 107(C10), 3165. doi: 10.1029/2000JC000581 Wu, R., & Kirtman, B. P., (2004). Impacts of the Indian Ocean on the Indian summer monsoon–ENSO relationship. Journal of Climate, 17, 3037–3054. Wu, R., Kirtman, B. P., & Krishnamurthy, V. (2008). An asym­ metric mode of tropical Indian Ocean rainfall variability in boreal spring, J. Geophys. Res., 113, D05104. doi:10.1029/ 2007JD009316. Xie, S. P., & Philander, S.G.H. 1994. A coupled ocean‐atmosphere model of relevance to the ITCZ in the eastern Pacific. Tellus A, 46. doi: 10.1034/j.1600‐0870.1994.t01‐1‐00001.x Xie, S.‐P., Annamalai, H., Schott, F. A., & McCreary, J. P. (2002). Structure and mechanisms of South Indian Ocean cli­ mate variability. J. Climate, 15, 864–878. Xie, S.‐P., Hu, K., Hafner, J., Du, Y., Huang, G., & Tokinaga, H. (2009). Indian Ocean capacitor effect on Indo‐western Pacific climate during the summer following El Niño. J. Climate, 22, 730–747. Xie, S. P., Kosaka, Y., Du, Y., Hu, K., Chowdary, J. S., & Huang, G., (2016). Indo‐western Pacific ocean capacitor and coherent climate anomalies in post‐ENSO summer: A review. Advances in Atmospheric Sciences, 33, 411–432.

Xue, Y., Cane, M. A., & Zebiak, S. E. (1997). Predictability of a coupled model of ENSO using singular vector analysis. Part I: Optimal growth in seasonal background and ENSO cycles. Mon. Wea. Rev., 125, 2043–2056. Yamagata, T., Behera, S. K., Luo, J. J., Masson, S., Jury, M. R., & Rao, S. A. (2004). Coupled ocean‐atmosphere vari­ ability in the tropical Indian Ocean. Earth’s Climate: The Ocean–Atmosphere Interaction, Geophys. Monogr, 147, 189–212. Yang, S., Li, Z. Yu, J.‐Y. Hu, X. Dong, W., & He, S. (2018). El Niño‐Southern Oscillation and its impact in the chang­ ing climate. National Science Review, nwy046, doi: 10.1093/ nsr/nwy046 Yeh, S. W., Kug, J. S., Dewitte, B., Kwon, M. H., Kirtman, B. P., & Jin, F. F. (2009). El Niño in a changing climate. Nature, 461(7263), 511. You, Y., & Furtado, J. C. (2017). The role of South Pacific atmospheric variability in the development of different types of ENSO, Geophysical Research Letters, 44 (14), 7438–7446. Yu, J.‐Y., & Fang, S.‐W. (2018). The distinct contributions of the seasonal footprinting and charged‐discharged mecha­ nisms to ENSO complexity. Geophysical Research Letters, 45, 6611–6618. doi:10.1029/2018GL077664 Yu J.‐Y., & Kao, H.‐Y. (2007). Decadal changes of ENSO per­ sistence barrier in SST and ocean heat content indices: 1958– 2001. J. Geophys. Res., 112. doi: 10.1029/2006JD007654 Yu, J.‐Y., Mechoso, C. R., McWilliams, J. C., & Arakawa, A. (2002). Impacts of the Indian Ocean on the ENSO cycle. Geophys. Res. Lett., 29, 1204. doi:10.1029/2001GL014098 Yu, J.‐Y., Kao, H.‐Y., & Lee, T. (2010). Subtropics‐related inter­ annual sea surface temperature variability in the equatorial central Pacific. Journal of Climate, 23, 2869–2884. doi: 10.1175/2010JCLI3171.1 Yu, J., Li, T., Tan, Z., & Zhu, Z. (2016). Effects of tropical North Atlantic SST on tropical cyclone genesis in the western North Pacific. Climate Dynamics, 46(3–4), 865–877. Yu, J.‐Y., Wang, X. Yang, S. Paek, H. & Chen, M. (2017). Changing El Niño‐Southern Oscillation and associated cli­ mate extremes. In S.‐Y. Wang, J.‐H. Yoon, C. Funk, & R. R. Gillies (Eds.), Climate extremes: Patterns and mechanisms (Vol. 226, pp. 3–38). AGU Geophysical Monograph Series. Yuan, D. L., et al. (2011). Forcing of the Indian Ocean dipole on the interannual variations of the tropical Pacific Ocean: Roles of the Indonesian throughflow. J Clim, 15, 3597–3608. Zebiak, S. E. (1993). Air–sea interaction in the equatorial Atlantic region. J. Climate, 6, 1567–1586. Zhang, H., Clement, A., & Di Nezio, P. (2014). The South Pacific meridional mode: A mechanism for ENSO‐like vari­ ability. Journal of Climate, 27, 769–783. https://doi.org/10. 1175/JCLI‐D‐13‐00082.1

12 The Effect of Strong Volcanic Eruptions on ENSO Shayne McGregor1,2, Myriam Khodri3, Nicola Maher4, Masamichi Ohba5, Francesco S. R. Pausata6, and Samantha Stevenson7

ABSTRACT The response of the El Niño–Southern Oscillation (ENSO) to tropical and extratropical volcanic eruptions has important worldwide implications for volcanically driven risk estimates. While there have been many studies on this subject using observations, paleoclimate archives, and model simulations, a comprehensive review of ENSO response to tropical and extratropical volcanic eruptions has not been presented to date. The relatively short observational record is suggestive of a relationship between tropical volcanism and El Niño events. Analyzing 17 previously defined reconstructions of ENSO, which on average span the past 550 years, we find that 70% of these reconstructions display a significant eastern Pacific warming (El Niño–like) response in the year of eruption, when a consistent set of volcanic events dates are used. There also appears to be an emerging consensus from models, with the overwhelming majority displaying a relative El Niño–like response in the eruption year. Thus, here we report that there is a clear consistency of evidence between observations, paleo-proxies, and models. Questions remain, however, over exactly how the near‐uniform radiative cooling of a tropical volcanic event projects onto ENSO. There is little observational and paleoclimatic evidence for the impact of extratropical volcanism on ENSO; however, models suggest that an extratropical Northern (Southern) Hemisphere eruption leads to a relative El Niño–like (La Niña–like) response. Despite the consistency in the evidence presented above, many subtle differences still exist among the modeled response to tropical and extratropical volcanic forcing that could be aided by the use of a consistent experimental protocol for general circulation model simulations (i.e., VolMIP). 12.1. INTRODUCTION Large volcanic eruptions can have major impacts on global climate, affecting both atmospheric and ocean circulation. 1  School of Earth Atmosphere and Environment, Monash University, Melbourne, VIC, Australia 2  ARC Centre of Excellence for Climate Extremes, Monash University, Melbourne, VIC, Australia 3   Laboratoire d’Océanographie et du Climat: Expérim­ entations et approches numériques, Sorbonne Universités, IPSL, UMR CNRS/IRD/MNHN, Paris, France 4  Max Planck Institute for Meteorology, Hamburg, Germany 5  Central Research Institute of Electric Power Industry, Chiba, Japan 6  Department of Earth and Atmospheric Sciences, University of Quebec in Montreal, Montreal, Quebec, Canada 7  Bren School of Environmental Science & Management, University of California Santa Barbara, Santa Barbara, CA, USA

These climatic impacts come via the injection of chemically and microphysically active gases and solid aerosol particles into the stratosphere by the eruptions. These particles scatter and absorb incoming solar radiation (e.g., Timmreck, 2012), resulting in a net reduction in surface radiation and hence a cooling at the surface (Harshvardhan, 1979; Rampino & Self, 1984; Robock & Mao, 1995). Global average surface temperatures have been shown to reach the maximum cooling 6–18 months after the eruptions peak optical depth and return to normal values approximately 5–6 years after the eruption (e.g., Thompson et al., 2009). The strong surface temperature cooling influence of these events has also seen them at least partly implicated in many past hiatuses of global surface warming (Maher et al., 2015), like the well documented hiatus of 1997–2013 (Fyfe et al., 2013; Haywood et  al., 2014; Medhaug et  al., 2017; Santer et  al., 2014).

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 267

268  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

These large volcanic eruptions are also shown to result in strong reductions in ocean heat content and sea level during the cooling period that follows the eruption (Church et  al., 2005; Fasullo et  al., 2017; Pausata et al., 2015). Volcanic eruptions have also been suggested to modulate natural modes of climate variability, for example, to provide conditions that lead to the development of a positive phase of the Arctic Oscillation during the first and second boreal winter following the eruption (Christiansen, 2008; Kodera, 1994; Shindell et al., 2004). Several studies have also shown the impacts of volcanic eruptions on the El Niño–Southern Oscillation (ENSO) (e.g., Emile‐Geay et al., 2008; McGregor & Timmermann, 2011). Given the profound influence of ENSO on global climate and its strong societal relevance (see chapters 1 and 14–20), it is important to understand the impact of volcanism on ENSO. If the modulation of ENSO by volcanic events is strong, there is also the possibility that future volcanic eruptions, after occurrence, may enhance predictability of subsequent El Niño/La Niña events. Observational SST analysis beginning in 1882 indicates that El Niño events followed/coincided with four out of five big eruptions during the historical period (Santa María in October 1902, Mt. Agung in March 1963, El Chichón in April 1982, and Mt. Pinatubo in June 1991). This apparent relationship between volcanism and ENSO was first identified in a controversial study by Handler (1984). This relationship was based on temporal correlations between both phenomena, but is also clearly visible  looking at observed sea surface temperature anomalies (SSTAs) composited around volcanic events (Figures 12.1). Questions over the Handler (1984) study arose because of a perceived lack of statistical robustness of the results, the volcanic chronology used, and the timing of volcanic events relative to ENSO event initiation (Nicholls, 1988; Robock, 2000; Sear et  al., 1987; Self et al., 1997). This latter point is apparent in Figure 12.1a, as it is clear that both the El Chichón (1982) and Pinatubo (1991) volcanic eruptions occurred after the initiation of El Niño events in the respective years, which is indicative of a more coincidental relationship. Hypotheses surrounding the impact of volcanism on ENSO have clearly evolved from the initial proposal that ENSO was entirely forced by volcanism, to volcanic forcing exerting a discernible influence on ENSO, whereby the likelihood and/or magnitude of events can be modulated. The motivation of the present chapter is to synthesize the current state of our understanding of how tropical and extratropical volcanic eruptions influence ENSO using information from paleoclimate archives, Coupled Model Intercomparison Project (CMIP) simulations, and other more idealized climate model experiments. This synthesis will include (i) tropical and extratropical eruptions of various sizes and a discussion

about how these eruptions impact radiation and surface temperatures, (ii) our dynamical understanding of how volcanically induced ENSO modulation occurs, and (iii) a discussion of the sensitivities of the results to the magnitude and timing of the eruption, and the current state of the Pacific at the time of the eruption. This chapter is laid out as follows: In section 12.2 we present details of the volcanic forcing. In section  12.3 paleoclimatic evidence of the volcanic influence on ENSO is presented, while in section 12.4 model evidence of the volcanic influence on ENSO is presented. Mechanisms proposed to understand the influence of volcanism on ENSO are also presented and reviewed in section 12.4. Implications, potential future work and conclusions of the chapter are all presented in section 12.5. 12.2. VOLCANIC FORCING OF CLIMATE Volcanic eruptions can inject large amounts of sulfur dioxide and hydrogen sulfide into the stratosphere. These gases are oxidized into sulfate aerosols that enhance the background aerosol mass load. For sufficiently large tropical eruptions, the resulting aerosols can then spread over the whole globe (Robock, 2000). These stratospheric volcanic aerosols then modify the radiative balance through scattering of solar shortwave radiation and increased stratospheric absorption of solar and terrestrial infrared radiation (Santer et al., 2014; Stenchikov et al., 1998). The eruption‐induced perturbation of the Earth’s radiative budget can persist for 2 to 3 years (Figure 12.2a–b), which corresponds to the lifetime of stratospheric aerosols (Robock, 2000). While the volcanic radiative forcing is relatively short lived, the large spatial extent and magnitude of the forcing makes eruptions an important source of variability for the Earth’s radiative balance (Myhre et al., 2013). For these reasons, tropical eruptions whose plumes reach the stratosphere lead to near‐global surface cooling that can persist for 3+ years and changes in ocean temperature and circulation that may last decades (Church et al., 2005), making them an active component of the climate system (Robock, 2000; Thompson & Solomon, 2009). Our level of understanding of volcanic aerosol microphysical processes and their influence on climate is, however, still limited. This is largely due to the limited number of stratospheric eruptions that occurred during the satellite era. The best‐observed tropical stratospheric eruption is the eruption of Mt. Pinatubo in June 1991 in the Philippines. This event injected approximately 15–20 Mt of SO2 gas into the lower stratosphere. A post-­ eruption global surface cooling of between 0.2°C and 0.3°C (e.g., Swingedouw et  al., 2017; Thompson et  al., 2009) and a global lower ­tropospheric cooling of ~0.5°C (e.g. Soden et al., 2002) persisted for the 2 years following the Mt. Pinatubo eruption,

EFFECT OF STRONG VOLCANIC ERUPTIONS ON ENSO  269

A S O N D

Pinatubo

F M A M J

J

El Chichón Mt Agung Santa Maria Krakatau

J F M A M J J

Year 0

A S O N D J

Year 1

(a) Observed Niño-3.4 Relative SSTA (b) Observed Equatorial Pacific Relative SSTA

–2.4 –1.2 –0.8

0 °C

0.8

1.2

2.4

150E

180E

–1.2 –0.8 –0.4

210E 0

240E

0.4 0.8 1.2

270E °C

Figure 12.1 Observed relative sea surface temperature (SST) response to the five main eruptions with stratospheric aerosol injection during 1870–2010. SST anomalies (SSTA) are calculated relative to the preceding 5‐year climatology, and the resulting anomalies are plotted relative to the tropical average (20°N–20°S). (a) Evolution of the composited Niño‐3.4 (5°S–5°N, 170°W–120°W) relative SSTA plotted over the 2‐year period including the five largest volcanic eruptions during 1870–2010 in HadISST observations (black line). The red dots denote when the five events (colored lines) have anomalies of the same sign. An arrow indicates the dates and relative magnitude of the selected eruptions, based on aerosol optical depth provided by Gao et al. (2008). (b) Same as (a) but shown as a longitude‐time section. The black rectangle denotes the Niño‐3.4 region during October‐November‐ December. Stippling highlights times and locations where the five eruptions display anomalies of the same sign and the contour corresponds to the 90% significance level of this anomaly according to a two‐tailed Student’s  t‐test. Note that the composites maintained a common seasonality (the calendar months of each eruption were aligned despite the differing eruption months) due to the seasonally synchronized nature of ENSO events (Nicholls, 2008).

with a gradual return to ­unperturbed conditions the third year (Figure  12.2c–f) (Soden et  al., 2002; Swingedouw et  al., 2017). It is, however, difficult to disentangle the volcanic forcing influence of individual events at regional scales due to climate phenomena, such as the Asian Monsoon and other natural modes of climate variability, having regional magnitudes that are much larger than the global mean temperature change induced by the eruption (Fischer et  al., 2007). While this recent event was well observed, the limited observations of past volcanic events also lead to large uncertainties in the aerosol size, distributions, and estimates of radiative forcing (Driscoll et al., 2012; Kravitz & Robock, 2011; Santer et al., 2014; Toohey et al., 2011) (Figure 12.2a). The current state‐of‐the‐art climate models that include interactive stratospheric aerosols are shown to accurately

capture the observed aerosol size distribution and climate response after the Mt. Pinatubo eruption (Niemeier et al., 2010). The Mt. Pinatubo eruption is the only large tropical event for which stratospheric aerosols have been reasonably well observed; as such, very little data are available to constrain eruptions of different magnitudes, and different latitudes, during different seasons. Several recent global aerosol‐modeling studies (e.g. Toohey et al., 2011) suggest that the latitude of the eruption, along with its season of occurrence and strength, can lead to differences in the aerosol concentration and evolution. (i) The differing radiative impacts of eruptions occurring at different latitudes are clearly highlighted by comparing tropical and extratropical eruptions. Aerosols formed after tropical stratospheric eruptions are often spread across the globe, while those that are formed after mid‐to‐high latitude

1.6 1.2

2.0

0.8

1.0

0.4

0.0 1850

0.0 1900

1950 Year

1.6

El chichón Mt Agung

1.2

Santa Maria

0.8

Krakatau

1.0

0.4 0.0

0.0

(d)

0

1 2 3 4 Years after eruption year

Temperature (°C)

0.0 –0.6

Land surface Temperature composite

1900

1950

0.0 –0.2 –0.4 –0.6

2000

–1

0

Year

(f)

Global mean SST 0.8

HadSST.3 ERSSTv5

0.0 –0.4 –0.8

5

SST composite

0.1 0.0 –0.1 –0.2

1850

1900

1950

–1

2000

0

1

2

3

4

5

Years after eruption year

Year Volcanic forcing over the last millennium

8.0

CMIP6 PMIP4

4.0

6.0

3.0

4.0

2.0

1.0

1.0

0.0

Aerosol optical depth

5.0

10.0

Estimated radiative forcing (W m–2)

(g)

1 2 3 4 Years after eruption year

0.2 Temperature (°C)

Temperature (°C)

(e) 0.4

5

0.2

0.6

1850

2.0

2.0

Pinatubo

–1

CRUTEM Berkeley GlSS

1.2

3.0

2000

Global mean Land surface Temperature 1.8

Volcanic forcing composite 4.0

Aerosol optical depth

3.0

2.0

SATO CMIP6

Aerosol optical depth

Estimated radiative forcing (W m–2)

4.0

(c) Temperature (°C)

(b)

Volcanic forcing

Estimated radiative forcing (W m–2)

(a)

0.0 1000

1200

1400

1600

1800

2000

Year

Figure 12.2  Volcanic forcing and surface temperatures. (a) Evolution of volcanic forcing estimates (in W m–2, left axis, following Gao et al. 2008 scaling method) based on stratospheric aerosol optical depth (right axis) provided by Sato et al. (1993) (blue line) and CMIP6 (red line) during 1850–2018. (b) Composite mean of volcanic forcing estimates using the five tropical volcanic eruptions in 1870–2010 with significant injection of sulfuric aerosol in the tropical stratosphere: Pinatubo (June 1991), El Chichón (April 1982), Mt. Agung (March 1963), Santa María (October 1902), and Krakatau (August 1883). (c) Annual global mean land surface temperature anomalies during 1850–2018 in CRUTEM (Morice et al., 2012, black line), Berkeley (Rohde et al., 2013, red line) and GISS (Hansen et al., 2010, blue line) data sets during 1850–2018. (d) Three product average (CRUTEM, Berkley, GISS) composite of annual mean land surface temperature anomalies around the five previously defined volcanic events. (e) Annual global mean sea surface temperature anomalies during 1850–2018 in HadISST3 (Rayner et al., 2003, black line) and ERSSTv5 (Huang et  al., 2017, blue line) data sets during 1850–2018. (f) Two product average (HadISST3, ERSSTv5) composite of annual mean sea surface temperature anomalies around the five previously defined volcanic events. (g) Evolution of volcanic forcing estimates (in W m–2, right axis, following Gao et al., 2018, scaling method) based on stratospheric aerosol optical depth (left axis) provided for PMIP4 (Toohey et al., 2016, blue line) during 850–1850 and CMIP6 (red line) during 1850–2018. The horizontal black line indicates the magnitude of the 1991 Pinitubo eruption for comparison.

EFFECT OF STRONG VOLCANIC ERUPTIONS ON ENSO  271

stratospheric eruptions remain confined within the eruption hemisphere (Kravitz & Robock, 2011; Oman et al., 2005). (ii) Simulations suggest that a Pinatubo‐like eruption (including the amount of SO2 injected and eruption latitude) occurring in the boreal winter will produce significantly larger global mean cumulative aerosol optical depth (AOD) than the same eruption occurring in any of the other seasons (e.g. Toohey et  al., 2011). The stronger wintertime stratospheric Brewer‐Dobson circulation leads to SO2 entrainment higher up in the stratosphere and stronger concentrations in the winter hemisphere (Stoffel et al., 2015), along with smaller effective radius, smaller sedimentation rates, and increased stratospheric aerosol lifetime compared with other seasons. For high‐latitude Northern Hemisphere eruptions of the same magnitude, the largest AODs are obtained when an eruption occurs in summer. This is due to higher concentrations of OH radical available to turn SO2 gas into sulphate aerosol compared to winter as well as slower removal and smaller wave activity during this season (Kravitz & Robock, 2011; Toohey et al., 2011). (iii) Model evidence also suggests that for very large tropical eruptions (i.e. those about 40 times the magnitude of the Pinatubo eruption), the effect of the Brewer‐Dobson circulation background seasonality on AOD is reduced by enhanced tropical stratospheric heating that is induced through absorption of infrared radiation (Toohey et al., 2011). Thus, aerosol size, distribution, and lifetime, along with the associated climatic effects of stratospheric volcanoes, appear to be heavily influenced by the eruption season, latitude, and magnitude. For volcanic eruptions of the recent and remote past, the reconstructed radiative properties (Figure  12.2g) are hampered by many assumptions used to deduce the stratospheric sulfate aerosol size distribution (Sigl et  al., 2015). Methods utilized to date have relied on linearity between volcanic radiative forcing and peak SO4 deposition concentrations in ice cores (Timmreck et al., 2009). The climatic impact of the largest volcanic events has been assessed in numerous modeling studies and tree‐ ring‐based hemispheric temperature reconstructions. However, volcanic surface cooling derived from PMIP3/ CMIP5 climate model simulations is systematically much stronger than the cooling seen in tree‐ring‐based proxies, suggesting that the proxies underestimate cooling and/or the modeled forcing is unrealistically high (Anchukaitis et  al., 2012; Briffa et  al., 1998; D’Arrigo et  al., 2006; Mann et al., 2012). Support of the latter is provided by the results of several studies that show that there is no universal linear relationship between the magnitude of the eruptions and the global cooling response (e.g., Pinto et  al., 2018; Timmreck et  al., 2009). This is due to the aerosol microphysical processes effectively self‐limiting

the impacts of major volcanic eruptions. Specifically, the use of a fixed aerosol size distribution for past eruptions does not allow for a realistic simulation of the climatic effects of very large Plinian eruptions, and its use may result in inaccurate estimates of volcanic forcing (Stoffel et al., 2015). Thus, while we have reasonable estimates of observed forcing for recent volcanic eruptions, accurately simulating the exact magnitude of the climate response of past volcanic events is a challenging task for climate models due to the limited availability of data required for accurate radiative forcing estimates. Information such as the eruption timing, latitude, and aerosol size and distribution are required for past and future eruptions for models to be able to produce accurate response magnitudes. 12.3. PALEOCLIMATE EVIDENCE The short observational record (from ~1850) and limited numbers of large volcanic eruptions during this period make it difficult to assess the statistical significance of the observed ENSO response to volcanic forcing. The uncertainty in the observed response is further amplified when considering that two of the El Niños that co‐occurred with volcanic eruptions during the observational record were already underway prior to the eruption (e.g. Nicholls, 1988). In an attempt to enhance the signal‐ to‐noise ratio by looking at ENSO’s response to more volcanic events, many studies have developed and utilized paleo-reconstructions of past ENSO activity. Paleoclimate evidence may also provide very important insights on how the ENSO response to volcanism depends on the properties of the eruption, as historical events have spanned a much larger range of aerosol loadings and other eruption characteristics than observed during the instrumental period (Figure 12.2g). Previous work has shown that the ENSO response to eruptions during the last millennium has been significant, although some uncertainty still remains. For instance, the studies of Adams et al. (2003) and McGregor et al. (2010) broadly agree with the instrumental record and suggest that an El Niño–like SST anomaly occurs in the year of the volcanic event. Adams et  al. (2003) data highlight anomalously warm conditions existing for several years after the eruption, not just a single year, as in the observed record. On the other hand, other studies suggest that tropical volcanism can lead to a La Niña–like response (Anchukaitis et  al., 2010; Li et  al., 2013). The terms El Niño–like and La Niña–like are used in the context of paleoclimatic proxies to respectively refer to an inferred warming and cooling in the eastern/central equatorial Pacific. This terminology is used out of convenience because ENSO events have distinct threshold‐based ­definitions and implied underlying dynamics that in some

272  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Table 12.1  Details of the ENSO reconstructions employed in this work, including their correlation with the observed ENSO variability as represented by the Southern‐Oscillation Index (SOI) over the period 1900–1977. Proxy Number

Start Year

End Year

 1  2  3  4  5  6  7  8  9 10 11 12 13

1706 1408 1650 1590 1800 1300 1727 1525 1650 1607 1540 900 1150

1997 1978 1990 1990 1990 1978 1982 1982 1977 1998 1998 2002 1998

14

1150

1998

15

1150

1998

16 17

1301 1607

2005 1997

Reference Stahle et al. (1998) Cook (2000) Mann et al. (2000) Evans et al. (2002) Evans et al. (2001) Cook et al. (2008) Braganza et al. (2009) Braganza et al. (2009) McGregor et al. (2010) Wilson et al. (2010) Wilson et al. (2010) Li et al. (2011) Emile‐Geay et al. (2013a, 2013b) Emile‐Geay et al. (2013a, 2013b) Emile‐Geay et al. (2013a, 2013b) Li et al. (2013) Tierney et al. (2015)

circumstances cannot be or are not diagnosed in the cited literature. There is also disagreement across paleo‐reconstructions with regard to the magnitude of the increased probability of an El Niño event in the year following major eruptions (Stevenson et al., 2017). To date, the reasons for these uncertainties have not been comprehensively assessed. Differences between previously defined ENSO reconstructions come from at least one of the following categories: source proxy type, source proxy geographical region, and reconstruction methods. Examples of proxy type and geographical region differences are best demonstrated between the reconstructions of Tierney et  al. (2015) and Li et  al. (2013) (Table  12.1). The study of Tierney et al. (2015) produced a reconstruction of eastern equatorial Pacific SST that spans 1607–1998 from eight coral records in the surrounding region. The study of Li et al. (2013), on the other hand, used tree rings from more than 2000 different locations around the Pacific basin to reconstruct eastern equatorial Pacific SSTs. Examples of differing reconstruction methods are most proficiently demonstrated between the studies of Emile‐Geay et  al. (2013a, 2013b) and Braganza et al. (2009). The study of Emile‐Geay et al. utilizes both the simple composite plus scaling method (e.g. Bradley & Jones, 1993), which standardizes the source proxies before averaging and scaling them to match an observed target time series. The study of Braganza et al., instead, uses an empirical orthogonal function analysis to effectively filter non‐climate‐related

Source Proxy Location

Source Proxy Type

Correlation with SOI

Pacific Basin North America Near global tropics Indo‐Pacific Basin America North America Pacific Basin Pacific Basin Pacific Basin Tropical Pacific Pacific Basin North America Near global tropics

Tree ring Tree ring Mixed Coral Tree ring Tree ring Mixed Mixed Mixed Mixed Mixed Tree ring Mixed

0.76 0.73 0.78 0.67 0.67 0.75 0.73 0.65 0.83 0.58 0.48 0.6 0.8

Near global tropics

Mixed

0.83

Near global tropics

Mixed

0.82

Pacific Basin Tropical Pacific

Tree ring Coral

0.67 0.74

noise in the source proxy network and identify the dominant, likely climate‐related signal that is evident across all proxies. Thus, while all studies have sought to reconstruct past ENSO variability, the differing methods and source data have led to each study producing differing ENSO reconstructions. Numerous data sets detailing the dates and magnitudes of past volcanic eruptions are available (Crowley et  al., 2008; e.g. Gao et al., 2008; Sigl et al., 2015), and analysis of these data sets shows conflicting dates of eruptions (Table 12.2). Volcanic years in the Gao et al. (2008) and Sigl et al. (2015) data sets are identified when the global radiative forcing reaches –3 W m–2, while those in the Crowley et  al. (2008) data set are identified when the tropical radiative forcing (average of forcing at 30°S and 30°N) reaches –3 W m–2. As in Gao et al., we convert the aerosol optical depths reconstructed by Crowley et al. to radiative forcing by multiplying by 20. Thus, differences in both the ENSO reconstructions and the identified years of volcanic events can lead to differences in the apparent response to volcanic forcing. As it is impossible to identify which is the most accurate ENSO reconstruction due to the lack of observational data, we choose to utilize all 17 reconstructions of ENSO variability detailed in Table  12.1 and look for response agreement for confidence. Each of these reconstructions has been shown to accurately depict ENSO variability over the entire 20th century and have a continuous record from near present back until at least the year 1800

EFFECT OF STRONG VOLCANIC ERUPTIONS ON ENSO  273 Table 12.2  The summary list of identified large tropical volcanic events from the three data sets, where large is defined as having a radiative forcing of less than –3 W m–2. The far right column lists large events that identified in at least two of the three data sets. Volcanic Event Years Gao et al., 2008 (IVI)

Crowley et al., 2008 (ICI)

Sigl et al., 2015

Agreed Event Years

1167 1176 1171 1191 1227 1258 1275 1284

1229 1230 1258–59 1286–87

1230 1258 1276

1230 1258 1275/76

1286

1286

1341

1641 1674

1345 1453 1458 1512 1585 1595 1601 1641 1673

1695–96

1695

1452 1456–57 1584 1600 1641

1452/53 1457/58 1584/85 1600/01 1641 1673/74

1693 1695

1719 1762 1783 1809 1815

1809 1815–17

1835

1835 1884

1991

1809 1815 1832 1836 1862 1884 1964 1992

1809 1815 1835 1884 1991/92

(Braganza et  al., 2009; Cook, 2000; Cook et  al., 2008; Emile‐Geay et al., 2013a, 2013b; Evans et al., 2001, 2002; Li et al., 2011, 2013; Mann et al., 2000; McGregor et al., 2010; Stahle et  al., 1998; Tierney et  al., 2015; Wilson et  al., 2010; see Table  12.1). The average of all 17 composite mean ENSO reconstructions around the volcanic events identified in each of the three data sets (Table 12.2) reveals vastly different responses in the year of volcanic event (Figure  12.3a). Thus, rather than focusing on any one index of past volcanic variability, we identify volcanic years that exist in at least two of the three data sets provided (Table 12.2).

Our results suggest that the majority of the 17 reconstructions display an El Niño–like response in the year of the volcanic event. We then calculate the significance of each ENSO reconstruction response using a Monte‐ Carlo sampling approach. That is, for each reconstruction, 1000 composite averages are calculated around a random set of years. The number of years/events composited around is determined by the number of volcanic events (Table 12.2) that occur during the period of each reconstruction. The 5th and 95th percentiles are then calculated from these random composites for each reconstruction. A significant El Niño–like response is identified when the ENSO reconstruction exceeds the 95% level, while a significant La Niña–like response is identified if the response is below the 5th percentile. We find that 12 of the 17 reconstructions produce a significant El Niño– like response in the year of the volcanic eruption, while none of the reconstructions produce a significant La Niña–like response. Two years after the event, we find that 8 of the 17 reconstructions produce a La Niña–like response that is significant above the 95% level. Thus, our results support the notion that volcanism can induce warmer SST in the central-eastern equatorial Pacific. Among the 17 reconstructions, 5 of them display a significant La Niña–like state 2 years prior to the eruption (Figure  12.3). Three of the five reconstructions that display a significant La Niña‐like state 2 years prior to the eruption display an El Niño–like state in the eruption year. Given the cyclic nature of ENSO, this suggests that at least some portion of the El Niño–like signal found in these three reconstructions in the year of the volcanic event may be unrelated to the forcing. Additionally, differences in the seasonality of eruptions may affect the representation of eruptions across proxy reconstructions due to, for instance, differential seasonal sensitivity of proxies or dynamical differences in the response as a function of eruption season (Stevenson et al., 2017). This is not something that can be resolved with further analysis of our current paleoclimatic information. As such, further model experimentation and analysis is required to better understand the contributions of dynamical ­modulations and proxy‐related considerations in generating variations across reconstructions. 12.4. MODEL EVIDENCE AND DYNAMICS 12.4.1. Tropical Volcanism At first glance, the effect of tropical explosive volcanic eruptions is not uniform across models. Initial experiments utilizing simple coupled models produce an increased probability for an El Niño–like state and the occurrence of El Niño events following stratospheric eruptions, agreeing with observations and paleoclimatic estimates (Emile‐Geay

274  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a)

El Nino-like

ENSO anomaly

0.5 Gao et al. (2008) Crowley et al. (2008) Sigl et al (2015)

0

–0.5 –2

(b)

La Nina-like –1

0

1

1

2

3

4

5 Stahle et al. 1998 Cook 2000 Mann et al. 2000 Evans et al. 2001 Evans et al. 2002 Cook et al. 2008 Braganza et al. 2009 (R5) Braganza et al. 2009 (R8) McGregor et al. 2010 Wilson et al. 2010 (COA) Wilson et al. 2010 (TEL) Li et al. 2011

El Nino-like

ENSO anomaly

0.5 0 –0.5

Emile-Geay et al.2013a,b (ERSST) Emile-Geay et al.2013a,b (HadSST) Emile-Geay et al.2013a,b (Kaplan)

–1 –1.5 –2

Li et al. 2013 Tiemey et al. 2015 Reconstruction mean

La Nina-like –1

0

1

2

3

4

5

Year after eruption

Figure 12.3  (a) Multireconstruction ENSO composite mean around volcanic events identified by three different data sets of past volcanic activity (see legend and Table 12.2). (b) Composite means for each individual ENSO reconstruction and the multireconstruction mean (see legend), which are all composited around volcanic event years that were identified in at least two data sets of past volcanic activity (Table 12.2, agreed event years).

et  al., 2008; Mann et  al., 2005). The simplicity of the  coupled model (Zebiak & Cane, 1987) utilized for the ­aforementioned studies, along with the applied uniform volcanic forcing, ensure that ocean dynamics are the only way that the forcing can project onto ENSO (i.e., the dynamical thermostat mechanism detailed in ­section 12.4.1.1). Further studies, using more complex atmosphere‐ocean coupled general circulation models (CGCMs), show numerous ways in which volcanic forcing can project onto ENSO (see sections 12.4.1.2 and 12.4.1.3)(e.g. McGregor & Timmermann, 2011; Pausata et al., 2020; Zanchettin et  al., 2012). The results of some of these CGCM‐type studies suggest that volcanic radiative forcing can lead to an initial cooling of the equatorial Pacific (e.g. McGregor & Timmermann, 2011; Zanchettin et al., 2012). However, this initial cooling may be a direct res­ponse to the volcanic forcing, which is dynamically distinct from a La Niña event (Stevenson et al., 2017). Despite the immediate cooling response to volcanic forcing, dynamical variables alluded to El Niño–like anomalies developing around 6–12 months after the eruption peak (McGregor & Timmermann, 2011). Analysis of historical CGCM simulations carried out for the Coupled Model Intercomparison Project Phase 3 and 5 (CMIP3 and CMIP5) initially suggested that the linkage between volcanoes and ENSO across CMIP models is weak (Stenchikov

et al., 2006; Ding et al., 2014). However, recent studies have highlighted that this is due to sea surface temperature signal cancellation, where the eastern equatorial Pacific surface warming of an El Niño event is offset via volcanically induced cooling (Khodri et al., 2017; Maher et al., 2015). The work of Maher et  al. (2015) and Khodri et  al. (2017) helped to reconcile the differences between previous modeling studies by focusing on dynamically relevant variables, such as SSH and relative SST (i.e. defined as deviations of SST anomalies from the tropical average 20°N–20°S) (Figure 12.4). Thus, here the terms relative El Niño–like and La Niña–like are respectively used throughout this section to refer to the warming and cooling in the eastern/central equatorial Pacific. The term relative is used to describe deviations from the tropical average (20°N–20°S; e.g. Khodri et  al., 2017). It is also noted that both the MIROC and NorESM models generate pairs of simulations, one member with volcanic forcing and the other without (e.g., Ohba et al. 2013; Pausata et al., 2015), and anomalies are calculated as differences between these simulations. The CMIP5 analysis Maher et al. (2015) and Khodri et al. (2017) suggests that a volcanic eruption produces a relative El Niño–like mean response and increases the likelihood of El Niño events occurring in the Pacific Ocean in the 6–18 month period following

EFFECT OF STRONG VOLCANIC ERUPTIONS ON ENSO  275

the peak optical depth of the eruption (Figures 12.4 and 12.5). An increased likelihood of relative La Niña events occurring in the third austral summer posteruption is also reported (i.e. 14 months after the peak of the El Niño), which may enhance the persistence of postvolcanic cooling seen in CMIP5 (Maher et al., 2015). The relative El Niño–like response of the tropical Pacific in the year after a large tropical volcanic eruption has also been confirmed in the modeling studies of Ohba et  al. (2013), Predybaylo et al. (2017), and Stevenson et al. (2017). Models can also be used to examine the climatic response to volcanic eruptions of differing magnitudes, locations, and the initial state of the Pacific prior to the eruption to help understand the dynamics underlying the modulation of ENSO (Khodri et  al., 2017; Lim et  al., 2016; McGregor & Timmermann, 2011; Ohba et  al., 2013; Predybaylo et  al., 2017; Stevenson et  al., 2017). Emile‐Geay et al. (2008) initially argued that a Pinatubo‐ sized or larger eruption is required to increase the likelihood of a posteruption El Niño occurring. A magnitude threshold was also recently presented by Lim et al. (2016), whose model results suggested that a relative El Niño–like response to large tropical eruptions only occurs past a threshold of 15 W/m2 (which is roughly two to three times the magnitude of Pinatubo). Both of these results appear to contrast the CMIP5 multimodel response (Khodri et al., 2017; Maher et al., 2015), which suggests that the smaller eruptions of the 20th century produce an El Niño–‐like response in the 6–18 months following the eruption peak (Figure 12.5). It is currently unclear whether discrepancy between model results is a signal‐to‐noise ratio issue caused by the characteristics of the modeled ENSO (i.e. models with a more variable ENSO will require a larger signal to overcome the model internal variability) and/or the experimental methods (i.e. namely the number of ensemble members of volcanic events included in the composites). The initial state of the Pacific has also shown to be very important in determining the Pacific response (Khodri et al., 2017; Ohba et  al., 2013; Predybaylo et  al., 2017). Predybaylo et  al. (2017) have modeled the dependence of ENSO’s response to a Pinatubo‐like eruption on the initial state of the Pacific, finding that a statistically significant relative El Niño–like warming (relative to an unperturbed simulation) occurs for all initial states, with the exception of a La Niña initial state. A weaker El Niño occurs in response to an eastern Pacific El Niño initial state compared to a central Pacific initial state. The results of Khodri et  al. (2017) qualitatively agree. They find that Pinatubo‐size eruptions induce an El Niño–like relative SST warming under all scenarios that tends to shorten La Niña events, lengthen El Niño events, and induce anomalous warming in the central equatorial Pacific when the initial condition is neutral. Predybaylo et  al. (2017) also investigated how the season of the eruption

affects the response, finding that the El Niño–like response is stronger for boreal summer eruptions compared to winter and spring eruptions. Thus, the El Niño–like warming seen in paleo‐proxy ENSO reconstruction in the year of the eruption (Figure 12.3) is broadly in agreement with CGCM output (Figures 12.4 and 12.5; Khodri et al., 2017; Lim et al., 2016; Ohba et al., 2013; Predybaylo et al., 2017; Stevenson et al., 2017). Idealized CGCM studies of volcanic impacts have identified numerous dynamical mechanisms to explain how the zonally uniform volcanic forcing can project onto the ENSO mode. In all cases, after this initial projection onto the ENSO mode, the Pacific basin’s Bjerknes feedback (Bjerknes, 1969) is assumed to have been initiated, which acts to reinforce the initial relative SST anomaly. Proposed mechanisms allowing volcanic forcing to modulate ENSO include dynamical thermostat, subtropical wind stress curl, and changes in land temperature. 12.4.1.1. Dynamical Thermostat The studies of Mann et  al. (2005) and Emile‐Geay et al. (2008) proposed that warming in the tropical eastern Pacific that occurs in response to a uniform surface cooling can be explained by the dynamical thermostat ­hypothesis (Clement et al., 1996). This hypothesis states that SSTs in the eastern Pacific are less sensitive to radiative forcing compared to the warm pool in the western Pacific, due to the fact that they are partly controlled by water upwelled from below the surface that is not subject to the volcanically induced cooling. Thus, uniform reductions in incoming surface solar radiation forced by a tropical volcanic eruption result in less cooling in the equatorial eastern Pacific relative to the western Pacific. This then acts to reduce the east‐west temperature gradient, which weakens the trade winds through the Bjerknes feedback and favors El Niño–like conditions. However this mechanism seems to play a negligible role when using more complex Earth System models (McGregor & Timmermann 2011; Pausata et al., 2020). 12.4.1.2. Subtropical Wind Stress Curl Both McGregor and Timmermann (2011) and Stevenson et al. (2017) emphasize the role of an initial eastern Pacific cooling that acts to generate a delayed ocean dynamical response. The enhanced cooling in the eastern equatorial Pacific during the first few months following an eruption leads to an easterly wind anomaly and associated negative wind stress curl on either side of the equator. These wind stress curl anomalies act to drive convergence of relatively warm subtropical waters onto the equator. This is similar to the recharge oscillator paradigm for ENSO (Jin, 1997; Meinen & McPhaden, 2000) and leads to an El Niño–like SST warming during the boreal winter in the year following the eruption. Stevenson et  al. (2017) further use a mixed‐layer heat budget approach to show that this

276  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

40N

Sfc air temp. anom. (K: relative to tropical mean) CMIP5 multi-model mean: Five strong volcanic eruptions composite SON(0)

20N EQ 20S 40S

0

60E

120E

180

120W

60W

0

60E

120E

180

120W

60W

0

60E

120E

180

120W

60W

0

60E

120E

180

120W

60W

0

DJF (0/1) 40N 20N EQ 20S 40S 0

MAM (1) 40N 20N EQ 20S 40S 0

JJA (1) 40N 20N EQ 20S 40S 0

–0.3

–0.2

–0.1

0

0.1

0.2

0.3

1

Figure 12.4  Composite ensemble mean seasonal average surface temperature anomalies (°C; relative to tropical mean: 20°N–20°S) and rainfall anomalies (contours at intervals of 0.2 mm/day; no zero contours) with surface wind (m/s; vector). The ensemble mean maps are calculated from 32 CMIP5 models that are each composited around five strong volcanic events (Krakatau, Santa María, Mt. Agung, El Chichón, and Pinatubo) where the eruption begins in year 0. DJF (0/1) is peak of the strong tropical volcanic forcing composite.

anomaly is indeed dynamically consistent with El Niño events in its balance of surface vs. advective heating terms. What differs between these two studies is the cause of the initial relative eastern equatorial Pacific cooling. McGregor and Timmermann (2011) identify several important mechanisms that can translate the uniform radiative forcing into a spatially inhomogeneous SST response pattern featuring eastern equatorial Pacific cooling. These are as follows:

a. Oceanic mixed layer depth. Areas with a shallower mean mixed layer depth will display a stronger cooling due to their lower heat capacity (i.e. the eastern equatorial Pacific), while regions with deep mean mixed layer depths will display a weaker cooling. This mechanism was identified by McGregor and Timmermann (2011) as playing the prominent role in the initial La Niña–like response of CCSM4 to volcanic forcing.

EFFECT OF STRONG VOLCANIC ERUPTIONS ON ENSO  277 CMIP5 multi-model mean: Five strong volcanic eruptions

0.25

Optical depth

Krakatau (1883) Santa Maria (1902) Mt Agung (1962) El Chichon (1982) Pinatubo (1991) Composite

0.2 0.15 0.1 0.05 0 0.4

Nino 3.4 index (K: relative to tropical mean TS)

0.3 0.2 0.1 0 –0.1 –0.2 0

3

6

9

12

15

18

21

24

27

30

33

36

39

42

45

months since Jan (Year 0)

Figure 12.5  Time series of CMIP5 model ensemble mean composite relative Niño‐3.4 index (bottom) in response to tropical volcanic forcing (top) for five strong tropical volcanic eruptions (Krakatau, Santa María, Mt. Agung, El Chichón, and Pinatubo) and their composite average. In addition to the normal calculation of SST anomalies, these anomalies are presented relative to the tropical mean (20°N–20°S).

b. Newtonian damping. This mechanism was introduced by Xie et al. (2010) to explain the larger warming in the central/eastern tropical Pacific in CMIP3 models in response to anthropogenic induced greenhouse warming. Under this mechanism, temperature changes in regions with cooler mean states are less damped by latent heating (i.e. the equatorial cool tongue region), which allows them to produce stronger cooling in response to volcanic forcing, relative to regions with warmer mean states (i.e. the western Pacific warm pool). The Newtonian damping component of latent heat was proposed to enhance the response of CCSM4 to volcanic forcing (McGregor & Timmermann, 2011). c. Cloud albedo impacts. Areas with high cloud cover and consequently high albedo (i.e the western Pacific warm pool) will display smaller surface solar radiation changes and surface cooling than regions with little or no cloud (i.e., the equatorial cool tongue region). This mechanism was first suggested to be of importance by the results of Stevenson et al. (2017) who identified that their modeled initial La Niña–like cooling was largely created by differential climatological‐mean cloud cover. 12.4.1.3. Changes in Land Temperature By carrying out a series of idealized volcanic forcing experiments in the Model for Interdisciplinary Research on Climate, version 5 (MIROC5), Ohba et al. (2013) first reported the emergence and extension of westerly wind anomalies in the tropical western Pacific following tropical

volcanic eruptions. The authors suggest that the source of the westerly wind anomalies in the western Pacific may be caused by the volcanically induced intense cooling of the Maritime Continent (Ohba et  al., 2013), a result that is supported by the study of Predybaylo et  al. (2017) and Eddebbar et al. (2019). In general, land surface temperature is more sensitive to volcanic radiative forcing than the ocean, due to the lower heat capacity and distinct surface and atmospheric feedbacks on land (e.g. Sutton et  al., 2007), which results in an anomalous land‐sea temperature gradient between the Maritime Continent and the western Pacific. The intensified relative cooling in response to volcanic forcing suppresses convective activity and induces westerly wind anomalies over the western Pacific. The resulting westerly wind anomalies over the western Pacific force changes in the eastern Pacific SST, via equatorial Kelvin waves, that initiate the Pacific basin Bjerknes feedback (Bjerknes, 1969), which acts to reinforce the initial relative SST anomaly. d. Khodri et  al. (2017) also identify a wind‐forced mechanism in the Pacific following a volcanic eruption that relies on strong land surface temperature cooling. However, experiments with the Institut Pierre Simon Laplace CGCM identify land surface cooling over the African continent as the prominent mechanism, rather than the Maritime Continent cooling identified in Ohba et al. (2013). The volcanically induced African continent cooling acts to weaken the convective activity of the western African monsoon, which creates atmospheric Kelvin

278  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

waves that suppress convective activity on their way and induce surface westerly wind anomalies that extend across the tropical Indian and Pacific oceans (known as Matsuno‐Gill response; Matsuno, 1966; Gill, 1980). Again, the resulting western Pacific westerly wind anomalies force changes in eastern Pacific SST that are reinforced by the initiation of the Pacific basin Bjerknes feedback (Bjerknes, 1969). The GFDL‐CM2.1 model study of Predybaylo et  al. (2017), along with the last millennium ERIK simulation study of Lim et al. (2016), also largely agree that volcanic forcing generates westerly wind anomalies in the tropical western Pacific, leading to an El Niño–like response. However, Predybaylo et al. (2017) noted that the mechanism does not appear to work with La Niña initial conditions, while they also surmise that the dynamical thermostat mechanism may aid the development of the El Niño–like anomaly after the volcanic eruption as the ocean responds more slowly to the radiative forcing compared to the continents. Results of the Lim et al. (2016) study suggest that the westerly wind anomalies that trigger the El Niño–like response result from reduced surface humidity in the Pacific basin and an associated weakened intensity of the Intertropical Convergence Zone (ITCZ). 12.4.2. Extratropical Volcanism Historically, extratropical volcanic eruptions have drawn less attention in the climate research community compared with tropical eruptions, as their radiative impacts were thought to be merely hemispheric, due to the aerosols being confined to the hemisphere of eruption, rather than global. Older studies have highlighted the potential regional impacts of large extratropical eruptions on the strength of the Asian Monsoon (Oman et al., 2005) and African rainfall (Oman et al., 2006). However, recent studies have shown the potential global impacts of extratropical eruptions through their impacts on ENSO (Liu et  al., 2018; Pausata et  al., 2015, 2016; Stevenson et  al., 2016; Zuo et  al., 2018) and oceanic circulation (Pausata et  al., 2015). The modulation of oceanic circulation via extratropical eruptions allows their effects to persist much longer than would otherwise be the case, with some work showing effects lasting for several decades (Pausata et al., 2015). Although ENSO is a key player in the climate response to eruptions, to date only two coupled climate models have been used to investigate the impacts of extratropical Northern Hemisphere (NH) volcanic eruptions on ENSO: the Norwegian Earth System Model version 1 (NorESM‐1) (Pausata et  al., 2015, 2016) and the Community Earth System Model version 1.1 (CESM1.1) (Stevenson et  al., 2016, 2017; Zuo et al., 2018) and version 1.0 (CESM1.0;

Liu et  al., 2018) (Figure  12.6). Only one model, the CESM1.1 model, has been used to examine the impact of extratropical Southern Hemisphere (SH) eruptions (Stevenson et  al., 2016). In this case, Stevenson et  al. (2016) identified a tendency for a La Niña–like response in the 12‐month period following an extratropical SH eruption (Figure 12.7b). For extratropical NH eruptions, both models identify a tendency for enhanced El Niño initiation in the boreal winter following the eruption but disagree on whether El Niño–like or La Niña–like conditions are expected the second winter (Pausata et  al., 2015; Stevenson et al., 2016; Zuo et al., 2018). 12.4.2.1. The Short-Term Response (0–1 years) Only one main mechanism has been highlighted as potentially driving the immediate ENSO response to extratropical volcanic eruptions: that is the migration of the ITCZ away from its climatological position. Extratropical volcanic eruptions are characterized by a sulfate aerosol distribution that is confined to the hemisphere in which the eruption occurs. It is well established that an interhemispherically asymmetric forcing pushes the ITCZ away from the hemisphere that is cooled (Kang et al., 2008; Schneider et al., 2014). This has been demonstrated in coupled modeling studies of extratropical eruptions, including simulations of Iceland’s Laki eruption in 1783 (Pausata et al., 2015), and simulations of extratropical eruptions covering the whole of the last millennium (Colose et al., 2016; Stevenson et al., 2017). An equatorward migration of the ITCZ, which occurs in response to a NH eruption, has a tendency to favor the initiation of El Niño events (Pausata et al., 2015). Because surface easterly winds are weakest in the proximity of the ITCZ, this equatorward shift implies a weakening of the easterly winds along the equator in the central and eastern equatorial Pacific. Through the Bjerknes feedback (Bjerknes, 1969), such weakening of the trades leads to a reduction in the east‐west temperature contrast across the tropical Pacific, thus favoring the development of an El Niño–like anomaly (Figure 12.7a). El Niño–like anomalies are also influenced by the preexisting ENSO state: a stronger El Niño–like response develops when the preexisting background state is La Niña compared with El Niño (Pausata et al., 2016). This asymmetry of the response can be easily understood since radiative‐convective equilibrium imposes an upper limit on the absolute intensity that an El Niño event can reach (Jin et al., 2003). Therefore, there is more room for large anomalies to develop when starting from neutral or La Niña conditions. Nevertheless, there are less obvious effects at play. For example, during an incipient La Niña or under neutral conditions, the ITCZ is farther north and the trades on the equator are stronger relative to the incipient El Niño case. Therefore, a nominal zonal wind

EFFECT OF STRONG VOLCANIC ERUPTIONS ON ENSO  279 (a)

(b)

CESM

42

°C (SSTA Relative to tropical mean)

NorESM

42

1

36

0.8

36

Months since Jan (year 0)

0.6 0.4

30

30

0.2 24

24

18

18

0 –0.2 –0.4

12

–0.6

12

–0.8 6

6 100E

150E

200E

250E

Longitude

100

150

200

250

–1

Longitude

Figure 12.6  Hovmoller plot of relative sea surface temperature anomalies (SSTA, shading) and zonal wind stress anomalies (contours) for (a) CESM‐LME and (b) NorESM Earth system models. The contour intervals follow intervals of 0.05 N/m2 (solid: positive anomalies; dashed: negative anomalies).

anomaly along the equator will trigger a larger change in wind stress under incipient La Niña than incipient El Niño conditions due to the quadratic relationship between wind speed and wind stress. The potential role of the ITCZ in driving the ENSO response in the first winter after extratropical NH eruptions was also suggested by Stevenson et al. (2016), who show an equatorward migration of ITCZ after an eruption and the consequent development of El Niño– like conditions. In other older studies (Highwood & Stevenson, 2003; Oman et al., 2005, 2006) that have investigated the impacts of extratropical volcanic eruptions on climate, an examination of the ENSO response was not possible due to the lack of a fully coupled ocean model and the resulting inability to diagnose coupled feedbacks. However, Oman et  al. (2005, 2006), using the Goddard Institute for Space Studies Model E coupled with a mixed layer ocean, show a weakening of the Asian and African monsoon following the Laki eruption, suggesting a southward shift of the ITCZ and hence favorable conditions for a development of an El Niño. The response to SH eruptions in CESM appears nearly opposite to the NH eruption response during the eruption year. That is, the SH eruptions cause a northward migration of the ITCZ along with a relatively strong initial cooling peaking in the boreal winter following the eruption

(Figure  12.7b). However, the results of Stevenson et  al. (2016) suggest that it is not likely that this response has dynamics consistent with a La Niña event based on the analyses of the full spatial structure of SST anomalies (Stevenson et al., 2016). A return to ENSO‐neutral conditions follows, lasting roughly for the next 18 months, at which point additional cooling in the tropical Pacific occurs. 12.4.2.2. The Medium-Term Response (2–5 years) As with the tropical volcanism, the longer‐term response of ENSO to extratropical eruptions has not been extensively studied. However, there is support from modeling‐based studies for a La Niña–like response during years 2–5 following the event (Pausata et  al., 2015, 2016; Stevenson et al., 2017). The only real disagreement between modeling studies is whether the La Niña–like response occurs in years 2–4 following the event (Pausata et  al., 2015, 2016) or years 3–5 following the event (Stevenson et  al., 2017; Figures  12.6 and 12.7). This difference stems from CESM producing unusual back‐to‐ back El Niño–like warming following the eruption while the NorESM model produces a more common 1‐year El Niño–like warming. Much of this La Niña–like response may represent a recovery from El Niño–like conditions during the previous year, as La Niña conditions generally follow El Niño events (Kessler, 2002; Ohba & Ueda,

280  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a)

NINO3.4 relative SSTA responses to high-latitude eruptions

NINO3.4 SSTA (°C)

2 1 0 –1

NorESM 90% range CESM 90% range NorESM mean CESM mean

–2 0

6

12

18

24

30

36

42

48

Months since Jan (Year 0)

(b)

NINO3.4 relative SSTA responses to high-latitude SH eruptions CESM 90% range CESM mean

NINO3.4 SSTA (°C)

2 1 0 –1 –2 0

6

12

18

24

30

36

42

48

Months since Jan (Year 0)

Figure 12.7  Percentile 90% ranges of model ensemble composites of the relative Niño‐3.4 index in response to strong extratropical (top) Northern Hemisphere and (bottom) Southern Hemisphere volcanic forcing.

2009; Okumura & Deser, 2010). Further modeling studies are needed to test the proposed dynamical processes and to verify whether such initial condition influences are sufficient to explain the entirety of this response. 12.4.2.3. Longer-Term Modulation of ENSO Behavior (5–50 years) The Atlantic meridional overturning circulation (AMOC) significantly intensifies 5 to 25 years after extratropical eruptions, mainly due to the large oceanic cooling, followed by an AMOC weakening lasting approximately two decades (Pausata et al., 2015). Several modeling and observation‐based studies have suggested that AMOC changes can modify the strength of ENSO (Dong & Sutton, 2002; Levine et al., 2017, 2018; Pausata et al., 2015; Sutton et al., 2007; Timmermann et al., 2007; Zhang & Delworth, 2005). This connection is thought to be largely due to large‐scale atmospheric circulation changes, initiated via AMOC‐induced SST gradient changes in the tropical Atlantic, modulating the tropical Pacific and ENSO with very little lag time (Polo et  al.,

2014; Ruprich‐Robert et  al., 2017). Additionally, on a timescale of a few decades, the thermocline signals associated with changes in AMOC strength can also be transmitted from the North Atlantic to the tropical Pacific through oceanic waves (e.g. Timmermann et al., 2005). The modeled interbasin connections suggesting that strong AMOC variability leads to weaker than normal ENSO variability are physically plausible (Timmermann et al., 2007). However, there is limited observational evidence to support this interbasin ENSO modulation, and a small number of modeling studies raise questions as to whether ENSO variability is positively or negatively correlated to AMOC strength (Atwood, 2015). In the case of extratropical eruptions, only one study is currently available, which looks at the modulation of ENSO after a volcanic eruption and under modern conditions (Pausata et al., 2015). Pausata et al. (2015) showed that an increase in ENSO variance occurs with AMOC strengthening, while decreases in ENSO variance occur with AMOC weakening (Table 12.3). The increased ENSO variability during strong AMOC periods is thought to be due to a

EFFECT OF STRONG VOLCANIC ERUPTIONS ON ENSO  281

shoaling and flattening of the equatorial Pacific thermocline, which boosts the strength of the Bjerknes feedback (Pausata et  al., 2015; Russell & Gnanadesikan, 2014). More studies investigating the link between AMOC changes and ENSO variability following volcanic eruptions are needed in order to better quantify the long‐­ lasting impact on climate beyond the lifetime of the injected stratospheric aerosols. 12.5. DISCUSSION AND CONCLUSIONS Volcanic eruptions have strong radiative and climate effects, and what has become apparent is that accurate details of this forcing are required to determine the magnitude of the climatic response. Observational platforms currently exist that provide sufficient details to allow for the accurate simulation of the volcanically induced climate response. For volcanic eruptions of the past (e.g. pre‐1990s), however, the eruption timing along with details of sulfate aerosol size and distribution in the stratosphere must be reconstructed (Sigl et al., 2015). Much progress has been made since PMIP3 and CMIP5 to improve past volcanic forcing reconstructions that rely on global aerosol models and improved accuracy in ice core analyses (Toohey et al., 2016). However, questions remain about how accurately the timing of the volcanic eruption and its forcing properties can be reconstructed from sulfates deposited on ice sheets (Marshall et  al., 2018). One way forward to aid better estimates of past volcanic forcing magnitudes is to compare results of standardized model experiments with a range of instrumental observations and proxy records (Timmreck et al., 2018; Zanchettin et al., 2016). Hence, to improve our understanding of the impact of volcanic forcing on ENSO, it is necessary to (i) continue refining reconstructions of past ENSO variability, which includes developing more independent reconstructions of ENSO‐teleconnected vs. “center of action” responses (Stevenson et al., 2016) and (ii) continue using models to explore how the magnitude and timing of the volcanic radiative forcing influence the response of ENSO. We also note that both the seasonally synchronized nature

of ENSO (e.g. Nicholls, 2008) and the seasonality of the radiative effects of volcanic forcing (e.g. Toohey et al., 2011) have the potential to modulate the magnitude of ENSO’s response to volcanic forcing, but these have either generally been explored in isolation to date, or the impacts of both have not been separated. Despite some discrepancies found analyzing studies  of the response of ENSO to volcanism with ­paleoclimate data (e.g. Adams et al., 2003; Li et al., 2013), we show that the majority of ENSO reconstructions (12 out of 17 reconstructions) display an El Niño–like warming in the year of eruption when provided with consistent dates of volcanic eruptions. Of equal importance, we also show that none of the ENSO reconstructions display a significant La Niña–like response when provided with consistent dates of volcanic eruptions. However, 3 of the 12 reconstructions that display a significant El Niño–like warming in the eruption year also display significant La Niña–like conditions 2 years prior. This suggests that at least a portion of this warming signal observed in these three reconstructions may be due to the oscillatory nature of ENSO rather than a response to volcanic forcing. We also identify what appears to be an emerging consensus from the numerous CGCM studies investigating the impact of tropical volcanism on ENSO, with the overwhelming majority displaying an El Niño–like warming occurring in the year following a tropical eruption. Thus, we can report that there appears to be a clear consistency of evidence between the model, paleoproxies, and observations. However, despite the consistent responses, the CMIP5 model response is notably weaker than that seen in the observational composite (compare Figures  12.1a and 12.5b). It is currently unclear whether the magnitude of the response is incorrectly modeled, or whether the coincident occurrence of several eruptions with El Niño events already underway underlies this difference (e.g. Nicholls, 1988). The former has the potential to decrease the signal‐to‐noise ratio, making the volcanic response harder to identify because of the large intrinsic variations in ENSO behavior.

Table 12.3  Ensemble average standard deviation change of monthly mean SST (°C) calculated in the Niño‐3.4 region for the period 5–25 yrs, 26–45 yrs, and 46–60 yrs after the eruption. As is commonly done when examining ENSO, we apply a 5‐month running mean on the SST anomalies to damp high‐frequency ocean variability unrelated to ENSO. The last column shows the composite of the changes in Niño‐3.4 SST standard deviation between strong and weak phases of the AMOC in simulations without volcanoes. The strong and weak AMOC periods are defined as periods in which the AMOC maximum anomalies are outside the 15th–85th percentiles for at least 48 consecutive months. This difference should be compared to the differences in the periods 5–25 years and 26–45 years after the eruption (i.e. the difference between the first [strong AMOC anomalies] and the second [weak AMOC anomalies] row [+0.213°C]). ΔSTDSST (°C)

5–25 yrs

26–45 yrs

46–60 yrs

AMOCstrong–AMOCweak

+0.068

−0.145

−0.033

+0.202

282  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

A La Niña–like cooling is seen to follow the initial volcanically induced El Niño–like warming in observations, paleo‐proxies, and model simulations (Figures 12.1, 12.3, and 12.5). This response is consistent with ENSO dynamics, whereby La Niña events tend to trail El Niño events (e.g. Ohba & Ueda, 2009; Okumura & Deser, 2010). This suggests that volcanic events not only impact ENSO in the year following the eruption but that their effects can continue up to 2–4 years after the eruption. Given that the prediction of ENSO’s climate impacts is vital for effective management of climate disasters, this lagged response of ENSO to tropical volcanic forcing suggests that strong volcanic events may provide the conditions required for long lead‐time prediction of climate variability (e.g. Luo et al., 2017). While the observed El Niño–like response is also seen in most CGCMs, questions remain over the exact dynamical causes of the response. As discussed in section 12.4.1, many mechanisms have been proposed to explain the projection of volcanically induced radiative cooling onto the ENSO mode, and different models appear to favor different mechanisms. It is currently unclear whether this apparent discrepancy stems from differing experimental design or differing model dynamics and parameterizations. Nevertheless, differing model dynamics and parameterizations will play some role in determining exactly how volcanic forcing will project onto ENSO, as there is a large diversity in ENSO properties and dynamics among the state‐of‐the‐art CGCMs included in the CMIP5 models (Bellenger et al., 2014; Collins et al., 2010; Guilyardi et  al., 2012; Vijayeta & Dommenget, 2018). The dynamical processes of ENSO are affected by modeled mean climate (i.e. zonal gradient of equatorial Pacific SST) and by multiple physical feedbacks (such as thermocline feedback and zonal advective feedback). What results is ENSO variability, that is, a delicate balance of amplifying and damping feedbacks, the combination of which results in the modeled ENSO characteristics (amplitude or frequency of ENSO events in the models) (Bellenger et al., 2014; Collins et al., 2010). It is possible for volcanic forcing to impact multiple feedbacks (e.g. altering SST gradients and the related surface wind feedback, while also altering ocean stratification, which may impact the connection between the ocean subsurface and surface); thus, the modeled response may be due to the sum or balance of these feedback projections. Given that the balance of feedbacks is different across models (e.g. Lloyd et al., 2011), exactly how ENSO responds to volcanism may also differ between models. Only a handful of studies have investigated the impact of extratropical volcanic eruptions on ENSO to date (Liu et al., 2018; Pausata et al., 2015, 2016; Stevenson et al., 2016; Zuo et  al., 2018). These studies show fairly con­ sistent results, simulating El Niño–like conditions

­eveloping within the first year after NH eruptions d (Pausata et al., 2015; Stevenson et al., 2016). Cool conditions are more likely in the equatorial Pacific after SH eruptions, although this response may not be dynamically a La Niña–like response (Stevenson et  al., 2016). Shifts in the ITCZ due to cooling of the eruption hemisphere have been proposed as main driver of the ENSO response (Pausata et al., 2015; Stevenson et al., 2016). A La Niña–like response in the years following volcanically induced El Niño–like warming is also seen in response to extratropical volcanism, which is consistent with expectations from ENSO dynamics (Jin, 1997) and observations (e.g. Ohba & Ueda, 2009; Okumura & Deser, 2010). The difference between the two studies, however, is whether the initial volcanically induced El Niño–like response persists for 1 or 2 years prior to the La Niña–like cooling. Regardless of this, these results suggest that extratropical volcanism can directly influence ENSO for about 1 to 4 years after an eruption. Pausata et al. (2015) also showed the potential for large high‐latitude eruptions to modulate ENSO variability out to roughly 50 years after the eruption, whereby amplitude is enhanced during the first half of the period, then damped in the remaining period. This modulation of ENSO variance is thought to be underpinned via interbasin connections between the Atlantic and Pacific basins that are initiated by volcanically induced changes in Atlantic Ocean circulation and heat content. As similar longer‐term changes in ocean circulation and heat content are reported to follow tropical eruptions (Church et  al., 2005; Swingedouw et  al., 2015; Zanchettin et  al., 2012), we would expect to see similar posteruption modulation of ENSO variability. However, this has not yet been investigated. There are also potentially intriguing interactions between anthropogenic greenhouse gas–induced ENSO changes (e.g., Cai et  al., 2015, 2018; Power et  al., 2013) and those induced by volcanism, which are yet to be investigated (e.g. Fasullo et  al., 2017). However, due to the short observational record, contradictory model responses to interbasin SST changes (e.g. Timmermann et al., 2007), and the possible impact of mean state biases on this relationship (e.g. McGregor et  al., 2018), questions remain over exactly if and how ENSO variability may be modulated after both extratropical and tropical eruptions. Despite the many similarities detailed above, many subtle differences still exist between the modeled responses to tropical and extratropical volcanic forcing (i.e. the mechanism(s) underlying the El Niño–like response to tropical volcanism). The causes of these discrepancies currently remain unclear due to intermodel physical and dynamical differences (e.g. Collins et  al., 2010); differing experimental designs, including the number of ensemble members and methods of constructing the ensembles (e.g. McGregor & Timmermann, 2011; Ohba et al., 2013);

EFFECT OF STRONG VOLCANIC ERUPTIONS ON ENSO  283

and the volcanic forcing details, including magnitude (Timmreck, 2012) and season (Kravitz & Robock, 2011; Stevenson et al., 2017; Toohey et al., 2011), as well as the aerosol optical properties associated with sulfate injection (Zanchettin et  al., 2016). A consistent experimental protocol with a large number of models, such as idealized volcanic‐perturbation experiments of VolMIP (Zanchettin et al., 2016), will allow future research to focus on response differences that stem from intermodel physical and dynamical differences. The VolMIP protocol is currently being carried out by many modeling centers as a CMIP6‐ endorsed project. As part of this project, ensemble simulations are conducted sampling a range of appropriate initial conditions and using the same idealized strong equatorial and hemispheric volcanic forcing data. We expect that analysis of the VolMIP project data will allow us to generate clearer understanding of the ENSO‐volcanic relationship and to better understand exact dynamical causes of the response and its intermodel diversity. ACKNOWLEDGMENTS The authors would like to thank two anonymous reviewers for their constructive comments. SM acknowledges funding support from the Australian Research Council through grant number FT160100162. NM was supported by the Max Planck Society for the Advancement of Science and the Alexander von Humboldt Foundation. SS acknowledges support from the US National Science Foundation through award number AGS-1805143. F.S.R.P. acknowledges the financial support from the Natural Sciences and Engineering Research Council of Canada (grant RGPIN-201804981) and the Fonds de recherche du Québec–Nature et technologies (2020-NC-268559). REFERENCES Adams, J. B., Mann, M. E., & Ammann, C. M. (2003). Proxy evidence for an El Niño‐like response to volcanic forcing. Nature, 426(6964), 274–278. https://doi.org/10.1038/ nature02101 Anchukaitis, K. J., Buckley, B. M., Cook, E. R., Cook, B. I., D’Arrigo, R. D., & Ammann, C. M. (2010). Influence of volcanic eruptions on the climate of the Asian monsoon region. Geophysical Research Letters, 37(22). https://doi. org/10.1029/2010GL044843 Anchukaitis, K. J., Breitenmoser, P., Briffa, K. R., Buchwal, A., Büntgen, U., Cook, E. R., et  al. (2012). Tree rings and volcanic cooling. Nature Geoscience, 5, 836. Retrieved from https://doi.org/10.1038/ngeo1645 Atwood, A. R. (2015). Mechanisms of tropical Pacific climate change during the Holocene. University of Washington. Bellenger, H., Guilyardi, E., Leloup, J., Lengaigne, M., & Vialard, J. (2014). ENSO representation in climate models:

From CMIP3 to CMIP5. Climate Dynamics, 42(7–8), 1999– 2018. https://doi.org/10.1007/s00382‐013‐1783‐z Bjerknes, J. (1969). Atmospheric teleconnections from the equatorial Pacific. Monthly Weather Review, 97(3), 163–172. https://doi.org/10.1175/1520‐0493(1969)0972.3.CO;2 Bradley, R. S., & Jones, P. D. (1993) “Little Ice Age” summer temperature variations: Their nature and relevance to recent global warming trends. Holocene, 3, 367–376. doi:10.1177/ 095968369300300409 Braganza, K., Gergis, J. L., Power, S. B., Risbey, J. S., & Fowler, A. M. (2009). A multiproxy index of the El Niño‐Southern Oscillation, A.D. 1525‐1982. Journal of Geophysical Research Atmospheres, 114(5), 1–17. https://doi.org/10.1029/2008JD 010896 Briffa, K. R., Jones, P. D., Schweingruber, F. H., & Osborn, T. J. (1998). Influence of volcanic eruptions on Northern Hemisphere summer temperature over the past 600 years. Nature, 393, 450. Retrieved from https://doi.org/10.1038/30943 Cai, W., Santoso, A., Wang, G., Yeh, S. W., An, S. Il, Cobb, K. M., et al. (2015). ENSO and greenhouse warming. Nature Climate Change, 5(9), 849–859. https://doi.org/10. 1038/nclimate2743 Cai, W., Wang, G., Dewitte, B., Wu, L., Santoso, A., Takahashi, K., et al. (2018). Increased variability of eastern Pacific El Niño under greenhouse warming. Nature, 564(7735), 201–206. https://doi.org/10.1038/s41586‐018‐0776‐9 Christiansen, B. (2008). Volcanic Eruptions, Large‐Scale Modes in the Northern Hemisphere, and the El Niño–Southern Oscillation. Journal of Climate, 21(5), 910–922. https://doi. org/10.1175/2007JCLI1657.1 Church, J. A., White, N. J., & Arblaster, J. M. (2005). Significant decadal‐scale impact of volcanic eruptions on sea level and ocean heat content. Nature. https://doi.org/10.1038/ nature04237 Clement, A. C., Seager, R., Cane, M. A., & Zebiak, S. E. (1996). An ocean dynamical thermostat, J. Clim., 9, 2190–2196. Collins, M., An, S.‐I., Cai, W., Ganachaud, A., Guilyardi, E., Jin, F.‐F., et al. (2010). The impact of global warming on the tropical Pacific Ocean and El Niño. Nature Geoscience, 3(6), 391–397. https://doi.org/10.1038/ngeo868 Colose, C. M., LeGrande, A. N., & Vuille, M. (2016). Hemispherically asymmetric volcanic forcing of tropical hydroclimate during the last millennium. Earth System Dynamics, 7(3), 681–696. https://doi.org/10.5194/esd‐7‐ 681‐2016 Cook, E. R. (2000). Niño 3 index reconstruction. In International Tree‐Ring Data Bank, IGBP PAGES/World Data Center for Paleoclimatology, Data Contribution Series #2000‐052. Boulder CO: NOAA/NGDC. Cook, E. R., D’Arrigo, R. D., & Anchukaitis, K. J. (2008). ENSO reconstructions from long tree ring chronologies: Unifying the differences? Presented at special workshop, “Reconciling ENSO Chronologies for the Past 500 Years,” Moorea, French Polynesia, 2–3 April 2008. Crowley, T. J., Zielinski, G. A., Vinther, B., Udisti, R., Kreutz, K., Cole‐Dai, J., & Castellano, E. (2008). Volcanism and the Little Ice Age. PAGES News, 16(2), 22–23. https://doi. org/10.1029/2002GL0166335.Hegerl

284  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE D’Arrigo, R., Wilson, R., & Jacoby, G. (2006). On the long‐ term context for late twentieth century warming. Journal of Geophysical Research: Atmospheres, 111(D3). https://doi. org/10.1029/2005JD006352 Ding, Y., Carton, J. A., Chepurin, G. A., Stenchikov, G., Robock, A., Sentman, L. T., & Krasting, J. P (2014). Ocean response to volcanic eruptions in Coupled Model Intercomp­ arison Project 5 simulations. J. Geophys. Res. Oceans,119, 5622–5637. doi:10.1002/2013JC009780 Dong, B.‐W., & Sutton, R. T. (2002). Adjustment of the coupled ocean‐atmosphere system to a sudden change in the Thermohaline Circulation. Geophysical Research Letters, 29(15), 18‐1‐18–4. https://doi.org/10.1029/2002GL015229 Driscoll, S., Bozzo, A., Gray, L. J., Robock, A., & Stenchikov, G. L. (2012). Coupled Model Intercomparison Project 5 (CMIP5) simulations of climate following volcanic eruptions. Journal of Geophysical Research Atmospheres, 117(17), n/a‐n/a. https://doi.org/10.1029/ 2012JD017607 Eddebbar, Y. A., Rodgers, K. B., Long, M. C., Subramanian, A. C., Xie, S.‐P., & Keeling, R. F. (2019). El Niño–Like Physical and Biogeochemical Ocean Response to Tropical Eruptions. Journal of Climate, 32(9), 2627–2649. https://doi.org/10.1175/ JCLI‐D‐18‐0458.1 Emile‐Geay, J., Seager, R., Cane, M. A., Cook, E. R., & Haug, G. H. (2008). Volcanoes and ENSO over the past millennium. Journal of Climate, 21(13), 3134–3148. https://doi. org/10.1175/2007JCLI1884.1 Emile‐Geay, J., Cobb, K. M., Mann, M. E., & Wittenberg, A. T. (2013a). Estimating central equatorial Pacific SST variability over the past millennium. Part I: Methodology and validation. Journal of Climate, 26(7), 2302–2328. https://doi. org/10.1175/JCLI‐D‐11‐00510.1 Emile‐Geay, J., Cobb, K. M., Mann, M. E., & Wittenberg, A. T. (2013b). Estimating central equatorial Pacific SST variability over the past millennium. Part II: Reconstructions and implications. Journal of Climate, 26(7), 2329–2352. https://doi. org/10.1175/JCLI‐D‐11‐00511.1 Evans, M. N., Cane, M. A, Schrag, D. P., Kaplan, A, Linsley, B. K., Villalba, R., & Wellington, G. M. (2001). Support for tropically driven Pacific decadal variability based on paleoproxy evidence. Geophys. Res. Lett, 28(19), 3689–3692. Evans, M. N., Kaplan, A, & Cane, M. A. (2002). Pacific sea surface temperature field reconstruction from coral δ18O data using reduced space objective analysis. Paleoceanography, 17(1), 1007. https://doi.org/10.1029/2000PA000590 Fasullo, J. T., Tomas, R., Stevenson, S., Otto‐Bliesner, B., Brady, E., & Wahl, E. (2017). The amplifying influence of increased ocean stratification on a future year without a summer. Nature Communications, 8(1). https://doi.org/10.1038/ s41467‐017‐01302‐z Fischer, E. M., Luterbacher, J., Zorita, E., Tett, S. F. B., Casty, C., & Wanner, H. (2007). European climate response to tropical volcanic eruptions over the last half millennium. Geophysical Research Letters, 34(5). https://doi.org/10.1029/ 2006GL027992 Fyfe, J. C., Gillett, N. P., & Zwiers, F. W. (2013). Overestimated global warming over the past 20 years. Nature Climate Change, 3(9), 767–769. https://doi.org/10.1038/nclimate1972

Gao, C., Robock, A., & Ammann, C. (2008). Volcanic forcing of climate over the past 1500 years: An improved ice core‐ based index for climate models. Journal of Geophysical Research Atmospheres, 113(23), 1–15. https://doi. org/10.1029/2008JD010239 Gill, A. E. (1980). Some simple solutions for heat induced tropical circulation. Quarterly Journal of the Royal Mete­ orological Society, 106(449), 447–462. https://doi.org/10.1002/ qj.49710644905 Guilyardi, E., Bellenger, H., Collins, M., Ferrett, S., Cai, W., & Wittenberg, A. (2012). A first look at ENSO in CMIP5. Clivar Exchanges, 17(58), 29–32. Retrieved from http://www. uib.no/People/ngfhd/EarthClim/Publications/Papers/ Guilyardi_etal_2012.pdf Handler, P. (1984). Possible association of stratospheric aerosols and El Nino type events. Geophysical Research Letters, 11, 1121–1124. https://doi.org/10.1029/GL011i011p01121 Harshvardhan. (1979). Perturbation of the zonal radiation balance by a stratospheric aerosol layer. Journal of the Atmospheric Sciences. https://doi.org/10.1175/1520‐0469(197 9)0362.0.CO;2 Haywood, J. M., Jones, A., & Jones, G. S. (2014). The impact of volcanic eruptions in the period 2000‐2013 on global mean temperature trends evaluated in the HadGEM2‐ES climate model. Atmospheric Science Letters, 15, 92–96. https://doi. org/10.1002/asl2.471 Highwood, E.‐J., & Stevenson, D. S. (2003). Atmospheric impact of the 1783–1784 Laki Eruption: Part II. Climatic effect of sulphate aerosol. Atmospheric Chemistry and Physics, 3(4), 1177–1189. https://doi.org/10.5194/acp‐3‐ 1177‐2003 Huang, B., Thorne, P. W., et al. (2017). Extended Reconstructed Sea Surface Temperature version 5 (ERSSTv5): Upgrades, validations, and intercomparisons. J. Climate. doi: 10.1175/ JCLI‐D‐16‐0836.1 Jin, F.‐F. (1997). An Equatorial ocean recharge paradigm for ENSO. Part II: A stripped‐down coupled model. Journal of the Atmospheric Sciences, 54(7), 830–847. https://doi.org/10.1 175/1520‐0469(1997)0542.0.CO;2 Jin, F.‐F., An, S.‐I., Timmermann, A., & Zhao, J. (2003). Strong El Niño events and nonlinear dynamical heating. Geophysical Research Letters, 30(3), 20–21. https://doi.org/10.1029/ 2002GL016356 Kang, S. M., Held, I. M., Frierson, D. M. W., & Zhao, M. (2008). The response of the ITCZ to extratropical thermal forcing: Idealized slab‐ocean experiments with a GCM. Journal of Climate, 21(14), 3521–3532. https://doi. org/10.1175/2007JCLI2146.1 Kessler, W. S. (2002). Is ENSO a cycle or a series of events? Geophysical Research Letters, 29(23), 40‐1‐40–4. https://doi. org/10.1029/2002GL015924 Khodri, M., Izumo, T., Vialard, J., Janicot, S., Cassou, C., Lengaigne, M., et al. (2017). Tropical explosive volcanic eruptions can trigger El Ninõ by cooling tropical Africa. Nature Communications, 8(1), 1–13. https://doi.org/10.1038/s41467‐ 017‐00755‐6 Kodera, K. (1994). Influence of volcanic eruptions on the troposphere through stratospheric dynamical processes in the northern hemisphere winter. Journal of Geophysical Research:

EFFECT OF STRONG VOLCANIC ERUPTIONS ON ENSO  285 Atmospheres, 99(D1), 1273–1282. https://doi.org/10.1029/ 93JD02731 Kravitz, B., & Robock, A. (2011). Climate effects of high‐latitude volcanic eruptions: Role of the time of year. Journal of Geophysical Research: Atmospheres, 116(D1). https://doi. org/10.1029/2010JD014448 Levine, A. F. Z., McPhaden, M. J., & Frierson, D. M. W. (2017). The impact of the AMO on multidecadal ENSO variability. Geophysical Research Letters, 44(8), 3877–3886. https://doi. org/10.1002/2017GL072524 Levine, A. F. Z., Frierson, D. M. W., & McPhaden, M. J. (2018). AMO forcing of multidecadal Pacific ITCZ variability. Journal of Climate, 31(14), 5749–5764. https://doi.org/10. 1175/JCLI‐D‐17‐0810.1 Li, J., Xie, S. P., Cook, E. R., Huang, G., D’arrigo, R., Liu, F., et al. (2011). Interdecadal modulation of El Niño amplitude during the past millennium. Nature Climate Change, 1(2), 114–118. https://doi.org/10.1038/nclimate1086 Li, J., Xie, S. P., Cook, E. R., Morales, M. S., Christie, D. A., Johnson, N. C., et al. (2013). El Niño modulations over the past seven centuries. Nature Climate Change, 3(9), 822–826. https://doi.org/10.1038/nclimate1936 Lim, H.‐G., Yeh, S.‐W., Kug, J.‐S., Park, Y.‐G., Park, J.‐H., Park, R., & Song, C.‐K. (2016). Threshold of the volcanic forcing that leads the El Niño‐like warming in the last ­millennium: Results from the ERIK simulation. Climate Dynamics, 46(11), 3725–3736. https://doi.org/10.1007/ s00382‐015‐2799‐3 Liu, F., Li, J., Wang, B., Liu, J., Li, T., Huang, G., & Wang, Z. (2018). Divergent El Niño responses to volcanic eruptions at different latitudes over the past millennium. Climate Dynamics, 50(9–10), 3799–3812. https://doi.org/10.1007/ s00382‐017‐3846‐z Lloyd, J., Guilyardi, E., & Weller, H. (2011). The role of atmosphere feedbacks during ENSO in the CMIP3 models. Part II: using AMIP runs to understand the heat flux feedback mechanisms. Climate Dynamics, 37(7), 1271–1292. https://doi.org/10.1007/s00382‐010‐0895‐y Luo, J.‐J., Liu, G., Hendon, H., Alves, O., & Yamagata, T. (2017). Inter‐basin sources for two‐year predictability of the multi‐year La Niña event in 2010–2012. Scientific Reports, 7(1), 2276. https://doi.org/10.1038/s41598‐017‐ 01479‐9 Maher, N., McGregor, S., England, M. H., & Gupta, A. S. (2015). Effects of volcanism on tropical variability. Geophysical Research Letters, 42(14). https://doi.org/10.1002/ 2015GL064751 Mann, M. E., Bradley, R. S., & Hughes, M. K. (2000). Long‐ term variability in the El Nino Southern Oscillation and associated teleconnections. In Diaz, H. F.., & Markgraf, V. (Eds.), El Nino and the Southern Oscillation: Multiscale variability and its impacts on natural ecosystems and society (Vol. 2, pp. 321–372). Cambridge University Press. Mann, M. E., Cane, M. A., Zebiak, S. E., & Clement, A. (2005). Volcanic and solar forcing of the tropical Pacific over the past 1000 years. Journal of Climate, 18(3), 447–456. https://doi. org/10.1175/JCLI‐3276.1 Mann, M. E., Fuentes, J. D., & Rutherford, S. (2012). Underestimation of volcanic cooling in tree‐ring‐based

reconstructions of hemispheric temperatures. Nature Geoscience, 5, 202. Retrieved from https://doi.org/10.1038/ ngeo1394 Marshall, L., Schmidt, A., Toohey, M., Carslaw, K. S., Mann, G. W., Sigl, M., et al. (2018). Multi‐model comparison of the volcanic sulfate deposition from the 1815 eruption of Mt. Tambora. Atmospheric Chemistry and Physics, 18(3), 2307– 2328. https://doi.org/10.5194/acp‐18‐2307‐2018 Matsuno, T. (1966). Quasi‐geostrophic motions in the equatorial area. J Meteor Soc Jpn, 44, 25–43. McGregor, S., & Timmermann, A. (2011). The effect of explosive tropical volcanism on ENSO. Journal of Climate, 24(8). https://doi.org/10.1175/2010JCLI3990.1 McGregor, S., Timmermann, A., & Timm, O. (2010). A unified proxy for ENSO and PDO variability since 1650. Climate of the Past, 6(1). https://doi.org/10.5194/cp‐6‐1‐2010 McGregor, S., Stuecker, M. F., Kajtar, J. B., England, M. H., & Collins, M. (2018). Model tropical Atlantic biases underpin diminished Pacific decadal variability. Nature Climate Change. https://doi.org/10.1038/s41558‐018‐0163‐4 Medhaug, I., Stolpe, M. B., Fischer, E. M., & Knutti, R. (2017). Reconciling controversies about the ‘global warming hiatus.’ Nature, 545, 41. Retrieved from http://dx.doi.org/10.1038/ nature22315 Meinen, C. S., & McPhaden, M. J. (2000). Observations of warm water volume changes in the equatorial Pacific and their relationship to El Niño and La Niña. Journal of Climate, 13(20), 3551–3559. https://doi.org/10.1175/1520‐0442(2000)0 132.0.CO;2 Morice, C. P., Kennedy, J. J., Rayner, N. A., & Jones, P. D. (2012). Quantifying uncertainties in global and regional temperature change using an ensemble of observational estimates: The HadCRUT4 dataset. Journal of Geophysical Research, 117, D08101. doi:10.1029/2011JD017187 Nicholls, N. (1988). Low latitude volcanic eruptions and the El Niño‐Southern Oscillation. Journal of Climatology, 8(1), 91– 95. https://doi.org/10.1002/joc.3370080109 Nicholls, N. (2008). Recent trends in the seasonal and temporal  behaviour of the El Niño–Southern Oscillation. Geophysical Research Letters, 35(19). https://doi.org/10.1029/ 2008GL034499 Niemeier, U., Schmidt, H., & Timmreck, C. (2010). The dependency of geoengineered sulfate aerosol on the emission strategy. Atmospheric Science Letters, 12(2), 189–194. https:// doi.org/10.1002/asl.304 Ohba, M., & Ueda, H. (2009). Role of nonlinear atmospheric response to SST on the asymmetric transition process of ENSO. Journal of Climate, 22(1), 177–192. https://doi. org/10.1175/2008JCLI2334.1 Ohba, M., Shiogama, H., Yokohata, T., & Watanabe, M. (2013). Impact of strong tropical volcanic eruptions on ENSO simulated in a coupled GCM. Journal of Climate, 26(14), 5169–5182. https://doi.org/10.1175/JCLI‐D‐ 12‐00471.1 Okumura, Y. M., & Deser, C. (2010). Asymmetry in the duration of El Niño and La Niña. Journal of Climate, 23(21), 5826–5843. https://doi.org/10.1175/2010JCLI3592.1 Oman, L., Robock, A., Stenchikov, G. L., Schmidt, G. A., & Ruedy, R. (2005). Climatic response to high‐latitude volcanic

286  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE eruptions. Journal of Geophysical Research: Atmospheres, 110(D13). https://doi.org/10.1029/2004JD005487 Oman, L., Robock, A., Stenchikov, G. L., & Thordarson, T. (2006). High‐latitude eruptions cast shadow over the African monsoon and the flow of the Nile. Geophysical Research Letters, 33(18). https://doi.org/10.1029/2006GL027665 Pausata, F. S. R., Chafik, L., Caballero, R., & Battisti, D. S. (2015). Impacts of high‐latitude volcanic eruptions on ENSO and AMOC. Proceedings of the National Academy of Sciences, 112(45), 13784–13788. https://doi.org/10.1073/ pnas.1509153112 Pausata, F. S. R., Karamperidou, C., Caballero, R., & Battisti, D. S. (2016). ENSO response to high‐latitude volcanic eruptions in the Northern Hemisphere: The role of the initial conditions. Geophysical Research Letters, 43(16), 8694–8702. https://doi.org/10.1002/2016GL069575 Pinto, J. P., Turco, R. P., & Toon, O. B. (2018). Self‐limiting physical and chemical effects in volcanic eruption clouds. Journal of Geophysical Research: Atmospheres, 94(D8), 11165–11174. https://doi.org/10.1029/JD094iD08p11165 Polo, I., Martin‐Rey, M., Rodriguez‐Fonseca, B., Kucharski, F., & Mechoso, C. R. (2014). Processes in the Pacific La Niña onset triggered by the Atlantic Niño. Climate Dynamics, 44(1–2), 115–131. https://doi.org/10.1007/s00382‐014‐2354‐7 Power, S., Delage, F., Chung, C., Kociuba, G., & Keay, K. (2013). Robust twenty‐first‐century projections of El Niño and related precipitation variability. Nature, 502(7472), 541– 545. https://doi.org/10.1038/nature12580 Predybaylo, E., Stenchikov, G. L., Wittenberg, A. T., & Zeng, F. (2017). Impacts of a Pinatubo‐size volcanic eruption on ENSO. Journal of Geophysical Research: Atmospheres, 122(2), 925–947. https://doi.org/10.1002/2016JD025796 Rampino, M. R., & Self, S. (1984). Sulphur‐rich volcanic eruptions and stratospheric aerosols. Nature. https://doi. org/10.1038/310677a0 Rayner, N. A., Parker, D. E., Horton, E. B., Folland, C. K., Alexander, L. V., Rowell, D. P., et al. (2003). Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res. Vol., 108(D14), 4407. 10.1029/2002JD002670 Robock, A. (2000). Volcanic eruptions and climate. Reviews of Geophysics. https://doi.org/10.1029/1998RG000054 Robock, A., & Mao, J. (1995). The volcanic signal in surface temperature observations. Journal of Climate. https://doi.org /10.1175/1520‐0442(1995)0082.0.CO;2 Rohde, R., Muller, R. A., Jacobsen, R., Muller, E., Perlmutter, S., et al. (2013). A new estimate of the average Earth surface land temperature spanning 1753 to 2011. Geoinfor Geostat: An Overview, 1, 1. Ruprich‐Robert, Y., Msadek, R., Castruccio, F., Yeager, S., Delworth, T., & Danabasoglu, G. (2017). Assessing the ­climate impacts of the observed atlantic multidecadal variability using the GFDL CM2.1 and NCAR CESM1 global coupled models. Journal of Climate, 30(8), 2785–2810. https:// doi.org/10.1175/JCLI‐D‐16‐0127.1 Russell, A. M., & Gnanadesikan, A. (2014). Understanding multidecadal variability in ENSO amplitude. Journal of Climate, 27(11), 4037–4051. https://doi.org/10.1175/ JCLI‐D‐13‐00147.1

Santer, B. D., Bonfils, C., Painter, J. F., Zelinka, M. D., Mears, C., Solomon, S., et al. (2014). Volcanic contribution to decadal changes in tropospheric temperature. Nature Geoscience. https://doi.org/10.1038/ngeo2098 Sato, M., Hansen, J. E., McCormick, M. P., & Pollack, J. B. (1993). Stratospheric aerosol optical depths, 1850–1990. J. Geophys. Res., 98(D12), 22,987–22,994. Schneider, T., Bischoff, T., & Haug, G. H. (2014). Migrations and dynamics of the intertropical convergence zone. Nature, 513, 45. Retrieved from http://dx.doi.org/10.1038/ nature13636 Sear, C. B., Kelly, P. M., Jones, P. D., & Goodess, C. M. (1987). Global surface‐temperature responses to major volcanic eruptions. Nature. https://doi.org/10.1038/330365a0 Self, S., Rampino, M. R., Zhao, J., & Katz, M. G. (1997). Volcanic aerosol perturbations and strong El Niño events: No general correlation. Geophysical Research Letters, 24(10), 1247–1250. https://doi.org/10.1029/97GL01127 Shindell, D. T., Schmidt, G. A., Mann, M. E., & Faluvegi, G. (2004). Dynamic winter climate response to large tropical volcanic eruptions since 1600. Journal of Geophysical Research. https://doi.org/10.1029/2003JD004151 Sigl, M., Winstrup, M., McConnell, J. R., Welten, K. C., Plunkett, G., Ludlow, F., et  al. (2015). Timing and climate forcing of volcanic eruptions for the past 2,500 years. Nature, 523, 543. Soden, B. J., Wetherald, R. T., Stenchikov, G. L., & Robock, A. (2002). Global cooling after the eruption of Mount Pinatubo: A test of climate feedback by water vapor. Science, 296(5568), 727 LP‐730. Retrieved from http://science.sciencemag.org/ content/296/5568/727.abstract Stahle, D. S., D’Arrigo, R. D., Krusic, P. J., Cleaveland, M. K., Cook, E. R., Allan, R. J., et al. (1998). Experimental dendroclimatic reconstruction of the Southern Oscillation. Bulletin of the American Meteorological Society, 79(10), 2137–2152. https://doi.org/10.1175/1520‐0477(1998)0792.0.CO;2 Stenchikov, G. L., Kirchner, I., Robock, A., Graf, H. F., Antuña, J. C., Grainger, R. G., et al. (1998). Radiative from the 1991 Mount Pinatubo volcanic eruption. Journal of Geophysical Research, 103(D12), 13,837‐13,857. Stenchikov, G., Hamilton, K., Stouffer, R. J., Robock, A., Ramaswamy, V., Santer, B., & Graf, H.‐F. (2006). ArcticOscillation response to volcanic eruptions in the IPCC AR4 climate models. J. Geophys. Res.,111, D07107. doi:10.1029/2005JD006286 Stenchikov, G. L., Delworth, T. L., Ramaswamy, V., Stouffer, R. J., Wittenberg, A., & Zeng, F. (2009). Volcanic signals in oceans. Journal of Geophysical Research Atmospheres, 114(16), 1–13. https://doi.org/10.1029/2008JD011673 Stevenson, S., Otto‐Bliesner, B., Fasullo, J., & Brady, E. (2016). “El Niño Like” hydroclimate responses to last millennium volcanic eruptions. Journal of Climate, 29(8), 2907–2921. https://doi.org/10.1175/JCLI‐D‐15‐0239.1 Stevenson, S., Fasullo, J. T., Otto‐Bliesner, B. L., Tomas, R. A., & Gao, C. (2017). Role of eruption season in reconciling model and proxy responses to tropical volcanism. Proceedings of the National Academy of Sciences, 114(8), 1822–1826. https://doi.org/10.1073/pnas.1612505114

EFFECT OF STRONG VOLCANIC ERUPTIONS ON ENSO  287 Stoffel, M., Khodri, M., Corona, C., Guillet, S., Poulain, V., Bekki, S., et al. (2015). Estimates of volcanic‐induced cooling in the Northern Hemisphere over the past 1,500 years. Nature Geoscience, 8, 784. Retrieved from https://doi.org/10.1038/ ngeo2526 Sutton, R. T., Dong, B., & Gregory, J. M. (2007). Land/sea warming ratio in response to climate change: IPCC AR4 model results and comparison with observations. Geophysical Research Letters, 34(2). https://doi.org/10.1029/2006GL028164 Swingedouw, D., Ortega, P., Mignot, J., Guilyardi, E., Masson‐ Delmotte, V., Butler, P. G., et  al. (2015). Bidecadal North Atlantic ocean circulation variability controlled by timing of volcanic eruptions. Nature Communications, 6, 6545. Retrieved from https://doi.org/10.1038/ncomms7545 Swingedouw, D., Mignot, J., Ortega, P., Khodri, M., Menegoz, M., Cassou, C., & Hanquiez, V. (2017). Impact of explosive volcanic eruptions on the main climate variability modes. Global and Planetary Change, 150, 24–45. https://doi.org/ https://doi.org/10.1016/j.gloplacha.2017.01.006 Thompson, D. W. J., & Solomon, S. (2009) Understanding recentstratospheric climate change.J. Climate, 22, 1934–1943. doi:10.1175/2008JCLI2482.1 Thompson, D. W. J., Wallace, J. M., Jones, P. D., & Kennedy, J. J. (2009). Identifying signatures of natural climate variability in time series of global‐mean surface temperature: Methodology and insights. Journal of Climate. https://doi. org/10.1175/2009JCLI3089.1 Tierney, J. E., Abram, N. J., Anchukaitis, K. J., Evans, M. N., Giry, C., Kilbourne, K. H., et al. (2015). Tropical sea surface temperatures for the past four centuries reconstructed from coral archives. Paleoceanography, 30(3), 226–252. https://doi. org/10.1002/2014PA002717 Timmermann, A., An, S. I., Krebs, U., & Goosse, H. (2005). ENSO suppression due to weakening of the North Atlantic thermohaline circulation. Journal of Climate, 18(16), 3122– 3139. https://doi.org/10.1175/JCLI3495.1 Timmermann, A., Okumura, Y., An, S. I., Clement, A., Dong, B., Guilyardi, E., et al. (2007). The influence of a weakening of the Atlantic meridional overturning circulation on ENSO. Journal of Climate, 20(19), 4899–4919. https://doi. org/10.1175/JCLI4283.1 Timmreck, C. (2012). Modeling the climatic effects of large explosive volcanic eruptions. Wiley Interdisciplinary Reviews: Climate Change, 3(6), 545–564. https://doi.org/10.1002/wcc.192 Timmreck, C., Lorenz, S. J., Crowley, T. J., Kinne, S., Raddatz, T. J., Thomas, M. A., & Jungclaus, J. H. (2009). Limited ­temperature response to the very large AD 1258 volcanic eruption. Geophysical Research Letters, 36(21). https://doi. org/10.1029/2009GL040083 Timmreck, C., Mann, G. W., Aquila, V., Hommel, R., Lee, L. A., Schmidt, A., et al. (2018). The Interactive Stratospheric

Aerosol Model Intercomparison Project (ISA‐MIP): Motivation and experimental design. Geoscientific Model Development, 11(7), 2581–2608. https://doi.org/10.5194/ gmd‐11‐2581‐2018 Toohey, M., Krüger, K., Niemeier, U., & Timmreck, C. (2011). The influence of eruption season on the global aerosol evolution and radiative impact of tropical volcanic eruptions. Atmospheric Chemistry and Physics, 11(23), 12351–12367. https://doi.org/10.5194/acp‐11‐12351‐2011 Toohey, M., Stevens, B., Schmidt, H., & Timmreck, C. (2016). Easy Volcanic Aerosol (EVA v1.0): An idealized forcing generator for climate simulations. Geoscientific Model ­ Development, 9(11), 4049–4070. https://doi.org/10.5194/ gmd‐9‐4049‐2016 Vijayeta, A., & Dommenget, D. (2018). An evaluation of ENSO dynamics in CMIP simulations in the framework of the recharge oscillator model. Climate Dynamics, 51(5), 1753– 1771. https://doi.org/10.1007/s00382‐017‐3981‐6 Wilson, R., Cook, E., D’Arrigo, R., Riedwyl, N., Evans, M. N., Tudhope, A., & Allan, R. (2010). Reconstructing ENSO: The influence of method, proxy data, climate forcing and teleconnections. Journal of Quaternary Science, 25(1), 62–78. https:// doi.org/10.1002/jqs.1297 Xie, S. P., Deser, C., Vecchi, G. A., Ma, J., Teng, H., & Wittenberg, A. T. (2010). Global warming pattern formation: Sea surface temperature and rainfall. Journal of Climate, 23(4), 966–986. https://doi.org/10.1175/2009JCLI3329.1 Zanchettin, D., Timmreck, C., Graf, H.‐F., Rubino, A., Lorenz, S., Lohmann, K., et al. (2012). Bi‐decadal variability excited in the coupled ocean–atmosphere system by strong tropical volcanic eruptions. Climate Dynamics, 39(1), 419–444. https://doi.org/10.1007/s00382‐011‐1167‐1 Zanchettin, D., Khodri, M., Timmreck, C., Toohey, M., Schmidt, A., Gerber, E. P., et  al. (2016). The Model Intercomparison Project on the climatic response to volcanic forcing (VolMIP): Experimental design and forcing input data for CMIP6. Geoscientific Model ­ Development, 9(8), 2701–2719. https://doi.org/10.5194/ gmd‐9‐2701‐2016 Zebiak, S. E., & Cane, M. A. (1987). A model El Nino‐Southern Oscillation. Monthly Weather Review. https://doi.org/10.1175 /1520‐0493(1987)1152.0.CO;2 Zhang, R., & Delworth, T. L. (2005). Simulated tropical response to a substantial weakening of the Atlantic thermohaline circulation. Journal of Climate, 18(12), 1853–1860. https://doi.org/10.1175/JCLI3460.1 Zuo, M., Man, W., Zhou, T., & Guo, Z. (2018). Different impacts of northern, tropical, and southern volcanic eruptions on the tropical Pacific SST in the last millennium. Journal of Climate, 31(17), 6729–6744. https://doi. org/10.1175/JCLI‐D‐17‐0571.1

13 ENSO Response to Greenhouse Forcing Wenju Cai1,2, Agus Santoso1,3, Guojian Wang1,2, Lixin Wu2, Mat Collins4, Matthieu Lengaigne5,6, Scott Power7,8, and Axel Timmermann9,10

ABSTRACT How ENSO responds to an increasing concentration of greenhouse gases in the atmosphere has remained an e­ lusive issue for decades. Climate models produce widely diverging results based on the traditional sea surface temperature (SST) metrics of ENSO. Some models project stronger ENSO SST variability, some weaker, some show no clear change. Steering away from these static measures, but more carefully examining the underlying processes and the associated key physical characteristics of ENSO, a clearer picture begins to emerge. Due to the nonlinear response of the atmosphere to SSTs, climate models project an increase in ENSO‐driven precipitation. Such a response tends to be robust across models linked to the relatively strong intermodel agreement in projected changes of the Pacific mean climate, marked by equatorially enhanced warming and weakened Walker Circulation. These mean‐state changes facilitate increased frequency of extreme El Niño events in models that are able to simulate nonlinear prop­ erties of ENSO closer to observations. In this ensemble of selected models, the frequency of extreme La Niña events is also projected to increase, as facilitated by faster warming of the Maritime Continent than the surrounding ocean waters. A projected increase in upper‐ocean stratification further favors increased variability and occurrences of Eastern Pacific El Niño. Uncertainties, however, remain due to persistent model biases, highlighting the need to further improve climate models, as well as sustain reliable observations to constrain model projections. Nonetheless, these projections underscore a possible heightened impact of ENSO‐driven changes in a warming climate.

13.1. INTRODUCTION Given the significant global‐scale impact of ENSO on society, economy, and the environment, understanding how ENSO responds to greenhouse forcing is an urgent critical

issue as global greenhouse‐gas emissions continue unabated. While ENSO as a naturally occurring climate phenomenon will continue to operate in the warmer future, crucial ques­ tions remain that are of relevance to climate and disaster risk management. Will it change in character? Will it become

1  Centre for Southern Hemisphere Oceans Research (CSHOR), CSIRO Oceans and Atmosphere, Hobart, TAS, Australia 2  Key Laboratory of Physical Oceanography/Institute for Advanced Ocean Studies, Ocean University of China and Qingdao National Laboratory for Marine Science and ­Tech­nology, Qingdao, China 3  Australian Research Council (ARC) Centre of Excellence for Climate Extremes and Climate Change Research Centre, University of New South Wales, Sydney, NSW, Australia 4  College of Engineering, Mathematics, and Physical Sciences, University of Exeter, Exeter, UK

5   LOCEAN-IPSL, Sorbonne Universités/UPMC-CNRS-IRDMNHN, Paris, France 6  MARBEC, University of Montpellier, CNRS, IFREMER, IRD, Sète, France 7  Australian Bureau of Meteorology, Melbourne, VIC, Australia 8  School of Earth, Atmosphere and Environment, and ARC Centre of Excellence for Climate Extremes, Monash University, Melbourne, VIC, Australia 9  Center for Climate Physics, Institute for Basic Science (IBS), Busan, South Korea 10  Pusan National University, Busan, South Korea

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 289

290  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

more or less active, stronger or weaker? And if so, what are the mechanisms? Are the climate models used to make the pro­ jections reliable? This chapter discusses the current state of understanding that is relevant to addressing these questions, indicative of the possibility that ENSO will respond to future anthropogenic greenhouse forcing in some significant ways. Considering the multitude of delicate processes gov­ erning ENSO characteristics and evolution as described in earlier chapters of this book, it is not surprising that projecting future ENSO behavior is a challenging under­ taking. Every element of the dynamical processes can respond directly to greenhouse forcing, or indirectly via changes in other components of the climate system. This complexity has also posed a great challenge in simulating a realistic ENSO in climate models that are used to make the projections (chapter 9). ENSO is tightly linked to the tropical Pacific mean climate upon which it evolves (chapter 8), so it is necessary to understand how the mean climate might change under greenhouse forcing. This will be first discussed in section  13.2. Section  13.3 explains the fundamental reasons for why at first impression there was no agreement among climate models in the projected ENSO changes. It is not until recently that more robust projections started to emerge, owing to further consider­ ations of the nonlinear nature of ENSO. This is to be covered in section 13.4 outlining the projected changes in atmosphere and oceanic aspects of ENSO dynamics along with the associated mechanisms. Section 13.5 dis­ cusses the uncertainties underlying the future projections tied to model biases. The chapter concludes in sec­ tion 13.6 with a summary and concluding remarks. 13.2. FORCED CHANGES IN BACKGROUND CLIMATE Under ENSO neutral conditions, the tropical Pacific climate is characterized by westward blowing trade winds that pile up warm waters in the western Pacific. The west­ ern Pacific warm pool exhibits annual mean SSTs above 28°C, a necessary condition for maintaining deep atmo­ spheric convection above the ocean (Graham & Barnett, 1987). The warm pool hence hosts an almost permanent deep atmospheric convection, in which mid-tropospheric heating drives the ascending branch of the Walker Circulation and surface easterlies over the eastern and central equatorial Pacific. The rising air flows eastward in the upper troposphere and subsides over the relatively cold and dry eastern equatorial Pacific (see Figure 1.1 in chapter  1). The equatorial trade winds generate strong upwelling in the eastern equatorial Pacific through Ekman divergence. The easterly momentum transfer from the equatorial trades to the ocean is balanced by a zonal oceanic pressure gradient, which in turn causes the thermocline to shoal (deepen) in the eastern (western) tropical Pacific. The off‐equatorial and equatorial trade

winds also generate subtropical meridional oceanic circulation cells (McCreary & Lu, 1994) that transport the upwelled equatorial water poleward while bringing back subducted subtropical waters towards the equator within the mean thermocline layer at about 100–200 m depths (e.g. Schott et al., 2004). In response to increasing greenhouse gas emissions, many climate models simulate stronger warming along the equator than off‐equator (Timmermann et  al., 1999; see schematic, Figure  13.1). This “enhanced equatorial warming” pattern (Liu et  al., 2005; Xie et  al., 2010; Cai et al., 2015b) is due to the fact that the evaporative damp­ ing of CO2‐induced warming is weaker in the equatorial strip compared to the off‐equatorial regions, because the latent heat flux scales with the mean wind speed, and the mean wind speed decreases towards the equator (Seager & Murtugudde, 1997; Xie et  al., 2010). In addition to the characteristic meridional warming structure, most CMIP3 and CMIP5 (Climate Model Intercomparison Project) models also simulate a slightly stronger warming in the eastern Pacific compared to the west (e.g. Xie et al. 2010; Power et al., 2013). Climate models with a thermodynamic mixed‐layer slab ocean can capture this response (e.g. Vecchi & Soden, 2007), sometimes even stronger than cou­ pled general circulation models (CGCMs). Factors that may contribute to this zonally asymmetric response include a shallower mixed layer in the eastern Pacific and asymmetric cloud feedbacks (stratus cloud feedback in the east, cumulus cloud/cirrus feedback in the west; Meehl & Washington, 1996), but also a deepening of the eastern equatorial thermocline in response to weaker equatorial trade winds (see discussion below). However, the fact that some cli­ mate models (e.g. Kohyama et al., 2017) instead simulate a negative zonal SST gradient change along the equator in response to transient CO2 radiative forcing suggests that the upwelling of relatively cold water in the eastern Pacific and the associated thermodynamic damping (“dynamical thermo­ stat”; Clement et al., 1996) can partly offset the CO2‐induced surface warming. This argument relies on the assumption that the older waters that upwell in the eastern equatorial Pacific had been exposed to lower atmospheric CO2‐induced warming the last time they were in contact with the surface. This non-equilibrium effect is likely to play an important role in transient climate change simulations, but not under equili­ brated conditions in which the amount of warming at subsur­ face has caught up with that at the surface. The enhanced equatorial warming plays a key role in intensifying rainfall in the equatorial region over the 21st century (e.g. Vecchi & Soden, 2007; Cai et al., 2014; Power & Delage, 2018), due to an increased low‐level moisture convergence. Such pattern of rainfall changes is consistent with a “warmer gets wetter” hypothesis (Xie et al., 2010) in which precipitation changes are closely tied to shifts in atmospheric circulation and moisture convergence (Chadwick et al., 2012; Widlansky et al., 2013).

ENSO Response to Greenhouse Forcing  291

Present state

Future changes

A

0m

10°N

B

5°N Eq

100m 0m 200m

C

100m

300m

200m

10°N 5°N

300m 120°E

150°E 0.6

180°E 1.2

150°W 1.8

2.4

Eq

120°W 5°S

°C 10°S

Figure 13.1  Schematic of tropical Pacific mean‐state changes due to greenhouse forcing. Red arrows and red outlined clouds indicate greenhouse‐induced changes. Present‐day mean states are featured in black or gray (e.g. gray arrows indicate mean ocean circulations). Color shading denotes temperature change from present to future. (A) The Walker Circulation (dashed arrow) slows down, resulting in weaker westward flowing trade winds and ocean currents. (B) The equatorial region warms faster than off equator, with the eastern region and Maritime Continent warming faster than in the central Pacific. Atmospheric convection shifts towards eastern equatorial Pacific due to the reduced meridional and zonal temperature gradients. (C) The vertical ocean temperature gradient increases in response to increased radiative forcing at the surface, leading to shoaled thermocline. Present‐ day thermocline is indicated by the black curve in the ocean interior which shoals eastward, shallowing in the future as indicated by the red curve. Adapted from Cai et al. (2015b).

The majority of climate models forced with increasing greenhouse gas concentrations show a slow‐down of the Walker Circulation (Vecchi et al., 2006; Power & Kociuba, 2011a; Kociuba & Power, 2015). Following the Clausius‐ Clapeyron relation (Clausius, 1850; Clapeyron, 1834), the saturated water vapor in the lower troposphere increases at a global mean rate of about 7% K–1 of global warming. Climate models support this thermodynamical relation­ ship, at least on a global scale (Collins et  al., 2010). However, the rate of precipitation increase is much lower (~2% K–1). Because the precipitation increase does not keep up with the increase in humidity, there must be a reduction in the mass flux from the moist boundary layer into the dryer air aloft (Held & Soden, 2006). As a consequence, atmospheric vertical motion over tropical convective regions, such as the western Pacific warm pool, is expected to decrease (~5‐10% K‐1), leading to a slowdown in the atmospheric overturning circulation (Zhang & Song, 2006). Even though this argument links the thermodynamics of global warming with atmospheric dynamics, it does not entirely explain how the atmo­ spheric circulation will slow down and whether the slow­

down will affect the Walker and Hadley circulation cells. Other factors, such as the enhanced equatorial warming (e.g. Liu et al., 2005) and off‐equatorial rainfall increase driven by the background sea surface warming (Kug et  al., 2011), can further contribute to the Walker Circulation slowdown. The wind change also leads to a reduction of Ekman divergence and equatorial upwelling (Vecchi & Soden, 2007; DiNezio et al., 2009; see Collins et al., 2010 and Cai et al., 2015b for a review). As a result, the east‐west tilted thermocline flattens, which can further amplify the initial equatorial warming, in particular in the eastern equatorial Pacific. Regarding the Hadley Circulation, the situation is more complex, because on top of an overall slowdown of the meridional mass stream function (e.g. Vecchi & Soden, 2007; Lu et al., 2007; Seo et al., 2014), there is also evi­ dence for a “deep tropical squeeze” (Lau & Kim, 2015), which tends to enhance the near‐equatorial circulation characteristics. Furthermore, given that the Hadley cells are influenced by atmospheric stratification, meridional surface temperature gradients, and extratropical eddy dynamics (Schneider, 1977; Seo et  al., 2014; Walker &

292  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Schneider, 2006), their future strength and properties can be influenced by changes in these various elements. It should be noted here that even though the tropical Pacific warming pattern and an El Niño bear some simi­ larities in terms of their surface temperature characteris­ tics, the underlying processes responsible for the pattern formation are very different. The greenhouse gas–induced westerly wind anomalies cover the entire equatorial Pacific, whereas westerly wind anomalies during an El Niño are confined to the western to central part of the Pacific basin. The weaker Walker Circulation under global warming reduces poleward heat transport, which is near symmetrical about the equator (e.g. Liu et  al., 2017). In contrast, there tends to be a net northward heat transport across the equator during an El Niño event (e.g. McGregor et al. 2014). The seasonality of the anomalous equatorial Pacific warming may also be different: For instance, in the GFDL CM2.1 model, during an El Niño it tends to peak toward boreal winter, whereas the greenhouse warming response peaks around mid–calendar year (Xie et  al., 2010). Further, sea level pressure over the tropical South Pacific decreases during an El Niño, but an increase may be detected under global warming (Xie et al., 2010). This would be reflected in the Southern Oscillation Index (SOI), which is negative during El Niño years but increases under global warming (Power & Kociuba, 2011b). This contrast arises because the SOI depends on mean sea level pressure (MSLP) spatial differences, and MSLP changes over the Pacific under global warming and during El Niño years tend to have a different spatial structure (Power & Kociuba, 2011b). While the projected mean‐state changes outlined above (Figure 13.1) tend to be consistent across models, confi­ dence in these projections is reduced by the existence of observational uncertainty and model biases. For instance, there can be disagreement in estimates of long‐term cli­ mate trends derived from different observational datasets. Disentangling long‐term observed SST trends in the tropical Pacific in terms of externally forced signals and corresponding feedbacks and natural variability has been hampered by the fact that the different SST observational products show opposing patterns, in particular in the eastern equatorial Pacific (Deser et  al., 2010; An et  al., 2012). A multivariate statistical decomposition that includes a removal of ENSO suggests a more robust strengthening of the prevailing SST gradients (Solomon & Newman, 2012) over the period 1900–2010, the extent of which is not captured by CMIP5 models (Coats & Karnauskas, 2017). This result would support the role of the “ocean dynamical thermostat” (Clement et al., 1996, Cane et  al., 1997), particularly during the boreal fall when the climatological SST gradient is the strongest (Karnauskas et al., 2009). This strengthening of the east‐ west SST gradient occurs despite a weakening of sea level

pressure gradient, which indicates a weakened Walker Circulation and trade winds (Karnauskas et  al., 2009). This near centennial trend could suggest a greenhouse forcing effect that may manifest in stronger zonal SST gradients, although uncertainties still exist in the observa­ tional data prior to 1950 (see chapter 3) and due to the fact that the greenhouse warming signal, natural inter­ decadal variability, and ENSO share some pattern char­ acteristics. In addition, these patterns may also vary over a shorter period. Over 1950–2009, for instance, the zonal SST gradient was found to weaken (Tokinaga et  al., 2012a), consistent with the weaker winds (Tokinaga et al., 2012b). However, these are inconsistent with analysis of sea level pressure based on several observational products that instead suggested a stronger Walker Circulation (L’Heureux et al., 2013). These inconsistent changes bet­ ween zonal SST gradient and Walker Circulation again differ from the case of El Niño in which weaker east‐west SST gradient associated with warmer eastern equatorial Pacific is accompanied by weaker trade winds through the Bjerknes feedback, a positive feedback loop that sus­ tains ENSO event development (chapters 1 and 2). Under greenhouse warming though, stronger zonal SST gra­ dient may not necessarily be accompanied by stronger Walker Circulation (An, 2011). Diagnosing greenhouse effect in the relatively short observational record (Capotondi & Sardeshmukh, 2017) is complicated by naturally occurring decadal variability, such as the Interdecadal Pacific Oscillation (IPO; Power et al., 1999). Shifts in the Pacific climate associated with the IPO were observed in the mid‐1970s, from a negative IPO to a positive IPO, followed by a shift to a negative phase in the late 1990s. The positive IPO phase was char­ acterized by an SST trend with stronger warming east of the Dateline and weaker trade winds (e.g. Meehl & Washington, 1996). The latter was marked by unprece­ dented acceleration of the Walker Circulation (Kociuba & Power, 2015), along with a cooler tropical Pacific that contributed to the global warming hiatus (Kosaka & Xie, 2013; England et al., 2014). Correspondingly, there were marked changes in ENSO properties. The positive IPO state saw stronger ENSO variability in the eastern equatorial Pacific marked by the 1982–1983 and 1997– 1998 extreme El Niño events, and the negative IPO state had stronger variability in the central Pacific (e.g. Wang & An, 2001; Lee & McPhaden, 2010; Santoso et al. 2017). The link between IPO and ENSO variability is still a topic of intense research and is further complicated by the fact that mean state changes can influence ENSO properties (e.g. Fedorov & Philander, 2000; Wang & An, 2002; Power et  al., 2013), while changes in ENSO vari­ ability can in turn influence multidecadal variability (Timmermann, 2003; Rodgers et  al., 2004; Power & Colman, 2006; Sun et al., 2014; Newman et al., 2016).

ENSO Response to Greenhouse Forcing  293

It is necessary to stress that climate models still suffer from persistent biases (oftentimes larger than their pro­ jected global warming responses), thus leaving uncer­ tainties in the projected mean‐state changes despite the reasonably strong intermodel consensus on the projections. A well‐known model deficiency is the Pacific “cold tongue” bias in which the ribbon of cool eastern equatorial Pacific water extends too far west into the Western Pacific Warm Pool, with the stronger than observed Trade Winds, and a double Intertropical Convergence Zone bias (chapter 9). It is not entirely clear what ramifications such biases have on future projections, but recent research (Li et al., 2016; Ying et al., 2018) indi­ cates that models with less severe cold‐tongue bias tend to project a warming pattern with stronger SST warming in the east Pacific (see Section 13.5 for a discussion). 13.3. ELUSIVE PROJECTIONS OF ENSO Despite relative intermodel agreement on the projected change in the 21st century mean climate, there is a lack of consensus on the change in ENSO as typically diagnosed in terms of SST variability at fixed locations in the equatorial Pacific, such as in the Niño‐3, Niño‐3.4, and Niño‐4 regions (Figure 13.2a, c, d). SST is a core variable for ENSO, as it is the main way through which ocean‐ atmosphere feedbacks are mediated; thus, the Niño indices have been widely used across research and predic­ tion platforms to characterize and monitor ENSO events. Investigations of how ENSO SST variability could change in response to global warming started in the 1990s, using climate models that were considered advanced at the time. These studies found little or no changes in future ENSO behavior (e.g. Meehl et al., 1993; Tett, 1995; Knutson et al., 1997). However, the reliability of the results was questionable given the models’ defi­ ciencies in simulating the complex interacting processes involved in ENSO. Using a model with a more realistic representation of ENSO in part due to higher resolution that could better resolve equatorial wave dynamics, Timmermann et al. (1999) found more frequent El Niños and stronger cold events in the eastern equatorial Pacific under a future emission scenario. The authors argued that a long‐term increase of vertical stratification in the eastern tropical Pacific enhanced the sensitivity of SST to ENSO‐related wind stress forcing. This higher sensitivity would strengthen the thermocline feedback and thus pos­ sibly contribute to ENSO amplitude changes. A follow‐ up study (Timmermann, 2001) then presented evidence for a major change of ENSO stability during this greenhouse warming simulation, which translated into rapid amplitude shifts. Stronger and more frequent ENSOs were also found in another model, the Hadley Centre coupled model version 2 (HadCM2), under

greenhouse forcing with CO2 concentration four times the preindustrial level (Collins, 2000a). In stark contrast, in spite of being forced with the same greenhouse gas concentration, no appreciable response was found in the third version of the Hadley Centre model (HadCM3), which had enhanced horizontal ocean resolution, exclusion of flux adjustments, and subtle changes in the subgrid scale parameterization schemes that can affect cloud formation (Collins, 2000b). This discrepancy is an early example of model‐based uncertainties in ENSO projections. Since the 1990s, the performance of climate models in simulating ENSO has notably improved, despite stub­ born common biases (see chapter 9). Facilitated through CMIP, an increasing body of studies have now analyzed ensembles of different climate models, run with the same forcing under equivalent emission scenarios. Results from the third phase of CMIP (CMIP3) still showed no inter­ model consensus in terms of amplitude and frequency of ENSO SST variability (van Oldenborgh et  al., 2005; Merryfield, 2006; Guilyardi, 2006; Yeh & Kirtman, 2007; Latif & Keenlyside, 2009; Collins et  al., 2010), nor did models that participated in the later intercomparison project (CMIP5) (Stevenson, 2012; Santoso et al., 2013; Bellenger et al., 2014; Taschetto et al., 2014; Kim et al., 2014a; Cai et al., 2015b; Chen et al., 2017). Some models showed an increase in the amplitude of ENSO SST vari­ ability, some a decrease, and some showed no appreciable change. One reason for the variety of responses can be expla­ ined  by the fact that ENSO SST variability essentially arises from an imbalance between positive and nega­ tive feedback processes. For instance, during the growth phase of an El Niño, positive feedback processes domi­ nate the negative ones (Stein et al., 2010), resulting in a positive temperature tendency or heating rate (Tt). If the negative feedback processes dominate, then Tt is negative, resulting in a cooling of the mixed layer. This can be expressed mathematically in terms of a mixed‐layer heat budget, which decomposes the rate of change of poten­ tial temperature T into the different contributing terms: 0

0

Tt dz Hm

v Ty

Q

u Tx

u Tx

u Tx

w Tz

wTz

w Tz

Hm

v Ty v Ty

dz Res, (13.1)

where Q is the net balance between shortwave and long­ wave radiations, and latent heat and sensible heat fluxes at the air‐sea interface (divided by the product of a refer­ ence density of ~1026 kg m-3, specific heat capacity of seawater [3986 J kg-1 K-1], and mixed layer depth [~50 m]).

(a)

(b)

Difference in s.d. of SST, multi-model mean 0.12 0.08 0.04 0 –0.04 –0.08 –0.12

10°N 5°N 0° 5°S 10°S 120°E

(c)

150°E

180°W

150°W

120°W

90°W

Difference in s.d. of rainfall, multi-model mean 1.2 0.8 0.4 0 –0.4 –0.8 –1.2

10°N 5°N 0° 5°S 10°S 120°E

150°E

180°W

150°W

120°W

90°W

Projected changes in variability of Nino4 SST index

°C

1.2

Control Climate change

0.8 0.4

(d)

Projected changes in variability of Nino3 SST index

°C

1.5 1 0.5

mm d–1

(e)

Projected changes in variability of Nino4 rainfall index 5 3 1

mm d–1

(f)

Projected changes in variability of Nino3 rainfall index 3 2 1

3 -M E ns. 5 R R 4 R R R 3 G M -H -R O C S 5 -0 H s2 M 5 M M S 4 C 2 -0 -3 -1 -m S1 SS1 m-1 -1-1 ESM SM -BG AM ES C-C -CM -CM 3-6 ART LS- -ES -CM M2 M2 -E2 -E2 2-A 2-C 2-E mcm A-L A-M B-L OC M-L M-M CM M1 1-M l e S C R C 5 S k e 5 G S C L 1 s I 5 M S M S M M M in M A IO D S S n C M C C M M -E E E 1M I-E -ES I-C rE ES od CE CCE c-c csm Ca C -CM -C S SM CC CM MC NR O- EC GO F GF DL- DL- GIS GIS dGE dGE dGE o or i-m I C c R P E L L P N c b F A A C IR C a F M C CE CM M M S SL S N ult Ha Ha H G GF bc IP IP IP CS M

Figure 13.2 Greenhouse‐warming induced change in sea surface temperature and precipitation variability over tropical Pacific. Results shown are based on CMIP5 outputs under historical and RCP8.5 emission scenario using 34 models, focusing on boreal winter (December– February average). (a) Multi‐model‐mean change in SST standard deviation between future (2000–2099) and present day (1900–1999). Niño‐4 (160°E–150°W, 5°S–5°N) and Niño‐3 (150°W–90°W, 5°S–5°N) regions are indicated by blue box and red box, respectively. Dotted areas mean that more than 70% of models (>~24 models) generate a change in phase with multi‐model‐mean value. (b) As in (a), but for rainfall (mm per day). Maritime Continent region (100°E–125°E, 5°S–5°N) used to construct zonal temperature gradient in Figure 13.3b is indicated by cyan box. (c) and (d) Projected changes in SST variability over Niño‐4 and Niño‐3 regions, respectively. Models that simulate a reduction are grayed out. Multimodel ensemble is also shown in both panels with error bars corresponding to the 95% confidence interval based on a bootstrap method. (e) and (f) As in (c) and (d), but for rainfall. Intermodel consensus is weak for grid‐point SST variability change (a, c, d), while it is strong for that of rainfall (b, e, f).

ENSO Response to Greenhouse Forcing  295 (a)

(b)

Observed relationship, 1979–2015

Zonal temperature gradient (°C)

Nino3 rainfall (mm d–1)

12 1997/98

9 1982/83

6

2015/16

3 0 –2

–1

1

0

2

Observed relationship, 1979–2015

2

1988/89

1

1998/99

0

–1

–2 –3

3

–2

Meridional SST gradient (°C)

(c)

Frequency of extreme El Nino: one event per 16.2 years Frequency of extreme sequence: one event per 131 years

12 9 6 3 0 –2

0

2

1

2000-2099 Frequency of extreme El Nino: one event per 6.93 years Frequency of extreme sequence: one event per 55.3 years

15 Nino3 rainfall (mm d–1)

Nino3 rainfall (mm d–1)

(d)

1900-1999 15

–1

Nino4 SST index (s.d.)

12 9 6 3 0

–1

0

1

2

3

4

5

6

Meridional SST gradient (°C)

–2

–1

0

1

2

3

4

5

6

Meridional SST gradient (°C)

Figure 13.3  Observed climate extremes and greenhouse‐warming induced changes. (a) Observed relationship between Niño‐3 rainfall and meridional SST gradient (Cai et  al 2014; 150°W–90°W, 5°N–10°N minus 150°W–90°W, 2.5°S–2.5°N). Black dots indicate extreme El Niño events, defined as when Niño‐3 rainfall is greater than 5 mm per day (marked by the horizontal line). Red star marks the 1997–1998 event, which was an extreme El Niño with concurrent eastward‐propagating SST anomalies (Santoso et al., 2013) that was also followed by an extreme La Niña the following year (Cai et al., 2015b). (b) Observed relationship between Niño‐4 SST and surface air temperature gradient between the Maritime Continent region (Figure 13.2b) and Niño‐4 (Cai et al., 2015a). Blue dots indicate extreme La Niña events, defined as when Niño‐4 SST is negative and greater than 1.75 s.d. in amplitude. (c) and (d) As in (a), but using 21 selected CMIP5 models, which can produce extreme El Niño events as in Cai et al. (2015b), under present day (1900–1999), and future (2000–2099), respectively. Gray dots indicate events that are not extreme El Niño. As indicated within each panel, extreme El Niño occurs more frequently in future climate, from approximately one event per 16 years to one event per 7 years. A rare sequence of extreme events as in 1997–1998 is also projected to occur more often, from approximately one event per 131 years to one in 55 years.

The ocean current variables, u, v, and w, denote currents in the zonal, meridional, and vertical direction. The sub­ scripts x, y, z are spatial derivatives in the zonal, meridi­ onal, and vertical direction, respectively, and prime indicates deviation from the climatological state denoted by the overbar. These collective terms are integrated across the surface mixed layer of depth Hm (50 m is a rea­ sonable estimate for the average mixed layer depth in the equatorial Pacific), and the residual term, Res, contains unresolved processes such as mixing and diffusion.

The main positive feedback processes for ENSO are contained in the square bracketed terms in Eq.  (1), of which the current and temperature anomalies are linked to anomalous winds and thermocline. The zonal advec­ tive feedback is contained in the u Tx term, where anomalous zonal current acts on the zonal gradient of the climatological mixed‐layer temperature. During a developing El Niño, this term is positive (anomalous east­ ward advection of warmer western tropical Pacific waters), as the climatological zonal temperature gradient

296  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

is negative (cooler temperature in the eastern Pacific than in the west) and u′ is positive due to westerly wind anomaly and deeper thermocline along equator than off‐ equator, following the Ekman and geostrophic relations (see chapter 6). The Ekman pumping feedback is associ­ ated with the w Tz term, which is also positive during an El Niño, since weaker zonal winds lead to weaker upwelling (i.e. w′ 12 10 8 6 4 2 0 –2 –4 –6 –8 –10 2.0.CO;2 Killworth, P. D., D. B. Chelton, & R. A. de Szoeke (1997). The speed of observed and theoretical long extratropical planetary waves. J. Phys. Oceanogr., 27, 1946–1966. https://doi.org/10.1 175/1520‐0485(1997)0272.0.CO;2 Kim, Y. Y., T. Qu, T. Jensen, T. Miyama, H. Mitsudera, H.‐W. Kang, & A. Ishida (2004). Seasonal and interannual variations of the North Equatorial Current bifurcation in a high‐ resolution OGCM. J. Geophys. Res., 109, C03040. doi: 10.1029/2003JC002013 King, J. R., V. N. Agostini, C. J. Harvey, G. A. McFarlane, M.G.G. Foreman, J. E. Overland, et al. (2011). Climate forcing and the California Current ecosystem. ICES Journal of Marine Science, 68(6), 1199–1216. https://doi.org/10.1093/ icesjms/fsr009 Koch‐Larrouy, A., G. Madec, P. Bouruet‐Aubertot, T. Gerkema, L. Bessières, & R. Molcard (2007). On the transformation of Pacific Water into Indonesian Throughflow Water by internal tidal mixing. Geophysical Research Letters, 34(4). LaCasce, J. H., & J. Pedlosky (2004). The instability of Rossby basin modes and the oceanic eddy field. J. Phys. Oceanogr., 34, 2027–2041. https://doi.org/10.1175/1520‐0485(2004)034< 2027:TIORBM>2.0.CO;2 Lambert, E., D. L. Bars, & W. P. de Ruijter (2016). The connection of the Indonesian Throughflow, South Indian Ocean Countercurrent and the Leeuwin Current. Ocean Science, 12(3), 771–780. Lau, N.‐C., & M. J. Nath (2003). Atmosphere–ocean variations in the Indo‐Pacific sector during ENSO episodes. Journal of Climate, 16(1), 3–20. doi: 10.1175/1520‐0442(2003)0162.0.Co;2

356  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Lee, T., I. Fukumori, D. Menemenlis, Z. Xing, & L. L. Fu (2002). Effects of the Indonesian Throughflow on the Pacific and Indian Oceans. Journal of Physical Oceanography, 32(5), 1404–1429. Lee, T., & I. Fukumori (2003). Interannual‐to‐decadal variations of tropical–subtropical exchange in the Pacific ocean: Boundary versus interior pycnocline transports. J. Clim., 16(24), 4022–4042. doi:10.1175/1520‐0442(2003)0162.0.CO;2 Lee, S.‐K., W. Park, M. O. Baringer, A. L. Gordon, B. Huber, & Y. Liu (2015). Pacific origin of the abrupt increase in Indian Ocean heat content during the warming hiatus. Nature Geosci, 8(6), 445–449. doi:10.1038/ngeo2438 Legeckis, R. (1977). Long waves in the eastern equatorial Pacific Ocean: A view from a geostationary satellite. Science, 197, 1179–1181. Liu, Q., M. Feng, & D. Wang (2011). ENSO‐induced interannual variability in the southeastern South China Sea. Journal of Oceanography, 67(1), 127–133. Liu, Q. Y., M. Feng, D. Wang, & S. Wijffels (2015). Interannual variability of the Indonesian Throughflow transport: A revisit based on 30 year expendable bathythermograph data. Journal of Geophysical Research: Oceans, 120(12), 8270–8282. Liu, Z. (1994). A simple model of the mass exchange between the subtropical and tropical ocean. J. Phys. Oceanogr., 24, 1153–1165. doi: 10.1175/1520‐0485(1994)024,1153:ASMOT M.2.0.CO;2 Liu, Z., & G. Philander (2001). Tropical‐extratropical oceanic exchange pathways. In G. Siedler, J. Church, & W. J. Gould (Eds.), Ocean circulation and climate: Observing and modeling the global ocean (pp. 247–254). San Diego, CA: Academic Press. Llanillo, P. J., J. Karstensen, J. L. Pelegrí, & L. Stramma (2013). Physical and biogeochemical forcing of oxygen and nitrate changes during El Niño/El Viejo and La Niña/La Vieja upper‐ocean phases in the tropical eastern South Pacific along 86°8W. Biogeosciences, 10(10), 6339–6355. https://doi. org/10.5194/bg‐10‐6339‐2013 Llanillo, P. J., J. L. Pelegrí, L. D. Talley, J. Peña‐Izquierdo, & R. R. Cordero (2018). Oxygen pathways and budget for the eastern South Pacific Oxygen Minimum Zone. J. Geophys. Res.‐Oceans, 123. https://doi.org/10.1002/2017JC013509 Lu, P., & J.P. McCreary Jr. (1995). Influence of the ITCZ on the flow of thermocline water from the subtropical to the equatorial Pacific Ocean. J. Phys. Oceanogr., 25, 3076–3088. Luo, J.‐J., & T. Yamagata. (2001). Long‐term El Niño‐Southern Oscillation (ENSO)‐like variation with special emphasis on the South Pacific. Journal of Geophysical Research: Oceans 106(C10), 22211–22227. https://doi.org/10.1029/2000J C000471 Lynn, R. J., & S. J. Bograd (2002). Dynamic evolution of the 1997‐1999 El Niño‐La Niña cycle in the southern California Current System. Prog. Oceanogr., 54, 59–75. Mackey, D. J., J. E. O’Sullivan, & R. J. Watson (2002). Iron in the Western Pacific: A riverine or hydrothermal source for iron in the equatorial undercurrent? Deep Sea Res. I. 49(5), 877–893.

Marchesiello, P., & P. Estrade (2010). Upwelling limitation by onshore geostrophic flow. Journal of Marine Research, 68(1), 37–62, https://doi.org/10.1357/002224010793079004 Matear, R. J., M. A. Chamberlain, C. Sun, & M. Feng. (2013). Climate change projection of the Tasman Sea from an eddy‐ resolving ocean model. Journal of Geophysical Research: Oceans, 118(6), 2961–2976. https://doi.org/10.1002/jgrc.20202 Mayer, M., M. Alonso Balmaseda, & L. Haimberger (2018). Unprecedented 2015/2016 Indo‐Pacific heat transfer speeds up tropical Pacific heat recharge. Geophysical Research Letters, 45(7), 3274–3284. McClean, J. L., D. P. Ivanova, & J. Sprintall (2005). Remote origins of interannual variability in the Indonesian Throughflow region from data and a global Parallel Ocean Program simulation. Journal of Geophysical Research, 110, C10013. doi: 10.1029/2004JC002477 McCreary, J. P., & P. Lu (1994). Interaction between the subtropical and equatorial ocean circulations: The subtropical cell. J. Phys. Oceanogr., 24, 466–497. https://doi.org/10.1175/ 1520‐0485(1994)0242.0.CO;2 Menezes, V. V., H. E. Phillips, A. Schiller, C. M. Domingues, & N. L. Bindoff (2013). Salinity dominance on the Indian Ocean Eastern Gyral current. Geophysical Research Letters, 40(21), 5716–5721. Melet, A., L. Gourdeau, J. Verron, & B. Djath (2013). Solomon Sea circulation and water mass modifications: Response at ENSO timescales. Ocean Dyn., 63(1), 1–19. doi: 10.1007/ s10236‐012‐0582‐0 Menkes, C. E., J. Vialard, S. C. Kennan, J.‐P. Boulanger, & G. Madec (2006). A modeling study of the impact of tropical instability waves on the heat budget of the eastern equatorial Pacific. Journal of Physical Oceanography, 36(5), 847–865. Meyers, G. (1996). Variation of Indonesian throughflow and the El Niño–Southern Oscillation. J. Geophys. Res., 101(C5), 12,255–12,263. Miller, A. J., H. Song, & A. C. Subramanian (2015). The physical oceanographic environment during the CCE‐LTER Years: Changes in climate and concepts. Deep Sea Res. Part II, 112, 6–17. doi: 10.1016/j.dsr2.2014.01.003 Montes, I., B. Dewitte, E. Gutknecht, A. Paulmier, I. Dadou, A. Oschlies, & V. Garçon (2014). High‐resolution modeling of the eastern tropical Pacific oxygen minimum zone: Sensitivity to the tropical oceanic circulation. J. Geophys. Res. Oceans, 119, 5515–5532. doi: 10.1002/2014JC009858 Montes, I., W. Schneider, F. Colas, B. Blanke, & V. Echevin (2011). Subsurface connections in the eastern tropical Pacific during La Niña 1999–2001 and El Niño 2002–2003, J. Geophys. Res., 116, C12022. doi:10.1029/2011JC007624 Morrow, R., & F. Birol (1998). Variability in the southeast Indian Ocean from altimetry: Forcing mechanisms for the Leeuwin Current. J. Geophys. Res., 103, 18,429–18,544. Murtugudde, R., & A. J. Busalacchi (1999). Interannual variability of the dynamics and thermodynamics of the tropical Indian Ocean. J. Climate, 12, 2300–2326. https://doi.org/10.1 175/1520‐0442(1999)0122.0.CO;2 Nieves, V., J. K. Willis, & W. C. Patzert (2015). Recent hiatus caused by decadal shift in Indo‐Pacific heating. Science, 349(6247), 532–535. https://doi.org/10.1126/science.aaa4521

ENSO Oceanic Teleconnections  357 Ohman, M. D., K. Barbeau, P.J.S. Franks, R. Goericke, M.  R.  Landry, A. J. Miller (2013). Ecological transitions in  a coastal upwelling ecosystem. Oceanography, 26, 210–219. Ohman, M. D., N. Mantua, J. Keister, M. Garcia‐Reyes, & S.  McClatchie (2017). ENSO impacts on ecosystem indicators in the California Current System. U.S. CLIVAR Variations, 15(1), 8–15. Oliver, E. C. J., M. G. Donat, M. T. Burrows, P. Moore, D. A. Smale, L. V. Alexander, et al. (2018). Longer and more frequent marine heatwaves over the past century. Nature Communications, 9, 1324. doi: 10.1038/s41467‐018‐03732‐9 Oschlies, A., O. Duteil, J. Getzlaff, W. Koeve, A. Landolfi, & S. Schmidtko (2017). Patterns of deoxygenation: Sensitivity to natural and anthropogenic drivers. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 375(2102), 20160325. doi: 10.1098/ rsta.2016.0325 Pariwono, J. I., J.A.T. Bye, & G. W. Lennon (1986). Long‐period variations of sea‐level in Australasia. Geophysical Journal International, 87(1), 43–54. Paulmier, A., & D. Ruiz‐Pino (2009). Oxygen minimum zones (OMZs) in the modern ocean. Progress in Oceanography, 80, 113–128. Pearce, A. F., & M. Feng (2013). The rise and fall of the “marine heat wave” off Western Australia during the summer of 2010/2011. Journal of Marine Systems, 111, 139–156. Pearce, A. F., & B. F. Phillips (1998). ENSO events, the Leeuwin Current and larval recruitment of the western rock lobster. Journal du conseil international pour l’exploration de la mer, 45, 13–21. Pedlosky, J. (1987). Geophysical fluid dynamics, 2nd ed. New York: Springer‐Verlag. Pennington, J. T., K. L. Mahoney, V. S. Kuwahara, D.  D.  Kolber,  R. Calienes, & F. P. Chavez (2006). Primary production in the eastern tropical Pacific: A review. Progress in Oceanography, A Review of Eastern Tropical Pacific Oceanography, 69(2), 285–317. https://doi.org/10.1016/j. pocean.2006.03.012 Philander, S. G. (1990). El Nino, La Nina, and the Southern Oscillation. San Diego: Academic Press. Phillips, H. E., S. E. Wijffels, & M. Feng (2005). Interannual variability in the freshwater content of the Indonesian‐ Australian Basin. Geophysical Research Letters, 32(3). Pizarro O., G. Shaffer, B. Dewitte, & M. Ramos (2002). Dynamics of seasonal and interannual variability of the Peru‐Chile undercurrent. Geophys. Res. Lett., 29(12), Art. No. 1581. Potemra, J. T. (2001). Contribution of equatorial Pacific winds to southern tropical Indian Ocean Rossby waves. Journal of Geophysical Research: Oceans, 106(C2), 2407–2422. Potemra, J. T., R. Lukas, & G. Mitchum (1997). Large‐scale estimation of transport from the Pacific to the Indian Ocean. J. Geophys. Res., 102, 27,795–27,812. Putrasahan, D., B. P. Kirtman, & L. M. Beal (2016). Modulation of SST interannual variability in the Agulhas leakage region associated with ENSO. J. Clim., 29, 7089–7102. doi: 10.1175/ JCLI‐D‐15‐0172.1

Qiao, L., & R. H. Weisberg (1995). Tropical instability wave kinematics: Observations from the Tropical Instability Wave Experiment (TIWE). J. Geophys. Res.,100, 8677–8693. Qin, X., A. Sen Gupta, & E. van Sebille (2015). Variability in the origins and pathways of Pacific Equatorial Undercurrent water. J. Geophys. Res. Oceans, 120. doi:10.1002/ 2014JC01054 Qin, X., L. Menviel, A. Sen Gupta, & E. van Sebille (2016). Iron sources and pathways into the Pacific Equatorial Undercurrent. Geophys. Res. Lett., 43, 9843–9851. doi:10.1002/2016GL070501 Qiu, B., & S. Chen (2010). Eddy‐mean flow interaction in the decadally modulating Kuroshio Extension system. Deep Sea Research Part II: Topical Studies in Oceanography, 57(13–14), 1098–1110. doi: 10.1016/j.dsr2.2008.11.036 Qiu, B., & R. Lukas (1996). Seasonal and interannual variability of the North Equatorial Current, the Mindanao Current, and the Kuroshio along the Pacific western boundary. Journal of Geophysical Research‐Oceans, 101(C5), 12315–12330. https://doi.org/10.1029/95JC03204 Qiu, B., W. Miao, & P. Muller (1997). Propagation and decay of forced and free baroclinic Rossby waves in off‐equatorial oceans. J. Phys. Oceanogr., 27, 2405–2417. Qiu, B., D. L. Rudnick, S. Chen, & Y. Kashino (2013). Quasi– stationary North Equatorial Undercurrent jets across the tropical North Pacific Ocean. Geophys. Res. Lett., 40, 2183– 2187. doi:10.1002/grl.50394 Qiu, B., D. L. Rudnick, I. Cerovecki, B. D. Cornuelle, S. Chen, M. C. Schonau, et al. (2015). The Pacific North Equatorial Current: New insights from the origins of the Kuroshio and Mindanao Currents (OKMC) project. Oceanography, 28(4), 24–33. Qu, T., & E. J. Lindstrom (2002). A climatological interpretation of the circulation in the western South Pacific, J. Phys. Oceanogr., 32(9), 2492–2508. doi:10.1175/1520‐0485‐32.9.2492 Qu, T., Y. Du, G. Meyers, A. Ishida, & D. Wang (2005). Connecting the tropical Pacific with Indian Ocean through South China Sea. Geophys. Res. Lett., 32, L24609. doi: 10.1029/2005GL024698 Qu, T., Y. Du, & H. Sasaki (2006). South China Sea throughflow: A heat and freshwater conveyor. Geophys. Res. Lett., 33, L23617. doi: 10.1029/2006GL028350 Ramos M., B. Dewitte, O. Pizarro, & G. Garric (2008). Vertical propagation of the extra‐tropical Rossby wave during the 1997/98 El Niño off the west coast of South America in a medium‐resolution OGCM simulation. J. Geophys. Res., 113, C08041, doi: 10.1029/2007JC004681 Ridgway, K. R., & J. R. Dunn (2007). Observational evidence for a Southern Hemisphere oceanic supergyre. Geophys. Res. Lett., 34, L13612. doi: 10.1029/2007GL030392 Ridgway, K. R., J. S. Godfrey, G. Meyers, & R. Bailey (1993). Sea level response to the 1986–87 ENSO event in the western Pacific in the vicinity of Papua New Guinea. J. Geophys. Res., 98, 16387–16395. Roemmich, D., J. Gilson, R. Davis, P. Sutton, S. Wijffels, & S. Riser (2007). Decadal spinup of the South Pacific subtropical gyre. Journal of Physical Oceanography, 37, 162–173. Ryan, J. P., I. Ueki, Y. Chao, H. Hang, P. S. Polito, & F. P. Chavez (2006). Western Pacific modulation of large phytoplankton

358  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE blooms in the central and eastern equatorial Pacific. J. Geophys. Res., 111, G02013 doi:10.1029/2005JG000084 Saji, N. H., B. H. Goswami, P. N. Vinayachandran, & T.  Yamagata (1999). A dipole mode in the tropical Indian Ocean. Nature, 401, 360–363. Santoso, A., H. Hendon, A. Watkins, S. Power, D. Dommenget, M. H. England, et al. (2019). Dynamics and predictability of the El Niño‐Southern Oscillation: An Australian perspective on progress and challenges. Bulletin of the American Meteorological Society. doi: 10.1175/BAMS‐D‐18‐0057.1 Sasaki, Y. N., S. Minobe, N. Schneider, T. Kagimoto, M. Nonaka, & H. Sasaki (2008). Decadal sea level variability in the South Pacific in a global eddy‐resolving ocean model hindcast. J. Phys. Oceanogr., 38, 1731–1747. doi:10.1175/ 2007JPO3915.1 Schott, F., J. McCreary, & G. C. Johnson. (2004). Shallow over­ turning circulations of the tropical‐subtropical oceans. Washington, DC: American Geophysical Union. Geophysical Monograph Series, 47, 261–304. doi: 10.1029/147GM15 Schouten, M. W., W.P.M. de Ruijter, P. J. van Leeuwen, & H. A. Dijkstra (2002). An oceanic teleconnection between the equatorial and southern Indian Ocean: An Equatorial‐South Indian Ocean teleconnection. Geophys. Res. Lett., 29, 59‐1– 59‐4. doi:10.1029/2001GL014542 Schwing, F., T. Murphree, L. DeWitt, & P. Green (2002). The evolution of oceanic and atmospheric anomalies in the northeast Pacific during the El Niño and La Niña events of 1995–2001. Prog. Oceanogr., 54, 459–491. doi: 10.1016/ S0079‐6611(02)00064‐2 Sen Gupta, A., A. Ganachaud, S. McGregor, J. N. Brown, & L.  Muir (2012). Drivers of the projected changes to the Pacific Ocean equatorial circulation. Geophys. Res. Lett., 39, L09605. doi:10.1029/2012GL051447 Sen Gupta, A., S. McGregor, E. van Sebille, A. Ganachaud, J.  Brown, & A. Santoso (2016). Future changes to the Indonesian Throughflow and Pacific circulation: The differing role of wind and deep circulation changes. Geophys. Res. Lett., 2016GL067757. doi:10.1002/2016GL067757 Simpson, J. J. (1984). El Niño‐induced onshore transport in the California Current during 1982–1983. Geophy. Res. Lett., 11, 241–242. doi: 10.1029/GL011i003p00233 Slemons, L. O., Murray, J. W., Gorgues, T., Aumont, O., & Menkes, C. (2009). Biogeochemical impact of a model western iron source in the Pacific Equatorial Undercurrent. Deep Sea Res. I. 56(12), 2115–2128. doi: 10.1016/j. dsr.2009.08.005 Slemons, L. O., J. W. Murray, J. Resing, B. Paul, & P. Dutrieux (2010). Western Pacific coastal sources of iron, manganese, and aluminum to the equatorial undercurrent. Global Biogeochem. Cycles. 24, 3. https://doi.org/10.1029/2009 GB003693 Sloyan, B. M., G. C. Johnson, & W. S. Kessler (2003). The Pacific cold tongue: A pathway for interhemispheric exchange. J. Phys. Oceanogr., 33, 1027–1043. https://doi.org /10.1175/1520‐0485(2003)0332.0.CO;2 Song, Q., A. L. Gordon, & M. Visbeck (2004). Spreading of the Indonesian throughflow in the Indian Ocean. Journal of Physical Oceanography, 34(4), 772–792. Spillane, M. C., D. B. Enfield, & J. S. Allen (1987). Intraseasonal oscillations in sea level along the West Coast of the Americas.

J. Phys. Oceanogr., 17, 313–325. https://doi.org/10.1175/1520 ‐0485(1987)0172.0.CO;2 Sprintall, J., & Révelard, A. (2014). The Indonesian Throughflow response to Indo‐Pacific climate variability. Journal of Geophysical Research: Oceans, 119(2), 1161–1175. Sprintall, J., S. E. Wijffels, R. Molcard, & I. Jaya (2009). Direct estimates of the Indonesian Throughflow entering the Indian Ocean: 2004–2006. J. Geophys. Res., 114, C07001. doi: 10.1029/2008JC005257 Sprintall, J., A. L. Gordon, A. Koch‐Larrouy, T. Lee, J. T. Potemra, K. Pujiana, & S. E. Wijffels (2014). The Indonesian seas and their role in the coupled ocean‐climate system. Nat. Geosci., 7, 487–492. doi:10.1038/ngeo2188 Sun, C., M. Feng, R. J. Matear, M. A. Chamberlain, P. Craig, K. R. Ridgway, & A. Schiller (2012). Marine downscaling of a future climate scenario for Australian boundary currents. Journal of Climate, 25:8, 2947–62. https://doi.org/10.1175/ JCLI‐D‐11‐00159.1. Susanto, R. D., & J. Marra (2005). The effect on 1997/98 El Niño on chlorophyll: A concentration along the southern coasts of Java and Sumatra. Oceanography, 18(4), 124–127. Susanto, R. D., T. Moore II, and J. Marra (2006). Ocean color variability in the Indonesian Seas during the SeaWiFS era, Geochemistry Geophysics Geosystems, 7(5), 1–16. doi: 10.1029/2005GC001009 Sverdrup, H. U. (1947). Wind‐driven currents in a baroclinic ocean; with application to the equatorial current of the eastern Pacific. Proc. Natl. Acad. Sci. USA, 33, 318–326. https:// doi.org/10.1073/pnas.33.11.318 Takahashi, K., A. Montecinos, K. Goubanova, & B. Dewitte (2011). ENSO regimes: Reinterpreting the canonical and Modoki El Niño. Geophys. Res. Lett., 38, L10704. doi: 10.1029/2011GL047364 Talley, L. D., & J. Sprintall (2005). Deep expression of the Indonesian Throughflow: Indonesian Intermediate Water in the South Equatorial Current. Journal of Geophysical Research: Oceans, 110(C10). Tan, S., & H. Zhou (2018). The observed impacts of the two types of El Niño on the North Equatorial Countercurrent in the Pacific Ocean. Geophysical Research Letters, 45, 10,493– 10,500. https://doi. org/10.1029/2018GL079273 Tanaka, K., M. Ikeda, & Y. Masumoto (2004). Predictability of interannual variability in the Kuroshio transport south of Japan based on wind stress data over the North Pacific. J. Oceanogr., 60, 283–291. doi: 10.1023/B:JOCE.0000038334. 86069.ea Thomson, R. E., & M. V. Krassovski (2010). Poleward reach of the California Undercurrent extension. J. Geophys. Res., 115, C09027. doi: 10.1029/2010JC006280 Tomczak, M., & J. S. Godfrey (1994). Regional oceanography. Pergamon. https://doi.org/10.1016/B978‐0‐08‐041021‐0.50003‐5 Tozuka, T., T. Kataoka, & T. Yamagata (2014). Locally and remotely forced atmospheric circulation anomalies of Ningaloo Niño/Niña. Climate Dynamics, 43(7–8), 2197–2205. Tsuchiya, M., R. Lukas, R. A. Fine, E. Firing, & E. Lindstrom (1989). Source waters of the Pacific Equatorial Undercurrent. Progress in Oceanography, 23(2), 101–147.

ENSO Oceanic Teleconnections  359 van Sebille, E., J. Sprintall, F. U. Schwarzkopf, A. S. Gupta, A. Santoso, M. H. England, et  al. (2014). Pacific‐to‐Indian Ocean connectivity: Tasman leakage, Indonesian Throughflow, and the role of ENSO. J. Geophys. Res. Oceans, 119, 1365–1382. doi:10.1002/2013JC009525 Vergara O., B. Dewitte, I. Montes, V. Garçon, M. Ramos, A. Paulmier, & O. Pizarro (2016). Seasonal variability of the oxygen minimum zone off Peru in a high‐resolution regional coupled model. Biogeosciences, 13, 4389–4410. doi: 10.5194/ bg‐13‐4389‐2016 Vergara, O., B. Dewitte, M. Ramos, & O. Pizarro (2017). Vertical energy flux at ENSO time scales in the subthermocline of the Southeastern Pacific. J. Geophys. Res. Oceans, 122. doi:10. 1002/2016JC012614 Wyrtki, K. (1979). The response of sea surface topography to the 1976 El Niño. J. Phys. Oceanogr., 9, 1223–1231. doi:10.1175/1520‐0485(1979) 0092.0.CO;2 Wyrtki, K. (1987). Indonesian Throughflow and the associated pressure gradient. J. Geophys. Res., 92, 12941–12946. doi: 10.1029/JC092iC12p12941 Wyrtki, K., & B. Kilonsky (1984). Mean water and current structure during the Hawaii‐to‐Tahiti Shuttle Experiment. Journal of Physical Oceanography, 14(2), 242–254, 10.1175/ 1520‐0485(1984)0142.0.CO;2 Wijffels, S., & Meyers, G. (2004). An intersection of oceanic waveguides: Variability in the Indonesian Throughflow region. Journal of Physical Oceanography, 34(5), 1232–1253. Wijffels, S. E., G. Meyers, & J. S. Godfrey (2008). A 20‐yr average of the Indonesian Throughflow: Regional currents and the interbasin exchange. Journal of Physical Oceano­ graphy, 38(9), 1965–1978. Yamagata T., Y. Shibao, S. Umatanit (1985). Interannual variability of the Kuroshio Extension and its relation to the Southern Oscillation/El Niño. Journal of the Oceanographical Society of Japan, 41(4), 274–281. doi: 10.1007/BF02109276 Yit Sen Bull, C., & Van Sebille, E. (2016). Sources, fate, and pathways of Leeuwin Current water in the Indian Ocean and Great Australian Bight: A Lagrangian study in an eddy‐ resolving ocean model. Journal of Geophysical Research: Oceans, 121(3), 1626–1639.

Yu, J.‐Y., & W. T. Liu (2003). A linear relationship between ENSO intensity and tropical instability wave activity in the eastern Pacific Ocean. Geophys. Res. Lett., 30, 1735. doi:10.1029/2003GL017176 Yuan, D., J. Wang, T. Xu, P. Xu, Z. Hui, X. Zhao, et al. (2011). Forcing of the Indian Ocean Dipole on the interannual variations of the tropical Pacific Ocean: Roles of the Indonesian Throughflow. Journal of Climate, 24(14), 3593–3608. doi: 10.1175/2011jcli3649.1 Yuan, D., H. Zhou, & X. Zhao (2013). Interannual climate variability over the tropical Pacific Ocean induced by the Indian Ocean Dipole through the Indonesian Throughflow. Journal of Climate, 26(9), 2845–2861. doi: 10.1175/jcli‐d‐12‐00117.1 Yuan, D., X. Li, Z. Wang, Y. Li, J. Wang, Y. Yang, et al. (2018). Observed transport variations in the Maluku Channel of the Indonesian Seas associated with western boundary current changes. Journal of Physical Oceanography, 48(8), 1803–1813. doi: 10.1175/jpo‐d‐17‐0120.1 Zhou, H., D. Yuan, L. Yang, X. Li, & W. Dewar (2018). Decadal variability of the meridional geostrophic transport in the upper tropical North Pacific Ocean. J. Climate, 31, 5891– 5910. https://doi.org/10.1175/JCLI‐D‐17‐0639.1 Zhou, Q., W. Duan, M. Mu, & R. Feng (2015). Influence of positive and negative Indian Ocean Dipoles on ENSO via the Indonesian Throughflow: Results from sensitivity experiments. Advances in Atmospheric Sciences, 32(6), 783–793. doi: 10.1007/s00376‐014‐4141‐0 Zhang, Y., J. M. Wallace, & D. S. Battisti (1997). ENSO‐like interdecadal variability: 1900–93. J. Clim., 10, 1004–1020. doi:10.1175/1520‐0442(1997)0102.0.CO;2 Zhang, Y., M. Feng Y. Du, H. E. Phillips, N. L. Bindoff, & M.  J.  McPhaden (2018). Strengthened Indonesian Throughflow drives decadal warming in the Southern Indian Ocean. Geophysical Research Letters, 45, 6167–6175. https:// doi.org/10.1029/2018GL078265 Zhuang, W., B. Qiu, & Y. Du (2013). Low‐frequency western Pacific Ocean sea level and circulation changes due to the connectivity of the Philippine Archipelago. J. Geophys. Res. Oceans., 118(12), 6759–6773. doi:10.1002/2013JC 009376

16 Impact of El Niño on Weather and Climate Extremes Lisa Goddard1 and Alexander Gershunov2

ABSTRACT El Niño influences weather and climate extremes in regions around the world. Direct impacts of El Niño on climate extremes include tropical drought and global mean temperature. In specific regions where El Niño is known to impact the probabilities of various weather and climate extremes, it is never the only factor. Rather, El Niño acts in concert with other influences occurring on longer or shorter timescales. This chapter describes the processes that can enhance El Niño’s expected impact in a region, resulting in a relatively rare outcome. The arguments are supported with regional examples, which make up the bulk of the chapter. Anthropogenic climate change may add to these extremes, though several processes are poorly understood, and constitute an area for future investigations. The prospects for predicting extremes are then discussed. Many of the devastating impacts of El Niño, as well as of climate change, are delivered through extremes. Prediction of these for management, preparedness, and planning thus constitutes a central part of adaptation and climate risk management. The silver lining of El Niño events is that the prediction skill of monthly to seasonal forecasts is substantial. But skillful forecasting of extremes still requires the ability to predict the details of the El Niño event, its effects on other tropical ocean basins, the atmospheric response to this changed pattern of tropics‐wide SSTs, and other contributing phenomena, such as the Madden‐Julian Oscillation. Additionally, forecasts should be considered in the context of the nonstationary climate that arises from both anthropogenic change and natural decadal variability.

16.1. INTRODUCTION The most devastating impacts of El Niño events are experienced through weather and climate extremes and the effects they have on food, water resources, health, and safety. Paradoxically, one of the most common misconceptions about El Niño is that it leads to more hydrometeorological extremes and associated disasters worldwide. While it is certainly true that some regions see extreme weather and climate associated with El Niño, these types of disasters occur in some parts of the world every year. The knowledge that El Niño and La Niña events have

1  International Research Institute for Climate and Society, Columbia University, Palisades, NY, USA 2   Scripps Institution of Oceanography, University of California–San Diego, La Jolla, CA, USA

expected climate impacts, some disastrous, has motivated enhanced scrutiny and accounting of the cost in human and economic terms during these events, which give the illusion that those costs are unusual. And while some weather and climate‐related disasters are on the rise due to climate change, as well as to the increase of population and assets that can be affected, and changes in surveillance and reporting of these impacts, El Niño events do not on their own cause more disasters overall (Goddard & Dilley, 2005). The difference is that during El Niño we have much more information about where those hydrometeorological disasters are likely to occur than during ENSO‐ neutral or La Niña conditions. Observations over the last century, or more, and proxy records for some parts of the world reveal where El Niño events are associated with floods or droughts or extreme temperatures, precipitation, etc. (e.g. Gergis & Ashcroft, 2013). Historical maps

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 361

362  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

of seasonal rainfall anomalies during El Niño events (Figure 16.1)1 provide a synthesis, though perhaps locally incomplete, of the global teleconnections that are statistically associated with ENSO events. Such maps of expected precipitation teleconnections are now regularly shown in newspapers and other media outlets during El Niño events. It was one of these teleconnections, namely the failure of the Indian monsoon in the late 19th century, that sparked the first inquiry into the globally connected nature of these climate anomalies. The catastrophic failure of the Indian monsoon in 1877, 1899, and 1918 provided great impetus to understand and predict the vagaries of the monsoon. Tasked with this challenge, Sir Gilbert Walker discovered the Southern Oscillation when he correlated all‐India rainfall (AIR) with pressure observations at the existing long‐term meteorological stations around the world (Walker, 1924; Walker & Bliss, 1932). ENSO teleconnections represent historically robust regional climate anomalies, but they are not altogether stationary or dependable. They are not realized for every El Niño–impacted region in every event, even during the strongest El Niño events. Examples include a normal Indian monsoon and South Africa summer rainfall season when dry conditions would have been expected during the 1997–1998 strong El Niño, and the absence of wet conditions for the southwestern United States during the 2015–2016 strong El Niño. In other El Niño events, these aforementioned teleconnections have resulted in extreme precipitation anomalies and crisis conditions for affected populations. ENSO teleconnections to AIR, in particular, are known to wax and wane on multidecadal timescales (e.g. Pant et al., 1988; Gershunov et al., 2001, and references therein). The timing of Sir Gilbert Walker’s studies, therefore, was fortunate as they were preceded by several decades of strong ENSO teleconnections to AIR. Phenomena that interfere with El Niño’s teleconnections in some years may enhance the anomalies in other years, leading to extreme weather and climate for specific regions. Sizable anomalies in seasonal mean precipitation or temperature are typically accompanied by shifts in expected risks of extremes (e.g. Supari et al., 2018). For example, with an expected warming in temperatures, one might expect an increase in the frequency or intensity of heat waves or a decrease in the frequency and intensity of cold spells. An expectation of more seasonal precipitation might indicate a reduction in dry spells and/or an increase in the frequency of rainfall of a given intensity or duration. On the other hand, exceptions to this rule have also been highlighted (e.g. Gershunov, 1998; Guirguis et al., 2015). The changes in frequency and/or 1  Although similar to Figure 1.2a in chapter  1, this figure importantly shows the seasonality of the impacts rather than the impacts over a specific season.

magnitude of extremes themselves affect the seasonal climate, however, such as more frequent storms brought to the southern United States from an enhanced subtropical jet. It is also possible that El Niño teleconnections may be enhanced by constructive interference with higher‐frequency weather (Muñoz et  al. 2015), lower‐frequency decadal variability (Gershunov & Barnett, 1998b), or by climate change (Power et al., 2017). The assumption that there are more extremes worldwide during El Niño events than at other times would obviously make El Niño a significant global threat. Because rainfall extremes worldwide are not significantly more frequent during El Niño, and given the greater predictive power during these events, El Niño should be viewed as a significant opportunity globally. Climate anomalies, and particularly extremes, orchestrated by El Niño teleconnections can be anticipated, and thus intervention strategies can be put in place to reduce adverse impacts. In this chapter, we summarize the state of knowledge about the connection between El Niño and extreme weather and climate events, spanning mainly precipitation and temperature extremes and some of their mechanistic causes, for impacted regions around the globe. This text is not exhaustive and is, admittedly, selective. Our treatment of El Niño teleconnections to extreme weather and climate is meant to be illustrative. We also describe historical predictability of extremes afforded by El Niño as well as the inherent probabilistic nature of such predictions. We acknowledge that La Niña impacts are also important and, in some cases, more predictable than El Niño impacts (e.g. Gershunov, 1998; Power et  al., 2006; Cai & van Rensch, 2012). However, in the interest of brevity, we intentionally avoid discussing La Niña teleconnections, which are not simply the opposite of El Niño teleconnections (especially as concerns extremes and predictability), as is commonly claimed and implicitly assumed in various linear analyses. Finally, we describe outstanding questions and research challenges, particularly with respect to seasonal climate predictability in the nonstationary climate of the present and future. 16.2. EXTREME CLIMATE IMPACTS 16.2.1. Direct Impacts 16.2.1.1. Tropical and Global Temperatures El Niño events are visible in the globally averaged temperature record. The estimated contribution of El Niño to globally averaged temperatures is about 0.1°C for each 1°C of the Niño‐3.4 SST anomaly index (Trenberth et al., 2002). Most of this warming manifests in the tropics (Yulaeva & Wallace, 1994) and is caused by additional

EI Niño and Rainfall El Niño conditions in the tropical Pacific are known to shift rainfall patterns in many different parts of the world. Although they vary somewhat from one El Niño to the next, the strongest shifts remain fairly consistent in the regions and seasons shown on the map below.

Figure 16.1  Expected changes in precipitation regionally during an El Niño event. The time of year these precipitation anomalies typically occur is indicated over each region. (Source: https://iri.columbia.edu/wp‐content/ uploads/2016/05/ElNino_Rainfall.pdf)

364  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

latent heat released in the midtroposphere as convection in the tropical Pacific shifts from Indonesia toward the equatorial central Pacific and expands. As the atmosphere in the deep tropics cannot maintain pressure gradients, the anomalous heating is quickly communicated throughout the tropics. 16.2.1.2. Tropical Drought  Like tropical temperature, the strength and extent of tropical drought also correlates strongly with the strength of El Niño (Figure 16.2). Lyon (2004) found that for all categories of drought2, moderate, intermediate, and severe, the amount of tropical land under drought increases with the magnitude of El Niño’s SST anomalies. In the weakest events, only a few percent of tropical

% Tropical land area

(a)

40

moderate (r = 0.94)

35

intermediate (r = 0.92)

30

severe (r = 0.89)

25 20 15 10 5 0 0.5

(b)

40

2.5

moderate (r = 0.87)

35 % Tropical land area

1 1.5 2 Nino 3.4 SST Anomaly (deg, C)

intermediate (r = 0.86) severe (r = 0.82)

30 25 20 15 10 5 0

0

0.5 1 1.5 2 Nino 3.4 SST Anomaly (deg, C)

2.5

Figure 16.2  Scatter plots of peak drought extent (% of tropical land area, for three levels of drought severity) versus maximum Niño‐3.4 SST anomaly (°C) during the associated El Niño, with (a) showing the 10 strongest events and (b) all events from 1950–2003. The filled symbols are for the 2002–2003 El Niño, where drought extent was computed using the CMAP data set. (from Lyon 2004) 2  If rainfall distributions were Gaussian, the standardized categories would correspond to moderate drought = 1-in-11-year event, intermediate = 1-in-23-year event, and severe = 1-in-43-year event.

land area experiences severe drought, but that proportion increases to about 10% for the strongest events. The amount of tropical land experiencing intermediate‐level drought in the strongest events is about 15%–20%, and about one‐third of the area experiences at least moderate drought. This study was based on a 12‐month standardized precipitation index, which agrees with other drought measures such as the Palmer Drought Severity Index (PDSI) (Lyon, 2004). The drought‐impacted areas are visible on the teleconnection map (Figure  16.1) and include northern South America, parts of southern Africa, India, the Maritime Continent, and much of Australia. Although the teleconnection map does not address the magnitude of the impacts or the range of variations due to characteristics of a specific El Niño, Lyon (2004) indicates that most of these areas do have a heightened risk of protracted drought during El Niño events. These El Niño‐induced droughts are spatially coherent and nearly synchronous, with a 3–4 month lag between the timing of the peak SST anomaly in the Niño‐3.4 region and the peak tropical drought extent. The linear connection between El Niño and synchronous tropical drought can be understood through the role of tropospheric heating on tropical precipitation. As mentioned above, the whole tropical free troposphere warms in response to El Niño; thermal damping of SST anomalies contributes some atmospheric warming, but midtropospheric heating dominates the warming as convection over the tropical Pacific ocean increases in strength and spatial extent (Yulaeva & Wallace, 1994; Sobel et  al., 2002). The atmospheric heating is rapidly distributed through the tropics. The increase of temperature aloft stabilizes the lapse rate and inhibits convection. Modeling studies of Chiang and Sobel (2002) and Neelin et  al. (2003) suggest that this stabilizing effect creates the large‐ scale tropical precipitation response during El Niño, which is then modified regionally depending on the character of the underlying surface, land‐atmosphere interaction, and varying aspects of the atmospheric circulation. For example, local maxima in tropical SSTs will encourage low‐level convergence and localized convection (Gill, 1980). Precipitation and latent heat release occur in these convergent areas, while descending, dry circulations are found throughout the rest of the tropics (e.g. Goddard & Graham, 1999). 16.2.2. Indirect and Multifactor Impacts The expected El Niño teleconnections are not extreme in the statistical sense, because El Niño events occur about every 3–7 years. Local extremes in teleconnection regions may occur due to weather phenomena indepen­ dent of El Niño, and during periods when ENSO is not driving anomalous regional climate. It is also possible

Impact of El Niño on Weather and Climate Extremes  365

that the constructive or destructive interference of large‐ scale climate modes acting at longer and shorter timescales can enhance or diminish the expected climate anomaly from El Niño. Interaction between intraseasonal variability and El Niño–driven climate anomalies may be described through a conceptually simple dice analogy (Gershunov, 1998). Imagine that each face of a die corresponds to one of the several preferred weather regimes. For example, over North America, one side would correspond to the well‐known Pacific–North American pattern and the surface weather patterns associated with it. Other faces of the die would correspond to other characteristic regimes (e.g. Michelangeli et al., 1995). The effect of the ENSO phenomenon is to change the loading in the regional climate die, so that during El Niño winters, most outcomes of a die throw are still possible but with probabilities that differ from the climatological ones depending on ENSO phase, intensity, seasonality, and other factors. The extratropical circulation is affected by  ENSO mainly through its effect on the frequencies of  occurrence of some typical weather regimes or weather types. Modulation of El Niño teleconnections is possible on longer timescales as well. The similarity in spatial pattern of the Pacific Decadal Oscillation3 (PDO; Mantua et al., 1997) to ENSO suggests an atmospheric signal that is similar. Studies that contrast North American climate anomalies (including occurrence of extreme precipitation) during the positive and negative phases of PDO have consistently revealed constructive interference of the positive and negative ENSO and PDO phases, and destructive interference when ENSO and PDO are out of phase (e.g. Gershunov & Barnett, 1998b). In other words, canonical and predictable El Niño signals tend to be associated with the positive phase of the PDO, which presents warm SST anomalies along the west coast of North America and cool SST anomalies in the central North Pacific, while El Niño events occurring during the negative phase of the PDO have less consistent and predictable impact on North American climate. 16.2.2.1.  Regional Precipitation and  Streamflow Extremes The spatial structure of temperature changes and shifts in location of atmospheric heating lead in turn to changes in atmospheric circulation that result in regional climate anomalies. During El Niño events, the increased tropical rainfall is mostly over ocean areas, but a considerable amount of moisture is fed into the extratropics. 3  Note, the term Pacific Decadal Oscillation (PDO) refers to robust patterns of decadal variability in the Pacific and is not distinct from Interdecadal Pacific Oscillation (IPO).

Extreme rainfall poses societal risks when it results in flooding, but it is not synonymous with flooding. Flooding can occur for several reasons: extreme seasonal rainfall, extreme daily rainfall within an overall normal rainfall season, or even persistent lower‐intensity rainfall that saturates the soil to the point of flooding. Additionally, flooding need not occur at the same location as the rainfall. Streamflow integrates the effect of rainfall, snow accumulation and melt, and rain‐on‐snow and warm pulses over a watershed or larger region. Changes in the risk of anomalously high or low streamflow in a given season (Figure 16.3a) are spatially very consistent with the rainfall anomalies identified as El Niño teleconnections (Figure  16.1). A measure of streamflow that considers conditions that are more extreme than the 1‐in‐10‐year low/high flow (Figure  16.3b) indicates that these flood‐ prone areas are central to the larger areas. Changes in risk of extremes can often be illustrated, at least historically, through a comparison of the frequency distributions of some variable, such as streamflow (Figure 16.4), between all years and El Niño years. The shifted distributions, like the loaded die, show an altered range or frequency of outcomes. It is a range. The risk of extremes increases dramatically relative to the magnitude of the climate anomaly, at least if an analytical distribution is assumed. Some El Niño years in a region that experiences wet conditions, like Tanzania (Figure 16.4a), were very wet, while other El Niño years were more moderate. Similar is the rare but evident occurrence of extreme drought in the dry teleconnection region over Zimbabwe. In most cases, the extreme outcomes were the result of multiple factors acting together. Madden‐Julian Oscillation (MJO). A relationship exists between El Niño and the MJO. Some studies suggest that enhanced MJO activity in boreal spring enhances the chance of a developing El Niño (Hendon et al., 2007), as westerly wind bursts create downwelling Kelvin waves that can depress the equatorial thermocline in the central and eastern Pacific. It may also be true that El Niño conditions change the characteristics of the MJO cycle. Coupled MJO activity may survive farther east in the Pacific when the warm pool and mean convection move eastward, even if the amplitude of the MJO doesn’t change in the far western Pacific or Indian ocean (Hendon et al., 1999; Lau, 2005). Still, they are unique phenomena, but their interaction, given relative phase, season, and location, can be an important factor in the risk of extreme precipitation. The 60–90 day MJO can lead to locally extreme rainfall and flooding in many parts of the world, as it propagates eastward through the tropics. The coincidence of a “wet phase” of MJO4 over a region with a wet El Niño 4  This wet phase may also be a teleconnection, rather than a local passing of the MJO.

366  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Probability of streamflow (DJF) in El-Niño years

(a)

SHADED AREAS ARE PROBABILITY BELOW AND ABOVE NORMAL UNSHADED ARES ARE NEAR-NORMAL A ARID AEAS (NO VALUE)

80% 70% 68% 50% 40% 40% 50% 60% 70% 80% BELOW-NORMAL

NORMAL

ABOVE-NORMAL

Probability of streamflow (DJF) in El-Niño years

(b)

SHADED AREAS ARE PROBABILITY BELOW-10% AND ABOVE-90% UNSHADED ARES ARE ONLY 10-90% A ARID AEAS (NO VALUE)

40% 35% 30% 25% 20% 20% 25% 30% 35% 40% BELOW-10%

10%-90%

ABOVE-90%

Figure 16.3  Probability of a change in simulated streamflow worldwide during the peak phase of El Niño (Dec‐ Jan‐Feb), (a) showing likelihood of above‐/below‐normal tercile streamflow, and (b) showing likelihood of being more extreme than the 10% tails of the historical distribution. The color of the shading indicates whether the risk increases (blue–pink) or decreases (yellow–red). White areas indicate no discernible change in flood risk. Streamflow simulations are described in Ward et al. (2013).

t­eleconnection can increase the risk of extreme rainfall. Similarly, the phase of the MJO may lead to dry spells over areas already at risk of below‐normal rainfall during El Niño. For example, while the MJO is in its convective

phase over the Indian Ocean, breaks in the monsoon rainfall are experienced over India. Over southeastern South America, extreme rainfall events are associated with the occurrence of three

Impact of El Niño on Weather and Climate Extremes  367 (a) All years El-Niño years BN: 03% NN: 25%

f(x)

AN: 72% B10: 00% A90: 24%

NN

BN 0

100

AN

200

300

400

500

600

700

DJF streamflow (cms)

(b) All years El-Niño years BN: 72% NN: 25%

f(x)

AN: 04% B10: 26% A90: 00%

BN 0

50

NN 100

AN

150

200

250

300

350

400

DJF streamflow (cms)

Figure 16.4  Streamflow distributions for climatology (43 years, black dots and line) and El Niño (10 years, blue dots and line) of (a) Location 1 (Tanzania) and (b) Location 2 (Zimbabwe). Red dotted lines are 10% and 90% streamflow percentiles. BN, NN, AN, B10, and A90 mean probability of below‐normal tercile, near‐normal tercile, above‐normal tercile, below 10%, and above 90%, respectively. Log‐normal distributions of seasonal streamflow are used for both the full record and El Niño years. The probabilities of below 10% and above 90% are only constructed on a limited number of El Niño years (10); thus, caution should be taken in interpretation.

f­ eatures: baroclinic fronts (Seluchi et al., 2006), heat and moisture advection (Salio et  al., 2002), and mesoscale convective complexes (Velasco & Fritsch, 1987; Ferreira et  al., 2003). These features can be synthesized into weather types. Muñoz et al. (2015) summarized Dec‐Jan‐ Feb rainfall characteristics into six weather types, two of which are associated with extreme rainfall. One of those is significantly more prevalent during El Niño conditions and appears to be influenced by the MJO. Other features also contribute to the attribution of extreme events, like the Southern Annular Mode, but MJO and ENSO capture most of the variability. The weather types may only persist for a matter of days, but the large‐scale climate drivers modulate the frequency of occurrence of

these patterns. Cross‐time‐scale (constructive) interference is found between the large‐scale drivers, such that their contemporaneous effect is greater than the effect that would conclude from the simple addition of their conditional probabilities (Muñoz et al. 2015). Atmospheric Rivers5 (ARs). Extreme and strongly orographic precipitation events and winter floods along the west coast of North America, as well as mountainous west coasts of other continents, are historically associated with atmospheric rivers. The warm, orographic precipitation associated with ARs tends to generate more runoff from topography and streamflow in elevated basins than other types of winter storms. Thus, intense rainfall delivered through ARs leads not only to extreme local precipitation, it also results in more remote downstream flooding than other, colder winter storms (e.g. Konrad & Dettinger, 2017). Although the frequency and intensity of these long, thin ephemeral bands of intense water vapor transport do not seem to be particularly sensitive to ENSO (Gershunov et al., 2017), their orientation at landfall does exhibit ENSO sensitivity (Guirguis et al., 2018), which influences precipitation patterns and contributes to the canonical El Niño and La Niña anomaly patterns. For example, Guirguis et  al. (2018) show that ARs landfalling at around 40°N, the fulcrum latitude at which ENSO‐precipitation anomalies change sign over western North America, tend to be more southerly oriented during El Niño conditions and result in heaviest orographic precipitation at, and south of, landfall (on the slopes of the coastal ranges of central California, Sierra Nevada, and the Transverse Ranges of southern California). Orientation of the mountains relative to the direction of moisture transport at landfall becomes key to defining regional patterns of extreme orographic precipitation due to landfalling ARs. It is not surprising, then, that flood damages, which are largely driven by ARs in the Western United States (Corringham et al., 2019b), are to a large extent orchestrated by ENSO (Corringham et al., 2019a). Cut‐off lows. Another archetypal meteorological cause of extratropical extreme precipitation events is the upper‐ level lows that detach from the dominant midlatitude westerly flow. Formation of cut‐off lows tends to be associated with Rossby wave breaking (Ndarana & Waugh, 2010), but once formed, they can operate independently of polar jet stream dynamics, sometimes stalling to produce copious precipitation persisting for days. Cut‐off lows are associated with extreme precipitation, and their behavior is partially coordinated by El Niño teleconnections over the South Pacific region (Favre et  al., 2012) and western North America (Oakley & Redmond, 2014).

5 http://glossary.ametsoc.org/wiki/Atmospheric_river

368  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Cut‐off lows are thus among the mechanisms by which ENSO impacts extreme precipitation remotely. SST patterns over other tropical oceans contributing to extremes. The “atmospheric bridge” was first described in the late 1990s to explain how SST anomalies in the Pacific can influence SST patterns on other parts of the global ocean through their effect on the atmospheric circulation (e.g. Klein et al., 1999; Alexander et al., 2002). El Niño’s large and persistent SST anomalies influence the location and strength of large‐scale convection and subsidence associated with the Walker Circulation and the other zonal overturning circulations. Those changes affect cloud cover and evaporation, which then can increase heat flux into remote oceans. Klein et al. (1999) find that increased heat flux is the primary factor in the surface warming of these oceans. For example, over the eastern Indian Ocean and South China Sea, enhanced subsidence during El Niño reduces cloud cover and increases the solar radiation absorbed by the ocean, thereby leading to enhanced SSTs there. In the tropical North Atlantic, a weakening of the trade winds during El Niño reduces surface evaporation, which increases SSTs. Changes in the low‐level wind patterns can also change remote ocean temperatures through advection, as well as through the triggering of coupled ocean‐atmosphere processes (Yu & Reinecker, 1999; Saji et al., 1999). Over the Indian Ocean, the pattern of SST response is key to the resulting climate anomalies. For example, if the Indian Ocean did not warm during El Niño events, the Greater Horn of Africa would be dry in those years (Goddard & Graham, 1999) rather than wet, as is typically experienced (Figure 16.1). The preferential warming of the western tropical Indian Ocean provides a zonal temperature gradient conducive to low‐level easterly winds over the Indian Ocean that converge, causing heavy precipitation, over eastern Africa. The western Indian Ocean warms preferentially because the shifting Walker Cell brings subsidence over the Maritime Continent, resulting in anomalous low‐level divergence (i.e. easterly flow to the west and anomalously westerly flow to the east). As an equatorial ocean basin, the Indian Ocean responds to the wind anomalies through equatorial wave adjustment, by which an equatorial Kelvin wave and off‐ equatorial Rossby waves will be generated (Yu & Renecker, 1999). Because these waves are forced by easterly winds, the Kelvin wave is upwelling, bringing colder conditions to the west near Indonesia, and the Rossby waves are downwelling, pushing down the thermocline in the western Indian Ocean. These waves generate the negative zonal SST gradient. The subsequent additional response from the winds reinforces the SST gradient pattern (Saji et al., 1999), which has been referred to as the Indian Ocean zonal mode and the Indian Ocean Dipole (IOD). As the equatorial Indian Ocean couples to the overlying

atmosphere for periods of weeks to a month or two, it enhances the El Niño teleconnection, causing flooding conditions in eastern Africa (Goddard & Graham, 1999) and drought in Indonesia (Pan et al., 2018). During the 1997–1998 event, the SST gradient was about five times the standard deviation of the IOD (Webster et al., 1999), and the October–December rain season was one of the wettest on record (Halpern & Bell, 1998). The tropical Atlantic SSTs also respond to Pacific variability; SSTs over the northern tropical Atlantic warm following the peak of El Niño. The shift of convection to central‐eastern Pacific leads to subsidence over the Caribbean and northern South America, which reduces cloudiness directly and reduces evaporation by slowing the northeasterly trade winds (Curtis & Hastenrath, 1995; Klein et  al., 1999). Both changes contribute to warming SSTs north of the equator. Additional reinforcement of the SST anomaly pattern through local air‐sea coupling may also contribute to the dry conditions (Chang et al., 1997). As the rainy season over northern South America, particularly northern Brazil, is determined by the seasonal migration of the Atlantic Intertropical Convergence Zone (ITCZ), the anomalous subsidence and the anomalous meridional SST gradient both influence that evolution (Lindzen & Nigam, 1987). During an El Niño event, a positive anomalous SST gradient (i.e. greater warming north of the equator) prevents the ITCZ from migrating as far south as it otherwise would, and keeps rain away from the region. This locally driven climate anomaly then contributes to the drying already contributed by El Niño’s subsidence. Droughts over northeastern Brazil are enhanced when El Niño events are accompanied by this response in Atlantic SSTs (Giannini et al., 2001). 16.2.2.2.  Correlated Extremes (e.g. Extratropical Drought and Temperatures) Drought may also serve as an additional driver of heat waves. In the extratropics, regional drought is often induced by remotely forced anomalous stationary waves. When the surface cooling effect of evaporation is replaced by sensible heating from the dry soil, temperatures increase much more rapidly, at which time the drought can enhance the heat wave even once the anomalous waves dissipate (Lyon & Dole, 1995). This also suggests that antecedent conditions can contribute to the severity of drought, and of associated heat spells, by reducing the buffering capacity of soil moisture. Particularly in semiarid regions that experience greater persistence of dry conditions during El Niño, heat waves are a concomitant risk. For some regions, the connection between drought and temperature extremes can be delayed by months. For example, during El Niño winters, the Pacific Northwest tends to receive lower snow accumulations, which leads to drier soils and hotter temperatures in the following spring

Impact of El Niño on Weather and Climate Extremes  369

and summer. Those environmental conditions can lead to elevated fire danger in montane forests of the western U.S. (e.g. Westerling et al., 2003). 16.2.3. Changing Risk of Extremes Due to Climate Change Higher global temperatures enhance the risks of local and regional extremes (Seneviratne et al., 2018), and climate change has already become a discernible factor in many weather extremes (NASEM, 2016), including heat waves, extreme precipitation, and to a lesser degree drought. As with the distributions shown in Figure 16.4, when temperatures increase, the whole distribution moves toward warmer outcomes, and the risk of extreme temperatures goes up. The tropical and regional temperature increases associated with El Niño are accompanied by anthropogenic warming and thus exacerbated. The warmer atmosphere holds more moisture, and many regions are becoming more arid (Dai & Zhao, 2017), even with little or no change in annual rainfall. Tropical drought severity and extent increases with the magnitude of El Niño (Lyon, 2004). Drier soil conditions ahead of the El Niño event, and higher temperatures throughout, intensify the drought conditions (Cayan et  al., 2010). Increased atmospheric water vapor in a warmer world can also lead to more intense rainfall events (Groisman et al., 2005) separated by longer dry spells (Groisman et  al., 2008). This is at least in part responsible for the very small relative trends in seasonal and annual precipitation compared to the natural variability. Most Mediterranean climate regions may be the exceptions, where drying trends consistent with projections (Polade et al., 2017) are already observed (Seager et al., 2019). These trends resulting from decreasing frequency of precipitation and increasing reliance on extremes are also associated with greater variability of total precipitation from year to year (Polade et  al., 2014, 2017), leading to increased probability of flood and drought, which may continue to be partially orchestrated by ENSO teleconnections in impacted regions (e.g. California). With increased atmospheric water vapor in a warmer world, atmospheric rivers are projected to become wetter, longer, fatter, and more impactful (Payne & Magnusdottir, 2015; Espinoza et al., 2018; Gershunov et al., 2019), mainly due to thermodynamic moistening. This leads to an increasing contribution of atmospheric rivers to the hydroclimate of impacted regions, like western North America, via increases in extreme precipitation (Gershunov et  al., 2019). El Niño teleconnections that result in extreme precipitation due to the action of atmospheric rivers would therefore result in stronger and warmer orographic extremes and greater challenges to water resource management, particularly in California and Baja California Norte.

Recent studies have attempted to describe the likely changes in El Niño teleconnection patterns due to climate change (e.g. Meehl & Teng, 2007; Power et  al., 2017). However, the results are not terribly conclusive. First, on the question whether El Niño itself will change, which could affect the teleconnections, there is no consensus from the ensembles of even individual models about how ENSO will change in a warmer world (Coelho & Goddard, 2009), given the tremendous variability inherent in ENSO (Wittenberg, 2009). While some recent studies show that the general circulation models’ (GCMs, or global climate models) projections for the end of century produce more of the very strong El Niño events, like 1997–1998 and 2015–2016, the overall frequency of El Niño events decreases (Cai et al., 2014). Additionally, within the tropics, climate change projection models do not indicate a change in the spatial pattern, severity, or extent of El Niño precipitation anomalies. However, increased temperatures can still enhance drought conditions, as mentioned earlier. The impact of climate change on extratropical teleconnections may be more nuanced and more difficult for the models to capture. Polar‐amplified global warming may impact ENSO teleconnections to the midlatitudes through a reduced equator‐to‐pole temperature gradient. A less vigorous midlatitude circulation can decrease the efficiency with which tropical signals propagate to the midlatitudes. On the other hand, weaker midlatitude westerlies and increased wave amplitude (Francis & Vavrus, 2012; Mann et al., 2017) could promote more persistent and therefore more impactful ENSO signals. These hypotheses are difficult to answer through observations to date, and even to verify in models. Over North America, many GCMs simulate the main observed relationship between patterns of Pacific SSTs and winter precipitation (Polade et  al., 2013). These models could be used in “Pacemaker” or other idealized experiments (e.g. Kosaka & Xie, 2016) to test the stability of the North American teleconnection pattern given observed or predicted large‐scale changes in the mean seasonal climate. The need for improved knowledge around these issues points to timely directions of future research. It should also be acknowledged that many of the known teleconnections (Figure  16.1) are nonstationary, meaning the strength of relationship between the regional climate anomalies for a given location and El Niño fluctuates over time. To determine whether long‐term changes are due to anthropogenic global warming or are likely natural bootstrapped observations, GCM simulations of the past century and/or long preindustrial simulations can be scrutinized to determine the expected level of natural variability in the teleconnection, and the level of agreement between what has been observed and the tools that are used to inform the future. Such studies would

370  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

help develop objective expectations about whether the recent changes in El Niño teleconnections, for example to western North America, can be expected to be temporary due to natural variability, or are likely to represent the evolving future of El Niño impacts in a warming world. This would shed much needed light on future expectations from ENSO‐based seasonal predictability. 16.3. PREDICTABILITY: HOW AND HOW WELL DO WE PREDICT ENSO’S EXTREME IMPACTS? From India and Australia to the Americas and Africa, ENSO has traditionally been the main source of seasonal climate predictability, with stronger events leading to greater predictability worldwide (Goddard & Dilley, 2005). For a given region, however, that relationship may appear to break down due to atmospheric stochasticity (Gershunov et al., 2001), the particular characteristics of individual El Niño events (Capotondi et al., 2015), and/ or the interference of other climate drivers (e.g. Muñoz et al., 2015; Gershunov & Barnett, 1998b). The Indian monsoon was first to be linked with ENSO (Walker, 1924; Walker & Bliss, 1932), and ENSO has since traditionally been the main predictor of AIR (Charney & Shukla, 1981). However, the teleconnection between ENSO and AIR seems to wax and wane (Gershunov et al., 2001). For example, the exceptionally strong 1997– 1998 El Niño was uncharacteristically associated with a stronger than average monsoon (Kumar et  al., 2006), while a very dry monsoon occurred in 2002 during weak to moderate La Nina conditions. These examples to the contrary notwithstanding, deficient monsoons, including some of the driest extremes, have coincided with El Niño onset, while copious monsoons have tended to be associated with developing La Nina conditions (Figure 16.5). El Niño teleconnections are expectations; they are probabilistic and are not guaranteed to materialize in every El Niño event. It is possible to recast teleconnection maps to show empirical likelihood based on how frequently the regional climate anomaly was present during past El Niño events (Mason & Goddard, 2001; Lenssen et al., 2019). Information about the robustness and seasonality of the teleconnections is very important for preparedness, and this represents a small step toward communicating ENSO impacts. However, information about the evolution of the particular El Niño event, as well as conditions in the other tropical oceans, also impacts the forecast. Lenssen et  al. (2019) found that seasonal dynamical forecasts provide higher skill globally than empirical forecasts based on ENSO state alone. Of course, statistical techniques that take other influences beyond the canonical ENSO state into account are also available regionally (e.g. Gershunov & Cayan, 2003). Whether dynamical or statistical, probabilistic forecasts during an El Niño event can and should be conditioned

on the specifics of the event and the general state of the climate system beyond just ENSO. Antecedent conditions in the environment and/or seasonal lags in the climate system can contribute to the predictability, as well as the extreme nature, of the El Niño impact. For some regions, it may be possible to predict a change in the risk of specific extremes up to a year in advance. Over the northwestern U.S., for example, where El Niño typically leads to deficient winter rainfall and snow accumulation, the resultant soil moisture deficiency the following summer is likely to result in enhanced probability of catastrophic montane forest fires (Westerling et al., 2002, 2003). Such signals are, in principle, predictable starting as far back as the emergence of the El Niño anomaly the previous summer, or even earlier with skillful El Niño prediction. This expectation would be reversed for the southwestern U.S., where El Niño is historically associated with excess winter and spring precipitation. In this case, La Niña would lead to the opposite impacts. North America is one place where seasonal climate prediction, even with respect to weather extremes, is possible thanks to ENSO’s evolution being largely predictable starting during its development phase in boreal summer, into its final mature phase in winter, when ENSO exhibits its strongest teleconnections to North America (Livezey & Timofeyeva, 2008). ENSO’s influences on extreme precipitation and hydrology have been demonstrated for North America (Gershunov & Barnett, 1998a; Cayan et al., 1999), primarily in concert with the PDO (Gershunov & Barnett, 1998b). The ENSO‐PDO constructive interference (i.e. El Niño/PDO+, La Niña/PDO–) is the observed combined leading teleconnection mode responsible for seasonal predictability of heavy daily precipitation frequency over the conterminous United States in association with sea surface temperature predictor patterns (Gershunov & Cayan, 2003). Although the PDO is an amalgam of influences including ENSO (Schneider & Cornuelle, 2005; Newman et al., 2016), for practical applications to seasonal prediction, since the PDO index does display strong multidecadal variability and tends to be quite persistent, it can and should be taken into account when making seasonal forecasts based on ENSO state. This is certainly true for predicting the probability of heavy and extreme precipitation occurrence over North America (Gershunov & Cayan, 2003). With similar caveats, the IPO, which measures changes in the North and South Pacific Ocean SST, is known to orchestrate ENSO teleconnections to precipitation and streamflow extremes over Australia (Kiem et al., 2003; King et al., 2013). Extremes are rare by definition, and forecasts must address the changing risks of extremes, rather than promising an accurate numerical forecast for an extreme weather or climate event. Studies have shown that extreme events become more likely with shifts in the mean conditions (e.g. Supari et al., 2018; Lee et al., 2015). Thus, their prediction is typically predicated on the prediction of

0 –100 –200

summer monsoon rainfall [mm]

100

Impact of El Niño on Weather and Climate Extremes  371

Jan 1860

–2.8°C –2.4°C

Jan 1880

–2°C

Jan 1900

Jan 1920

–1.6°C –1.2°C –0.8°C –0.4°C

Jan 1940

0°C

Jan 1960

0.4°C

0.8°C

Jan 1980

1.2°C

1.6°C

Jan 2000

2°C

2.4°C

2.8°C

Figure 16.5  This plot shows the relationship between summer monsoonal rainfall in India and ENSO. The height of the bars indicates the rainfall anomaly in the India monsoon: up means greater than normal, and down means less than normal. The rainfall data used are the June–September All‐India rainfall index from the India Meteorological Department (https://mausam.imd.gov.in), using the entire 1856–2016 as the climatology reference period. The color of the bars indicates the strength of ENSO during October–January for that year: red is a warm (El Niño) event, blue is a cold (La Niña) event, and off‐white indicates near‐normal sea surface temperatures in the east‐central equatorial Pacific Ocean (Niño‐3.4 region; 5°N–5°S, 120°–170°W; based on Kaplan SST; https://iridl.ldeo.columbia.edu/SOURCES/.KAPLAN/.EXTENDED/.v2/.ssta/). (Source: http://iri.columbia.edu)

seasonal means in the sense that if average seasonal temperature is predicted to be anomalously warm over a region, it is assumed to translate into greater (lesser) probability for warm (cold) extremes. It is worth noting, however, that an increase of risk in one tail of the distribution may not be accompanied by a decrease of risk in the opposite tail (e.g. Guirguis et al., 2015). As an example, during the extreme 2015–2016 El Niño event, much of the Philippines experienced severe drought, but the country was also hit during that time with two tropical cyclones that caused considerable local flooding. So part of predicting a changing risk of extremes should be to consider the background risks that may remain, as well as the risk of extremes that may be enhanced or reduced because of El Niño. Muñoz et al. (2015) showed that a set of daily atmospheric circulation regimes, or “weather types,” is sensitive to interferences of large‐scale climate drivers acting on different time scales, namely ENSO and the MJO. Further, predicted activity of these weather types can be used as potential predictors for the occurrence of extreme rainfall over south-

eastern South America. At the seasonal time scale, a combination of those weather types outperforms alternative predictors, such as SST patterns, phases of the MJO, and combinations of both (Muñoz et al. 2016). This is not, however, just a higher temporal update of the seasonal forecast with MJO information, or a summation of the odds. With consideration of the sequencing of weather types, including seasonality, one can build up a forecast that indicates risks of extreme precipitation occurring during specific parts of the season (i.e. early, mid, late). This type of forecast could offer useful subseasonal‐to‐seasonal climate information to those interested not only in whether extreme events will be more or less prevalent in the coming season, but also in how, when, and where those events are likely to occur. 16.4. CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE RESEARCH Globally, the weather and climate are not more extreme during El Niño events, but the extremes are more predictable.

372  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Certain extremes in particular regions are more or less likely to occur during El Niño. Extremes, whose probability of occurrence is bolstered by El Niño, include the following: Drought  –  tropics‐wide, as well as more punctuated droughts for India, Indonesia and South Africa Extreme precipitation  –  southeastern South America, parts of southern United States, especially the southwestern U.S., Greater Horn of Africa, and southeastern China Extreme streamflow  –  over many of the regions with extreme precipitation, including the seasonally lagged effect of snow melt in some locations Extreme wildfire behavior  –  e.g. northwestern United States, due to dry winters, deficient snowpack and soil moisture, leading to more likely montane forest fires during the following dry season (i.e. summer). 1. ENSO teleconnections are not necessarily extremes in year‐to‐year climate. This is because El Niño occurs pretty frequently: about every 3–7 years. However, El Niño events can selectively modulate the probabilities of extremes in various areas around the globe. As the range of possible seasonal climate outcomes shifts under the influence of El Niño, the extreme tails of the distribution can also change, though we note that the shape of the distribution may change as well. On occasions when the weather or climate anomaly reaches an extreme value, perhaps even causing crisis, there are usually additional factors contributing to the anomaly. 2. Other factors can enhance teleconnections, leading to extremes. Other phenomena in the climate system that are conducive to predictability can modulate the risk of extremes related to El Niño teleconnections, sometimes enhancing them, other times offsetting them. These can be higher‐ frequency phenomena like the Madden Julian Oscillation: El Niño typically brings wet conditions to southeastern South America, but within that season the extreme rainfall almost always occurs during just a couple phases of the MJO. The teleconnections can also be affected by lower‐frequency phenomena, such as the decadal variability in the Pacific and/or Atlantic oceans: El Niño’s influence on heavy precipitation over North America is sensitive to the phase of the Pacific Decadal Oscillation, showing greater robustness when El Niño and the PDO are in phase. Teleconnections that are moderated through the atmospheric bridge to other tropical oceans will be enhanced or muted according to the strength and persistence of the remote SST pattern driven by ENSO. Additionally, many types of extremes associated with ENSO teleconnections, as well as extremes in all years, are increasingly exacerbated by anthropogenic climate change. Significant changes, consistent with global warming, have been observed worldwide in temperatures, drought prevalence, and precipitation intensity on

extremely wet days. However, global warming is not uniform around the globe (e.g. polar amplification), and disruptions or enhancements may occur in the teleconnections and/or the weather extremes themselves. The scientific community is beginning to understand the nuanced regional impacts of global warming on specific weather extremes, such as heat waves, cold spells, droughts, and atmospheric rivers. The impact of global warming on ENSO teleconnections themselves is much less well understood. All of these factors, natural and anthropogenic, provide challenges to ENSO‐related predictability as well as opportunities for future research to result in enhanced predictability of extremes. 3. Climate is more predictable during El Niño, and there’s opportunity for improved prediction of extremes. Skillful predictions of El Niño can provide advanced warning to anticipate the regional climate anomalies 1–12 months in advance, including anomalous probabilities of extremes. Seasonal forecasts using dynamical climate models outperform empirical ENSO‐only outlooks, though the model predictions must be bias corrected and calibrated. Once an El Niño event begins, its further evolution becomes more predictable, and the lead time of early warnings for the risks of extremes can lengthen to a couple of seasons. For example, an increased risk of heavy wintertime precipitation and related extreme streamflow in the southwestern U.S. and northwestern Mexico can be skillfully anticipated at least as far back as the previous summer when El Niño conditions begin to clearly emerge. Newly emerging capacity for subseasonal prediction can potentially bridge the extreme weather forecasts with seasonal forecasts for greater specificity on impending extremes. Extended‐range subseasonal weather prediction can be made conditional on climate state, including ENSO. The MJO, for example, is a target of subseasonal prediction, and as skill improves, the weather and climate prediction communities will be able to make better use of the cross‐time‐scale interference of the MJO and ENSO in risk and hazard outlooks. The even more experimental area of decadal prediction may promote quantification of longer‐term risks, as it considers both the role of climate variability and change. The low‐frequency variability and persistence associated with the Pacific Decadal Oscillation, for example, provide an opportunity to anticipate whether the climate state reflected in the PDO phase will be conducive to stronger or diminished teleconnected signals from El Niño. Furthermore, polar‐amplified global warming impacts the atmospheric dynamics in which ENSO teleconnections operate and independently promotes specific extremes. Understanding how ENSO teleconnections, particularly to the midlatitudes, will evolve in a nonuniformly warmer future is a major emerging challenge for research into predictability of weather and climate extremes.

Impact of El Niño on Weather and Climate Extremes  373

ACKNOWLEDGMENTS The authors would like to thank Laurel DiSera and Weston Anderson for help with the literature search. REFERENCES Alexander, M. A., I. Bladé, M. Newman, J. R. Lanzante, N.‐C. Lau, & J. D. Scott (2002). The atmospheric bridge: The influence of ENSO teleconnections on air–sea interaction over the global oceans, J. Climate, 15, 2205–2231. Barnston, A. G., M. Chelliah, & S. B. Goldenberg (1997). Documentation of a highly ENSO‐related SST region in the equatorial Pacific. Atmos.–Ocean, 35, 367–383. Cai, W., & P. van Rensch (2012). The 2011 southeast Queensland extreme summer rainfall: A confirmation of a negative Pacific Decadal Oscillation phase? Geophysical Research Letters, 39(8). https://doi.org/10.1029/2011GL050820 Cai, W., S. Borlace, M. Lengaigne, P. van Rensch, M. Collins, G. Vecchi, et al. (2014). Increasing frequency of extreme El Niño events due to greenhouse warming, Nature Climate Change, 4, 111–116. doi: 10.1038/NCLIMATE2100 Capotondi, A., et  al. (2015). Understanding ENSO diversity. Bull. Amer. Meteor. Soc., 96, 921–938. https://doi.org/10.1175/ BAMS‐D‐13‐00117.1 Cayan, D. R., K. T. Redmond, & L. G. Riddle (1999). ENSO and hydrologic extremes in the western United States. J. Climate, 12, 2881–2893. Cayan, D.R., T. Das, D. W. Pierce, T. P. Barnett, M. Tyree, & A.  Gershunov (2010). Future dryness in the southwest US and the hydrology of the early 21st century drought. PNAS, 107, 21271–21276. doi:10.1073/pnas.0912391107 Chang, P., J. Link, & H. Li (1997). A decadal climate variation in the tropical Atlantic Ocean from thermodynamic air‐sea interactions. Nature, 385, 516–518. Charney, J. G. ,& J. Shukla (1981). Predictability of monsoons. In Monsoon dynamics (pp. 99–109). Cambridge Univ. Press. Chiang, J.C.H., & A. Sobel (2002). Tropical tropospheric temperature variations caused by ENSO and their influence on the remote tropical climate. J. Clim., 15, 2616–2631. Coelho, C.A.S., & L. Goddard (2009). El Niño–induced tropical droughts in climate change projections. J. Climate. doi: 10.1175/2009JCLI3185.1 Corringham, T. W., & D. R. Cayan (2019a). The effect of El Niño on flood damages in the western United States. Wea. Climate Soc., 11, 489–504. Corringham, T. W., F. M. Ralph, A. Gershunov, D. R. Cayan & C. A. Talbot (2019b). Atmospheric rivers drive flood damages in the western United States. Science Advances, 5, 12, eaax4631, DOI: 10.1126/sciadv.aax4631 Curtis, S., & S. Hastenrath (1995). Forcing of anomalous sea‐ surface temperature evolution in the tropical Atlantic during Pacific warm events. J. Geophys. Res., 100, 15835–15847. Dai, A., & T. Zhao (2017. Uncertainties in historical changes and future projections of drought. Part I: Estimates of historical drought changes. Climatic Change, 144, 519–522. https:// doi.org/10.1007/s10584‐016‐1705‐2

Espinoza, V., Waliser, D. E., Guan, B., Lavers, D. & Ralph, F. M. (2018). Global analysis of climate change projection effects on atmospheric rivers. Geophys. Res. Lett., 45, 4299–4308. Favre, A., & A. Gershunov (2006). Extra‐tropical cyclonic/anticyclonic activity in north‐eastern Pacific and air temperature extremes in western North America. Climate Dynamics, 26, 617–629. Favre, A., B. Hewitson, M. Tadross, C. Lennard, & R. Cerezo‐ Mota (2012). Relationships between cut‐off lows and the semiannual and southern oscillations. Clim Dyn, 41, 2331. doi: 10.1007/s00382‐011‐1030‐4 Favre, A., B. Hewitson, C. Lennard, R. Cerezo‐Mota, & M. Tadross (2013). Cut‐off lows in the South Africa region and their contribution to precipitation. Clim Dyn, 41, 2331. https://doi.org/10.1007/s00382‐012‐1579‐6 Ferreira, R. N., T. M. Rickenbach, D. L. Herdies, & L.M.V. Carvalho (2003). Variability of South American convective cloud systems and tropospheric circulation during January– March 1998 and 1999. Mon. Weather Rev., 131, 961–973. doi:10.1175/1520‐0493(2003)131 2.0.CO;2 Francis, J. A., & Vavrus, S. J. (2012). Evidence linking Arctic amplification to extreme weather in mid‐latitudes. Geophys. Res. Lett., 39, 06801. doi: 10.1029/2012GL051000 Gergis, J., & L. Ashcroft (2013). Rainfall variations in south‐ eastern Australia. Part 2: A comparison of documentary, early instrumental and palaeoclimate records, 1788–2008. Int. J. Climatol, 33, 2973–2987. https://doi.org/10.1002/joc.3639 Gershunov, A. (1998). ENSO influence on intraseasonal extreme rainfall and temperature frequencies in the contiguous US: Implications for long‐range predictability. Journal of Climate, 11, 3192–3203. Gershunov, A., & T. Barnett (1998a). ENSO influence on intraseasonal extreme rainfall and temperature frequencies in the contiguous US: Observations and model results. Journal of Climate, 11, 1575–1586. Gershunov, A., & T. Barnett (1998b). Inter‐decadal modulation of ENSO teleconnections. Bulletin of the American Meteo­ rological Society, 79, 2715–2725. Gershunov, A., N. Schneider, & T. Barnett (2001). Low frequency modulation of the ENSO‐Indian monsoon rainfall relationship: Signal or noise. J. Climate, 14, 2486–2492. Gershunov, A., & D. Cayan (2003). Heavy daily precipitation frequency over the contiguous United States: Sources of climatic variability and seasonal predictability. J. Climate, 16, 2752–2765. Gershunov A., T. M. Shulgina, F. M. Ralph, D. Lavers, & J.  J.  Rutz (2017). Assessing climate‐scale variability of atmospheric rivers affecting western North America. Geophys. Res. Lett., 44, doi: 10.1002/2017GL074175 Gershunov, A., T. M. Shulgina, R.E.S. Clemesha, K. Guirguis, D. W. Pierce, M. D. Dettinger, et  al. (2019). Precipitation regime change in western North America: The role of atmospheric rivers. Nature Scientific Reports, 9, 9944. doi: 10.1038/ s41598‐019‐46169‐w Giannini, A., J.C.H. Chiang, M. A. Cane, Y. Kushnir, & R. Seager (2001). The ENSO teleconnection to the tropical Atlantic Ocean: Contributions of the remote and local SSTs to rainfall variability in the tropical Americas. J. Climate, 14, 4530–4544.

374  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Gill, A. E. (1980). Some simple solutions for the heat‐induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106, 447–462. Goddard, L., & M. Dilley. (2005). El Niño: Catastrophe or opportunity. J. Climate, 18, 651–665. Goddard, L., & N. E. Graham (1999). The importance of the Indian Ocean for simulating precipitation anomalies over  eastern and southern Africa. J. Geophys. Res., 104, 19099–19116. Groisman, P. Ya., R. W. Knight, D. R. Easterling, T. R. Karl, G. C. Hegerl, & V. N. Razuvaev (2005). Trends in precipitation intensity in the climate record. J. Climate, 18, 1326–1350. Groisman, P. Ya., & R. W. Knight (2008). Prolonged dry episodes over the conterminous United States: New tendencies emerging during the last 40 years. J. Climate, 21, 1850–1862. doi:https://doi.org/10.1175/2007JCLI2013.1 Guirguis, K., A. Gershunov, & D. R. Cayan (2015). Interannual variability in associations between seasonal climate, weather and extremes: Wintertime temperature over the Southwestern United States. Env. Res. Lett., 10, 124023. doi:10.1088/ 1748–9326/10/12/124023 Guirguis, K., A. Gershunov, T. M. Shulgina, R.E.S. Clemesha, & F. M. Ralph (2018). Atmospheric rivers impacting northern California and their modulation by a variable climate. Clim. Dyn. https://doi.org/10.1007/s00382‐018‐4532‐5 Guzman Morales, J., A. Gershunov, J. Theiss, H. Li, & D. R. Cayan (2016.) Santa Ana winds of southern California: Their climatology, extremes, and behavior spanning six and a half decades. Geophys. Res. Lett., 43. doi:10.1002/2016GL067887 Halpern, M., & G. Bell (1998). Climate assessments for 1997. Bull. Am. Met. Soc., 79, s1–s50. Hendon, H. H., M. C. Wheeler, & C. Zhang (2007). Seasonal dependence of the MJO‐ENSO relationship. Journal of Climate, 20, 531–543. https://doi.org/10.1175/JCLI4003.1 Hendon, H. H., C. Zhang, & J. D. Glick (1999). Interannual variation of the Madden–Julian Oscillation during austral summer. Journal of Climate, 12, 2538–2550. Kiem, A. S., S. W. Franks, & G. Kuczera (2003). Multi‐decadal variability of flood risk. Geophysical Research Letters, 30(2). https://doi.org/10.1029/2002GL015992 King, A. D., L. V. Alexander, & M. G. Donat (2013). Asymmetry in the response of eastern Australia extreme rainfall to low‐ frequency Pacific variability. Geophysical Research Letters, 40(10), 2271–2277. https://doi.org/10.1002/grl.50427 Klein, S. A., B. J. Soden, & N‐C. Lau (1999). Remote sea surface temperature variations during ENSO: Evidence for a tropical atmospheric bridge. J. Climate, 12, 917–932. Konrad, C. P., & M. D. Dettinger (2017). Flood runoff in relation to water vapor transport by atmospheric rivers over the western United States 1945‐2015. Geophys. Res. Lett., 44, 11456–11462. Kosaka, Y., & S.-P. Xie (2016). The tropical Pacific as a key pacemaker of the variable rates of global warming. Nature Geoscience, 9, 669–673. Krishnamurthy, V., & B. N. Goswami (2000). Indian monsoon– ENSO relationship on interdecadal timescale. J. Climate, 13, 579–594. Krishna Kumar, K., B. Rajagopolan, & M. A. Cane (1999). On the weakening relationship between the Indian Monsoon and ENSO. Science, 284, 2156–2159.

Kumar, K. K., B. Rajagopalan, M. Hoerling, G. Bates, & M. Cane (2006). Unraveling the mystery of Indian monsoon failure during El Niño. Science, 314, 115–119. Lau, W.K.M. (2005). El Niño southern oscillation connection. In Intraseasonal variability in the atmosphere‐ocean climate system (pp. 271–305). Berlin, Heidelberg: Springer. Lee, D., P. J. Ward, & P. Block (2015). A preliminary evaluation of season‐ahead flood risks globally. 2015 AGU Fall Meeting, NH51B‐1870. Lenssen, N., L. Goddard, & S. J. Mason (2019). Observed Global ENSO teleconnections as a baseline for seasonal forecast verification. J. Weather and Forecasting, submitted. Lindzen, R. S., & S. Nigam (1987). On the role of sea surface temperature gradients in forcing low‐level winds and convergence in the Tropics. J. Atmos. Sci, 44, 2418–2436. Livezey, R. E., & M. M. Timofeyeva (2008). The first decade of long‐lead U.S. seasonal forecasts. Bull. Amer. Meteor. Soc., 89, 843–854. https://doi.org/10.1175/2008BAMS2488.1 Lyon, B. (2004). The strength of El Niño and the spatial extent of tropical drought. Geophys. Res. Lett., 3, L21204. doi: 10.1029/2004GL020901 Lyon, B., & R. M. Dole (1995). A diagnostic comparison of the 1980 and 1988 U.S. summer heat wave‐droughts. J. Climate, 8, 1658–1675. https://doi.org/10.1175/1520‐0442(1995)0082.0.CO;2 Mann, M. E., et al. (2017). Influence of anthropogenic climate change on planetary wave resonance and extreme weather events. Sci. Rep., 7, 45242. doi: 10.1038/srep45242 Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, & R. C. Francis (1997). A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78, 1069–1079. Mason, S. J., & L. Goddard (2001). Probabilistic precipitation anomalies associated with ENSO. Bull. Amer. Meteor. Soc., 82, 619–638. Meehl, G. A, & H. Teng (2007). Multi‐model changes in El Niño teleconnections over North America in a future warmer world. Climate Dynamics, 29, 779–790. https://doi.org/10.1007/ s00382‐007‐0268‐3 Mehta, V. M., & K‐M. Lau (1997). Influence of solar irradiance on the Indian monsoon–ENSO relationship at decadal‐multidecadal time scales. Geophys. Res. Lett., 24, 159–162. Michelangeli, P.‐A., R. Vautard, & B. Legras (1995). Weather regimes: Recurrence and quasi‐stationarity, J. Atmos. Sci., 52, 1237–1256. https://doi.org/10.1175/1520‐0469(1995)052< 1237:WRRAQS>2.0.CO;2 Muñoz, A. G., L. Goddard, S. J. Mason, & A. W. Robertson (2016). Cross‐timescale interactions and rainfall extreme events in South East South America for the austral summer. Part II: Predictive skill. J. Climate. doi: http://dx.doi. org/10.1175/JCLI‐D‐15‐0699.1 Muñoz, A., L. Goddard, A. Robertson, Y. Kushnir, & W. Baethgen (2015). Cross–time scale interactions and rainfall extreme events in southeast South America for the austral summer. Part I: Potential predictors. J. Climate, 28, 7894– 7913. doi: http://10.1175/JCLI‐D‐14‐00693.1 National Academies of Sciences, Engineering, and Medicine (NASEM). 2016. Attribution of extreme weather events in the context of climate change. Washington, DC: National Academies Press. doi: 10.17226/21852

Impact of El Niño on Weather and Climate Extremes  375 Ndarana, T., & D.W. Waugh (2010). The link between cut‐off lows and Rossby wave breaking in the Southern Hemisphere. Q. J. R. Meteorol. Soc., 136, 869–885. doi:10.1002/qj.627 Neelin, J. D., C. Chou, & H. Su (2003). Tropical drought regions in global warming and El Niño teleconnections, Geophysical Research Letters, 30(24), 2275. doi: 10.1029/2003GL018625 Newman, M., M. A. Alexander, T. R. Ault, K. M.,Cobb C . Deser, E. Di Lorenzo, et  al. (2016). The Pacific Decadal Oscillation, revisited. Journal of Climate, 29, 4399–4427. Oakley, N., & K. Redmond (2014). A climatology of 500‐hPa closed lows in the northeastern Pacific Ocean 1948–2011. Journal of Applied Meteorology and Climatology, 53, 1578–1592. Pant, G. B., K. Rupa Kumar, B. Parthasarathy, & H. P. Borgaonkar (1988). Long‐term variability of the Indian summer monsoon and related parameters. Adv. Atmos. Sci., 5, 469–481. Payne, A. E., & G. Magnusdottir (2015). An evaluation of atmospheric rivers over the North Pacific in CMIP5 and their response to warming under RCP 8.5. Journal of Geophysical Research, 120, 11173–11190. Pan, X., M. Chin, C. M. Ichoku, & R. D. Field (2018). Connecting Indonesian fires and drought with the type of El Niño and phase of the Indian Ocean dipole during 1979– 2016. Journal of Geophysical Research: Atmospheres, 123, 7974–7988. https://doi.org/10.1029/2018JD028402 Pant, G. B., K. Rupa Kumar, B. Parthasarathy, & H. P. Borgaonkar (1988). Long‐term variability of the Indian summer monsoon and related parameters. Adv. Atmos. Sci., 5, 469–481. Polade, S. D., A. Gershunov, D. R. Cayan, M. D. Dettinger, & D. W. Pierce (2013). Natural climate variability and teleconnections to precipitation over the Pacific‐North American region in CMIP3 and CMIP5 models. Geophys. Res. Lett., 40. doi:10.1002/grl.50491 Polade, S. D., D. W. Pierce, D. R. Cayan, A. Gershunov, & M. D. Dettinger (2014). The key role of dry days in changing regional climate and precipitation regimes. Nature Scientific Reports, 4, 4364. doi:10.1038/srep04364 Polade, S. D., A. Gershunov, D. R. Cayan, M. D. Dettinger, & D. W. Pierce (2017). Precipitation in a warming world: Assessing projected hydro‐climate of California and other Mediterranean climate regions. Nature Scientific Reports, 7, 10783. doi:10.1038/s41598‐017‐11285‐y Power, S., M. Haylock, R. Colman, & X. Wang (2006). The predictability of interdecadal changes in ENSO Activity and ENSO teleconnections. J. Climate 19(19), 4755–4771. https:// doi.org/10.1175 /JCLI3868.1 Power, S., F. Delage, C. Chung, G. Kociuba, & K. Keay (2017). Robust twenty‐first‐century projections of El Niño and related precipitation variability, Nature, 502, 541–545. Saji, N. H., B. N. Goswami, P. N. Vinayachandran, & T. Yamagata (1999). A dipole mode in the tropical Indian Ocean. Nature, 401(6751), 360–363. https://doi.org/10.1038/43854 Salio, P., M. Nicolini, & A. C. Saulo (2002). Chaco low‐level jet events characterization during the austral summer season. J. Geophys. Res., 107, 4816,.doi:10.1029/2001JD001315 Saravanan, R., & P. Chang (2000). Interaction between tropical Atlantic variability and El Niño–Southern Oscillation. J. Climate, 13, 2177–2194.

Schneider, N., & B. D. Cornuelle (2005). The forcing of the Pacific decadal oscillation. J. Climate, 18, 4355–4373. Seager, R., T. J. Osborn, Y. Kushnir, I. R. Simpson, J. Nakamura, & H. Liu (2019). Climate variability and change of Mediterranean‐type climates. J. Climate, 32, 2887–2915. Seluchi, M. E., R. Garreaud, F. A. Norte, & A. C. Saulo (2006). Influence of the subtropical Andes on baroclinic disturbances: A cold front case study. Mon. Wea. Rev., 134, 3317–3335. doi:10.1175/MWR3247.1 Seneviratne, S. I., J. Rogelj, R. Séférian, R. Wartenburger, M. R. Allen, M. Cain, et  al. (2018). The many possible climates from the Paris Agreement’s aim of 1.5°C warming. Nature, 558, 41–49. https://doi.org/10.1038/ s41586‐018‐0181‐4 Sobel, A. H., I. M. Held, & C. S. Bretherton (2002). The ENSO signal in tropical tropospheric temperature, J. Clim., 15, 2702–2706. Supari, F. Tangang, E. Salimun, E. Aldrian, A. Sopaheluwakan, & L. Juneng (2018). ENSO modulation of seasonal rainfall and extremes in Indonesia, Clim. Dyn., 51, 2559–2580. https://doi.org/10.1007/s00382‐017‐4028‐8 Trenberth, K. E., J. M. Caron, D. P. Stepaniak, & S. Worley (2002). The evolution of ENSO and global atmospheric surface temperatures J. Geophys. Res., 107, D8. 10.1029/ 2000JD000298 Velasco, I., & J. M. Fritsch (1987). Mesoscale convective complexes in the Americas. J. Geophys. Res., 92, 9591. doi:10.1029/JD092iD08p09591 Walker, G. T. (1924). Correlation in seasonal variations of weather. IV: A further study of world weather. Mem. Indian Meteor. Dept., 24, 275–332. Walker, G. T., & E. W. Bliss (1932). World weather. V. Mem. Roy. Meteor. Soc., 4, 53–84. Ward, P. J., B. Jongman, F. S. Weiland, A. Bouwman, R. van Beek, M.F.P. Bierkens, et  al. (2013). Assessing flood risk at the global scale: Model setup, results, and sensitivity. Environ. Res. Lett., 8, 044019. Webster, P. J., A. M. Moore, J. P. Loschnigg, & R. R. Leben (1999). Coupled ocean‐atmosphere dynamics in the Indian Ocean during 1997‐98. Nature, 401, 356–360. doi: 10.1038/43848 Wittenberg, A. T. (2009). Are historical records sufficient to constrain ENSO simulations? Geophys. Res. Lett., 36, L12702. doi:10.1029/2009GL038710 Westerling, A. L., A. Gershunov, D. R. Cayan, & T. P. Barnett (2002). Long lead statistical forecasts of area burned in western U.S. wildfires by ecosystem province. Int. Journal of Wildland Fire, 11, 257–266. Westerling, A. L., A. Gershunov, T. Brown, D. Cayan, & M. Dettinger (2003). Climate and wildfire in the western United States. Bulletin of the American Meteorological Society, 84, 595–604. Yu, L., & M. M. Rienecker (1999). Mechanisms for the Indian Ocean warming during the 1997–98 El Niño. Geophys. Res. Lett., 26, 735–738. Yulaeva, E., & J. M. Wallace (1994). The signature of ENSO in global temperature and precipitation fields derived from the microwave sounding unit. J. Climate, 7, 1719–1736.

17 ENSO and Tropical Cyclones I‐I Lin1, Suzana J. Camargo2, Christina M. Patricola3,4, Julien Boucharel5, Savin Chand6, Phil Klotzbach7, Johnny C. L. Chan8, Bin Wang9, Ping Chang10, Tim Li9, and Fei‐Fei Jin9

ABSTRACT One of ENSO’s most important influences is its worldwide modulation of tropical cyclone (TC) activity. TCs impact millions of people annually and can devastate life and property. Because TC attributes (e.g. genesis, track and landfall locations, intensity) are largely controlled by large‐scale environmental conditions, TC activity can be substantially altered by ENSO, via ENSO’s strong modulation on both the atmosphere and ocean. Atmospheric modulations include changes in vertical wind shear, humidity, low‐level vorticity, and the strength and position of subtropical highs. The ocean influences TCs via changes in sea‐surface temperature and upper‐ocean heat content and structure. This chapter will focus on ENSO’s influences on TC basins around the globe, including local effects and remote influences via teleconnections. These basins include the western North Pacific, central eastern North Pacific, North Atlantic, North Indian Ocean, and Southern Hemisphere (South Pacific and South Indian Ocean). We will also discuss additional factors that, together with ENSO, are important for TC prediction and projection, including other modes of natural climate variability and anthropogenic climate change.

17.1. INTRODUCTION Tropical cyclones (TCs, also known as typhoons and hurricanes) threaten 1–2 billion people worldwide each Department of Atmospheric Sciences, National Taiwan University, Taipei, Taiwan 2  Lamont-Doherty Earth Observatory, Columbia University, Palisades, NY, USA 3  Climate and Ecosystem Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA 4  Iowa State University, Ames, IA, USA 5  LEGOS-CNRS, University of Toulouse, Toulouse, France 6  Centre for Informatics and Applied Optimization, Federation University Australia, Ballarat, VIC, Australia 7  Department of Atmospheric Science, Colorado State University, Fort Collins, CO, USA 8  Guy Carpenter Asia-Pacific Climate Impact Centre, School of Energy and Environment, City University of Hong Kong, Hong Kong, China 9  Department of Atmospheric Sciences, University of Hawai’i at Mānoa, Honolulu, HI, USA 1 

year (Peduzzi et  al., 2012). On average, approximately 80–90 named TCs1 form annually (Frank & Young, 2007), of which about one‐third make landfall2, making TCs one of the most common weather extremes (Figure 17.1). Needless to say, it is highly important to understand TCs and their relationship to climate variability and change. At a conceptual level, a TC may be understood as an atmospheric Carnot heat engine that transfers energy from the warmer planetary boundary layer to the cooler free 1  In this chapter, the term TC (or storm) is used as a general term that includes TC (≥ category 1), tropical storm (TS), and tropical depression (TD). The Saffir–Simpson tropical cyclone scale based on the 1-min maximum sustained winds: category 1: 64–82 kts, category 2: 83–95 kts, category 3: 96–112 kts, category 4: 113–136 kts, and category 5: ≥ 137 kts. Related classifications for TD: ≤ 33 kts, TS: 34–63 kts. 2  Based on the IBTrACSv03r10 TC data set. 10  Department of Oceanography and Department of Atmospheric Sciences, Texas A&M University, College Station, TX, USA

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 377

378  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a) Best track data from NHC and JTWC, 1980~2018

TD

TS

C1

C2

C3

C4

C5

Wind Speed(knots) C5

≥137

C4

113–136

C3

96–112

C2 C1

83–95 64–82

TS

34–63

TD

≤ 33

(b)

Figure 17.1  (a) Historic global TC tracks (based on the best track data of the US Joint Typhoon Warning Center [JTWC] and the National Hurricane Center [NHC] from 1980 to 2018), produced following the original style from NASA (https://earthobservatory.nasa.gov/images/7079/historic‐tropical‐cyclone‐tracks), with locations of major TC basins overlaid.(b) Tropical cyclone genesis locations in the global tropics during 1990 and 2017 (dots). Data source: IBTrACSv03r10 data set. Distribution of the warmest tropical ocean areas where sea surface temperature exceeds 26.5°C (red shade) during May–November in the Northern Hemisphere and November–May in the Southern Hemisphere. Mean rainfall distribution averaged over May–November for the Northern Hemisphere and November–May for the Southern Hemisphere (pink shade: ≥5 mm/day). (Courtesy of Dan Fu)

atmosphere aloft and, in due course, converts thermal energy to mechanical energy (Emanuel, 2005). The primary energy source of TCs is air‐sea enthalpy (latent and sensible heat) fluxes from the warm tropical ocean surface (Emanuel, 1995, 1999; Bister & Emanuel, 1998). In general, TC genesis and intensification requires a number of favorable environmental atmospheric and ocean conditions, including warm sea‐­ surface temperature (SST, typically greater than 26°C in the present climate3, Dare & McBride, 2011), a sufficiently deep and warm subsurface ocean layer (i.e. high ocean heat content, OHC4), low atmospheric stability, high humidity in the lower‐

to‐mid troposphere, weak vertical wind shear (VWS), and favorable background vorticity (Gray, 1979; Ritchie & Holland, 1999; Shay et  al., 2000; DeMaria et  al., 2005; Emanuel, 2005; Lin et al., 2008; Goni et al., 2009; Price, 2009; Chih & Wu, 2020). Because of these requirements, TCs form over tropical/subtropical oceans. Once formed, the movement of a TC, i.e. its track, is determined by its own “beta drift” and by the “steering flow,” i.e. steering and advection by the large‐ scale atmospheric environmental flow (Chan, 2005). As the El Niño‐Southern Oscillation (ENSO)5 can p­ rofoundly change both the atmospheric and oceanic

3  Besides the above-mentioned general conditions, there also exists much regional variability (Defforge & Merlis, 2017). 4  Defined as the depth-integrated heat content from ocean surface down to the 26°C isotherm (Shay et al., 2000).

5  For this book, ENSO is used as a general term including both El Niño and La Niña events. El Niño is for the warm ENSO phase and La Niña is for the cold ENSO phase.

ENSO and Tropical Cyclones  379

environments, it can effectively modulate global TC activity (Chan, 1985; Camargo et al., 2010; Wang et al., 2010; Aiyyer & Thorncroft, 2011; Camargo & Hsiang, 2015; Zheng et al., 2015). ENSO can redistribute upper OHC in the tropical Pacific, producing anomalies in SST and upper ocean structures, and change the locations of tropical convection and the Intertropical Convergence Zone (ITCZ), generating anomalous Walker and Hadley circulations. Through atmospheric teleconnections, the influence of ENSO reaches far beyond the tropical Pacific, resulting in atmospheric circulation changes and hence influencing TCs globally. Figure  17.2 illustrates ENSO’s global influence on TCs; for example, during El Niño (developing) years, there is a significant drop in Atlantic TC activity. Over both the eastern and western North Pacific, TCs tend to form closer to the international date line (R. Bell et al., 2014). The global total number of tropical storm days, which gives an integrated measure of TC frequency and life span each year, is strongly influenced by ENSO (and the Interdecadal Pacific Oscillation, IPO) (Wang et al., 2010). There have been good reviews on the ENSO‐TC relationship (e.g. Landsea, 2000, with focus on Atlantic; Chu, 2004). With the development of the field, this chapter provides an update, with additional topics including ENSO diversity (chapter 4), new teleconnections pathways, and other climate modes. We include some discussions on other coexisting modes of climate variability (in subsections for each basin) and anthropogenic climate change (in section  17.7), because despite its strong influence, ENSO alone is often inadequate to fully describe variability and change in TC activity. This chapter is thus written from TC’s (instead of ENSO’s) perspective. Like ENSO (interannual scale), other climate modes, e.g. subseasonal‐scale Madden‐Julian Oscillation (MJO; Madden & Julian, 1972; 1994), multidecadal‐scale Pacific Decadal Oscillation (PDO; Mantua et al., 1997), or centennial‐ scale climate change (Knutson et  al., 2010), can influence TCs independently or work with ENSO to influence TCs. Because El Niño events generally peak in boreal winter and decay in the following spring/summer, for consistent terminology, El Niño year is defined here as the “El Niño developing year.” The following year is termed the “El Niño decaying year” (usually also a normal or La Niña developing year). In addition, for the three El Niño “flavors,” the classical type (e.g. 1982– 1983 and 1997–1998) will be called EP (Eastern Pacific) El Niño or simply El Niño; the Modoki type (e.g. 2009– 2010) will be called CP (Central Pacific) El Niño (Ashok et al., 2007; Kao & Yu, 2009; Kug et al., 2009), and the EP/CP type (e.g. 2015–2016) will be called mixed type (Paek et  al., 2017; Santoso et  al., 2017; Williams & Patricola, 2018).

17.2. WESTERN NORTH PACIFIC (WNP) TROPICAL CYCLONES 17.2.1. Introduction The WNP is the largest and the most active (for both frequency and intensity) TC basin in the world. Each year there are ~30 storms, with 16 reaching TC intensity (i.e. ≥ Category 1) and 9 major TCs (i.e. ≥ Category 3)6. These TCs severely impact the densely populated Asian countries (e.g. Philippines, Taiwan, Japan, China, Korea, and Vietnam). For example, in 2013, TC Haiyan devastated the Philippines (Lin et al., 2014). The main TC season in the WNP is from July to October, although it is possible for TCs to form throughout the year (Ritchie & Holland, 1999; Yoshida & Ishikawa, 2013; Chan, 2015). Once formed, TCs tend to intensify over the main developing region (MDR) at the center of the WNP (typically 125°–160°E, 10°–25°N), then move toward the land or higher‐latitude oceans (Figure 17.1a).

17.2.2. The Influence of ENSO on WNP TCs 17.2.2.1. TC Frequency TC frequency is typically normal or above normal during El Niño years and normal or below normal during El Niño decaying years (e.g. Chan, 1985; T. Li, 2012) (Figure  17.3a 7). The first investigation of the relationship between ENSO and WNP TC frequency was by Chan (1985). He found that annual TC frequency has an interannual oscillation with a period of about 3–3.5 years, similar to that of El Niño. A cross‐spectral analysis between the Southern Oscillation Index (SOI) and the TC frequency time series reveals a high coherence, with the TC time series lagging behind by about 1 year. Because the SOI generally reaches a minimum in the boreal winter, this result suggests that TC activity in the following summer is reduced. In terms of mechanisms, Chan (1985) first proposed that the decrease in TC frequency during El Niño decaying years was related to changes in the Walker circulation, enhancing subsidence in the WNP, hence making the atmosphere less favorable for TC genesis. This mechanism was simulated in a low‐resolution general circulation model by Wu and Lau (1992), and later expanded by Wang and Chan (2002) to include the anticyclonic circulation over the Philippines that typically forms after the occurrence of an El Niño event (Wang et al., 2000). 6  Based on the average of 1980–2018 best track data from the U.S. Joint Typhoon Warning Center (JTWC). 7  Except before 1976, it may be attributed to the lack of systematic satellite observations during the earlier period.

380  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a)

(b)

IBTrACS El Niño - clim

60N

60N

30N

30N

0

0

30S

30S 60S

60S 0

30E 60E 90E 120E 150E 180 150W120W 90W 60W 30W –3

–1.7 –2.2

(c)

–1 –1.4

0

0

–0.3 –0.6

–0.1

30N

30N

0

0

30S

30S

30E 60E 90E 120E 150E 180 150W120W 90W 60W 30W –1.7 –2.2

–1 –1.4

1.4 1

(d)

IBTrACS EP El Niño - clim

–3

0.6 0.3

60N

0

30E 60E 90E 120E 150E 180 150W120W 90W 60W 30W

0.1

60N

60S

IBTrACS La Niña - clim

0

60S

–0.3 –0.6

0

1.7

3

IBTrACS CP El Niño - clim

30E 60E 90E 120E 150E 180 150W120W 90W 60W 30W

0.1 –0.1

0

2.2

0.6 0.3

1.4 1

0

2.2 1.7

3

Figure 17.2  Tropical cyclone track density anomaly maps during May–November in the Northern Hemisphere and October–May in the Southern Hemisphere. Data source: IBTrACSv03r10. Track density is based on storm transits/month/106 km2 or equivalent to a 5° radius). (a) El Niño years minus 1979–2010 climatology. (b) La Niña years minus 1979–2010 climatology. (c) EP El Niño years minus 1979–2010 climatology. (d) CP El Niño years minus 1979–2010 climatology.(a) and (b) are reproduced following the original figure and style of Bell et al. (2014) (© American Meteorological Society. Used with permission). (c) and (d) are added to illustrate the different impact from EP and CP El Niño events. The definition of different ENSO years is from Patricola et al. (2016). El Niño years are 1982, 1986, 1987, 1991, 1994, 1997, 2001, 2002, 2004. La Niña years are 1988, 1998, 2010. EP El Niño years are 1982, 1987, 1997. CP El Niño years are 1986, 1991, 1994, 2001, 2002, 2004. (Courtesy of Dan Fu)

More recently, Du et al. (2011) attributed the reduction of TCs during El Niño decaying years to a teleconnection from the Indian Ocean. Because El Niño can induce warming in the tropical Indian Ocean, a warm tropospheric Kelvin wave can propagate back into the WNP to contribute to the development of the anomalous anticyclone mentioned above (Wang et al., 2000) and an increase in VWS to suppress TC formation. Yu et al. (2016) and Li et  al. (2017) suggested another possible route via the tropical North Atlantic Ocean. An El Niño event can first warm the tropical North Atlantic and then influence the tropical Indian Ocean, and subsequently affect WNP TCs in the El Niño decaying year. The increase in TC frequency during El Niño years (Figures 17.2 and 17.3a) is associated with the increase in TC activity and genesis in the southeast (SE) quadrant of WNP. Chan (1985) found that TC activity in this region is anticorrelated with the SOI, suggesting that more TCs

are expected to form during El Niño years (Figure 17.3bc), as corroborated in subsequent studies (Dong, 1988; Lander, 1994; Chen et al., 1998; Chan, 2000; Chia & Ropelewski, 2002; Wang & Chan, 2002; Li & Zhou, 2012; Wang et al., 2013). The enhancement over SE WNP was attributed to favorable atmospheric conditions, i.e. increased low‐level vorticity (Wang & Chan, 2002), reduced VWS, and an expanded monsoon trough (Chen et al., 1998; Chia & Ropelewski, 2002; Li & Zhou, 2012). Many of the above studies also showed that for La Niña years, the area with increased TC genesis is shifted toward the northwest (NW) WNP (Figure 17.3bc). 17.2.2.2. TC Genesis, Track, and Life Span The aforementioned SE‐NW shift in the TC genesis locations (Chan, 2000; Wang & Chan, 2002; Wang et al., 2013) also contributes to notable track changes. In El Niño years, TCs form in the southeast, moving toward

ENSO and Tropical Cyclones  381

the northwest and then recurving close to the direction of Japan (Chan, 2000; Wu et al., 2004). This generally leads to fewer TCs tracking toward the southern countries, i.e. Philippines and the South China Sea. On the other hand, with TCs forming toward the west during La Niña years, more TCs track westward and enter the South China Sea, increasing the chance of landfall in China (Wu et  al., 2004). Camargo et al. (2007c, 2007d) performed a cluster analysis on WNP TC tracks and found that they can be grouped into seven types, three of which are strongly modulated by ENSO (Figure  17.4). During La Niña years, short tracks (cluster A) that form at higher latitudes tend to have a higher occurrence, with increased landfall over the Asian continent (53%). In contrast, during El Niño years, TCs form closer to the equator and the international date line (clusters E and G), with long recurving tracks and a lower landfall rate (42% and 13%, respectively). Therefore, during El Niño years, TCs can spend a longer time over ocean before making landfall. During La Niña years, TCs are formed toward land and have a shorter life span. The average TC life span during El Niño (La Niña) is 7 (4) days (Chia & Ropelewski, 2002; Wang & Chan, 2002).

(a) 40

Annual number of tropical storms and typhoons

30 20 10 1960

1970

(b)

1980

1990

2000

2010

Main season

16°N

8°N 120°E

130°E

140°E

(c)

150°E

160°E

Late season

16°N

8°N 120°E

130°E

(d)

140°E

150°E

160°E

1997 ASON UOHC anomaly

30°N 15°N 0 120°E –60

150°E –40

180 –20

150°W 0

20

120°W 40

kJ/cm2

60

Figure 17.3 (a) Annual number of TCs in the WNP (1960– 2016). Red circles: El Niño; blue squares: La Niña; green triangles: CP El Niño; empty circles: neutral years. Data are from the Joint Typhoon Warning Center. Definitions of El Niño and La Niña years are as follows. A year is defined as an El Niño (La Niña) year if the mean Jul–Nov adjusted Niño‐3.4 index is ≥0.5 (≤–0.5). The Niño‐3.4 index is adjusted using the new centered 30‐year base periods. http://origin.cpc.ncep.noaa.gov/products/ analysis_monitoring/ensostuff/ONI_change.shtml. A year is defined as an CP El Niño year if the mean Jul–Nov El Niño Modoki Index (EMI) is ≥0.5.http://www.jamstec.go.jp/frcgc/ research/d1/iod/enmodoki_home_s.html.en. EMI = SSTA(Box A) – 0.5*SSTA(Box B) – 0.5*SSTA(Box C); Box A: (165°E–140°W, 10°S–10°N), Box B: (110°W–70°W, 15°S–5°N), Box C: (125°E–145°E, 10°S–20°N), http://www.jamstec.go.jp/frcgc/ research/d1/iod/e/elnmodoki/about_elnm.html. Note that the mean Jul–Nov EMI index in 1966 and 1967 is ≥0.5, but these years are not considered as the CP El Niño because the high EMI index is mainly due to the negative SSTA in box B and C but not the warming in box A.(b)(c) Annual average of TC genesis locations in the WNP (1960–2016). Red circles: El Niño; blue squares: La Niña; green triangle: CP El Niño; and empty circles: neutral years. Data are from the Joint Typhoon Warning Center. Definitions of the years are given in (a). (d) Observed change in ocean heat content in August–November 1997 El Niño year. Corresponding TC tracks is shown in green. Yellow box denotes the domain of (a,b,c). UOHC = upper ocean heat content. (Revised and reproduced following the original figure of Zheng et al., 2015)

17.2.2.3. TC Intensity: A Complex Situation Because of the SE shift of the genesis location and longer development time over ocean, TCs tend to be more intense during El Niño years (Wang & Chan, 2002; Camargo & Sobel, 2005; Li & Zhou, 2012). Favorable atmospheric conditions (e.g. greater vorticity and weaker VWS) also exist in the SE WNP during El Niño years to favor intensification (Li & Zhou, 2012). However, recent research reveals that the influence of ENSO in WNP TC intensity is not so simple. Although atmospheric factors maybe favorable, ocean conditions are more unfavorable. Zheng et  al. (2015) showed that during El Niño years, upper OHC decreases by as much as 20%–30% (as compared to normal) over the WNP (Figure 17.3d). This is a negative influence to offset the favorable atmospheric conditions and longer tracks, and the end result is moderately above normal TC intensity during El Niño years (Zheng et al., 2015). On the other hand, during La Niña or La Niña–like conditions, there is an increase in OHC to favor intensification. In fact, the most intense, record‐breaking category “6”8 type of TCs (e.g. Haiyan in 2013 and Megi in 2010) formed under La Niña–like or La Niña conditions (Lin et al., 2013a, 2014; Huang et al., 2017; Braun et al., 2018). Thus, it is not possible to use a simplistic view to generalize TC intensity 8  From Lin et al. (2014), category “6” is suggested to be ≥ 160kts in 1-min maximum sustained wind speed, i.e. 25 kts above the existing category 5 intensity. There is increasing interest on this subject, e.g. in the EOS news in 2017 (Hornyak, 2017).

382  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Cluster A – La Niña

Cluster A – El Niño 60N

60N 52 TCs

89 TCs

40N

40N

20N

20N

0

120E

140E

160E

180

160W

Cluster E – El Niño

60N

0

40N

20N

20N

140E

160E

180

160W

0

120E

Cluster G – El Niño

180

160W

140E

160E

180

160W

180

160W

Cluster G – La Niña 60N

60N 53 TCs

5 TCs

40N

40N

20N

20N

0

160E

20 TCs

40N

120E

140E

Cluster E – La Niña

60N

53 TCs

0

120E

120E

140E

160E

180

160W

0

120E

140E

160E

Figure 17.4  Tropical cyclone tracks in El Niño years (left panels, in red) and La Niña years (right panels, in blue). (From Camargo et al., 2007d, © American Meteorological Society. Used with permission).

change under ENSO. Ultimately, the fate of an individual TC will need to be determined holistically by the possible offset from its instantaneous atmospheric and ocean conditions. Besides intensity, accumulated cyclone energy (ACE; Bell et al., 2000) is another TC attribute of interest. The ACE of a TC is defined as the square of TC’s intensity integrated over its lifetime (in 6 hour intervals). Annual ACE is the integration of ACE over all TC cases in a year. Here for simplicity, ACE is referred to as annual ACE. Camargo and Sobel (2005) found that ACE in the WNP is strongly correlated with ENSO, with high values during El Niño years, due to a higher occurrence of intense and long‐lived TCs in the region. Another impor-

tant parameter for disaster mitigation is rapid intensification (defined as exceeding 30 kts TC intensification within 24 hours, Kaplan & DeMaria, 2003). Wang and Zhou (2008) reported that there is a higher rapid intensification rate during El Niño (53%) than during La Niña years (37%). 17.2.2.4. TC Size Based on a small sample (1999–2002), Chan and Yip (2003) suggested that the average TC size (defined by the azimuthal average of the radius of 15 m s‐1 winds), is likely larger during El Niño years. They suggest that more TCs tend to form in the SE WNP, thus having a longer track (Wang & Chan, 2002) and a greater chance to grow

ENSO and Tropical Cyclones  383

in size. Using longer datasets, Yuan et  al. (2007) and Chan and Chan (2012) confirmed this finding. 17.2.2.5. ENSO Diversity and WNP TCs In addition to increased TC frequency during EP El Niño years, Chen and Tam (2010) reported that TC frequency during CP El Niño years is also above normal (Figures 17.2 and 17.3a). Kim et al. (2011) reported that during CP years, TC activity is more westward and extended more toward the NW, as compared to the EP years (Kim et al., 2011; Ha et al., 2012; Chung & Li, 2015; C. Hu et al., 2018; Chen & Lian, 2018) (Figure 17.3bc). Due to the westward shift of the heating position from CP El Niño, there are anomalous westerly winds and an extension of the monsoon trough toward the NW to favor TC genesis. This pattern change also increases the landfall probability in East Asian countries (e.g. Korea and Japan; Kim et al., 2011). From the ocean side, similar to EP El Niño, Zheng et  al. (2015) also reported a considerable drop of 10%–20% OHC over the WNP during CP years. Patricola et al. (2018) found an increase in observed WNP ACE that is comparable in magnitude for EP and CP El Niño events, despite considerably weaker SST anomalies during CP El Niño. Using climate model simulations, they suggest that for a similar amount of SST warming, CP El Niño is more effective at enhancing WNP TC activity. This closer relationship between TC and CP El Niño is also observed in Zhao and Wang (2018). Besides CP studies, Y.‐K. Wu et al. (2018) compared the 1997–1998 (EP type) and the 2015–20169 (mixed type) events and found that although TC frequency is expected to decrease in EP El Niño decaying years (see earlier discussions), there is little decrease in the decaying year of the 2015–2016 El Niño. They suggest an additional factor is present during the 2015–2016 El Niño, i.e. the eastern North Pacific SST anomaly associated with the Pacific Meridional Mode (PMM). This SST anomaly may contribute to a low‐level cyclonic circulation anomaly over the WNP that offsets the TC‐suppressive anticyclone in the WNP, thus returning TC activity to normal in the 2015–2016 decaying year. 17.2.3. The Influence of Other Modes of Climate Variability on WNP TCs PDO is an important non‐ENSO influence on WNP TCs. During the PDO cold phase in 1998–2014, the Pacific was characterized by a multidecadal La Niña–like

9  The 2015–2016 El Niño event exhibited interesting characteristics in that although the magnitude of SST warming was comparable to the extreme 1982–1983 and 1997–1998 EP events, it also exhibited CP characteristics (e.g. PMM related feature; Paek et al., 2017; Santoso et al., 2017). The associated eastward shift in atmospheric deep convection was also substantially less (L’Heureux et al., 2017; Williams & Patricola, 2018).

condition (Kosaka & Xie, 2013; Wang et al., 2013; England et al., 2014) with more CP El Niño and La Niña events (C. Hu et al., 2018; Zhao & Wang, 2018). Due to unfavorable atmospheric conditions, TC frequency significantly reduced by ~40%, as compared to the early 1990s (Liu & Chan, 2013; Hsu et  al., 2014; Lin & Chan, 2015; F. Hu et al., 2018; Zhao & Wang, 2018). Superimposed over this multidecadal background, the 2010 La Niña produced only 14 storms, the lowest in historical records. PDO also works with ENSO to modify late‐season TC activity (Zhao & Wang, 2016), TC landfall along the China coast (Chan et al., 2012; Yang et al., 2018), and TC rapid intensification (Wang et al., 2015; Wang & Liu, 2016). At a shorter time scale, the subseasonal MJO can enhance (suppress) WNP TC genesis during the convectively active (inactive) local MJO phase (Gray, 1979; Liebmann et al., 1994). The MJO also works with ENSO to modify WNP TC activity; e.g. during El Niño years, the MJO has a stronger modulation on TC genesis (Kim et al., 2008; Li et al., 2012).

17.3. CENTRAL AND EASTERN NORTH PACIFIC (CEP) TCs 17.3.1. Introduction CEP is the second most active TC basin in the world, with an annual average of 19 named storms (9 TCs and 5 major TCs; Neumann, 1993; Figure 17.5a). Yet much remains unknown about the environmental factors regulating CEP TCs (Dong & Holland, 1994; Wang & Lee, 2009; Peduzzi et al., 2012; Caron et al., 2015). CEP TCs can have important economic consequences on the southwestern U.S., Mexico, the Hawaiian Islands, and military and commercial maritime routes between these areas (Court, 1980; Englehart & Douglas, 2001; Jauregui, 2003; Corbosiero et al., 2009; Ritchie et al., 2011; Raga et al., 2013; Wood & Ritchie, 2013; Martinez‐Sanchez & Cavazos, 2014). There is increasing interest in understanding CEP TCs, especially because four of the last five TC seasons (i.e. 2013–2017) have exceeded the 75th percentile of ACE over the period 1975–2017 (Figure 17.5b). In September 2015 (a mixed‐type El Niño season), an unprecedented occurrence of three distinct category 4 TCs surrounded Hawaii, although none caused significant damage to the region (Murakami et  al., 2017a). A few weeks later, Hurricane Patricia developed rapidly off the coast of Mexico into the most intense TC ever recorded, with peak intensity estimated at 185 knots (Foltz & Balaguru, 2016; Huang et  al., 2017; Rogers et al., 2017). A deeper than normal thermocline and low surface salinity both contributed to Patricia’s

(a)

(b)

TC trajectories in the Central Eastern Pacific, 1975-2017

60°N

ACE 300 250

10–4 Knots2

50°N Category5 Category4 Category3

40°N 30°N 20°N

ACE Central Pacific ACE East. Pacific ACE Central East. Pacific

200 150 100 50

10°N 180°W

140°W

100°W

0 1975

60°W

(c) Subsurface Temp. averaged over periods of El Niño meridional recharge (d)

28

20°N

1980

1985

1990

1995

2000

2005

2010

2015

Subsurface Temp. averaged over periods of El Niño meridional discharge

26 24 22

0°

20 18 16 14

20°S 180°W

140°W

(e)

100°W

60°W

180°W

140°W

100°W

60°W

Ocean/Atmosphere characteristics averaged in the CEP during summer post El Niño events

2

Post Central Pacific events Post Eastern Pacific events Post Neutral conditions Post Niña conditions

1

0

–1 ACEA

SSTA

T80A

VWSA - NCEP

VWSA - ERA40/interim

ERHA - NCEP

ERHA - ERA40/interim

Figure 17.5  (a) Trajectories of tropical cyclones in the central (left of the blue bold dashed line) and eastern (right of the blue bold dashed line) Pacific basins. Red lines for category 3, magenta lines for category 4, and black lines for category 5 hurricanes. (b) Accumulated cyclone energy (ACE) calculated over the central only (blue bars), eastern only (orange bars), and central‐eastern (yellow bars) Pacific basins. The horizontal black dashed line represents the value of the central‐eastern Pacific ACE’s 75th percentile for the period 1975–2017.The panels (c) and (d) explain the “hurricane fueling mechanism.” This process is represented by the second principal component (PC2) of the interannual anomalies of subsurface temperature (5–105 m average). In short, when PC2 > std(PC2), the El Niño equatorial upper‐ocean heat is discharged meridionally toward the CEP hurricane region after the wintertime event’s peak. Conversely, when PC2 < –std(PC2), the heat is confined (recharged) in the equatorial region. The shading denotes summer (June–November) mean subsurface temperature averaged over recharged periods (PC2 < –std[PC2]) in (c) and discharged periods (PC2 > std[PC2]) in (d). The solid blue line denotes the 26°C isotherm to SST averaged over the corresponding composite periods. In each panel, the colored lines represent the trajectories of major tropical cyclones that occurred during the corresponding composite periods (same color code as in panel [a]). (After Jin et al., 2014)Panel (e) shows the hurricane season (May–November) average of different oceanic and atmospheric variables and ACE anomalies for composite post‐CP (blue) and post‐EP (orange) El Niño, post‐neutral (yellow), and post–La Niña (purple) conditions and for different atmospheric reanalysis products. The different events are defined as follows: EP events: 1983, 1987, 1988, 1992, 1998. CP events: 1991, 1995, 2003, 2005, 2010. Neutral events: 1979, 1980, 1986, 2001, 2012. La Niña events: 1989, 1999, 2000, 2008, 2011. ACEA = ACE anomalies, SSTA = sea surface temperature anomalies, T80A = the anomalies of upper‐ocean temperature averaged in the first 80 m, VWSA = vertical wind shear anomalies, and ERHA = environmental relative humidity anomalies. (After Boucharel et al., 2016c)

ENSO and Tropical Cyclones  385

unusual intensification10 (Foltz & Balaguru, 2016; Huang et  al., 2017). The strong 2015–2016 event and other climate factors (see footnotes 10 and 12) favored both the 2015 El Niño developing season and the 2016 decaying season to be very active, with increased intensity of TCs in the central Pacific (Boucharel et  al., 2016b; Murakami et  al., 2017a). This emphasizes the value of understanding the environmental controls of CEP TC activity in general and related to ENSO in particular. 17.3.2. The Influence of ENSO on CEP TCs In general, the strength of the CEP cross‐equatorial pressure gradient determines the latitudinal position and intensity of the mean ITCZ and its subsequent subweekly variability, which is responsible for the amplification of deep convection and the formation of the initial tropical depression (Toma & Webster, 2010). TC genesis in the CEP can also be triggered by African easterly waves from the North Atlantic, topographical effects, ITCZ breakdown, upper­level potential vorticity, and the confluence between the monsoon westerlies and trade easterlies (Avila, 1991; Bosart & Bartlo, 1991; Zehnder, 1991; Bister & Emanuel, 1997; Ferreira & Schubert, 1997; Rappaport et  al., 1998; Zehnder et  al., 1999; Molinari et  al., 2000; Vincent & Fink, 2001).

(developing) years12,13 because the warming signal is largely confined along the equator (over the cold tongue region and thus far away from the active TC region, which stands mostly northward of 10°N; Jin et  al., 2014; Murakami et al., 2017a). Therefore, during El Niño years, the increase in TC activity is primarily due to changes in atmospheric conditions (reduction in VWS, increase in vorticity; Chu & Wang, 1997; Whitney & Hobgood, 1997; Chu, 2004; Collins, 2007), which tend to promote an overall increase of intense TC counts (Gray & Sheaffer, 1991), as well as a westward displacement of both TC tracks and the main genesis region (Irwin & Davis, 1999; Kimberlain, 1999; Chu & Zhao, 2007; Wu & Chu, 2007; Camargo et al., 2008). In addition, the topographic Central American gap winds can also drive strong intrabasin variability in CEP TCs, with activity enhanced in the western part of the basin but reduced in the eastern near shore region, during El Niño years (Fu et al., 2017). Using composites of a genesis potential index for different ENSO phases, Camargo et al. (2007b) concluded that VWS is the main contributor for the ENSO modulation of TC activity in this region, with potential intensity (i.e. the maximum sustainable intensity of TCs based on the thermodynamic state of the atmosphere and sea surface, measured in terms of the maximum winds or minimum central pressure; Emanuel, 1995) also playing a role. Overall, weaker (stronger) VWS during El Niño (La Niña) years leads to enhanced (decreased) TC activity in the western CEP.

17.3.2.1.  Westward Shift of  TC Activity During El Niño Developing Years El Niño can strongly modulate CEP TC activity by shifting the active regions toward the dateline (i.e. westward) and promoting enhanced seasonal TC activity during El Niño years, as well as a longer TC lifetime (Chu & Wang, 1997; Kimberlain 1999; Chu, 2004; Toma & Webster, 2010; Wood & Ritchie, 2013; Bell et  al., 2014; Caron et al., 2015; Jien et al., 2015; Fu et al., 2017; Huang et  al., 2017). From Collins and Mason (2000) and Camargo et al. (2008), the western CEP subregion (typically west of 116°W) is more sensitive to El Niño than the eastern subregion11, though the eastern subregion usually has greater TC counts. Oceanic conditions and in particular SST do not have a major influence on CEP TC activity during El Niño

17.3.2.2. Oceanic Control of TC Activity in the CEP ENSO‐related anomalies in ocean thermocline depth (and associated OHC) can affect the interannual variability of TC activity over the entire CEP (Collins & Mason, 2000; Collins, 2007; Balaguru et al., 2013; Vincent et al., 2014). The region of TC influence in this basin stands above a shallow “heat‐depressed” thermocline ridge (i.e. strong stratification in the upper ocean thermal structure) and is sensitive to slight variations in the upper‐ocean vertical structure (Shay & Brewster, 2010; Jin et al., 2014; Boucharel et al., 2015). Even small variations in thermocline depth (e.g. from ENSO or other environmental origins, such as the passage of an ocean eddy) can contribute to subsurface OHC anomalies and potentially influence TC intensification (Lin et al., 2005; Wu et al., 2007).

10 Favorable ocean (deeper thermocline [higher OHC] and lower salinity) condition is one factor contributing to Patricia’s extraordinary intensification. There is much ongoing research to further understand Patricia’s intensification, including Patricia’s own internal structure and atmospheric environment (Rogers et al., 2017). 11 This is referring mainly to atmospheric impact. Atmospheric impact from El Niño is more evident over the west CEP than the east CEP. Ocean impact (see following discussions) can also impact the east CEP (Jin et al., 2014; cluster 3 in Caron et al., 2015).

12  An exception was 2015. It was featured by much warmer SST over TC active region, as compared to other El Niño years, e.g. 1997 (see Figure 3ab in Murakami et al., 2017a). Additional features present in 2015, such as the “warm blob” feature (Bond et  al., 2015) and the PMM (Murakami et  al., 2017a) are suggested to contribute to the unusually warm SST in the CEP TC active region (Huang et al., 2017). 13  During El Niño decaying years, ocean conditions (especially subsurface OHC and structure) can significantly influence TC activity; see later discussion.

386  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Recent studies have highlighted the considerable oceanic control of TC activity in the CEP during the decaying phase of El Niño. Jin et  al. (2014) showed that the classic ENSO recharge‐discharge theory (Jin, 1997; chapter  6, this volume) has an important influence on seasonal variations of TC activity. Specifically, the warm water in the upper ocean along the eastern equatorial Pacific associated with the wintertime peak of El Niño can be transported meridionally toward higher latitudes and reach the CEP TC active region in about 6–9 months, right in time for the following hurricane season. Jin et al. (2014) suggested that the redistribution of this subsurface heat reservoir (i.e. OHC) could promote a significant increase in major TCs (≥ category 3) during the El Niño decaying years14. Overall, this “fueling mechanism” explains 20%–40% of the total interannual variability of TC activity in the region (Figures 17.5cd). 17.3.2.3. ENSO Diversity and CEP TCs This ocean discharge mechanism plays a crucial role during the TC season following EP El Niño events but is only marginal after CP El Niño events (Ren & Jin, 2013; Jin et al., 2015; Boucharel et al., 2016c). On the atmosphere side, the westward shift of anomalous SST warming associated with CP events induces a descending motion of dry air in the eastern CEP and therefore a reduction of TC activity in this region during the developing phase of CP El Niño (Figure 17.2 and Figure 5 of Kim et al., 2011). The TC seasons following the peak of CP El Niño (i.e. in CP El Niño decaying year) are near normal due to the compensating effects of low heat content and reduced VWS (Figure 17.5e; Boucharel et al., 2016c). 17.3.3. The Influence of Other Modes of Climate Variability on CEP TCs Akin to ENSO’s ability to remotely influence Atlantic TC activity, there is a teleconnection between tropical Atlantic SST and CEP TC activity (Caron et  al., 2015; Patricola et  al., 2017). Cool northern tropical Atlantic SST anomalies tend to support active CEP seasons due to reduced VWS, and vice versa (Caron et al., 2015; Patricola et al., 2017). In addition, TC‐permitting climate simulations suggest that extremely active or inactive CEP TC seasons are driven by SST variability in the Pacific and Atlantic (and associated large‐scale atmospheric circulation changes), rather than internal atmospheric variability (Patricola et al., 2017). 14  It is important to note that this mechanism is specifically related to the aftermath of El Niño, i.e. in El Niño decaying years. Usually, El Niño decaying years can be either neutral or La Niña years. If an El Niño decaying year is also a La Niña year, the situation is more complex (due to other coexisting ocean processes) and may not be explained only by the recharge-discharge theory (Jin, 1997).

MJO modulation is clear in the CEP (especially over the eastern CEP; Camargo et al., 2008), with an increase in TC frequency (Maloney & Hartmann, 2000) and tracks closer to Central America (Camargo et al., 2009). There is also a link between atmospheric Kelvin waves, the MJO, and the probability of CEP TC genesis (Molinari et  al., 1997; Maloney & Hartmann, 1998, 2000; Molinari et al., 2000; Molinari & Vollaro, 2000; Aiyyer & Molinari, 2008; Schreck & Molinari, 2011; Jiang et al., 2012; Schreck et al., 2012). The MJO also imposes its imprint on the tropical Pacific Ocean. In particular, oceanic equatorial Kelvin waves, which are generated by MJO‐like atmospheric variability and participate in the El Niño buildup, also exert evident control on seasonal TC intensity. These oceanic waves and their reflection as off‐equatorial Rossby waves have a marked signature on the CEP upper‐ocean properties in boreal summer and autumn and can explain up to 15% of TC activity seasonal variations (Boucharel et al., 2016a). At multidecadal timescales, the frequency and intensity of CEP TCs are modulated by the PDO (Chu & Clark, 1999; Chu, 2002; Mantua, 2002; Chu & Zhao, 2004). During PDO warm phase, favorable conditions including warmer local SST, low sea level pressure, reduction in VWS, and increased total precipitable water tend to promote increased TC activity (Chu, 2002). The PDO transitioned from a cool to a warm phase in the late 1970s, concurrently with an increase in CEP TC activity, and then to a cool and rather inactive TC phase in the late 1990s (Zhao & Chu, 2006). It seems that 2014 marks the return to a warm PDO phase, along with a significant increase in CEP TC activity (cf. Figure 17.5b), despite a marginal El Niño event in 2014 (Sobel et al., 2016a). 17.4. NORTH ATLANTIC TCs 17.4.1. Introduction The North Atlantic (NA) basin averages 12 named storms, 6 hurricanes, and 3 major hurricanes each year (Landsea & Franklin, 2013). These hurricanes can cause devastation in the United States, the Caribbean, and Central America. For example, hurricanes Harvey, Irma, and Maria in 2017 caused over $200 billion USD in damage and were responsible for thousands of fatalities. ENSO plays a significant role in driving both basin‐wide and landfalling TC activity levels, as will be discussed in this section. 17.4.2. The Influence of ENSO on NA TCs 17.4.2.1. TC Frequency The influence of ENSO on NA TCs has been documented in many studies, starting with the pioneering

ENSO and Tropical Cyclones  387 (a)

(b)

45N 30N 15N EQ 15S 30S 45S 120E

(d)

(c)

180

150E –1.5

–1.25

150W

–1

–0.75 –0.5 –0.25 0.25

–1

–0.75 –0.5 –0.25 0.25

120W 0.5

0.75

90W 1

1.25

1

1.25

60W

1.5

45N 30N 15N EQ 15S 30S 45S 120E

180

150E –1.5

–1.25

150W

120W 0.5

0.75

90W

60W

1.5

Figure 17.6  (a) Tracks of Atlantic hurricanes at category 3+ intensity in the 10 strongest ASO El Niño events since 1950 (15 category 3+ hurricanes). (b) August–October–averaged SST anomaly (°C) composite in the ten strongest ASO El Niño events since 1950. (c) As in (a) but for the 10 strongest ASO La Niña events since 1950 (33 category 3+ hurricanes). (d) As in (b) but for the 10 strongest ASO La Niña events since 1950.

work of Gray (1984a). He noted that when El Niño c­onditions were developing during boreal summer, the NA had fewer hurricanes. Conversely, La Niña years favored more hurricanes in the NA (Figure  17.6). The reduction in TC activity during El Niño years was attributed to increased vertical wind shear associated with anomalous upper‐level westerly winds driven by an eastward‐shifted and weaker Walker Circulation. Conversely, La Niña events were associated with a stronger Walker Circulation and anomalous upper‐level easterly winds in the Caribbean. These observed relationships have been confirmed in many studies that followed (e.g. Gray et al., 1994; Goldenberg & Shapiro, 1996; Landsea et al., 1999; Tartaglione et al., 2003; Camargo et al., 2007a; Klotzbach, 2011; Patricola et  al., 2014; Boudreault et  al., 2017; Klotzbach et al., 2017). More recently, atmospheric thermodynamic conditions have also been found to be less favorable in the NA during El Niño years, due to anomalous upper‐tropospheric warming (Tang & Neelin, 2004), which causes the atmosphere in the NA to be drier and more stable (Camargo et al., 2007a). In addition, during El Niño events, the rest of the tropics tends to be warmer than the NA, causing the

relative SST15 to favor TC activity in basins other than the NA, as can be explained by simple physical arguments (Latif et  al., 2007; Vecchi, et  al., 2007; Ramsay & Sobel, 2011). La Niña years tend to favor NA TC activity due to a generally cooler tropics relative to tropical NA SSTs16. The strength of El Niño is also important, with stronger El Niño events imparting stronger shear and reducing NA TC activity by a larger degree than weaker events (Patricola et al., 2014). For example, the two most recent strong El Niño events (Oceanic Niño Index ≥ 1.5°C) during August–October were in 1997 and 2015. Both of these 15  Relative SST is the difference in SSTs between the tropical Atlantic and rest of the global tropics. When relative SST is positive (e.g. the Atlantic is warm relative to the rest of the tropics), Atlantic TC activity is favored due to increased convection and decreased vertical wind shear, whereas when relative SST is negative, Atlantic TC activity is suppressed. 16  When the rest of the global tropics is relatively warmer than the Atlantic, as is typically the case in developing El Niño years, it tends to favor suppressed vertical motion and increased vertical wind shear in the tropical Atlantic. In developing La Niña seasons, the tropical Atlantic is warmer relative to the rest of the global tropics, thereby favoring enhanced vertical motion and decreased vertical wind shear in the tropical Atlantic.

388  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

met the National Oceanic and Atmospheric Administration (NOAA) definition for below‐normal NA hurricane seasons (ACE ≤ 66). However, 2004 was classified as a weak El Niño during August–October, and the NA had a hyperactive season, with six major hurricanes (1981–2010 average is three) and an ACE of 227 (1981–2010 average is 106). 17.4.2.2. Landfall and Regional Impacts While early studies focused on ENSO’s influence on basin‐wide NA hurricane activity (e.g. Gray, 1984a, 1984b; Gray et  al., 1994; Goldenberg & Shapiro, 1996), more recent research has also implicated ENSO at regional scales. Tartaglione et al. (2003) and Klotzbach (2011) have noted that Caribbean TC activity is especially reduced relative to the rest of the basin during El Niño seasons, while Kossin and Vimont (2007) noted that the influence of El Niño was less than the Atlantic Meridional Mode (AMM) in terms of its influence on shear in the eastern portion of the Atlantic basin. El Niño has also been shown to reduce landfalling hurricane activity in the continental U.S., while La Niña increases landfalling continental U.S. hurricane likelihood (Bove et al., 1998; Klotzbach, 2011; Klotzbach et al., 2018). These differences in landfall between El Niño and La Niña are significant for hurricanes making landfall along Florida and the East Coast but are insignificant for the Gulf Coast (Klotzbach et  al., 2018). As would be expected given these changes in landfalling hurricane likelihood, normalized damage from continental U.S. hurricanes is significantly reduced during El Niño seasons and increased during La Niña seasons (Pielke & Landsea, 1999; Pielke et  al., 2008; Klotzbach et  al., 2018). At a regional level, Florida and East Coast hurricane normalized damage increases during La Niña seasons and decreases during El Niño seasons, while no significant differences are observed for Gulf Coast hurricanes (Klotzbach et al., 2018). 17.4.2.3. TC Tracks Several studies have also examined the influence of ENSO on NA TC formation locations and tracks. Kossin et  al. (2010) used track cluster analysis to show that La Niña tends to favor TC genesis in the deep tropics (as opposed to the subtropics and the Gulf of Mexico), which makes physical sense since the influence of ENSO on shear is strongest in the tropics. Colbert and Soden (2012) demonstrated that El Niño tends to weaken the NA subtropical high, promoting recurvature of Atlantic TCs. This implies that El Niño reduces continental U.S. landfall likelihood not only due to fewer basin‐wide TCs, but also due to atmospheric steering currents that inhibit TC landfall in the continental U.S. 17.4.2.4. ENSO Diversity and NA TCs In recent decades, a change in the spatial patterns of El Niño has been observed, with maximum SST warming occurring more frequently and with greater intensity in

the central equatorial Pacific (e.g. CP El Niño) compared to the EP El Niño (Ashok et al., 2007; Kug et al., 2009; Lee & McPhaden, 2010). These different so‐called flavors of El Niño have substantially different influences on NA TC activity, with CP El Niño significantly more effective at suppressing NA TC activity compared to EP El Niño, for equal magnitudes of SST warming (Patricola et  al., 2016). This is because less warming is required near the warm pool to satisfy the SST threshold for deep convection (e.g. Williams et  al., 2009; Johnson & Xie, 2010), which influences the strength of the Walker Circulation response and its associated influence on NA vertical wind shear (Patricola et al., 2016). 17.4.3. The Influence of Other Modes of Climate Variability on NA TCs In addition to ENSO, NA TC activity is strongly influenced by other interannual to multidecadal Atlantic SST variability, including the AMM, which is characterized by interannual‐to‐decadal variability in the meridional gradient of tropical Atlantic SST (Chang et al., 1997; Servain et  al., 1999; Chiang & Vimont, 2004), and the Atlantic Multidecadal Oscillation (AMO), which describes multidecadal variations in NA SST. In particular, the positive phase of the AMM (anomalously warm northern and anomalously cool southern tropical Atlantic SST) and the warm phase of the AMO favor active Atlantic hurricane seasons (Landsea et  al., 1999; Goldenberg et  al., 2001; Kossin & Vimont, 2007; Vimont & Kossin, 2007). In combination, ENSO and the AMO or AMM can produce constructive or compensating influences on Atlantic TC activity. For example, La Niña and a positive AMO or AMM together tend to support the most active seasons, whereas El Niño is not a guarantee of an inactive season, especially if paired with a positive AMM, such as was the case in 2004 (Klotzbach, 2011; Patricola et al., 2014). In addition, NA TC tracks are connected to the Indian ­monsoon, with an anomalously strong Indian monsoon driving an anomalously strong NA subtropical high when controlling for ENSO (Kelly et al., 2018). Finally, at subseasonal timescales, the MJO modulates NA TC activity due to changes in both dynamic and thermodynamic large‐scale conditions (e.g. Mo, 2000; Klotzbach, 2010). When convection is enhanced over Africa and the western Indian Ocean (enhanced over the western North Pacific), vertical wind shear is reduced (increased), low‐level vorticity is increased (reduced), midlevel humidity is increased (reduced), and vertical motion is increased (reduced) (Camargo et  al., 2009; Klotzbach, 2010). Klotzbach and Oliver (2015) showed that El Niño and suppressed phases of the MJO are associated with extremely quiet periods for Atlantic hurricane activity, while La Niña and enhanced phases of the MJO are associated with extremely active periods for Atlantic

ENSO and Tropical Cyclones  389

hurricane activity, with up to 10 times as much ACE occurring with La Niña and enhanced phases of the MJO than with El Niño and suppressed phases of the MJO. 17.5. NORTH INDIAN OCEAN (NIO) TCs 17.5.1. Introduction Although the NIO does not have as many TCs as the other basins, the average of 4.3 TCs that occur each year can cause devastating impacts on southern Asian countries, including India, Bangladesh, Pakistan, Myanmar, and Oman. For example, in 2008, TC Nargis claimed over 130,000 lives in Myanmar (Webster, 2008; Lin et al., 2009; McPhaden et al., 2009). Given the dense population and inadequate infrastructure, strong TCs are a major threat to livelihoods in south and Southeast Asian countries. Yet current understanding is still unsatisfactory on the formation, development, and movement of TCs in the NIO. Due to the unique setting of the NIO, with strong monsoonal influences, the situation here is different from other basins. It is important to first understand the NIO setting in detail before exploring ENSO’s influence on NIO TCs. Also, because ENSO’s influence is not as evident to NIO TCs as compared to other basins, this section provides a shorter review on ENSO’s influence, with a longer discussion on the influence from other important modes of variability. TC formation in the NIO exhibits a double peak, with both a premonsoon (May–June) and a postmonsoon (October–December) season (Kikuchi et al., 2009; Evan & Camargo, 2011). The absence of summer TC activity is mainly due to the presence of strong easterly VWS driven by the monsoon (Kikuchi et  al., 2009), although the ambient relative humidity is favorable for TC activity (Li et  al., 2013). Owing to the complex baroclinic structure of the monsoonal flows, the effects of steering flows on TC movement are not as prominent as in other TC basins. The easterly waves that provide incipient disturbances for TC genesis in the NA, CEP, and western North Pacific are virtually absent in the NIO17. The land‐sea configuration of the NIO confines TCs to the Bay of Bengal (BoB) and Arabian Sea. In the postmonsoon season, about 80% of NIO TCs form in the BoB, with the highest frequency of severe TCs occurring during November (Singh et  al., 2001; Mahala et  al., 2015). TC maximum intensity is limited by the relatively small size of the ocean basin and the associated short life spans

17  There are other sources of incipient disturbances in the NIO, e.g. from intraseasonal oscillation events (Kikuchi et al., 2009).

of NIO TCs. As such, the maximum intensity of TCs over the NIO tends to be weaker than those over the Atlantic and Pacific. The terminology used for classification of TC intensity in the NIO also differs from other basins. Details are discussed in Kikuchi et  al. (2009). Whereas TC genesis occurs more frequently during the postmonsoon period, super cyclones (category 4 or above) occur more frequently during the premonsoon period due to enhanced OHC and intraseasonal variability (e.g. MJO and other intraseasonal oscillation; Lin et al., 2009; Li et al., 2013). About 90% of the Arabian Sea premonsoon TCs occur from mid‐May to mid‐June, during which the climatological mean VWS in the southern Arabian Sea increases from approximately 12 to 25 m s‐1 (Wang et  al., 2012), leaving a narrow window for TC development. Thus, an abnormal monsoon onset can have a large influence on interannual variability of NIO TC activity, with May (June) Arabian Sea TCs being associated with an early (late) onset of the monsoon (Evan & Camargo, 2011). 17.5.2. The Influence of ENSO on NIO TCs ENSO affects NIO TCs mainly during the postmonsoon season as ENSO events mature toward the end of the calendar year. During an El Niño (developing) year, fewer intense TCs are observed in the BoB (Singh et al., 2000; Ng & Chan, 2012). During La Niña developing years, weak VWS over the central and northern BoB and increased relative humidity due to enhanced moisture transport favor TC development (Felton et  al., 2013; Mahala et al., 2015). On the other hand, Li et al. (2016) find insignificant differences in TC frequency between El Niño and La Niña years. During the premonsoon season, La Niña tends to increase lower troposphere cyclonic vorticity and midtroposphere humidity in the northern BoB, favoring TC intensification. Meanwhile, enhanced VWS in the southern BoB tends to reduce TC development (Balaguru et al., 2016). 17.5.3. The Influence of Other Modes of Climate Variability on NIO TCs Indian Ocean Dipole (IOD) SST anomalies reach their maximum during autumn (Saji et al., 1999), which significantly influences postmonsoon TCs in the BoB. During a negative phase of the IOD, warmer SST anomalies over the southeastern tropical NIO and cooler SST anomalies over the western tropical NIO enhance TC activity in the BoB during November (Singh, 2008) by generating an anomalous cyclonic circulation over the BoB (Kripalani & Kumar, 2004). Changes in relative humidity contribute more to the TC difference than vorticity (Li et al., 2016).

390  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

In addition, the influences of ENSO and IOD may enhance each other. Specifically, some El Niño (La Niña) events are coupled with a positive (negative) IOD, reducing (enhancing) postmonsoon TC activity over the BoB. This is because El Niño can induce a positive IOD with southeast Indian Ocean cooling (e.g. the 1997–1998 event). The induced positive IOD can reduce the postmonsoon TC activity, which is similar to the ENSO influence. Therefore, the influences of ENSO and IOD can enhance each other. The PDO can regulate the influence of ENSO on BoB TC activity during October– December (Girishkumar et al., 2015). The El Niño and La Niña events have opposite effects on TC genesis. The warm phase of PDO amplifies this difference while the cold phase of PDO does not. The MJO and associated equatorial waves often provide favorable large‐scale conditions for TC genesis and development (e.g. Bessafi & Wheeler, 2006; Frank & Roundy, 2006; Zhou & Wang, 2007). The deadly TC Nargis was initiated by the emanation of a Rossby wave vortex when the MJO reached the Maritime Continent (Kikuchi et al., 2009). The northeastward shift of MJO convection in the NIO (Kemball‐Cook & Wang, 2001) is critical for rapid development of NIO TCs. The most important factors controlling intraseasonal TC genesis are 500 hPa vertical motion, 850 hPa relative vorticity weighted by the Coriolis parameter, and zonal VWS (i.e. VWS at E‐W direction) (Wang & Moon, 2017; Moon et al., 2018). TC activity during May–June in the Arabian Sea has significantly intensified since 1997 in association with significant reductions of storm‐ambient VWS that was attributed to the dimming effects of increased anthropogenic black carbon and sulphate emissions (Evan et al., 2011). However, Wang et al. (2012) found that the VWS change is due to an advance (by 15 days) in TC occurrences and a corresponding rapid decrease of the climatological monsoon VWS. The increased TC activity was likely caused by the swift Interdecadal Pacific Oscillation (IPO)18 phase transition to a long‐lasting La Niña–like state in the late 1990s (Xiang & Wang, 2013). Finally, Wahiduzzaman et al. (2019) also suggest the influence on NIO TCs by quasi‐biennial oscillation. 17.6. SOUTHERN HEMISPHERE TCs 17.6.1. Introduction About one‐third of the average global number of TCs form in the Southern Hemisphere, most of which impact 18 IPO and PDO are both decadal climate modes. The IPO is defined using basin-wide SST anomaly while PDO is defined using SST anomaly north of 20°N.

the coastal population and infrastructure on the vulnerable island nations of the South Pacific Ocean, as well as on the northern states of Australia. However, the seasonal number of TCs can vary considerably from year to year, and it is well established that ENSO is a major contributor to such variability, as first established by Nicholls (1979) for the Australian region. In this section, we review how ENSO modulates TC activity in the Southern Hemisphere with emphasis on genesis locations, intensity, and tracks. For ease of interpretation, we divide the Southern Hemisphere into two basins and discuss ENSO’s influence on each basin separately: the South Pacific Ocean basin (east of 135°E, which includes eastern Australia and the South Pacific island countries) and the South Indian Ocean basin (west of 135°E, which includes central and western Australia). 17.6.2. The Influence of ENSO on South Pacific Ocean TCs 17.6.2.1. TC Frequency Initially, some contradictory views existed on the relationship between ENSO and TC numbers in the South Pacific Ocean basin (e.g. Ramage & Hori, 1981; Revell & Goulter, 1986a, 1986b; Hastings, 1990), primarily due to incomplete and inconsistent observed TC data sets used in earlier studies. However, using improved and quality‐ controlled data sets from the postsatellite era (i.e. after the mid‐1960s), Basher and Zheng (1995) and subsequent studies (e.g. Chand & Walsh, 2009; Kuleshov et al., 2009; Vincent et  al., 2011; Dowdy & Kuleshov, 2012; Dowdy et  al., 2012; Ramsay et  al., 2012; Magee et  al., 2017) established the existence of a strong link between ENSO and TC activity in the South Pacific. 17.6.2.2. TC Genesis and Tracks A well‐documented influence of ENSO on TC activity in the South Pacific Ocean is the geographical distribution of TC genesis and tracks (e.g. Kuleshov et  al., 2008; Ramsay et al., 2012). During El Niño years, TC activity is enhanced east of about 170°E, extending northeastward to the Cook Islands and French Polynesia (Figures 17.2 and 17.7). Simultaneously, low activity dominates west of 170°E in the Coral Sea and Australian regions. In contrast, the reverse occurs during La Niña years when TC activity occurs primarily west of 170°E with relatively low activity east of about 170°E. These changes in TC activity due to ENSO are physically controlled by the associated shifts in the South Pacific Convergence Zone (SPCZ), which is a major feature of the convective activity in the South Pacific Ocean (Vincent, 1994). A southwestward shift in the SPCZ during La Niña corresponds to more TCs forming farther west in the South Pacific, and a

ENSO and Tropical Cyclones  391

northeastward shift in the SPCZ during El Niño corresponds to more TCs forming around the central and eastern South Pacific regions (Vincent et  al., 2011; Dowdy et  al., 2012). Ramsay et  al. (2012) concluded that the ENSO‐modulating effect on genesis is primarily caused by changes in low‐level zonal flow between the equator and 10°S and associated relative vorticity changes in the main development regions. Furthermore, they found a significant modulation in genesis location by ENSO in only three specific track types in the Southern Hemisphere, with TCs forming further equatorward (poleward) during El Niño (La Niña) in addition to the large shifts in longitude. 17.6.2.3. ENSO Diversity and South Pacific Ocean TCs In addition, Chand et al. (2013a, 2013b) identified two more ENSO regimes that resembled CP El Niño features in the South Pacific, termed “positive‐neutral” and

“n­egative‐neutral” regimes. The positive‐neutral regime showed an El Niño–type influence with enhanced TC activity in the central and eastern South Pacific, whereas the negative‐neutral regime showed a La Niña–type influence with enhanced activity in the Coral Sea and eastern Australian regions. Magee et al. (2017) confirmed this link, particularly showing that CP El Niño has the largest influence on TC formation in the later part of the Southern Hemisphere TC season. 17.6.2.4. TC Intensity Several studies have also examined the influence of ENSO on mean TC intensity in the South Pacific Ocean (e.g. Chand & Walsh, 2011; Dowdy et al., 2012; Ramsay et al., 2012 and others). They found that during El Niño years, TC intensity reaches its maximum by around 20°S and decreases rapidly thereafter as TCs track poleward. However, the maximum intensity can be maintained

(a)

10S 20S 30S 40S 50S 60S 30E 40E 50E 60E 70E 80E 90E 100E 110E 120E 130E 140E 150E 160E 170E 180 170W 160W 150W 140W 130W (b)

10S 20S 30S 40S 50S 60S

30E 40E 50E 60E 70E 80E 90E 100E 110E 120E 130E 140E 150E 160E 170E 180 170W 160W 150W 140W 130W

Figure 17.7  Tropical cyclone track density in 2° × 2° gridded boxes for (a) El Niño and (b) La Niña events over the period 1949–1950 to 2005–2006 cyclone seasons in the Southern Hemisphere. (After Kuleshov et al., 2008)

392  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

p­oleward of 20°S during La Niña years. This is consistent with the equatorward (poleward) shift in favorable large‐ scale environmental conditions that affect intensity during El Niño (La Niña) years, in particular, those associated with environmental VWS, midlevel relative humidity, and sea surface temperature (Chand & Walsh, 2011). 17.6.3. The Influence of ENSO on South Indian Ocean TCs TC activity in the South Indian Ocean basin, defined here as the region west of 135°E, is also modulated substantially by ENSO (e.g. Ho et al., 2006; Camargo et al., 2007b; Kuleshov et al., 2008, 2009; Dowdy & Kuleshov, 2012). During El Niño developing years, TC activity is enhanced west of 75°E, extending as far west as the East African coast, and suppressed east of 75°E in the South Indian Ocean basin. However, the opposite happens during La Niña developing years when TC activity is suppressed west of 75°E and enhanced east of 75°E (Figures  17.2 and 17.7). This enhanced (suppressed) activity east of 75°E to about 170°E during La Niña (El Niño) consequently accounts for the increased (decreased) activity over the Australian region as first uncovered by Nicholls (1979) and subsequently confirmed by several studies (e.g. Evans & Allan, 1992; Ramsay et  al., 2008). This relationship has since been used to develop seasonal forecast models to predict TC activity for Australia several months in advance (e.g. McDonnell & Holbrook, 2004; Werner & Holbrook, 2011; Liu & Chan, 2012; Ramsay et  al., 2014). Several large‐scale environmental factors critical for South Indian Ocean cyclone development have been identified. For example, Kuleshov et al. (2009) attributed shifts in TC activity during ENSO phases to changes in vorticity and relative humidity, and Camargo et  al. (2007b) showed that the shifts are primarily due to changes in VWS. Such changes in the environmental conditions during different phases of ENSO are likely to be due to shifts in convective activity associated with the Walker Circulation (Ho et al., 2006; Yoo et al., 2006). 17.6.4. The Influence of Other Modes of Climate Variability on Southern Hemisphere TCs Other modes of climate variability that operate at different time scales can also influence TCs in the Southern Hemisphere. For example, the MJO and associated tropical waves can enhance (suppress) TC activity during their active (inactive) phases as they propagate eastward in the tropical Pacific (e.g. Bessafi & Wheeler, 2006; Frank & Roundy, 2006; Leroy & Wheeler, 2008). The number of TCs reaching TC (≥ category 1) or major TC (category 3 to 5) intensity can also undergo significant enhancement during active compared to inactive phases of the MJO

(Chand & Walsh, 2010; Klotzbach, 2014). The relationship between TC activity and the MJO can be further strengthened during coincident El Niño events. For example, if enhanced phases of MJO occur during El Niño, more TCs are likely to form in the Australian and South Pacific regions when compared to enhanced phases of MJO occurring during La Niña events (Chand & Walsh, 2010). In addition, the IOD can influence Southern Hemisphere TC activity. For example, the negative phase of IOD led to an increase in the likelihood of TCs forming in the eastern Australian region, and vice versa during the positive IOD phase (Liu & Chan, 2012). Similarly, the IOD Index was the most commonly occurring predictor of the best‐ performing seasonal prediction models for the Australian region (Wijnands et  al., 2015). In addition, seasonal TC activity in the Southern Hemisphere can also vary at decadal or multidecadal timescales (e.g. Callaghan & Power, 2011), for example, through the influences of PDO, but such variability remains relatively unexplored due to a lack of consistent long‐term data records. 17.7. TROPICAL CYCLONES AND CLIMATE CHANGE 17.7.1. Introduction Like natural climate variability, anthropogenic climate change can strongly influence TC activity. The influence of climate change on TCs has been a focal point of much research over the last few decades, as summarized in several review papers (Knutson et  al., 2010; Grossmann & Morgan, 2011; Sobel et  al., 2016b; Walsh et  al., 2016; Knutson et  al., 2019a, 2019b). There are two separate aspects of this issue: (i) Is it possible to detect changes in TC activity from the existing historical data, and can they be attributed to climate change? (ii) How much confidence do we have in climate model projections of future TC activity? This section will review recent research on the influence of climate change on TCs, as well as the links between ENSO, climate change, and TCs. 17.7.2. Observed Trends in the Historical Climate The ability to detect historical trends in various measures of TC activity is complicated by two factors: the large interannual‐to‐decadal variations of TC activity associated with natural climate variability and the length and quality of historical TC data sets. Therefore, it remains uncertain if past changes in global TC characteristics, such as frequency, intensity, and rainfall, exceed natural variability, after considering changes in observing capabilities (Landsea et al., 2006; Knutson et al., 2010; Landsea et al., 2010). In particular, the global best‐track datasets for TCs lack temporal and interregion homogeneity, which can be

ENSO and Tropical Cyclones  393

problematic when attempting to detect trends in TC metrics. To address this issue, Kossin et al. (2013) developed a new homogenized record of TC intensity by applying the Dvorak technique (Dvorak, 1975) to a homogenized satellite data set that was sampled spatially and temporally to match the early period. The homogenized observations indicate that strong TCs (category 4 and above) became more intense globally during the 1982–2009 period, although the statistical significance of this trend is marginal (Kossin et al., 2013). On the other hand, Klotzbach and Landsea (2015) found a small insignificant downward trend of category 4 and 5 TCs over the 1990–2014 period using standard historical datasets. The contrast between these two analyses illustrates the difficulties in detecting trends in the historical record of TC activity. Trends in other aspects of TC activity have also been investigated. For instance, a poleward shift in the latitude of lifetime maximum intensity of TCs has been identified (Kossin et al., 2014), associated with a poleward shift in genesis location (Daloz & Camargo, 2018). These shifts are thought to be due to the expansion of the Hadley cell (Sharmila & Walsh, 2018). Moon et al. (2015) pointed out that the statistical significance of these poleward trends is sensitive to the specific subset of the TC data considered. Recently, Kossin (2018) identified a global slowdown in the translation speed of TCs. 17.7.3. Attribution Studies Whereas trend analyses focus on metrics of aggregated TC activity, attribution studies quantify how climate change may have influenced specific TC events and seasons. Following Hurricane Harvey’s record rainfall and devastating floods in South Texas, several studies estimated the contribution of climate change to this event using a variety of methods, including statistical analysis of observed rainfall data (Risser & Wehner, 2017), a budget analysis (Trenberth et al., 2018), a combination of modeling and data analysis (van Oldenborgh et al., 2017), and climate modeling (Emanuel, 2017; Wang et  al., 2018). Combined with convection‐permitting climate model experiments of hurricanes Katrina, Irma, and Maria (Patricola & Wehner, 2018), these studies provide evidence that climate change has already begun to enhance TC rainfall. In addition, various studies have quantified the influence of climate change on specific TC basins and seasons, highlighting the delicate balance between anthropogenic effects, climate variability, and stochastic processes (e.g. Murakami et al., 2015, 2017a, 2017b, 2018).

generation climate models, depending on the TC characteristics. The most robust projections are for an increase in TC intensity and rainfall. Specifically, models and theory project an increase in the global mean maximum wind speed of TCs (2%–11%), with the frequency of the most intense storms expected to increase. In addition, the maximum wind speed of intense TC events may increase by 10–15 knots by the end of the 21st century (Patricola & Wehner, 2018). Furthermore, rainfall rates are also likely to increase ~20% within 100 km of the TC center (Knutson et al., 2010), with a substantial range of uncertainty in total storm rainfall (Scoccimarro et  al., 2014; Patricola & Wehner, 2018). Finally, flooding by TCs will almost certainly increase as a result of accelerated sea‐ level rise (Woodruff et al., 2013). On the other hand, climate models disagree on whether TCs are expected to become more or less frequent in the future (e.g. Knutson et  al., 2010; Camargo, 2013; Emanuel, 2013; Walsh et  al., 2016; Bhatia et  al., 2018). Other aspects of TC characteristics, such as tracks and duration, have lower confidence projections. For instance, changes in NA TC tracks are highly dependent on future scenarios (Daloz et al., 2015). A shift from the western to eastern Atlantic has also been reported (Murakami & Wang, 2010). In contrast, in the WNP (e.g. Nakamura et al., 2017), two robust track changes can be identified: an eastward shift near the central Pacific (Li et al., 2010) and a poleward shift (Kossin et al., 2016). Future changes in environmental genesis parameters19 are not uniform across basins. For instance, although TC genesis parameters change significantly under global warming in the NA, that is not the case in the WNP (Murakami et al., 2013a). In contrast to the present‐day climate, in which thermodynamic genesis parameters are dominant, in the future climate, dynamical factors will become more important in controlling genesis in the NA (Murakami et al., 2013a). In order to make reliable future TC projections, it is necessary for climate models to resolve TCs. Low‐­ resolution models are unable to accurately simulate TCs (Camargo & Wing, 2016), leading to uncertain projections (e.g. Camargo, 2013; Tory et al., 2013). The most reliable TC projections have been done using high‐resolution (~0.25°) climate models either coupled (e.g. Bhatia et al., 2018) or forced with fixed projected SSTs (e.g. Roberts et al., 2015; Wehner et al., 2015), or downscaled (Emanuel, 2013; Knutson et al., 2015). An accurate representation of SST is also critical for TC projections, as the tropical SST biases common to coupled models (Richter, 2015; Zuidema et  al., 2016) can cause substantial errors in

17.7.4. Future Projections Regarding late 21st‐century projections of TC activity, there is large uncertainty in the projections by current‐

19  TC genesis parameters typically include thermodynamic (SST or potential intensity), dynamic (vertical wind shear, vorticity), and humidity (e.g. relative humidity at midlevels) variables.

394  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

simulated TC activity (Hsu et  al., 2018). Uncoupled atmospheric models with prescribed SST can be used to avoid SST biases; however, the lack of air‐sea interaction introduces uncertainty in representing TC intensity (Zarzycki, 2016; Li & Sriver, 2018). Due to the challenges of using coupled and uncoupled climate models to project TC activity, it is useful to understand ocean subsurface changes under global warming and how they may influence TCs. As anthropogenic climate change drives faster warming of SST compared to the underlying subsurface ocean, the future ocean thermal gradient sharpens and stratification increases (Emanuel, 2015; Huang et al., 2015). As a result, the TC‐ induced ocean cooling effect is expected to increase in the future and can potentially weaken the TC potential intensity increasing trends (Huang et al., 2015). If this negative subsurface factor is taken into account, the projections for the occurrence of category 5 TCs are ~15% lower than projections without this process (Emanuel, 2015). 17.7.5. Projections of ENSO and TC Activity At the time of this review, there were relatively few studies that considered the influence of ENSO on TC projections. For instance, Chand et al. (2017) showed that while the general trend is toward fewer TCs in a warming climate globally, some Pacific island nations are likely to experience robust changes in the ENSO‐driven variability of TC occurrence by the late 21st century. In particular, they showed that TCs are highly likely to become more frequent (~20%–40%) during future‐climate El Niño events compared with present‐climate El Niño events (and less frequent during future‐climate La Niña events [~20%–60%]) around a group of small island nations in the Pacific (for example, Fiji and Vanuatu in the South Pacific and the Marshall Islands and Hawaii in the North Pacific; see Figure 17.8). They attributed these changes to the likely disruptions in the Walker Circulation and associated convective zones as a result of global warming. In another study, Murakami et al. (2013b) showed that the frequency of intense TCs near Hawaii is expected to increase drastically with climate change, which is a typical TC response to El Niño in that region. Jin et  al. (2014) estimated an increased probability, between 25% and 45% compared with present climate conditions, for more major TCs (category 3 and above) in the CEP under global warming scenarios associated with the ENSO delayed ocean transport mechanism (section  17.3.2; Jin et  al., 2014; chapter  6). A substantial increase in TC activity over the CEP due to global warming might favor a reverse mechanism, i.e. TCs could influence ENSO and hence global climate, as suggested by Sobel and Camargo (2005). Using the Pliocene epoch as an analogy to global warming, Fedorov et  al. (2010) suggested that an

increased CEP TC activity could promote warmer surface water parcels (due to TC‐induced ocean mixing) to be transported to the equatorial cold tongue region via ocean circulation. This would help maintain a more permanent El Niño–like state. This is another intriguing TC‐ENSO‐climate interaction process that needs to be further explored in the future. One important factor contributing to the uncertainty of TC projections is the large uncertainty on how ENSO will respond to climate change (Yeh et al., 2009; Collins et  al., 2010; Vecchi & Wittenberg, 2010; Power et  al., 2013; Cai et  al., 2014; Williams & Patricola, 2018; chapter 13). Given that most TC projections are based on high‐resolution models forced with projected SST, such uncertainties are inherently a part of the TC projections. Since ENSO is a major driver of interannual variability of TC activity globally, any change in ENSO in a warming climate is likely to have significant implications for regional TC activity (e.g. Camargo et  al., 2010). Furthermore, it is critical to understand future changes in the diversity of ENSO events, or “ENSO flavors” (Ashok et  al., 2007; Kao & Yu, 2009; Capotondi et  al., 2014; Timmermann et  al., 2018; chapter  4), as TC activity in several basins depends strongly on the spatial pattern of warming during El Niño (Patricola et  al., 2016, 2018; L.  Wu et  al., 2018). A new ENSO metric, the ENSO Longitude Index, uniquely describes the diversity of ENSO events by tracking east‐west variations in the average longitude of deep convection and accounting for the nonlinear response of deep convection to SST (Williams & Patricola, 2018). Furthermore, the ENSO Longitude Index includes changes in the background climate state, which are unaccounted for in traditional ENSO metrics based on SST anomalies, and reveals future increases in La Niña, El Niño, and CP events during boreal summer (Williams & Patricola, 2018). Cai et al. (2018) and chapter 13 similarly found an increase in the variability of eastern Pacific El Niño with climate change. Future increases in La Niña, El Niño, and CP events suggest that both active and inactive TC seasons could become more frequent at the expense of average seasons in the Atlantic, CEP, and WNP basins, although factors unrelated to ENSO are also expected to play an important role. 17.8. CONCLUSION AND DISCUSSION ENSO significantly alters the large‐scale atmospheric and oceanic environment globally, thus profoundly influencing global TC activity. This chapter reviews our knowledge about ENSO‐TC relationships in individual basins (sections 17.2–17.6) and the relationship with global warming (section  17.7). Key points are summarized below.

ENSO and Tropical Cyclones  395 Projected changes between future and current-climate overall TCs

(a) 30°N 0° 30°S

60°E

120°E

180°

120°W

60°W

Projected changes between future and current-climate El Niño TCs

(b)

30°N 0° 30°S 60°E

120°E

180°

120°W

60°W

Projected changes between future and current-climate La Niña TCs

(c)

30°N 0° 30°S 120°E

60°E –60

–50

–40

–30

180° –20

20

120°W 30

60°W 40

50

60

(Percentage future change of TCs per year and per 2.5° × 2.5° grid)

Figure 17.8  Projected future changes in TC density defined as the difference between future (2070–2100) and current (1970–2000) climates: (a) overall climatology; (b) El Niño; (c) La Niña. Red shading indicates projected future increases in TC frequency. Stippling denotes changes that are statistically significant at the 95% level, with at least 9 out of 12 models agreeing on the sign of change. Four subregions in the Pacific, as well as the horseshoe‐ shaped region associated with ENSO, are also indicated. (After Chand et al., 2017)

••Over the western North Pacific, during El Niño (developing) years, TC seasons tend to be characterized by longer‐lived TCs that are larger in size and generate greater ACE. These TCs also tend to track away from the Asian coast and form closer to the international date line. During El Niño decaying years, TC frequency drops considerably over the WNP. During El Niño years, upper OHC can decrease by as much as 30% and has a negative impact on TC intensification. This change in OHC can counteract with other positive factors (e.g. longer track) for TC intensification during El Niño, and the final sign after offset is moderately positive (as compared to normal). ••Over the central eastern Pacific, El Niño shifts TCs away from the Mexican coast and toward the international date line. TC activity increases with more major TCs observed during both El Niño and decaying years. During El Niño decaying years, large‐scale oceanic movement via the subsurface recharge‐discharge mechanism brings additional heat to the CEP, causing an increase in numbers of major TCs (category 3 and above; Jin et al., 2014). Due to the higher predictability of this

mechanism as compared to other factors, it has potential to improve intense TC seasonal forecast by up to a 6‐ month lead time. ••Over the North Atlantic, El Niño significantly suppresses TC activity, with a reduction of landfall along the U.S. coastline, while TC activity and U.S. landfall likelihood are increased during La Niña. The combination of La Niña and a positive AMO or AMM presents a potential worst‐case scenario for the most active Atlantic TC seasons. Furthermore, El Niño is not a guarantee of an inactive TC season, especially when occurring with a positive AMO or AMM. The location and magnitude of maximum warming during El Niño has important implications for Atlantic TC activity through its influence on the Walker Circulation. ••The seasonality of TCs in the North Indian Ocean is different from other basins and is characterized by two TC seasons (pre‐ and postmonsoon). In this region, TCs are influenced substantially by the monsoon and MJO, though ENSO still plays a role. In particular, El Niño mainly affects the postmonsoon season, with fewer intense TCs in the Bay of Bengal, whereas TC activity is

396  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

increased during La Niña. An improvement in forecasting over this region, which has co‐occurring influences from the monsoon, MJO, and ENSO, is important. ••ENSO’s influence on the Southern Hemisphere can be divided into two regions: the South Pacific Ocean and the South Indian Ocean. In the South Pacific region, El Niño enhances TC activity east of 170°E toward the Cook Islands and French Polynesia, but reduces it over the Coral Sea and Australia (west of 170°E). The opposite is found during La Niña years. In the South Indian Ocean, TC activity is enhanced (suppressed) west (east) of 75°E and can extend westward as far as the African east coast. Although not covered extensively in this chapter, ENSO can influence TC‐related rainfall. TC‐induced extreme rainfall tends to occur more in East Asia and the northwestern Pacific islands during El Niño (Lyon & Camargo, 2009; Khouakhi et al., 2017). During La Niña, TC‐induced rainfall is enhanced in Australia and along the U.S. East Coast (Lyon & Camargo, 2009; Khouakhi et al., 2017). Another related issue is seasonal TC forecasting. In tandem with the progress on ENSO‐TC research described in this chapter, there are certainly ongoing efforts to improve TC seasonal forecasting. After all, seasonal forecasting can directly benefit society (Gray, 1984a, Chan et al., 1998; Camargo & Zebiak, 2002; Vitart et al., 2007; Klotzbach & Gray, 2009; Vecchi et al., 2011; Werner & Holbrook, 2011; Chen & Lin, 2012; Camp et al., 2015). Given ENSO’s strong link with TC activities, ENSO information has been used extensively over different TC basins (e.g. WNP, CEP, NA, Southern Hemisphere) in the latest forecasting systems for TC annual frequency, ACE, track density, major TCs, and landfall likelihood (Caron et al., 2015; Zhan & Wang, 2016; Klotzbach et al., 2017; Kim & Chan, 2018; Davis & Zeng, 2019). The new results on the ENSO‐TC relationship discussed in this chapter (e.g. the stronger linkage between CP El Niño and WNP TCs; Zhao & Wang, 2018) or the ocean subsurface fueling mechanism for CEP major TCs also may potentially contribute to future improvements in TC seasonal forecasting (Jin et al., 2014). Certainly, ENSO prediction itself is imperative to seasonal TC forecasting. At the moment, it is still difficult to overcome ENSO’s forecasting skill in boreal spring (i.e. the springtime predictability barrier, due to low signal‐to‐noise ratio (chapter  10; Santoso et  al., 2019). Therefore, TC forecasts issued in July/early August generally have more skill than the forecasts issued in April/May (Camargo et  al., 2007a; Camargo & Barnston, 2009; Klotzbach & Gray, 2009; Klotzbach et al., 2017). Besides the aforementioned, there are many other related interesting issues, e.g. ENSO’s relationship with sea‐level rise (chapter 18). Because there is large sea‐level change during an ENSO event, it is relevant for landfall-

ing TCs. For example, during La Niña or La Niña–like conditions, sea level is much higher in the WNP (England et al., 2014). This is compounded with much higher OHC, which fuels record‐breaking TCs (e.g. Haiyan in 2013, second most intense TC in global record; Lin et al., 2014). Due to higher TC intensity and the preexisting higher sea level from La Niña, the surge and flood damage upon landfall can be much more severe, as in the devastation from Haiyan to the Philippines (Lin et al., 2013b, 2014). During El Niño events, there is a large drop in sea level over the WNP. Although TC intensity may increase (Camargo & Sobel 2005; Zheng et  al., 2015), the surge damage from TCs may be offset from the preexisting sea level drop. Conversely, over the CEP, there is evident sea level rise from El Niño events. How this preexisting rise may contribute to the storm surge and flood of landfalling TCs’ surge impact is also intriguing. Besides ENSO’s influence on TCs, it may also be possible for TCs to influence ENSO (see also section 17.7). For example, the role of westerly wind bursts (WWBs, chapter 7) is intriguing. Chen and Lian (2018) and Lian et al. (2018) suggested the association of WWBs with TC genesis and thus may contribute to possible triggering for El Niño events. Finally, linking TC activity to ENSO under climate change is a new field and depends strongly on how ENSO itself will behave under global warming. Latest results and large‐ensemble simulations (Williams & Patricola, 2018) suggest an increase in extreme El Niño (i.e. strong EP type) and extreme La Niña events under global warming, based on CMIP5 modeling (Cai et al., 2014, 2015, 2018; chapter  13) and large‐ensemble simulations (Williams & Patricola, 2018). Modeling efforts to improve simulations of both ENSO and TCs in present and future climates, as well as careful uncertainty considerations, are ongoing and imperative (e.g. Knutson et al., 2007; Bell et al., 2014; Kim et al., 2014). The current understanding is that global warming may reduce TC frequency. However, some southern Pacific islands are likely to experience increased TC frequency during future El Niño events in a global warming scenario as compared to current‐climate El Niño. An increased probability for major (≥ category 3) TC occurrences over CEP is also likely (e.g. Jin et  al., 2014). With more extreme La Niña events (Cai et  al., 2015) associated with greater OHC (and higher TC intensity), together with higher sea level rise over the WNP, the occurrence of category “6” TCs like Haiyan with severe surge damage (Lin et al., 2013ab; 2014) may also increase. Clearly, this new field of research about ENSO’s influence on TCs in a warming world requires more exploration, and many intriguing questions remain open. For example, will ENSO still strongly modulate TC activity in the future? Also, in the present climate, we have observed many ENSO‐related teleconnection pathways

ENSO and Tropical Cyclones  397

that have the ability to impact TCs in other basins (e.g. ENSO’s influence on NA TCs). Will such teleconnection pathways remain, or will new pathways emerge in the future climate to modulate TCs differently in other basins? How will the influence from other modes of natural climate variability on TC activity evolve to enhance or diminish ENSO’s influence in the future? As in the old Chinese saying “toss out a brick to get gems,” we hope this chapter will serve as a brick to invite many new research gems in this highly intriguing field. ACKNOWLEDGMENTS We are grateful to Agus Santoso for excellent editorial advice and comments. Thanks to Tom Knutson, Jinyi Yu, and John Knaff for helpful advice. Special thanks to Dan Fu for constructive suggestion and graphic assistance (Figures 17.1b and 17.2), to Jaison Kurian for constructive suggestions, to Chun‐An Shih, Chun‐Chi Lien, and Hsiao‐Ching Huang for editorial and graphic assistance. We thank two reviewers and chapter meeting participants for very constructive comments and feedback. I.L. acknowledges support from the Taiwan Ministry of Science and Technology. C.M.P acknowledges support from the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, Climate and Environmental Sciences Division, Regional & Global Climate Modeling Program, under Award Number DE‐ AC02‐05CH11231. J.B. acknowledges support from the French Agence Nationale de la Recherche project MOPGA “Trocodyn,” grant number ANR‐17‐ MPGA‐0018. S.C. acknowledges funding support from the Earth Systems and Climate Change Hub of the Australian Government’s National Environmental Science Programme (NESP). T. Li was supported by NSF grant AGS-2006553 and NOAA grant NA18OAR4310298. REFERENCES Aiyyer, A., & Molinari, J. (2008). MJO and tropical cyclogenesis in the Gulf of Mexico and Eastern Pacific: Case study and idealized numerical modeling. Journal of the Atmospheric Sciences,65,2691–2704.https://doi.org/10.1175/2007JAS2348.1 Aiyyer, A., & Thorncroft, C. (2011). Interannual‐to‐multidecadal variability of vertical shear and tropical cyclone activity. Journal of Climate, 24(12), 2949–2962. https://doi. org/10.1175/2010JCLI3698.1 Ashok, K., Behera, S. K., Rao, S. A., Weng, H., & Yamagata, T. (2007). El Niño Modoki and its possible teleconnection. Journal of Geophysical Research, 112, C11007. https://doi. org/10.1029/2006JC003798 Avila, L. A. (1991). Atlantic tropical systems of 1990. Monthly Weather Review, 119, 2027–2033. https://doi.org/10.1175/152 0‐0493(1991)1192.0.CO;2

Balaguru, K., Leung, L. R., & Yoon, J.‐H. (2013). Oceanic control of northeast Pacific hurricane activity at interannual timescales. Environmental Research Letters, 8(4), 044009. https://doi.org/10.1088/1748‐9326/8/4/044009 Balaguru, K., Leung, L. R., Lu, J., & Foltz, G. R. (2016). A meridional dipole in premonsoon Bay of Bengal tropical cyclone activity induced by ENSO, Journal of Geophysical Research: Atmospheres, 121, 6954–6968. https://doi. org/10.1002/2016JD024936 Basher, R. E., & Zheng, X. (1995). Tropical cyclones in the southwest Pacific: Spatial patterns and relationships to Southern Oscillation and sea surface temperature. Journal of Climate, 8, 1249–1260. https://doi.org/10.1175/1520‐0442(19 95)0082.0.CO;2 Bell, G. D., Halpert, M. S., Schnell, R. C., Higgins, R.  W.,  Lawrimore, J., Kousky, V. E., et  al. (2000). Climate assessment for 1999. Bulletin of the American Meteorological Society, 81, S1–S50. https://doi.org/10.1175/1520‐0477(2000) 81[S1:CAF]2.0.CO;2 Bell, R., Hodges, K., Vidale, P.L., Strachan, J., Roberts, M., et al. (2014). Simulation of the global ENSO–tropical cyclone teleconnection by a high‐resolution coupled general circulation model. Journal of Climate, 27(17), 6404–6422. https://doi.org/10.1175/JCLI‐D‐13‐00559.1 Bessafi, M., & Wheeler, M. C. (2006). Modulation of south Indian Ocean tropical cyclones by the Madden‐Julian oscillation and convectively coupled equatorial waves. Monthly Weather Review, 134, 638–656. https://doi.org/10.1175/ MWR3087.1 Bhatia, K., Vecchi, G., Murakami, H., Underwood, S., & Kossin, J. (2018). Projected response of tropical cyclone intensity and intensification in a global climate model. Journal of Climate, 31, 8281–8303. https://doi.org/10.1175/ JCLI‐D‐17‐0898.1 Bister, M., & Emanuel, K. A. (1997). The genesis of hurricane Guillermo: TEXMEX analyses and a modeling study. Monthly Weather Review, 125, 2662–2682. https://doi.org/10. 1175/1520‐0493(1997)1252.0.CO;2 Bister, M., & Emanuel, K. A. (1998). Dissipative heating and hurricane intensity. Meteorology and Atmospheric Physics, 65(3–4), 233–240. https://doi.org/10.1007/BF01030791 Bond, N. A., Cronin, M. F., Freeland, H., & Mantua, N. (2015). Causes and impacts of the 2014 warm anomaly in the NE Pacific. Geophysical Research Letters, 42(9), 3414–3420. https://doi.org/10.1002/2015GL063306 Bosart, L. F., & Bartlo, J. A. (1991). Tropical storm formation in a baroclinic environment. Monthly Weather Review, 119, 1979–2013. https://doi.org/10.1175/1520‐0493(1991)1192.0.CO;2 Boucharel, J., Timmermann, A., Santoso, A., England, M. H., Jin, F.‐F., & Balmaseda, M. A. (2015). A surface layer variance heat budget for ENSO. Geophysical Research Letters, 42, 3529–3537. https://doi.org/10.1002/2015 GL063843 Boucharel, J., Jin, F.‐F., England, M. H., Dewitte, B., Lin, I. I., Huang, H.‐C., et al. (2016a). Influence of intraseasonal oceanic Kelvin waves on the Eastern Pacific hurricane activity. Journal of Climate, 29(22), 7941–7955. https://doi. org/10.1175/JCLI‐D‐16‐0112.1

398  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Boucharel, J., Jin, F.‐F., England, M. H., & Lin, I. I. (2016b). Modes of hurricane activity variability in the Eastern Pacific: implications for the 2016 season. Geophysical Research Letters, 43. https://doi.org/10.1002/2016GL070847 Boucharel, J., Jin, F.‐F., Lin, I. I., Huang, H.‐C., & England, M. H. (2016c). Different control of tropical cyclone activity in the eastern Pacific for two types of El Niño. Geophysical Research Letters, 43, 1679–1686. https://doi.org/10.1002/ 2016GL067728 Boudreault, M., Caron, L.‐P., & Camargo, S. J. (2017). Reanalysis of climate influences on Atlantic tropical cyclone activity using cluster analysis. Journal of Geophysical Research, 122, 4258–4280. https://doi.org/10.1002/2016 JD026103 Bove, M. C., O’Brien, J. J., Elsner, J. B., Landsea, C. W., & Niu, X. (1998). Effect of El Niño on U.S. landfalling hurricanes, revisited. Bulletin of the American Meteorological Society, 79, 2477–2482. https://doi.org/10.1175/1520‐0477(19 98)0792.0.CO;2 Braun, S. A., Archambault, H. M., Lin, I. I., Miyamoto, Y., Riemer, M., Rios‐Berrios, R., et  al. (2018). Ninth WMO International Workshop on Tropical Cyclones: Intensity Change: External Influences, IWTC‐9 report. Cai, W., Borlace, S., Lengaigne, M., vanRensch, P., Collins, M., Vecchi, G., et al. (2014). Increasing frequency of extreme El Niño events due to greenhouse warming. Nature Climate Change, 4, 111–116. https://doi.org/10.1038/nclimate2100 Cai, W., Wang, G., Santoso, A., McPhaden, M. J., Wu, L., Jin,  F.‐F. et  al. (2015). Increased frequency of extreme La Niña events under greenhouse warming. Nature Climate Change, 5(2), 132–137. https://doi.org/10.1038/nclimate2492 Cai, W., Wang, G., Dewitte, B., Wu, L., Santoso, A., Takahashi, K., et al. (2018). Increased variability of eastern Pacific El Niño under greenhouse warming. Nature, 564(7735), 201–206. https://doi.org/10.1038/s41586‐018‐0776‐9 Callaghan, J., & Power, S. B. (2011). Variability and decline in the number of severe tropical cyclones making lad‐fall over eastern Australia since the late nineteenth century. Climate Dynamics, 37, 647–662. Camargo, S. J., & Zebiak, S. E. (2002). Improving the detection and tracking of tropical stroms in atmospheric general circulation models. Weather Forecasting, 17, 1152–1162. Camargo, S. J., & Sobel, A. H. (2005). Western North Pacific tropical cyclone intensity and ENSO. Journal of Climate, 18(15), 2996–3006. https://doi.org/10.1175/JCLI3457.1 Camargo, S. J., Barnston, A. G., Klotzbach, P. J., & Landsea, C.  W. (2007a). Seasonal tropical cyclone forecasts. World Meteorological Organization Bulletin, 56, 297–309. Camargo, S. J., Emanuel, K. A., & Sobel, A. H. (2007b). Use of a genesis potential index to diagnose ENSO effects on tropical cyclone genesis. Journal of Climate, 20(19), 4819–4834. https://doi.org/10.1175/JCLI4282.1 Camargo, S. J., Robertson, A. W., Gaffney, S. J., Smyth, P., & Ghil, M. (2007c). Cluster analysis of typhoon tracks. Part I: General properties. Journal of Climate, 20(14), 3635–3653. https://doi.org/10.1175/JCLI4188.1 Camargo, S. J., Robertson, A. W., Gaffney, S. J., Smyth, P., & Ghil, M. (2007d). Cluster analysis of typhoon tracks. Part II:

Large‐scale circulation and ENSO. Journal of Climate, 20(14), 3654–3676. https://doi.org/10.1175/JCLI4203.1 Camargo, S. J., Robertson, A. W., Barnston, A. G., & Ghil, M. (2008). Clustering of eastern North Pacific tropical cyclone tracks: ENSO and MJO effects. Geochemistry, Geophysics, Geosystems, 9(6). https://doi.org/10.1029/2007GC001861 Camargo, S. J., & Barnston, A. G. (2009). Experimental dynamical seasonal forecasts of tropical cyclone activity at IRI. Weather and Forecasting, 24(2), 472–491. https://doi.org/10. 1175/2008WAF2007099.1 Camargo, S. J., Wheeler, M. C., & Sobel, A. H (2009). Diagnosis of the MJO modulation of tropical cyclogenesis using an empirical index. Journal of the Atmospheric Sciences, 66(10), 3061–3074. https://doi.org/10.1175/2009JAS3101.1 Camargo, S. J., Sobel, A. H., Barnston, A. G., & Klotzbach, P. J. (2010). The influence of natural climate variability on tropical cyclones, and seasonal forecasts of tropical cyclone activity. In Chan, J.C.I., & Kepert, J. D. Global perspectives on tropical cyclones (pp. 325–360). https://doi.org/10.1142/9789814293488_0011 Camargo, S. J. (2013). Global and regional aspects of tropical cyclone activity in the CMIP5 models. Journal of Climate, 26(24), 9880–9902. https://doi.org/10.1175/JCLI‐D‐12‐00549.1 Camargo, S. J., & Hsiang, S.M. (2015). Tropical cyclones. In Chaves, M., Ghil, M., & Urrutia‐Fucugauchi, J. Extreme events: Observations, modeling, and economics (pp. 303–342). American Geophysical Union (AGU). https://doi. org/10.1002/9781119157052.ch18 Camargo, S. J., & Wing, A. A. (2016). Tropical cyclones in climate models. WIREs Climate Change, 7, 211–237. Camp, J., Roberts, M., MacLachlan, C., Wallace, E., Hermanson, L., Brookshaw, A., et al. (2015). Seasonal forecasting of tropical storms using the Met Office GloSea5 seasonal forecast system. Quarterly Journal of the Royal Meteorological Society, 141, 2206–2219. Capotondi, A., Wittenberg, A.T., Newman, M., Di Lorenzo, E., Yu, J.‐Y., Braconnot, P., et al. (2014) Understanding ENSO diversity. Bulletin of the American Meteorological Society, 96, 921–938. Caron, L.‐P., Boudreault, M., & Camargo, S. J. (2015). On the variability and predictability of eastern Pacific tropical cyclone activity. Journal of Climate, 28(24), 9678–9696. https://doi.org/10.1175/JCLI‐D‐15‐0377.1 Chan, J.C.L. (1985). Tropical cyclone activity in the northwest Pacific in relation to the El Niño/Southern Oscillation pheno­ menon. Monthly Weather Review, 113, 599–606. Chan, J.C.L., Shi, J., & Lam, C. (1998). Seasonal forecasting of tropical cyclone activity over the western North Pacific and the South China Sea. Weather and Forecasting, 13(4), 997–1004. Chan, J.C.L. (2000). Tropical cyclone activity over the western North Pacific associated with El Niño and La Niña events. Journal of Climate, 13, 2960–2927. Chan, J.C.L., & Yip, C.K.M. (2003). Interannual variations of tropical cyclone size over the western North Pacific. Geo­ physical Research Letters, 30, 2267. https://doi.org/10. 1029/2003GL018522 Chan, J.C.L. (2005). The physics of tropical cyclone motion. Annual Review of Fluid Mechanics, 37, 99–128.

ENSO and Tropical Cyclones  399 Chan, J.C.L., Liu, K.‐S., Xu, M., & Yang Q. (2012). Variations of frequency of landfalling typhoons in East China 1450– 1949. International Journal of Climatology, 32, 1946–1950. Chan, J.C.L. (2015). Tropical cyclones in the western North Pacific. In North, G. R., Pyle, J., & Zhang, F. (Eds.), Encyclopedia of Atmospheric Sciences (2nd ed. Vol. 6, pp. 77–84). Chan, K.T.F., & Chan, J.C.L. (2012). Size and strength of tropical cyclones as inferred from QuikSCAT data. Monthly Weather Review, 140, 811–824. https://doi.org/10.1175/ MWR‐D‐10‐05062.1 Chand, S. S., & Walsh, K.J.E. (2009). Tropical cyclone activity in the Fiji region: Spatial patterns and relationship to large‐ scale circulation. Journal of Climate, 22, 3877–3893. Chand, S. S., & Walsh, K.J.E. (2010). The influence of the Madden‐Julian Oscillation on tropical cyclone activity in the Fiji region. Journal of Climate, 23, 868–886. Chand, S. S., & Walsh, K.J.E. (2011). Influence of ENSO on tropical cyclone intensity in the Fiji region. Journal of Climate, 24, 4096–4108. Chand, S. S., McBride, J. L., Tory, K. J., Wheeler, M. C., & Walsh, K.J.E. (2013a). Impact of different ENSO regimes on southwest Pacific tropical cyclones. Journal of Climate, 26, 600–608. Chand, S. S., Tory, K. J., McBride, J. L., Wheeler, M. C., Dare, R. A., & Walsh, K.J.E. (2013b). The different impact of positive‐ neutral and negative neutral ENSO regimes on Australian tropical cyclones. Journal of Climate, 26, 8008–8016. Chand, S. S., Tory, K. J., Ye, H., & Walsh, K.J.E. (2017). Projected increase in El Niño‐driven tropical cyclone frequency in the Pacific. Nature Climate Change, 7, 123–127. https://doi.org/10.1038/nclimate3181 Chang, P., Ji, L., & Li, H. (1997). A decadal climate variation in the tropical Atlantic Ocean from thermodynamic air–sea interactions. Nature, 385, 516–518. http://dx.doi.org/10. 1038/385516a0 Chen, D., & Lian, T. (2018). Interaction of western‐Pacific tropical cyclones with El Niño diversity. National Science Review, 5(6), 803–804. https://doi.org/10.1093/nsr/nwy126 Chen, G., & Tam, C.‐Y. (2010). Different impacts of two kinds of Pacific Ocean warming on tropical cyclone frequency over the western North Pacific. Geophysical Research Letters, 37, L01803. https://doi.org/10.1029/2009GL041708 Chen, J.‐H., & Lin, S.‐J. (2012). Seasonal predictions of tropical cyclones using a 25‐km‐resolution general circulation model. Journal of Climate, 26(2), 380–398. https://doi.org/10.1175/ JCLI‐D‐12‐00061.1 Chen, T.‐C., Weng, S.‐P., Yamazaki, N., & Kiehne, S. (1998). Interannual variation in the tropical cyclone formation over the western North Pacific. Monthly Weather Review, 126, 1080–1090. Chia, H.‐H., & Ropelewski, C. F. (2002). The interannual variability in the genesis location of tropical cyclones in the Northwest Pacific. Journal of Climate, 15, 2934–2944. Chiang, J.C.H., & Vimont, D. J. (2004). Analogous Pacific and Atlantic meridional modes of tropical atmosphere–ocean variability. Journal of Climate, 17, 4143–4158. https://doi. org/10.1175/JCLI4953.1

Chih, C.‐H., & Wu, C.‐C. (2020). Exploratory analysis of upper ocean heat content and sea surface temperature underlying tropical cyclone rapid intensification in the western North Pacific. Journal of Climate, 33, 1031–1050. Chu, P. S., & Wang, J. (1997). Tropical cyclone occurrences in the vicinity of Hawaii: Are the differences between El Niño and non‐El Niño years significant? Journal of Climate, 10, 2683–2689. Chu, P. S., & Clark, J. D. (1999). Decadal variations of tropical cyclone activity over the Central North Pacific. Bulletin of the American Meteorological Society, 80, 1875–1881. Chu, P.S. (2002). Largescale circulation features associated with decadal variations of tropical cyclone activity over the Central North Pacific. Journal of Climate, 15, 2678–2689. Chu, P. S. (2004). ENSO and tropical cyclone activity. In Murnane, R. J., & Liu, K.‐B. (Eds.), Hurricanes and typhoons: Past, present, and future (pp. 297–332). Columbia University Press, New York. Chu, P. S., & Zhao, X. (2004). Bayesian changepoint analysis of tropical cyclone activity: The Central North Pacific case. Journal of Climate, 17, 4893–4901. Chu, P. S., & Zhao, X. (2007). A Bayesian regression approach for predicting seasonal tropical cyclone activity over the Central North Pacific. Journal of Climate, 20, 4002–4013. Chung, P.‐H., & Li, T. (2015). Characteristics of tropical cyclone genesis in the western North Pacific during the developing and decaying phases of two types of El Niño. Journal of Tropical Meteorology, 21, 14–22. Colbert, A. J., & Soden, B. J. (2012). Climatological variations in North Atlantic tropical cyclone tracks. Journal of Climate, 25, 657–673. https://doi.org/10.1175/JCLI‐D‐11‐00034.1 Collins, J. M., & Mason, I. M. (2000). Local environmental conditions related to seasonal tropical cyclone activity in the northeast Pacific basin. Geophysical Research Letters, 27, 3881–3884. Collins, J. M. (2007). The relationship of ENSO and relative humidity to interannual variations of hurricane frequency in the north‐east Pacific Ocean. Papers of the Applied Geography Conferences, 30, 324–333. Collins, M., An, S.‐I., Cai, W., Ganachaud, A., Guilyardi, E., Jin, F.‐F., et al. (2010). The impact of global warming on the tropical Pacific Ocean and El Niño. Nature Geoscience, 3, 391–397. https://doi.org/10.1038/ngeo868 Corbosiero, K. L., Dickinson, M. J., & Bosart, L. F. (2009). The contribution of eastern North Pacific tropical cyclones to the rainfall climatology of the Southwest Unite States. Monthly Weather Review, 137, 2415–2435. Court, A. (1980). Tropical cyclone effects on California. National Oceanic and Atmospheric Administration Tech. Memo, NWS WR‐159. Daloz, A. S., Camargo, S. J., Kossin, J. P., Emanuel, K., Horn, M., Jonas, J. A., et al. (2015). Cluster analysis of explicitly and downscaled simulated North Atlantic tropical cyclone tracks. Journal of Climate, 28, 1333–1361. https://doi.org/10.1175/JCLI‐D‐13‐ 00646.1 Daloz, A. S., & Camargo, S. J. (2018). Is the poleward migration of tropical cyclone maximum intensity associated with a poleward migration of tropical cyclone genesis? Climate Dynamics, 50, 705–715. https://doi.org/10.1007/s00382‐017‐3636‐7

400  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Dare, R. A., & McBride, J. L. (2011). The threshold sea surface temperature condition for tropical cyclogenesis. Journal of Climate, 24(17), 4570–4576. https://doi.org/10.1175/JCLI‐D‐10‐05006.1 Davis, K., & Zeng, X. (2019). Seasonal prediction of North Atlantic accumulated cyclone energy and major hurricane activity. Weather and Forecasting, 34(1), 221–232. https://doi. org/10.1175/WAF‐D‐18‐0125.1 Defforge, C. L., & Merlis, T. M. (2017). Evaluating the evidence of a global sea surface temperature threshold for tropical cyclone genesis. Journal of Climate, 30(22), 9133–9145. https://doi.org/10.1175/JCLI‐D‐16‐0737.1 DeMaria, M., Mainelli, M., Shay, L. K., Knaff, J. A., & Kaplan, J. (2005). Further improvements to the Statistical Hurricane Intensity Prediction Scheme (SHIPS). Weather and Forecasting, 20(4), 531–543. https://doi.org/10.1175/WAF862.1 Dong, K. (1988). El Niño and tropical cyclone frequency in the Australian region and the northwest Pacific. Australian Meteorological Magazine, 36, 219–255. Dong, K., & Holland, G. J. (1994). A global view of the relationships between ENSO and tropical cyclone frequencies. Journal of Meteorological Research, 8, 19–29. Dowdy, A., & Kuleshov, Y. (2012). An analysis of tropical cyclone occurrence in the Southern Hemisphere derived from a new satellite‐era data set. International Journal of Remote Sensing, 33, 7382–7397. Dowdy, A. J., Qi, L., Jones, D., Ramsay, H., Fawcett, R., & Kuleshov, Y. (2012). Tropical cyclone climatology of the South Pacific Ocean and its relationship to El Niño–Southern Oscillation. Journal of Climate, 25, 6108–6122. Du, Y., Yang, L., & Xie, S.‐P. (2011). Tropical Indian ocean influence on northwest Pacific tropical cyclones in summer following strong El Niño. Journal of Climate, 24(1), 315–322. https://doi.org/10.1175/2010JCLI3890.1 Dvorak, V. F. (1975). Tropical cyclone intensity analysis and forecasting form satellite imagery. Monthly Weather Review, 103, 420–430. Emanuel, K. A. (1995). Sensitivity of tropical cyclones to surface exchange coefficients and a revised steady‐state model incorporating eye dynamics. Journal of the Atmospheric Sciences. https://doi.org/10.1175/1520‐0469(1995)0522.0.CO;2 Emanuel, K. A. (1999). Thermodynamic control of hurricane intensity. Nature, 401(6754), 665–669. https://doi.org/10.1038/44326 Emanuel, K. A. (2005). Divine wind. Oxford University Press. Emanuel, K. A. (2013). Downscaling CMIP5 climate models shows increased tropical cyclone activity over the 21st century. Proceedings of the National Academy of Sciences of the United States of America, 110, 12,219–12,224. https://doi. org/10.1073/pnas.1301293110 Emanuel, K. A. (2015). Effect of upper‐ocean evolution on projected trends in tropical cyclone activity. Journal of Climate, 28, 8165–8170. https://doi.org/10.1175/JCLI‐D‐15‐0401.1 Emanuel, K. A. (2017). Assessing the present and future probability of Hurricane Harvey’s rainfall. Proceedings of the National Academy of Sciences of the United States of America, 114, 12,681–12,684. https://doi.org/10.1073/pnas.1716222114 England, M. H., McGregor S., Spence, P., Meehl, G.A., Timmermann, A., Cai, W., et al. (2014). Recent intensifica-

tion of wind‐driven circulation in the Pacific and the ongoing warming hiatus, Nature Climate Change, 4, 222–227. Englehart, P. J., & Douglas, A. V. (2001). The role of eastern North Pacific tropical storms in rainfall climatology of western Mexico. International Journal of Climatology, 21(11), 1357–1370. https://doi.org/10.1002/joc.637 Evan, A. T., & Camargo, S. J. (2011). A climatology of Arabian Sea cyclonic storms. Journal of Climate, 24, 140–158. Evan, A. T., Kossin, J. P., Chung, C. E. & Ramanathan, V. (2011). Arabian sea tropical cyclones intensified by emissions of black carbon and other aerosols. Nature, 479, 94–97. Evans, J. L., & Allan, R. J. (1992). El‐Niño/Southern Oscillation modification to the structure of the monsoon and tropical cyclone activity in the Australasian region. International Journal of Climatology, 12, 611–623. Fedorov, A. V., Brierley, C. M. & Emanuel, K. (2010). Tropical cyclones and permanent El Niño in the early Pliocene epoch. Nature, 463, 1066–1070. Felton, C. S., Subrahmanyam, B., & Murty, V.S.N. (2013). ENSO‐modulated cyclogenesis over the Bay of Bengal. Journal of Climate, 26(24), 9806–9818. https://doi. org/10.1175/JCLI‐D‐13‐00134.1 Ferreira, R. N., & Schubert, W. H. (1997). Barotropic aspects of ITCZ breakdown. Journal of the Atmospheric Sciences, 54, 261–285. Foltz, G. R., & Balaguru, K. (2016). Prolonged El Niño conditions in 2014–2015 and the rapid intensification of Hurricane Patricia in the eastern Pacific. Geophysical Research Letters, 43, 10,347–10,355. https://doi.org/10.1002/2016GL070274 Frank, W. M., & Roundy, P. E. (2006). The role of tropical waves in tropical cyclogenesis. Monthly Weather Review, 134, 2397–2417. Frank, W. M., & Young, G. S. (2007). The interannual variability of tropical cyclones. Monthly Weather Review, 135, 3587–3598. Fu, D., Chang, P., & Patricola, C. M. (2017). Impact of Central American gap‐winds on intrabasin variability of the eastern North Pacific Tropical cyclones during ENSO. Scientific Reports, 7, 1658. Girishkumar, M. S., Thanga Prakash, V. P., & Ravichandran, M. (2015). Influence of Pacific Decadal Oscillation on the relationship between ENSO and tropical cyclone activity in the Bay of Bengal during October–December. Climate Dynamics, 44, 3469–3479. https://doi.org/10.1007/s00382‐014‐2282‐6 Goldenberg, S. B., & Shapiro, L. J. (1996). Physical mechanisms for the association of El Niño and West African rainfall with Atlantic major hurricane activity. Journal of Climate, 9, 1169–1187. https://doi.org/10.1175/1520‐0442(1996)0092.0.CO;2 Goldenberg, S. B., Landsea, C. W., Mestas‐Nuñez, A. M., & Gray, W. M. (2001). The recent increase in Atlantic hurricane activity: Causes and implications. Science, 293, 474–479. https://doi.org/10.1126/science.1060040 Goni, G., DeMaria, M., Knaff, J., Sampson, C., Ginis, I., Bringas, F., et  al. (2009). Applications of satellite‐derived ocean measurements to tropical cyclone intensity forecasting. Oceanography, 22(3), 190–197. https://doi.org/10.5670/ oceanog.2009.78

ENSO and Tropical Cyclones  401 Gray, W. M. (1979). Hurricanes: Their formation, structure and likely role in the tropical circulation. In D. B. Shaw (Ed.), Meteorology over the tropical oceans (pp. 155–218). Royal Meteorological Society. Gray, W. M. (1984a). Atlantic seasonal hurricane frequency. Part I: El Niño and 30 mb quasi‐biennial oscillation influences. Monthly Weather Review, 112, 1649–1668. https://doi.org/10.1 175/1520‐0493(1984)112%3C1649:ASHFPI%3E2.0.CO;2 Gray, W. M. (1984b). Atlantic seasonal hurricane frequency. Part II: Forecasting its variability. Monthly Weather Review, 112, 1669–1683. https://doi.org/10.1175/1520‐0493(1984)112 2.0.CO;2 Gray, W. M., & Sheaffer, J. D. (1991). El Niño and QBO influences on tropical cyclone activity. In Glanz, M. H., Katz, R. W., & Nicholls, N. (Eds.), Teleconnections linking worldwide anomalies (pp. 257–284). New York: Cambridge University Press. Gray, W. M., Landsea, C. W., Mielke, P. W., & Berry, K. J. (1994). Predicting Atlantic basin seasonal tropical cyclone activity by 1 June. Weather and Forecasting, 9, 103–115. https://doi.org/1 0.1175/1520‐0434(1994)0092.0.CO;2 Grossmann, I., & Morgan, M. G. (2011). Tropical cyclones, climate change, and scientific uncertainty: What do we know, what does it mean, and what should be done? Climatic Change, 108, 543–579. Ha, K.‐J., Yoon, S.‐J., Yun, K.‐S., Kun, J.‐S., Jang, Y.‐S., & Chan, J. C. L. (2012). Dependency of typhoon intensity and genesis locations on El Niño phase and SST shift over the western North Pacific. Theoretical and Applied Climatology, 109, 383–395. Hastings, P. A. (1990). Southern Oscillation influence on tropical cyclone activity in the Australian/south‐west Pacific region. International Journal of Climatology, 10, 291–298. Ho, C. H., Kim, J.‐H., Jeong, J.‐H., Kim, H.‐S., & Chen, D. (2006). Variation of tropical cyclone activity in the South Indian Ocean: El Niño–Southern Oscillation and Madden‐ Julian Oscillation effects. Journal of Geophysical Research, 111, D22191. https://doi.org/10.1029/2006JD007289 Hornyak, T. (2017). Probing the power of Pacific supertyphoons, Eos, 98, https://doi.org/10.1029/2017EO077563 Hsu, P.‐C., Chu, P.‐S., Murakami, H., & Zhao, X. (2014). An abrupt decrease in the late‐season typhoon activity over the western North Pacific. Journal of Climate, 27(11), 4296–4312. https://doi.org/10.1175/JCLI‐D‐13‐00417.1 Hsu, W.‐C., Patricola, C. M., & Chang, P. (2018). The impact of climate model sea surface temperature biases on tropical cyclone simulations. Climate Dynamics, 1–20. https://doi. org/10.1007/s00382‐018‐4577‐5 Hu, C., Zhang, C., Yang, S., Chen, D., & He, S. (2018). Perspective on the northwestward shift of autumn tropical cyclogenesis locations over the western North Pacific from shifting ENSO. Climate Dynamics, 51(7–8), 2455–2465. https://doi.org/10.1007/s00382‐017‐4022‐1 Hu, F., Li, T., Liu, J., & Peng, M. (2018). Cause of Interdecadal change of tropical cyclone controlling parameter in the western North Pacific. Climate Dynamics, 51, 719–732. https:// doi.org/10.1007/s00382‐017‐3951‐z Huang, P., Lin, I.‐I., Chou, C., & Huang, R.‐H. (2015). Change in ocean subsurface environment to suppress tropical cyclone

intensification under global warming. Nature Communi­ cations, 6, 7188. https://doi.org/10.1038/ncomms8188 Huang, H.‐C., Boucharel, J., Lin, I. I., Jin, F.‐F., Lien, C.‐C., & Pun, I.‐F. (2017). Air‐sea fluxes for Hurricane Patricia (2015): Comparison with supertyphoon Haiyan (2013) and under different ENSO conditions. Journal of Geophysical Research: Oceans, 122(8), 6076–6089. https://doi.org/10.1002/ 2017JC012741 Irwin, R. P., & Davis, R. E. (1999). The relationship between the Southern Oscillation index and tropical cyclone tracks in the eastern North Pacific. Geophysical Research Letters, 26, 2251–2254. https://doi.org/10.1029/1999GL900533 Jauregui, E. (2003). Climatology of landfalling hurricanes and tropical storms in Mexico. Atmosfera, 16(4), 193–204. Jiang, X., Zhao, M., & Waliser, D. E. (2012). Modulation of tropical cyclones over the eastern Pacific by intraseasonal variability simulated in an AGCM. Journal of Climate, 25, 6524–6538. Jien, J. Y., Gough, W. A., & Butler, K., (2015). The influence of El Niño‐Southern Oscillation on tropical cyclone activity in the eastern North Pacific basin. Journal of Climate, 28, 2459–2474. Jin, F.‐F. (1997). An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model. Journal of the Atmospheric Sciences, 54, 811–829. Jin, F.‐F., Boucharel, J., & Lin, I.‐I. (2014). Eastern Pacific tropical cyclones intensified by El Niño delivery of subsurface ocean heat. Nature, 516, 82–85. Jin, F.‐F., Boucharel, J., & Lin, I.‐I., (2015). Jin et  al. reply. Nature, 526, E5–E6. https://doi.org/10.1038/nature15547 Johnson, N. C., & Xie, S.‐P. (2010). Changes in the sea surface temperature threshold for tropical convection. Nature Geoscience, 3, 842–845. http://dx.doi.org/10.1038/ngeo1008 Kao, H.‐Y., & Yu, J.‐Y. (2009). Contrasting eastern‐Pacific and central‐Pacific types of ENSO. Journal of Climate, 22(3), 615–632. https://doi.org/10.1175/2008JCLI2309.1 Kaplan, J., & DeMaria, M. (2003). Large‐scale characteristics of rapidly intensifying tropical cyclones in the North Atlantic Basin. Weather and Forecasting, 18(6), 1093–1108. https://doi. org/10.1175/1520‐0434(2003)0182.0.CO;2 Kelly, P., Leung, L. R., Balaguru, K., Xu, W., Mapes, B., & Soden, B. (2018). Shape of Atlantic tropical cyclone tracks and the Indian monsoon. Geophysical Research Letters, 45. https://doi.org/10.1029/2018GL080098 Kemball‐Cook, S., & Wang, B. (2001). Equatorial waves and air‐sea interaction in the boreal summer intraseasonal oscillation. Journal of Climate, 14, 2923–2942. Khouakhi, A., Villarini, G., & Vecchi, G. A. (2017). Contribution of tropical cyclones to rainfall at the global scale. Journal of Climate, 30(1), 359–372. https://doi.org/10.1175/JCLI‐D‐ 16‐0298.1 Kikuchi, K., Wang, B., & Fudeyasu, H. (2009). Genesis of tropical cyclone Nargis revealed by multiple satellite observations. Geophysical Research Letters, 36, L06811. https://doi. org/10.1029/2009GL037296 Kim, H.‐M., Webster, P. J., & Curry, J. A. (2011). A modulation of North Pacific tropical cyclone activity by three phases of ENSO. Journal of Climate, 24, 1839–1849.

402  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Kim, H.‐S., Vecchi, G. A., Knutson, T. R., Anderson, W. G., Delworth, T. L., Rosati, A., et  al. (2014). Tropical cyclone simulation and response to CO2 doubling in the GFDL CM2.5 High‐Resolution Coupled Climate Model. Journal of Climate, 27(21), 8034–8054. https://doi.org/10.1175/ JCLI‐D‐13‐00475.1 Kim, J.‐H., Ho, C.‐H., Kim, H.‐S., Sui, C.‐H., & Park, S. K. (2008). Systematic variation of summertime tropical cyclone activity in the western North Pacific in relation to the Madden–Julian Oscillation. Journal of Climate, 21(6), 1171– 1191. https://doi.org/10.1175/2007JCLI1493.1 Kim, O.‐Y., & Chan, J. C. L. (2018). Cyclone‐track based seasonal prediction for South Pacific tropical cyclone activity using APCC multi‐model ensemble prediction. Climate Dynamics, 51, 3209. https://doi.org/10.1007/s00382‐018 ‐4075‐9 Kimberlain, T. B. (1999). The effects of ENSO on North Pacific and North Atlantic tropical cyclone activity. In Proc. 23rd Conference on Hurricanes and Tropical Meteorology, Dallas, Boston: AMS. Klotzbach, P. J., & Gray, W. M. (2009). 25 years of Atlantic basin seasonal hurricane forecasts. Geophysical Research Letters, 36, L09711. https://doi.org/10.1029/2009GL037580 Klotzbach, P. J. (2010). On the Madden–Julian Oscillation– Atlantic Hurricane Relationship. Journal of Climate, 23(2), 282–293. https://doi.org/10.1175/2009JCLI2978.1 Klotzbach, P. J. (2011). The influence of El Niño–Southern Oscillation and the Atlantic Multi‐decadal Oscillation on Caribbean tropical cyclone activity. Journal of Climate, 24, 721–731. https://doi.org/10.1175/2010JCLI3705.1 Klotzbach, P. J. (2014). The Madden‐Julian oscillation’s impacts on worldwide tropical cyclone activity, Journal of Climate, 27, 2317–2330. Klotzbach, P. J., & Landsea, C. W. (2015). Extremely intense hurricanes: Revisiting Webster et  al. (2005) after 10 years. Journal of Climate, 28, 7621–7629. https://doi.org/10.1175/ JCLI‐D‐15‐0188.1 Klotzbach, P. J., & Oliver, E. C. J. (2015). Modulation of Atlantic basin tropical cyclone activity by the Madden–Julian Oscillation (MJO) from 1905 to 2011. Journal of Climate, 28(1), 204–217. https://doi.org/10.1175/JCLI‐D‐14‐00509.1 Klotzbach, P. J., Saunders, M. A., Bell, G. D., & Blake, E. S. (2017). North Atlantic seasonal hurricane prediction: Underlying science and an evaluation of statistical models. In Wang, S. et  al. (Eds), Climate Extremes: Patterns and Mechanisms, Geophysical Monograph Series (Vol. 226, pp. 315–328). Washington DC: American Geophysical Union. https://doi.org/10.1002/9781119068020.ch19 Klotzbach, P. J., Bowen, S. G., Pielke, R. A., & Bell, M. M. (2018). Continental U.S. hurricane landfall frequency and associated damage: Observations and future risks. Bulletin of the American Meteorological Society, 99, 1359–1376. https:// doi.org/10.1175/BAMS‐D‐17‐0184.1 Knutson, T. R., Sirutis, J. J., Garner, S. T., Held, I. M., & Tuleya, R. E. (2007). Simulation of the recent multidecadal increase of Atlantic hurricane activity using an 18‐km‐grid regional model. Bulletin of the American Meteorological Society, 88(10), 1549– 1565. https://doi.org/10.1175/BAMS‐88‐10‐1549

Knutson, T. R., McBride, J. L., Chan, J., Emanuel, K., Holland, G., Landsea, C., et al. (2010). Tropical cyclones and climate change. Nature Geoscience, 3, 157–163. https://doi. org/10.1038/ngeo779 Knutson, T. R., Sirutis, J. J., Zhao, M., Tuleva, R. E., Bender, M., Vecchi, G. A., et  al. (2015). Global projections of intense tropical cyclone activity for the late twenty‐first century from dynamical downscaling of CMIP5/RCP4.5 Scenarios. Journal of Climate, 28, 7203–7224. Knutson, T. R., Camargo, S. J., Chan, J. C. L., Emanuel, K., Ho, C.‐H., Kossin, J., et  al. (2019a). Tropical cyclones and climate change assessment: Part I. Detection and attribution. Bulletin of the American Meteorological Society, 100, 1987–2007, https://doi.org/10.1175/BAMS-D-18-0189.1. Knutson, T. R., Camargo, S. J., Chan, J. C. L., Emanuel, K., Ho, C.‐H., Kossin, J., et  al. (2019b). Tropical cyclones and climate change assessment: Part II: Projected Response to Anthropogenic Warming. Bulletin of the American Meteorological Society, 101, E303–E322, https://doi. org/10.1175/BAMS-D-18-0194.1. Kosaka, Y., & Xie, S. P. (2013). Recent global warming hiatus tied to equatorial Pacific surface cooling, Nature, 501, 403–407. Kossin, J. P., & Vimont, D. J. (2007). A more general framework for understanding Atlantic hurricane variability and trends. Bulletin of the American Meteorological Society, 88, 1767– 1781. https://doi.org/10.1175/BAMS‐88‐11‐1767 Kossin, J. P., Camargo, S. J., & Sitkowski, M. (2010). Climate modulation of North Atlantic hurricane tracks. Journal of Climate, 23, 3057–3076. https://doi.org/10.1175/2010JCLI3497.1 Kossin, J. P., Olander, T. L., & Knapp, K. R. (2013). Trend analysis with a new global record of tropical cyclone intensity. Journal of Climate, 26, 9960–9976. https://doi.org/10.1175/ JCLI‐D‐13‐00262.1 Kossin, J. P., Emanuel, K. A., & Vecchi, G. A. (2014). The poleward migration of the location of tropical cyclone maximum intensity. Nature, 509, 349–352. Kossin, J. P., Emanuel, K. A., & Camargo, S. J. (2016). Past and projected changes in western North Pacific tropical cyclone exposure. Journal of Climate, 29, 5725–5739. https://doi. org/10.1175/JCLI‐D‐16‐0076.1 Kossin, J. P. (2018). A global slowdown of tropical‐cyclone translation speed. Nature, 558, 104–107. https://doi. org/10.1038/s41586‐018‐0158‐3 Kripalani, R. H., & Kumar, P. (2004). Northeast monsoon rainfall variability over south peninsular India vis‐à‐vis the Indian Ocean dipole mode. International Journal of Climatology, 24, 1267–1282. Kug, J.‐S., Jin, F.‐F., & An, S.‐I. (2009). Two types of El Niño events: Cold tongue El Niño and warm pool El Niño. Journal of Climate, 22, 1499–1515. https://doi.org/10.1175/2008JCLI2624.1 Kuleshov, Y., Qi, L., Fawcett, R., & Jones, D. (2008). On tropical cyclone activity in the Southern Hemisphere: Trends and the ENSO connection. Geophysical Research Letters, 35, L14S08. https://doi.org/10.1029/2007GL032983 Kuleshov, Y., Chane Ming, F., Qi, L., Chouaibou, I., Hoareau, C., & Roux, F. (2009). Tropical cyclone genesis in the Southern Hemisphere and its relationship with the ENSO. Annales Geophysicae, 27, 2523–2538.

ENSO and Tropical Cyclones  403 Lander, M. A. (1994). An exploratory analysis of the relationship between tropical storm formation in the western North Pacific and ENSO. Monthly Weather Review, 122, 636–651. Landsea, C. W., Pielke, R. A., Mestas‐Nuñez, A. M., & Knaff, J. A. (1999). Atlantic basin hurricanes: Indices of climatic change. Climatic Change, 42, 89–129. https://doi.org/10.1023/ A:1005416332322 Landsea, C. W. (2000). El Niño‐Southern Oscillation and the seasonal predictability of tropical cyclones. In Diaz, H. F., & Markgraf, V. (Eds.), El Niño and the Southern Oscillation: Multiscale variability and global and regional impacts (pp. 149–181). New York: Columbia University Press. Landsea, C. W., Harper, B. A., Hoarau, K. & Knaff, J. A. (2006). Can we detect trends in extreme tropical cyclones? Science, 313, 452–454. Landsea, C. W., Vecchi, G. A., Bengtsson, L. & Knutson, T. R. (2010). Impact of duration thresholds on Atlantic tropical cyclone counts. Journal of Climate, 23, 2508–2519. Landsea, C. W., & Franklin, J. L. (2013). Atlantic hurricane database uncertainty and presentation of a new database format. Monthly Weather Review, 141(10), 3576–3592. https://doi.org/10.1175/MWR‐D‐12‐00254.1 Latif, M., Keenlyside, K., & Bader, J. (2007). Tropical sea surface temperature, vertical wind shear, and hurricane development. Geophysical Research Letters, 34, L01710. https://doi.org/10.1029/2006GL027969 Lee, T., & McPhaden, M. J. (2010). Increasing intensity of El Niño in the central‐equatorial Pacific. Geophysical Research Letters, 37, L14603. https://doi.org/10.1029/2010GL044007 Leroy, A., & Wheeler, M. C. (2008). Statistical prediction of weekly tropical cyclone activity in the southern hemisphere. Monthly Weather Review, 136, 3637–3654. L’Heureux, M. L., Takahashi, K., Watkins, A. B., Barnston, A. G., Becker, E. J., DiLiberto, T. E., et al. (2017). Observing and predicting the 2015/2016 El Niño. Bulletin of the American Meteorological Society, 98, 1363–1382. https://doi. org/10.1175/BAMS‐D‐16‐0009.1 Li, H., & Sriver, R. L. (2018). Tropical cyclone activity in the high‐resolution Community Earth System Model and the impact of ocean coupling. Journal of Advances in Modeling Earth Systems, 10, 165–186. Li, R. C. Y. & Zhou W. (2012). Changes in western Pacific tropical cyclones associated with the El Niño–Southern Oscillation cycle. Journal of Climate, 25, 5864–5878. Li, T., Kwon, M., Zhao, M., Kug, J., Luo, J., & Yu, W. (2010). Global warming shifts Pacific tropical cyclone location. Geophysical Research Letters, 37, L21804. https://doi. org/10.1029/2010GL045124 Li, T. (2012). Synoptic and climatic aspects of tropical cyclogenesis in western North Pacific. In K. Oouchi & H. Fudeyasu (Eds.), Cyclones: Formation, triggers, and control (pp. 61–94). Hauppage, NY: Nova Science Publishers, Inc. Li, T., Wang, B., Wu, B., Zhou, T., Chang, C.‐P., & Zhang, R. (2017). Theories on formation of an anomalous anticyclone in western North Pacific during El Niño: A review. Journal of Meteorological Research, 31, 987–1006. https://doi. org/10.1007/s13351‐017‐7147‐6

Li, Z., Yu, W., Li, T., Murty, V. S. N., & Tangang, F. (2013). Bimodal character of cyclone climatology in the Bay of Bengal modulated by monsoon seasonal cycle. Journal of Climate, 26(3), 1033–1046. https://doi.org/10.1175/JCLI‐D‐ 11‐00627.1 Li, Z., Li, T., Yu, W., Li, K., & Liu, Y. (2016). What controls the interannual variation of tropical cyclone genesis frequency over Bay of Bengal in the post‐monsoon peak season? Atmospheric Science Letters, 17(2), 148–154. https://doi. org/10.1002/asl.636 Lian, T., Chen, D., Tang, Y., Liu, X., Feng, J., & Zhou, L. (2018). Linkage between westerly wind bursts and tropical cyclones. Geophysical Research Letters, 45(20), 11,411– 431,438. https://doi.org/10.1029/2018GL079745 Liebmann, H., Hendon, H., & Glick, J. D. (1994). The relationship between tropical cyclones of the western Pacific and Indian Oceans and the Madden‐Julian Oscillation. Journal of the Meteorological Society of Japan. Ser. II, 72(3), 401–412. https://doi.org/10.2151/jmsj1965.72.3_401 Lin, I. I., Wu, C.‐C., Emanuel, K. A., Lee, I.‐H., Wu, C.‐R., & Pun, I.‐F. (2005). The interaction of Supertyphoon Maemi (2003) with a warm ocean eddy. Monthly Weather Review, 133(9), 2635–2649. https://doi.org/10.1175/MWR3005.1 Lin, I. I., Wu, C.‐C., Pun, I.‐F., & Ko, D.‐S. (2008). Upper‐ ocean thermal structure and the western North Pacific category 5 typhoons. Part I: Ocean features and the category 5 typhoons’ intensification. Monthly Weather Review, 136(9), 3288–3306. https://doi.org/10.1175/2008MWR 2277.1 Lin, I. I., Chen, C. H., Pun, I. F., Liu, W. T., & Wu, C. C. (2009). Warm ocean anomaly, air sea fluxes, and the rapid intensification of tropical cyclone Nargis (2008). Geophysical Research Letters, 36(3), 1–5. https://doi.org/10.1029/ 2008GL035815 Lin, I. I., Black, P., Price, J. F., Yang, C. Y., Chen, S. S., Lien, C. C., et al. (2013a). An ocean coupling potential intensity index for tropical cyclones. Geophysical Research Letters, 40(9), 1878–1882. https://doi.org/10.1002/grl.50091 Lin, I. I., Goni, G. J., Knaff, J. A., Forbes, C., & Ali, M. M. (2013b). Ocean heat content for tropical cyclone intensity forecasting and its impact on storm surge. Natural Hazards, 66(3), 1481–1500. https://doi.org/10.1007/s11069‐012‐0214‐5 Lin, I. I., Pun, I. F., & Lien, C. C. (2014). “Category‐6” supertyphoon Haiyan in global warming hiatus: Contribution from subsurface ocean warming. Geophysical Research Letters, 41(23), 8547–8553. https://doi.org/10.1002/2014GL061281 Lin I. I., & Chan, J. C. L. (2015). Recent decrease in typhoon destructive potential and global warming implications. Nature communications, 6, 7182. https://doi.org/10.1038/ ncomms8182 Liu, K. S., & Chan, J. C. L. (2012). Interannual variation of Southern Hemisphere tropical cyclone activity and seasonal forecast of tropical cyclone number in the Australian region. International Journal of Climatology, 32, 190–202. Liu, K. S., & Chan, J. C. L. (2013). Inactive period of western North Pacific tropical cyclone activity in 1998–2011. Journal of Climate, 26(8), 2614–2630. https://doi.org/10.1175/JCLI‐D‐12‐ 00053.1

404  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Lyon, B., & Camargo, S. J. (2009). The seasonally‐varying influence of ENSO on rainfall and tropical cyclone activity in the Philippines. Climate Dynamics, 32(1), 125–141. https:// doi.org/10.1007/s00382‐008‐0380‐z Madden, R. A., & Julian, P. R. (1972). Description of global‐ scale circulation cells in the tropics with a 40–50 day period. Journal of the Atmospheric Sciences, 29(6), 1109–1123. https://doi.org/10.1175/1520‐0469(1972)0292.0.CO;2 Madden, R. A., & Julian, P. R. (1994). Observations of the 40–50 day tropical oscillation—A review. Monthly Weather Review, 122, 814–837. Magee, A. D., Verdon‐Kidd, D. C., Diamond, H. J., & Kiem, A.  S. (2017). Influence of ENSO, ENSO Modoki, and the IPO on tropical cyclogenesis: A spatial analysis of the southwest Pacific region. International Journal of Climatology, 37, 1118–1137. Mahala, B. K., Nayak, B. K., & Mohanty, P. K. (2015). Impact of ENSO and IDO on tropical cyclone activity in the Bay of Bengal. Natural Hazards, 75, 1105–1125. https://doi. org/10.1007/s11069‐014‐1360‐8 Maloney, E. D., & Hartmann, D. L. (1998). Frictional moisture convergence in a composite life cycle of the Madden–Julian oscillation. Journal of Climate, 11, 2387–2403. Maloney, E. D., & Hartmann, D. L. (2000). Modulation of Eastern North Pacific hurricanes by the Madden–Julian Oscillation. Journal of Climate, 13, 1451–1460. Mantua, N. J., Hare, S. R., Zhang, Y., Wallace, J. M., & Francis, R. C. (1997). A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society, 78(6), 1069–1080. https://doi.org/10. 1175/1520‐0477(1997)0782.0.CO;2 Mantua, N.J. (2002). The Pacific decadal oscillation. Journal of Oceanography, 58, 35–44. Martinez‐Sanchez, J. N., & Cavazos, T. (2014). Eastern tropical Pacific hurricane variability and landfalls on Mexican coasts. Climate Research, 58(3), 221–234. https://doi.org/10.3354/ cr01192 McDonnell, K. A., & Holbrook, N. J. (2004). A Poisson regression model of tropical cyclogenesis for the Australian– southwest Pacific Ocean region. Weather and Forecasting, 19, 440–455. McPhaden, M. J., Foltz, G. R., Lee, T., Murty, V.S.N., Ravichandran, M., Vecchi, G. A., et  al. (2009). Ocean‐ atmosphere interactions during Cyclone Nargis. Eos, Transactions American Geophysical Union, 90(7), 53–54. https://doi.org/10.1029/2009EO070001 Mo, K. C. (2000). The association between intraseasonal oscillations and tropical storms in the Atlantic basin. Monthly Weather Review, 128(12), 4097–4107. https://doi.org/10.1175 /1520‐0493(2000)1292.0.CO;2 Molinari, J., Knight, D., Dickinson, M., Vollaro, D., & Skubis, S. (1997). Potential vorticity, easterly waves, and eastern Pacific tropical cyclo‐genesis. Monthly Weather Review, 125, 2699–2708. Molinari, J., & Vollaro, D. (2000). Planetary‐ and synoptic‐scale influences on eastern Pacific cyclogenesis. Monthly Weather Review, 128, 3296–3307.

Molinari, J., Vollaro, D., Skubis, S., & Dickinson, M. (2000). Origins and mechanisms of eastern Pacific cyclogenesis: A case study. Monthly Weather Review, 128, 125–139. Moon, I‐J, Kim, S.‐H., Klotzbach, P., & Chan, J. C. L. (2015). Roles of interbasin frequency changes in the poleward shift of the maximum intensity location of tropical cyclones. Environmental Research Letters, 10, 104004. https://doi.org/1 0.1088/1748‐9326/10/10/104004 Moon, J.‐Y., Wang, B., Lee, S.‐S., & Ha, K.‐J. (2018). An intraseasonal genesis potential index for tropical cyclones during Northern Hemisphere summer, Journal of Climate, 31(22), 9055–9071. https://doi.org/10.1175/JCLI‐D‐18‐0515.1 Murakami, H., & Wang, B. (2010). Future change of North Atlantic tropical cyclone tracks: Projection by a 20‐km‐mesh global atmospheric model. Journal of Climate, 23(10), 2699– 2721. https://doi.org/10.1175/2010JCLI3338.1 Murakami, H., Li, T., & Peng, M. (2013a). Changes to environmental parameters that control tropical cyclone genesis under global warming. Geophysical Research Letters, 40(10), 2265– 2270. https://doi.org/10.1002/grl.50393 Murakami, H., Wang, B., Li, T., & Kitoh, A. (2013b). Projected increase in tropical cyclones near Hawaii. Nature Climate Change, 3, 749–754. Murakami, H., Vecchi, G. A., Delworth, T. L., Paffendorf, K., Jia, L., Gudgel, R., et al. (2015). Investigating the influence of forcing and natural variability on the 2014 Hawaiian hurricane season. Bulletin of the American Meteorological Society, 96, S115–S119. https://doi.org/10.1175/BAMS‐D‐ 15‐00119.1 Murakami, H., Vecchi, G. A., Delworth, T. L., Wittenberg, A. T., Underwood, S., Gudgel, R., et al. (2017a). Dominant role of subtropical Pacific warming in extreme eastern Pacific hurricane seasons: 2015 and the future. Journal of Climate, 30, 243–264. https://doi.org/10.1175/JCLI‐D‐16‐0424.1 Murakami, H., Vecchi, G. A., & Underwood, S. (2017b). Increasing frequency of extremely severe cyclonic storms over the Arabian Sea. Nature Climate Change, 7, 885–889. Murakami, H., Levin, E., Delworth, T. L., Gudgel, R., & Hsu,  P.‐C. (2018). Dominant effect of relative tropical Atlantic warming on major hurricane occurrence. Science, early online. https://doi.org/10.1126/science.aat6711 Nakamura, J., Camargo, S. J., Sobel, A. H., Henderson, N., Emanuel, K. A., Kumar, A., et  al. (2017). Western North Pacific tropical cyclone model tracks in present and future climates. Journal of Geophysical Research, 122, 9721–9744. https://doi.org/10.1002/2017JD027007 Neumann, C. J. (1993). Global overview in Global guide to tropical cyclone forecasting, edited by G. J. Holland, 1.1– 1.56. Technical Document WMO/TC‐No. 560, Report No. TCP‐31. Geneva: World Meteorological Organization. Ng, E.K.W. & Chan, J.C.L. (2012). Interannual variation of tropical cyclone activity over the north Indian Ocean. International Journal of Climatology, 32, 819–830. Nicholls, N. (1979). A possible method for predicting seasonal tropical cyclone activity in the Australian region. Monthly Weather Review, 107, 1221–1224. Paek, H., Yu, J.‐Y., & Qian, C. (2017). Why were the 2015/2016 and 1997/1998 extreme El Niños different? Geophysical

ENSO and Tropical Cyclones  405 Research Letters, 44(4), 1848–1856. https://doi.org/10.1002/ 2016GL071515 Patricola, C. M., Saravanan, R., & Chang, P. (2014). The impact of the El Niño‐Southern Oscillation and Atlantic Meridional Mode on seasonal Atlantic tropical cyclone activity. Journal of Climate, 27, 5311–5328. https://doi.org/10.1175/JCLI‐D‐13‐ 00687.1 Patricola, C. M., Chang, P., & Saravanan, R. (2016). Degree of simulated suppression of Atlantic tropical cyclones modulated by flavour of El Niño. Nature Geoscience, 9, 155. http:// dx.doi.org/10.1038/ngeo2624 Patricola, C. M., Saravanan, R., & Chang, P. (2017). A teleconnection between Atlantic seas surface temperature and eastern and central North Pacific tropical cyclones. Geophysical Research Letters, 44, 1167–1174. Patricola, C. M., Camargo, S. J., Klotzbach, P. J., Saravanan, R., & Chang, P. (2018). The influence of ENSO flavors on western North Pacific tropical cyclone activity. Journal of Climate, 31, 5395–5416. Patricola, C. M., & Wehner, M. F. (2018). Anthropogenic influences on major tropical cyclone events. Nature, 563(7731), 339–346. https://doi.org/10.1038/s41586‐018‐0673‐2 Peduzzi, P. B., Chatenoux, B., Dao, H., DeBono, A., Herold, C., Kossin, J., et al. (2012). Global trends in tropical cyclone risk. Nature Climate Change, 2, 289–294. Pielke Jr., R. A., & Landsea, C. W. (1999). La Niña, El Niño and Atlantic hurricane damages in the United States. Bulletin of the American Meteorological Society, 80, 2027–2034. https://doi.org/10.1175/1520‐0477(1999)0802.0.CO;2 Pielke Jr., R. A., Gratz, J., Landsea, C. W., Collins, D., Saunders, M. A., & Musulin, R. (2008). Normalized hurricane damages in the United States: 1900–2005. Natural Hazards Review, 9, 29–42. https://doi.org/10.1061/(ASCE)1527‐6988(2008)9:1(29) Power, S., Delage, F., Chung, C., Kociuba, G., & Keay, K. (2013). Robust twenty‐first‐century projections of El Niño and related precipitation variability. Nature, 502, 541–545. https://doi.org/10.1038/nature12580 Price, J. F. (2009). Metrics of hurricane‐ocean interaction: Vertically‐integrated or vertically‐averaged ocean temperature? Ocean Science, 5(3), 351–368. https://doi.org/10.5194/ os‐5‐351‐2009 Raga, G. B., Bracamontes‐Ceballos, L. M.,Farfan, & Romero‐ Centeno, R. (2013). Landfalling tropical cyclones on the Pacific coast of Mexico: 1850–2010. Atmósfera, 26(2), 209–220. Ramage, C. S., & Hori, A. M. (1981). Meteorological aspects of El Niño. Monthly Weather Review, 109, 1827–1835. Ramsay, H. A., Leslie, L. M., Lamb, P. J., Richman, M. B., & Leplastrier, M. (2008). Interannual variability of tropical cyclones in the Australian region: Role of large‐scale environment. Journal of Climate, 21, 1083–1103. Ramsay, H. A., & Sobel, A. H. (2011). Effects of relative and absolute sea surface temperature on tropical cyclone potential intensity using a single‐column model. Journal of Climate, 24, 183–193. https://doi.org/10.1175/2010JCLI3690.1 Ramsay, H. A., Camargo, S. J., & Kim, D. (2012). Cluster analysis of tropical cyclone tracks in the Southern Hemisphere. Climate Dynamics, 39, 897–917.

Ramsay, H. A., Richman, M. B., & Leslie, L. M. (2014). Seasonal tropical cyclone predictions using optimized combinations of ENSO Regions: Application to the Coral Sea Basin. Journal of Climate, 27, 8527–8542. Rappaport, E. N., Avila, L. A., Lawrence, M. B., Mayfield, M., & Pasch, R. J. (1998). Eastern North Pacific hurricane season of 1995. Monthly Weather Review, 126, 1152–1162. Ren, H. L., & Jin, F.‐F. (2013). Recharge oscillator mechanisms in two types of ENSO. Journal of Climate, 26, 6506–6523. Revell, C. G., & Goulter, S. W. (1986a). South Pacific tropical cyclones and the Southern Oscillation. Monthly Weather Review, 114, 1138–1145. Revell, C. G., & Goulter, S. W. (1986b). Lagged relationships between Southern Oscillation and numbers of tropical cyclones in the South Pacific region. Monthly Weather Review, 114, 2669–2670. Richter, I. (2015). Climate model biases in the eastern tropical oceans: Causes, impacts and ways forward. WIREs Climate Change, 6(3), 345–358. https://doi.org/10.1002/wcc.338 Risser, M. D., & Wehner, M. F. (2017). Attributable human‐ induced changes in the likelihood and magnitude of the observed extreme precipitation during Hurricane Harvey. Geophysical Research Letters, 44, 12,457–12,464. https://doi. org/10.1002/2017GL075888 Ritchie, E. A., & Holland, G. J. (1999). Large‐scale patterns associated with tropical cyclogenesis in the western Pacific. Monthly Weather Review, 127, 2027–2043. Ritchie, E. A., Wood, K. M., Gutzler, D. S., & White, S. R. (2011). The influence of eastern Pacific tropical cyclone remnants on the southwestern United States. Monthly Weather Review, 139, 192–210. Roberts, M. J., Vidale, P. L., Mizielinski, M. S., Demory, M.‐E., Schiemann, R., Strachan, J., et al. (2015). Tropical cyclones in the UPSCALE ensemble of high‐resolution global climate models. Journal of Climate, 28, 574–596. Rogers, R. F., Aberson, S., Bell, M. M., Cecil, D. J., Doyle, J. D., Kimberlain, T. B., et al. (2017). Rewriting the tropical record books: The extraordinary intensification of Hurricane Patricia (2015). Bulletin of the American Meteorological Society, 98(10), 2091–2112. https://doi.org/10.1175/BAMS‐D‐ 16‐0039.1 Saji, N. H., Goswami, B. N., Vinayachandran, P. N., & Yamagata, T. (1999). A dipole mode in the tropical Indian Ocean. Nature, 401(6751), 360–363. https://doi.org/10.1038/ 43854 Santoso, A., Mcphaden, M. J., & Cai, W. (2017). The defining characteristics of ENSO extremes and the strong 2015/2016 El Niño. Reviews of Geophysics, 55(4), 1079–1129. https:// doi.org/10.1002/2017RG000560 Santoso, A., Hendon, H., Watkins, A., Power, S., Dommenget, D., England, M. H., et al. (2019). Dynamics and predictability of El Niño–Southern Oscillation: An Australian perspective on progress and challenges. Bulletin of the American Mete­ orological Society, 100(3), 403–420. https://doi.org/10.1175/ BAMS‐D‐18‐0057.1 Schreck, C. J., & Molinari, J. (2011). Tropical cyclogenesis associated with Kelvin waves and the Madden‐Julian oscillation. Monthly Weather Review, 139, 2732–2734.

406  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Schreck, C. J., Molinari, J., & Aiyyer, A. (2012). A global view of equatorial waves and tropical cyclogenesis. Monthly Weather Review, 140, 774–788. Scoccimarro, E., Gualdi, S., Villarini, G., Vecchi, G. A., Zhao,  M., Walsh, K., et  al. (2014). Intense precipitation events associated with landfalling tropical cyclones in response to a warmer climate and increased CO2. Journal of Climate, 27, 4642–4654. https://doi.org/10.1175/ JCLI‐D‐14‐00065.1 Servain, J., Wainer, I., McCreary, J. P., & Dessier, A. (1999). Relationship between the equatorial and meridional modes of climatic variability in the tropical Atlantic. Geophysical Research Letters, 26, 485–488. https://doi.org/10.1029/ 1999GL900014 Sharmila, S., & Walsh, K.J.E. (2018). Recent poleward shift of tropical cyclone formation linked to Hadley cell expansion. Nature Climate Change, 8, 730–736. https://doi.org/10.1038/ s41558‐018‐0227‐5 Shay, L. K., Goni, G. J., & Black, P. G. (2000). Effects of a warm oceanic feature on Hurricane Opal. Monthly Weather Review, 128(5), 1366–1383. https://doi.org/10.1175/1520‐049 3(2000)1282.0.CO;2 Shay, L. K., & Brewster, J. K. (2010). Oceanic heat content variability in the eastern Pacific Ocean for hurricane intensity forecasting. Monthly Weather Review, 138(6), 2110–2131. https://doi.org/10.1175/2010MWR3189.1 Singh, O. P., Khan, T.M.A., & Rahman, M. S. (2000). Changes in the frequency of tropical cyclones over the North Indian Ocean. Meteorology and Atmospheric Physics, 75, 11–20. Singh, O. P., Khan, T.M.A., & Rahman, M. S. (2001). Has the frequency of intense tropical cyclones increased in the north Indian Ocean? Current Science, 80, 575–580. https://doi. org/10.2307/24104250 Singh, O. P. (2008). Indian Ocean dipole mode and tropical cyclone frequency. Current Science, 94, 29–31. Sobel, A. H., & Camargo, S. J. (2005). Influence of western North Pacific tropical cyclones on their large‐scale environment. Journal of the Atmospheric Sciences, 62(9), 3396–3407. https://doi.org/10.1175/JAS3539.1 Sobel, A. H., Camargo, S. J., Barnston, A. G., & Tippett, M. K. (2016a). Northern Hemisphere tropical cyclones during the quasi–El Niño of late 2014. Natural Hazards, 83, 1717–1729. https://doi.org/10.1007/s11069‐016‐2389‐7 Sobel, A. H., Camargo, S. J., Hall, T. M., Lee, C.‐Y., Tippett, M. K., & Wing, A. A. (2016b). Human influence on tropical cyclone intensity. Science, 353, 242–246. https://doi.org/10.1126/ science.aaf6574 Tang, B. H., & Neelin, J. D. (2004). ENSO influence on Atlantic hurricanes via tropospheric warming. Geophysical Research Letters, 31, L24204. https://doi.org/10.1029/2004 GL021072 Tartaglione, C. A, Smith, S. R., & O’Brien, J. J. (2003). ENSO impact on hurricane landfall probabilities for the Caribbean. Journal of Climate, 16, 2925–2931. https://doi.org/10.1175/15 20‐0442(2003)0162.0.CO;2 Timmermann, A., An, S.‐I., Kug, J.‐S., Jin, F.‐F., Cai, W., Capotondi, A., et  al. (2018). El Niño–Southern Oscillation complexity. Nature, 559, 535–545.

Toma, V. & Webster, P. J. (2010). Oscillations of the intertropical convergence zone and the genesis of easterly waves. Part I: Diagnostics and theory. Climate Dynamics, 34, 587–604. Tory, K., Chand, S. S., McBride, J. L., Ye, H., & Dare, R. A. (2013). Projected changes in late 21st century tropical cyclone frequency in thirteen coupled climate models from the Coupled Model Intercomparison Project Phase 5. Journal of Climate, 26, 9946–9959. Trenberth, K. E., Cheng, L., Jacobs, P., Zhang, Y., & Fasullo, J. (2018). Hurricane Harvey links to ocean heat content and climate change adaptation. Earth’s Future, 6, 730–744. https:// doi.org/10.1029/2018EF000825 van Oldenborgh, G. J., van derWiel, K., Sebastian, A., Singh, R., Arrighi, J., Otto, F., et al. (2017). Attribution of extreme rainfall from Hurricane Harvey, August 2017. Environmental Research Letters, 12, 124009. https://doi. org/10.1088/1748‐9326/aa9ef2 Vecchi, G. A., Swanson, K. L., & Soden, B. J. (2007). Whither hurricane activity? Science, 322, 687–689. https://doi.org/10. 1126/science.1164396 Vecchi, G. A., & Wittenberg, A. T. (2010). El Niño and our future climate: Where do we stand? Wiley Interdisciplinary Reviews: Climate Change, 1(2), 260–270. https://doi. org/10.1002/wcc.33 Vecchi, G. A., Zhao, M., Wang, H., Villarini, G., Rosati, A., Kumar, A., et al. (2011). Statistical–dynamical predictions of seasonal North Atlantic hurricane activity. Monthly Weather Review, 139(4), 1070–1082. https://doi.org/10.1175/ 2010MWR3499.1 Vimont, D. J., & Kossin, J. P. (2007). The Atlantic meridional mode and hurricane activity. Geophysical Research Letters, 34, L07709. https://doi.org/10.1029/2007GL029683 Vincent, D. G. (1994). The South Pacific Convergence Zone (SPCZ): A review. Monthly Weather Review, 122, 1949–1970. Vincent, D. G., & Fink, A. H. (2001). Tropical cyclone ­environments over the northeastern and northwestern Pacific based on ERA15 analyses. Monthly Weather Review, 129, 1928–1948. Vincent, E. M., Lengaigne, M., Menkes, C. E., Jourdain, N. C., Marchesiello, P., & Madec, G. (2011). Interannual variability of the South Pacific convergence zone and implications for tropical cyclone genesis. Climate Dynamics, 36, 1881–1896. Vincent, E. M., Emanuel, K. A., Lengaigne, M., Vialard, J., & Madec, G. (2014). Influence of upper‐ocean stratification interannual variability on tropical cyclones. Journal of Advances in Modeling Earth Systems, 6, 680–699. https://doi. org/10.1002/2014MS000327 Vitart, F., Huddleston, M. R., Déqué, M., Peake, D., Palmer, T.  N., Stockdale, T. N., et  al. (2007). Dynamically‐based seasonal forecasts of Atlantic tropical storm activity issued in June by EUROSIP. Geophysical Research Letters, 34(16). https://doi.org/10.1029/2007GL030740 Wahiduzzaman, M., Oliver, E.C.J., Klotzbach, P. J., Wotherspoon, S. J., & Holbrook, N. J. (2019). A statistical seasonal forecast model of North Indian Ocean tropical cyclones using the quasi‐biennial oscillation. International Journal of Climatology, 39(2), 934–952. https://doi. org/10.1002/joc.5853

ENSO and Tropical Cyclones  407 Walsh, K.J.E., McBride, J. L., Klotzbach, P. J., Balachandran, S., Camargo, S. J., Holland, G., et al. (2016). Tropical cyclones and climate change. Wiley Interdisciplinary Reviews: Climate Change, 7, 65–89. https://doi.org/10.1002/wcc.371 Wang, B., Wu, R., & Fu, X. (2000). Pacific–East Asian teleconnection: How does ENSO affect East Asian climate? Journal of Climate, 13, 1517–1536. Wang, B., & Chan, J.C.L. (2002). How strong ENSO events affect tropical storm activity over the western North Pacific. Journal of Climate, 15, 1643–1658. Wang, B., & Zhou, X. (2008). Climate variation and prediction of rapid intensification in tropical cyclones in the western North Pacific. Meteorology and Atmospheric Physics. 99(1– 2), 1–16. https://doi.org/10.1007/s00703‐006‐0238‐z Wang, B., Yang, Y., Ding, Q., Murakami, H., & Huang, F. (2010). Climate control of the global tropical storm days (1965–2008). Geophysical Research Letters, 37, L07704. https://doi.org/10.1029/2010GL042487 Wang, B., Xu, S., & Wu, L. (2012). Intensified Arabian Sea tropical storms. Nature, 489(7416). https://doi.org/10.1038/ nature11470 Wang, B., & Moon, J.‐Y. (2017). An anomalous genesis potential index for MJO modulation of tropical cyclones. Journal of Climate, 30(11), 4021–4035. Wang, C., & Lee, S. ‐K. (2009). Co‐variability of tropical cyclones in the north Atlantic and the eastern North Pacific. Geophysical Research Letters, 36, L24702. Wang, C., Li, C., Mu, M. & Duan, W. (2013). Seasonal modulations of different impacts of two types of ENSO events on tropical cyclone activity in the western North Pacific. Climate Dynamics, 40, 2887–2902. https://doi.org/10.1007/s00382‐ 012‐1434‐9 Wang, S.‐Y. S., Zhao, L., Yoon, J.‐H., Klotzbach, P., & Gillies, R. R. (2018). Quantitative attribution of climate effects on Hurricane Harvey’s extreme rainfall in Texas. Environmental Research Letters, 13, 054014. https://doi.org/10.1088/ 1748‐9326/aabb85 Wang, X., Wang, C., Zhang, L., & Wang, X. (2015). Multidecadal variability of tropical cyclone rapid intensification in the Western North Pacific. Journal of Climate, 28, 3806‐3820. Wang, X., & Liu, H. (2016). PDO modulation of ENSO effect on tropical cyclone rapid intensification in the western North Pacific. Climate Dynamics, 46, 15–28. Webster, P. J. (2008). Myanmar’s deadly daffodil. Nature Geoscience, 1(8), 488–490. https://doi.org/10.1038/ngeo257 Wehner, M., Prabhat, Reed, K. A., Stone, D., Collins, W. D., & Bacmeister, J. (2015). Resolution dependence on future tropical cyclone projections of CAM5.1 in the U.S. CLIVAR Hurricane Working Group idealized configurations. Journal of Climate, 28, 3905–3925. Werner, A., & Holbrook, N. J. (2011). A Bayesian forecast model of Australian region tropical cyclone formation. Journal of Climate, 24, 6114–6131. Whitney, L. D., & Hobgood, J. (1997). The relationship between sea surface temperatures and maximum intensities of tropical cyclones in the eastern North Pacific Ocean. Journal of Climate, 10, 2921–2930.

Wijnands, J. S., Qian, G., Shelton, K. L., Fawcett, R. J. B., Chan,  J.  C.  L., & Kuleshov, Y. (2015). Seasonal forecasting of tropical cyclone activity in the Australian and the South Pacific Ocean regions. Mathematics of Climate and Weather Forecasting, 1, 21–42. Williams, I. N., Pierrehumbert, R. T., & Huber, M. (2009). Global warming, convective threshold and false thermostats. Geophysical Research Letters, 36, L05702. https://doi. org/10.1029/2009GL039849 Williams, I. N., & Patricola, C. M. (2018). Diversity of ENSO events unified by convective threshold sea surface temperature: A nonlinear ENSO Index. Geophysical Research Letters, 45, 9236–9244. Wood, K. M., & Ritchie, E. A. (2013). An updated climatology of tropical cyclone impacts on the southwestern United States. Monthly Weather Review, 141, 4322–4336. Woodruff, J. D., Irish, J. L., & Camargo, S. J. (2013). Coastal flooding by tropical cyclones and sea‐level rise. Nature, 504, 44–52. https://doi.org/10.1038/nature12855 Wu, C.‐C., Lee, C.‐Y., & Lin, I. I. (2007). The effect of the ocean  eddy on tropical cyclone intensity. Journal of the Atmospheric Sciences, 64(10), 3562–3578. https://doi. org/10.1175/JAS4051.1 Wu, G., & Lau, N.‐C. (1992). A GCM simulation of the relationship between tropical storm formation and ENSO. Monthly Weather Review, 120, 958–977. Wu, L., Zhang, H., Chen, J.‐M., & Feng, T. (2018) Impact of two types of El Niño on tropical cyclones over the western North Pacific: Sensitivity to location and intensity of Pacific warming. Journal of Climate, 31, 1725–1742. https://doi. org/10.1175/JCLI‐D‐17‐0298.1 Wu, M. C., Chang, W. L., & Leung, W. M. (2004). Impacts of El Niño–Southern Oscillation events on tropical cyclone landfalling activity in the western North Pacific. Journal of Climate, 17(6), 1419–1428. https://doi.org/10.1175/1520‐0442 (2004)0172.0.CO;2 Wu, P., & Chu, P.‐S. (2007). Characteristics of tropical cyclone activity over the eastern North Pacific: The extremely active 1992 and the inactive 1977. Tellus A: Dynamic Meteorology and Oceanography, 59, 444–454. Wu, Y.‐K., Hong, C.‐C., & Chen, C.‐T. (2018). Distinct effects of the two strong El Niño Events in 2015–2016 and 1997– 1998 on the western North Pacific Monsoon and tropical cyclone activity: Role of subtropical eastern North Pacific warm SSTA. Journal of Geophysical Research: Oceans, 123(5), 3603–3618. https://doi.org/10.1002/2018JC013798 Xiang, B., & Wang, B. (2013). Mechanisms for the advanced Asian Summer Monsoon onset since the mid‐to‐late 1990s. Journal of Climate, 26, 1993–2009. https://doi.org/10.1175/ JCLI‐D‐12‐00445.1 Yang, L., Chen, S., Wang, C., Wang, D., & Wang, X. (2018). Potential impact of the Pacific Decadal Oscillation and sea surface temperature in the tropical Indian Ocean–Western Pacific on the variability of typhoon landfall on the China coast. Climate Dynamics, 51, 2695–2705. Yeh, S.‐W., Kug, J.‐S., Dewitte, B., Kwon, M.‐H., Kirtman, B.  P., & Jin, F.‐F. (2009). El Niño in a changing climate. Nature, 461, 511–514.

408  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Yoo, S.‐H., Yang, S., & Ho, C.‐H. (2006). Variability of the Indian Ocean sea surface temperature and its impacts on Asian‐Australian monsoon climate. Journal of Geophysical Research, 111, D03108. https://doi.org/10.1029/2005 JD006001 Yoshida, R., & Ishikawa, H. (2013). Environmental factors contributing to tropical cyclone genesis over the western North Pacific. Monthly Weather Review, 141, 451–467. Yu, J.‐H., Li, T., Tan, Z., & Zhu, Z. (2016). Effects of tropical North Atlantic SST on tropical cyclone genesis in the western North Pacific. Climate Dynamics, 46, 865–877. Yuan, J., Wang, D., Wan, Q., & Liu, C. (2007). A 28‐year climatological analysis of size parameters for Northwestern Pacific tropical cyclones. Advances in Atmospheric Sciences, 24, 24–34. Zarzycki, C. M. (2016). Tropical cyclone intensity errors associated with lack of two‐way ocean coupling in high‐resolution global simulations. Journal of Climate, 29(23), 8589–8610. https://doi.org/10.1175/JCLI‐D‐16‐0273.1 Zehnder, J. A. (1991). The interaction of planetary­scale tropical easterly waves with topography: A mechanism for the initiation of tropical cyclones. Journal of the Atmospheric Sciences, 48, 1217–1230. Zehnder, J. A., Powell, D. M., & Ropp, D. L. (1999). The interaction of easterly waves, orography, and the intertropical convergence zone in the genesis of eastern Pacific tropical cyclones. Monthly Weather Review, 127, 1566–1585. Zhan, R.‐F., & Wang, Y. (2016). CFSv2‐based statistical prediction for seasonal Accumulated Cyclone Energy (ACE) over the Northwest Pacific. Journal of Climate, 29, 525–541.

Zhao, H., & Wang, C. (2016). Interdecadal modulation on the relationship between ENSO and typhoon activity during the late season in the western North Pacific. Climate Dynamics, 47, 315–328. Zhao, H., & Wang, C. (2018). On the relationship between ENSO and tropical cyclones in the western North Pacific during the boreal summer. Climate Dynamics. https://doi. org/10.1007/s00382‐018‐4136‐0 Zhao, X., & Chu, P.‐S. (2006). Bayesian multiple changepoint analysis of hurricane activity in the eastern North Pacific: A Markov Chain Monte Carlo approach. Journal of Climate, 19(4), 564–578. https://doi.org/10.1175/JCLI3628.1 Zheng, Z.‐W., Lin, I. I., Wang, B., Huang, H. ‐C., & Chen, C. H. (2015). A long neglected damper in the El Niño–typhoon relationship: A ‘Gaia‐like’ process. Scientific Reports, 5, 11103. https://doi.org/10.1038/srep11103 Zhou, X., & Wang, B. (2007). Transition from an eastern Pacific upper‐level mixed Rossby‐gravity wave to a western Pacific tropical cyclone. Geophysical Research Letters, 34, L24801. https://doi.org/10.1029/2007GL031831 Zuidema, P., Chang, P., Medeiros, B., Kirtman, B. P., Mechoso, R., Schneider, E. K., et al. (2016). Challenges and prospects for reducing coupled climate model SST biases in the eastern tropical Atlantic and Pacific oceans: The U.S. CLIVAR Eastern Tropical Oceans Synthesis Working Group. Bulletin of the American Meteorological Society, 97(12), 2305–2328. https://doi.org/10.1175/ BAMS‐D‐15‐00274.1

18 ENSO-Driven Ocean Extremes and Their Ecosystem Impacts Neil J. Holbrook1,2, Danielle C. Claar3, Alistair J. Hobday4, Kathleen L. McInnes5, Eric C. J. Oliver6, Alex Sen Gupta7,8, Matthew J. Widlansky9, and Xuebin Zhang4 ABSTRACT El Niño–Southern Oscillation (ENSO) events can cause extremes in the ocean environment that have substantial impacts on marine ecosystems. In the shallow‐water/coastal marine environment, ENSO‐related extremes in sea level and seawater temperature have been found to impact coral, kelp, seagrass, and mangrove ecosystems. Coastal impacts from sea level extremes include exposure of shallow‐water ecosystems and inundation in low‐lying areas. Ocean temperature extremes, including marine heatwaves, cause coral bleaching and can impact kelp and seagrass density. This chapter reviews knowledge and understanding of ENSO’s role in sea level extremes and ocean ­temperature extremes, and their impacts on these key shallow‐water/coastal marine ecosystems.

18.1. INTRODUCTION El Niño–Southern Oscillation (ENSO) events are due to coupled ocean‐atmosphere dynamics operating in the equatorial Pacific Ocean (chapters 2 and 3). Nevertheless, this globally dominant mode of interannual climate variability has important teleconnections and impacts (section VI). In particular, extremes in sea level and ocean temperature associated with ENSO can be highly detrimental to marine ecosystems. Importantly, shallow‐water marine ecosystems, which have high biodiversity and represent critical regions for fisheries and aquaculture, can be vulnerable to these ocean extremes (Smale et al., 2019).

This chapter reviews our knowledge of the role of ENSO events on extremes in sea level and seawater temperature, and concomitant impacts on shallow‐water/coastal marine ecosystems. For seawater temperature, we only examine warm water extremes in relation to ENSO events that appear to be driving significant changes to ecosystems against the background of rising ocean temperatures. For sea level, we examine both high and low extremes (which are also increasing against the backdrop of rising global sea levels) in relation to ENSO events, as both can be critically important to ecosystems. Our examination of marine ecosystem impacts is primarily focused on the important foundational species of coral, kelp, seagrass, and mangrove.

 Institute for Marine and Antarctic Studies, University of Tasmania, Hobart, TAS, Australia 2 Australian Research Council Centre of Excellence for Climate Extremes, University of Tasmania, Hobart, TAS, Australia 3  School of Aquatic and Fisheries Sciences, University of Washington, Seattle, WA, USA 4  CSIRO Oceans and Atmosphere, Hobart, TAS, Australia 5 Climate Science Centre, CSIRO Oceans and Atmosphere, Aspendale, VIC, Australia

  Department of Oceanography, Dalhousie University, Halifax, Nova Scotia, Canada 7  Climate Change Research Centre, University of New South Wales, Sydney, NSW, Australia 8   Australian Research Council Centre of Excellence for Climate Extremes, University of New South Wales, Sydney, NSW, Australia 9    Joint Institute for Marine and Atmospheric Research, School of Ocean and Earth Science and Technology, University of Hawai‘i at Mˉanoa, Honolulu, HI, USA

1

6

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 409

410  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

18.2. EXTREMES IN SEA LEVEL AND SEAWATER TEMPERATURE Here we provide background understanding of what constitutes extremes in sea level and seawater temperature. We also provide a review of the role that ENSO plays in causing or modulating these extremes. 18.2.1. Sea Level Extremes 18.2.1.1. Background Extreme sea levels (either high or low) occur as a result of various processes that operate on a wide range of time and space scales. These range from tides, which operate on subdaily time scales, to weather and synoptic systems operating on daily to multiday time scales, to interannual changes associated with ENSO, to sea level rise associated with greenhouse warming. Moreover, multiple processes may combine to produce the most severe sea level extremes. Contributions from seasonal and interannual factors such as the variations in ocean temperature, salinity, and circulation also play a role by influencing mean regional sea level and/or by influencing the weather conditions (e.g. tropical cyclones, monsoons, and extratropical depressions) that can cause sea level extremes from storm surges and surface waves. Extreme high sea levels are of concern because they cause coastal flooding, saltwater contamination, and erosion in the coastal zone (Barnard et al., 2015). Extreme low sea levels can cause coastal ecosystems to be exposed to other stressors. Due to different ways in which sea levels can affect urban and natural environments, there is no single definition for an extreme sea level. Thresholds above which sea levels may be considered extreme include tidal metrics such as the Highest Astronomical Tide or Mean High Water Springs (i.e. near the annual high‐ water mark), or statistical estimates of heights associated with particular return periods (e.g. Average Recurrence Intervals) that are required for coastal planning or engineering applications. Scientific applications may also consider high percentile thresholds as a means to define extreme sea levels. In addition to sea level height, the duration of coastal flooding or strength of coastal currents may be important in determining the severity of coastal impacts. Here, we consider extreme sea level characteristics in a general sense as those associated with coastal inundations or exposures of shallow ecosystems (i.e. not fully submerged in water anymore). Growing evidence suggests that the characteristics of extreme sea level are experiencing long‐term changes as global sea levels gradually rise (Menéndez & Woodworth, 2010; Woodworth & Menéndez, 2015). Changes in the frequency and intensity of these extremes can be related to multiple factors. Most directly, climate change is expected to alter extreme sea level frequency and inten-

sity via a number of pathways. For example, rising mean sea level due to ocean thermal expansion and land ice melt will alter regionally the background sea level on which extremes occur. Higher mean sea levels can also interact nonlinearly with other processes such as tides, waves, and storm surges to alter extreme water levels. Hoeke et  al. (2015) undertook a numerical modeling study in Apia, Samoa, and showed that sea level rise reduced the amount of wind and wave setup occurring during tropical cyclone–induced storm surges. The deeper water from sea level rise reduced the hydraulic roughness, which led to a reduction in coastal sea levels caused by wind and wave setup, but it also increased the wave energy reaching the coast by 200% due to the decreased wave ­dissipation on outer reefs. Climate change is also expected to affect the intensity, frequency, seasonality, and tracks of weather systems that cause storm surges and waves and therefore extreme sea levels (Chand & Walsh, 2009; Walsh et al., 2012). For example, McInnes et al. (2014, 2016) employed numerical modeling to show that for Samoa and Fiji, the projected future increase in tropical cyclone intensity together with long return periods (200 years and longer) would slightly lower storm tide 50‐year return periods. However, the projected rise in sea levels increased the total height of storm tides for Fiji overall. Finally, climate change may also affect the behavior of ENSO (see chapter 13; e.g. Cai et al., 2015). This will in turn influence extreme sea levels, particularly in the Indo‐ Pacific region through its influence on regional sea level and wind and weather events such as tropical cyclones (see chapter  17), with consequent effects on waves and storm surges (the effect of climate change on ENSO is investigated in chapter 13). 18.2.1.2. Role of ENSO In the tropical Pacific Ocean, ENSO is the dominant process affecting sea level on monthly‐to‐interannual timescales (Zhang & Church, 2012; Figure  18.1), with changes exceeding ±30 cm over several months recorded by satellites and tide gauges, especially during and after strong El Niño and La Niña events (e.g. Wyrtki, 1984; Merrifield et al., 1999). The anomalously high regional sea level in the central and eastern (western) tropical Pacific during El Niño (La Niña) events can increase the likelihood of extreme water levels impacting the coast, since the background sea level is one of the key factors affecting frequency and intensity of water level extremes (Menéndez & Woodworth, 2010; Woodworth & Menéndez, 2015). Pacific sea level anomalies are mostly related to wind‐ driven shifts of the tropical thermocline, which can propagate across the basin either eastward as equatorial Kelvin waves or westward as oceanic Rossby (planetary) waves. Ocean Kelvin waves also propagate to the

ENSO-Driven Ocean Extremes and Their Ecosystem Impacts  411 (a)

(b)

EOF1 of SLA (cm)

60°N

10 8 6 4 2 0 –2 –4 –6 –8 –10

40°N 20°N 0° 20°S 40°S 60°S 0°

60°E

120°E

(c)

180°

120°W 60°W

EOF2 of SLA (cm)

60°N

10 8 6 4 2 0 –2 –4 –6 –8 –10

40°N 20°N 0° 20°S 40°S 60°S

0°

0°

60°E

120°E

(d)

PC1 of SLA

180°

120°W 60°W

0°

PC1 of SLA

4

4

3

3

2

2

1

1

0

0

–1

–1

–2

–2 –3

–3 MEI

PC1

–4 1995

2000

2005

2010

2015

–4

MEI

PC2 1995

2000

2005

2010

2015

Figure 18.1  Empirical orthogonal function (EOF) analysis of global sea level anomalies (SLAs). (a, b) The first and second EOFs based on monthly 1x1o satellite altimetry observations over 1993–2018 provided by Australia CSIRO (https://research.csiro.au/slrwavescoast/sea‐level/measurements‐and‐data/sea‐level‐data/). (c, d) Corres­ ponding principal components (PCs) in comparison with the Multivariate ENSO Index. To focus on natural vari­ ability, monthly global mean sea level has been removed and the altimetry records have been detrended locally before EOF analysis. All time series in (c, d) are smoothed by a 5‐month running mean filter.

e­ xtratropics along coastal waveguides at the eastern sides of ocean basins, for example, along the west coast of North and South America and west coast of Australia, affecting coastal sea level and circulation (chapter  15). Seesaws in east‐west sea level occur across the Pacific mostly in‐phase with ENSO (Wyrtki, 1984), with sea level increasing (decreasing) in the east (west) during El Niño (La Niña) (Figure  18.1a). Occasionally, a weaker north‐south seesaw in sea level (e.g. Delcroix & Rual, 1997) also takes place in the tropics that typically persists long after a strong El Niño event ends (Figure  18.1b; Widlansky et  al., 2014). Particularly during strong El Niño events, sea level drops around tropical western Pacific islands, first in the Northern Hemisphere and later in the Southern Hemisphere, but rises in the eastern Pacific. Around the peak of an El Niño, high sea levels occur in the central and eastern equatorial Pacific and along the eastern boundary of the Pacific in both hemispheres (i.e. the North and South American coasts).

Coastal high sea level anomalies during El Niño have been observed as far north and south as Alaska and the southern tip of South America, respectively (Chelton & Davis, 2002; Enfield & Allen, 2002; Barnard et al., 2017). During La Niña events, the sea level pattern mostly reverses, although nowhere are the anomalies typically as large as during strong El Niño. Sea levels outside of the Pacific Ocean are also affected by ENSO via atmospheric teleconnections which cause oceanic anomalies in other basins such as the Atlantic (Sweet & Park, 2014; Hamlington et al., 2015). El Niño and La Niña events also sometimes indirectly affect sea level through changes in the occurrence or intensity of storms such as tropical cyclones in the Pacific (Chand et al., 2013) as well as in other ocean basins (e.g. Camargo et al., 2007), which cause localized storm surges, surface wind waves, and also swells of wave energy that can propagate across the Pacific (unless refracted by an island; see chapter  17). These higher‐frequency and

412  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

mostly localized storm phenomena cause more extreme water level variations compared to the sea level fluctuations directly associated with ENSO. Conversely, the occurrence of El Niño and La Niña events can influence the coastal impacts of severe storms when they occur. For example, in December 2008, a major inundation event, triggered by swell waves from an intense low pressure system in the North Pacific Ocean, caused significant inundation across a number of islands within six nations in the western tropical Pacific Ocean from Wake Atoll in the north to Papua New Guinea and the Solomon Islands in the south. A contributing factor to the severity of the impacts was shown to be the positive sea level anomalies of about 10–20 cm across the region caused by a La Niña event that was occurring at the time (Hoeke et al., 2013). The coastal impacts of sea level variability can be exacerbated by large astronomical tides which compound the sea level anomalies to produce either extremely high or low water‐level events. High sea levels, combined with powerful storm surges or waves, can wreak havoc at the coast in the form of inundations of low‐lying areas (Barnard et al., 2015), especially around the time of highest tides (Hoeke et al., 2013). Conversely, below‐normal sea levels make the lowest tides even lower, which sometimes exposes shallow‐water ecosystems such as reef flats to air and may cause coral die‐offs (Ampou et al., 2017). Hemer et  al. (2010) showed that wave direction in the Tasman Sea is influenced by ENSO with waves in this region that are mainly from the southeast undergoing a rotation toward being from the south during El Niño and from the east during La Niña. The rotation in wave direction with ENSO events has been found to influence erosion along Australia’s east coast (Ranasinghe et al., 2004; Harley et al., 2011), such that during El Niño the northern end of a beach widens (accretes) while the southern end erodes, resulting in a net clockwise rotation of the beach. During La Niña events an anticlockwise rotation occurs. Climate change is likely to affect the behaviour of extreme ENSO events (chapter 13; Cai et al., 2015), which would likely cause associated effects on sea level variability, especially in the Indo‐Pacific region. Using global coupled general circulation models, future projections suggest that climate change will enhance ENSO‐related sea level extremes by up to 25% in the tropical Pacific (Widlansky et  al., 2015). The future change pattern is consistent with more extreme swings of the major wind convergence zones such as the South Pacific Convergence Zone (Cai et al., 2012). The effects of future changes to ENSO and extreme wave energy reaching the coast have been investigated by Mentaschi et al. (2017), who find a significant increase in wave energy along 29% and a significant decrease along 26% of the global coastline in the future. The positive trends in the southern tropical Pacific region and the

northeast Pacific, together with the decreases projected for the western Pacific, are shown to be linked to an intensification of the ENSO pattern and a shift of climate toward El Niño conditions. For the southern tropical Pacific, the increases are driven by storms and convective activity in the eastern Pacific, while the positive link between El Niño and the Aleutian Low pressure in the northeast Pacific is linked to positive trends in wave energy there. For the western Pacific, generally colder, drier and less stormy conditions during El Niño episodes lead to a reduced frequency of severe wind and wave conditions in the future (Mentaschi et al., 2017). ENSO‐related regional sea level anomalies and tropical cyclone behaviour and their links with extreme sea levels have been studied to some extent for particular coastlines (Feng & Tsimplis, 2014; Sweet & Park, 2014), and some studies have attempted to quantify the effects of ENSO on extreme sea levels from storm surges at the island scale (McInnes et al., 2014, 2016). However, the role of ENSO on extreme waves is more challenging. Coastal wave extremes are sensitive to local coastal morphology, particularly shoreline slope (Hoeke et al., 2013; Aucan et al., 2019). Along much of the global coastline, particularly in small islands, such information is lacking, as are in situ observations of extreme coastal wave and sea levels to quantify the local impacts of extreme sea levels, waves, and their relationship to ENSO. Due to the consequences to vulnerable coasts, associated with changing sea levels and ENSO conditions, a number of scientific challenges must be addressed to produce more robust future predictions and projections. Already, seasonal forecasts of sea level extremes (e.g. Miles et al., 2014), while still experimental, are helping coastal communities adapt to the impacts of rising sea levels as well as the shorter‐term variability associated with ENSO. Yet current‐generation global climate models are challenged in resolving sea level dynamical processes, especially around shallow coastal regions (Church et al., 2013). Furthermore, remote teleconnections (i.e. ENSO‐related fluctuations outside of the tropical Pacific) are not fully understood, either in the present climate or future projections. Next‐ generation higher‐resolution models (Griffies et al., 2015), or dynamical downscaling with regional high‐resolution ocean models (Liu et  al., 2016; Zhang et  al., 2017), may provide improved predictions of sea level extremes. 18.2.2. Ocean Temperature Extremes 18.2.2.1. Background Ocean temperature extremes (OTEs) occur when ocean temperatures exceed a suitably defined extreme threshold (either hot or cold). The temperature threshold can be either an absolute one (e.g. 1°C above normal summer

ENSO-Driven Ocean Extremes and Their Ecosystem Impacts  413

maximum temperature in the case of coral bleaching risk (Donner, 2011) or 12°C winter temperature in the case of the spiny sea urchin off southeastern Australia (below which larval development is poor; Ling et al., 2008)) or a relative one (e.g. the 90th percentile of the temperature distribution; Hobday et  al., 2016). OTEs can have moderate to severe impacts on marine ecosystems, which will be discussed in more detail in the following section. Periods of prolonged and sustained OTEs at a particular location may be considered marine heatwaves (MHWs; Hobday et al., 2016) or marine cold spells (Schlegel et al., 2017). A number of prominent MHWs have occurred in recent years, including in the northern Mediterranean Sea in 2003 (Sparnocchia et al., 2006; Olita et al., 2007), off Western Australia in 2011 (Pearce & Feng, 2013), in the northwest Atlantic in 2012 (Chen et al., 2014), and in the northeast Pacific from 2013 to 2016 (Bond et  al., 2015; Di Lorenzo & Mantua, 2016). MHWs have received considerable attention recently due to the historical record indicating significant increases in MHW frequency and duration since the early 20th century (Oliver et al., 2018a), as well as future projections indicating that these trends will likely accelerate (Frölicher et al., 2018). OTEs may occur at any depth through the water column. However, most research to date has focused on sea surface temperatures (SSTs), primarily due to the lack of high spatial and temporal resolution subsurface temperature data. Given the focus on surface temperatures, the understanding of local physical drivers of OTEs has primarily been in the context of mixed‐layer heat budget analyses (Benthuysen et  al., 2014; Chen et al., 2014; Oliver et al., 2017). Local drivers of OTEs include (i) anomalous air‐sea heat fluxes associated with changes in cloud cover and radiation or winds that affect latent and sensible heat fluxes; (ii) anomalous horizontal advection by changes in the large‐scale circulation, mesoscale structures, or surface Ekman currents; and (iii)  vertical processes including mixing or upwelling (Holbrook et al., 2019). The resulting changes in temperature will also be modulated by the background ocean state, for example, an anomalously shallow mixed layer will be susceptible to larger temperature variations. Individual OTEs will be triggered, sustained, and terminated by different combinations of these terms. For example, the 2015–2016 Tasman Sea MHW (Oliver et  al., 2017) and 2011 Ningaloo Niño off Western Australia (Feng et al., 2013) were primarily initiated by greater heat transport associated with western and eastern boundary current intensification, respectively, although air‐sea interactions also played a role (Benthuysen et al., 2014). Conversely, the 2017–2018 Tasman Sea MHW (Perkins‐Kirkpatrick et  al., 2019) and the Blob (Bond et al., 2015) were triggered by atmospheric systems that primarily modified air‐sea heat fluxes.

The timing, intensity, and frequency of OTEs may be modulated by large‐scale modes of climate variability, such as ENSO, the Indian Ocean Dipole, or the North Atlantic Oscillation. This can occur by these modes’ modulating ocean temperatures directly, as in the case of ENSO in the tropical Pacific, or indirectly by driving complex atmospheric (chapter 14) or oceanic (chapter 15) teleconnections, which occur remotely from the main center of action of the climate mode. A number of studies have linked specific events to such remote ENSO teleconnections. These are described in the following section. Recently, Holbrook et al. (2019) used a common framework to identify MHW characteristics and to examine the influence of large‐scale climate modes on MHWs around the world. They showed that the different flavors of ENSO (refer chapter 4) appear as the dominant drivers of MHW characteristics in much of the tropical Indian Ocean and many parts of the extratropical Pacific Ocean, in addition to the tropical Pacific. For example, the number of days experiencing MHW conditions in the northeast Pacific almost doubles (halves) when the central equatorial Pacific temperature anomalies (described by the Niño‐3.4 index) is positive (negative). In many other parts of the extratropical Pacific, low‐frequency expressions of ENSO like the Pacific Decadal Oscillation (PDO) or Interdecadal Pacific Oscillation (IPO), which modulate background SST on decadal timescales, are also significant predictors of MHW occurrence. 18.2.2.2. Role of ENSO Coupled ocean‐atmosphere feedbacks associated with ENSO generate large SST anomalies in the eastern and central tropical Pacific that can give rise to OTEs lasting for a season or longer. In addition, tropical ENSO disturbances can transmit to remote regions via atmospheric (chapter  14) or oceanic (chapter  15) teleconnections, including planetary waves and large‐scale circulation, providing the conditions favorable for MHWs. ENSO has the strongest effect on MHW properties in the eastern and central tropical Pacific (Oliver et  al., 2018a; Figure  18.2). In those regions, El Niño events drive long‐duration MHW events with high intensities (Figure  18.2a,d,g). Average durations are as long as 60 days in parts of the ENSO‐dominated eastern tropical Pacific (Figure 18.2g), but these events occur infrequently, typically less than one event per year (Figure 18.2a). In other parts of the tropical oceans, MHW durations are typically 5–10 days with one to three events per year. ENSO tends to increase the frequency of MHWs in all ocean basins in the tropics, but particularly across the Pacific (Figure  18.2b). Outside of the tropics, El Niño and La Niña events are both associated with increases in the mean MHW duration and intensity in the northeast

2.0

30°S

1.5 1.0

60°S

0.5

(d)

Intensity

0°

2

30°S

4

Duration

15

30°S

10 5

60°S 60°E

120°E

180°

120°W

60°W

0°

[days]

20

0°

Duration

60°N

30

30°N

2.5 El Niño period La Niña period

Intensity

30°N 0° 30°S 60°S 60°E

120°E

180°

120°W

60°W

0°

2.0 1.5 2.4 2.3

0.75

2.2

0.25

2.1 2.0 1.9 1.8 1.7

–1.50

(h)

60

60°N

3.0

–1.00

60°S

0

(g)

3.5

1.25

0°

4.5 4.0

–0.50

30°S 60°S

(f)

Intensity

30°N [°C]

3

60°S

60°N

4

30°N

30°S

(e)

5

60°N

0°

Global average Excluding ENSO

55.0 35.0 15.0 7.5 2.5 –2.5 –7.5 –15.0 –35.0 –55.0

[count]

0°

30°N

Global average Frequency

[°C]

2.5

(c)

(i)

Duration 25 20 15 10 1985 1990 1995 2000 2005 2010 2015

Figure 18.2  Marine heatwave (MHW) properties and the impact of ENSO. Local MHW (a) frequency, (d) intensity, and (g) duration are shown averaged over the 1982–2016 period based on NOAA OI SST data. The contribution of ENSO to these average MHW properties is shown in (b, e, h). This was calculated by first removing the influence of ENSO from the underlying SST, statistically using the Multivariate ENSO Index (MEI), and then subtracting the resultant ENSO-independent MHW patterns from the original patterns shown in (a, d, g). Associated globally averaged time series of MHW (c) fre­ quency, (f) intensity, and (i) duration are shown before (black lines) and after (red lines) removing the influence of ENSO. Light red and blue shading indicate El Niño and La Niña periods, respectively, defined by +/‐1 std. dev. of the MEI for at least 3 months consecutively. (Figure adapted from Oliver et al., 2018a)

[days]

30°N

60°N

2.5 2.0 1.5 1.0 0.5 –0.5 –1.0 –1.5 –2.0 –2.5

Annual number [count]

3.0

Annual number [count]

60°N

Mean of difference (w/ENSO - w/o ENSO) Frequency

[°C]

(b)

3.5

[days]

Mean (w/ENSO) Frequency

(a)

ENSO-Driven Ocean Extremes and Their Ecosystem Impacts  415

Pacific Ocean, off western Australia, and coastal California (Figure 18.2e,h). ENSO also tends to increase MHW frequency in the mid‐ and high latitudes of the Pacific Ocean. At global scales, ENSO also drives significant interannual variations in MHW properties. During the strongest El Niño events (i.e. 1982–1983, 1997–1998, and 2015–2016), when averaged globally, there was typically 0.5–1 extra MHW event (Figure 18.2c), with intensity increases of 0.1°C–0.2°C (Figure  18.3f) and annual number of MHW days increasing by 5–10 days (Figure 18.2i). A number of specific high‐impact MHWs outside of the central and eastern tropical Pacific have been linked to ENSO events. To describe these events, we have applied the new categorization scheme of Hobday et  al. (2018) that expresses the severity reached by events. Various mechanisms come into play in generating remote temperature responses that can take part in initiating, maintain(a)

South Central Pacific 27-12-2009

ing, and terminating MHWs. In the atmosphere, changes in the strength and location of tropical convection associated with the zonal displacement of the western Pacific warm pool can affect the Walker Circulation and the regions of convergence/divergence across the tropical oceans of all basins. Convection changes also trigger planetary Rossby waves that drive remote changes in atmospheric circulation, including wind speed and direction and cloudiness that can influence ocean mixed layer temperatures (Lee et  al., 2010; Di Lorenzo & Mantua, 2016; e.g. chapter  14). Oceanic Kelvin waves also propagate to the extratropics along coastal waveguides at the eastern sides of ocean basins, for example, along the west coast of North and South America and the west coast of Australia, affecting coastal temperatures, sea level, and circulation (chapter 15). Changes in the Walker Circulation link ENSO changes to the Indian Ocean, helping to trigger Indian Ocean Dipole (b)

Nigaloo Nino 6-3-2011

0° 20°S 20°S 30°S 40°S 40°S

60°S 140°E 160°E

(c)

180°

160°W 140°W 120°W 100°W 80°W

80°E

90°E

100°E

110°E

120°E

130°E

The Blob 15-2-2015

60°N

Marine Heatwave Category Extreme

40°N

Severe Strong

20°N Moderate

0°

180°

160°W

140°W

120°W

100°W

80°W

Figure 18.3  Categories for three high‐impact marine heatwaves associated with ENSO events: (a) South central Pacific during the 2009–2010 central Pacific El Niño, (b) Ningaloo Niño, associated with the 2010–2011 La Niña, and (c) the latter stages of the North Pacific “Blob,” associated with the weak El Niño conditions of 2014–2015. Category definitions are given in Hobday et al. (2018).

140°E

416  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

(IOD) events in the months leading up to the ENSO peak (Saji et  al., 1999), and a basin‐wide warming across the tropical Indian Ocean during and after the peak of ENSO events. For example, the IOD event associated with the extreme 2015–2016 El Niño was implicated in an extended MHW in the waters to the north of Australia (Oliver et al., 2018b), and the broad‐scale warming of the Indian Ocean associated with the 1997–1998 extreme El Niño was linked to an intense MHW and extensive coral mortalities in the Seychelles. More generally, MHWs and associated bleaching events are more likely in the tropical southeast Indian Ocean as a result of increased insolation and a weaker Australian monsoon during El Niño (Zhang et al., 2017). A number of unprecedented warming events outside of the tropics are also believed to have been triggered by both El Niño and La Niña events. In the central South Pacific during the austral summer of 2009–2010, SST anomalies exceeded five standard deviations (Lee et  al., 2010) concurrent with a record‐breaking central Pacific El Niño (Figure  18.3a). The event was linked to an atmospheric Rossby wave train stretching into the subtropics that formed a blocking high‐pressure system over the region. This led to an anomalous inflow of warmer air and weaker wind speeds with an associated decrease in turbulent heat losses from the ocean. Wind changes also weakened the normal northward flow of cold surface waters in the Ekman layer (Lee et al., 2010). As such, multiple atmospherically forced processes combined to generate this extreme event. The west coast of Australia experienced an unprecedented MHW that peaked in February 2011 when temperatures were 2°C–4°C warmer than normal for about 3 months (Figure  18.3b). The event led to significant ecosystem impacts and damage that has persisted to the present day (Wernberg et al., 2013, 2016). This event has also been linked to one of the strongest La Niña events on record. Enhanced Pacific trade winds led to sea level increases in the western Pacific that propagated through the Indonesian Archipelago and along the west coast of Australia as coastally trapped oceanic Kelvin waves. The enhanced cross‐shore sea level gradient intensified the Leeuwin Current, advecting more warm water southward. In addition, an atmospheric teleconnection to the La Niña (a Gill‐Matsuno response) suppressed the seasonal southerlies that normally weaken the Leeuwin Current and reduced the regional wind strength, thereby suppressing heat losses from the ocean (Feng et al., 2013; Caputi et al., 2016). Thus, a combination of remote ocean and atmospheric teleconnections combined to ­generate this unprecedented event. ENSO may have also played a role in the extreme persistence of “The Blob” (Figure 18.3c): a multiyear MHW in the northeast Pacific. The event had important ecological implications, with large numbers of sea mammal and bird deaths and harmful algal blooms that caused closures of important shellfish fisheries along the western U.S. coastline. The initi-

ation of the MHW has been linked to a persistent high‐ pressure ridge that occurred between October 2013 and January 2014 and an associated reduction in wind speeds (leading to reduced ocean heat loss) and anomalous warm northward advection in the Ekman layer (Bond et al., 2015). Di Lorenzo and Mantua (2016) suggest that the high‐ pressure system was part of the basin‐wide North Pacific Oscillation that favors the development of an El Niño. They suggest that the resulting warming of the tropical Pacific generated an atmospheric Rossby wave response that manifested as a deep low‐pressure system in the open northeast Pacific Ocean. This caused the warm SST anomalies to shift toward the North American coastal region and allowed the MHW to persist through the summer of 2015. Future projections of MHWs (Frölicher et  al., 2018) indicate a significant increase in the probability of MHWs due to anthropogenically driven global warming, particularly in the tropical oceans. This may raise the background likelihood of MHWs in this region and therefore exacerbate the risk due to ENSO‐driven extreme MHWs. Given the influence of ENSO on MHWs, changes in ENSO intensity could also affect MHW characteristics. However, there is little consensus around how ENSO will change in the future, except perhaps for suggested increases in the frequency of the most extreme events (chapter 13). 18.3. IMPACTS ON SHALLOW-WATER MARINE ECOSYSTEMS Shallow‐water marine ecosystems are vulnerable to extremes in sea level and seawater temperature. In the following subsections, we introduce the key shallow‐water marine ecosystem foundation species of coral reefs, rocky reef kelp forests, sea grasses, and mangroves. We examine the impacts on these shallow‐water marine ecosystems from sea level and ocean temperature extremes, and more specifically associated with ENSO event occurrences. All of these ecosystems are sensitive to environmental change and extremes. The specific sensitivity varies between ecosystems. In the case of seagrass and kelps, they have limited propagule dispersal and require sunlight; thus, they cannot simply avoid warming waters by retreating into deeper (cooler but darker) waters. Corals have a narrow thermal tolerance but can persist in deeper waters. Mangroves occupy the interface between land and sea and are thus impacted by both marine and atmospheric conditions. Flow‐on effects from ecosystem impacts to humans are discussed in chapter 19. 18.3.1. Tropical Coral Reef Ecosystems 18.3.1.1. Background Coral reefs, some of the most biodiverse ecosystems on Earth, are threatened by climate change and episodic

ENSO-Driven Ocean Extremes and Their Ecosystem Impacts  417

warming events like El Niño (Hoegh‐Guldberg et  al., 2007; Ainsworth et al., 2016). Severe stress events are already affecting corals and are expected to increase in the near future, even under moderate warming scenarios (van Hooidonk et  al., 2016), threatening fundamental services provided by coral reefs, including food and economic stability, that support tens of millions of people in over 100 countries (Salvat, 1992; Moberg & Folke, 1999). The health of hard (scleractinian) corals is fundamentally important to the persistence and biodiversity of coral reefs, since corals form complex reef structures providing habitat for many thousands of reef‐associated species. The impact of episodic warming events can be further exacerbated by local stressors, including overfishing, pollution, and disease (Hughes et  al., 2003; Wiedenmann et  al., 2012; Vega Thurber et  al., 2014). Warming events that affect corals may occur due to a number of cyclic and stochastic climatic events, such as El Niño, La Niña, or anomalous summer conditions (Zhang et  al., 2017; Eakin et  al., 2018). In particular, severe warming events, such as a strong El Niño, can decimate even remote and protected reefs (e.g. Aeby et  al., 2003; Hughes et al., 2017). Therefore, understanding the dynamics of El Niño severity is necessary to enable prediction of coral resilience over the next century. Corals form an obligate nutritional symbiosis with Symbiodiniaceae (the symbiont, previously called zooxanthellae or Symbiodinium; Muscatine & Cernichiari, 1969). This symbiosis provides photosynthetic metabolites to host corals, ultimately allowing corals to grow and build the carbonate skeletons fast enough to withstand the natural forces of physical and biological erosion to create coral reef habitat (Muscatine & Porter, 1977). Originally, it was thought that there was only one type of symbiont, but advances in molecular analysis have revealed that there are multiple genera within Symbiodiniaceae that can change in abundance over time (Rowan, 1995; Baker, 2003; LaJeunesse et  al., 2018). These different symbionts vary in their ecological functions and response to stress (Stat & Gates, 2011; Baker et al., 2018). Since symbioses can vary in time and space, they are a source of local, regional, and species‐ specific variability in coral susceptibility to episodic heat stress events such as El Niño. During periods of intense and prolonged warming, this coral symbiosis can break down and Symbiodinaceae are expelled from the coral tissue causing the host coral to appear white or “bleached” (Brown, 1997). Although other sources of stress can cause coral bleaching (e.g. cool temperatures, anomalous salinity, and pollution), we focus this section on bleaching caused by high‐temperature stress. The link between temperature stress and mortality was forged through reef observations during El Niño warming. Although one of the first reports of mass

coral bleaching was associated with the 1982–1983 El Niño (Glynn, 1983), the link to El Niño warming was not made at that time. However, as research into this early event continued, evidence built for the connection between El Niño warming and mass coral bleaching (Glynn & D’Croz, 1990; Williams & Bunkley‐Williams, 1990). Additional research has shown that the probability of coral bleaching on any particular reef is not dependent on a specific temperature but rather local thermal thresholds (Gleeson & Strong, 1995). In areas with seasonal temperature change, corals have been more susceptible to thermal stress during the local summer warm season, when thermal thresholds are more likely to be crossed. Coral bleaching risk has been commonly predicted and described using the metric “degree‐heating weeks” (DHW, unit °C‐weeks; NOAA Coral Reef Watch, 2000; Liu et al., 2014), which sums temperature anomalies (>1°C) above the maximum annual mean accumulated over the previous 12 weeks as a measure of cumulative heat stress. Due to timing and local conditions, each El Niño event has its own spatial and temporal signature of heat stress, and the DHW metric captures the specific duration and intensity of warming affecting coral reefs at any one location. Depending on the amount of heat stress, there is some capacity for corals to persist and recover after bleaching. Although if warming is extreme or prolonged, corals may die (McClanahan, 2004). While warming can fundamentally transform coral communities and decimate reefs (Hughes et  al., 2019), the relatively short‐term nature of El Niño warming may promote coral adaptation to warming and increase resilience to future events in some cases by selecting for resilient genotypes (Guest et al., 2012). However, as the return times between bleaching events decrease (Hughes et al., 2018b), corals will likely become more threatened with each new El Niño event. 18.3.1.2. Impacts from ENSO-Related Ocean Extremes El Niño events have instigated several coral bleaching events since intensive reef monitoring began in the early 1980s (Heron et  al., 2016). In particular, three “global” bleaching events occurred during the 1997–1998, 2009– 2010, and 2015–2016 El Niño events, while more regionally confined bleaching events occurred during the 1982–1983, 1986–1987, 2002–2003, and 2005 El Niño events (Oliver et  al., 2018). El Niño warming is now superimposed on the warming of the global oceans, increasing the probability of exceeding coral thermal thresholds compared to the past (Hughes et  al., 2018b; Lough et  al., 2018). This is also evident in the overwhelming dominance of ocean warming that occurred during the 2015–2016 El Niño event (Figures  18.4 and 18.5). Overall, El Niño–associated warming significantly increases coral bleaching and mortality, although the

418  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

intensity of effects can be variable across regions and events (Figure 18.5; reviewed in Claar et al., 2018). Here, we note that the warming caused by La Niña tends to pale in comparison to warming caused by El Niño events. Therefore, while La Niña events can also cause some coral bleaching due to localized or regional warming, El Niño is the primary ENSO‐related driver of bleaching at the large scale. The first well‐documented case of large‐scale coral bleaching occurred during the 1982–1983 El Niño (Glynn, 1983, 1984, 1988; Williams & Bunkley‐Williams, 1990). Bleaching intensity was regionally variable, with some locations experiencing extreme bleaching. For example, 95% coral mortality was observed during this event on some eastern Pacific reefs (Glynn, 1990), and up to 100% mortality was documented at some sites in Indonesia (Brown & Suharsono, 1990). These losses were severe, and El Niño events of this magnitude can even cause extinction of endemic coral species. In Panama, this El Niño caused the probable extinction of two hydrocoral species (Glynn & de Weerdt, 1991). Reefs in Panama were further impacted by low‐tide exposure that occurred during anomalous low sea levels associated with the subsequent La Niña events in 1988–1989 and 1993, causing significant reef erosion that had not recovered 17 years later (Eakin, 2001), although nearby Costa Rican reefs were well on their way to recovery 20 years after the event (Guzman & Cortes, 2006). Exposures of shallow reefs in the northwestern and south‐ central Pacific have also been observed, respectively, during and after strong El Niño events (Widlansky et al., 2014). In the Samoan Islands, for example, such events are referred to as taimasa (kai‐ma‐sa), which means smelly reef. The first recorded global coral bleaching event coincided with the 1997–1998 El Niño and the subsequent strong La Niña (Wilkinson & Science, 1998). A cascade of mass bleaching events during this El Niño caused serious degradation of more than 16% of the world’s tropical coral reefs (Wilkinson, 2002). For example, during this global bleaching event there was 90% mortality on some Indian Ocean reefs (Wilkinson & Hodgson, 1999). Coral recovery after El Niño can be slow. After the 1997–1998 El Niño, Brazilian reefs took more than a decade to recover to prebleaching levels (Kelmo & Attrill, 2013), and an isolated reef system off western Australia took 12 years to recover to prebleaching coral cover (Gilmour et  al., 2013). Following this strong El Niño, scientists proposed that increased mass coral bleaching was linked to increased El Niño activity (Stone et al., 1999). The moderate 2002–2003 El Niño caused mass bleaching and mortality in the Phoenix Islands (central Pacific; Obura & Mangubhai, 2011). Islands in the central equatorial Pacific may face a double risk during El Niño: this region tends to be strongly influenced by El Niño–

associated warming, and corals in warmer regions (measured by long‐term mean SST, which is typically highest near the equator) may be more susceptible to heat stress (Claar et al., 2018). This variability in susceptibility may be due to adaptation of higher‐latitude reefs to seasonal temperature variability (Donner, 2011), since equatorial reefs generally experience relatively stable thermal climates on an intra‐annual scale and therefore may be less prepared for episodic temperature extremes associated with El Niño. Central Pacific reefs experienced El Niño–associated bleaching conditions 10 times between 1960 and 2016, with the 2015–2016 El Niño event being unprecedented in magnitude during that time (Barkley et al., 2018). The 2009–2010 El Niño impacted reefs across the Pacific Ocean and Caribbean Sea. Southeast Asian reefs were among the most impacted, losing 18% of their coral during the 2010 bleaching event (Tun et  al., 2010). The 2009–2010 El Niño also caused repeat bleaching at several locations that had been affected by previous El Niño events. In Lakshadweep (Indian Ocean), reefs that bleached during the previous 1997–1998 and the 2009– 2010 El Niño events were more resistant to bleaching and mortality but slower to recover after the latter event (Yadav et al., 2018). The largest global bleaching event to date was associated with the 2015–2016 El Niño. During this event, some locations exceeded a thermal stress threshold (24°C‐ weeks) that was not expected to occur on any reef until approximately 2050 (Hoegh‐Guldberg, 2011), reaching an unprecedented 35°C‐weeks on Jarvis Island in the central Pacific (Brainard et  al., 2018). Heat stress persisted in the central Pacific for more than a year, making 2015–2016 the longest duration thermal stress measured to date. Although the central Pacific accumulated the largest amount of heat stress during this event, intense warming occurred in many other regions as well (Claar et  al., 2018). In Australia, the 2015–2016 El Niño catastrophically transformed nearly one third of the Great Barrier Reef (29% of 3,863 individual reefs), dramatically diminishing coral communities (Hughes et  al., 2018a). Researchers are currently integrating and analyzing coral bleaching data from this event, and upcoming research will likely give us an even clearer understanding of how coral reefs are influenced by a combination of El Niño and anthropogenic warming. With long recovery times and decreasing return times between coral bleaching events (Hughes et al., 2018b), El Niño–associated warming paints a grim picture for the future of coral reefs (Langlais et  al., 2017). However, some corals may be able to acclimate or adaptively respond to warming events, making them better prepared for the next bleaching event (Guest et al., 2012; Palumbi et  al., 2014; Dziedzic et  al., 2019). With coral reef conservation in mind, further research is needed to deter-

ENSO-Driven Ocean Extremes and Their Ecosystem Impacts  419

Figure 18.4  Photo images of coral bleaching and mortality in the central equatorial Pacific (Kiritimati Island, Kiribati) during the 2015–2016 El Niño event. Top panel shows a bleached coral colony (Porites) in the center of the photo as well as widespread mortality of nearly all reef‐building coral colonies in March 2016. Bottom panel shows the start of coral bleaching in July 2015 for large plating corals (Acropora), and variability in bleaching severity from entirely bleached to only moderately affected colonies. (Photo credits: Danielle Claar [top] and Kristina Tietjen [bottom], Baum Lab, University of Victoria)

mine how ENSO variability influences coral survival, as well as when, and why, some corals adapt to El Niño. 18.3.2. Kelp Ecosystems 18.3.2.1. Background Kelps are photosynthetic macroalgae belonging to the group generally known as seaweeds. Seaweeds are sensitive to environmental change because they are sessile, with

limited propagule dispersal, and sensitive to temperature (Wahl et  al., 2015). As they require sunlight, seaweeds cannot simply avoid warming waters by retreating into deeper (cooler but darker) waters. Kelps are the best known of the seaweeds, as they form large forests that may rise up to 40 meters from the seafloor to the surface (e.g. Macrocystis) or form dense meter‐high canopies over rocky reefs (e.g. Ecklonia, Laminaria). The more than 100 species of kelps are

420  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a) 2010

30°N

2007 15°N

1998

0°

1995

El Niño Year

2003

1992

15°S

1987 30°S 0°

120°E

180°W

120°W

60°W

0°

1983 2016

30°N

2010 2007

15°N

2003 1998

0°

1995 15°S

El Niño Year

(b)

60°E

1992 1987

30°S 0°

60°E

120°E

180°W

120°W

60°W

0°

1983

Figure 18.5 El Niño events with the greatest heat stress. Both panels show which El Niño event caused the greatest maximum degree‐heating weeks for each area. Note that this figure does not demonstrate bleaching response, only maximum cumulative heat stress per El Niño event. The events are color‐coded by year. The 1997–1998 El Niño event was the most severe event in the eastern Pacific around the South American coast. (a) All El Niño events from 1982 to 2010, showing how much heterogeneity there is in the geographic distribution of the most extreme heat stress. (b) All El Niño events since 1982, including the 2015–2016 El Niño event, dem­ onstrating the coral heat stress homogenization that occurred during this most recent warming event. (Figure and caption reprinted from Claar et al., 2018)

­ istributed along temperate coastlines worldwide, with a d limited number of species also occurring in subpolar waters of both hemispheres (Graham et al., 2007). These forests are home to a diverse range of species, including commercially important abalone, rock lobster, and other shellfish and provide a wide range of other ecosystem services (Bennett et al., 2016). Loss of kelps from rocky reefs subsequently leads to the absence of hundreds of associated species (Ling et al., 2008; Vergés et al., 2014). Seaweeds such as kelps directly respond to changes in variables such as temperature and nutrients that are affected by ENSO events. Kelps are indirectly affected by changes in other ecosystem processes that are affected by ENSO events, such as competition, consumption, and fouling by other organisms that grow on their fronds (Vega et al., 2005; Wahl et al., 2015). Warm surface waters are generally depleted in nutrients, and as kelps have

limited storage of nutrients, effects are felt after short periods of above‐average temperatures (Graham et  al., 2007). Recovery of kelps following El Niño events occurs relatively quickly in California (Graham et  al., 2007). However, long‐term declines in Tasmania have been attributed to anthropogenic warming, increased presence of low‐nutrient waters along the coast (Wahl et al., 2015), and the influx of herbivores from lower latitudes (Ling et al., 2008). 18.3.2.2. Impacts from ENSO-Related Ocean Extremes The influence of ENSO events on rocky reefs and their kelp forests is clear in many regions of the world. Best known are the El Niño–related changes in temperature and nutrient availability that dramatically reduced the size of Macrocystis kelp forests in California during the 1982– 1983 event (Dayton & Tegner, 1984). In southern

ENSO-Driven Ocean Extremes and Their Ecosystem Impacts  421

California, for example, it has been shown that kelp growth becomes nutrient limited below approximately 1 μM nitrate, which typically occurs when water temperatures rise above 16°C (Graham et al., 2007). El Niño events lead to depression of the thermocline off California, which shuts down nutrient replenishment via coastal upwelling. Dramatic loss of these kelp forests occurs during major El Niño events (Tegner & Dayton, 1987), with mortality of 100% reported in many Macrocystis forests in southern and Baja California following the 1982–1983 and 1997– 1998 El Niño events (Graham et al., 2007). Loss of kelp forests can also be exacerbated by large storms associated with these extreme events (Dayton & Tegner, 1984). Other oceanographic processes can limit the impact of El Niños. In northern Chile, forests of Lessonia trabecu­ lata and Macrocystis integrifolia were maintained during the 1997–1998 El Niño due to persistence of coastal upwelling (Vega et  al., 2005). These cells of coastal upwelling along a coastline can transport kelp propagules following an El Niño and repopulate areas that have suffered local kelp extinctions. The opposite conditions, while expected to be favorable for kelps, can also bring surprises. In the same region of northern Chile, intensification of upwelling associated with the 1998–2000 La Niña led to increased surface nutrient availability. However, increased abundance of a grazing sea urchin species led to local extinction and range reduction for both species of kelps (Vega et  al., 2005). The examples show that both direct and indirect effects play a role in mediating the effect of ENSO phases, and generality across space and time is limited (Wahl et al., 2015). Kelp forests have been shown to be resilient to disturbance from El Niño warming in the eastern Pacific (Tegner et  al., 1997), in that kelp forests recover after these events. However, long‐term warming and extreme events have led to dramatic changes in their distribution and abundance off southern Australia (Johnson et  al., 2011; Wernberg et al., 2011). Increased frequency of warm ocean extremes coupled with reduced surface nutrients is likely to lead to widespread decline of kelp forests and their associated communities (Wernberg et al., 2016). 18.3.3. Seagrass Ecosystems 18.3.3.1. Background Seagrasses form a critical component of many nearshore environments, helping to stabilize sediments and store carbon. The approximately 70 species of seagrass are widely distributed around the globe but are at their most extensive and diverse in Australia (Poloczanska et al., 2007). They occur in shallow marine waters such as estuaries, protected bays, lagoons, and reef platforms protected from strong water movement, but also in deeper

waters (to 70 m) in areas where water clarity is high (Connolly, 2012). The critical factors for seagrass growth are light, temperature, CO2, nutrients, and suitable substrate. Seagrasses living in shallow waters are subject to wide temperature fluctuations seasonally and interannually. They are considered more vulnerable to changes in water quality than changes in temperature, including salinity and sediments that directly smother plants or indirectly reduce the availability of light (Connolly, 2012). They provide food for megafauna such as turtles and dugongs, as well as critical habitat for birds and recreationally and commercially important fish and other species (Waycott et al., 2009). There is limited evidence regarding environmental drivers of seagrass dynamics, due in part to the lack of long time series. Recent extreme events have shown that, like other coastal habitats, seagrasses and their associated communities are also affected. 18.3.3.2. Impacts from ENSO-Related Ocean Extremes A significant impact of ENSO on seagrasses is associated with heavy rainfall events which can lead to turbid flood plumes that reduce available light for seagrasses. Elevated levels of nutrients transported to the nearshore environment by floods can promote phytoplankton and epiphyte growth, further limiting the light and oxygen available to seagrasses (McKenzie et  al., 2012). For example, during the strong 2010–2011 La Niña, a series of tropical cyclones crossed the coast of Queensland, Australia (Hodgkinson et al., 2014). This, in combination with monsoonal conditions and generally heavy rainfall, resulted in record river flows and large sediment inputs over the seagrass beds of the adjacent Great Barrier Reef, in addition to the direct loss of seagrass beds in the path of the cyclones. Severe declines in seagrasses were reported, with >70% of all seagrass beds declining by >20% and seagrass condition in all regions rated as “very poor” and at historically low levels (McKenzie et  al., 2014). Following these floods, there were high levels of mortality of the seagrass‐dependent green turtles and dugongs (Meager & Limpus, 2014). This long‐term analysis showed that peak mortality of dugongs (and inshore dolphins) followed sustained periods of elevated freshwater discharge and low air temperature, with a strong relationship between annual mortality and the Southern Oscillation index (Meager & Limpus, 2014), an often‐used indicator of ENSO strength. On the other side of Australia, the marine heatwave in 2011 (La Niña conditions) in Western Australia resulted in damage to about 36% of Shark Bay’s seagrass meadows (Arias‐Ortiz et al., 2018). These losses in seagrass habitat were estimated to lead to between 2 and 9 Tg of CO2 released to the atmosphere during the following 3 years,

422  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

increasing emissions from land‐use change in Australia by 4% to 21% per annum. The reduction in seagrass habitat and quality corresponded with a decline in the health status of largely herbivorous green turtles (Chelonia mydas) over the following 2 years, providing evidence of long‐ term, community‐level impacts of the event (Thomson et al., 2015). The impact of El Niño and La Niña on seagrass beds is an indirect one, based on flooding and sediment loads in coastal waters rather than direct temperature effects. Marine heatwaves have nevertheless led to dramatic declines in seagrass meadows. When compounded by other stressors, these events cause extensive seagrass mortality, with subsequent starvation for species that feed on seagrass (Thomson et al., 2015). 18.3.4. Mangrove Ecosystems 18.3.4.1. Background Tropical coasts around the world support mangrove forests which provide a wide range of ecosystem services (Halpern et al., 2008). Mangrove forests are responsible for major components of the primary productivity of coastal habitats that support a diverse assemblage of fish and fisheries. The physical structure of mangrove forests provides both shelter and a stable substrate for flora and fauna, as well as increasing soil stability on which subterranean communities of fish, crustaceans, and detritus recyclers rely (Nagelkerken et al., 2008). The poleward limit of mangrove distribution is set by low temperatures, and mangroves are absent from coastal areas where mean temperatures in winter months fall below 4°C. They extend poleward where sufficiently warm ocean currents permit and frost damage is reduced. Mangrove forests have highest diversity in the wet tropics, with diversity decreasing on temperate coasts and in arid regions (Lovelock & Skilleter, 2012). Mangrove forests require fresh water for survival. Although they live in saline waters, many mangrove species live close to their salinity tolerance levels. Lowered sea levels, increased salinity, and decreases in freshwater runoff can lead to deaths of individual plants and stands of mangroves (Lovelock et al., 2017). 18.3.4.2. Impacts from ENSO-Related Ocean Extremes Impacts of ENSO events on mangroves have received relatively little attention until recent times. An earlier study by Drexler and Ewel (2007) showed that a mangrove ecosystem in Micronesia responded negatively to the 1997–1998 El Niño–related drought and increased salinity in this coastal region, and that mangrove forest structure and functioning may be potentially affected by such disturbances for repeated drought cycles. Further, the study shows that mangrove ecosystems are vulnerable to impacts from such short‐term climatic fluctuations.

Most recently, both the role of ENSO and climate change have been considered for mangrove areas around Australia (Lovelock et al., 2017) and the Pacific coast of Columbia (Riascos et  al., 2018). Based on a study of mangrove seedlings and forests of Colombia, Riascos et  al. (2018) suggest that mangrove forests might be expected to show little change in regions where precipitation is projected to increase. Additionally, they suggest that climatic variations associated with ENSO phases may be important for mangrove reproduction, dispersion, and recruitment across the tropical eastern Pacific, including Panama, Costa Rica, and Ecuador (Riascos et al., 2018). The 2015–2016 El Niño was associated with dramatic dieback of mangroves across more than 1000 km in the Gulf of Carpentaria, Australia (Lovelock et al., 2017; Babcock et al., 2019). The dieback affected up to 6% of the overall Gulf of Carpentaria mangrove community, and in some regions the dieback affected up to 26% of the mangrove stands, with 100% mortality in some of these stands (Babcock et al., 2019). The dieback was attributed to both abnormally low El Niño–related sea level, which was compounded by lower‐than‐average rainfall over recent years, and record high temperatures in the preceding 6 months (Duke et al., 2017). Modeling analysis demonstrates that the 2015–2016 El Niño was the dominant factor affecting the extremely low sea level in the Gulf at the time (Harris et al., 2017). It is likely that the mangrove forests suffered from a combination of soil moisture stress, loss of monthly inundation of the upper intertidal zone during neap tides, and abnormally hot temperatures over an extended period. The combination of global warming and extreme events is likely to threaten mangrove habitats due to both direct and indirect effects. Mangrove areas killed or weakened during El Niño events may be subject to additional erosion as a result of tidal action and tropical cyclones (Duke et  al., 2017). Mangrove forests sequester large amounts of carbon, which could be liberated to the atmosphere following diebacks and thus act as a positive feedback for climate warming. 18.4. DISCUSSION AND CONCLUSIONS This chapter has reviewed current knowledge regarding ENSO‐related extremes in sea level and ocean temperature and their impacts on the shallow‐water marine ecosystem foundation species of coral, kelp, seagrass, and mangrove. As the dominant global mode of interannual climate variability, ENSO plays a critical role in sea level and sea temperature fluctuations, influencing marine ­ecosystems. These ocean extremes may be directly related to dynamical processes in the tropical Pacific or the consequence of atmospheric or oceanic teleconnections

ENSO-Driven Ocean Extremes and Their Ecosystem Impacts  423

and modulated regional processes in the extratropics or other basins. Finally, climate change represents an additional “wicked” dimension that is already increasing sea level extremes and marine heatwave frequency and intensity, and may also be affecting ENSO variability. In combination with large‐scale natural ENSO variations, climate change influences on ENSO, and current and future projected coastal population growth, we fully expect that marine ecosystems will continue to experience unprecedented environmental pressures and impacts. This review highlights the following: ••ENSO plays a critical role in both modulating and triggering marine heatwaves and sea level extremes, which impact shallow‐water marine ecosystems through their exposure to thermal stress, sea level change and coastal inundation. ••Tropical coral ecosystems live near the upper limits of their thermal tolerances. As such, exposure to relatively mild but persistent marine heatwaves can cause bleaching: i.e. when coral expel their symbiotic algae (zooxanthellae; Symbiodiniaceae), potentially leading to mortality. ENSO events play a substantial role in the prevalence of marine heatwaves globally, with the most severe coral bleaching events typically occurring during El Niño event periods. With current and projected global warming trends, increased ocean temperatures and extremes pose a critical threat to the future of coral reefs around the world. ••Kelp ecosystems respond directly to changes in temperature and nutrients, both factors that are influenced by ENSO events. Increased frequency of marine heatwaves coupled with reduced surface nutrients is likely to result in widespread decline of kelp forests and their associated communities. ••Seagrass ecosystems tend to be indirectly but significantly impacted by El Niño/La Niña due to heavy rainfall events, freshwater river flood plumes and increased sediment loads, and resultant lowered salinity and light levels. ••Mangrove ecosystems exist close to their salinity tolerance levels. While mangroves live in saline coastal waters, they also require fresh water for their survival. Studies of mangrove ecosystems in Micronesia, the Pacific coast of Columbia, and Australia have shown that mangrove ecosystems are sensitive to ENSO variations via local changes in temperature, rainfall, and sea level. ••Impacts of ENSO through extremes in sea level and ocean temperature are becoming more severe due to background warming and sea‐level rise. ACKNOWLEDGMENTS We would like to gratefully acknowledge the two anonymous reviewers for their very constructive comments and suggestions which helped to improve this book chapter. N.J.H. acknowledges funding received from the ARC

Centre of Excellence for Climate Extremes (CE170100023) and the NESP Earth Systems and Climate Change Hub. D.C.C. acknowledges support from the NOAA Climate and Global Change Postdoctoral Fellowship Program, administered by UCAR’s Cooperative Programs for the Advancement of Earth System Science (CPAESS) under award #NA18NWS4620043B. E.C.J.O. acknowledges support from the National Sciences and Engineering Research Council of Canada Discovery Grant RGPINand from the Marine Environmental 2018-05255 ​ Observation, Prediction and Response Network (MEOAR) project “Drivers, predictability and fisheries impacts of ocean temperature extremes”. M.J.W. was supported by NOAA grant NA17OAR4310110. X.Z. was supported by the Centre for Southern Hemisphere Oceans Research (CSHOR), a joint research center between QNLM and CSIRO. We would like to thank Jules Kajtar and Axel Durand for assistance with reference formatting. REFERENCES Aeby, G. S., Kenyon, J. C., Maragos, J. E., & Potts, D. C. (2003). First record of mass coral bleaching in the Northwestern Hawaiian Islands. Coral Reefs, 22(3), 256. Ainsworth, T. D., Heron, S. F., Ortiz, J. C., Mumby, P. J., Grech, A., Ogawa, D., et  al. (2016). Climate change disables coral bleaching protection on the Great Barrier Reef. Science, 352(6283), 338–342. https://doi.org/10.1126/science.aac7125 Arias‐Ortiz, A., Serrano, O., Masqué, P., Lavery, P. S., Mueller, U., Kendrick, G. A., et al. (2018). A marine heatwave drives massive losses from the world’s largest seagrass carbon stocks. Nature Climate Change, 8, 338–344. https://doi.org/10.1038/ s41558‐018‐0096‐y Aucan, J., Hoeke, R. K., Storlazzi, C. D., Stopa, J., Wandres, M., & Lowe, R. (2019). Waves do not contribute to global sea‐level rise. Nature Climate Change, 9(1), 2. https://doi. org/10.1038/s41558‐018‐0377‐5 Babcock, R. C., Bustamante, R. H., Fulton, E. A., Fulton, D. J., Haywood, M. D. E., Hobday, A. J., et al. (2019). Severe and extensive climate change impacts are happening now: Extreme climate events impact marine habitat forming communities along 45% of Australia’s coast. Frontiers in Marine Science, 6, Article 411. doi:10.3389/fmars.2019.00411 Baker, A. C. (2003). Flexibility and specificity in coral‐algal symbiosis: Diversity, ecology, and biogeography of Symbiodinium. Annual Review of Ecology, Evolution, and Systematics, 34(1), 661–689. https://doi.org/10.1146/annurev. ecolsys.34.011802.132417 Baker, D. M., Freeman, C. J., Wong, J. C. Y., Fogel, M. L., & Knowlton, N. (2018). Climate change promotes parasitism in a coral symbiosis. ISME J., 12(3), 921–930. Barkley, H. C., Cohen, A. L., Mollica, N. R., Brainard, R. E., Rivera, H. E., DeCarlo, T. M., et al. (2018). Repeat bleaching of a central Pacific coral reef over the past six decades (1960– 2016). Communications Biology, 1. https://doi.org/10.1038/ s42003‐018‐0183‐7

424  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Barnard, P. L., Hoover, D., Hubbard, D. M., Snyder, A., Ludka, B. C., Allan, J., et al. (2017). Extreme oceanographic forcing and coastal response due to the 2015‐2016 El Niño. Nature Communications, 8. https://doi.org/10.1038/ncomms14365 Barnard, P. L., Short, A. D., Harley, M. D., Splinter, K. D., Vitousek, S., Turner, I. L., et al. (2015). Coastal vulnerability across the Pacific dominated by El Niño/Southern Oscillation. Nature Geoscience, 8, 801–807. https://doi.org/10.1038/ ngeo2539 Bennett, S., Wernberg, T., Connell, S. D., Hobday, A. J., Johnson, C. R., & Poloczanska, E. S. (2016). The “Great Southern Reef ”: Social, ecological and economic value of Australia’s neglected kelp forests. Marine and Freshwater Research, 67(1), 47–56. https://doi.org/10.1071/MF15232 Benthuysen, J., Feng, M., & Zhong, L. (2014). Spatial patterns of warming off Western Australia during the 2011 Ningaloo Niño: Quantifying impacts of remote and local forcing. Continental Shelf Research, 91, 232–246. https://doi. org/10.1016/j.csr.2014.09.014 Bond, N. A., Cronin, M. F., Freeland, H., & Mantua, N. (2015). Causes and impacts of the 2014 warm anomaly in the NE Pacific. Geophysical Research Letters, 42(9), 3414–3420. https://doi.org/10.1002/2015GL063306 Brainard, R. E., Oliver, T., McPhaden, M. J., Cohen, A., Venegas, R., Heenan, A., et al. (2018). Ecological impacts of the 2015/16 El Niño in the central equatorial Pacific. Bulletin of the American Meteorological Society, 99(1), S21–S26. Brown, B. E. (1997). Coral bleaching: Causes and consequences. Coral Reefs, 16, S129–S138. https://doi.org/10.1007/ s003380050249 Brown, B. E., & Suharsono. (1990). Damage and recovery of coral reefs affected by El Niño related seawater warming in the Thousand Islands, Indonesia. Coral Reefs, 8(4), 163–170. https://doi.org/10.1007/BF00265007 Cai, W., Santoso, A., Wang, G., Yeh, S. W., An, S. Il, Cobb, K. M., et al. (2015). ENSO and greenhouse warming. Nature Climate Change. https://doi.org/10.1038/nclimate2743 Camargo, S. J., Emanuel, K. A., & Sobel, A. H. (2007). Use of a genesis potential index to diagnose ENSO effects on tropical cyclone genesis. Journal of Climate, 20(19), 4819–4834. https://doi.org/10.1175/JCLI4282.1 Caputi, N., Kangas, M., Denham, A., Feng, M., Pearce, A., Hetzel, Y., & Chandrapavan, A. (2016). Management adaptation of invertebrate fisheries to an extreme marine heat wave event at a global warming hot spot. Ecology and Evolution, 6(11), 3583–3593. https://doi.org/10.1002/ece3.2137 Chand, S. S., & Walsh, K. J. E. (2009). Tropical cyclone activity in the Fiji region: Spatial patterns and relationship to large‐ scale circulation. Journal of Climate, 22(14), 3877–3893. https://doi.org/10.1175/2009jcli2880.1 Chand, S. S., McBride, J. L., Tory, K. J., Wheeler, M. C., & Walsh, K. J. E. (2013). Impact of Different ENSO regimes on southwest Pacific tropical cyclones. Journal of Climate, 26(2), 600–608. https://doi.org/10.1175/jcli‐d‐12‐00114.1 Chelton, D. B., & Davis, R. E. (2002). Monthly mean sea‐level variability along the west coast of North America. Journal of Physical Oceanography, 12(8), 757–784. https://doi.org/10.11 75/1520‐0485(1982)0122.0.co;2

Chen, K., Gawarkiewicz, G. G., Lentz, S. J., & Bane, J. M. (2014). Diagnosing the warming of the northeastern U.S. coastal ocean in 2012: A linkage between the atmospheric jet stream variability and ocean response. Journal of Geophysical Research: Oceans, 119, 218–227. https://doi.org/10.1002/ 2013JC009393 Church, J. A., Clark, P. U., Cazenave, A., Gregory, J. M., Jevrejeva, S., Levermann, A., et al. (2013). Sea‐level rise by 2100. Science, 342(6165), 1445. Claar, D. C., Szostek, L., McDevitt‐Irwin, J. M., Schanze, J. J., & Baum, J. K. (2018). Global patterns and impacts of El Niño events on coral reefs: A meta‐analysis. PLoS One, 13(2), e0190957. Connolly, R. (2012). Seagrass. In A Marine Climate Change Impacts and Adaptation Report Card for Australia 2012. https://doi.org/10.1017/CBO9781107415324.004 Dayton, P. K., & Tegner, M. J. (1984). Catastrophic storms, El Niño, and patch stability in a southern California kelp community. Science, 224(4646), 283–285. https://doi. org/10.1126/science.224.4646.283 Delcroix, T., & Rual, P. (1997). On sea level changes in the tropical Pacific during El Niño Southern Oscillation events. Proceedings of the Third SPREP Meeting on Climate Change and Sea Level Rise, Noumea, 104–111. Di Lorenzo, E., & Mantua, N. (2016). Multi‐year persistence of the 2014/15 North Pacific marine heatwave. Nature Climate Change, 6(11), 1042–1047. https://doi.org/10.1038/ nclimate3082 Donner, S. D. (2011). An evaluation of the effect of recent temperature variability on the prediction of coral bleaching events. Ecological Applications, 21(5), 1718–1730. Drexler, J. Z., & Ewel, K. C. (2007). Effect of the 1997‐1998 ENSO‐related drought on hydrology and salinity in a Micronesian wetland complex. Estuaries, 24(3), 347. https:// doi.org/10.2307/1353237 Duke, N. C., Kovacs, J. M., Griffiths, A. D., Preece, L., Hill, D. J. E., Van Oosterzee, P., et al. (2017). Large‐scale dieback of mangroves in Australia’s Gulf of Carpentaria: A severe ecosystem response, coincidental with an unusually extreme weather event. Marine and Freshwater Research, 68(10), 1816–1829. https://doi.org/10.1071/MF16322 Dziedzic, K. E., Elder, H., Tavalire, H., & Meyer, E. (2019). Heritable variation in bleaching responses and its functional genomic basis in reef‐building corals (Orbicella faveolata). Molecular Ecology, 28, 2238–2253. https://doi.org/10.1111/ mec.15081 Eakin, C. M. (2001). A tale of two ENSO events: Carbonate budgets and the influence of two warming disturbances and intervening variability, Uva Island, Panama. Bulletin of Marine Science, 69(1), 171–186. Eakin, C. M., Lough, J. M., Heron, S. F., & Liu, G. (2018). Climate variability and change: Monitoring data and evidence for increased coral bleaching stress. In van Oppen, M. J. H., & Lough, J. M (Eds.), Coral bleaching, patterns, processes, causes and consequences (2nd ed., chap. 4). Springer. Elvan Ampou, E., Johan, O., Menkes, C. E., Niño, F., Birol, F., Ouillon, S., & Andrefouet, S. (2017). Coral mortality induced

ENSO-Driven Ocean Extremes and Their Ecosystem Impacts  425 by the 2015‐2016 El‐Niño in Indonesia: The effect of rapid sea level fall. Biogeosciences, 14(4), 817–826. https://doi. org/10.5194/bg‐14‐817‐2017 Enfield, D. B., & Allen, J. S. (2002). On the structure and dynamics of monthly mean sea level anomalies along the Pacific Coast of North and South America. Journal of Physical Oceanography, 10(4), 557–578. https://doi.org/10.11 75/1520‐0485(1980)0102.0.co;2 Feng, M., McPhaden, M. J., Xie, S. P., & Hafner, J. (2013). La Niña forces unprecedented Leeuwin Current warming in 2011. Scientific Reports, 3, 1277. https://doi.org/10.1038/ srep01277 Feng, X., & Tsimplis, M. N. (2014). Sea level extremes at the coasts of China. Journal of Geophysical Research: Oceans, 119(3), 1593–1608. https://doi.org/10.1002/2013JC009607 Frölicher, T. L., Fischer, E. M., & Gruber, N. (2018). Marine heatwaves under global warming. Nature, 560(7718), 360– 364. https://doi.org/10.1038/s41586‐018‐0383‐9 Gilmour, J. P., Smith, L. D., Heyward, A. J., Baird, A. H., & Pratchett, M. S. (2013). Recovery of an isolated coral reef system following severe disturbance. Science, 340, 69–71. https://doi.org/10.1126/science.1232310 Gleeson, M. W., & Strong, A. E. (1995). Applying MCSST to coral reef bleaching. Advances in Space Research, 16(10), 151–154. Glynn, P. W. (1983). Extensive ‘bleaching’ and death of reef corals on the Pacific coast of Panamá. Environmental Conservation, 10, 149–154. https://doi.org/10.1017/S037689 2900012248 Glynn, P. W. (1984). Widespread coral mortality and the 1982‐83 El Niño warming event. Environmental Conservation, 11(02), 133–146. Glynn, P. W. (1988). El Niño Southern Oscillation 1982‐1983: Nearshore population, community, and ecosystem responses. Annual Review of Ecology and Systematics, 19, 309–345. Glynn, P. W. (1990). Global ecological consequences of the 1982‐83 El Niño‐Southern Oscillation. Elsevier. Glynn, P. W., & D’Croz, L. (1990). Experimental evidence for high temperature stress as the cause of El Niño‐coincident coral mortality. Coral Reefs, 8(4), 181–191. https://doi. org/10.1007/BF00265009 Glynn, P. W., & de Weerdt, W. (1991). Elimination of two reef‐ building hydrocorals following the 1982‐83 El Niño warming event. Science, 253(5015), 69–71. https://doi.org/10.1126/ science.253.5015.69 Graham, M., Buschmann, A., & Vasquez, J. (2007). Global ecology of the giant kelp Macrocystis. Oceanography and Marine Biology, 45, 39–88. https://doi.org/10.1201/97814200 50943.ch2 Griffies, S. M., Winton, M., Anderson, W. G., Benson, R., Delworth, T. L., Dufour, C. O., et  al. (2015). Impacts on ocean heat from transient mesoscale eddies in a hierarchy of climate models. Journal of Climate, 28(3), 952–977. https:// doi.org/10.1175/JCLI‐D‐14‐00353.1 Guest, J. R., Baird, A. H., Maynard, J. A., Muttaqin, E., Edwards, A. J., Campbell, S. J., et al. (2012). Contrasting patterns of coral bleaching susceptibility in 2010 suggest an adaptive response to thermal stress. PLoS ONE, 7(3), e33353–e33353. https://doi.org/10.1371/journal.pone.0033353

Guzman, H. M., & Cortes, J. (2006). Reef recovery 20 years after the 1982‐1983 El Niño massive mortality. Marine Biology, 151(2), 401–411. https://doi.org/10.1007/s00227‐006‐0495‐x Halpern, B. S., Walbridge, S., Selkoe, K. A., Kappel, C. V., Micheli, F., D’Agrosa, C., et  al. (2008). A global map of human impact on marine ecosystems. Science, 319(5865), 948–952. https://doi.org/10.1126/science.1149345 Hamlington, B. D., Leben, R. R., Kim, K. Y., Nerem, R. S., Atkinson, L. P., & Thompson, P. R. (2015). The effect of the El Niño‐Southern Oscillation on U.S. regional and coastal sea level. Journal of Geophysical Research: Oceans, 120, 3970–3986. https://doi.org/10.1002/2014JC010602 Harley, M. D., Turner, I., Short, A., & Ranasinghe, R. (2011). A re‐evaluation of coastal embayment rotation in SE Australia: The dominance of cross‐shore versus alongshore sediment transport processes. Journal of Geophysical Research, 116(F4). https://doi.org/doi:10.1029/2011JF001989 Harris, T., Hope, P., Oliver, E., Smalley, R., Arblaster, J., Holbrook, N., et al. (2017). Climate drivers of the 2015 Gulf of Carpentaria mangrove dieback, NESP ESCC Hub Report No. 2. Hemer, M. A., Church, J. A., & Hunter, J. R. (2010). Variability and trends in the directional wave climate of the Southern Hemisphere. International Journal of Climatology, 30(4), 475–491. https://doi.org/doi:10.1002/joc.1900 Heron, S. F., Maynard, J. A., van Hooidonk, R., & Eakin,  C. M. (2016). Warming trends and bleaching stress of the world’s coral reefs 1985‐2012. Scientific Reports, 6, 38402. https:// doi.org/10.1038/srep38402 Hobday, A. J., Alexander, L. V., Perkins, S. E., Smale, D. A., Straub, S. C., Oliver, E. C. J., (2016). A hierarchical approach to defining marine heatwaves. Progress in Oceanography, 141, 227–238. https://doi.org/10.1016/j.pocean.2015.12.014 Hobday, A., Oliver, E., Sen Gupta, A., Benthuysen, J., Burrows, M., Donat, M., et al. (2018). Categorizing and naming marine heatwaves. Oceanography, 31(2), 161–173. https:// doi.org/10.5670/oceanog.2018.205 Hodgkinson, J. H., Hobday, A. J., & Pinkard, E. A. (2014). Climate adaptation in Australia’s resource‐extraction industries: Ready or not? Regional Environmental Change, 14(4), 1663–1678. https://doi.org/10.1007/s10113‐014‐0618‐8 Hoegh‐Guldberg, O. (2011). Coral reef ecosystems and anthropogenic climate change. Regional Environmental Change Change, 11(1), 215–227. https://doi.org/10.1007/s10113‐010‐0189‐2 Hoegh‐Guldberg, O., Mumby, P. J., Hooten, A. J., Steneck, R. S., Greenfield, P., Gomez, E., et al. (2007). Coral reefs under rapid climate change and ocean acidification. Science, 318(5857), 1737–1742. https://doi.org/10.1126/science.1152509 Hoeke, R. K., McInnes, K. L., Kruger, J. C., McNaught, R. J., Hunter, J. R., & Smithers, S. G. (2013). Widespread inundation of Pacific islands triggered by distant‐source wind‐waves. Global and Planetary Change, 108, 128–138. https://doi. org/10.1016/j.gloplacha.2013.06.006 Hoeke, R., McInnes, K., & O’Grady, J. (2015). Wind and wave setup contributions to extreme sea levels at a tropical high island: A stochastic cyclone simulation study for Apia, Samoa. Journal of Marine Science and Engineering, 3(3), 1117–1135. https://doi.org/10.3390/jmse3031117

426  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Holbrook, N. J., Scannell, H. A., Sen Gupta, A., Benthuysen, J. A., Feng, M., Oliver, E. C. J., et al. (2019). A global assessment of marine heatwaves and their drivers. Nature Communications, 10 (2624). https://doi.org.10.1038/s41467‐019‐10206‐z Hughes, T. P., Baird, A. H., Bellwood, D. R., Card, M., Connolly, S. R., Folke, C., et  al. (2003). Climate change, human impacts, and the resilience of coral reefs. Science, 301(5635), 929–933. https://doi.org/10.1126/science.1085046 Hughes, T. P., Kerry, J. T., Álvarez‐noriega, M., Álvarez‐ romero, J. G., Anderson, K. D., Baird, A. H., et al. (2017). Global warming and recurrent mass bleaching of corals. Nature, 543, 373–378. https://doi.org/10.1038/nature21707 Hughes, T. P., Kerry, J. T., Baird, A. H., Connolly, S. R., Dietzel, A., Eakin, C. M., et al. (2018a). Global warming transforms coral reef assemblages. Nature, 556, 492–496. Hughes, T. P., Anderson, K. D., Connolly, S. R., Heron, S. F., Kerry, J. T., Lough, J. M., et al. (2018b). Spatial and temporal patterns of mass bleaching of corals in the Anthropocene. Science, 359(6371), 80–83. Hughes, T. P., Kerry, J. T., Connolly, S. R., Baird, A. H., Eakin, C. M., Heron, S. F., et al. (2019). Ecological memory modifies the cumulative impact of recurrent climate extremes. Nature Climate Change, 9, 40–43. https://doi.org/10.1038/ s41558‐018‐0351‐2 Johnson, C. R., Banks, S. C., Barrett, N. S., Cazassus, F., Dunstan, P. K., Edgar, G. J., et al. (2011). Climate change cascades: Shifts in oceanography, species’ ranges and subtidal marine community dynamics in eastern Tasmania. Journal of Experimental Marine Biology and Ecology, 400, 17–32. https://doi.org/10.1016/j.jembe.2011.02.032 Kelmo, F., & Attrill, M. J. (2013). Severe impact and subsequent recovery of a coral assemblage following the 1997‐8 El Niño event: A 17‐year study from Bahia, Brazil. PLoS One, 8(5), e65073. LaJeunesse, T. C., Parkinson, J. E., Gabrielson, P. W., Jeong, H. J., Reimer, J. D., Voolstra, C. R., & Santos, S. R. (2018). Systematic revision of Symbiodiniaceae highlights the antiquity and diversity of coral endosymbionts. Current Biology, 28(16), 2570–2580. Langlais, C. E., Lenton, A., Heron, S. F., Evenhuis, C., Sen Gupta, A., Brown, J. N., & Kuchinke, M. (2017). Coral bleaching pathways under the control of regional temperature variability. Nature Climate Change, 7(11), 839. Lee, T., Hobbs, W. R., Willis, J. K., Halkides, D., Fukumori, I., Armstrong, E. M., et al. (2010). Record warming in the South Pacific and western Antarctica associated with the strong central‐Pacific El Niño in 2009‐10. Geophysical Research Letters, 37, 1–6. https://doi.org/10.1029/2010GL044865 Ling, S. D., Johnson, C. R., Frusher, S., & King, C. K. (2008). Reproductive potential of a marine ecosystem engineer at the edge of a newly expanded range. Global Change Biology, 14(4), 907–915. https://doi.org/10.1111/j.1365‐2486.2008.01543.x Liu, G., Heron, S. F., Eakin, C. M., Muller‐Karger, F. E., Vega‐Rodriguez, M., Guild, L. S., et  al. (2014). Reef‐scale thermal stress monitoring of coral ecosystems: New 5‐km bal products from NOAA coral reef watch. Remote glo­ Sensing, 6(11), 11579–11606. https://doi.org/10.3390/ rs61111579

Liu, Z.‐J., Minobe, S., Sasaki, Y. N., & Terada, M. (2016). Dynamical downscaling of future sea level change in the western North Pacific using ROMS. Journal of Oceanography, 72(6), 905–922. https://doi.org/10.1007/s10872‐016‐0390‐0 Lough, J. M., Anderson, K. D., & Hughes, T. P. (2018). Increasing thermal stress for tropical coral reefs: 1871–2017. Scientific Reports, 8(1), 6079. Lovelock, C.E., G. Skilleter, N. S. (2012). Tidal wetlands. In Hobday, A. J., Richardson, A. J., & Poloczanska, E. S. (Eds.), Marine Climate Change Impacts and Adaptation Report Card for Australia. Lovelock, C. E., Feller, I. C., Reef, R., Hickey, S., & Ball, M. C. (2017). Mangrove dieback during fluctuating sea levels. Scientific Reports, 7(1). https://doi.org/10.1038/s41598‐017‐ 01927‐6 McClanahan, T. R. (2004). The relationship between bleaching and mortality of common corals. Marine Biology, 144(6), 1239–1245. McInnes, K. L., Walsh, K. J. E., Hoeke, R. K., O’Grady, J. G., Colberg, F., & Hubbert, G. D. (2014). Quantifying storm tide risk in Fiji due to climate variability and change. Global and Planetary Change, 116, 115–129. https://doi.org/https://doi. org/10.1016/j.gloplacha.2014.02.004 McInnes, K. L., Hoeke, R. K., Walsh, K. J. E., O’Grady, J. G., & Hubbert, G. D. (2016). Application of a synthetic cyclone method for assessment of tropical cyclone storm tides in Samoa. Natural Hazards, 80(1), 425–444. https://doi. org/10.1007/s11069‐015‐1975‐4 McKenzie, L. J., Collier, C., & Waycott, M. (2012). Reef Resource Marine Monitoring Program: Inshore seagrass, annual report for the sampling period 1st July 2010 – 31st May 2011. Cairns. McKenzie, L. J., Collier, C., & Waycott, M. (2014). Reef Rescue Marine Monitoring Program: Inshore seagrass, annual report for the sampling period 1st July 2011  –  31st May 2012. Cairns. Meager, J. J., & Limpus, C. (2014). Mortality of inshore marine mammals in eastern Australia is predicted by freshwater discharge and air temperature. PLoS ONE, 9(4). https://doi. org/10.1371/journal.pone.0094849 Menéndez, M., & Woodworth, P. L. (2010). Changes in extreme high water levels based on a quasi‐global tide‐gauge data set. Journal of Geophysical Research: Oceans, 115(10). https://doi. org/10.1029/2009JC005997 Mentaschi, L., Vousdoukas, M. I., Voukouvalas, E., Dosio, A., & Feyen, L. (2017). Global changes of extreme coastal wave energy fluxes triggered by intensified teleconnection patterns. Geophysical Research Letters, 44, 2416–2426. https://doi. org/10.1002/2016GL072488 Merrifield, M., Kilonsky, B., & Nakahara, S. (1999). Interannual sea level changes in the tropical Pacific associated with ENSO. Geophysical Research Letters, 26(21), 3317–3320. https://doi.org/10.1029/1999GL010485 Miles, E. R., Spillman, C. M., Church, J. A., & McIntosh, P. C. (2014). Seasonal prediction of global sea level anomalies using an ocean–atmosphere dynamical model. Climate Dynamics, 43(7–8), 2131–2145. https://doi.org/10.1007/ s00382‐013‐2039‐7

ENSO-Driven Ocean Extremes and Their Ecosystem Impacts  427 Moberg, F., & Folke, C. (1999). Ecological goods and services of coral reef ecosystems. Ecological Economics, 29(2), 215–233. Muscatine, L., & Cernichiari, E. (1969). Assimilation of photosynthetic products of zooxanthellae by a reef coral. The Biological Bulletin, 137(3), 506–523. Muscatine, L., & Porter, J. W. (1977). Reef corals: Mutualistic symbioses adapted to nutrient‐poor environments. Bioscience, 27(7), 454–460. https://doi.org/10.2307/1297526. Nagelkerken, I., Blaber, S. J. M., Bouillon, S., Green, P., Haywood, M., Kirton, L. G., et  al. (2008). The habitat function of mangroves for terrestrial and marine fauna: A review. Aquatic Botany, 89, 155–185. https://doi.org/10.1016/j. aquabot.2007.12.007 NOAA Coral Reef Watch. (2000). NOAA Coral Reef Watch Operational 50‐km satellite coral bleaching degree heating weeks product. Silver Spring, Maryland, USA: NOAA Coral Reef Watch. Obura, D. O., & Mangubhai, S. (2011). Coral mortality associated with thermal fluctuations in the Phoenix Islands, 2002– 2005. Coral Reefs, 30(3), 607–619. Olita, A., Sorgente, R., Natale, S., Gabersek, S., Ribotti, A., Bonanno, A. et al. (2007). Effects of the 2003 European ­heatwave on the Central Mediterranean Sea: surface fluxes and the dynamical response. Ocean Science, 3, 273–289. https://doi.org/10.5194/os-3-273-2007 Oliver, E. C. J., Benthuysen, J. A., Bindoff, N. L., Hobday, A. J., Holbrook, N. J., Mundy, C. N., & Perkins‐Kirkpatrick, S. E. (2017). The unprecedented 2015/16 Tasman Sea marine heat­ wave. Nature Communications, 8. https://doi.org/10.1038/ ncomms16101 Oliver, E. C. J., Donat, M. G., Burrows, M. T., Moore, P. J., Smale, D. A., Alexander, L. V., et  al. (2018a). Longer and more frequent marine heatwaves over the past century. Nature Communications, 9(1). https://doi.org/10.1038/s41467‐018‐ 03732‐9 Oliver, E. C. J., Perkins‐Kirkpatrick, S. E., Holbrook, N. J., & Bindoff, N. L. (2018b). Anthropogenic and natural influences on record 2016 marine heat waves. Bulletin of the American Meteorological Society, 99, S44–S48. https://doi.org/10.1175/ bams‐d‐17‐0093.1 Oliver, J. K., Berkelmans, R., & Eakin, C. M. (2018). Coral bleaching in space and time. In van Oppen, M. J. H., & Lough, J. M (Eds.), Coral bleaching, patterns, processes, causes and consequences (2nd ed, chap. 3). Springer. Palumbi, S. R., Barshis, D. J., Traylor‐Knowles, N., & Bay, R. A. (2014). Mechanisms of reef coral resistance to future climate change. Science, 344(6186). https://doi.org/10.1126/ science.1251336 Pearce, A. F., & Feng, M. (2013). The rise and fall of the “marine heat wave” off Western Australia during the summer of 2010/2011. Journal of Marine Systems, 111–112, 139–156. https://doi.org/10.1016/j.jmarsys.2012.10.009 Perkins‐Kirkpatrick, S. E., King, A. D., Cougnon, E. A., Grose, M. R., Oliver, E. C. J., & Holbrook, N. J., et al. (2019). The role of natural variability and anthropogenic climate change in the 2017/2018 Tasman Sea marine heatwave. Bulletin of the American Meteorological Society, 100, S105–S110.

Poloczanska, E. S., Babcock, R. C., Butler, A., Hobday, A. J., Hoegh‐Guldberg, O., Kunz, T. J., et  al. (2007). Climate change and Australian marine life. Oceanography and Marine Biology, 45, 407. Ranasinghe, R., McLoughlin, R., Short, A., & Symonds, G. (2004). The Southern Oscillation Index, wave climate, and beach rotation. Marine Geology, 204(3), 273–287. https://doi. org/https://doi.org/10.1016/S0025‐3227(04)00002‐7 Riascos, J. M., Cantera, J. R., & Blanco‐Libreros, J. F. (2018). Growth and mortality of mangrove seedlings in the wettest neotropical mangrove forests during ENSO: Implications for vulnerability to climate change. Aquatic Botany, 147, 34–42. https://doi.org/10.1016/j.aquabot.2018.03.002 Rowan, R. (1995). Intraspecific diversity and ecological zonation in coral‐algal symbiosis. Proceedings of the National Academy of Sciences, 92(7), 2850–2853. https://doi.org/10. 1073/pnas.92.7.2850 Saji, N. H., Goswami, B. N., Vinayachandran, P. N., & Yamagata, T. (1999). A dipole mode in the tropical Indian Ocean. Nature, 401(6751), 360–363. https://doi.org/10. 1038/43854 Salvat, B. (1992). Coral reefs: A challenging ecosystem for human societies. Global Environmental Change, 2(1), 12–18. https://doi.org/10.1016/0959‐3780(92)90032‐3 Schlegel, R. W., Oliver, E. C. J., Wernberg, T., & Smit, A. J. (2017). Nearshore and offshore co‐occurrence of marine heatwaves and cold‐spells. Progress in Oceanography, 151, 189–205. https://doi.org/10.1016/j.pocean.2017.01.004 Smale, D. A., Wernberg, T., Oliver, E. C. J., Thomsen, M., Harvey, B. P., Straub, S. C., et al. (2019). Marine heatwaves threaten global biodiversity and the provision of ecosystem services. Nature Climate Change, 9, 306–312. https://doi. org/10.1038/s41558‐019‐0412‐1 Sparnocchia, S., Schiano, M. E., Picco, P., Bozzano, R., & Cappelletti, A. (2006). The anomalous warming of summer 2003 in the surface layer of the Central Ligurian Sea (Western Mediterranean). Annales Geophysicae, 24(2), 443–452. https://doi.org/10.5194/angeo‐24‐443‐2006 Stat, M., & Gates, R. D. (2011). Clade D Symbiodinium in scleractinian corals: A “nugget” of hope, a selfish opportunist, an ominous sign, or all of the above? Journal of Marine Biology, 730715, 1–9. https://doi.org/10.1007/s00227‐009‐ 1263‐5 Stone, L., Rajagopalan, B., Bhasin, H., & Loya, Y. (1999). Mass coral reef bleaching: A recent outcome of increased El Niño activity? Ecology Letters, 2(5), 325–330. https://doi. org/10.1046/j.1461‐0248.1999.00092.x Sweet, W. V., & Park, J. (2014). From the extreme to the mean: Acceleration and tipping points of coastal inundation from sea level rise. Earth’s Future, 2(12), 579–600. https://doi. org/10.1002/2014EF000272 Tegner, M. J., & Dayton, P. K. (1987). El Niño effects on Southern California kelp forest communities. Advances in Ecological Research, 17(C), 243–279. https://doi.org/10.1016/ S0065‐2504(08)60247‐0 Tegner, M. J., Dayton, P. K., Edwards, P. B., & Riser, K. L. (1997). Large‐scale, low‐frequency oceanographic effects on kelp forest succession: A tale of two cohorts. Marine Ecology

428  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Progress Series, 146(1–3), 117–134. https://doi.org/10.3354/ meps146117 Thomson, J. A., Burkholder, D. A., Heithaus, M. R., Fourqurean, J. W., Fraser, M. W., Statton, J., & Kendrick, G. A. (2015). Extreme temperatures, foundation species, and abrupt ecosystem change: An example from an iconic seagrass ecosystem. Global Change Biology, 21(4), 1463–1474. https://doi. org/10.1111/gcb.12694 Tun, K., Chou, L. M., Low, J., Yeemin, T., Phongsuwan, N., Setiasih, N., et al. (2010). Status of Coral Reefs in East Asian Seas Region: 2010. van Hooidonk, R., Maynard, J., Tamelander, J., Gove, J., Ahmadia, G., Raymundo, L., et  al. (2016). Local‐scale projections of coral reef futures and implications of the Paris Agreement. Scientific Reports, 6, 39666. Vega, J. M. A., Vásquez, J. A., & Buschmann, A. H. (2005). Population biology of the subtidal kelps Macrocystis integrifolia and Lessonia trabeculata (Laminariales, Phaeophyceae) in an upwelling ecosystem of northern Chile: Interannual variability and El Niño 1997‐1998. Revista Chilena de Historia Natural, 78(1), 33–50. https://doi.org/10.4067/S0716‐078X 2005000100004 Vega Thurber, R. L., Burkepile, D. E., Fuchs, C., Shantz, A. A., Mcminds, R., & Zaneveld, J. R. (2014). Chronic nutrient enrichment increases prevalence and severity of coral disease and bleaching. Global Change Biology, 20(2), 544–554. https://doi.org/10.1111/gcb.12450 Vergés, A., Steinberg, P. D., Hay, M. E., Poore, A. G. B., Campbell, A. H., Ballesteros, E., et al. (2014). The tropicalization of temperate marine ecosystems: Climate‐mediated changes in herbivory and community phase shifts. Proceedings of the Royal Society B: Biological Sciences, 281(1789), 20140846. Wahl, M., Molis, M., Hobday, A. J., Dudgeon, S., Neumann, R., Steinberg, P., et  al. (2015). The responses of brown ­macroalgae to environmental change from local to global scales: Direct versus ecologically mediated effects. Pers­ pectives in Phycology, 2(1), 11–29. https://doi.org/10.1 127/pip/2015/0019 Walsh, K. J. E., McInnes, K. L., & McBride, J. L. (2012). Climate change impacts on tropical cyclones and extreme sea levels in the South Pacific: A regional assessment. Global and Planetary Change, 80–81, 149–164. https://doi.org/https:// doi.org/10.1016/j.gloplacha.2011.10.006 Waycott, M., Duarte, C. M., Carruthers, T. J. B., Orth, R. J., Dennison, W. C., Olyarnik, S., et al. (2009). Accelerating loss of seagrasses across the globe threatens coastal ecosystems. Proceedings of the National Academy of Sciences, 106(30), 12377–12381. https://doi.org/10.1073/pnas.0905620106 Wernberg, T., Bennett, S., Babcock, R. C., de Bettignies, T., Cure, K., Depczynski, M., et  al. (2016). Climate‐driven regime shift of a temperate marine ecosystem. Science, 353(6295), 169–172. Wernberg, T., Russell, B. D., Thomsen, M. S., Gurgel, C. F. D., Bradshaw, C. J. A., Poloczanska, E. S., & Connell, S. D. (2011). Seaweed communities in retreat from ocean warming.

Current Biology, 21(21), 1828–1832. https://doi.org/10.1016/j. cub.2011.09.028 Wernberg, T., Smale, D. A., Tuya, F., Thomsen, M. S., Langlois, T. J., de Bettignies, T., et al. (2013). An extreme climatic event alters marine ecosystem structure in a global biodiversity hotspot. Nature Climate Change, 3, 78–82. https://doi. org/10.1038/nclimate1627 Widlansky, M. J., Timmermann, A., McGregor, S., Stuecker, M. F., & Cai, W. (2014). An interhemispheric tropical sea level seesaw due to El Niño Taimasa. Journal of Climate, 27, 1070–1081. https://doi.org/10.1175/JCLI‐D‐13‐00276.1 Widlansky, M. J., Timmermann, A., & Cai, W (2015). Future extreme sea level seesaws in the tropical Pacific. Science Advances, 1(8), e1500560. Wiedenmann, J., D’Angelo, C., Smith, E. G., Hunt, A. N., Legiret, F.‐E., Postle, A. D., & Achterberg, E. P. (2012). Nutrient enrichment can increase the susceptibility of reef corals to bleaching. Nature Climate Change, 3(2), 160–164. Wilkinson, C. (2002). Coral bleaching and mortality: The 1998 event 4 years later and bleaching to 2002. Status of Coral Reefs of the World: 2002, 12–44. Wilkinson, C., & Hodgson, G. (1999). Coral reefs and the 1997‐1998 mass bleaching and mortality. Nature & Resources, 35(2), 16–25. Wilkinson, C., & Science, A. I. of M. (1998). Status of coral reefs of the world: 1998. Global Coral Reef Monitoring Network, 2, 181. Williams, E. H., & Bunkley‐Williams, L. (1990). The world‐ wide coral reef bleaching cycle and related sources of coral mortality. Atoll Research Bulletin, 335, 1‐71. Woodworth, P. L., & Menéndez, M. (2015). Changes in the mesoscale variability and in extreme sea levels over two decades as observed by satellite altimetry. Journal of Geophysical Research: Oceans, 120(1), 64–77. https://doi. org/10.1002/2014JC010363 Wyrtki, K. (1984). The slope of sea level along the equator during the 1982/1983 El Nino. Journal of Geophysical Research, 89(C6), 10419–10424. https://doi.org/10.1029/JC089iC0 6p10419 Yadav, S., Alcoverro, T., & Arthur, R. (2018). Coral reefs respond to repeated ENSO events with increasing resistance but reduced recovery capacities in the Lakshadweep Archipelago. Coral Reefs, 37, 1245–1257. https://doi. org/10.1007/s00338‐018‐1735‐5 Zhang, N., Feng, M., Hendon, H. H., Hobday, A. J., & Zinke, J. (2017). Opposite polarities of ENSO drive distinct patterns of coral bleaching potentials in the southeast Indian Ocean. Scientific Reports, 7(1). https://doi.org/10.1038/ s41598‐017‐02688‐y Zhang, X., & Church, J. A. (2012). Sea level trends, interannual and decadal variability in the Pacific Ocean. Geophysical Research Letters, 39(21). https://doi.org/10.1029/2012G L053240 Zhang, X., Church, J. A., Monselesan, D., & McInnes, K. L. (2017). Sea level projections for the Australian region in the 21st century. Geophysical Research Letters, 44(16), 8481– 8491. https://doi.org/10.1002/2017GL074176

19 ENSO Impact on Marine Fisheries and Ecosystems Patrick Lehodey1, Arnaud Bertrand2, Alistair J. Hobday3, Hidetada Kiyofuji4, Sam McClatchie5, Christophe E. Menkès6, Graham Pilling7, Jeffrey Polovina8, and Desiree Tommasi9

ABSTRACT El Niño events were first perceived several centuries ago as a dramatic change in the marine resources along the Peruvian coast. It is now recognized as part of the world’s largest natural climate fluctuation: the El Niño Southern Oscillation (ENSO). There is a rapidly growing body of scientific literature showing that ENSO has physical and ecological impacts throughout the Pacific Ocean and more broadly across the other oceanic basins through atmospheric teleconnections. This review details a range of these examples in all major ecosystems impacted by ENSO in the Pacific Ocean. Teleconnections with other basins are also discussed, as are the diversity of changes associated with ENSO phases and their consequences on fisheries sustained by these ecosystems. Information is provided on the emerging complexity of the connection between ENSO and the ocean ecosystems, and particularly the diversity of El Niño types, characterized by eastern and central spatial patterns and differences in intensity. As these mechanisms become better understood, useful predictive capacity for ecosystem and fisheries management will result. However, growing evidences suggest that climate change may have already started interacting with ENSO dynamics and effects, complicating mechanistic understanding.

19.1. INTRODUCTION Ocean ecosystems and fish stocks dynamics are strongly affected by multiple scales of climate variability (e.g., Lehodey et al., 2006; Drinkwater et al., 2010). Impacts on marine species can be on movements and migration patterns or on biological and environmental conditions that affect the survival of individuals, either directly (e.g., starving larvae and juveniles) or through cascading and delayed events in the food web. Effects on growth,

reproduction, and behavior are also important and ubiquitous. The El Niño Southern Oscillation (ENSO) originates in the equatorial Pacific, where it is the dominant mode of interannual variability. It leads to physical and ecological impacts throughout the Pacific basin, with important connections in the other oceanic basins (chapters 14 and 15). El Niño was originally recognized by South American fishers in the 1600s as an unusual warm ocean current off the coast of Peru (Garcia‐Herrera et al., 2008; Grove &

  Collecte Localisation Satellite, Ramonville St Agne, France    Institut de Recherche pour le Développement (IRD), MARBEC, Univ Montpellier, CNRS, Ifremer, IRD, Sète, France 3   CSIRO Oceans and Atmosphere, Hobart, TAS, Australia 4  National Research Institute of Far Seas Fisheries, Japan Fisheries Research and Education Agency, Shimizu, Shizuoka, Japan 5 38 Upland Rd, Huia, Auckland, 0604, New Zealand

   Institut de Recherche pour le Développement (IRD), ENTROPIE  (IRD/CNRS/Univ. La Réunion) Nouméa, New Caledonia 7    The Pacific Community (SPC), Noumea, New Caledonia 8   196 Pauahilani Pl., Kailua, HI, USA 9   Institute of Marine Sciences, University of California Santa Cruz, Santa Cruz, CA, USA, and NOAA Southwest Fisheries Science Center, La Jolla, CA, USA

1 2

6

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 429

430  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Adamson, 2018). The name (Christ Child in Spanish), first used in scientific literature in 1893, was linked to the time of year (around December) during which these warm water events tended to occur. The understanding that El Niño was a basin‐scale phenomenon involving an alternate La Niña phase and coupling between the atmosphere and the ocean came later (Bjerknes, 1966 chapter 2). Analyses of the spatial patterns of El Niño events reveals two main types. Eastern Pacific (EP) El Niños were first recognized, while central Pacific (CP, also sometimes called Modoki) El Niños have increased in frequency since the 1980s. Although the development of each ENSO phase involves common ocean‐atmosphere feedback processes, each event varies in intensity and impact (Timmermann et  al., 2018). Consequently, the impacts of ENSO on the biology and ecology of marine species, including many exploited species, are diverse, complex, and often difficult to distinguish from other drivers of variability, e.g. the Pacific Decadal Oscillation (PDO) or Indian Ocean Dipole (Saji et al., 1999; Webster et  al., 1999). This chapter provides a series of regional case studies covering the main large Pacific ecosystems and highlights the diversity of changes associated with ENSO phases and their consequences on the fisheries sustained in each region. Key questions and future needs to progress understanding of this complex multidisciplinary research are then discussed, with particular attention to developing useful predictive capacity for ecosystem and fisheries management. 19.2. THE HUMBOLDT CURRENT SYSTEM It is in the northern Humboldt Current System (HCS), off Peru, that the impacts of canonical (EP) El Niño events on ecosystem and fisheries are most notable (Chavez et  al., 2008). Located along the coast, this upwelling region presently supports more fish catch per unit area than any other region in the oceans; it represents 0.1% of the world ocean but produces up to 10% of the world fish catch (Chavez et al., 2008). For a long time, it was thought that the impact of El Niño events on important fishery resources was straightforward, with a negative impact on anchovy and a positive one on sardine, as demonstrated by the variability in catch landings (e.g. Bakun & Broad, 2003). More recently, this simple vision has been questioned because various factors occurring at different spatiotemporal scales need to be considered to understand the impact of each El Niño (Bertrand et al., 2004). First, it is important to discriminate between the eastern Pacific (EP) and the central Pacific (CP) El Niños. The former can dramatically affect the HCS; however, the latter does not always result in HCS impacts. Surface temperature anomalies can even be negative in

coastal Peru during CP El Niño. For this reason, in this section we will only focus on EP El Niño, as a clear link with the HCS has been demonstrated. El Niño events are typically triggered by anomalies in the wind field in the western equatorial Pacific. These anomalies generate downwelling Kelvin waves that propagate eastward and subsequently poleward along the coasts of North and South America (see chapters 14 and 15), deepening the thermocline and making coastal upwelling “inefficient” in terms of nutrient enrichment. The usually colder and nutrient‐enriched surface waters in the HCS are replaced by warmer and nutrient‐depleted waters. The area of productive coastal water is reduced dramatically. The weaker upwelling allows the less productive oceanic ecosystem to extend toward the coast (Bertrand et al., 2008; Chavez et al., 2002) (Figure 19.1). The HCS ecosystem responses can be very different across different ENSO events. Upwelling and the associated biological system recovered rapidly after the 1997– 1998 El Niño, suggesting weaker ecological impacts than those observed in the seasons after the 1972–1973 and 1982–1983 events (Bertrand et al., 2004; Escribano et al., 2004). For instance, in the case of pelagic fish, several phenomena at different spatiotemporal scales are thought to affect the diversity of responses (Bertrand et al., 2004), including, by decreasing order on the time scale: (i) the decadal regime; (ii) the frequency, intensity and duration of ENSO events; (iii) the conditions of species populations before the event, especially those that are heavily exploited; (iv) the fishing pressure and other mortality sources, e.g. predation; (v) reproductive behavioural responses by the fish, including timing and location of spawning; and (vi) the presence of local efficient upwellings. 19.2.1. Impact on Benthic Species The “tropicalization” of the Humboldt Current during strong El Niño events has different impacts on benthic species, with some winners and some losers. For crustaceans, high temperature during El Niño events can lead to mass mortality of the brachurian crabs Romaleon seto­ sus and Platyxanthus oribigny off Peru and a migration to deeper waters of Cancer porter, with negative impacts on the artisanal crab fishery (Fischer & Thatje, 2016). In contrast, the range of penaeid shrimps (e.g. Xiphopenaeus riveti and Penaeus stylirostris) and the spiny lobster Panuliris gracilis extends southward. Peruvian fishers adjusted their fishing methods to catch greater quantities of shrimp after the El Niño of 1982–1983 (Barber & Chavez, 1986; Arntz et  al., 2006; Thatje et  al., 2008). However, this tropicalization effect was less pronounced during the two following extreme El Niño events in 1997– 1998 and 2015–2016.

Col d

Coa stal Wa te

r

Epipelagic layer

Oceanic water

Thermocline

ental

n Conti

shelf

1

Oxygen minimum zone

Relative scale

Nutrients Phytoplankton Zooplankton

0 Distance from the shore (km)

250

Col d

Epipelagic layer

Oceanic water

0

Coa stal wa

ter

Thermocline

ental

n Conti

shelf

Oxygen minimum zone

1

250

Relative scale

Nutrients Phytoplankton Zooplankton

Distance from the shore (km)

0

0

Figure 19.1  Conceptual model of pelagic ecosystem changes associated with El Niño in the Humboldt Current Ecosystem. In non–El Niño years (top panel), the thermocline is shallow, so the wind‐driven upwelling is highly efficient to supply nutrients. The coastal ecosystem exhibits high biomass and primary productivity that extend far from shore. It is dominated by large phytoplankton, supports a food web with large zooplankton, small pelagic fish, seabirds, marine mammals, and fishers. An oceanic low biomass and primary productivity ecosystem is found offshore of the coastal ecosystem when nutrients are depleted. It is dominated by picophytoplankton, whose grazers are protists with similar growth rates, creating an efficient recycling system. A complex food web evolves with a smaller proportion of the primary production reaching the upper trophic levels composed of ­oceanic species such as tuna or dolphin fish. Note that the presence of mesoscale to submesoscale eddies can lead to the presence of several peaks of nutriments or phytoplankton. During eastern Pacific El Niño years (bottom panel) the productive coastal area is reduced dramatically, and the oceanic ecosystem impinges close to the shore. (Inspired from Chavez et al., 2002, and Guevara‐Carrasco & Bertrand, 2017)

432  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Strong El Niño events are synonymous with high temperatures, floods, and heavy river discharges of sediment‐ laden freshwater in northern Peru. These impacts negatively affect mollusc species, for example, reducing biomass of the scallop Agropecten purpulatus. In contrast, in the south, a drastic (1982–1983), significant (1997–1998), or moderate (1991–1992) increase in scallop stock biomass was roughly proportional to the intensity of warm anomalies in sea surface temperatures observed during these events. These increases had a positive net effect for Peruvian scallop fisheries that compensated for the economic loss in the north. The impact of La Niña events is opposite; however, the increase in scallop landings in northern Peru does not compensate for the decrease in the south (Badjeck et al., 2009). Similarly, El Niño events are associated with reduced surf clam Mesodesma donacium landings in Peru and northern Chile but increased landings in southern Chile (Ortega et al., 2016). Finally, in Peru and northern Chile, increasing sea surface temperature (SST) enhances the recruitment and availability of octopus prey items during El Niño events, which leads to a dramatic increase in Octopus mimus abundance (Arntz et al., 2006). 19.2.2. Impact on Demersal and Pelagic Species The Peruvian hake (Merluccius gayi peruanus) is the most abundant commercially exploited demersal fish in the northern HCS. During an El Niño event, its availability, in particular juvenile abundance, increases near the Peruvian coast. This is due to higher oxygen content and possibly a temperature effect (Guevara‐Carrasco & Lleonart, 2008). During El Niño events, the oxygen content increases in coastal waters (Figure 19.1) due to the deepening of the oxygen minimum zone upper limit under the action of coastal trapped waves forced by intense downwelling equatorial Kelvin waves (Gutierrez et  al., 2008, Espinoza Morriberón et al., 2018). Also, the El Niño conditions allow warmer and more oxygenated waters from the equatorial region to reach the coast of Peru (Montes et al., 2011; Espinoza Morriberón et al., 2018). However, this intrusion into the coastal domain is accompanied by a drastic increase in metabolic cost associated with warmer waters and thus a decrease in female hake fecundity, with an overall negative impact on the Peruvian hake population (Ballón et al., 2008). By contrast, El Niño events have a positive effect on the recruitment of common Merluccius gayi gayi and southern Merluccius australis hake in central‐southern Chile (Payá & Ehrhardt, 2005). Strong EP El Niños affect the distribution of small and medium pelagic fish (anchovy, sardine, mackerel, and jack mackerel). They generally concentrate closer to the coast during El Niños, avoid the warm waters in Northern Peru,

and in some cases, move into deeper waters (Barber & Chávez, 1986; Alheit & Niquen, 2004; Bertrand et  al., 2004). The increased accessibility to fish close to the coast results in elevated catches at the beginning of the events (Alheit & Niquen, 2004). Furthermore, El Niño events were reputed to produce massive die‐offs of anchovy (Engraulis ringens) off the shores of Peru and northern Chile but to favor other pelagic species such as sardine (Sardinops sagax) or jack mackerel (Trachurus murphyi) (Pauly & Tsukayama, 1987; Arntz & Fahrbach, 1996; Bakun & Broad, 2003). However, more recent studies have nuanced this classical view. In most cases, EP El Niño events effectively produced a reduction in anchovy biomass, but subsequent recovery did not always follow the same pattern (Bakun & Broad, 2003; Alheit & Niquen, 2004). Recovery was slow after El Niño events in 1972– 1973, 1977–1978, and 1982–1983, but rapid after the El Niño of 1987 and 1997–1998. Finally, the El Niños of 1991–1992 or 2002–2003 had no perceptible impact on anchovy biomass (Bertrand et al., 2004). Similarly, if sardine catch can increase during El Niños, it does not mean that these events are favorable for the sardine population. There is no clear spatial pattern emerging from the various observed events (Gutiérrez et  al., 2012). Overall, there is apparently rather negative effects due to a reduction in fish condition factor and gonosomatic index (Barber & Chávez, 1986; Bertrand et al., 2004; Cárdenas, 2009). Finally, oceanic predators such as the bonito Sarda chilensis, the dolphin fish Coryphaena hippurus, and the yellowfin tuna Thunnus albacares follow the warm water and move closer to the coast, increasing their availability to fishers (Barber & Chávez, 1986). In central‐south Chile, El Niño impacts are less clear for most species. For instance, catch variability in the anchovy fishery is seemingly not related with ENSO (Cubillos et  al., 2007), although El Niños seem to negatively impact the recruitment of the common sardine Strangomera bentincki in this region (Cubillos & Arcos, 2002). To conclude, while ENSO events can dramatically affect pelagic fish populations of the HCS and their fisheries, they do not seem to play a major role in their long‐ term dynamics, which instead seem to be controlled by decadal ocean conditions and fishing pressure (Alheit & Niquen, 2004; Bertrand et  al., 2004; Gutierrez et  al., 2007; Salvatteci et al., 2018). 19.3. THE EQUATORIAL PACIFIC AND TROPICAL TUNA FISHERIES The equatorial and tropical Pacific Ocean support the world’s largest tuna fisheries, dominated by four tuna species: skipjack (Katsuwonus pelamis), yellowfin (Thunnus albacares), bigeye (T. obesus), and albacore (T.  alalunga). These represent >90% of the total catch

ENSO Impact on Marine Fisheries and Ecosystems  433

taken by industrial fleets. Skipjack and yellowfin spend their entire lives in tropical waters, or tropical waters transported into more temperate latitudes by currents (Kuroshio and East Australian currents). They are fast growing and reach first maturity early (~11 and 20  months, respectively, for skipjack and yellowfin). In contrast, bigeye tuna and albacore have longer life spans and mature later (~3 and 4.5 years respectively) and are thus less productive stocks and respond on longer time scales to environmentally driven variability in juvenile recruitment. They have extended habitats ranging from equatorial waters, where they spawn, to temperate waters, with a well‐observed feeding migration for albacore. Skipjack and yellowfin are caught mainly by surface fishing gear (pole‐and‐line and purse seine), while large yellowfin, bigeye, and albacore are targets of longline fishing gear in subsurface waters. El Niño or La Niña events directly affect horizontal movements and vertical distributions of these tuna species. Spatial distributions of purse seine catch and tagging data in the central western Pacific have revealed a spatial shift in abundance that follows the eastward extension of the warm pool during an El Niño event (Lehodey et  al., 1997). These large east‐west displacements of skipjack in the equatorial region associated with ENSO can be simulated with the spatially explicit ecosystem and fish population dynamics model SEAPODYM (Lehodey et al., 2008; Senina et al., 2008). Model results are consistent with the observed changes in purse seine fishing grounds and tagging data (Figure 19.2). They suggest that the eastward extension (westward contraction) of the species and fisheries distributions during El Niño (La Niña) phases are driven by changes in temperature, prey, and dissolved oxygen concentration. As discussed previously, not all ENSO events are equal. For example, despite being one of the largest events (Figure 19.2), the 2015–2016 El Niño did not strongly impact primary production in the eastern Pacific Ocean, as did other (EP) El Niño events. The strong temperature gradient of the thermocline is a physical barrier for skipjack and juvenile yellowfin and bigeye tuna, while adult yellowfin and bigeye tuna can dive below the thermocline to chase mesopelagic prey. Therefore, changes in the vertical thermal structure of the ocean associated with ENSO can potentially impact the catchability of tuna species by different fishing gears. Purse seiners targeting surface tuna use the top of the thermocline as a lower barrier to trap tuna schools. Typically, the thermocline in the western equatorial Pacific is shallower (deeper) during El Niño (La Niña) than in neutral conditions. The opposite pattern occurs in the eastern Pacific. With modern purse‐seine nets that can reach depths greater than 200 m, these changes in thermocline depths have now limited impact. However, it

was not the case in the 1980s when the purse‐seine fleet started to develop and explore the western Pacific with nets adapted to the eastern region. The fishermen discovered that tuna schools were able to escape below their shallow nets. The contraction and extension of adult bigeye and yellowfin vertical habitat associated with ENSO phases and the depth of the thermocline has been shown from electronic tagging data (e.g., Brill et  al., 1999; Schaefer & Fuller, 2002). The impact on their catchability by longline has been observed or detected from analyses of catch data (Bertrand et  al., 2002; Bigelow et al., 2002) but may be also linked to change in horizontal habitats and migrations associated with ENSO. The environmental variability associated with ENSO can impact the survival of tuna larvae and produce high and low peaks in their abundance. Unfortunately, there are no direct abundance surveys, such as the eggs and larvae sampling commonly used for coastal small pelagic stocks, to monitor such large‐scale variability of tropical tuna species larval densities. However, this variability propagates through the population structure and can be detected with some delay in the exploited stock, either through the analysis of catch rates and size frequencies of catch or as inferred from model and stock assessment analyses. For long‐living species, e.g. bigeye and albacore, the decreasing growth rate with age and its natural variability over time and space leads to a cohort (age) signal more and more difficult to detect in larger fish. Therefore, the recruitment variability associated with ENSO, i.e. low or high peaks of abundance in the first cohort, is smoothed and damped while it combines with the internal dynamic processes of the species propagating toward the older cohorts (Lehodey et  al., 2010; Sibert et  al., 2012; Senina et al., 2017). Thus, the signal is easier to detect in short‐living species like skipjack. The first evidence of an ENSO impact on skipjack larvae recruitment was obtained from simulations with the model SEAPODYM (Lehodey et  al., 2003, 2006; Senina et  al., 2008). El Niño events were shown to be favorable to strong larvae recruitment as illustrated by a (negative) correlation with the Southern Oscillation Index (SOI) (www.cpc.ncep.noaa.gov/data/indices/soi). In this model, the spawning and subsequent recruitment of surviving larvae are predicted from the local biomass of adult fish (spawners) with a density dependence function (Beverton & Holt) and environmental conditions that can be more or less favorable for the concerned species. This is expressed using a spawning index that combines the species water temperature preference and the abundances of prey (i.e. zooplankton) and predators (micronekton) of larvae. The parameters that control these processes are estimated through a quantitative approach using catch data, size frequencies of catch, and

4.0

–2.5 –1.5

2.0

–0.5

0.0 0.5

–2.0

1.5

–4.0 SOI

WCPFC-2016

–6.0

SOI (moving average)

R MFCL detrended (moving average)

6.0

J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98 J-00 J-02 J-04 J-06 J-08 J-10 J-12 J-14 J-16 J-18

(a)

2.5

(b) Mid-Dec 2007 (La Niña) 45 32.5 20 7.5 –5 –17.5 –30 100

135

165

195

225

255

285

225

255

285

mid-Dec 2015 (El Niño) 45 32.5 20 7.5 –5 –17.5 –30 100

135

165

195

Skipjack biomass in metric tonnes per km2 0

0.5

1

Figure 19.2  Impact of ENSO on Pacific skipjack tuna population and fisheries. (a) Comparison of skipjack tuna recruitment index estimated with the model MUTIFAN‐CL (MFCL) for the Western Central Pacific Fisheries Commission (WCPFC) stock assessment study (McKechnie et al., 2016) and the SOI index (reversed axis). Note that the recruitment index has been detrended as the estimate indicates a linear increase. (b) Biomass distribution of skipjack tuna predicted with SEAPODYM (t per km2) and observed catch (black circles) during typical La Niña and El Niño situations.

ENSO Impact on Marine Fisheries and Ecosystems  435

tagging data (Senina et al., 2008). This ENSO‐larvae link is confirmed by an independent estimate of recruitment (Figure  19.2) of the Western Central Pacific Fisheries Commission (McKechnie et  al., 2016). In that case, the recruitment series is estimated from catch and tagging data, without any oceanographic information. Similar positive effects of El Niños on early life stages were detected with SEAPODYM in bigeye and yellowfin tuna species, mainly in the eastern Pacific Ocean. Favorable conditions for larvae survival increase during El Niño events in the eastern Pacific Ocean and decrease in the central region. These species with longer life spans are also more susceptible to present decadal regimes of high and low productivity due to the accumulation of successive low or high peaks of recruitment driven by the decadal modulation of ENSO. A dominance of either El Niño or La Niña events is observed over multiyear periods, possibly in correlation with the Pacific‐scale Interdecadal Pacific Oscillation (IPO). ENSO also impacts the far western Pacific tropical oceanic ecosystem (East of Indonesia, Philippines, and Vietnam), where the variability is complicated by the additional influence of the Indian Ocean Dipole (IOD), another interannual mode of variability developing in the India Ocean (Saji et al., 1999; Webster et al., 1999). IOD and ENSO are partially independent climate modes. Their different phases occur together about 50% of the time (Meyers et al., 2007). A positive IOD is associated with a cold SST anomaly in the southeastern equatorial Indian Ocean. In the Indonesian region, the recent years have seen contrasting climatic conditions. In 2014, ENSO and IOD conditions were neutral. Then in 2015, an El Niño developed in parallel with a positive phase of IOD. The peak intensity in both El Niño and the positive IOD at the end of 2015 coincided with a strong cold anomaly in the southern Indonesian region, associated with a productive coastal upwelling south of Java and Sumatra and in the Banda Sea. In early 2016, a strong negative IOD developed together with a weak La Niña. The temperature anomaly south of Java and Sumatra became positive, and the primary productivity became very weak in the absence of coastal upwelling. The impact of this variability on the bigeye larval recruitment has been explored with the model SEAPODYM (Lehodey et al., 2018). Although there are no observations to validate the results, the simulation suggested that oceanographic conditions during the combined warm phases of ENSO and IOD in September 2015 were highly favorable to tuna larvae survival but reversed to become unfavorable in September 2016 during the cold phase. Future simulations and stock assessment studies may reveal if such low and high peaks of recruitment can be detected in the catch and adult population cohorts.

19.4. THE CENTRAL NORTH PACIFIC The central North Pacific includes most of the North Pacific Subtropical Gyre, a clockwise rotating gyre with warm, low‐nutrient surface waters, bounded on the east by the California Current, the south by the North Equatorial Current, the west by the Kuroshio Current, and the north by the Kuroshio Extension and North Pacific currents (Howell et al., 2012). In spite of its low productivity, it supports complex pelagic and insular marine ecosystems, including tunas, billfishes, cetaceans, sea turtles, seabirds, and coral reef fishes. The ecosystem supports fisheries harvests by commercial longline vessels, local recreational fishers, and artisanal subsistence fishers. The human communities in the region are located in the center of the gyre in the State of Hawaii and in the west in Micronesia, a region composed of five nations: the federated states of Micronesia, Palau, Kiribati, Marshall Islands, and Nauru, and three U.S. territories: the Northern Mariana Islands, Guam, and Wake Island. The physical impacts from El Niño events vary spatially in this region. The southern portion of the gyre, south of about 15°N latitude, experiences the typical equatorial ENSO impacts. The northern portion of the gyre, north of about 25°N latitude, is influenced by midlatitude teleconnections that during an El Niño consist of an intensified and southward‐shifted Aleutian Low pressure system, especially during winter and spring, resulting in a southward expansion of the westerly winds and productive subtropical fronts (Howell et  al., 2012). During a La Niña, the Aleutian Low weakens, and westerlies and the subtropical fronts shift northward (Howell et al., 2012). The Pacific Decadal Oscillation is a climate mode that also manifests as an intensification or weakening of the Aleutian Low with a similar spatial structure to ENSO in the northern gyre but operating on decadal scales (Mantua et al., 1997; Di Lorenzo & Ohman, 2013; Newman et al., 2016). One of the longest time series for an apex species in the Hawaiian Archipelago consists of population and demographic estimates for the endangered Hawaiian monk seal (Monachus schauinslandi). Monk seal pup survival estimates from 1984 to the present show the greatest interannual variations occuring in populations located at the northern islands and atolls (Lisianski Island, Pearl and Hermes Reef, and Midway and Kure atolls) in the northern portion of the Hawaiian Archipelago impacted by the north‐south expansion or contraction of the westerly winds and productive fronts (Baker et al., 2007). In particular, the latitude of the Transition Zone Chlorophyll Front (TZCF; Polovina et al., 2017), a productive front marking the center of a sharp north‐south gradient in surface chlorophyll, shifts south, reaching these northern atolls during the winter season impacted by a strong El

436  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

Niño, but it remains north of these northern atolls during neutral or La Niña conditions (Baker et al., 2007). During neutral or La Niña conditions, when the TZCF was north of the atolls, annual new pup survival 1–2 years later was about 50% compared with 80%–90% 1–2 years after the TZCF reached the atolls during a strong El Niño (Baker et al., 2007). When the TZCF reaches the northern atolls, it may enhance the recruitment of prey for the pups that forage on benthic fishes, while when the TZCF remains north of the atolls, prey recruitment is low (Baker et al., 2007). The 1‐to‐2‐year time lag between the position of the TZCF and monk seal pup survival is thought to be the time required between enhanced productivity and prey recruitment (Baker et al., 2007). Increased entanglement of monk seals in marine debris and subsequent mortality has been observed during El Niños (Donohue & Foley, 2007). This observation can be linked to the increased southward Ekman transport from the more intense and southward‐shifted westerlies during strong El Niños that deliver marine debris from the convergence zone north of Hawaii to the Hawaiian Archipelago (Howell et al., 2012). The Hawaii longline fishery for tuna and swordfish largely operates between 15°N and 30°N, north of the equatorial ENSO impacts and mostly south of the midlatitude ENSO teleconnection impacts, which likely explains why ENSO impacts in the fishery have generally not been observed. However, during the 1997–1998 El Niño, unusually high catches for bigeye tuna were documented at Palmyra Atoll (6°N latitude), a region that is normally not part of the fishing grounds (Howell & Kobayashi, 2006). These catches may have been due to tuna moving eastward from the western Pacific and foraging in favorable habitat that developed around Palmyra during El Niño conditions (Howell & Kobayashi, 2006). In the populated islands and atolls of Micronesia, impacts on local communities from a strong El Niño can be severe (Rupic et  al., 2018). During the developing phase of El Niños, the region experiences strong westerly wind bursts, heavy rainfall, and an increase in tropical cyclones, together with a drop in SST and sea level (Figure  19.3). During the mature and decaying phase, easterlies return, drought conditions develop, and SST and sea level increase (Figure  19.3). The more frequent cyclones with heavy rain and strong storm surge damage infrastructure and increase coastal erosion and saltwater intrusion. During the decaying phase, drought conditions may result in water shortages impacting human communities and agriculture (Rupic et  al., 2018). During the 2014–2016 El Niño, drought conditions resulted in declarations of states of emergencies in Palau, Marshall Islands, Guam, and the Northern Mariana Islands (Rupic et al., 2018). The drop in sea level exposed shallow coral reefs to the air, and intense solar radiation resulted in

coral mortality, while warmer SST resulted in coral bleaching (Rupic et al., 2018). 19.5. THE CALIFORNIA CURRENT ECOSYSTEM The California Current Ecosystem (CCE), between the southernmost point of California and Oregon, is a moderately productive upwelling current system that supports important fisheries. Historically they were dominated by small pelagic species (anchovies and sardines) that exhibit pronounced fluctuations in biomass over decadal periods. In the 1990s, Pacific hake and market squid fisheries developed and became dominant fisheries after the recent collapse of the sardine stock. Superimposed on decadal trends, high‐frequency variability is driven by ENSO events that can have dramatic effects. They have been reviewed recently (McClatchie, 2014) and synthesized in this section. El Niño events in the CCE are associated with anomalous higher temperature and lower salinity, deepening of the mixed layer, increase in coastal dynamic height, broadening and intensifying of northward coastal flow in the Southern California Bight, temporary reversal of the  net southward flow, movement of the core of the California Current further offshore, and reduction in the intensity of coastal upwelling when trade winds weaken (e.g., Chelton et al., 1982; Lynn & Bograd, 2002). Both the offshore extent and the gradient in dynamic height (i.e. the strength of this flow) vary between El Niño events (Hayward, 2000). The anomalies associated with El Niño off southern California and Baja California can be spatially patchy and variable from one event to another. For example, during the 1982–1983 El Niño event, surface temperature anomaly off southern California was in the range of 0.5°C–3°C warmer on January 1983 but more than 4°C warmer closer to the coast (Fiedler, 1984). This large short‐term/spatial heterogeneity in SST was also evident in the coastal sea surface height (Fiedler, 1984), which is a more consistent indicator of El Niño than SST, being less sensitive to high‐frequency variability. Biological communities do not respond in the same way to each El Niño or La Niña. El Niño events impact the fish community assemblages of the Southern California Bight as reported in studies back to the 1940s (Hubbs, 1948; Radovich, 1960). Warm‐water species were found further north than usual during both the strong 1983 and 1992 El Niño events. But the most spectacular change was observed following the extreme El Niño of 1997/98, with a huge shift toward subtropical communities off southern California due to advective processes (Chavez et  al., 2002; Checkley & Barth, 2009; Lea & Rosenblatt, 2000). Shifts in the northern ranges of 29 families of eastern Pacific tropical fishes were reported into Southern Californian waters (Lea & Rosenblatt, 2000).

ENSO Impact on Marine Fisheries and Ecosystems  437

El Niño Developing J

F

M

A

Decaying

Mature

M J

J

A

S

O

N

D

J

F

M

A

M

J

J A

S

O

N

D

O

N

D

Surface Winds Heavy Rainfall Drought

Tropical Cyclones Sea Level Sea Surface Temperature

El Niño Developing J

F

M

A

M

Decaying

Mature J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

Coastal Flooding Rain Flooding Coastal Erosion Wildfire Coral Bleaching Public Health

Figure 19.3  Summary of forcings (top) and impacts (bottom) during a strong El Niño in Micronesia. (Reprinted from Rupic et al., 2018)

These included species not recorded for almost a century in this region, corroborating the physical evidence that the 1997–1998 event was unusually intense. The authors speculated that many arrivals would have been as ichthyoplankton (i.e. fish larvae), or juveniles perhaps associated with flotsam, but that larger fishes may have arrived simply by swimming in suitable water masses. 19.5.1. Anchovy Fiedler et  al. (1986) found no consistent relationship between El Niño and recruitment of anchovy in the CCE in a 17‐year record covering three El Niño events. However, their analysis showed that the extreme 1982–

1983 Eastern Pacific El Niño had a pervasively negative effect on anchovy, affecting the growth, mortality, size‐at‐ age, fecundity, spawning distribution, and the movements of the juveniles and adults. Growth was 47% slower on average for the 1982 and 1983 year‐classes during the El Niño compared to the 1978–1981 year‐classes. Both adults and juveniles were smaller by as much as 10–25  mm (over a mean size range of 10–15 cm) in El Niño years. Fecundity was lower in 1983 and 1984 compared to 1980–1982, and this was attributed to smaller female size‐ at‐age (Fiedler et  al., 1986). Lower fecundity did not translate directly into proportional decrease in fish recruitment due to density‐dependence mechanisms that compensated for lower fecundity in 1983. In addition to

438  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

impacts on growth and fecundity, it was observed that smaller anchovy moved into the nearshore (shallower than 100 m) Southern California Bight from the south during the 1982–1983 El Niño, while larger anchovy moved north and west of Point Conception. 19.5.2. Sardine In contrast to anchovy, sardine appear to recruit more successfully in El Niño years. The increase in abundance of sardines in the northwest of the CCE in 1992 was, for example, attributed to the 1991–1992 El Niño event (Hargreaves et al., 1994). Prior to this event, fishing reports indicated only occasional catches in this region (Emmett et al., 2005). The change in sardine abundance was associated with changes in the entire small pelagic fish community (Emmett et  al., 2005). The 1992 resurgence may not be entirely due to El Niño because none of the previous events back to 1957 produced a similar impact. A possible explanation is that El Niño may not have a detectable effect on the sardine until a population threshold is exceeded. High‐frequency variability in spawning and recruitment is rather typical of small pelagic stocks. This is the case for sardines present along the U.S. West Coast (Lo et al., 2005; Reiss et al., 2008; Weber & McClatchie, 2010), and this has been frequently investigated (Smith, 1990; McFarlane et al., 2002; Bakun & Broad, 2003; Emmett et  al., 2005; Agostini et  al., 2007; Lo et  al., 2010). The spawning habitat is associated with isotherms of 14°C–15°C (Emmett et al., 2005) and sometimes, but not always, shifts northward of San Francisco in El Niño years (Bjorkstedt et  al., 2010). Sampling of sardine egg densities from research cruises between 1997 and 2012 showed high contrast between the 2002 La Niña and the 2003 El Niño (Bjorkstedt et al., 2010). Similarly, spawning habitat area was an order of magnitude larger (Reiss et  al., 2008), and daily egg production was lower (Bjorkstedt et  al., 2010) during the 1999 La Niña compared to the 1998 El Niño. However, recent work shows that a northward shift of sardine spawning also occurs with non‐ENSO warm events (Auth et al., 2018). During the 2006 El Niño and 2007 La Niña, the ocean states showed significant differences, but densities of sardine eggs were not as dramatically different compared to the 2002–2003 El Niño to La Niña transition. Rather than simply driven by change in temperature fields, Weber & McClatchie (2010) suggest that variability in density of sardine eggs and larvae associated to ENSO is likely due to the combination of changes in favorability, predicted from temperature, salinity and chlorophyll‐a concentration, and extension of spawning habitat. These changes occurred under the influence of offshore transport of upwelled nutrient‐rich water, driven by the main wind patterns that vary with ENSO.

The sardine stock in the CCE has continuously declined since the mid‐2000s and has reached its lowest level currently, leading to a closure of the fishery in 2015. It has not reopened since. Recent results based on statistical modeling suggest that concomitant decline of adult sardine, anchovy, and hake are occurring concurrently with tropicalization of the southern CCE due to increased presence of Pacific equatorial‐influenced water in the inshore Southern California region (McClatchie et  al., 2018). This collapse has consequences for the ecosystem as sardines are an important food source for several marine species, including sea lions, salmon, brown pelicans, dolphins, and whales. 19.5.3. Market Squid The market squid, Doryteuthis (formerly Loligo) opal­ escens, fishery off southern and central California is highly variable, and catches have declined to as little as 10% of the catch quota during past El Niño events (Vojkovich, 1998; Marinovic, et  al., 2002; Jackson & Domeir, 2003). During the recent 2015–2016 El Niño, California market squid landings dropped by 65%, leading to a $48 million decline in revenue for California fishers (NMFS, 2017). Hypotheses proposed to explain the decline associated with El Niño events include reduced krill densities (Ish et al., 2004), since krill can make up as much as 65% of the diet of market squid (Karpov & Cailliet, 1979), or changes in the growth rate of squid paralarvae in the month after hatching (Reiss et al., 2004, 2008). The species has a short life span, estimated between 6 and 9 months (Butler et al., 1999; Zeidberg et al., 2006) to a maximum of 18 months (Spratt, 1979; Jackson, 1998), and the population can increase by orders of magnitude in a few generations during periods of rapid growth (Reiss et  al., 2004, 2008). This seems to be an advantage for quick recovery after El Niño–related crashes, especially in Southern California (Ish et  al., 2004). The rate of recovery may differ from one ENSO event to another, and the mechanisms are complex and incompletely understood (McInnis & Broenkow, 1978; Zeidberg et al., 2006; Koslow & Allen, 2011). 19.6. THE NORTHEAST PACIFIC SUBPOLAR GYRE The ocean circulation of the northeast Pacific region is dominated by two oceanic gyres, the North Pacific Subpolar Gyre to the north and the North Pacific Subtropical Gyre to the south. The two circulation systems are separated by the North Pacific Current flowing west to east from Japan to Canada. Off the British Columbia coast, this slow and warm current splits into the equatorward California Current and the poleward Alaska Current (Batchelder & Powell, 2002). Here the

ENSO Impact on Marine Fisheries and Ecosystems  439

focus is on ENSO impacts on fisheries in the Alaska Current and North Pacific Subpolar Gyre, including the Gulf of Alaska. This biologically productive region supports subsistence and tribal fisheries, as well as lucrative commercial and recreational fisheries. The Alaska commercial fishery alone generated more than $4 billion in sales in 2015, with landings revenue being dominated by groundfish, salmon, and crab (NMFS 2017). ENSO, via its effect on large‐scale atmospheric circulation, is a major driver for interannual variability of the North Pacific and also influences the decadal variability in this region (Di Lorenzo et al., 2013; Newman et al., 2016). ENSO impacts can be transmitted from the equatorial region to the northern region via atmospheric teleconnections, particularly during winter (Schwing et  al., 2002; Alexander 2002; see also chapter  14). Anomalous heating at the equator during an El Niño event generates planetary waves that propagate to high latitudes and influence the Aleutian Low (see chapter 14). This triggers a change in atmospheric circulation at higher latitudes: in particular, an intensification, deepening, and eastward movement of the Aleutian Low pressure system, which results in a change in surface winds and ocean transport pathways (Schwing et  al., 2002). A canonical eastern North Pacific response to an El Niño event is characterized by cyclonic (counterclockwise) wind anomalies, an intensification of poleward winds along the North America coast, stronger downwelling, anomalously negative sea‐level pressure anomalies, warmer ocean temperatures, increased ocean stratification, and stronger poleward flow along the west coast of North America (Schwing et  al., 2002). The opposite patterns are true for La Niña. Remote ocean forcing, whereby changes in the water column structure at the equator excite trapped waves that move poleward along the North American coast, can also generate anomalous North Pacific sea surface temperature (see also chapter 14). However, their effect is largely relegated to southern coastal regions of the northeast Pacific (Newman et al., 2016). It is important to point out that because North Pacific ENSO effects are mediated by atmospheric teleconnections, which are also affected by random atmospheric noise, not all ENSO events have the same strong impacts on North Pacific atmospheric circulation and ocean dynamics (Newman et al., 2016). One of the most striking consequences of the strengthening of poleward transport and increased ocean temperatures observed during El Niño events is the dramatic change in distribution and range expansions of many fish and invertebrate species. Fishers may need to move away from their usual fishing grounds as a result of these dramatic changes in fish availability. For instance, during the 1982–1983 El Niño, triggerfish were observed in Alaska, 2800 km north of their previous northern record

(Pearcy & Schoener, 1987). During that same event, market squid abundance increased in Alaska, but they disappeared from their usual fishing grounds in southern California (Pearcy & Schoener, 1987). Changes in distributions are also evident for planktonic organisms, with more tropical plankton species present in the Alaska Current region during El Niño events (Mackas & Gailbraith, 2002). Changes in food web structure, brought about by anomalous advection or changes in ocean mixing and nutrient availability, can affect small pelagic fish and the predators that feed upon them. El Niño events have been associated with reduced availability of forage fish and extensive seabird die‐offs from starvation in the Gulf of Alaska (Bailey et  al., 1995; Morgan, 1999). Variations in ocean transport associated with El Niño also impact fish recruitment. Pacific halibut (Hippoglossus stenolepis) is a winter offshore spawning groundfish. To ensure their survival, its larvae must make their way from offshore spawning areas in the Gulf of Alaska to nursery grounds on the coastal shelf. Doing so requires crossing the fast‐flowing Alaska coastal stream current (Bailey & Picquelle, 2002). Submarine canyons traversing the shelf act as larval transport corridors, but the efficacy of such transport pathways is modulated by large‐scale atmospheric forcing. Pacific halibut larval abundance in nursery areas and year‐class recruitment strength are higher during El Niño events, as the increased coastal current speed acts to entrain more offshore waters, and larvae, onto the shelf (Bailey & Piquelle, 2002). Variability in cross‐shelf transport driven by large‐scale atmospheric forcing also controls Greenland halibut (Reinhardtius hippoglossoides) and Pacific halibut recruitment in the eastern Bering Sea (Duffy‐Anderson et al., 2013; Vestfals et al., 2014). Recent work also demonstrates that different types of El Niño events (EP or CP), by eliciting different teleconnections, have distinct influences on North Pacific decadal scale oceanic and ecosystem variability (Kilduff et al., 2015). Both the PDO and the North Pacific Gyre Oscillation (NPGO), two indices of low frequency oceanic variability in the North Pacific Ocean, are influenced by El Niño (Newman et  al., 2016, Di Lorenzo et al., 2010). EP events, via atmospheric teleconnections, force variability in the Aleutian Low pressure system and hence affect the PDO (Alexander et al., 2002; Newman et  al., 2016). By contrast, CP events impact the North Pacific Oscillation and the NPGO (Di Lorenzo et  al., 2010). A positive PDO pattern is characterized by anomalous warming all along the coast of North America (Mantua et  al., 1997; Figure  19.4). Variability in the PDO index was associated with fluctuations in salmon catch, with Alaska stocks positively, and California, Oregon, and Washington stocks negatively correlated

440  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

with the PDO (Mantua et  al., 1997). The contrasting response of salmon survival rates to increased ocean temperature is a result of a differential response of food web productivity to increased ocean warming in the two systems (Malik et al., 2015, Kilduff et al., 2015). By contrast, CP teleconnections influence low‐frequency variability of the NPGO (Di Lorenzo et  al., 2010). The NPGO, unlike the PDO, is associated with anomalously cold coastal waters in the Gulf of Alaska (Kilduff et al., 2015; Figure 19.4). With the increase of CP events since the 1980s, the NPGO has intensified and now appears to be a better index of salmon survival than the PDO (Kilduff et al., 2015; Figure 19.4). This intensification of the NPGO is also associated with increased coherence of survival rates among different salmon stocks (Kilduff et al., 2015; Figure 19.4).

(a)

19.7. THE NORTHWEST PACIFIC The northwest Pacific (NWP) is one of the most productive fishing areas in the Pacific Ocean. This region supports many commercial fisheries (e.g., Yasuda et al., 2003), including small pelagic species such as Japanese sardine (Sardinops melanostictus), Japanese anchovy (Engraulis japonicus), Pacific saury (Cololabis saira), chub mackerel (Scomber japonicus), and Japanese common squids (Todarades pacificus), as well as highly migratory species, including albacore tuna (Thunnus alalunga), skipjack tuna (Katsuwonous pelamis), Pacific bluefin tuna (Thunnus orientalis), and the neon flying squid (Ommastrephes bartramii). The major oceanographic features in the NWP are (Figure 19.5) the North Equatorial Current (NEC), the

(c)

SPATIAL NPGO & PDO SSTA PATTERNS

PC1 OF SALMON SURVIVAL

60N

COHO PC1 NPGO

40N

1 0

20N

–1

0

R = 0.71

20S

(d)

(b) 60N

CHINOOK PC1 NPGO

40N

1 0

20N

–1

0

R = 0.75 20S

1980 200W –0.6

–0.4

–0.2

150W 0 R

0.4

1990

1995

2000

2005

Year

100W 0.2

1985

0.6

Figure 19.4  Adapted from Kilduff et  al. (2015). Spatial correlation maps of the (a) annual North Pacific Gyre Oscillation (NPGO) and (b) annual PDO with wintertime (January–March) sea surface temperature anomalies. Colored rectangles and ovals highlight differences between the NPGO and PDO spatial signals. Correlation of dominant mode of variability (first principal component) of (c) coho salmon (red line) and (d) chinook salmon (blue line) survival rates with the NPGO (black).

ENSO Impact on Marine Fisheries and Ecosystems  441

warm Kuroshio Current formed off the southern part of Japan (Tsujino et al., 2006), and the Kuroshio Extension (KE), which lies at approximately 35°N (Qiu & Chen, 2005). The cold Oyashio current flowing along the northern coast of Japan joins the Kuroshio in a transition zone between 35°N and 40°N, creating a productive area characterized by high surface chlorophyll‐a concentration and defined as the Transition Zone Chlorophyll Front (TZCF; Polovina et al., 2001, 2017; Figure 19.5). The KE delineates the northern boundary of the subtropical mode water (Holbrook & Maharaj, 2008; Oka, 2009). It has been shown to shift between extended and contracted regimes with less energetic conditions at interannual time scales, likely in relation to the PDO (Qiu & Chen, 2005). Studies on climate variability and fisheries oceanography of the NWP have focused more on the PDO and multidecadal regimes in species biomass and total catches than on ENSO interannual variability (Mantua et  al., 1997; Yatsu et al., 2013; Wang et al., 2012). Decadal fluctuations have been detected in total zooplankton biomass and significantly correlated to wintertime PDO (Chiba et al., 2006). Decadal variations potentially driven by the PDO phases have also been found in distributions of small pelagic fishes such as sardine, anchovy, and saury

(a) 110° 50°

(Noto & Yasuda, 1999; Tian et  al., 2004). More recent research (Ichii et  al., 2018) indicated that saury recruitment variability is related to the winter SST in the Kuroshio or the spring chlorophyll‐a concentration in the Kuroshio‐Oyashio transition area. Therefore, the observed multiyear extended and contracted regimes of the KE can easily produce different decadal phases of high and low recruitment. However, the mechanisms explaining the relationships with climate indices as the PDO index and fluctuation in abundance of marine species or shift in ecosystem regimes are still far from being fully understood. For instance, theoretical modeling (e.g., Di Lorenzo & Ohman, 2013) suggests that the cumulative integrations of white‐noise (high‐frequency) atmospheric forcing can generate red‐noise (low‐­ frequency) responses in oceanographic variables (as for the PDO) and thus generate marine population responses that are characterized by different regimes and strong transitions. The influence of ENSO has been detected among the large, highly migratory species inhabiting the NWP. Based on Japanese longline fisheries data, the North Pacific albacore migration patterns seem more widely dispersed in El Niño years (Kimura et al., 1997). Also, given that all tuna

(b) 120°

130°

140°

150°

160°

Oyashio 40° KBF

AN JAP

170°

110°E

120°E

130°E

140°E

150°E

160°E

50°N

40°N

TZCF

WCR

Kuroshio Extension

30°N

30° Kuroshio

Kuroshio recirculation

20°N

20°

NEC 10°N

10°

ME

MC

0°

–10°

NECC HE

NGCC

0°

10°S

Figure 19.5  (a) Major oceanographic feature in the northwest Pacific Ocean. NEC: North Equatorial Current; NECC: North Equatorial Counter Current; MC: Mindanao Current; ME: Mindanao Eddy; HE: Halmahera Eddy; NGCC: New Guinea Counter Current; KBF: Kuroshio Bifurcation; WCR: Warm Core Ring. (b) Seasonal climatological Chl‐a distribution from NASA Ocean Color Web (https://oceandata.sci.gsfc.nasa.gov/MODIS‐Aqua). TZCF: Transition Zone Chlorophyll Front defined as 0.2 mg/m3 (Polovina et al., 2001).

170°E

442  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

species spawn in tropical warm waters under the influence of ENSO, it can be assumed that their spawning habitats and the subsequent fish recruitment are all impacted by ENSO variability, as has been demonstrated for skipjack tuna (cf section above). The result can be a delayed fluctuation in abundance of juvenile and adult tuna moving to the NWP. This also seems to be the case for the Japanese eels (Anguilla japonica) that spawn in the subtropical NWP. Larvae and juvenile are transported by currents, especially the NEC, which diverges before reaching the coast of the Philippines (Figure 19.5). The NEC bifurcation to the north is at the origin of the Kuroshio. The intensity of the flow and position of the bifurcation change with ENSO and therefore impact larval transport and the recruitment of juveniles in the NWP (Hsiung et al., 2018). The impact of ENSO has been shown to be also favorable for spawning and nursery grounds of the neon flying squid (Ommastrephes bartramii), which supports a major fishery in the North Pacific. The stock was exploited by an international driftnet fishery between 1978 and 1992, with total annual catches reaching more than 350,000 t from the 1980s. The driftnet fishery has been prohibited and the squid is now targeted by jigging vessels from Japan, China, South Korea, and Taiwan at a much lower level (Bower & Ichii, 2005). Neon flying squid is a large oceanic squid distributed in the Pacific between 20°N and 50°N. The distribution of paralarvae suggests that they hatch where the SST ranges from 21°C to 25° C (Bower & Ichii, 2005). Population dynamics are linked with the basin‐wide oceanic circulation and with life history stages that are very responsive to the changes in oceanographic regimes (Ichii et  al., 2009, 2011). The species undergo seasonal migration between spawning (subtropical region) and foraging grounds (transition and subarctic regions), and like most cephalopods, neon flying squids are opportunistic species that prey on zooplankton, other squids, and myctophid fishes (Watanabe et al., 2002). Using habitat models, Alabia et al. (2016) found that the potential squid spawning and nursery habitats were largely influenced by ENSO‐forced environmental changes during the period of reproduction. These changes result in a substantial reduction/enhancement of available habitats in the summers after CP El Niño/La Niña, where the latter leads to an expansion of favorable spawning and nursery grounds. However, the autumn–winter periods of weaker and short‐lived EP El Niño showed elevated potential habitats for these species due to warmer than average sea surface temperature and better feeding conditions. 19.8. THE SOUTHWEST PACIFIC The southwest Pacific Ocean is defined here by a northern boundary at 5°S coinciding with the westward‐ flowing South Equatorial Current, and a southern

boundary that is the subtropical convergence with the Southern Ocean to the south of New Zealand. This region is also influenced by currents in the western limb of the South Pacific subtropical gyre. The East Australian Current is the major western boundary current of the gyre, flowing from the southern Coral Sea and along the coast of northern New South Wales before separating at the Tasman Front and flowing east to New Zealand, or south toward Tasmania as a series of eddies (Suthers et al., 2011). On the east coast of Australia, ENSO has a relatively weak but nevertheless significant influence on the southward‐flowing East Australian Current (Holbrook et al., 2011; Suthers et al., 2011). On the large scale, historical temperature records indicate that an ENSO response can be identified over most of the upper southwest Pacific Ocean, with the strongest signal in the tropics but with significant signals also evident in the subtropical gyre and south Tasman Sea (Holbrook & Bindoff, 1997). 19.8.1. Impacts on Pelagic Species The tuna species found in the southwest Pacific are part of the same stocks as the central and western tropical Pacific regions (section 19.2). The variability of albacore tuna (Thunnus alalunga) longline catch per unit effort (CPUE) in New Caledonia’s exclusive economic zone (EEZ) has been explained by seasonal and interannual influence of ENSO, with highest CPUEs recorded from 1986 to 1998, which corresponds to a period with frequent El Niño events (Briand et al., 2011). The analysis was extended to Samoa and French Polynesia using the SOI index (http://www.bom.gov.au/climate/glossary/soi. shtml) as an explanatory variable. While it confirmed that higher albacore CPUE occurs with El Niños in New Caledonia, lower CPUEs were found in Samoa and French Polynesia, with the reverse situation encountered during La Niña events (Figure 19.6). This east‐west dipole effect is confirmed in other EEZs with a transition region approximately located around Fiji (~180°E). The immediate effect of ENSO on albacore CPUE suggests that catchability may be the main factor to explain this CPUE variability. During El Niño, the vertical habitat of albacore in New Caledonian waters may be compressed to the surface due to shallowing of the vertical thermal structure, illustrated by the change in the 20°C isotherm in the west (Figure 19.7). This increases catchability for the surface fishery. In contrast, regions such as French Polynesia experience a deepening of that habitat during El Niño events, reducing albacore catchability (Jurado‐Molina et al., 2011). The impact of ENSO on the South Pacific albacore abundance is less clear, although a link between recruitment and La Niña phases has been proposed from population dynamics model simulations (Lehodey et al. 2006).

ENSO Impact on Marine Fisheries and Ecosystems  443

Much below average

Much above average

Average

Below average

Above average

NC

S

FP El Niño

Neutral

La Niña

Figure 19.6  Results of the general linear models (GLMs) fitting catch per unit effort (CPUE) in NC (New Caledonia), S (Samoa), and FP (French Polynesia) versus SOI stratified into five classes. SOI values were stratified into five classes representative of Strongest Niño (–15 and 5 and 15). Three GLMs were fitted for EEZs. The CPUE response was also stratified into five classes.

Other large pelagic species also respond to ENSO signals in the southwest Pacific. Off northeast Australia, for example, there is evidence for a greater abundance of black marlin (Makaira indica) during El Niño years (Williams et al., 1994), with Hill et al. (2016) showing that suitable habitat extended up to ~300 km further south during La Nina events. Further south in the East Australian Current, there is little evidence of ENSO phases influencing the distribution or abundance of pelagic species (Holbrook et al., 2009). In Australia’s largest pelagic fishery, the east coast

longline fishery, studies seeking ENSO links to distribution and abundance of the target tuna and billfish have not revealed strong signals. This is due in part to weak temperature anomalies in the region in either ENSO phase. 19.8.2. Impacts on Coastal Benthic and Demersal Species There is currently little evidence for direct ENSO impacts on benthic or demersal species in the southwest Pacific. An

0 –0

5S

2 .1

–0.1 8

8 –0.1 –0.12

–0.24

S

0.12

FP

15S

0.06

NC

20S

0.00

0.00

25S 140E

0.06

.06

6

–0.

–0

–0.0

12

0.00

–0.12

0. 06

Latitude

0.18

10S

160E

180

140W

160W

Longitude

.30

–0.24

–0.18

–0.12

–0.06

0.00

0.06

0.12

0.18

0.24

Figure 19.7 First EOF of 20°C isotherm depth (explaining 20% of the total variance) from the 2004–2017 ARGO product in colors and contours. (ftp://kakapo.ucsd.edu/pub/gilson/argo_climatology/RG_ArgoClim_Temperature_2017. nc.gz) The three exclusive economic zones hashed are for New Caledonia (NC), Samoa (S), and French Polynesia (FP) as labeled on the map.

0.30

444  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

exception is along Australia’s north coast, where enhanced catches of banana prawns (Panaeus merguiensis) in the Gulf of Carpentaria occur during high rainfall/river flow during La Niña years (Vance et al., 1985). In the northwest portion of the southwest Pacific, greater likelihood of unusually warm waters during El Niño years leads to coral bleaching in the Great Barrier Reef, with dramatic bleaching reported in 2015–2016 (Hughes et al., 2017). Bleaching and cyclone damage led to loss of habitat, which subsequently affects the fisheries of the Great Barrier Reef, particularly the coral trout fishery (Hodgkinson et al. 2014). Along the Queensland coast, Meynecke and Yee (2011) showed region‐ and species‐specific environmental relationships for a number of commercially important fish species. The SOI was a significant explanatory variable in many cases; however, the overall effect was relatively minor, and overall explanatory power for the full catch‐environment models was weak to moderate (most model R2 < 0.7). In South Australia and Victoria and midlatitudes of New Zealand, settlement of southern rock lobster (Jasus edwardsii) is higher during El Niño years (Hinojosa et al., 2017). La Niña years were associated with higher settlement in Tasmania and southern New Zealand. In both cases, the relationships were relatively weak. The larval stage for this species spends up to 18 months in the pelagic environment of the Tasman Sea, and exposure to different environmental conditions is likely to lead to these disparate relationships between relatively close regions. 19.9. DISCUSSION Classically linked to the Peruvian anchoveta fishery, we now know that ENSO has an impact on a large number of ecosystems and marine resources of the Pacific Ocean and other basins as well. While we have described the impact of ENSO on many major ecosystems and fisheries, other case studies are discussed in the literature, including for example effects on seabirds, jellyfish, or cetaceans. In addition, through atmospheric and oceanic teleconnections (Yeh et al., 2018; see also chapters 14 and 15), ENSO influences all the other oceanic basins, including the Antarctic, where ENSO has been strongly linked to change in sea ice concentration (Kwok et  al., 2016), which has a central role in the dynamics of the Antarctica oceanic ecosystem. In the Indian Ocean, single or conjugate effects of ENSO and IOD have been shown to impact tropical tuna distributions and recruitment in the eastern Indian Ocean basin (cf section 19.3). Similar mechanisms effect reef ecosystems and tuna fisheries in the western Indian Ocean (Moustahfid et  al., 2018). ENSO has also a strong impact on the intensity of the southward‐flowing Leeuwin Current along Australia’s west coast in the Indian Ocean, being transmitted by

ENSO‐generated planetary waves that propagate from the western Pacific Ocean through the Indonesian Throughflow. Therefore, the ENSO signal in the Leeuwin Current is further transmitted along the south coast of Australia (Holbrook et al., 2009). These physical changes result in impacts on regional fisheries. Along Australia’s west coast, for example, La Niña events have been found to enhance the transport of western rock lobster (Panulirus cygnus) larvae, while El Niño events enhance scallop recruitment. The strength of the Leeuwin Current, linked to the equatorial Pacific circulation through the Indonesian Throughflow, also influences recruitment of pilchard, whitebait, Australian salmon, and herring along Australia’s south coast (see Holbrook et al., 2009). Teleconnections between the Pacific and the tropical Atlantic Ocean are responsible for the variability of upwelling intensity off the West African coast that supports a productive ecosystem and multiple fisheries (Roy & Reason, 2001), as well as the growth of coral and associated changes of coral reef fauna off the Brazilian coast (Kelmo et  al., 2004, 2014; Evangelista et  al., 2007). The research effort to understand mechanisms leading to the development of Atlantic ENSO is growing but still relatively small compared to what has been deployed in the Pacific Ocean. There is no doubt that with the rapid development of knowledge, especially on the timing of the ENSO–tropical Atlantic connection, more biologists will start to investigate in detail the influence of this variability on fisheries and marine species and ecosystem dynamics. It is well demonstrated that despite several classical phases of development, each ENSO event is unique in terms of its intensity and impact, and the sequence of cold, neutral, and warm phases (Timmermann et  al., 2018). The interaction with decadal signal and the effect of long‐term climate change add to ENSO complexity. It is still difficult to attribute recent observed changes in the typology of El Niño events to anthropogenic‐induced climate change. Lee & McPhaden (2010) have reported increasing amplitudes of El Niño events in the Niño‐4 (central equatorial) region, and the last El Niño event (2015–2016) generated an unprecedented warm temperature anomaly in the central equatorial region. Its extreme intensity has been attributed in part to unusually warm conditions in 2014 and to long‐term background warming (Santoso et  al., 2017; Newman & Wittenberg, 2018; Brainard et al., 2018). Given the relatively limited period of modern Earth observation, it is possible that such an extreme event still belongs to the range of natural variability that occurred for ENSO in the last few centuries. Nevertheless, the superimposition of ENSO on a general warming trend will certainly lead to a higher frequency of such extreme events. The biological consequences of the extreme 2015–2016 El Niño event were dramatic on the ecosystems of small,

ENSO Impact on Marine Fisheries and Ecosystems  445

remote Pacific islands in this central region, especially in Jarvis Island (0°22′S, 160°01′W), on the equator south of Hawaii. Unlike in previous strong El Niño events, the 2015–2016 event was not followed by a strong La Niña phase, depriving this region of a strong subsequent recovery of the equatorial upwelling and high productivity associated with it. Consequently, the longest and most widespread coral bleaching event was recorded in Jarvis Island, with massive mortality, i.e. 95% of Jarvis corals were killed (see also chapter 18 on coral reef habitats). Although it was not the first catastrophic bleaching event on Jarvis, it was unprecedented in magnitude (Barkley et  al., 2018). In the meantime, the biomass of planktivore and reef fishes significantly declined, as did the seabird abundance (Brainard et al., 2018). These recent observations pose the decisive question regarding the evolution of ENSO under the influence of climate change, and how it will modify the impacts on ecosystem diversity as described in this review. A better knowledge and modeling of this evolution would make it possible to predict its impacts and thus to prevent and limit the most harmful economical and societal ones. The latest projections of ENSO under the IPCC business‐as‐ usual emission scenarios suggest more frequent extreme EP El Niño events (Cai et  al., 2014, 2018), as well as extreme La Nina events (Cai et al., 2015) associated with the mean‐state changes under greenhouse warming (see chapter 13). Projection uncertainties, however, remain due to model biases (Chen et al., 2017). Therefore, to consider this uncertainty, models of ecosystem or key population species should be coupled to a diversity of climate models, allowing us to explore the diversity of responses to future ENSO patterns forecasted by these projections. The interest for shorter‐term forecasts of ecosystem and marine resources at seasonal and interannual time scales is also rapidly growing (Salinger et al. 2016; Payne et al., 2017; Tommasi et al., 2017) because they can offer a large range of applications for management issues. The last two decades have seen substantial progress in the development of operational ocean models that simulate the state of the ocean in real time, at high resolution, and with assimilation of data from multiple platforms (e.g. satellite, moorings, drifters, and argo floats). It is now possible to envisage ocean forecasts ranging from a few months to several years. These ocean forecasts can be used to drive statistical or ecosystem models to assist in the management of living marine resources. With short‐ term predictions, the model skills are rapidly evaluated in the following months of the forecast, allowing a faster loop of development and progress. A simple statistical relationship between fish recruitment and ENSO as shown for skipjack (Figure 19.2) can be used in conjunction with ENSO forecasts to provide a useful monitoring index of the stock, taking advantage

of the predictability linked to the recruitment index propagating over time in the whole population. Similarly, forecasts of species habitat maps could be provided to assist in marine spatial planning and fisheries monitoring. However, unless they are based only on physical variables, such forecasts would likely require the development of coupled ocean‐biogeochemical forecast systems to provide productivity indices, such as surface chlorophyll concentration or total primary production, which are often used as explanatory variables in species habitat modeling. Several of these systems are now implemented in various ocean forecasting centers (Gehlen et  al., 2015). A study on the predictability of primary production in the tropics (Séférian et al., 2014) suggests a predictive skill of 3 years, which is higher than that of sea surface temperature (1 year). This higher predictability is attributed to the poleward advection of nutrient anomalies (nitrate and iron), which sustain fluctuations in phytoplankton productivity over several years. Therefore, multiyear forecasting of marine ecosystems and fish dynamics models that have the potential to support strategic and investment scale decisions (Salinger et al., 2016) can be envisaged. This represents considerable potential to improve sustainability of marine resource management and industry decisions, improving resilience to ENSO and other climate variabilities. ACKNOWLEDGMENTS The authors are extremely grateful to the anonymous reviewers for constructive suggestions to improve this manuscript. REFERENCES Agostini, V. N., Bakun, A., & Francis, R. C. (2007). Larval stage controls on Pacific sardine recruitment variability: High zooplankton abundance linked to poor reproductive success. Marine Ecology Progress Series, 345, 237–244. doi.org/10. 3354/meps06992 Alabia, I. D., Saitoh, S.‐I., Hirawake, T., Igarashi, H., Ishikawa, Y., Usui, N., et  al. (2016). Elucidating the potential squid habitat responses in the central North Pacific to the recent ENSO flavors. Hydrobiologia, 772, 215–227. Alexander, M. A. (2002). The atmospheric bridge: The influence of ENSO teleconnections on air–sea interactions over the global ocean. Journal of Climate, 15, 2205–2231. Alheit, J., & Niquen, M. (2004). Regime shifts in the Humboldt Current ecosystem. Progress in Oceanography, 60, 201–222. Arntz, W. E., & Fahrbach, E. (1996). El Niño: Experimento climático de la naturaleza. Mexico: Fondo de cultura económica. Arntz, W. E., Gallardo, V. A., Gutiérrez, D., Isla, E., Levin, L. A., Mendo, J., et al. (2006). El Niño and similar perturbation

446  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE effects on the benthos of the Humboldt, California, and Benguela Current upwelling ecosystems. Advances in Geosciences, 6, 243–265. Auth, T. D., Daly, E. A., Brodeur, R. D., & Fisher, J. L. (2018). Phenological and distributional shifts in ichthyoplankton associated with recent warming in the northeast Pacific Ocean. Global Change Biology, 24, 259–272. Badjeck, M.‐C., Mendo, J., Wolff, M., & Lange, H. (2009). Climate variability and the Peruvian scallop fishery: The role of formal institutions in resilience building. Climate Change, 94, 211–232. Bailey, K. M, & Picquelle, S. J. (2002). Larval distribution of offshore spawning flatfish in the Gulf of Alaska: Potential transport pathways and enhanced onshore transport during ENSO events. Marine Ecology Progress Series, 236, 205–217. Bailey, K. M., Macklin, S. A., Reed, R. K., Brodeur, R. D., Ingraham, W. D., Piatt, J. F., et al. (1995). ENSO events in the Northern Gulf of Alaska, and effects on selected marine fisheries. California Cooperative Oceanic Fisheries Investigations Reports, 36, 78–96. Baker, J. D., Polovina, J. J., & Howell, E. A. (2007). Effect of variable oceanic productivity on the survival of an upper trophic predator, the Hawaiian monk seal Monachus schauin­ slandi. Marine Ecology Progress Series, 346, 277–283. Bakun, A., & Broad, K. (2003). Environmental ‘loopholes’ and fish population dynamics: Comparative pattern recognition with focus on El Nino effects in the Pacific. Fisheries Oceanography, 12(4/5), 458–473. Ballón, M., Wosnitza‐Mendo, C., Guevara‐Carrasco, R., & Bertrand, A. (2008). The impact of overfishing and El Niño on the condition factor and reproductive success of Peruvian hake, Merluccius gayi peruanus. Progress in Oceanography, 79, 300–307. https://doi.org/10.1016/j.pocean.2008.10.016 Barber, R. T., & Chávez, F. P. (1986). Ocean variability in relation to living resources during the 1982–83 El Niño. Nature, 319(6051), 279–285. Barkley, H. C., Cohen, A. L., Mollica, N. R., Brainard, R. E., Rivera, H. E., DeCarlo, T., et al. (2018). Repeat bleaching of a central Pacific coral reef over the past six decades (1960– 2016). Communications Biology, 177, 1‐10. https://doi.org/10. 1038/s42003‐018‐0183‐7 Batchelder, H. P., & Powell, T. M. (2002). Physical and biological conditions and processes in the northeast Pacific Ocean. Progress in Oceanography, 53, 105–114. Bertrand, A., Josse, E., Bach, P., Gros, P., & Dagorn, L. (2002). Hydrological and trophic characteristics of tuna habitat: Consequences on tuna distribution and longline catchability. Canadian Journal of Fisheries and Aquaculture Sciences, 59, 1002–1013. Bertrand, A., Segura, M., Gutiérrez, M., & Vasquez, L. (2004). From small‐scale habitat loopholes to decadal cycles: A habitat‐based hypothesis explaining fluctuation in pelagic fish populations off Peru. Fish and Fisheries, 5, 296–316. Bertrand, S., Dewitte, B., Tam, J., Díaz, E., & Bertrand, A. (2008). Impacts of Kelvin wave forcing in the Peru Humboldt Current system: Scenarios of spatial reorganizations from physics to fishers. Progress in Oceanography, 79, 278–289.

Bigelow, K.A., Hampton, J., Miyabe, N., (2002). Application of a habitat‐based model to estimate effective longline fishing effort and relative abundance of Pacific bigeye tuna (Thunnus obesus). Fisheries Oceanography, 11(3), 143–155. Bjerknes, J. (1966). A possible response of the atmospheric Hadley circulation to equatorial anomalies of ocean temperature. Tellus, 18(4), 820–829. doi: 10.3402/tellusa.v18i4.9712 Bjorkstedt, E., Goericke, R., McClatchie, S., Weber, E., Watson, W., Lo, N., Peterson, W., et al. (2010). State of the California Current 2009–2010: Regional variation persists through transition from La Nina to El Nino (and back?). California Cooperative Oceanic Fisheries Investigations Reports, 51, 39–69. Bower J. R., & Ichii, T. (2005). The red flying squid (Ommastrephes bartramii): A review of recent research and the fishery in Japan. Fisheries Research, 76(1), 39–55. doi. org/10.1016/j.fishres.2005.05.009 Brainard R. E., Oliver T., McPhaden M. J., Cohen A., Venegas R., Heenan A., Vargas‐Angel et al. (2018). Ecological impacts of the 2015‐16 El Niño in the central equatorial Pacific. In “Explaining Extreme Events of 2016 from a Climate Perspective.” Bulletin of American Meteorological Society, 99 (1), S21 –S26. doi:10.1175/BAMS‐D‐17‐0118.1 Briand, K., Molony, B. & Lehodey, P. (2011). A study on the variability of albacore (Thunnus alalunga) longline catch rates in the south‐west Pacific Ocean. Fisheries Oceanography, 20, 517–529. Brill, R. W., Block, B. A., Boggs, C. H., Bigelow, K. A., Freund,  E. V. & Marcinek, D. J. (1999). Horizontal movements and depth distribution of large adult yellowfin tuna (Thunnus albacares) near the Hawaiian Islands, recorded using ultrasonic telemetry: Implications for the physiological ecology of pelagic fishes. Marine Biology, 133, 395–408. Butler, J., Fuller, D., & Yaremko, M. (1999). Age and growth of market squid (Loligo opalescens) off California during 1998. California Cooperative Oceanic Fisheries Investigations Reports, 40, 191–195. Cai, W., S. Borlace, M. Lengaigne, P. van Rensch, M. Collins, G. Vecchi, et al. (2014). Increasing frequency of extreme El Niño events due to greenhouse warming. Nature Climate Change, 4, 111–116. http://dx.doi.org/10.1038/nclimate2100 Cai, W., Wang, G., Santoso, A., McPhaden, M. J., Wu, L., Jin, F.‐F., et al. (2015). Increased frequency of extreme La Niña events under greenhouse warming. Nature Climate Change, 5, 132–137. http://dx.doi.org/10.1038/nclimate2492 Cai, W., et al. (2018). More enhanced and more frequent EP El Niños in the future. Nature, 564, 201–206. Cárdenas, G. (2009). Análisis de series de tiempo de los indicado­ res biológicos, pesqueros y poblacionales de la sardina, Sardinops sagax sagax (jenyns, 1842), en función de la variabi­ lidad ambiental y la pesca (PhD Thesis). University Nacional Mayor de San Marcos, Lima, Peru. Chavez, F. P., Pennington, J. T., Castro, C. G., Ryan, J. P., Michisaki, R. P., Schlining, B., et al. (2002). Biological and chemical consequences of the 1997–1998 El Niño in central California waters. Progress in Oceanography, 54, 205–232. Chavez, F. P., Bertrand, A., Guevara, R., Soler, P., & Csirke, J. (2008). The northern Humboldt Current System: Brief

ENSO Impact on Marine Fisheries and Ecosystems  447 ­ istory, present status and a view towards the future. Progress h in Oceanography, 79(2–4), 95–105. Checkley, D. M., & Barth, J. A. (2009). Patterns and processes in the California Current System. Progress in Oceanography, 83, 49–64. doi.org/10.1016/j.pocean.2009.07.028 Chelton, D., Bernal, P., & McGowan, J. (1982). Large‐scale interannual physical and biological interaction in the California Current. Journal of Marine Research, 40, 1095–1125. Chen, C., Cane, M. A., Wittenberg, A. T., & Chen, D. (2017). ENSO in the CMIP5 simulations: Life cycles, diversity, and responses to climate change. Journal of Climate, 30, 775–801. doi/10.1175/JCLI‐D‐15‐0901.1 Chiba, S., Tadokoro, K., Sugisaki, H. & Saino, H. (2006). Effects of decadal climate change on zooplankton over the past 50 years in the western North Pacific. Global Change Biology, 12, 907–920. Cubillos, L. A., & Arcos, D. F. (2002). Recruitment of common sardine (Strangomera bentincki) and anchovy (Engraulis rin­ gens) off central‐south Chile in the 1990s and the impact of the 1997–1998 El Niño. Aquatic Living Resources, 15, 87–94. Cubillos, L., Serra, R., & Fréon, P. (2007). Synchronous pattern of fluctuation in three anchovy fisheries in the Humboldt Current System. Aquatic Living Resources, 20(1), 69–75. doi:10.1051/alr:2007017 Di Lorenzo, E., & Ohman, M. D. (2013). A double‐integration hypothesis to explain ocean ecosystem response to climate forcing. Proceedings of the National Academy of Sciences, 110 (7), 2496–2499. Di Lorenzo, E., Cobb, K. M., Furtado, J. C., Schneider, C., Anderson, D. T., Bracco, A., et al. (2010). Central Pacific El Niño and decadal climate change in the North Pacific Ocean. Nature Geoscience, 3, 762–765. Di Lorenzo, E., Combes, V., Keister, J., & Strub, P. T. (2013). Synthesis of Pacific Ocean climate and ecosystem dynamics. Oceanography, 26, 32–45. Donohue, M. & Foley, D. G. (2007). Remote sensing reveals links among the endangered Hawaiian monk seal, marine debris, and El Nino. Marine Mammal Science, 23(2), 468–473. Drinkwater, K. F., Beaugrand, G., Kaeriyama, M., Kim, S., Ottersen, G., Perry, R. , I. et  al. (2010). On the processes linking climate to ecosystem changes. Journal of Marine Systems, 79, 374–388. doi: 10.1016/j.jmarsys.2008.12.014 Duffy‐Anderson, J. T., Blood, D. M., Cheng, W., Ciannelli, L., Matarese, A. C., Sohn, D., et  al. (2013). Combining field observations and modeling approaches to examine Greenland halibut (Reinhardtius hippoglossoides) early life ecology in the southeastern Bering Sea. Journal of Sea Research, 75, 96–109. Emmett, R., Brodeur, R., Miller, T., Pool, S., Krutzikowsky, G., Bentley, P., & McCrae, J. (2005). Pacific sardine (Sardinops sagax) abundance, distribution, and ecological relationships in the Pacific Northwest. California Cooperative Oceanic Fisheries Investigations Reports, 46, 122–143. Escribano, R., Daneri, G., Farías, L., Gallardo, V. A., González, H. E., Gutiérrez, D., et  al. (2004). Biological and chemical consequences of the 1997‐1998 El Niño in the Chilean coastal

upwelling system: A synthesis. Deep Sea Research II, 51, 2389–2411. Espinoza Morriberón, D., Echevin, V., Colas, F., Tam Málaga, J. L., Gutiérrez Aguilar, D. A., Graco, M. I. et  al. (2018). Oxygen variability during ENSO in the tropical south eastern Pacific. Frontiers in Marine Science, 5, 526. Evangelista, H., Godiva, D., Sifeddine, A., Leao, Z.M.A.N., Rigozo, N. R., Segal, B. et  al. (2007). Evidences linking ENSO and coral growth in the southwestern‐South Atlantic. Climate Dynamics, 29, 869–880. Fiedler, P. (1984). Satellite observations of the 1982–83 El Niño along the U.S. Pacific coast. Science, 224, 1251–1254. Fiedler, P. C., Methot, R. D., & Hewitt, R. P. (1986). Effects of California El Niño 1982–1984 on the northern anchovy. Journal of Marine Research, 44, 317–338. doi.org/10. 1357/002224086788405365 Fischer, S., & Thatje, S. (2016). Temperature effects on life‐history traits cause challenges to the management of brachyuran crab fisheries in the Humboldt Current: A review. Fisheries Research, 183, 461–468. Garcia‐Herrera, R., Diaz, H. F., Garcia, R. R., Prieto, M. R., Barriopedro, R., Moyano, R., et al. (2008). A chronology of El Nino events from primary documentary sources in Northern Peru. Journal of Climate, 21, 1948–1962. Gehlen, M., Barciela, R., Bertino, L., Brasseur, P., Butenschön, M., Chai, F., et  al. (2015). Building the capacity for forecasting marine biogeochemistry and ecosystems: Recent advances and future developments. Journal of Operational Oceanography, 8, s168–s187. doi: 10.1080/1755876X.2015 .1022350 Grove, R., & Adamson, G. (2018). El Niño in World History. London: Palgrave Macmillan UK. https://doi.org/10. 1057/978‐1‐137‐45740‐0 Guevara‐Carrasco, R., & Bertrand, A. (2017). Atlas de la pesca artesanal del mar del Perú. Lima, Peru: IMARPE‐IRD. Guevara‐Carrasco, R., & Lleonart, J. (2008). Dynamics and fishery of the Peruvian hake: Between nature and man. Journal of Marine Systems, 71, 249–259. Gutierrez, M., Swartzman, G., Bertrand, A., & Bertrand, S. (2007). Anchovy (Engraulis ringens) and sardine (Sardinops sagax) spatial dynamics and aggregation patterns in the Humboldt Current ecosystem, Peru, from 1983–2003. Fisheries Oceanography, 16(2), 155–168. Gutiérrez, D., Enríquez, E., Purca, S., Quipúzcoa, L., Marquina, R., Flores, G., et  al. (2008). Oxygenation episodes on the continental shelf of central Peru: Remote forcing and benthic ecosystem response. Prog. Oceanogr., 79, 177–189. doi: 10.1016/j.pocean.2008.10.025 Gutiérrez, M., Castillo, R., Segura, M., Peraltilla, S., & Flores, M. (2012). Trends in spatio‐temporal distribution of Peruvian anchovy and other small pelagic fish biomass from 1966–2009. Latin American Journal of Aquatic Research, 40, 633–648. Hargreaves, N. B., Ware, D., & McFarlane, G. (1994). Return of Pacific sardine (Sardinops sagax) to the British Columbia coast in 1992. Canadian Journal of Fisheries and Aquatic Sciences, 51, 460–463. Hayward, T. (2000). El Niño 1997–98 in the coastal waters of southern California: A timeline of events. California

448  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Cooperative Oceanic Fisheries Investigations Reports, 41, 98–116. Hill, N. J., Tobin, A. J., Reside, A. E., Pepperell, J. G., & Bridge, T.C.L. (2016). Dynamic habitat suitability modelling reveals rapid poleward distribution shift in a mobile apex predator. Global Change Biology, 22, 1086–1109. Hinojosa, I. A., Gardner, C., Green, B. S., Jeffs, A., Leon, R., & Linnane, A. (2017). Differing environmental drivers of settlement across the range of southern rock lobster (Jasus edwardsii) suggest resilience of the fishery to climate change. Fisheries Oceanography, 26(1), 49–64. Hodgkinson, J. A., Hobday, A. J., & Pinkard, E. A. (2014). Climate adaptation in Australia’s resource‐extraction industries: Ready or not? Regional Environmental Change, 14(4), 1663–1678. Holbrook, N. L., & Bindoff, N. L. (1997). Interannual and decadal temperature variability in the southwest Pacific Ocean between 1955 and 1988. Journal of Climate, 10, 1035–1049. Holbrook, N. J., & Maharaj, A. M. (2008). Southwest Pacific subtropical mode water: A climatology. Progress in Oceanography, 77, 298–315. Holbrook, N. J., Davidson, J., Feng, M., Hobday, A. J., Lough, J. M., McGregor, S., & Risbey, J. S. (2009). El Niño‐Southern Oscillation. In E. S. Poloczanska, A. J. Hobday, & A. J. Richardson (Eds.), A Marine Climate Change Impacts and Adaptation Report Card for Australia 2009. NCCARF Publication 05/09, ISBN 978‐1‐921609‐03‐9. Holbrook, N. J., Goodwin, I. D., McGregor, S., Molina, E., & Power S. ,B. (2011). ENSO to multi‐decadal time scale changes in East Australian Current transports and Fort Denison sea level: Oceanic Rossby waves as the connecting mechanism. Deep Sea Research Part II, 58(5), 547–558. doi. org/10.1016/j.dsr2.2010.06.007 Howell, E. A., & Kobayashi, D. R. (2006). El Nino effects in the Palmyra Atoll region: Oceanographic changes and bigeye tuna (Thunnus obesus) catch rate variability. Fisheries Oceanography, 15(6), 477–489. Howell, E. A., Bograd, S. J., Morishige, C., Seki, M. P., & Polovina, J. J. (2012). On North Pacific circulation and associated marine debris concentration. Marine Pollution Bulletin, 65(1–3), 16–22. Hsiung, K.‐M., Kimura, S., Han, Y.‐S., Takeshige, A. & Iizuka, Y. (2018). Effect of ENSO events on larval and juvenile duration and transport of Japanese eel (Anguilla japonica). PLoS ONE, 13(4), e0195544. doi.org/10.1371/journal.pone.0195544 Hubbs, C. (1948). Changes in the fish fauna of western North America correlated with changes in ocean temperature. Journal of Marine Research, 7, 459–482. Hughes, T. P., Kerry, J. T., Álvarez‐Noriega, M., Álvarez‐ Romero, J. G., Anderson, K. D., Baird, A. H., et al. (2017). Global warming and recurrent mass bleaching of corals. Nature, 543, 373–377. Ichii, T., Mahapatra, K., Sakai, M. & Okada, Y. (2009). Life history of the neon flying squid: Effect of the oceanographic regime in the North Pacific Ocean. Marine Ecological Progress Series, 378, 1–11.

Ichii, T., Mahapatra, K., Sakai, M., Wakabayashi, T., Okamura, H., Igarashi, H., & Okada, Y. (2011). Changes in abundance of the neon flying squid Ommastrephes bartramii in relation to climate change in the central North Pacific Ocean. Marine Ecological Progress Series, 441, 151–164. Ichii, T., Nishikawa, H., Mahapatra, K., Okamira, Igarashi, H., Sakai, M., et  al. (2018). Oceanographic factors affecting interannual recruitment variability of Pacific saury (Cololabis saira) in the central and western North Pacific. Fisheries Oceanography, 27, 445–457. doi.org/10.1111/fog.12265 Ish, T., Dick, E., Switzer, P., & Mangel, M. (2004). Environment, krill and squid in Monterey Bay: From fisheries to life histories and back again. Deep‐Sea Res., 51, 849–862. Jackson, G. (1998). Research into the life history of Loligo opal­ escens: Where to from here? California Cooperative Oceanic Fisheries Investigations Reports, 39, 101–107. Jackson, G., & Domeir, M. (2003). The effects of an extraordinary El Niño/La Niña event on the size and growth of the squid Loligo opalescens off Southern California. Marine Biology, 142(5), 925–935. Jurado‐Molina, J., Bigelow, K., Hoyle, S., Nicol, S., & Briand, K. (2011). Developing a spatially adjusted CPUE for the albacore fishery in the South Pacific. WCPFC‐SC7‐2011/EB‐WP‐ IP‐05. https://www.wcpfc.int/node/2852 Karpov, K., & Cailliet, G. (1979). Prey composition of the market squid, Loligo opalescens Berry, in relation to depth and location of capture, size of squid, and sex of spawning squid. California Cooperative Oceanic Fisheries Investigations Reports, 20, 51–57. Kelmo, F., Attrill, M. J., Gomes R.C.T., & Jones, M. B. (2004). El Niño induced local extinction of coral reef bryozoan species from Northern Bahia, Brazil. Biological Conservation, 118(5), 609–617. doi.org/10.1016/j.biocon.2003.10.010 Kelmo, F., Bell, J. J., Moraes, S. S., Gomes, R.C.T., Mariano‐ Neto, E., & Atrill, M. J. (2014). Differential responses of emergent intertidal coral reef fauna to a large‐scale El Nino Southern Oscillation Event: Sponge and coral resilience. PLosONE, 9(3), e93209. doi:10.1371/journal.pone.0093209 Kilduff, D. P., Di Lorenzo, E., Botsford, L. W., & Teo, S.L.H. (2015). Changing central Pacific El Ninos reduce stability of North American salmon survival rates . Proceedings of the National Academy of Sciences, 112, 10962–10966. Kimura, S., Nakai, M., & Sugimoto, T. (1997). Migration of albacore, Thunnus alalonga, in the North Pacific Ocean in relation to large oceanic phenomena. Fisheries Oceanography, 6(2), 51–57. Koslow, J., & Allen, C. (2011). The influence of the ocean environment on the abundance of market squid, Doryteuthis (= Loligo) opalescens, paralarvae in the Southern California Bight. California Cooperative Oceanic Fisheries Investigations Reports, 52, 205–213. Kwok, R., Comiso, J. C., Lee, T., & Holland, P. R. (2016). Linked trends in the South Pacific sea ice edge and Southern Oscillation Index. Geophysical Research Letters, 43, 10,295– 10,302. https://doi.org/10.1002/2016GL070655 Lea, R., & Rosenblatt, R. (2000). Observations on fishes associated with the 1997–98 El Niño off California. California

ENSO Impact on Marine Fisheries and Ecosystems  449 Cooperative Oceanic Fisheries Investigations Reports, 41, 117–129. Lee, T., & McPhaden, M. J. (2010). Increasing intensity of El Niño in the central equatorial Pacific. Geophysical Research Letter, 37, L14603. doi:10.1029/ 2010GL044007 Lehodey, P., Bertignac, M., Hampton, J., Lewis, T., & Picaut J. (1997). El Niño Southern Oscillation and tuna in the western Pacific. Nature, 389, 715–718. Lehodey, P., Chai, F., & Hampton, J. (2003). Modelling climate‐ related variability of tuna populations from a coupled ocean‐ biogeochemical‐populations dynamics model. Fisheries Oceanography, 12(4), 483–494. Lehodey, P., Alheit, J., Barange, M., Baumgartner, T., Beaugrand, G., Drinkwater, K., et  al. (2006). Climate variability, fish and fisheries. Journal of Climate, 19(20), 5009–5030. Lehodey, P., Senina, I., & Murtugudde, R. (2008). A Spatial Ecosystem And Populations Dynamics Model (SEAPODYM): Modelling of tuna and tuna‐like populations. Progress in Oceanography, 78, 304–318. Lehodey, P., Senina, I., Sibert, J., Bopp, L, Calmettes, B., Hampton, J., & Murtugudde, R. (2010). Preliminary forecasts of population trends for Pacific bigeye tuna under the A2 IPCC scenario. Progress in Oceanography, 86, 302–315. Lehodey, P., Senina, I., Wibawa, T. A., Titaud, O., Calmettes, B., Tranchant, B., & Gaspar, P. (2018). Operational modelling of bigeye tuna (Thunnus obesus) spatial dynamics in the Indonesian region. Marine Pollution Bulletin, 131, 19–32. Lo, C., Macewicz, B., & Griffith, D. (2010). Biomass and reproduction of Pacific sardine (Sardinops sagax) off the Pacific northwestern United States, 2003–2005. Fishery Bulletin U. S., 108(2), 174–192. Lo, N., Griffith, D., & Macewicz, B. (2005). Spawning biomass of Pacific sardine (Sardinops sagax) from 1994–2004 off California. California Cooperative Oceanic Fisheries Inves­ tigations Reports, 46, 93–112. Lynn, R., & Bograd, S. (2002). Dynamic evolution of the 1997– 1999 El Niño‐La Niña cycle in the southern California Current System. Progress in Oceanography, 54, 59–75. Mackas, D. L., & Galbraith, M. (2002). Zooplankton community composition along the inner portion of Line P during the 1997– 98 El Niño event. Progress in Oceanography, 54, 423–437. Malick, M. J., Cox, S. P., Mueter, F. J., & Peterman, R. M. (2015). Linking phytoplankton phenology to salmon productivity along a north‐south gradient in the Northeast Pacific Ocean. Canadian Journal of Fisheries & Aquatic Sciences, 72(5), 697–708. Mantua, N. J., Hare, S. R., Zhang, Y., Wallace, J. M., & Francis, R. C. (1997). A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society, 78, 1069–1080. Marinovic, B., Croll, D. A., Gong, N., Benson, S. R., & Chavez F. P. (2002). Effects of the 1997–1999 El Niño and La Niña events on zooplankton abundance and euphausiid community composition within the Monterey Bay coastal upwelling system. Deep‐Sea Research, 54, 265–277.

McClatchie, S. (2014). Regional fisheries oceanography of the California Current System: the CalCOFI program. Springer. ISBN 978‐94‐007‐7222‐9. McClatchie, S., Gao J., Drenkard, E. J., Thompson, A. R., Watson, W., Ciannelli, L., et  al. (2018). Interannual and secular variability of larvae of mesopelagic and forage fishes in the Southern California Current System. Journal of Geophysical Research, 123(9), 6277–6295. McFarlane, G., & Beamish, R. J. (1999). Sardines return to British Columbia waters. In Freeland, H. J., Peterson, W. T., & Tyler, A. (Eds.), Proceedings of the 1998 Science Board Symposium on the impacts of the 1997/98 El Niño event on the North Pacific Ocean and its marginal seas. PICES Scientific Report, 10. McFarlane, G., Smith, P., Baumgartner, T., & Hunter, J. (2002). Climate variability and Pacific sardine populations and fisheries. In McGinn, N. (Ed.), Fisheries in a Changing Climate, Symposium 32 (pp. 195–214). New York: American Fisheries Society. McInnis, R. R., & Broenkow, W. M. (1978). Correlations between squid catches and oceanography conditions in Monterey Bay, California. California Department Fish Game and Fishery Bulletin, 169, 161–170. McKechnie, S., Hampton, J., Pilling, G. M., & Davies N. (2016). Stock assessment of skipjack tuna in the western and central Pacific Ocean. 12th regular session of the Scientific Committee of the Western Central Pacific Fisheries Commission, 3–11 Aug 2016, Bali, Indonesia. WCPFC‐ SC12‐2016/SA‐WP‐04, 120 pp. Meyers, G., McIntosh, P., Pigot, L., & Pook, M. (2007). The years of El Niño, La Niña, and interactions with the tropical Indian Ocean. Journal of Climate, 20, 2847–2880. Meynecke, J.‐O., & Lee, S. Y. (2011). Climate‐coastal fisheries relationships and their spatial variation in Queensland, Australia. Fisheries Research, 110, 365–376. Montes, I., Wolfgang, S., Colas, F., Blanke, B., & Echevin, V. (2011). Subsurface connections in the eastern tropical Pacific during La Niña 1999–2001 and El Niño 2002–2003. J. Geophys. Res., 11, :C12022. doi: 10.1029/2011JC007624 Morgan, K. H. (1999). Impact of the 1997/98 El Niño on seabirds of the North East Pacific. In Freeland, H. J., Peterson, W. T., & Tyler, A. (Eds.), Proceedings of the 1998 Science Board Symposium on the impacts of the 1997/98 El Nino event on the North Pacific Ocean and its marginal seas. PICES Scientific Report, 10. Moustahfid, H., Marsac, F., & Gangopadhyay, A. (2018) Climate change impacts, vulnerabilities and adaptations: Western Indian Ocean marine fisheries. In Barange, M., Bahri, T., Beveridge, M.C.M., Cochrane, K. L., Funge‐Smith, S., & Poulain, F. (Eds.), Impacts of climate change on fisheries and aquaculture: Synthesis of current knowledge, adaptation and mitigation options (chap. 12). FAO Fisheries and Aquaculture Technical Paper No. 627. Rome: FAO. Newman, M., & Wittenberg, A. (2018). The extreme 2015/16 El Niño in the context of historical climate variability and change. In “Explaining Extreme Events of 2016 from a Climate Perspective”. Bulletin of American Meteorological Society, 99(1), S16–S20. doi:10.1175 /BAMS‐D‐17‐0116.1

450  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Newman, M., Alexander, M. A., Ault, T. R., Cobb, K. M., Deser, C., Di Lorenzo, E., et al. (2016). The Pacific Decadal Oscillation, Revisited. Journal of Climate, 29, 4399–4427. NMFS, National Marine Fisheries Service (2017). Fisheries Economics of the United States, 2015. NOAA Technical Memorandum (NMFS‐F/SPO‐170, 247p.). U.S. Dept. of Commerce. Noto, M., & Yasuda, I. (1999). Population decline of the Japanese sardine, Sardinops melanostictus, in relation to sea surface temperature in the Kuroshio Extension. Canadian Journal of Fisheries and Aquatic Sciences, 56, 973–983. Oka, E. (2009). Seasonal and interannual variation of North Pacific Subtropical mode water in 2003‐2006. Journal of Oceanography, 65, 151–164. Ortega, L., Celentano, E., Delgado, E., & Defeo, O. (2016). Climate change influences on abundance, individual size and body abnormalities in a sandy beach clam. Marine Ecology Progress Series, 545, 203–216. Pauly, D., & Tsukayama, I. (1987). On the implementation of management‐oriented fishery research: The case of the Peruvian anchoveta. In Pauly, D., & Tsukayama, I. (Eds.), The Peruvian anchoveta and its upwelling ecosystem: Three decades of change (pp. 1–13). Callao, Peru: Instituto del Mar del Peru. Payá, V., & Ehrhardt, N. M. (2005). Comparative sustainability mechanisms of two hake (Merluccius gayi gayi and Merluccius australis) populations subjected to exploitation in chile. Bulletin of Marine Science, 76, 261–286. Payne, M. R., Hobday, A. J., MacKenzie, B. R., Tommasi, D., Dempsey, D. P., Fässler, S.M.M., et al. (2017). Lessons from the first generation of marine ecological forecast products. Frontiers in Marine Sciences, 4, 289. doi: 10.3389/fmars. 2017.00289 Pearcy, W. G., & Schoener, A. (1987). Changes in the marine biota coincident with the 1982‐1983 El Niño in the Northeastern Subarctic Pacific Ocean. Journal of Geophysical Research, 92, 417–428. Polovina, J., Howell, E., Kobayashi, D. R., & Seki, M. P. (2001). The transition zone chlorophyll front, a dynamic global feature defining migration and forage habitat for marine resource. Progress in Oceanography, 49, 469–483. Polovina, J. J., Howell, E. A., Kobayashi, D. R., & Seki, M. P. (2017). The Transition Zone Chlorophyll Front updated: Advances from a decade of research. Progress in Oceanography, 150, 79–85. Qiu, B. & Chen, S. (2005). Variability of the Kuroshio Extension Jet, Recirculation Gyre, and Mesoscale Eddies on decadal time scales. Journal of Physical Oceanography, 35, 2090–2103. Radovich, J. (1960). Redistribution of fishes in the eastern North Pacific Ocean in 1957 and 1958. California Cooperative Oceanic Fisheries Investigations Reports, 7, 163–171. Reiss, C. S., Maxwell, M. R., & Hunter, J. R. (2004). Investigating environmental effects on population dynamics of Loligo opalescens in the Southern California Bight. CalCOFI Reports, 45, 87–97. http://calcofi.org/publications/calcofi reports/v45/Vol_45_Reiss.pdf Reiss, C., Checkley Jr, D., & Bograd, S. (2008). Remotely sensed spawning habitat of Pacific sardine (Sardinops sagax) and

Northern anchovy (Engraulis mordax) within the California Current. Fisheries Oceanography, 17(2), 126–136. Roy, C., & Reason, C. (2001). ENSO related modulation of coastal upwelling in the eastern Atlantic. Progress in Oceanography, 49, 245–255. Rupic, M., Wetzell, L., Marra, J. J., & Balwani, S. (2018). 2014‐2016 El Niño assessment report: An overview of the impacts of the 2014‐16 El Niño on the U.S.‐affiliated Pacific Islands (USAPI). Report to the NOAA Pacific Region ENSO Tiger Team. Saji, N. H., Goswami, B. N., Vinayachandran, P. N., & Yamagata, T. (1999). A dipole mode in the tropical Indian Ocean. Nature, 401, 360–363. Salinger, J., Hobday, A. J., Matear, R. J., O’Kane, T. J., Risbey, J. S., Dunstan, P. K., et al. (2016). Decadal‐scale forecasting of climate drivers for marine applications. Advances in Marine Biology, 74, 1–68. doi:10.1016/bs.amb.2016.1004.1002 Salvatteci, R., Field, D., Gutiérrez, D., Baumgartner, T., Ferreira, V., Ortlieb, L., et al. (2018). Multifarious anchovy and sardine regimes in the Humboldt Current System during the last 150 years. Global Change Biology, 24(3), 1055–1068. https://doi.org/10.1111/gcb.13991 Santoso, A., Mcphaden, M. J., & Cai, W. (2017). The defining characteristics of ENSO extremes and the strong 2015/2016 El Niño. Reviews of Geophysics, 55, 1079–1129. https://doi. org/10.1002/2017RG000560 Schaefer, K. M., & Fuller, S. W. (2002). Movements, behavior, and habitat selection of bigeye tuna (Thunnus obesus) in the eastern equatorial Pacific, ascertained through archival tags. Fishery Bulletin, 100, 765–788. Schwing, F. B., Murphree, T., deWitt, L., & Green, P. M. (2002). The evolution of oceanic and atmospheric anomalies in the northeast Pacific during the El Niño and La Niña events of 1995‐2001. Progress in Oceanography, 54, 459–491. Séférian, R., Bopp, L., Gehlen, M., Swingedouw, D., Mignot, J., Guilyardi, E., & Servonnat, J. (2014). Multiyear predictability of tropical marine productivity. Proceeding of the National Academy of Sciences, 111(32), 11646–11651. Senina, I., Sibert, J., & Lehodey, P. (2008). Parameter estimation for basin‐scale ecosystem‐linked population models of large pelagic predators: Application to skipjack tuna. Progress in Oceanography, 78, 319–335. Senina, I., Lehodey, P., Kiyofuji, H., Masujima, M., Hampton, J., Smith, N., & Williams, P. (2017). Skipjack Japan impacts of recent high catches of skipjack on fisheries on the margins of  the WCPFC convention area. 13th Scientific Committee of the Western Central Pacific Fisheries Commission, Rarotonga, Cook Islands, 9–17 August 2017. Working Paper WCPFC‐SC13‐2017/SA‐WP‐07, 42 pp. https://www.wcpfc. int/node/29520 Sibert, J., Senina, I., Lehodey, P., & Hampton, J. (2012). Shifting from marine reserves to maritime zoning for conservation of Pacific bigeye tuna (Thunnus obesus). Proceedings of the National Academy of Sciences, 109(44), 18221–18225. Smith, P. (1990). Monitoring interannual changes in spawning area of Pacific sardine (Sardinops sagax). California Cooperative Oceanic Fisheries Investigations Reports, 31, 145–151.

ENSO Impact on Marine Fisheries and Ecosystems  451 Spratt, J. (1979). Age and growth of the market squid, Loligo opalescens Berry, from statoliths. California Cooperative Oceanic Fisheries Investigations Reports, 20, 58–64. Suthers, I. M., Everett, J. D., Roughan, M., Young, J. W., Oke, P. R., Condie, S. A., et  al. (2011). The strengthening East Australian Current, its eddies and biological effects: An introduction and overview. Deep Sea Research Part II: Topical Studies in Oceanography, 58, 538–546. Thatje, S., Heilmayer, O., & Laudien, J. (2008). Climate variability and El Niño Southern Oscillation: Implications for natural coastal resources and management. Helgoland Marine Research, 62(S1), 5–14. https://doi.org/10.1007/s10152‐008‐0104‐0 Tian, Y., Ueno, Y., Suda, M. & Akamine, T. (2004). Decadal variability in the abundance of Pacific saury and its response to climate/oceanic regime shifts in the northwestern subtropical Pacific during the last half century. Journal of Marine Systems, 52, 235–257. Timmermann, A., An, S.‐I., Kug, J.‐S., Jin, F.‐F., Cai, W., Capotondi, A., et  al. (2018). El Niño–Southern Oscillation complexity. Nature, 559(7715), 535–545. Tommasi, D., Stock, C. , A., Hobday, A. J., Methot R., Kaplan I. , C., Eveson, J. , P., et al. (2017). Managing living marine resources in a dynamic environment: The role of seasonal to decadal climate forecasts. Progress in Oceanography, 152 (2017) 15–49. http://dx.doi.org/10.1016/j.pocean.2016.12.011 Tsujino, H., Usui, N. & Nakano, H. (2006). Dynamics of Kuroshio path variations in a high‐resolution general circulation model. Journal of Geophysical Research, 111, C11001. doi:10.1029/2005JC003118 Vance, D. J., Staples, D. J., & Kerr, J. D. (1985). Factors affecting year‐to‐year variation in the catch of banana prawns (Penaeus merguiensis) in the Gulf of Carpentaria, Australia. Journal du Conseil International pour l’Exploration de la Mer, 42, 83–97. Vestfals, C. D., Ciannelli, L., Duffy‐Anderson, J. T., & Ladd, C. (2014). Effects of seasonal and interannual variability in along‐shelf and cross‐shelf transport on ground fish recruitment in the eastern Bering Sea. Deep‐Sea Research Part II, 109, 190–203.

Vojkovich, M. (1998). The California fishery for market squid (Loligo opalescens). California Cooperative Oceanic Fisheries Investigations Report, 39, 55–60. Wang, H., Kumar, A., Wang, W., & Xue, Y. (2012). Influence of ENSO on Pacific decadal variability: An analysis based on the NCEP climate forecast system. Journal of Climate, 25, 6136–6151. Watanabe, H., Kawaguchi, K., & Hayashi, A. (2002). Feeding habits of juvenile surface‐migratory myctophid fishes (family Myctophidae) in the Kuroshio region of the western North Pacific. Marine Ecological Progress Series, 236, 263–272. Weber, E., & McClatchie, S. (2010). Predictive models of northern anchovy (Engraulis mordax) and Pacific sardine (Sardinops sagax) spawning habitat in the California Current. Marine Ecological Progress Series, 406, 251–263. Webster, P. J., Moore, A. M., Loschnigg, J. P., & Leben, R. R., (1999). Coupled oceanic atmospheric dynamics in the Indian Ocean during 1997–98. Nature, 401, 356–360. Williams, D. McB., Cappo, M. C., & Speare, P. J. (1994). Coral Sea region billfish atlas: Seasonal distribution and abundance of billfish species around the Coral Sea rim, Solomon Islands, Papua New Guinea, Vanuatu. Townsville: Australian Institute of Marine Science, 90 p. Yasuda, I. (2003). Hydrographic structure and variability in the Kuroshio‐Oyashio transition area. Journal of Oceanography, 59, 389–402. Yatsu, A., Chiba, S., Yamanaka, Y., Ito, S., Ito, S., Shimizu, Y., et al. (2013). Climate forcing and the Kuroshio/Oyashio ecosystem. ICES Journal of Marine Sciences, 75, 922–933. Yeh, S.‐W., Cai, W., Min, S.‐K., McPhaden, M. J., Dommenget, D., Dewitte, B., et al. (2018). ENSO atmospheric teleconnections and their response to greenhouse gas forcing. Reviews of Geophysics, 56, 185–206. doi.org/10.1002/2017RG000568 Zeidberg, L., Hamner, W., Nezlin, N., Henry, A. (2006). The fishery for California market squid (Loligo opalescens) (Cephalopoda: Myopsida), from 1981 through 2003. Fishery Bulletin U.S., 104, 46–59.

20 ENSO and the Carbon Cycle Richard A. Betts1,2, Chantelle A. Burton1, Richard A. Feely3, Mat Collins4, Chris D. Jones1, and Andy J. Wiltshire1,2

ABSTRACT Observational studies of atmospheric CO2, land ecosystems, and ocean processes show that variability in the carbon cycle is closely related with ENSO. Years with a warm anomaly in the tropical Pacific show a faster CO2 rise due to weaker land carbon sinks, particularly in the tropics, with a partial offset by stronger net uptake by oceans. The opposite happens in years with cool Pacific SST anomalies. This relationship holds for small ENSO SST anomalies as well as large ones and is robust enough for the annual CO2 growth rate anomaly to be highly predictable on the basis of SST observations and forecasts. Generally, variability in the land‐atmosphere carbon flux is mainly driven by physiological processes (photosynthesis and/or respiration), with a smaller contribution from fire. Fire was important in the 1997–1998 El Niño, making a major contribution to the CO2 rise, which can be viewed as anthropogenic in nature since the ignition was caused by humans. However, in the 2015–2016 El Niño event, the change in land carbon flux was mainly due to physiological processes, particularly reduced pro­ ductivity. In the oceans, El Niño conditions involve decreased upwelling of carbon in the equatorial Pacific due to a weakening of the trade winds, causing this region to become a weaker sink of CO2, or near neutral if the El Niño event is strong. The year‐to‐year variations in the rate of CO2 rise can be successfully reconstructed and predicted on the basis of sea surface temperatures in the Pacific. ENSO‐CO2 relationships may also provide an emergent constraint on the strength of climate‐carbon cycle feedbacks on future anthropogenic climate change.

20.1. INTRODUCTION Climate variability related to ENSO plays an impor­ tant part in global carbon and influences the rate at which anthropogenic emissions of CO2 build up in the atmos­ phere year to year. The long‐term increase in atmospheric CO2 concentra­ tions is entirely the result of human‐caused emissions of carbon dioxide into the atmosphere; more than enough CO2 is being emitted by fossil fuel burning, cement pro­   Met Office Hadley Centre, Exeter, UK Global Systems Institute, University of Exeter, Exeter, UK 3   NOAA Pacific Marine Environmental Laboratory, Seattle, WA, USA 4    College of Engineering, Mathematics, and Physical Sciences, University of Exeter, Exeter, UK 1 2

duction, and deforestation to account for the increase measured in the atmosphere (Figure 20.1). CO2 concen­ trations have increased from 278 parts per million (ppm) (Keeling et al., 2001) and passed 410 ppm in 2019. This increase would have been even larger if some CO2 had not been removed from the atmosphere by global vegetation and the oceans. On average, these natural sinks together offset approximately half of anthropogenic emissions (Figure  20.1), although their magnitude varies substan­ tially from year to year (Rayner & Law, 1999). In 2015, for example, natural land and ocean sinks offset less than 40% of anthropogenic emissions (Figure 20.1). It has long been established that the strong interannual variability in the global carbon cycle is related to ENSO (Bacastow, 1976; Bacastow et al., 1980; Keeling & Revelle, 1985; Jones et al., 2001). There are clear correlations bet­ ween the growth rate of atmospheric CO2 and sea surface

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 453

454  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Anthropogenic figures 2008–2017 average PgC per year Carbon cycling PgC per year Stocks PgC

The global carbon cycle Atmospheric CO2

+4.7 [+6.2]

[1.6] 1.5

Fossil CO2

(0.8–2.2)

3.2

0.1

2.4

[1.8]

860 PgC

(2.5–3.9)

9.4

(1.9–2.9)

Ocean uptake

(8.9–9.9)

120

90

450–650 PgC

Permafrost 175–265 PgC

1700 PgC

Soils 1500–2400 PgC

Budget imbalance

Dissolved inorganic carbon

385–1135 PgC

Oil reserves

Atmospheric increase Uncertainity values

Vegetation Gas reserves

Land uptake

+

120

[9.7]

Land-use change

[2.6]

Organic carbon

Rivers and lakes

700 PgC

Coasts 10–45 PgC

Surface sediments

38,000 PgC

90 Marine biota 3 PgC

1750 PgC

Coal reserves 445–540 PgC

Budget imbalance +0.5

Figure 20.1 Components of the global carbon cycle on average and in a year with a strong El Niño event. Colored numbers at heads of arrows are means over 2008–2017. Red numbers in square brackets are values for 2015. 1 petagram of carbon (PgC) is equivalent to 1 gigatonne of carbon (GtC). (Modified from original figure by Le Quéré et al., 2018)

temperatures in the regions of the Pacific used as metrics for ENSO, and there is a long history of study of this relationship and the reasons behind it. The tropical regions, particularly land ecosystems, have been identi­ fied as a key part of the process (Rayner et al., 1999). More recently, measurements of other components of the carbon cycle from an increasing range and variety of sources have provided further evidence for relationships between the carbon cycle and ENSO, and have helped improve understanding of the processes involved. Much of this has been on tropical land ecosystems, although the ocean and other land regions are also covered in some data sets. The large El Niño of 2015–2016 provided a particu­ larly important opportunity to study carbon cycle responses to ENSO, and this has resulted in a wealth of new studies, including a number that arose from a discussion meeting at the Royal Society, London, in November 2017 (Malhi et al., 2018). Many of these pro­ vide important input to this chapter. Furthermore, com­ parisons with the 1997–1998 El Niño show differences in the processes involved in the carbon cycle response to these events. For example, in 1997–1998, widespread dry and warm conditions across the tropics caused reduced gross primary productivity (GPP) and increased eco­ system respiration, whereas in 2015–2016, wetter condi­ tions in parts of Africa caused increased GPP, which partly offset decreases elsewhere (Wang et  al., 2018). Also, fire emissions from Southeast Asia played a major

role in 1997–1998, whereas they were much less substan­ tial in 2015–2016. In addition to contributing to our understanding of the processes of the global carbon cycle and its interactions with the atmosphere and oceans (and their coupled behavior), insights into ENSO–carbon cycle interactions may also be important for our ability to predict future cli­ mate change. For example, if the frequency and/or magni­ tude of ENSO changes as a result of anthropogenic climate change, this could act as a feedback on CO2 rise and climate change. More generally, even if ENSO itself does not change, similar processes, teleconnections, and regional climate changes may nevertheless be associated with a long‐term climate change trend, and hence insights into their impacts on the carbon cycle will be important. Early work with Earth system models (ESMs) on cli­ mate–carbon cycle feedbacks (Cox et al., 2000) discussed the concept of “El Niño–like climate change” (Meehl & Washington., 1996) and the reproduction of observed relationships SSTs in the equatorial Pacific and regional climate responses relevant to carbon cycle impacts (such as precipitation in the Amazon region) was used as sup­ porting evidence for the credibility of models projecting strong carbon cycle feedbacks (Cox et  al., 2004). Moreover, changes in the ENSO regime are also pro­ posed as a potential tipping point in the climate system, with possible implications for tropical land ecosystems and the carbon cycle (Kriegler et  al., 2009). The performance of ESMs in simulating ENSO–carbon cycle

ENSO and the Carbon Cycle  455

interactions on interannual timescales may therefore pro­ vide insight into the role of ENSO‐related processes in carbon cycle feedbacks on anthropogenic climate change in the longer term. In this chapter, we give an overview of variability in the global carbon cycle in relation to ENSO. We begin by describing and quantifying the observed interannual and decadal variability in the global carbon cycle, including atmosphere, land, and ocean components, and how these correlate with variability in metrics for ENSO. We then discuss the processes through which ENSO impacts the carbon cycle on interannual timescales. We present an estimate of the impacts of two major ENSO events, 1997–1998 and 2015–2016, on the annual rise in atmo­ spheric CO2, and discuss how knowledge of the relation­ ship between the CO2 rise and ENSO is now being used to make accurate forecasts of the annual CO2 rise. Finally, we discuss the use of ENSO–carbon cycle inter­ actions on interannual timescales as a constraint on carbon cycle feedbacks on long‐term anthropogenic cli­ mate change. 20.2. CARBON CYCLE VARIABILITY AND ITS CORRELATION WITH ENSO 20.2.1. Atmospheric CO2 Growth Rate Measurements of the atmospheric CO2 concentration have been taken continuously at Mauna Loa Observatory, Hawaii, since 1958 and are good indicator of global mean CO2 on annual timescales. When measurements began, the annual mean CO2 concentration at Mauna Loa was 316 ppm. In 2019, annual mean concentrations passed 410 ppm and continue to rise at an accelerating rate as a result of anthropogenic emissions. Comparison of the year‐by‐year increment in atmo­ spheric CO2 with the rate of anthropogenic emissions reveals two important features: a clear increase in both emissions and concentrations growth in the long term, but also a large short‐term variability in the concentra­ tions growth rate that is not seen in the emissions (Figure  20.2). There are very few years when the emis­ sions rate has been smaller than that of the previous year, and such cases have only been a few tenths of a GtC. In contrast, the annual CO2 increment varies by between 1 and 2 GtC from year to year. It has long been noted that the interannual variability in CO2 concentration incre­ ments relates closely to ENSO, with peaks in the CO2 increment generally coinciding with warm sea surface temperature (SST) anomalies in the tropical Pacific (Figure  20.2; Bacastow, 1976; Bacastow et  al., 1980; Keeling & Revelle, 1985). The exception is in years fol­ lowing large volcanic eruptions, particularly Pinatubo in 1991 (Jones et al., 2001)

There is a strong correlation between the annual CO2 increment and SSTs anomalies in the equatorial Pacific for both large and small variations. The correlation is strongest when annual mean SSTs are taken from April to March (Jones et al., 2001). There are obvious large spikes in CO2 growth coinciding with El Niño years, and troughs coinciding with La Niña years (Figure  20.2). The exception to this is when El Niño events coincide with a major volcanic eruption, which temporarily cools the global climate due to the injection of aerosol particles into the atmosphere that reflect some of the incoming solar radiation back to space. It is also clear that gener­ ally the CO2 growth rate relates to Pacific SST anomalies, even in years with smaller SST anomalies that are not for­ mally identified as “El Niño” or “La Niña” (Figure 20.2). Mechanisms of the relationship with the carbon cycle must therefore still be present even when a full ENSO event has not emerged. As well as the correlation between annual means, apparent relationships between the CO2 increment and SSTs can also be seen on longer timescales. For example, after the concentration increment increased at approxi­ mately half the rate of increase of emissions for the first four decades of the record, from the mid‐2000s onwards the annual CO2 increment ceased to increase for around a decade. The CO2 increment varied around an average of approximately 2.1 ppm yr‐1, despite emissions rising more rapidly at that time (Keenan et al., 2016). This coincided with a period of relatively cool tropical Pacific SSTs, with La Niña episodes in 2007–2009 and 2010–2012 and with only small positive SST anomalies in between. The time series of CO2 increments can be reconstructed from a multiple linear regression of annual CO2 incre­ ments (ΔCO2) against annual anthropogenic emissions (ε) and the annual mean anomaly in sea surface tempera­ tures in the region of the equatorial Pacific Ocean char­ acterising ENSO activity (N) (Jones & Cox, 2005):

co2

1

2

N

3

, (20.1)

where α1, α2, and α3 are regression coefficients calculated using the observed records of ΔCO2, N, and ε. Table 20.1 shows values of these coefficients calculated using observed records up to 2017. This method can successfully reproduce the variability in the observed CO2 rise (Figure 20.2), although it has limita­ tions, particularly concerning the choice of the specific region of the Pacific for N, because the carbon cycle can respond differently to ENSO temperature anomalies occurring in dif­ ferent parts of the equatorial region (Chylek et al., 2018). For example, the 1997–1998 El Niño was different in character to that of 2015–2016, with the temperature anomaly being focused more in the central Pacific (Niño‐3 region) in

456  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a) 12 10

PgC yr–1

Emissions

6

3

Hindcast

4

ppm yr–1

4

8

2 Annual increment

2

1

0 EI Niño events

1966

1973

Volcanic Eruptions (b) 3 Niño 3.4 index

1983

1988

EI Chichon

1998

2016

Pinatubo

2

°C

1 0 –1 Observed annual –2 1960

1970

1980

2000

1990 year

2010

2020

Figure 20.2  Relationships between annual CO2 increments at Mauna Loa and sea surface temperatures in the Niño‐3.4 region of the Pacific, including a statistical reconstruction of the annual increment based on SSTs. (a) Anthropogenic emissions from the Global Carbon Project data set (Le Quéré et al., 2018) and CO2 concentrations at Mauna Loa from observations (black) and reconstruction from emissions and SSTs using equation (20.1) (blue). (b) Annual mean April–March SSTs from the HadSST3.1.1.0 data set (Kennedy et al., 2011a, 2011b), with anomalies in °C calculated relative to the 1961–1990 mean.

1­997–1998 and more in the eastern Pacific (Niño‐3.4) in 2015–2016 (Santoso et  al., 2017). Jones and Cox (2005) applied Equation 20.1 using Niño‐3 SST, whereas Betts et al. (2016) used Niño‐3.4. The hindcast reconstruction in Figure  20.2 here also uses Niño‐3.4. Nevertheless, use of either region gives a good reconstruction overall. 20.2.2. Flows of CO2 Between the Atmosphere and Surface The rate of uptake of CO2 from the atmosphere by the surfaces of the global land and ocean has been increasing over recent decades (Figure  20.3), with the current rate being approximately 4 PgC yr‐1. There is considerable year‐to‐year variability in this, and while the long‐term trend in uptake is similar for both the land and ocean fluxes, the main contribution to interannual variability

comes from the land flux. Generally, the atmosphere‐land flux temporarily becomes smaller (i.e. a weaker net land sink) in El Niño years while the atmosphere‐ocean flux temporarily becomes slightly larger (i.e. a stronger net ocean sink). Much of the research on quantifying and understanding variability in the carbon cycle in relation to ENSO therefore focuses on the atmosphere‐land fluxes, although studying the atmosphere‐ocean fluxes is also important. Table 20.1  Regression coefficients for Equation (20.1) calculated using observed records up to 2017 (Betts et al., 2018). α1 (ppm yr‐1)

0.045

α2 (ppm yr °C ) α3 (ppm PgC‐1) ‐1

‐1

0.426 0.214

ENSO and the Carbon Cycle  457

20.2.2.1. Atmosphere-land CO2 Flows The flows of CO2 between the atmosphere and surface can be quantified with inversion modeling, which com­ bines measurements of atmospheric CO2 concentrations with atmospheric transport models constrained by obser­ vations (e.g. Röden beck et al., 2018). These simulate the variations in atmosphere‐land fluxes both geographically and over time. The net flux of carbon from the atmosphere to land is generally increasing over time, indicating a long‐term increase in carbon uptake by land ecosystems. However, the flux becomes stronger in La Niña events and weaker, or even negative, in El Niño events (Figure 20.3). Broadly similar behavior is seen both in northern extratropical land and tropical land but with both the trend and inter­ annual variability more marked in the tropics. Overall, land regions take up less CO2 (or emit more) during El Niño and take up more CO2 (or emit less) during La Niña. In most years since the 1960s, global land ecosystems have been a net sink of carbon, with the mean rate of uptake increasing to over 2 PgC yr‐1 by the 2010s. However, until 2003, years with El Niño events or smaller warm SST anomalies saw global land ecosystems became a net source of carbon. In 1997–1998, the terrestrial biosphere released approximately 2 PgC yr‐1 to the atmosphere, representing an anomaly of approximately 3 PgC yr‐1 relative to the long‐term mean sink. Most of this was accounted for by emissions from the tropics. After 2003, the long‐term increase in global sink strength meant that decreases in the sink strength associated with warm SST anomalies were not enough to turn the terrestrial biosphere into a net source of carbon. Even though the large El Niño of 2015– 2016 reduced the sink strength substantially, the net land‐ atmosphere carbon flux remained approximately neutral rather than becoming a net carbon source. The maximum global net land sink strength has so far occurred in 2011– 2012 and 2013–2014, reaching approximately 3 PgC yr‐1. At regional scales, the ENSO response of the atmosphere‐land CO2 flux varies substantially. During El Niño, land south of 25° becomes an overall net source of carbon, and almost all areas of tropical land show positive anomalies (i.e. more outgassing of CO2 (or reduced uptake) from land to the atmosphere; Figure  20.4). South America includes a particularly strong relative source (or weaker sink), with eastern Amazonia and northeast Brazil showing a major anomaly, especially in the strong El Niño events of 1997 and, to a lesser extent, 2015. Land north of 25° remains an overall carbon sink but is weaker than average, and some mid‐ and high‐latitude regions show negative anom­ alies (i.e. less CO2 outgassing, a stronger net carbon sink locally; Figure  20.4). These general patterns are seen in most El Niño events, but the magnitude of the local anomalies varies between events.

20.2.2.2. Atmosphere-ocean CO2 Flows The flux of CO2 between the atmosphere and ocean is estimated with a number of methods. One method uses a self-organizing neural network combined with the effec­ tive partial pressure of CO2 (the partial pressure pCO2 corrected to account for CO2 not behaving as an ideal gas) database provided by the Surface Ocean CO2 Atlas (SOCAT; Bakker et al., 2016, Landschutzer et al., 2016). Another method uses a combined measure of the ratio of oxygen and nitrogen concentrations and CO2 concentra­ tions. The ratio, known as atmospheric potential oxygen, is insensitive to changes in CO2 from the terrestrial bio­ sphere and hence provides a quantification of changes in CO2 associated with the oceans (Battle et  al., 2006; Keeling & Manning, 2014). An inversion method com­ bines data‐based estimates of anthropogenic CO2 in the ocean with simulated ocean transport and mixing from ocean general circulation models (Mikaloff Fletcher et  al., 2006). The distribution of chlorofluorocarbons (CFCs) in the ocean can also be used to infer CO2 uptake (McNeil et  al., 2003). The Global Carbon Project (Le Quéré et  al. 2018) compiles estimates from all these methods along with estimates from global ocean biogeo­ chemistry models to produce a combined assessment of atmosphere‐ocean CO2 fluxes (Figure 20.3c). The interannual variability in atmosphere‐ocean CO2 fluxes is of the order of a few tenths of a PgC, much less than the variability in atmosphere‐land fluxes. Nevertheless, correlations with ENSO are apparent. The global mean atmosphere‐ocean CO2 fluxes respond to ENSO in generally the opposite direction to the atmosphere‐land flux response, with an increase in the flux of CO2 from atmosphere to ocean in El Niño years, and a decrease in La Niña years. 20.3. PROCESSES INVOLVED IN ENSO-CARBON CYCLE INTERACTIONS 20.3.1. Terrestrial Ecosystem Processes The net flow of carbon between the atmosphere and land is known as net ecosystem exchange (NEE) or alter­ natively net biosphere productivity (NBP). This consists of several components. Gross primary productivity (GPP) is the uptake of carbon by plants through photo­ synthesis. Plants also release carbon via respiration, and the difference between GPP and plant respiration is net primary productivity (NPP). Carbon is also released by respiration by microbes in the soil, and by animal life. The difference between NPP and nonplant respiration is net ecosystem productivity (NEP). Carbon is also released through disturbance processes such as fire. The difference between NEP and disturbance emissions is NBP or NEE. NBP/NEE represent the net flux of carbon

458  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a)

Emissions and Sinks

12.00 10.00

PgC yr–1

8.00 6.00 4.00

Fossil fuel and industry emissions

(b)

Land-use change emissions

Ocean sink

2017

2015

2011

2013

2009

2007

2005

2001

2003

1999

1997

1995

1993

1991

1989

1987

1985

1983

1981

1979

1977

1975

1973

1971

1969

1967

1965

1963

1961

0.00

1959

2.00

Land sink

Niño3.4 index

2.5 2 1.5 1

°C

0.5 0 –0.5 –1

2018

2016

2014

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1972

1970

1968

1966

1964

1960

–2

1962

–1.5

Figure 20.3  Relationships between components of change in the carbon cycle and sea surface temperatures (SSTs) in the Niño‐3.4 region of the Pacific. (a) Emissions from fossil fuel burning and land use change separately, and land and ocean carbon sinks (Le Quéré et al., 2018). (b) Annual mean April–March SSTs from the HadSST3.1.1.0 dataset (Kennedy et al., 2011a, 2011b), with anomalies in °C calculated relative to the 1961–1990 mean.

between the atmosphere and land. The following sections discuss the changes in GPP, NPP, and fire emissions in relation to ENSO, as well as some aspects of regional cli­ mate variability that appear to drive these. 20.3.1.1. Vegetation Productivity NPP is the net carbon uptake by plants, including GPP and plant respiration. NPP can be estimated for individual trees by measuring their change in stem diameter, and this can be used to estimate large‐scale NPP, scaling up large numbers of measurements from individual sites.

Using measurements of wood allocation to tree stems (NPPstem) of 8725 tropical trees in 50 sites across 14   regions in the Global Ecosystems Monitoring net­ work and a set of site‐specific statistical relationships with meteorological variables, Rifai et  al. (2018) esti­ mated the impacts of climate variability on pan‐tropical woody production from 1996 to 2016. NPPstem shows clear relationships with ENSO in all three major tropical forest areas (Figure  20.5). With the long‐ term increasing trend removed for clarity, the 12‐month running mean anomaly generally shows declines during

ENSO and the Carbon Cycle  459 (a) 1965

1982

1972

1997

1986

2015

2009

(b) mean

150 more CO2 outgassing 0 more CO2 uptake –150

Figure 20.4  Anomalies in regional land‐atmosphere CO2 fluxes (gC m‐2 yr‐1) during El Niño events (a) and the mean (b), inferred from atmospheric CO2 observations using a statistical relationship between station CO2 data and the Multivariate ENSO Index. (Reproduced from Rödenbeck et al., 2018)

periods of warm Niño‐3.4 SSTs and increases during periods of cool Niño‐3.4 SSTs. This is evident for periods of small SST anomalies as well as the major anomalies in El Niño and La Niña events. The impacts are generally largest in the Americas and smallest in Africa. All three major tropical forest areas were impacted to some degree by the major El Niño events in 1997–1998 and 2015–2016, but the impacts were much larger in the Americas and quite minimal in Africa (Figure  20.5). NPPstem in the Americas and the Asia‐Pacific region were also impacted to some extent by smaller positive Niño‐3.4 anomalies, particularly 2010–2011. Both of these regions also showed positive NPPstem anomalies in La Niña events, as did Africa in the 2008–2009 La Niña. GPP can be estimated with satellite remote sensing with a variety of methods based on the absorption of solar radiation by plants as part of photosynthesis, in combination with other quantities. These often use the fraction of absorbed photosynthetically active radiation, derived from the Normalized Difference Vegetation Index (NDVI), which is a measure of the relative ratios of reflected and incoming radiation in red and near infrared wave bands. Live vegetation absorbs and reflects red and near infrared radiation very differently to other surfaces. NDVI is provided by several satellite instruments, including the MODerate Resolution Imaging Spectrometer

(MODIS) and Advanced Very High Resolution Radio­ meter (AVHRR). These data show a general rising trend in GPP from 2000 to 2016, with a reduction in 2015 associated with the El Niño and an increase in 2011 associated with the La Niña (Luo et al., 2018) After removing the rising trend in global GPP due to CO2 fertilization, increased growing season length, and other long‐term climate trends, Luo et al. (2018) found that global GPP decreased by 0.70 ± 1.20 PgC in 2015, accounting for 60% of the reduction in NEP (net ecosystem productivity). However, in 2016, there was a small increase in GPP of 0.05 ± 0.89 PgC. Because the overall carbon sink still declined in 2016, which cannot be explained by decreased productivity, this implies that the continuation of the weaker net land sink in 2016 was due to increases in carbon release due to res­ piration or fire. Variability in large‐scale photosynthesis therefore appears to play an important role in the variability of the atmosphere‐land CO2 flux, but it is not the only process responsible. 20.3.1.2. Fire Global CO2 emissions from wildfire show a relation­ ship with ENSO, driven predominantly by variability in emissions from tropical regions. Although the details of

460  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE tropical forest NPPstem anomaly Africa 0.1 0 –0.1 Americas NPPstem PgC yr–1

0.1 0 –0.1 Asia-Pacific 0.1 0 –0.1 pantropical 0.1 0 –0.1 1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017

Niño-3.4 index –1

0

1

2

Figure 20.5  Twelve‐month detrended, running mean anomalies of NPPstem (black line) in the three major tropical regions from 1996 to 2017 estimated with a statistical model relating NPPstem to meteorological conditions. Vertical colored bars show the Niño‐3.4 index. (Reproduced from Rifai et al., 2018)

regional weather in individual events play an important role, other factors such as direct human impact via land use, drainage of wetlands, and human ignition of fires play a very large role. The nature of the ecosystems affected by fire is also crucial, with fire in peatlands being particularly important due to the very large carbon stocks they contain and hence potential for a high rate of emissions. Emissions from fire featured prominently in the 1997– 1998 El Niño event, exceeding 3 PgC yr‐1, and since then global fire emissions have fluctuated between approxi­ mately 1.8 and 2.3 PgC yr‐1, with minima generally in La Niña events and maxima in El Niño events (Figure 20.6a). Generally, global fire emissions are showing a down­ ward trend. The very large spike in global fire emissions in 1997– 1998 was almost entirely from equatorial Asia, more spe­ cifically from peatlands in Southeast Asia, which were under severe drought and subject to direct human distur­ bance. Page et al. (2002) estimated that between 0.81 and 2.57 PgC were released from Indonesia in 1997, with the upper estimate representing a very large proportion of the global CO2 flux anomaly estimated from inversion studies. However, the Global Fire Emissions Database (GFED) gives emissions of 0.53 PgC from equatorial Asian peatlands in 1997 (Giglio et al., 2013), and this is

used in the global carbon budget (e.g. Le Quéré et  al. 2018). In contrast, in 2015, fires in Indonesia resulted in much lower emissions of 0.35–0.60 PgC (Nechita‐Banda et al., 2018). One reason for the lower emissions compared to 1998–1998 is that in 2015 the rains returned in November. GFED may have overestimated fire emissions in Indonesia in 2015, as it used the same relationships bet­ ween burnt area and emissions as were used in 1997–1998, but in the areas that have been repeatedly burnt, there is less carbon per unit area to be released (Sue Page, personal communication) However, ground‐based studies in Amazonia suggest GFED underestimates emissions for this region (Withey et al., 2018). Global fire emissions were 2.4 PgC yr‐1 in 2015 and 1.9 PgC yr‐1 in 2016, compared to an average of 2.0 PgC yr‐1 from 2010 to 2014 (Bastos et al., 2018). Regional contributions to global fire emissions vary substantially (Figure  20.6b). Total fire emissions from Africa are consistently high (between 600 and 700 TgC yr‐1 from Northern Hemisphere Africa, and 400 to 600 TgC yr‐1 from Southern Hemisphere Africa) and show relatively little interannual variability. Emissions from Southern Hemisphere Africa show a long‐term decline. Most other regions have total annual emissions below 200 TgC yr‐1, with relatively small interannual variability.

ENSO and the Carbon Cycle  461

The exceptions are equatorial Asia and Southern Hemisphere South America, which show large interan­ nual variability with high emissions in El Niño years. Notably, the Global Carbon Budget data set of anthro­ pogenic CO2 emissions shows a clear spike in 1997–1998 (Figure  20.2a), which arises from land use emissions (Figure 20.3a). This features in the data set because the peatland fire emissions from Southeast Asia are classed as anthropogenic (Houghton & Nassikas, 2017) since

they were ignited by humans. However, the climatic con­ ditions were suitable for fires to run out of control, so the El Niño event played a part, too. 20.3.1.3.  Regional Climate Drivers of  Land Carbon Cycle Variability The global land‐atmosphere carbon fluxes are clearly related to ENSO, but what are the climate processes that lead to this connection? Atmospheric CO2 growth rate is related

Global CO2 emissions from fire (PgC yr–1)

(a) 3 2.8 2.6 2.4 2.2 2 1.8 1.6 1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017

CO2 emissions from fire by region (TgC yr–1)

(b) 1200 1000 800 600 400 200 0 1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

BONA

TENA

CEAM

NHSA

SHSA

EURO

MIDE

NHAF

SHAF

BOAS

TEAS

SEAS

EQAS

AUST

2017

Figure 20.6  Annual emissions of CO2 from fire (a) globally and (b) by region. BONA: Boreal North America; TENA: Temperate North America; CEAM: Central America; NHSA: Northern Hemisphere South America; SHSA: Southern Hemisphere South America; EURO: Europe; MIDE: Middle East; NHAF: Northern Hemisphere Africa; SHAF: Southern Hemisphere Africa; BOAS: Boreal Asia; TEAS: Temperate Asia; SEAS: South East Asia; EQAS: equatorial Asia; AUST: Australia and New Zealand. Source: GFED.

462  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

at the global scale to both land temperature and water avail­ ability (Humphrey et  al., 2018), with higher growth rates corresponding to warm, dry anomalies (Humphrey et  al., 2018). Links to temperature have long been established (Wang et al., 2013). More recently, large‐scale estimates of changes in terrestrial water storage (TWS) variability from 2002 to 2017 have become available from satellite remote sensing with GRACE (Gravity Recovery And Climate Experiment), which measured local changes in the Earth’s gravity with a pair of satellites. TWS anomalies have also been reconstructed for before the start of the gravimetric measurement period through a modeled relationship with other observed climate variables (Humphrey et  al., 2018). The resulting TWS time series shows a strong inverse corre­ lation with CO2 growth rate (Figure  20.7), with generally higher growth rates in years with smaller TWS, which coin­ cide with periods of warm ENSO SST anomalies. The major exception to this was in 1992–1993, when despite a large TWS anomaly showing widespread dry conditions, the CO2 growth rate was anomalously small. This can be accounted for by the temporary cooling effect of the Mount Pinatubo eruption, which would have been expected to reduce both plant respiration and soil respi­ ration, hence reducing outgassing of CO2. Although the TWS‐CO2 relationship is robust, there are some notable differences in the relationship between the two major El Niño events of 1997–1998 and 2015–2016. In 2015–2016 the negative TWS anomaly is the largest on record at over –2 Tt H2O, and the CO2 growth rate is the second largest in the record by the measure used in this study. In 1997–1998, the global mean TWS anomaly is not substantially larger than in many other years, whereas the CO2 growth rate was the largest in the record by the measure used here. This may suggest that while soil moisture played an important role in 2015–2016, other important factors in 1997–1998 either amplified the response to soil moisture changes or added to it. For example, the very large store of carbon in Indonesian peatlands meant that the drought‐ related wildfires released substantial quantities of carbon. In 2015–2016, the reductions in GRACE soil moisture were largest in tropical South America and smaller in southern Africa and Southeast Asia (Gloor et al., 2018). However, soil moisture may not have been the sole driver of terrestrial carbon cycle changes in 2015–2016. Koren et al. (2018) suggested that in the Amazon, humidity may have been the strongest driver of GPP impact in 2016, as productivity recovered early in 2016 as atmospheric mois­ ture demand returned to normal levels even though soil moisture drought conditions persisted. 20.3.2. Ocean Processes The interannual variability of the sea‐air exchange fluxes in tropical oceans is largely controlled by two factors: (i) the variability in partial pressure of CO2

(pCO2) in the surface waters and (ii) the wind forcing, influenced by the nature and phasing of ENSO events, particularly in the tropical Pacific (e.g. Feely et al., 1999, 2002, 2006; Ishii et al., 2009, 2014; Takahashi et al., 2009; Wanninkhof et al., 2013; Landschutzer et al. 2014, 2016). During non–El Niño and La Niña periods, the central and eastern equatorial Pacific is the major oceanic source of CO2 to the atmosphere. It is near neutral during strong El Niño periods and a weak source during weak El Niño periods. The La Niña phase of the ENSO cycle is charac­ terized by strong trade winds, cold tropical SSTs, and enhanced upwelling along the equator. The mean sea‐air CO2 flux based on all observational data is 0.51 ± 0.24 PgC yr‐1 for the tropical Pacific, 0.10 PgC yr‐1 for the tropical Atlantic, and 0.1 PgC yr‐1 for the tropical Indian Ocean (Takahashi et al., 2009; Ishii et al., 2014). The warm El Niño phase of the ENSO cycle is charac­ terized by a large‐scale weakening of the trade winds, decrease in upwelling of CO2 and nutrient‐rich subsur­ face waters, and a corresponding warming of SST in the eastern and central equatorial Pacific. The cold tongue in the eastern tropics extends to the west past the interna­ tional date line during the cold La Niña events and retreats to the east during the warm El Niño events. ENSO drives changes in the distributions of DIC, SST, and salinity in surface water as well as the surface wind field, and causes large perturbations to surface water pCO2 and significant temporal variability in the CO2 ­outgassing from the tropical Pacific. During the strong eastern Pacific El Niño events of 1982–1983, 1997–1998, and 2015–2016, the cold waters of the eastern equatorial Pacific disappear and pCO2 values are close to equilibrium with the atmosphere. During the weaker central Pacific El Niños of 1991–1994, 2002–2005, and 2006–2007, the equatorial cold tongue is present but less pronounced, and pCO2 values are higher than atmospheric values but lower than corresponding values for non–El Niño periods. These findings are consistent with the westward shift of SST anomalies toward the central Pacific during central Pacific El Niños resulting from weaker upwelling there compared with the stronger SST anomalies in the far eastern Pacific during a strong eastern Pacific El Niño (Landschutzer et al., 2016). Figure 20.8, reproduced from Feely et al. (2019), presents time‐longitude plots of SST and pCO2 for the region from 5°N to 10°S and 130°E to 95°W, and the Oceanic Niño Index for the 36‐yr period from 1982 to 2018. The strongest El Niño event of 1997– 1998 featured SST anomalies exceeding 4°C and the lowest pCO2 values throughout most of the equatorial Pacific. In contrast, the 2015–2016 El Niño event fea­ tured SST anomalies that were similar in magnitude to the 1997–1998 event, although the pCO2 values were sig­ nificantly higher because the upwelling‐favorable winds were stronger in the easternmost and westernmost parts of the region. Recent satellite OCO‐2 CO2 data and

ENSO and the Carbon Cycle  463 GRACE (satellite) GRACE-REC (model) CO2 growth rate (reversed axis)

(a)

(b) 1.5

0

2 4 1980

1985

1990

–2 –4 1995

2000

2005

2010

Time

2015

CGR (PgC yr–1)

0

TWS (Tr H2O)

2

Mt Pinatubo

CGR (PgC yr–1)

–2

1

2016 2015 2003

0.5 0 –0.5

2005 2012 2010 2013 2007 2004

–1 –1.5 –1.5

2002

2006 2014 2008 2009

2011

y = –1.33x (r = –0.85) –1

–0.5 0 0.5 TWS (Tt H2O)

1

1.5

Figure 20.7  (a) Time series of CO2 growth rate (CGR: black) and terrestrial water storage (TWS) from GRACE satellite measurements (dark blue) and reconstructed from observed climate variables using relationships with GRACE measurements (light blue). (b) Annual CGR versus TWS. (Reproduced from Humphrey et al., 2018)

mooring‐based observations by Chatterjee et  al. (2017) show that during the early stages of the 2015–2016 ENSO event, atmospheric CO2 concentrations in the tropical Pacific indicated a flux reduction by 26% to 54% that was due to the decrease in the sea‐air CO2 flux in the tropical Pacific. This was followed by a dramatic increase in atmo­ spheric CO2 after August 2015 attributed to terrestrial processes as described above. 20.4. IMPACTS OF MAJOR EL NIÑO EVENTS ON THE GLOBAL CARBON CYCLE The observed annual CO2 increment from 2015 to 2016 was 3.39 ppm, the largest in the Mauna Loa record. Prior to this, the previous largest annual increment was 2.9 ppm between 1997 and 1998. The observed CO2 annual incre­ ment in 2015–2016 was therefore substantially larger than that following the previous large El Niño event. How much did each El Niño event contribute to the CO2 rise, and how much was due to human emissions? Why was the 2015–2016 rise larger than that of 1997–1998? We can make a simple estimate of the contribution of specific El Niño events to the annual CO2 rise by calcu­ lating Equation (20.1) both with the observed SST anomaly for N and with no anomaly (i.e. N = 0; Betts et al., 2018). Using the observed emissions and SST anomaly, the reconstructed CO2 increment for 2015–2016 is 3.21 ppm (Table 20.2). In comparison, the observed CO2 increment was 3.39 ppm. Using a zero SST anomaly, we estimate that the 2016 annual CO2 increment without El Niño would have been 2.42 ppm, so this method suggests that the El Niño increased the 2016 annual CO2 increment by 0.79 ppm. We therefore estimate that the El Niño contrib­ uted approximately 25% to the record annual CO2 rise

between 2015 and 2016, with the other 75% being due to anthropogenic CO2 emissions. For 1997–1998 the reconstructed CO2 increment is 2.59 ppm, and the observed rise was 2.9 ppm. This underestimate might be due to the Niño‐3.4 index’s being less representative of the magnitude of the 1997–1998 El Niño; the Niño‐3 region might be more representative. The estimated “no El Niño” CO2 rise is 1.82 ppm, implying an El Niño contribution of 0.77 ppm. This is similar to that for 2015–2016 in terms of the absolute contribution, but it is a larger proportion (30%) of the overall rise that year. An important further point for the 1997–1998 El Niño is that the anthropogenic emission itself includes an ENSO‐related contribution, as much of the land use emissions were from burning peatlands in Southeast Asia (Le Quéré et  al., 2018). Although the fires were ignited by humans and hence are counted as anthropo­ genic in the global carbon budget, the El Niño–related drought provided the conditions for the fires to run out of control and produce very large emissions (Page et al., 2002). The total anthropogenic emissions in 1997 were 8.31 PgC, with 6.53 PgC from fossil fuel combustion and 1.78 PgC from land use change, including the Asian peatland fires (Le Quéré et al., 2018). By linearly interpolating bet­ ween the emissions for the previous and following years (1.31 PgC for 1996 and 1.23 PgC for 1998), we estimate that without the El Niño–related peatland fires, global land use emissions in 1997 would have been 1.27 PgC, so total anthropogenic emissions would have been 7.80 PgC. This interpolation implies that the contribution of the Asian peatland fires to anthropogenic land use emissions was 0.51 PgC, close to the total value for Asian peatland fires used in GFED (Giglio et al., 2013).

464  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE (a)

SST (°C) 32

105°W

30

135°W

28

165°W

26 24

165°E

22

135°E 1985

(b)

1990

1995

2000

2005

2010

2015

SST (°C)

pCO2 (μatm)

20

525

105°W

505 485

135°W

465

165°W

425

165°E

385

405 365

135°E 1985

(c)

1990

1995

2000

2005

2010

2015

345 pCO2 (μatm)

Oceanic Niño Index (ONI) 2 1 0 –1 –2 1985

1990

1995

2000

2005

2010

2015

Year

Figure 20.8  Time‐longitude plots of (a) SST, (b) pCO2, and the (c) Oceanic Niño Index from 1982 thru 2018 in the equatorial Pacific. Significant reductions in surface water pCO2 values (low CO2 outgassing) correspond with the El Niño events of 1982–1983, 1986–1987, 1991–1994, 1997–1998, 2002–2005, 2006–2007, 2009–2010, and 2015–2016. Significant enhancement of the pCO2 values (high CO2 outgassing) occurred with the strong La Niña events in 1984–1985, 1998–1999, 1995–1996, 1998–2000, 2007–2008, and 2011–2012. 2018 was a normal non–El Niño year. (Reproduced from Feely et al., 2019)

Repeating the estimate of the “no El Niño” CO2 rise for 1997–1998, including this emissions contribution, gives a revised figure of 1.71 ppm, so the revised estimate of the El Niño–related contribution to the annual CO2 increment is 0.88 ppm. This is 34% of the total reconstructed CO2 rise, so it is estimated that the reduced net carbon sink associated with the 1997–1998 El Niño contributed approximately one third of the CO2 rise between 1997 and 1998. The observed mean CO2 rise for the decade prior to 2015 was steady at approximately 2.1 ppm yr–1, so the rise of 3.39 ppm in 1 year was a substantial increase. Before 2015, the growth rate did not rise despite an increase in

anthropogenic emissions, and this has been attributed to an increased net uptake of carbon by the terrestrial bio­ sphere due to increased CO2 fertilization, accompanied by a lack of increase in respiration resulting from the temporary slowdown in the rate of global warming (Keenan et al., 2106). We note that our reconstruction of CO2 increments using Equation (20.1) captures this hiatus in the rate of CO2 rise between approximately 2003 and 2014 (Figure 20.1). Since the only climate‐related term in Equation (20.1) is the Niño‐3.4 SST anomaly, this sug­ gests that the relatively cool conditions in the equatorial Pacific in several of these years may have played a role in the hiatus in the CO2 rise. La Niña conditions are associated

ENSO and the Carbon Cycle  465 Table 20.2  Estimating the contribution of El Niño to the annual CO2 increment in 2015–2016. N (°C) ε (PgC) ΔCO2 (ppm)

Period

Observed

Calculation With El Niño SSTs

Calculation Without El Niño SSTs

April 2015–March 2016 January–December 2015 2015 – 2016

1.85 ± 0.19 11.1 3.39

1.85 ± 0.19 11.1 3.21

0 11.1 2.42

Table 20.3  Estimating the contribution of El Niño to the annual CO2 increment in 1997–1998. Period N (°C) ε (PgC) ΔCO2 (ppm)

April 1997–March 1998 January– December 1997 1997–1998

Calculation With El Observed Niño SSTs

Calculation Without El Niño SSTs

Calculation Without El Niño SSTs or Land Use Emissions

1.81

1.81

0

0

8.3

8.3

8.3

7.8

2.9

2.59

1.82

1.71

with smaller annual CO2 rises (Figures  20.2 and 20.3), with generally wetter and cooler conditions in many areas. This may be consistent with an emerging under­ standing of the role of Pacific decadal variability in the global warming hiatus (Meehl et al., 2011; Kosaka & Xie, 2013; England et al., 2014). Therefore, although the large increase in CO2 rise in 2015–2016 was largely associated with El Niño, there was also probably a contribution from the cessation of the anomalously slow rate of rise associated with cooler con­ ditions in the tropical Pacific. It therefore appears that the main reason why the CO2 annual increment in 2015–2016 was larger than 1997– 1998 is that anthropogenic emissions from fossil fuel burning increased substantially in that period. Although land use emissions including fire made a smaller contri­ bution to the CO2 rise in 2015–2016 compared to 1997– 1998, total anthropogenic emissions including fossil fuel burning had risen by 2.7 PgC between 1997 and 2015 (an increase of over 30%). 20.5. ROLE OF ENSO IN PREDICTING THE FUTURE BEHAVIOR OF THE EARTH SYSTEM 20.5.1. Forecasting the CO2 Rise on Annual Timescales Since the annual CO 2 growth rate is closely linked to ENSO, and ENSO has some predictability (see chapter 10), this means that it is possible to make success­ ful forecasts of the annual CO2 rise ahead of time (Betts et al., 2016; Betts et al., 2018). Such forecasts are now a routine product from the Met Office (https://www. metoffice.gov.uk/news/releases/2019/2019‐carb­ondioxide‐ forecast). These provide a useful learning experience for testing understanding of climate–carbon cycle links:

making a specific prediction and then looking closely at the reasons for success or failure can be instructive. Also, the ongoing rise in CO2 is of interest to those members of the public, media, and policy world who are concerned with anthropogenic climate change, so making and test­ ing predictions of the CO2 concentration can provide a tool for raising awareness. The Met Office forecast method uses the statistical rela­ tionship between the annual CO2 increment, emissions, and equatorial Pacific SSTs in Equation (20.1). The fore­ cast is calculated in November using annual emissions from the Global Carbon Project for the current calendar year (which includes a projection for the end of the year), and Niño‐3.4 SSTs from HadISST observations (Kennedy et al., 2011a, 2011b) from the preceeding April to October, combined with a forecast of SSTs from the GloSea cli­ mate model (MacLachlan et  al., 2015) for the coming November to March. These are used to calculate the pre­ dicted increment in annual mean atmospheric CO2 concentration at Mauna Loa from the current year to the following year. This is then used to forecast the annual mean CO2 concentration across the following year. Using an assumption of a uniform seasonal cycle, the monthly mean CO2 concentrations are also predicted, with a particular focus on the annual maximum monthly concentration in May and the annual minimum monthly concentration in September (Betts et al., 2018). The Mauna Loa measurements are chosen as the focus of the Met Office CO2 forecast, as they provide a very specific, precisely measured quantity; in contrast, the global mean CO2 concentration relies on estimates and assumptions, which introduce additional uncertainties. Although CO2 is measured at other sites, Mauna Loa is the original measurement site and provides the longest record as well as being of historic interest.

466  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

The Met Office CO2 forecast was first produced in 2015, predicting the annual increment between that year and 2016 (Betts et  al., 2016). This was of particular interest for two reasons: (i) the large El Niño emerging in late 2015 and (ii) atmospheric CO2 concentrations being around 400 ppm, not a threshold of physical significance but nevertheless a symbolic threshold as rising concentra­ tions passed from the 300s to the 400s. Since then, the CO2 forecast has been issued on an annual basis. The CO2 rise in 2016 was successfully forecast to be the largest on record (Table  20.4), due to the amplifying effects of the 2015–2016 El Niño event (Betts et al., 2018). The rises in 2017 and 2018 were successfully forecast to be more moderate, due to neutral Niño‐3.4 SSTs in 2016– 2017 and La Niña conditions in 2017–2018. The rise in 2019 was successfully forecast to be a return to a larger annual increment (but not as large as 2015–2016) due to the predicted moderate El Niño–like conditions. An area for possible methodological improvement of the CO2 forecast concerns the assumption of a stationary seasonal cycle. Since the shape of the seasonal CO2 cycle varies, this is a potential shortcoming with the monthly forecasts. However, the amplitude of the seasonal cycle shows less variation, so the prediction of annual maxima and minima appears to be robust. 20.5.2. Carbon Cycle–ENSO Relationships in Earth System Models Since coupled ocean‐atmosphere general circulation models (GCMs) project ENSO events to become more frequent and intense (Cai et al., 2015; see chapter 13), it is important to understand the potential implications of this for the carbon cycle and hence as a feedback process on anthropogenic climate change. Now that the carbon cycle is now routinely included in GCMs as part of their evolution into ESMs (Earth system models), this pro­ vides an opportunity to examine the role of ENSO– carbon cycle interactions in long‐term climate change. The relationship between ENSO and interannual vari­ ability in the carbon cycle emerges in ESMs (Jones et al., 2001; Kim et al., 2016) and hence is evidence of realistic

behavior of such models on these timescales. The CMIP5 generation of ESMs simulate the observed ENSO– carbon cycle relationship reasonably well in terms of magnitude, timing, and the processes involved. Considering the onset and decline of the atmospheric CO2 growth rate anomaly within an El Niño event, the multimodel mean overestimates the maximum growth rate by approximately 20%, and the peak growth rate is correctly simulated to occur in the boreal spring but in around April rather than January as observed (Kim et al., 2016). The growth rate anomalies are caused mainly by variability in NPP, in agreement with the indications from observationally based studies, as described earlier in this chapter. In the CMIP5 ESMs, the sensitivity of the terrestrial carbon flux to ENSO is projected to increase by 44% (± 15%) in a scenario of CO2 concentrations stabilized at around 540 ppm, i.e. nearly double the preindustrial concentration (Kim et  al., 2017). The response of land temperature to ENSO and the response of GPP to land temperature are both projected to increase under the warmer climate, with the depletion of soil moisture increasing the temperature response to ENSO events. Cox et  al. (2013) and Wenzel et  al. (2014) suggested that the relationship between interannual variability in CO2 growth rate and tropical temperature could provide a constraint on a key component of carbon cycle feed­ backs on climate change (Figure 20.9), although clearly this depends on whether similar processes are involved on interannual and longer‐term timescales. In two genera­ tions of ESMs (C4MIP and CMIP5), the sensitivity of annual growth rate to temperature on annual timescales correlates strongly with the response of tropical land carbon storage to climate change (Figure  20.9a). Comparison with the observed CO2‐temperature rela­ tionship on annual timescales can therefore be used to estimate the probability distribution of tropical land carbon responses to climate change (Figure 20.9b). Without the ENSO constraint, there is a large spread in the projected loss of tropical land carbon (–49 ± 40 PgC per °C of tropical land warming; Figure  20.8b). The ENSO constraints narrow this to –44 ± 14 PgC °C–1.

Table 20.4  Summary of forecast and observed CO2 concentrations and rises for 2016, 2017, 2018, and 2019. Annual Mean CO2 Concentration (ppm)

Increase From Previous Year (ppm)

May Maximum (ppm)

September Minimum (ppm)

Year

Forecast

Observed

Forecast

Observed

Forecast

Observed

Forecast

Observed

2019 2018 2017 2016

411.30 ± 0.58 408.94 ± 0.59 406.75 ± 0.61 404.45 ± 0.53

411.49 408.59 406.59 404.28

2.74 ± 0.58 2.29 ± 0.59 2.46 ± 0.61 3.15 ± 0.53

2.90 2.00 2.31 3.39

414.70 ± 0.58 412.20 ± 0.59 409.86 ± 0.61 407.57 ± 0.53

414.83 411.31 409.91 407.65

408.10 ± 0.58 405.79 ± 0.59 403.72 ± 0.61 401.48 ± 0.53

408.55 405.59 403.27 401.01

ENSO and the Carbon Cycle  467 (a)

(b) 0

0.040

–20

γLT [PgC °C–1]

–40 –60 –80

LOOP FRCGC CCSM1 MPI IPSL HadCM3LC

–100 –120 –140 –15

–12

–9

–6

–3

0

3

γIAV [PgC °C–1 yr–1]

0.030 Probability density

NorESM1-ME MPI-ESM-LR MIROC-ESM IPSL-CM5A-LR HadGEM2-ES GFDL-ESM2M CESM1-BGC CanESM2

0.020

0.010

0.000 –150 –120 –90

–60

–30

0

30

γLT [PgC °C–1]

Figure 20.9  Use of interannual variability in CO2 growth rate and tropical temperature as a constraint on carbon cycle feedbacks on climate change. (a) Strength of feedback on climate change vs. sensitivity of annual growth rate to temperature on annual timescales in Earth system models from C4MIP (black symbols, best fit shown with dot‐dashed black line) and CMIP5 (colored stars, best fit shown with solid red line), with observational constraint on the interannual growth rate (vertical black dashed lines). (b) Probability distribution of feedback strengths from unconstrained ESMs (dashed curves) and after constraint by comparison with observed interannual variability (solid curves), for C4MIP (red) and CMIP5 (black) ESMs. (Reproduced from Wenzel et al., 2014)

20.6. SUMMARY AND CONCLUSIONS A wealth of observational studies of atmospheric CO2, land ecosystem, and ocean processes show that variability in the carbon cycle is closely related with ENSO. Years with a warm anomaly in the tropical Pacific show a faster CO2 rise due to weaker land carbon sinks, with a partial offset by stronger net uptake by oceans. The opposite happens in years with cool Pacific SST anomalies. This relationship holds for small ENSO SST anomalies as well as large ones and is robust enough for the annual CO2 growth rate anomaly to be highly predictable on the basis of SST observations and forecasts. Generally, variability in the land‐atmosphere carbon flux is mainly driven by physiological processes (photo­ synthesis and/or respiration responses to temperature and precipitation), with a smaller contribution from fire. Fire was important in the 1997–1998 El Niño, with equatorial Asian peatland fires making a major contribu­ tion to the CO2 rise, which can be viewed as anthropo­ genic in nature since the ignition was caused by humans. However, in the 2015–2016 El Niño event, the change in land carbon flux was mainly due to physiological processes, particularly reduced productivity, although this does not provide a complete explanation, so increased respiration may also have played a role. In El Niño years, net CO2 release from tropical land is the most immediate impact, followed by release from northern ecosystems a

few months later. The lag between peak SST anomaly and peak CO2 growth rate anomaly varies according to whether the SST anomaly is in the central or eastern Pacific. In the oceans, El Niño conditions involve decreased upwelling of carbon in the equatorial Pacific due to a weakening of the trade winds, causing this region to become a weaker sink of CO2 or near‐neutral if the El Niño event is strong. ENSO–carbon cycle relationships beyond individual years are also seen. In the late 2000s to early 2010, the annual increase in CO2 concentration remained approxi­ mately steady rather than generally rising, as it had done before and has done since. A statistical relationship with tropical Pacific SSTs reproduces this, with a dominance of La Niña conditions in the late 2000s to early 2010s, and global land carbon uptake anomalously high in that period, suggesting the relationship with ENSO acting via land ecosystem carbon sinks was responsible for this hiatus in the rise of the global CO2 growth rate. Earth system models also reproduce the ENSO–carbon cycle relationship well, and they have been proposed as a useful constraint on carbon cycle feedbacks in future cli­ mate change projections. ESMs project the carbon cycle to become more responsive to ENSO under future cli­ mate change, and with ENSO events projected to become more frequent and intense, this may lead to carbon cycle feedbacks accelerating anthropogenic climate change.

468  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

ACKNOWLEDGEMENTS RAB, CAB, CDJ and AJW were supported by the Newton Fund of the UK Department of Business, Energy and Industrial Strategy via the project Climate Science for Service Partnership Brazil (CSSP Brazil). RAF was supported by the NOAA Global Ocean Monitoring and Observing Program (GOMO), contribu­ tion number 5107 from the NOAA Pacific Marine Environmental Laboratory. REFERENCES Bacastow, R. B. (1976). Modulation of atmospheric carbon dioxide by the Southern Oscillation. Nature, 261, 116–118. Bacastow, R. B., Adams, J. A., Keeling, C. D., Moss, D. J., Whorf, T. P., & Wong, C. S. (1980). Atmospheric carbon dioxide, the Southern Oscillation, and the weak 1975 El Niño. Science, 210, 66–68. Bakker, D.C.E. (2016). A multi‐decade record of high‐quality fCO2 data in version 3 of the Surface Ocean CO2 Atlas (SOCAT), Earth Syst. Sci. Data, 8, 383–413, https://doi. org/10.5194/essd‐8‐383‐2016  Bastos, A., Friedlingstein, P., Sitch, S., Chen, C., Mialon, A., Wigneron, J‐P., et  al. (2018). Impact of the 2015/2016 El Niño on the terrestrial carbon cycle constrained by bottom‐ up and top‐down approaches. Phil. Trans. R. Soc. B, 373, 20170304. Battle, M., Fletcher, S.E.M., Bender, M. L., Keeling, R. F., Manning, A. C., Gruber, N., et  al. (2006). Atmospheric potential oxygen: New observations and their implications for some atmospheric and oceanic models. Global Biogeochemical Cycles, 20. Betts, R. A., Jones, C. D., Knight, J. R., Keeling, R. F., Kennedy, J. J. (2016). El Niño and a record CO2 rise. Nature Climate Change, 6, 806–810. doi:10.1038/nclimate3063 Betts, R. A, Jones, C. D., Knight, J. R., Keeling, R. F., Kennedy, J. J., Wiltshire, A. J., Andrew, R. M., Aragao LEOC. (2018). A successful prediction of the record CO2 rise associated with the 2015/2016 El Niño. Phil. Trans. R. Soc. B, 373, 20170301. Cai, W., Santoso, A., Wang, G., Yeh, S‐W., An, S‐I., Cobb, K. M., et  al. (2015). ENSO and greenhouse warming. Nature Climate Change, 5, 849–859. Chatterjee, A., Gierach, M. M., Sutton, A. J., Feely, R. A., Crisp, D., Eldering, A. et al., (2017). Influence of El Niño on atmospheric CO2 over the tropical Pacific Ocean: Findings from NASA’s OCO‐2 mission. Science, 358(6360), eaam5776. doi: 10.1126/science.aam5776 Chylek, P., Tans, P., Christy, J., & Dubey, M. K. (2018). The carbon cycle response to two El Nino types: An observational study. Environ. Res. Lett., 13, 024001. Cox, P. M., Betts, R. A., Jones, C. D., Spall, S. A., & Totterdell, I. J. (2000). Acceleration of global warming due to carbon‐cycle feedbacks in a coupled climate model. Nature, 408, 184–187. Cox, P. M., Betts, R. A., Collins, M., Harris, P. P., Huntingford, C., & Jones, C. D. (2004). Amazonian forest dieback under

c­ limate‐carbon cycle projections for the 21st century. Theor Appl Climatol, 78, 137–156. https://doi.org/10.1007/s00704‐ 004‐0049‐4 Cox, P. M., Pearson, D., Booth, B. B., Friedlingstein, P., Huntingford, C., Jones, C. D., Luke, C. M. (2013). Sensitivity of tropical carbon to climate change constrained by carbon dioxide variability. Nature, 494, 341–344. England, M. H., McGregor, S., Spence, P., Meehl, G. A., Timmermann, A., Cai, W., et al. (2014). Recent intensifica­ tion of wind‐driven circulation in the Pacific and the ongoing warming hiatus, Nature Climate Change, 4, 222‐227. doi: 10.1038/nclimate2106 Feely, R. A., Wanninkhof, R., Takahashi, T., & Tans, P. (1999). Influence of El Niño on the equatorial Pacific contribution of atmospheric CO2 accumulation. Nature, 398, 597–601. doi: 10.1038/19273 Feely, R. A., Boutin, J., Cosca, C. E., Dandonneau, Y., Etcheto, J., Inoue, H. Y., et al. (2002). Seasonal and interannual variability of CO2 in the equatorial Pacific. Deep‐Sea Res. Pt. II, 49(13– 14), 2443–2469. doi: 10.1016/S0967‐0645(02)00044‐9 Feely, R. A., Takahashi, T., Wanninkhof, R., McPhaden, M. J., Cosca, C. E., Sutherland, S. C., & Carr, M.‐E. (2006). Decadal variability of the air‐sea CO2 fluxes in the equatorial Pacific Ocean. J. Geophys. Res., 111(C08), C08S90. doi: 10.1029/ 2005JC003129 Feely, R. A., Wanninkhof, R., Carter, B. R., Landschützer, P., Sutton, A. J., Cosca, C., & Triñanes, J. A. (2019). Global ocean carbon cycle. In State of the Climate in 2018, Global Oceans. Bull. Am. Meteorol. Soc., 100(9), S94–S99. doi: 10.11 75/2019BAMSStateoftheClimate.1 Giglio, L., Randerson, J. T., & van der Werf, G. R. (2013). Analysis of daily, monthly, and annual burned area using the fourth‐generation global fire emissions database (GFED4). Journal of Geophysical Research: Biogeosciences, 118, 317– 328. doi:10.1002/jgrg.20042, 2013 Gloor, E., Wilson, C., Chipperfield, M. P., Chevallier, F., Buermann, W., Boesch, H., et  al., (2018). Tropical land carbon cycle responses to 2015/16 El Niño as recorded by atmospheric greenhouse gas and remote sensing data. Phil. Trans. R. Soc. B, 373, 20170302. Houghton, R. A., & Nassikas, A. A. (2017). Global and regional fluxes of carbon from land use and land cover change 1850– 2015. Global Biogeochem. Cycles, 31, 456–472. doi:10.1002/ 2016GB005546 Humphrey, V., Zscheischler, J., Ciais, P., Gudmundsson, L., Sitch, S., & Seneviratne, S. I. (2018). Sensitivity of atmo­ spheric CO2 growth rate to observed changes in terrestrial water storage. Nature, 560(7720): 628–631. Ishii, M., Inoue, H. Y., Midorikawa, T., Saito, S., Tokieda T., Sasano, D., et al. (2009). Spatial variability and decadal trend of the oceanic CO2 in the western equatorial Pacific warm/ fresh water. Deep‐Sea Res. II, 56(8–10), 591–606. doi: 10.1016/j.dsr2.2009.01.002 Ishii, M., Feely, R. A., Rodgers, K. B., Park, G.‐H, Wanninkhof, R., Sasano, D. et al. (2014). Air‐sea CO2 flux in the Pacific Ocean for the period 1990‐2009. Biogeosciences, 11, 709–734. doi: 10.5194/bg‐11‐709‐2014

ENSO and the Carbon Cycle  469 Jones, C. D., & Cox, P. M. (2005). On the significance of atmo­ spheric CO2 growth rate anomalies in 2002–2003, Geophys. Res. Lett., 32, L14816. doi:10.1029/2005GL023027 Jones, C. D., Collins, M., Cox, P. M., & Spall, S. A. (2001). The carbon cycle response to ENSO: A coupled climate‐carbon cycle model study, J. Clim., 14, 4113–4129. Keeling, R. F., & Manning, A. C. (2014). Studies of recent changes in atmospheric O2 content. In Treatise on geochemistry: Reference module in Earth systems and environmental sciences (Vol. 5, pp. 385‐404). Amsterdam: Elsevier. Keeling, C. D., & Revelle, R. (1985). Effects of El‐Niño southern oscillation on the atmospheric content of carbon‐dioxide. Meteoritics, 20, 437–450. Keeling, C. D., Piper, S. C., Bacastow, R. B., Wahlen, M., Whorf, T. P., Heimann, M. & Meijer, H. A. (2001). Exchanges of atmospheric CO2 and 13CO2 with the terrestrial biosphere and oceans from 1978 to 2000: Part I. Global aspects. SIO Reference Series, No. 01‐06, San Diego: Scripps Institution of Oceanography, 88 pp. Keenan, T. F., Prentice, I. C., Canadell, J. G., Williams, C. A., Wang, H., Raupach, M., Collatz, G. J. (2016). Recent pause in the growth rate of atmospheric CO2 due to enhanced ter­ restrial carbon uptake. Nature Communications, 7, 13428. doi:10.1038/ncomms13428 Kennedy, J. J., Rayner, N. A., Smith, R. O., Saunby, M., & Parker, D. E. (2011a). Reassessing biases and other uncer­ tainties in sea‐surface temperature observations since 1850: Part 1. Measurement and sampling errors. J. Geophys. Res., 116, D14103. doi:10.1029/2010JD015218 Kennedy, J. J., Rayner, N. A., Smith, R. O., Saunby, M., & Parker, D. E. (2011b). Reassessing biases and other uncer­ tainties in sea‐surface temperature observations since 1850: Part 2. Biases and homogenisation. J. Geophys. Res., 116, D14104. doi:10.1029/2010JD015220 Kim, J. S., Kug, J. S., Yoon, J.‐H., & Jeong, S.‐J. (2016). Increased atmospheric CO2 growth rate during El Niño driven by reduced terrestrial productivity in the CMIP5 ESMs. Journal of Climate, 29, 8783–8805. Kim, J‐S., Kug, J‐S., & Jeong, S‐J. (2017). Intensification of ter­ restrial carbon cycle related to El Niño–Southern Oscillation under greenhouse warming. Nature Communications, 8, Art. No. 1674. Koren, G., van Schaik, E., Araújo, A. C., Boersma, K. F., Gärtner, A., Killaars, L., et al. (2018). Widespread reduction in sun‐induced fluorescence from the Amazon during the 2015/2016 El Niño. Phil. Trans. R. Soc. B, 373, 20170408. http://doi.org/10.1098/rstb.2017.0408 Kosaka, Y., Xie, S.‐P. (2013). Recent global‐warming hiatus tied to equatorial Pacific surface cooling. Nature, 501, 403–407. Kriegler, E., Hall, J. W., Held, H., Dawson, R., & Schellnhuber, H.‐J. (2009). Imprecise probability assessment of tipping points in the climate system. PNAS, 106(13), 5041–5046. https://doi.org/10.1073/pnas.0809117106 Landschützer, P., Gruber, N., Bakker, D.C.E., & Schuster, U. (2014) Recent variability of the global ocean carbon sink. Global Biogeochemical Cycles, 28, 927–949. doi: 10.1002/ 2014gb004853

Landschützer, P., Gruber, N., & Bakker, D.C.E. (2016) Decadal variations and trends of the global ocean carbon sink. Global Biogeochem. Cycles, 30, 1396–1417. doi:10.1002/2015 GB0 05359 Le Quéré, C., Andrew, R. M., Friedlingstein, P., Sitch, S., Hauck, J., Pongratz, J., et al. (2018). Global carbon budget 2018. Earth System Science Data, 10, 1–54. doi: 10.5194/ essd‐10‐2141‐2018. Luo, X., Keenan, T. F., Fisher, J. B., Jiménez‐Muñoz, J.‐C., Chen, J. M., Jiang, C., et  al. (2018). The impact of the 2015/2016 El Niño on global photosynthesis using satellite remote sensing. Phil. Trans. R. Soc. B, 373, 20170409. http:// doi.org/10.1098/rstb.2017.0409 MacLachlan, C., Arribas, A., Peterson, K. A., Maidens, A., Fereday, D., Scaife, A. A., et  al. (2015). Description of GloSea5: The Met Office high resolution seasonal forecast system. Q. J. R. Met. Soc. doi: 10.1002/qj.2396 Malhi, Y., Rowland, L., Aragão, L.E.O.C, & Fisher, R. A. (2018). New insights into the variability of the tropical land carbon cycle from the El Niño of 2015/2016. Phil. Trans. R. Soc. B, 373, 20170298. http://dx.doi.org/10.1098/rstb.2017. 0298 McNeil, B. I., Matear, R. J., Key, R. M., Bullister, J. L., & Sarmiento, J. L. (2003). Anthropogenic CO2 uptake by the ocean based on the global chlorofluorocarbon data set. Science, 299, 235–239. https://doi.org/10.1126/science. 1077429 Meehl, G. A., & Washington, W. M. (1996) El Niño‐like climate change in a model with increased atmospheric CO2 concen­ trations. Nature, 382, 56–60. Meehl, G. A., Arblaster, J. M., Fasullo, J. M., Hu, A. & Trenberth, K. E. (2011). Model‐based evidence of deep‐ ocean heat uptake during surface‐temperature hiatus periods. Nature Climate Change. doi:10.1038/nclimate1229 Mikaloff Fletcher, S. E., Gruber, N., Jacobson, A. R., Doney, S. C., Dutkiewicz, S., & Gerber, M. (2006). Inverse estimates of anthropogenic CO2 uptake, transport, and storage by the oceans. Global Biogeochem. Cy., 20, GB2002. https://doi. org/10.1029/2005GB002530 Nechita‐Banda, N., Krol, M., van der Werf, G. R., Kaiser, J. W., Pandey, S., Huijnen, V., et al. (2018). Monitoring emissions from the 2015 Indonesian fires using CO satellite data. Phil. Trans. R. Soc. B, 373, 20170307. http://doi.org/10.1098/rstb. 2017.0307 Page, S. E. Siegert, F., Rieley, J. O., Boehm, H‐D.V., Jayak, A. & Limink, S. (2002). The amount of carbon released from peat and forest fires in Indonesia during 1997. Nature, 420, 61–65. Rayner, P. J., & Law, R. M. (1999). The interannual variability of the global carbon cycle. Tellus, 51B, 210–212. doi:https:// doi.org/10.1034/j.1600‐0889.1999.t01‐1‐00007.x Rayner, P. J., Law, R. M., & Dargaville, R. (1999). The relation­ ship between tropical CO2 fluxes and the El Niño–Southern Oscillation. Geophys. Res. Lett., 26, 493–496. https://doi. org/10.1029/1999GL900008 Rifai, S. W., Girardin, C.A.J., Berenguer, E., del Aguila‐Pasquel, J., Dahlsjö, C.A.L., Doughty, C. E., et  al. (2018). ENSO

470  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Drives interannual variation of forest woody growth across the tropics. Phil. Trans. R. Soc. B, 373, 20170410. Rödenbeck, C., Zaehle, S., Keeling, R., & Heimann, M. (2018) History of El Niño impacts on the global carbon cycle 1957– 2017: A quantification from atmospheric CO2 data. Phil. Trans. R. Soc. B, 373, 20170303. Santoso, A., Mcphaden, M. J., & Cai, W. (2017). The defining characteristics of ENSO extremes and the strong 2015/2016 El Niño. Reviews of Geophysics, 55, 1079–1129. https://doi. org/10.1002/2017RG000560 Takahashi, T., Sutherland, S. C., Wanninkhof, R., Sweeney, C., Feely, R. A., Chipman, D. W., et  al. (2009). Climatological mean and decadal change in surface ocean pCO2, and net sea‐air CO2 flux over the global oceans. Deep‐Sea Res. II, 56(8–10), 554–577. doi: 10.1016/j.dsr2.2008.12.009 Wang, W., Ciais, P., Nemani, R. R., Canadell, J. P., Piao, S., Sitch, S., et al. (2013). Variations in atmospheric CO2 growth rates coupled with tropical temperature. Proc. Natl Acad. Sci. USA 110, 13061–13066.

Wang, J., Zeng, N., Wang, M., Jiang, F., Wang, H., & Jiang, Z. (2018) Contrasting terrestrial carbon cycle responses to the 1997/98 and 2015/16 extreme El Niño events. Earth System Dynamics, 9(1), 1–14. https://doi.org/10.5194/esd‐ 9‐1‐2018 Wanninkhof, R., Park, G.‐H., Takahashi, T., Sweeney, C., Feely, R., Nojiri, Y., et  al. (2013). Global ocean carbon uptake: Magnitude, variability, and trends. Biogeosciences, 10, 1983–2000. doi: 10.5194/bg‐10‐1983‐2013 Wenzel, S., Cox, P. M., Eyring, V., & Friedlingstein, P. (2014). Emergent constraints on climate‐carbon cycle feedbacks in the CMIP5 Earth system models. J. Geophys. Res. Biogeosci., 119, 794–807. doi:10.1002/2013JG002591 Withey, K., Berenguer, E., Palmeira, A. F., Espírito‐Santo, F.D.B., Lennox, G. D., Silva, C.V. J., et al. (2018). Quantifying immediate carbon emissions from El Niño‐mediated wildfires in humid tropical forests. Philosophical Transactions of the Royal Society B: Biological Sciences, 373. http://doi. org/10.1098/rstb.2017.0312

Section VII Closing

21 ENSO in a Changing Climate: Challenges, Paleo-Perspectives, and Outlook Christina Karamperidou1, Malte F. Stuecker2,3,4,5, Axel Timmermann4,5, Kyung‐Sook Yun4,5, Sun‐Seon Lee4,5, Fei‐Fei Jin1, Agus Santoso6,7, Michael J. McPhaden8, and Wenju Cai6,9

ABSTRACT The El Niño Southern Oscillation (ENSO) phenomenon is a dominant force driving year‐to‐year climate variability with ecological and socioeconomic impacts that reverberate around the globe. The complex processes that govern ENSO and its impacts have generated intense research over the past decades, reviewed in previous chapters: a better understanding of how ENSO responds to anthropogenic climate change requires effort in resolving how ENSO responds to and interacts with a multitude of factors such as weather‐scale phenomena, volcanic eruptions, orbital forcing, etc. This chapter highlights some key unresolved issues in ENSO, supplemented by analysis of paleoclimate data and past and future state‐of‐the‐art climate model simulations. First, paleo‐ENSO reconstructions indicate a weakening of ENSO variability accompanying a weaker seasonal cycle, albeit lacking a clear orbital signal. This apparent positive correlation between changes in the magnitude of the seasonal cycle and ENSO amplitude seems to hold in future greenhouse‐gas forcing scenarios. Yet the mechanisms behind this relationship remain unclear, as accelerated paleoclimate model simulations under orbital forcing show the opposite relationship, in accordance with the idea of frequency entrainment in nonlinear oscillatory systems. These results underscore another prominent unresolved question: is ENSO a nonlinear system exhibiting regime‐like behavior (internally generated or in response to external forcing), or is ENSO a stochastically forced linear system whose behavior is modulated by noise? The community’s efforts to answer this question face the limitations imposed by the short instrumental record. Paleoclimate reconstructions provide extensions of this record although there persist sampling issues and uncertainties surrounding the manifestation of the ENSO signal in hydroclimate records which are influenced by local and regional processes. In its conclusion, this chapter highlights recent research directions and underscores the need for sustained and improved observations, paleo‐proxy reconstructions, hierarchical climate modeling, theories, and collaboration across disciplines toward addressing the open ENSO questions. 21.1. INTRODUCTION Tropical Pacific interannual variability has been a property of Earth’s climate system for at least several million 1    Department of Atmospheric Sciences, University of Hawai‘i at Mˉanoa, Honolulu, HI, USA 2   Department of Oceanography, University of Hawai‘i at Mˉanoa, Honolulu, HI, USA 3   International Pacific Research Center, University of Hawai‘i at Mˉanoa, Honolulu, HI, USA 4   Center for Climate Physics, Institute for Basic Science, Busan, Republic of Korea

years (Watanabe et  al., 2011). The interactions between ENSO and a plethora of processes in the atmosphere,   Pusan National University, Busan, Republic of Korea   Centre for Southern Hemisphere Oceans Research, CSIRO Oceans and Atmosphere, Hobart, TAS, Australia 7   ARC Centre of Excellence for Climate Extremes, Climate Change Research Centre, University of New South Wales, Sydney, NSW, Australia 8   NOAA/Pacific Marine Environmental Laboratory, Seattle, WA, USA 9   Key Laboratory of Physical Oceanography/Institute for Advanced Ocean Studies, Ocean University of China and Qingdao National Laboratory for Marine Science and Technology, Qingdao, China 5 6

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 473

474  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

hydrosphere, and biosphere are fascinating to study (section I), while its impacts on human societies cannot be overstated, thus motivating in the recent decades an expansion of research on the past and future of the phenomenon and its fundamental properties (section III). In assessing ENSO’s sensitivity to external radiative forcing, much research has focused on identifying forced changes in ENSO properties related to changes in Earth’s orbit, volcanic eruptions, and greenhouse gas forcing (section V), as captured in paleoclimate proxy records, instrumental data, and model simulations (section II). Understanding ENSO’s sensitivity to external forcing ultimately improves modeling and forecasting of ENSO (section IV) and enables the mitigation of risks associated with ENSO events and their socioeconomic impacts (section VI). The preceding chapters of this book aimed at presenting a comprehensive review of historical and recent advancements in ENSO research. The purpose of this final chapter is to identify and highlight some key issues in ENSO research that have not been fully addressed in previous chapters and that may require further scientific attention in future studies. 21.2. SEASONAL CYCLE–ENSO INTERACTIONS ENSO can be viewed as an instability of the seasonally varying background state (Jin et al., 1994). Its seasonal phase‐locking (i.e. the tendency for ENSO to peak in boreal winter; section I) is testimony to the fact that ENSO and the seasonal cycle of temperature, winds, and ocean currents in the equatorial Pacific are closely coupled (Wang, 1994; Xie, 1995; Stein et al., 2011; An et al., 2010; Stein et al., 2014). As a result, we would expect that a change in the amplitude or phase of one will likely change characteristics of the other. For example, changes in Earth’s orbital characteristics during the Holocene modulate the seasonal cycle of the tropical Pacific, which in turn may impact ENSO properties, via changes in the background state and the feedbacks that determine the growth and decay of ENSO events. Recent paleoclimate reconstructions of ENSO variability spanning the last 7,000 years (Carré et  al., 2014; Grothe et  al., 2020) find evidence for a large, sustained reduction in ENSO variability 3,000–5,000 years ago, relative to adjacent intervals (Figure 21.1a). Paleo‐ENSO reconstructions have motivated a suite of modeling studies over the years (see chapter 5) which are based on models of various complexity (chapter 9), ranging from more simplified intermediate models of the tropical Pacific, like the Cane‐ Zebiak Model (e.g. Clement et al., 2000), to comprehensive global climate models exposed to Holocene orbital conditions. The majority of fifth phase Climate Model Intercomparison Project (CMIP5) coupled general circulation models (CGCMs) simulates slightly weaker eastern tropical Pacific

SST variability during the mid‐Holocene (~6,000 years ago), both on annual and interannual timescales (Masson‐ Delmotte et al., 2013). In addition, analysis of recent climate model simulations suggests the possibility for a shift toward more central Pacific (CP) ENSO events during the mid‐Holocene owing to seasonal changes in ocean currents and stratification (Karamperidou et al., 2015). Considering the differential impacts of eastern Pacific (EP) and CP ENSO events on hydroclimate in regions rich with paleoclimate proxies, the interpretation of these proxy records should take into consideration potential changes in ENSO diversity in conjunction with changes in the seasonal cycle (Karamperidou et al., 2015; Kiefer & Karamperidou, 2019). Orbital forcing during the mid‐Holocene presents mostly a meridionally asymmetric change in the amplitude of the seasonal cycle. Any change in ENSO character has to therefore emerge from an interaction between annual and ENSO timescales (Timmermann et  al., 2007a) through mechanisms such as frequency entrainment (Chang et  al., 1994; Liu, 2002). Frequency entrainment occurs in nonlinear oscillatory systems, subject to strong external (periodic) forcing, when the internal frequency of the system gets “entrained” into the forcing frequency. The entrainment often follows a pathway in which the forced system adopts integer periods of the forcing period. With increasing forcing, the internal dynamics will jump between mode‐locked solutions (multiples of the forcing frequency) and quasi‐ periodic and sometimes chaotic states. One of the key predictions of this scenario, which has been invoked to explain ENSO’s response, both to orbital and millennial‐ scale forcing (Timmermann et al., 2007b; Liu et al., 2014), is the presence of mode‐locked parameter regimes. To test if this scenario really holds, we conducted an 11‐ member ensemble of orbitally forced simulations using the Community Earth System Model (version 1.2, T31 atmosphere resolution and 3x3 degree ocean horizontal resolution; Hurrell et al., 2013), covering the past 408,000 years. Orbital forcing was applied with an acceleration factor of 250, which squeezes one precessional cycle (period ~21,000 years) into 84 years. This is enough time for upper ocean thermodynamics and dynamics to adjust to the fastest external forcing. Running an 11-member ensemble of these orbitally forced simulations further minimizes uncertainties in the adjustment, and removes effects of unforced variability. Figure 21.2 shows the changes in simulated ENSO and seasonal cycle variance as a function of the longitude of perihelion, which indicates the timing of Earth’s closest approach to the sun with respect to the Northern Hemisphere fall equinox (see figure caption for details). For strong seasonal cycle amplitude (upper quadrants), we find generally weaker ENSO variability. In the upper left quadrant, we find evidence for variability peaking at periods close to 3 years. This feature could be indicative of ­frequency

Change in SD of ENSO proxy (%)

Proxy-inferred variance of ENSO (wrt 1971-2000CE) 60

(b) Pseudocoral

(a) Proxy (coral)

SST+SSS

30

SST

0 –30 –60 –90 7000

6000

5000

4000

3000

2000

1000

Change in SD of ENSO proxy (%)

Year (BP) 60

0 (1950CE)

(c) Proxy reconstructions

2000

2040 2080 Year (CE)

(d) Rainfall

30 0 –30 –60 –90 1400

1500

1600

1700

1800

(e) Proxy location

2040 2080

NADA

MADA

20N

2000 2000

Year (CE)

Year (CE) 40N

1900

Liu et al. (2017): 1400-2000 Liu et al. (2013): 1400-2000 McGregor et al. (2013): 1400-2000 Palmyra: 1886-2006 Kiritimati: 1938-2016 Fanning: 1949-2005 Pseudoproxy (CMIP5): 1900-2100

Central Pacific

0 20S 40S 60E

120E

180

120W

60W

Figure 21.1  Past and projected changes in ENSO (or interannual) variability at ENSO‐sensitive sites. (a) Changes in interannual variability (plotted as the standard deviation of 1.5–10 yr bandpassed data) recorded in coral oxygen isotopic (δ18O) data (Cobb et al., 2013; Grothe et al., 2020) from the central tropical Pacific, relative to the 1977–2007 reference period (dashed line). Data from Kiritimati (green), Fanning (blue), and Palmyra (orange), including 20th‐ century records (stars) and fossil corals (circles, where open circles represent sequences shorter than 30 years in length). Vertical bars represent the full range of interannual variability captured in the sequences >30 years. Grey shading indicates the quartile range of late 20th‐century interannual variability from CMIP5 historical runs. (b) Changes in interannual variability (s.d. of 1.5–10 yr bandpass, averaged across 30‐yr sliding windows) of SST (thick blue) and pseudo‐coral δ18O time series (thick red) constructed from SST and SSS output averaged across the red box plotted in panel (e) from 29 models in the CMIP5 archive, using the equation δ18Ocoral = –0.21 SSTA + 0.27 SSSA, plotted with quartile spread as thin lines of corresponding color, relative to the 1971–2000 CE. SSSA corresponds to the simulated sea surface salinity anomalies. (c) Changes in interannual variability (s.d. of 1.5–10 year bandpass, averaged across 30‐year sliding windows) of 16 paleo‐ENSO proxy records, relative to the 1977–2007 period, plotted with quartile spread (McGregor et al., 2013; Li et al., 2013; Liu et al., 2017). (d) Same as (b), but for simulated changes in precipitation from 16 individual model gridpoints where precipitation is significantly correlated to ENSO, selected at random, with 10 terrestrial gridpoints and 6 oceanic gridpoints, similar to the marine/terrestrial ratio of the proxy datasets shown in (c). (e) Map of the locations of ENSO‐sensitive proxy records shown in panels (a) through (d).

476  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE 10

7 6

Period (years)

5

5

0 ka 0

11 ka

1

2

3

4 3

4

–5

5

6

2

7

1

–10 –10

6 ka –5

0

5

10

0

Period (years)

Figure 21.2  ENSO/seasonal cycle interaction in an orbitally forced climate model. The ENSO/seasonal cycle phase locking is quantified by Morlet wavelet spectral power (unitless) of normalized Niño‐3.4 SST simulated by an 11‐member ensemble of orbitally forced (and accelerated) Community Earth System Model simulations. Instead of showing the time evolution of the ensemble mean wavelet power of the simulated Niño‐3.4 SST, we show the corresponding phase wheel, where the phase (angle) through the center point represents the longitude of perihelion λ (sin(λ) corresponds to the precessional parameter): λ = 0° when the perihelion occurs at NH fall equinox, λ = 90° when the perihelion occurs at NH winter solstice, λ = 180° when the perihelion occurs at NH vernal equinox, and λ = 270° when perihelion occurs at NH summer solstice. In present day, perihelion is in January and its longitude is ~103° . The angles of the dashed red lines correspond to the longitude of perihelion of the early Holocene (11 ka; ~279°), the mid Holocene (6 ka, ~0.87°), and present‐day conditions (~102.9°). One revolution on this phase wheel takes about 21,000 years. Forward time‐direction corresponds to an anticlockwise movement on the phase wheel. The distance from the center point characterizes the period of the wavelet‐transformed Niño‐3.4 SST signal, and white‐labeled dashed circles correspond to the integer periods of 1 year (i.e. 1, 2, 3, ... years). The solid blue line shows an unscaled estimate of the overall wavelet amplitude (distance from center) in the frequency band of 2–8 years as a function of the longitude of perihelion.

locking between ENSO and the seasonal cycle. For weaker seasonal cycle amplitude in SST (i.e. from 8 to 3 ka), we see an enhancement of simulated interannual ENSO variability in the period range of 2–6 years, but no clear sign for power peaking at multiples of the seasonal cycle. This analysis provides more evidence for the fact that even in a coupled general circulation model, ENSO frequency entrainment to the seasonal cycle is active. However, this

specific scenario of orbital amplitude and frequency control appears to be at odds with the observational findings (Figure  21.1a), which suggest suppressed ENSO variance from 6 to 1 ka. To further reconcile paleoclimate evidence with climate model simulations, we need to obtain a better understanding of the large uncertainties in simulated ENSO variance changes during the mid‐Holocene, which range from 35% suppression to 20% enhancement (Masson‐ Delmotte et al., 2013). Moreover, even though ENSO variance may have been reduced during preindustrial times (Figure  21.1a), a clear orbital signal is still not apparent, and much of the secular variability in ENSO amplitude may also be related to processes internal to the climate system. Such processes are clearly manifest in ENSO reconstructions covering the past few hundred years, that show an ENSO system that can vary in amplitude by up to 30% (Figure 21.1c) without apparent external drivers. The yet‐to‐be‐resolved ENSO–seasonal cycle interactions discussed above further imply that biases in the simulation of the seasonal cycle in climate models may also impact the representation of ENSO and its diversity. Indeed, much like there is disagreement among coupled climate models on projected changes in ENSO variance and diversity in response to greenhouse gas forcing (Taschetto et al., 2014; Karamperidou et al., 2017), there is also disagreement on projected changes in the magnitude of the seasonal cycle in the eastern Pacific (Figure 21.3). Nearly half of the CMIP5 models analyzed here (19 of 34) show a decrease in seasonal cycle magnitude in a high emissions scenario (RCP8.5), while the rest project an increase. Interestingly, there is an apparent and overlooked positive relationship between ENSO variance changes and seasonal cycle change across the CMIP5 models, with models that project an increase in the magnitude of the seasonal cycle also simulating an increase in ENSO variance. This relationship could be partially reflecting the impact of seasonal cycle changes on ENSO as well as the modulation of the seasonal cycle by ENSO (e.g. as discussed in Stein et al., 2011). These results seem consistent with the majority model behavior in mid‐ Holocene experiments but are at odds with the suggestion of the accelerated paleoclimate simulations presented in Figure 21.2. Hence, a systematic diagnosis of these interactions in climate models is needed in order to elucidate the mechanisms behind the apparent relationship shown in Figure 21.3 and whether the future sensitivity is distinct from the one that controlled ENSO’s amplitude in response to orbital forcing (as presented in Figure 21.2). This understanding would be highly beneficial to determine whether paleo‐ENSO records indeed provide a constraint on projections of future ENSO variability. It should be mentioned here that the seasonal cycle in the eastern equatorial Pacific also emerges from coupled atmosphere/ocean dynamics (Mitchell & Wallace, 1991;

ENSO IN A CHANGING CLIMATE  477 3.0

32

ENSO variance change [RCP8.5:piControl]

cor = 0.51

15

2.5

31 33

2.0

9

34 4 13

16

28

1.5

11

21

12 2 26 6

4 14

1.0

5

36 3 6 1 17

7

19 2

29 4 22 130 335

8

18 20

25 10 27

0.8 1.0 1.2 1.4 seasonal cycle magnitude change [RCP8.5:piControl]

Figure 21.3  Seasonal cycle and ENSO variance changes in future CMIP5 simulations. Here the seasonal cycle is defined as the range of climatological SST in the Niño‐3 region, while ENSO variance is defined as the variance of the SST anomalies in the Niño‐3.4 region. The change is defined as the ratio of the high emissions scenario (RCP8.5) value over the preindustrial control value (piControl). Numbers indicate CMIP5 models as ordered in Karamperidou et al. (2017). The shaded area indicates the 95% confidence intervals for the linear regression fit. The clear positive relationship, which is consistent with the changes described in Timmermann et al. (2004), is indicative of the fact that coupled processes control both the magnitude of seasonal cycle and ENSO variability.

Wang, 1993; Xie, 1995). Chen and Jin (2017, 2018) developed a coupled dynamic framework that is similar to the Bjerknes stability index for ENSO (see chapter 6) and uses simplified linear approximations to analyze the feedback processes associated with the seasonal cycle of the eastern tropical Pacific. In this work, they showed that the main contributor to seasonal cycle biases in CMIP5 models is the shortwave thermodynamic feedback, much like it is for the simulation of ENSO nonlinearities and centennial‐scale warming of the cold tongue, as discussed, for instance, in Bellenger et al. (2014) and Karamperidou et al. (2017). And as with their simulation of ENSO, models may appear to adequately simulate the amplitude of the seasonal cycle as a result of a compensation of biases among its various contributing coupled processes; thus, models may simulate the “right” amplitude of the seasonal cycle as well as the “right” amplitude of ENSO for the “wrong” reasons. Furthermore, it has been shown that common mean state biases in climate models are responsible for too‐weak pan‐tropical interbasin connectivity and thus an underestimation of internal variability on decadal timescales in many models (McGregor et al., 2018; Cai et al., 2019). It is unclear at this point if this underestimation of decadal variability might further influence simulated ENSO characteristics and its phase‐locking with the seasonal cycle.

It follows that to fully understand how ENSO might change in response to external forcings, it is necessary to understand the mechanisms behind changes in the seasonal cycle of the equatorial Pacific and its two‐way interactions with ENSO. Given the fact that both the seasonal cycle and ENSO explain a similar level of SST variability in the eastern equatorial Pacific, it is crucial to simulate both of them realistically in state‐of‐the‐art coupled general circulation models. Moreover, nonlinear interactions between ENSO and the seasonal cycle also govern its global teleconnections, for instance, to the East Asian monsoon (Stuecker et al., 2015a), and generate deterministic variability on a wide range of timescales (Stuecker et al., 2015b). With this perspective in mind, it is surprising to see that only a few studies (Timmermann et al., 2004) have investigated the sensitivity of the seasonal cycle in the eastern equatorial Pacific to greenhouse warming. 21.3. FORCED ENSO CHANGES VS. INTERNAL VARIABILITY, AND THE POTENTIAL FOR INCREASING CONFIDENCE IN ENSO PROJECTIONS The projected magnitude and pattern of changes in tropical or extratropical annual mean climate, seasonal cycle, and ENSO carry uncertainties associated with

478  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

three factors (Hawkins & Sutton, 2009): (i) the magnitude of internal climate variability; (ii) model biases and intermodel disagreements due to differences in their dynamical assumptions, numerical frameworks, and parameterizations of subgrid processes that are either not resolved by models or poorly understood; and (iii) the anthropogenic forcing scenario that the global community will eventually follow (including greenhouse gas emissions, aerosol forcing, and land‐use changes). While the latter factor is a function of complex socioeconomic conditions, assessing and potentially reducing uncertainties in ENSO projections associated with internal variability and model disagreement may be within reach of the ENSO scientific community. To date, the relative extent to which millennial or decadal changes in ENSO can be attributed to external forcing or arise from internal dynamics remains elusive. Indeed, rich internal ENSO variability at multiple timescales has been shown to exist in multicentennial CGCM simulations in the absence of any external influences, such as volcanic, orbital, greenhouse gas forcing and land‐use changes (Wittenberg, 2009; Deser et al., 2012; chapter 8). Whether such internal variability results from deterministic (weakly chaotic) processes (Munnich et al., 1991; Jin et al., 1994; Tziperman et al., 1994; Timmermann & Jin, 2002; Ghil et al., 2008) or variations in stochastic forcing (Vecchi et  al., 2006; Gebbie et  al., 2007; Zavala‐Garay et al., 2008; Kleeman, 2008; Newman, 2007) or both also remains an open question. An answer to this question is closely linked to our ability to address the issues highlighted in section 21.2 above: e.g. if ENSO exhibits regime‐ like behavior stemming from a nonlinear nature, then its response in past or future climates may result from a combination of nonlinear behavior (e.g. frequency entrainment) and linear responses to external factors (e.g. via changes in the seasonal cycle). Of course, the magnitude of internal ENSO variability, whatever its mechanisms may be, does not preclude sensitivity of ENSO to external forcing: the potential impacts of orbital forcing were discussed in section  21.2 of this chapter and in chapter 5. The response of ENSO to volcanic forcing was reviewed in chapter  12; proxy reconstructions and most CGCM simulations suggest that strong tropical volcanic forcing can increase the probability of El Niño events in the year following an eruption (Adams et  al., 2003; McGregor et  al., 2010). Given the small magnitude of solar irradiance changes, a discernible ENSO response to solar forcing has not been established, either in observational records or in CGCM simulations (Le, 2017). The response of ENSO to greenhouse gas forcing remains a matter of active research, particularly due to biases in climate models, although segregating models based on certain physical considerations may lead to stronger intermodel consensus (see chapter  13), e.g. in terms of changes in eastern Pacific ENSO variance

(Karamperidou et  al., 2017; Cai et  al., 2018). Paleo‐ reconstructions also offer some clues, although uncertainties related with proxy reconstructions need to be borne in mind (chapter 5). Multiple paleo‐ENSO proxy data sets point to an intensification of ENSO variability during the late 20th century relative to the preindustrial period (McGregor et al., 2013; Cobb et al., 2013; Li et al., 2013; Liu et  al., 2017; Figure  21.1a, c). These proxy records reflect ENSO‐related temperature and hydrological influences, such that the observed increase in interannual variance in these records likely reflects an increase in ENSO‐related temperature and/or hydrological variability. Interestingly, a CMIP5‐based δ18O pseudo coral proxy network shows an increase in interannual variance that seems to continue the reconstructed trend from the real coral data (Figure 21.1b). Similarly, changes in future simulated rainfall variance over a network of proxy locations (Figure  21.1d; see caption for details) are qualitatively consistent with the reconstructed ENSO‐related hydroclimate variance trends. This finding raises the issue as to whether 20th century changes in ENSO characteristics are already impacted by greenhouse warming and whether observed trends are already emergent against natural variability. However, one still has to note that a significant portion of the paleoclimate proxy records that inform these reconstructions have individual lengths spanning decades to centuries; therefore, it is possible that they be sampling to an extent natural decadal ENSO variability independent of external forcing. With the additional consideration of the impacts of ENSO diversity and seasonal cycle changes in these records, as discussed above, the issue of separating these impacts in proxy records and assessing confidence in the changes they reflect becomes central in our efforts to increase our confidence in ENSO simulations and projections. The same sample‐size uncertainties plague the instrumental record, especially considering that the observing system in the tropical Pacific was not well established until the 1980s, thus providing less than 40 years of accurate records for tropical Pacific variability (chapter 3). To illustrate the limitations to detecting changes in the standard deviation of ENSO sea surface temperature anomalies imposed by the length of the instrumental record, we can determine the sample size required to detect a percentage change in SST standard deviation using classical statistical power analysis (Cohen, 1988). Let us assume that we have an original sample of 40 values of SST anomalies averaged over December‐ January‐February (DJF; i.e. 40 years); we can empirically estimate the cumulative probability distribution function from these values, which is positively skewed (large El Niño events are greater in magnitude than large La Niña events). If we would like to detect a 60% decrease in standard deviation (similar to that indicated

ENSO IN A CHANGING CLIMATE  479

in Figure 21.1) with 75% power at a 95% confidence level (i.e. having a 75% chance of detecting the decrease at the 95% confidence level), we would need at least 20 more years of data (DJF values).1 Attempting to answer more involved and arguably more critical questions, such as whether ENSO is a chaotic phenomenon or a stable noise‐driven system, require much longer records: Cane et  al. (1995) showed that at least 500–1000 years of Niño‐3 SST anomalies are needed in order to distinguish between data from the chaotic CZ model and data from a noise‐driven stable model statistically derived from CZ simulations (Sarachik & Cane, 2010). And ultimately, the statistical tests appropriate to detect changes in ENSO variance or extremes depend on answering this question, i.e. confirming the underlying hypotheses about the nature of ENSO as a dynamical phenomenon [see Trenberth & Hoar (1996, 1997) and the energetic scientific exchange that followed in Rajagopalan et  al. (1997, 1999), Trenberth and Hurrell (1999a, 1997b), and Wunsch (1999a, 1997b), inter alia]. It is thus obvious that an irregular and skewedly distributed phenomenon like ENSO might very well be presenting a catch‐222, and we are still far from being able to confidently detect and hence attribute changes in its properties based on the available length of the instrumental record. Finally, reducing model uncertainty in projected ENSO changes goes hand in hand with reducing model uncertainty in projections of tropical Pacific mean‐state changes (chapter  8). There is still disagreement among reanalysis datasets and among CGCMs on the pattern of tropical Pacific temperature change (Vecchi et  al., 2008, Karnauskas et al., 2009; Deser et al., 2010; DiNezio et  al., 2013); some models project a larger relative increase of east Pacific temperatures relative to west Pacific temperatures, while others project more uniform changes or less warming of the cold tongue compared to the warm pool (chapter  13; Kohyama et  al., 2017;

1  Under the null hypothesis, two samples of size n and m have equal 2 2 variances ( 2 1 2 ), and the F-ratio statistic follows an F distribution with m - 1, n - 1 degrees of freedom (F  m - 1, n - 1). If we wish to detect an effect ρ (change in standard deviation) with a power of 1 - β, we need Pr(ρ2F ≤ α) ≥ 1 - β, where α is the desired p-value and 1 - β is the desired statistical power of the test (here, 75%). Here we use the empirical distribution function of DJF Niño-3.4 SST anomalies estimated from the last 40 years of NCEP data (Reynolds et al., 2002) to produce samples of size n and m. To estimate the sample size m that is required to detect a 60% decrease in variance compared to a sample of size n = 40 years, we calculate the ratio of the two samples’ variances (F-statistic) and its distribution from 10,000 simulations. All calculations are done using R. 2  A catch-22 is a paradoxical situation “for which the only solution is denied by a circumstance inherent in the problem or by a rule” (Merriam-Webster Dictionary). The phrase was coined by Joseph Heller in his 1961 satirical novel Catch-22, in which a World War II bombardier is caught in a bureaucratic loophole whereby he tries to avoid dangerous combat missions by requesting a mental evaluation for insanity, but his desire to avoid them is taken to prove his sanity.

Karamperidou et al., 2017). Karamperidou et al. (2017) showed that models with increased ENSO nonlinearity project less relative warming of the cold tongue owing to their simulation of atmospheric damping in the east Pacific, and the difference in projected SST changes between the models that exhibit different degrees of ENSO nonlinearity can be more than one degree Celsius per century. As with ENSO–seasonal cycle interactions, the two‐way interactions between ENSO and the mean state, which include the impact of residual nonlinear dynamical heating associated with ENSO (Jin et  al., 2003; Hayashi & Jin, 2017), require further and continuous attention. 21.4. CONCLUDING REMARKS AND FUTURE PERSPECTIVES Advances in computational power and the addition of new components in the current generation of Earth system models (which now incorporate, for example, interactive carbon, the sulfur and ozone cycles, aerosols, and vegetation) come with promises for improved understanding of ENSO and facilitate new or renewed focus on ENSO interactions with a vast number of processes. The list of fascinating problems that remain unsolved includes and is not limited to the following: the limits of ENSO predictability, tropical and extratropical interbasin interactions, ENSO impacts on marine heatwaves and associated ecological processes, ENSO interactions with short time and small space scale oceanic and atmospheric processes (e.g. tropical instability waves, vertical mixing, moist convection, and regional cloud processes), the relationship between ENSO and aerosol forcing, the importance of biogeochemical processes in the ocean (e.g. algae blooms and their feedback on ocean physics), and land‐use/vegetation changes, as well as the role of thermodynamical feedbacks associated with marine stratus clouds in the eastern Pacific. Some of these are illustrated in Figure 21.4, while many of them are touched upon in this book. The development of coordinated efforts by the scientific community to diagnose and better understand a wide array of processes via model intercomparison projects encourages hope for new and exciting insights into a number of (also wide‐) open ENSO questions. The current list of intercomparison projects’ efforts in the sixth phase of the Climate Model Intercomparison Project (CMIP6) includes strategic foci on aerosols and chemistry, the carbon cycle, cloud feedbacks, decadal ­climate, global monsoons, high‐resolution models, land‐ use and land‐surface processes, ocean models, paleoclimate, volcanic forcing, sea ice, ice sheets, polar amplification, and more; all of them bearing a relevance for better understanding certain aspects of ENSO and its impacts on a wide variety of phenomena. Importantly, such targeted experiments provide the opportunity

480  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

to  explore the relationship of all these processes (background state changes, tropical‐extratropical ­interactions, natural and anthropogenic external forcing, seasonal cycle changes) with the limits of ENSO predictability. The sources of ENSO predictability and the cause of the spring predictability barrier remain an open question (chapter  10) that is accompanied by active scientific debates on whether eastern or central Pacific El Niño or La Niña is more predictable (e.g. Kim et  al., 2009; McPhaden, 2012; Dommenget et al, 2013; Imada et al., 2015; Chen et al., 2016; Ren et al., 2019; chapter 7), on the mechanisms of decadal modulation of predictability (Karamperidou et al., 2014; Ramesh et al., 2016), on the relationship between ENSO asymmetry and predictability (chapter  7), and on whether there is asymmetry in predictability between the onset and decay phase of El Niño (Cai et al., 2003, Cheng et al., 2010a, 2010b; Hou et al., 2018). Ultimately, improving the accuracy of ENSO predictions and increasing the lead time of prediction ahead of the spring predictability barrier Tropical cyclones

(see chapter  10) is crucial for seasonal predictions of temperature, rainfall, extreme events, and ecological impacts around the globe (section VI) and are hence ­crucial for mitigation of ENSO impacts on life, property, and the economy of many regions. One of the greatest challenges that arguably permeates all aspects of current and future ENSO research is the ­detection and attribution of changes in ENSO properties as they relate to changes in the tropical Pacific mean state, including ENSO diversity and predictability. While advances in ­computational power and statistical methods, including sophisticated machine‐learning algorithms, aid toward better separating signals from noise, it is doubtful that any leaps in the detection and attribution problem can be made without the expansion and improvement of the ocean and atmosphere observing systems (especially in the far eastern Pacific, where strong El Niño events develop), which will provide longer, more accurate, and spatially extended instrumental records (see section 21.3 above for discussion). Ecosystem feedbacks

Tropical instability waves

Interbasin connections

Marine heatwaves

Cloud processes

Figure 21.4  Open ENSO problems and future directions. Indicative list of potential foci of future ENSO research in light of advancements in observing systems and climate model resolution and complexity, including but not limited to (i) interbasin connections, (ii) tropical cyclones, (iii) ecosystem feedbacks, (iv) tropical instability waves, (v) cloud processes, and (vi) marine heatwaves. The center panel shows observed sea surface temperature variance calculated from HadISST data (Rayner et  al., 2003). The tropical cyclone image is published by NASA Visible Earth (credit Jesse Allen) using VIIRS data from the Suomi National Polar‐orbiting Partnership. The remaining depicted processes are from a coupled high‐resolution (~1/4° horizontal atmosphere resolution and ~1/10° horizontal ocean resolution) present‐day control simulation using CESM 1.2 conducted by the IBS Center for Climate Physics (ICCP).

ENSO IN A CHANGING CLIMATE  481 Energy

re ltu icu r Ag

Disast er r isk red uc tio n

Weather noise, remote effects Biosphere

Atmosphere

c hea

ea

ra t i

Fo

od

se

cu

em

en

t

on

Impacts

M ig

Linear atmosphereocean modes

Pub li

li n r

lt h

s

Non

H u m an

Orbital changes, aerosols, and greenhouse gas forcing

ENSO complexity

cts

pa

Im m ic

cts

na

pa

dy

Im

Annual cycle

Hydrosphere r it y

te Wa

a rm

na

g

Figure 21.5  ENSO impacts and interactions. Illustration of the core concepts of the unified framework for ENSO and its impacts on atmosphere, hydrosphere, biosphere, and sectors of the anthroposphere, including water management, public health, disaster risk reduction, energy, agriculture, human migration, and food security.

In addition, history has shown that increases in resolution and complexity of models inadvertently pushes us deeper into the conceptual abyss (Bony et al., 2013) and is not always accompanied by commensurate leaps in our understanding of the fundamental processes governing the climate system. Historically, there has been deep appreciation in the ENSO community for continuing and advancing a hierarchical approach to modeling of ENSO: The development and use of conceptual and intermediate complexity models has led to improved theoretical understanding of ENSO and to new ways of interpreting the results of comprehensive climate models (section III). It is this combination of observations and models spanning the full range of complexity (chapter 9) that may ultimately answer remaining critical questions, such as, Is ENSO diversity part of a continuum or a manifestation of distinct linear modes or regimes (chapter 4; Capotondi et al., 2015, Takahashi et al., 2018, Timmermann et al., 2018)? Is ENSO linearly subcritical or nonlinearly supercritical (i.e. is its growth rate slightly negative or slightly positive), and what are the causes of ENSO asymmetry, periodicity, and its broad spectrum (chapters 6–7; Jin 1997, 1998)? Is there determinism in the ENSO system or does the role of stochastic forcing overwhelm determinism, thus limiting its predictability (chapter  10; Karamperidou et  al., 2014; Newman & Sardeshmukh, 2017)? Finally, will ENSO’s characteristics and/or its impacts change in a future warmer world,

and when will these potential changes be detectable (chapter 13)? On a final note, the Sirens of ENSO have lured and continue to lure a richly diverse community of scientists owing to the phenomenon’s unifying coupled nature that operates across the atmosphere, the ocean, and the earth (McPhaden et al., 2006), much like the mythological creatures who sought to unite the souls with their celestial, earthly, and oceanic hosts3 (Plato, Republic, 10.614– 10.621). Indeed, ENSO interactions and impacts span the atmosphere, hydrosphere, biosphere, and anthroposphere (Figure 21.5). In the end, ENSO exemplifies the need for and successes of inter‐ and cross‐disciplinary research. This book is a testament to that, having brought together a diverse community of more than 90 researchers, representing several generations of ENSO research. We can only be optimistic that it is with such broad and cross‐­ disciplinary approaches, along with improved proxy records, sustained observational networks, and a hierarchy of models to guide ENSO theory, that significant progress in future ENSO research will be achieved. 3  The nature and role of Sirens appears ambiguous across Greek mythology. Proclus (412–185 CE), the most prominent Greek Neoplatonist philosopher, attempts to reconcile their ambiguous nature by advancing three categories of Sirens: celestial, terrestrial/oceanic (Homeric Sirens), and chthonic (Proclus, In Platonis rem publicam commentarii).

482  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE

ACKNOWLEDGMENTS The authors would like to acknowledge Pamela Grothe, Kim M. Cobb, Karl Stein, Elke Zeller, and Christian Wengel for sharing their data and/or expertise, and Mark Cane for the inspiring discussions. C.K. is supported by U.S. NSF Award #AGS‐1602097. A.T., M.F.S., S.‐S.L., and K.S.Y. acknowledge support from the Institute for Basic Science (project code IBS‐R028‐D1). W.C. and A.S. are supported by the Centre for Southern Hemisphere Oceans Research (CSHOR), a joint research center between QNLM and CSIRO, as well as the Earth Systems and Climate Change Hub of the Australian Government’s National Environmental Science Program (NESP). We thank the participants of the ENSO complexity workshop hosted by ICCP in Busan (as listed in https://www. nature.com/articles/s41586‐018‐0252‐6) for their contribution to the discussions that led to the generation of Figures. 21.1 and 21.5. This is PMEL contribution 5022. REFERENCES Adams, J. B., M. E. Mann, & C. M. Ammann (2003). Proxy evidence for an El Niño‐like response to volcanic forcing. Nature, 426, 274–278. doi:10.1038/nature02101 (2003) An, S.-I., Y. Ham, J. Kug, A. Timmermann, J. Choi, & I. Kang (2010). The inverse effect of annual‐mean state and annual‐ cycle changes on ENSO. J. Climate, 23, 1095–1110. https:// doi.org/10.1175/2009JCLI2895.1 Bellenger, H., E. Guilyardi, J. Leloup, M. Lengaigne, & J. Vialard (2014). ENSO representation in climate models: From CMIP3 to CMIP5. Clim. Dyn., 42(7–8), 1999–2018. Bony, S., B. Stevens, I. H. Held, J. F. Mitchell, J.‐L. Dufresne, K. A. Emanuel, et  al. (2013). Carbon dioxide and climate: Perspectives on a scientific assessment. In G. Asrar & J. Hurrell (Eds.), Climate science for serving society (pp. 391– 413). Dordrecht: Springer‐Verlag. Cai, M., E. Kalnay, & Z.Toth (2003). Bred vectors of the Zebiak–Cane model and their potential application to ENSO predictions. Journal of Climate, 16(1), 40–56. https://doi.org/ 10.1175/1520‐0442(2003)0162.0.CO;2 Cai, W., G. Wang, B. Dewitte, L. Wu, A. Santoso, K. Takahashi, Y. Yang, et al. (2018). Increased variability of eastern Pacific El Niño under greenhouse warming. Nature, 564, 201–206. Cai, W., L. Wu, M. Lengaigne, T. Li, S. McGregor, J.‐S. Kug, et  al. (2019). Pantropical climate interactions. Science, 363. doi: 10.1126/science.aav4236 Cane, M. A., S. Zebiak, & Y. Xue (1995). Model studies of the long‐term behavior of ENSO. In D.G. Martinson, et  al. (Eds.), Natural climate variability on decade‐to‐century time scales (pp. 442–457). DEC‐CEN Workshop, Irvine, CA. Washington, DC: National Academy Press. Capotondi, A., A. T. Wittenberg, M. Newman, E. Di Lorenzo, J.‐Y. Yu, P. Braconnot, et  al. (2015). Understanding ENSO Diversity. Bulletin of the American Meteorological Society, 96(6) 921–938. doi:10.1175/bams‐d‐13‐00117.1

Carré, M., J. P. Sachs, S. Purca, A. J. Schauer, P. Braconnot, R. A. Falcón, et al. (2014). Holocene history of ENSO variance and asymmetry in the eastern tropical Pacific. Science, 345, 1045–1048. doi:10.1126/science.1252220 Chang, P., B. Wang, T. Li, & L. Ji (1994). Interactions between the seasonal cycle and the Southern Oscillation: Frequency entrainment and chaos in a coupled ocean‐atmosphere model. Geophysical Research Letters, 21(25), 2817–2820. Chen, C., M. A. Cane, N. Henderson, D. E. Lee, D. Chapman, D. Kondrashov, & M. D. Chekroun (2016). Diversity, nonlinearity, seasonality, and memory effect in ENSO simulation and prediction using empirical model reduction. Journal of Climate, 29(5), 1809–1830. Chen, Y. Y. & F.‐F. Jin (2017). Dynamical diagnostics of the SST annual cycle in the eastern equatorial Pacific: Part II. Analysis of CMIP5 simulations. Clim Dyn, 49. https://doi. org/10.1007/s00382‐017‐3550‐z Chen, Y. Y. & F.‐F. Jin (2018). Dynamical diagnostics of the SST eannual cycle in the eastern equatorial Pacific: Part I. A linear coupled framework. Clim Dyn 50. https://doi. org/10.1007/s00382‐017‐3725‐7 Cheng, Y., Y. Tang, X. Zhou, P. Jackson, & D. Chen (2010a). Further analysis of singular vector and ENSO predictability in the Lamont model: Part I. Singular vector & the control factors. Climate Dynamics, 35(5), 807–826. https://doi. org/10.1007/s00382‐009‐0595‐7 Cheng, Y., Y. Tang, P. Jackson, D. Chen, X. Zhou, & Z. Deng (2010b). Further analysis of singular vector and ENSO predictability in the Lamont model: Part II. Singular value and predictability. Climate Dynamics, 35(5), 827–840. https://doi. org/10.1007/s00382‐009‐0728‐z Clement, A., R. Seager, & M. A. Cane (2000). Suppression of El Niño during the mid‐Holocene by changes in the Earth’s orbit. Paleoceanography, 15(6), 731–737. Cobb, K. M., N. Westphal, H. R. Sayani, J. T. Watson, E. Di Lorenzo, H. Cheng, R. L. Edwards, & C. D. Charles (2013), Highly variable El Niño‐Southern Oscillation throughout the Holocene. Science, 339(6115), 67–70. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.), Mahwah, NJ: Erlbaum. Deser, C., A. S Phillips, & M. A. Alexander (2010). Twentieth century tropical sea surface temperature trends revisited. Geophys. Res. Lett., 37, L10701. Deser, C., A. S. Phillips, R. A. Tomas, Y. M. Okumura, M. A. Alexander, A. Capotondi, et  al. (2012). ENSO and Pacific decadal variability in the Community Climate System Model Version 4. J. Clim., 25, 2622–2651. DiNezio, P. N., G. A. Vecchi, & A. C. Clement (2013). Detectability of changes in the Walker Circulation in response to global warming. J. Climate, 26, 4038–e. doi: http://dx.doi. org/10.1175/JCLI‐D‐12‐00531.1 Dommenget, D., T. Bayr, & C. Frauen (2013). Analysis of the non‐linearity in the pattern and time evolution of El Niño Southern Oscillation. Climate Dyn., 40, 2825–2847. doi:10.1007/s00382‐012‐1475‐0 Gebbie, G., I. Eisenman, A. T. Wittenberg, & E. Tziperman (2007). Modulation of westerly wind bursts by sea surface temperature: A semistochastic feedback for ENSO. J Atmos Sci, 64, 3281–3295. doi:10.1175/JAS4029.1

ENSO IN A CHANGING CLIMATE  483 Ghil M., M. Chekroun, & E. Simonnet (2008). Climate dynamics and fluid mechanics: Natural variability and related uncertainties. Physica D, 237, 2111–2126. Grothe, P. R., Cobb, K. M., Liguori, G., Di Lorenzo, E., Capotondi, A., Lu, Y., et al. ( 2020). Enhanced El Niño–Southern oscillation variability in recent decades. Geophysical Research Letters, 47, e2019GL083906. https://doi.org/10.1029/2019GL083906 Hawkins, E., & R. Sutton (2009). The potential to narrow uncertainty in regional climate predictions. Bull. Amer. Meteor. Soc., 90, 1095–1107. Hayashi, M., & F.‐F. Jin (2017). Subsurface nonlinear dynamical heating and ENSO asymmetry. Geophysical Research Letters, 44, 12427–12435. https://doi.org/10.1002/2017 GL075771 Hou, Z., J. Li, R. Ding, C. Karamperidou, W. Duan, T. Liu, & J. Feng, J. (2018). Asymmetry of the predictability limit of the warm ENSO phase. Geophysical Research Letters, 45. https:// doi.org/ 10.1029/2018GL077880 Hurrell, J. W., M. M. Holland, P. R. Gent, S. Ghan, J. E. Kay, P. J. Kushner, et  al. (2013). The Community Earth System Model: A framework for collaborative research. Bull. Amer. Meteor. Soc., 94, 1339–1360. https://doi.org/10.1175/ BAMS‐D‐12‐00121.1 Imada, Y., H. Tatebe, M. Ishii, Y. Chikamoto, M. Mori, M. Arai, et  al. (2015). Predictability of two types of El Niño assessed using an extended seasonal prediction system by MIROC. Mon. Wea. Rev., 143, 4597–4617. https://doi. org/10.1175/MWR‐D‐15‐0007.1 Jin, F.-F., J. D. Neelin, & M. Ghil (1994). El‐Niño on the Devil’s Staircase: Annual subharmonic steps to chaos. Science, 264(5155), 70–72. Jin, F.–F. (1997). An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model. Journal of the Atmospheric Sciences, 54(7), 811‐829. https://doi.org/10.1175/1520‐0469(1 997)054%3C0811:AEORPF%3E2.0.CO;2 Jin, F.‐F. (1998). A simple model for the Pacific cold tongue and ENSO. Journal of the Atmospheric Sciences, 55(14), 2458– 2469. https://doi.org/10.1175/1520‐0469(1998)055%3C2458: ASMFTP%3E2.0.CO;2 Jin, F.‐F., S.-I. An, A. Timmermann, & J. Zhao (2003). Strong El Niño events and nonlinear dynamical heating. Geophys. Res. Lett., 30(3), 1120. doi:10.1029/2002GL016356 Karamperidou, C., M. A. Cane, U. Lall, & A. T. Wittenberg (2014). Intrinsic modulation of ENSO predictability viewed through a local Lyapunov lens. Climate Dynamics, 42, 253– 270. doi: 10.1007/s00382‐013‐1759‐z Karamperidou, C., P. N. Di Nezio, A. Timmermann, F.‐F. Jin, & K. M. Cobb (2015). The response of ENSO flavors to mid‐ Holocene climate: Implications for proxy interpretation. Paleoceanography, 30(5), 527–547. Karamperidou, C., F.‐F. Jin, & J. L. Conroy (2017). The importance of ENSO nonlinearities in tropical Pacific response to external forcing. Clim Dyn, 49, 2695. https://doi.org/10.1007/ s00382‐016‐3475‐y Karnauskas, K. B., R. Seager, A. Kaplan, Y. Kushnir, & M. A. Cane (2009). Observed strengthening of the zonal sea surface temperature gradient across the equatorial Pacific Ocean. Journal of Climate, 22, 4316–4321.

Kiefer, J. & C. Karamperidou (2019). High‐resolution modeling of ENSO‐induced pre‐ cipitation in the tropical Andes: Implications for proxy interpretation. Paleoceanography and Paleoclimatology, 34. https://doi.org/10.1029/2018PA003423 Kim, H. M., P. J. Webster, & J. A. Curry (2009). Impact of shifting patterns of Pacific Ocean warming on North Atlantic tropical cyclones. Science, 325, 77–80. Kleeman, R. (2008). Stochastic theories for the irregularity of ENSO. Philos Trans R Soc A 366(1875), 2509–2524. Kohyama, T., D. L. Hartmann & D. S. Battisti (2017). La Nina‐ like mean‐state response to global warming and potential oceanic roles. J. Climate, 30, 4207‐4225. Le, T. (2017). ENSO response to external forcing in CMIP5 simulations of the last millennium. Global and Planetary Change, 148, 105‐112. doi:10.1016/j.gloplacha.2016.12.002 (2017) Li, J., S.‐P. Xie, E. R. Cook, M. S. Morales, D. A. Christie, N. C Johnson, et  al. (2013). El Niño modulations over the past seven centuries. Nature Climate Change, 3, 822–826. doi:10.1038/nclimate1936 Liu, Y., K. M. Cobb, H. Song, Q. Li, C.‐Y. Li, T. Nakatsuka, et al. (2017). Recent enhancement of central Pacific El Niño variability relative to last eight centuries. Nature Communications, 8. doi:10.1038/ncomms15386 Liu, Z. G. (2002). A simple model study of ENSO suppression by external periodic forcing. Journal of Climate, 15(9), 1088‐1098. Liu, Z. Y., Z. Lu, X. Wen, B. L. Otto‐Bliesner, A. Timmermann, & A.K.M. Cobb (2014). Evolution and forcing mechanisms of El Niño over the past 21,000 years. Nature, 515(7528), 550–553. Masson‐Delmotte, V. et al. (2013). Information from paleoclimatic archives. In Stocker, T.F., Qin, D., Plattner, G.‐K., Tignor, M., Allen, S.K., Boschung, J., et al., Climate Change 2013: The physical basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press. McGregor, S., A. Timmermann, M. H. England, O. E. Timm, & A. T. Wittenberg (2013). Inferred changes in El Niño‐ Southern Oscillation variance over the past six centuries. Climate of the Past, 9, 2269–2284. doi:10.5194/cp‐9‐2269‐2013 McGregor, S., A. Timmermann, & O. Timm (2010). A unified proxy for ENSO and PDO variability since 1650. Climate of the Past, 6, 1–17. McGregor, S., M. F. Stuecker, J. B. Kajtar, M. H. England, & M. Collins (2018). Model tropical Atlantic biases underpin diminished Pacific decadal variability. Nature Climate Change. doi:10.1038/s41558‐018‐0163‐4 McPhaden, M. J., S. E. Zebiak, & M. H. Glantz (2006). ENSO as an integrating concept in Earth science. Science, 314(5806), 1740‐1745. doi: 10.1126/science.1132588 McPhaden, M. J. (2012). A 21st century shift in the relationship between ENSO SST and warm water volume anomalies. Geophys. Res. Lett., 39, L09706. doi:10.1029/2012GL051826 Mitchell, T. P., & J. M. Wallace (1991). On the annual cycle in equatorial convection and sea surface temperature. J. Climate, 5, 1140–1156. Munnich, M., M. A. Cane, & S. Zebiak (1991). A study of self‐ excited oscillations of the tropical ocean‐atmosphere system. Part II: Nonlinear cases. J Atmos Sci, 48, 1238–1248.

484  EL NIÑO SOUTHERN OSCILLATION IN A CHANGING CLIMATE Newman, M. (2007). Interannual to decadal predictability of tropical and North Pacific sea surface temperatures. J. Climate, 20, 2333–2356. https://doi.org/10.1175/JCLI4165.1 Newman, M., & P. D. Sardeshmukh (2017). Are we near the predictability limit of tropical sea surface temperatures? Geophys. Res. Lett., 44. doi: 10.1002/2017GL074088 Rajagopalan, B., U. Lall, & M. A. Cane (1997). Anomalous ENSO occurrences: An alternative view. J. Clim., 10, 2351–2357. Rajagopalan, B., U. Lall, & M. A. Cane (1999). Comment on "Reply to the comments of Trenberth and Hurrell." Bull. Am. Meteorol. Soc., 80, 2724–2726. Ramesh, N., M. A. Cane, & R. Seager (2016). Predictability and prediction of persistent cool states of the tropical Pacific Ocean. Clim. Dyn. doi 10.1007/s00382‐016‐3446‐3 Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, et al. (2003). Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108(D14), 4407. doi: 10.1029/2002JD002670 Ren, H.‐L., J. Zuo, & Y. Deng (2019). Statistical predictability of Niño indices for two types of ENSO. Climate Dynamics, 52(9‐10), 5361–5382. Reynolds, R. W., N. A. Rayner, T. M. Smith, D. C. Stokes, & W. Wang (2002). An improved in situ and satellite SST analysis for climate. J. Climate, 15, 1609–1625. Sarachik, E. S., & M. A. Cane (2010). The El Niño‐Southern Oscillation phenomenon. London: Cambridge University Press. Stein, K., A. Timmermann, & N. Schneider (2011). Phase synchronization of the El Niño‐Southern Oscillation with the annual cycle. Physical Review Letters, 107(12). Stein, K., A. Timmermann, N. Schneider, F. F. Jin, M. F. Stuecker (2014). ENSO seasonal synchronization theory. Journal of Climate, 27(14), 5285–5310. Stuecker, M. F., F.‐F. Jin, A. Timmermann, & S. McGregor (2015a). Combination mode dynamics of the anomalous Northwest Pacific anticyclone. Journal of Climate, 28, 1093‐1111. doi: 10.1175/JCLI‐D‐14‐00225.1 Stuecker, M. F., F.‐F. Jin, & A. Timmermann (2015b). El Niño‐ Southern Oscillation frequency cascade. Proceedings of the National Academies of the Sciences, 112, 13490–13495. doi: 10.1073/pnas.1508622112 Takahashi, K., C. Karamperidou, & B. Dewitte (2018). A theoretical model of strong and moderate El Niño regimes. Climate Dynamics, 1–17. https://doi.org/10.1007/s00382‐018‐4100‐z Taschetto, A. S., A. S. Gupta, N. C. Jourdain, A. Santoso, C. C. Ummenhofer, & M. H. England (2014). Cold Tongue and warm pool ENSO Events in CMIP5: Mean state and future projections. J. Climate, 27, 2861–2885. https://doi. org/10.1175/JCLI‐D‐13‐00437.1 Timmermann, A., & F.‐F. Jin (2002). A nonlinear mechanism for decadal El Niño amplitude changes. Geophys Res Lett, 29(1), 1003. doi: 10.1029/2001GL013369 Timmermann, A., Jin, F., Collins, M. (2004). Intensification of the annual cycle in the tropical Pacific due to greenhouse warming. Geophysical Research Letters, 31(12).

Timmermann, A., S. J. Lorenz, S.‐I. An, A. Clement, & S. P. Xie (2007a). The effect of orbital forcing on the mean climate and variability of the tropical Pacific. Journal of Climate, 20(16), 4147–4159. Timmermann, A., et al. (2007b). The influence of a weakening of the Atlantic meridional overturning circulation on ENSO. Journal of Climate, 20(19), 4899–4919. Timmermann, A., S.-I. An, J. S. Kug, F. F. Jin, W. Cai, A.Capotondi, et  al. (2018). El Niño‐Southern Oscillation complexity. Nature, 559(7715), 535–545. https://doi. org/10.1038/s41586‐018‐0252‐6 Trenberth, K. E., & T. J. Hoar (1996). The 1990–1995 El Niño– Southern Oscillation event: Longest on record. Geophys. Res. Lett., 23, 57–60. Trenberth, K. E., & T. J. Hoar (1997). El Niño and climate change. Geophys. Res. Lett., 24, 3057–3060. Trenberth, K. E., & J. W. Hurrell (1999a). Comments on “The interpretation of short climate records, with comments on the North Atlantic and Southern Oscillations,” Bull. Am. Meteorol. Soc., 80, 2721–2722. Trenberth, K. E., & J. W. Hurrell (1999b). Reply to Rajagopalan, Lall, and Cane’s comment about “The interpretation of short climate records, with comments on the North Atlantic and Southern Oscillations,” Bull. Am. Meteorol. Soc., 80, 2726–2728. Tziperman, E., L. Stone, M. A. Cane, & H. Jarosh (1994). El Niño chaos: Overlapping of resonances between the seasonal cycle and the Pacific Ocean‐atmosphere oscillator. Science 264, 72–74. Vecchi, G. A., A. T. Wittenberg, & A. Rosati (2006). Reassessing the role of stochastic forcing in the 1997–1998 El Niño. Geophys Res Lett, 33, L01706. doi:10.1029/2005GL024738 Vecchi, G. A., A. C. Clement, & B. J. Soden (2008). Examining the tropical Pacific’s response to global warming. Eos, Transactions American Geophysical Union, 89(9), 81–83. Wang, B. (1993). On the annual cycle in the equatorial Pacific cold tongue. J. Climate, 6, 2351–2361. Wang, X.L. (1994). The coupling of the annual cycle and ENSO over the tropical Pacific. J. Atmos. Sci., 51, 1115–1136. https:// doi.org/10.1175/1520‐0469(1994)0512.0.Co;2 Watanabe, T., et  al. (2011). Permanent El Niño during the Pliocene warm period not supported by coral evidence. Nature, 471(7337), 209–211. Wittenberg, A. (2009). Are historical records sufficient to constrain ENSO simulations? Geophys. Res. Lett., 36, L12702. doi:10.1029/2009GL038710 Wunsch, C. (1999a). The interpretation of short climate records, with comments on the North Atlantic and Southern Oscillations. Bull. Am. Meteorol. Soc., 80, 245–256. Wunsch, C. (1999b). Reply. Bull. Am. Meteorol. Soc., 80, 2723. Xie, S. P. (1995). Interaction between the annual and interannual variations in the equatorial Pacific. Journal of Physical Oceanography, 25(9), 1930–1941. Zavala‐Garay, J., C. Zhang, A. M. Moore, A. T. Wittenberg, M. J. Harrison, A. Rosati, et al. (2008). Sensitivity of hybrid ENSO models to unresolved atmospheric variability. J Clim, 21, 3704–3721. doi:10.1175/2007JCLI1188.1

GLOSSARY This nonexhaustive list of terms and their definitions may be supplemented with other glossaries, such as http:// glossary.ametsoc.org/wiki/. Terms in italics that appear in definitions are themselves defined in the glossary. Accumulated Cyclone Energy (ACE):  a measure of tropical cyclone activity, defined as the square of tropical cyclone intensity integrated over its lifetime (in 6‐hourly intervals). Annual ACE of a specific tropical cyclone region is the integration of ACE over all tropical cyclones in that region in a year. Aerosols:  a collection of airborne solid or liquid parti­ cles, with a typical size between 0.01 and 10 μm residing in the atmosphere for at least several hours. Aerosols, which can be of either natural or anthropogenic origin, can influence climate directly through scattering and absorbing solar radiation, and indirectly through act­ ing as condensation nuclei for cloud formation or modifying cloud properties. Anomaly:  a deviation from normal, e.g. difference between a given value in a time series and the long‐term average. Atlantic Meridional Mode (AMM):  interannual vari­ ability in the tropical Atlantic that arises from air‐sea coupled dynamics. It consists of an interhemispheric gradient in sea surface temperature (SST) in the tropical Atlantic, often co‐occurring with ENSO events. This surface temperature pattern modulates the location of the Intertropical Convergence Zone in the Atlantic, thereby affecting rainfall over northeast Brazil and West Africa. Atlantic Niño:  an ENSO‐like climate phenomenon in the equatorial Atlantic Ocean. Its positive phase is charac­ terized by anomalous surface warming, and negative phase by anomalous cooling. Atmospheric river:  a long and narrow stream of strong water vapor transport typically associated with extra­ tropical cyclones, often leading to heavy precipitation due to topographical uplift or heating. Atmospheric convection:  vertical movement of air due to heating at the surface. As the surface heats up (e.g. due to incoming shortwave radiation), air parcels rise as they become less dense than the surroundings, often leading to cloud formation and precipitation.

Bjerknes feedback:  a positive feedback loop named after Jacob Bjerknes describing ocean‐atmospheric interactions over the tropical Pacific that govern development of El Niño and La Niña events. It involves reinforcing variations between surface winds, SST, and thermocline depth. Brewer‐Dobson circulation: meridional overturning circulation in the stratosphere transporting air from the equator to polar latitudes. Carbon cycle: the flow of carbon between the atmosphere, land, and oceans, involving biological, chemical, and physical processes. Central Pacific (CP) El Niño:  sometimes referred to as “El Niño Modoki”; an El Niño event characterized by maximum warm surface anomalies in the central equatorial Pacific. Coriolis force:  inertial force exerted by the Earth’s rota­ tion, deflecting moving fluid parcels to the right in the Northern Hemisphere and to the left in the Southern Hemisphere. Coupled Model Intercomparison Project (CMIP):  an international effort of several modeling groups to pro­ duce coordinated climate model experiments subject to common protocols to address specific questions posed by the Intergovernmental Panel on Climate Change. δ18O:  oxygen isotope composition in “delta” notation, referring to the relative departure of oxygen isotopic ratios from a standard (most commonly, the Vienna Standard Mean Ocean Water, or Pee Dee Belemnite). The δ18O of calcium carbonate reflects the combined influences of SST, seawater δ18O, and (in some taxa) a biological offset. Detrital:  referring to sedimentary material that origi­ nates from terrestrial rock fragments accumulating in lake or marine environments; typically indicative of high‐energy hydrodynamic conditions, and often used as proxies for runoff. Synonymous with “clastic.” Diabatic heating:  heating from processes that can change air temperature. Sources of diabatic heating can come from latent heat release, precipitation, moist convection, and radiation.

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 485

486 GLOSSARY

Easterly wind surge (EWS):  sometimes referred to as easterly wind burst (EWB); an abrupt strengthening of the equatorial trade winds, typically associated with weather systems in the western and central Pacific. An EWS can generate upwelling Kelvin waves and a west­ ward contraction of the Western Pacific Warm Pool. Eastern Pacific (EP) El Niño:  sometimes referred to as “cold tongue” El Niño, characterized by maximum surface warm anomalies in the equatorial cold tongue region of the eastern Pacific. Ekman drift (or Ekman current):  the lateral movement of water in the frictional boundary layer of a fluid. At the ocean surface, this flow is to the right of the wind in the Northern Hemisphere and to the left of the wind in the Southern Hemisphere because of the Coriolis force. Ekman layer:  the depth ranges over which frictional flow occurs in a fluid. Ekman pumping:  the vertical transport that balances lateral movement of water due to Ekman drift. Ekman feedback:  a positive feedback process in which warm sea SST anomalies during a developing El Niño weaken the equatorial trade winds, reducing the upwelling of climatologically cold subsurface water in the eastern equatorial Pacific, thus reinforcing the initial surface warming. The converse occurs during a developing La Niña. El Niño Southern Oscillation (ENSO):  the strongest year‐to‐year climate variability on the planet, origi­ nating in the equatorial Pacific Ocean through cou­ pled ocean‐atmosphere interactions. ENSO manifests itself in anomalous surface warming or cooling that tends to peak in boreal winter. El Niño:  the warm phase of ENSO, characterized by anomalous surface warming and weaker trade winds in the equatorial Pacific Ocean. El Niño Modoki:  a form of El Niño with unusually warm surface temperatures peaking in the central Pacific (see also Central Pacific El Niño). “Modoki” is a Japanese word for “similar but different.” Empirical orthogonal function (EOF):  a statistical tech­ nique for decomposing data into a series of compo­ nents each with characteristic spatial and temporal patterns that are orthogonal to and uncorrelated with another. Equatorial cold tongue:  a region in the eastern equatorial Pacific characterized by wind‐driven upwelling of cold subsurface waters. The cold tongue warms consider­ ably during eastern Pacific El Niño events and cools during La Niña events. A similar though weaker cold tongue exists in the equatorial Atlantic. External forcing:  forcing external to the climate system, such as volcanic eruptions, anthropogenic greenhouse

gas emissions, and variations in solar radiation due to either changes in the tilt of earth’s axis or changes in solar energy output. Foraminifera:  zooplankton whose calcium carbonate shells contain geochemical indicators of the envi­ ronment in which they grew (e.g. δ18O, elemental ratios). Future projection:  a simulation of the climate system subject to future emission scenarios for atmospheric greenhouse gases. Plausible scenarios include the Representative Concentration Pathways used by the fifth IPCC report. Geostrophic current:  an oceanic current resulting from a balance between pressure differences and the Coriolis force. Gill‐Matsuno response:  the atmospheric circulation response to a heating source in the tropics. In the case of a symmetric heat source about the equator, the tropical atmosphere generates an equatorially trapped response that produces a low‐level easterly flow via an equatorial Kelvin wave and an off‐ equator cyclonic flow in both hemispheres on the western flank of the heating area caused by Rossby waves. Glacial:  extended cold periods of widespread ice on a global scale, hence an ice age. Greenhouse effect: the heat‐trapping effect of greenhouse gases, such as carbon dioxide, water vapor, and methane, that effectively absorb and reradiate infrared radiation from the sun, heating the earth’s surface. Gross primary productivity (GPP):  the uptake of carbon from the environment by plants through photosynthesis. Hadley Circulation: thermally driven meridional circulation in the atmosphere consisting of poleward flow in the upper troposphere, subsiding air over high pressure regions of the subtropics, a return surface flow as part of the trade winds, and rising air in the Intertropical Convergence Zone. Hiatus:  a multiyear period with relatively little change in globally averaged surface temperatures compared to the long‐term global warming trend. The period from 1999 to 2013 was the most recent hiatus. Holocene:  the last 11,650 years of Earth’s history, a relatively warm period following the last glacial maximum. Indonesian Throughflow (ITF):  oceanic transport from the Pacific to the Indian Ocean through the passages of the Indonesian Archipelago. Indian Ocean basin mode (IOBM):  surface warming (cooling) that occurs throughout the Indian Ocean fol­ lowing the peak of El Niño (La Niña) events.

GLOSSARY  487

Indian Ocean Dipole (IOD):  an ENSO‐like climate phenomenon in the tropical Indian Ocean that peaks in boreal summer-fall. Its positive phase is character­ ized by anomalous surface cooling in the southeastern tropical Indian Ocean and anomalous warming over the western side of the basin. Interannual:  referring to variations on time scales longer than a year but shorter than a decade. Interdecadal:  referring to variations on time scales longer than a decade but shorter than a century. “Decadal” is sometimes used instead of interdecadal. Interdecadal Pacific Oscillation (IPO):  a decadal mode of climate variability occurring in the Pacific. Its positive phase features a meridionally broad El Niño– like surface temperature anomaly pattern with cooling at higher latitudes in the North and South Pacific. It is similar to the Pacific Decadal Oscillation, which exhibits a more prominent Northern Hemisphere signal than the IPO. Interglacial:  warm period between glaciations. The previous interglacial, dating from approximately from 129,000 to 115,000 years ago, is referred to as “last interglacial (LIG).” Intergovernmental Panel on Climate Change (IPCC):  United Nations body for assessing the science related to climate change. Intertropical Convergence Zone (ITCZ):  a basin‐scale belt of low pressure just north of the equator where the northeast trade winds meet the southeast trade winds. As these winds converge, uplifted moist air forms a band of heavy precipitation that migrates lati­ tudinally with the seasons. Intraseasonal:  referring to variations on time scales shorter than a season (e.g. weekly, monthly). Kelvin waves:  large‐scale waves in the atmosphere and ocean that propagate eastward along the equator. In the equatorial Pacific Ocean, westerly wind anomalies generate downwelling Kelvin waves, deepening the thermocline in the eastern Pacific. The converse occurs for easterly wind anomalies. Kelvin waves can also propagate along coastlines in the ocean and along mountain ranges in the atmosphere. Jet stream:  a narrow meandering band of very strong predominantly westerly air currents encircling the globe several kilometers above the earth. La Niña:  the cold phase of ENSO, characterized by anomalous surface cooling and stronger trade winds in the equatorial Pacific Ocean. Madden‐Julian Oscillation (MJO):  a dominant climate variability operating on intraseasonal time scales pre­ dominantly over the tropical Indian and Pacific Oceans, characterized by eastward progression of large regions of atmospheric convection.

Marine heatwave (MHW):  A discrete and prolonged period of anomalously warm water in a particular region. Monsoon:  seasonal reversal in surface winds and associ­ ated precipitation due to land‐sea heating contrasts. In summer, monsoon rains occur over land, while drying occurs over adjacent oceans. Net ecosystem productivity (NEP):  the difference between net primary productivity and nonplant respiration. Net primary productivity (NPP):  the overall uptake of carbon by plants, defined as gross primary productivity minus plant respiration. Ningaloo Niño:  persistent extreme ocean warming occurring off the western coast of Australia. Northern Annular Mode (NAM):  an atmospheric vari­ ability manifesting in a ring‐like vacillation in pressure between the Arctic and Northern Hemisphere midlat­ itudes. NAM is the Northern Hemisphere counterpart of SAM (see Southern Annular Mode). North Atlantic Oscillation (NAO):  the primary mode of internal atmospheric variability in the North Atlantic characterized by a north‐south dipole of alternating sea level pressure anomalies between the subtropics and high latitudes. North Pacific Oscillation (NPO):  north‐south seesaw in winter sea level pressure anomalies over the North Pacific on monthly (and shorter) time scales. Ocean heat content (OHC):  the quantity of heat stored in the upper ocean, proportional to temperature integrated vertically over a prescribed range of depths. Pacific Decadal Oscillation (PDO):  an SST pattern that varies on decadal timescales, characterized by pro­ nounced SST anomalies in the North Pacific and SST anomalies of opposite sign in the tropical Pacific. Pacific Meridional Mode (PMM):  interannual variability in the tropical Pacific arising from wind‐­evaporation‐ SST (WES) feedback. The PMM manifests itself as an SST dipole in a north‐south orientation extending from near the equator to the subtropics in either hemisphere. Pacific North American (PNA) pattern:  an internal mode of atmospheric variability characterized by a wave train of geopotential height anomalies from the central North Pacific arching toward North America. Centers of action are located in the subtropical north­ eastern Pacific, the Gulf of Alaska, northwestern North America, and the southeastern United States. It is often energized by ENSO events. Pacific South American (PSA) pattern:  a mode of atmo­ spheric variability analogous to the Pacific North American pattern, but in the Southern Hemisphere. It is often energized by ENSO events. Paleoclimate:  climate during periods prior to the development of measuring instruments, including

488 GLOSSARY

­ istoric and geologic time, for which only paleo-proxy h climate records are available. Paleo‐proxy:  An indirect measure of climate variability in the distant past such as inferred from tree rings, coral skeletons, ice cores, and sediments. Parameterization:  in climate modeling, a technique for representing processes that cannot be explicitly resolved at the spatial and temporal resolution of the model. Pliocene:  the interval of geological time between ~2.6 and 5.3 million years ago. Porites:  genus of reef‐building coral, includes many species used for paleoclimate reconstructions in the tropical Indo‐Pacific. Rapid Intensification (RI):  an intensification in the maximum sustained winds of a tropical cyclone exceeding 30 knots within 24 hours. Recharge‐discharge:  the result of meridional transport of heat into (recharge) and out of (discharge) the upper equatorial Pacific Ocean. A recharged state often precedes an El Niño while a discharged state often precedes La Niña. Remote forcing:  factors influencing a physical variable in a certain region that originate outside the region. For instance, ENSO can remotely force climate variations in the Indian and Atlantic Oceans, and vice versa. Representative Concentration Pathway (RCP):  a hypothet­ ical greenhouse gas concentration trajectory describing one of several plausible climate futures, adopted by the IPCC for its fifth Assessment Report in 2013–14. Respiration:  the release of carbon by organisms including animals, plants, and microbes. Rossby waves:  planetary waves in the atmosphere and ocean that arise due to the Earth’s rotation, typically propagating westward with phase speeds that decrease with increasing latitude. They play an important role in shaping weather patterns and ENSO evolution. Seasonal cycle:  also sometimes referred to as “mean seasonal cycle.” The average evolution of conditions over the 12 months of the calendar year. The seasonal cycle is the baseline use to compute interannual anom­ alies associated with ENSO. Seasonal phase locking:  the tendency of ENSO surface temperature anomalies to be strongest in boreal winter and weakest in boreal spring. South Atlantic Convergence Zone (SACZ):  a band of large‐scale enhanced convection, wind convergence, and cloudiness across South America from the Amazon basin toward the southwest Atlantic. The SACZ is most prominent during austral summer and defines the rainy season in southeast Brazil. South Atlantic Subtropical Dipole (SASD):  an air‐sea cou­ pled mode characterized by SST variability in the South Atlantic associated with modulations in the strength and position of the South Atlantic subtropical high.

Southern Annular Mode (SAM):  the leading mode of large‐scale atmospheric variability in the Southern Hemisphere, characterized by an anomalous pressure center over Antarctica and a zonally symmetric pressure anomaly of opposite sign at midlatitudes. The positive and negative phases of the SAM are respectively associated with a poleward and equator­ ward displacement of the midlatitude westerly winds. South Pacific Convergence Zone (SPCZ):  a large‐ scale band of strong convective activity that forms in the South Pacific in a northwest‐southeast orienta­ tion from the western Pacific toward the French Polynesia. Southern Oscillation Index (SOI):  a sea level pressure (SLP)‐based ENSO index defined in terms of Tahiti minus Darwin SLP anomalies. It captures the see‐saw of atmospheric pressure between the western and central Pacific during ENSO events. Spring predictability barrier:  a feature of the ENSO cycle in which forecast skill sharply drops in boreal spring, more so than in any other season of the year. Also sometimes referred to as the austral autumn pre­ diction barrier. Storm tracks:  main tracks of extratropical atmospheric disturbances occurring as sequences of low‐ and high‐ pressure systems propagating eastward with the pre­ vailing midlatitude westerly winds. State dependent noise:  also referred to as “multiplicative noise.” Random fluctuations in the atmosphere on time scales of a few days to weeks that are unpredict­ able on seasonal time scales but that can influence, and be influenced by, the development of El Niño and La Niña events. Surface heat flux:  the transfer of heat across the Earth’s surface consisting primarily of shortwave and long­ wave radiation, latent heat, and sensible heat. Teleconnection:  a connection between changes in atmo­ spheric or oceanic circulation over widely separated parts of the world. In physical terms, teleconnections are often a consequence of large‐scale wave motions, whereby energy is transferred from source regions along preferred paths in the atmosphere and ocean. For instance, atmospheric teleconnection refers to coherent atmospheric response to remote SST anom­ alies (e.g. during ENSO events), leading to far-field changes in environmental variables (e.g. air tempera­ ture, precipitation, SST, sea‐ice extent). Thermal damping:  a negative feedback to ENSO SST anomalies due to surface heat flux typically resulting from changes in cloud cover and latent heat. Thermocline:  a zone of maximum vertical temperature gradient, separating warm and cold layers of water. The 20°C isotherm is often used as an indicator of thermocline depth in the equatorial Pacific Ocean.

GLOSSARY  489

Thermocline feedback: a self‐reinforcing feedback involving surface warming, weakened trade winds, and thermocline deepening in the eastern equatorial Pacific during El Niño. This feedback works in the opposite sense to cool the surface during La Niña. Trade winds:  relatively steady easterly winds that domi­ nate most of the tropics and subtropics throughout the world, deflected westward by the Coriolis force as air flows from higher pressure subtropical regions toward the equator. Tropical instability waves (TIW):  wave‐like variations in the frontal boundary between warm and cold SSTs, most prominently occurring in the equatorial cold tongue regions of the Pacific and Atlantic oceans. They tend to be more prominent in the Pacific during La Niña events and very weak during El Niño events. Upwelling:  a process in which deep, cold water rises toward the surface, often related to alongshore wind forcing in coastal zones and trade wind forcing in equatorial regions. The reverse process of surface water forced downward into the ocean interior is referred to as “downwelling.” Walker Circulation:  thermally driven zonal overturning atmospheric circulation in the tropics associated with rising air in the Western Pacific Warm Pool and sinking air in the eastern Pacific. These rising and sinking air masses undergo significant longitudinal shifts in loca­

tion during ENSO events linked to SST anomaly development. Westerly wind burst (WWB):  often referred to as “west­ erly wind event” (WWE); refers to an abrupt relaxation of the equatorial trade winds, typically associated with weather systems in the western and central Pacific. WWBs generate downwelling Kelvin waves and an east­ ward expansion of the Western Pacific Warm Pool. Western Hemisphere Warm Pool (WHWP):  a region of SST warmer than 28.5°C that covers the eastern tropical North Pacific to the Gulf of Mexico and the Caribbean. Western Pacific Warm Pool:  a vast area of warm waters surrounding the Maritime Continent with tempera­ tures exceeding 28°C. Wind‐Evaporation‐SST (WES) feedback:  a thermody­ namical positive feedback loop involving changes in surface winds, surface evaporation, and sea surface temperature. Zonal advective feedback:  a process involving surface warming during El Niño when westward flowing sur­ face equatorial currents weaken because the trade winds weaken. This results in less water coming from the cooler eastern equatorial Pacific, which leads to further surface warming. The process works in reverse during La Niña for strengthening winds, surface zonal cur­ rents, and surface cooling.

INDEX Accumulated cyclone energy (ACE), 382 NA TCs and, 389 Adams, J. B., 271 Advanced Very High Resolution Radiometer (AVHRR), 43, 459 Aerosol optical depth (AOD), 271 Aerosols CMIP6 and, 479 GMST and, 33 from volcanoes, 268–71, 278 of El Chichón, 9 Africa droughts in, 7 ENSO atmospheric teleconnections in, 325 extreme climate in, 372 Indian Ocean El Niño and, 368 AGCM. See Atmosphere‐only GCM Agricultural production, 3, 41 AIR. See All‐India rainfall Alabia, I. D., 442 Alaska Current, 438–39 Aleutian Low in El Niño, 69 fisheries and, 435 North Pacific Subpolar Gyre and, 439 Alexander, M. A., 31 All‐India rainfall (AIR), 362 AMM. See Atlantic Meridional Mode AMO. See Atlantic Multidecadal Oscillation AMOC. See Atlantic meridonial overturning circulation Amplitude, of ENSO asymmetry of, 157–63, 158f, 160f, 161f El Niño and, 154 ENSO low‐frequency modulation and, 182 ENSO simple models for, 138–41 biases with, 212–13 AMV. See Atlantic Multidecadal Variability An, S.‐I., 101, 124, 155, 158, 165

Anchovy, 437–38 Antarctica ENSO atmospheric teleconnections in, 325–26 ice sheets of, 22–23 SAM and, 33 Anthropocene, 4–5 ENSO past variability and, 103 AOD. See Aerosol optical depth APE. See Available potential energy APEC. See Asia‐Pacific Economic Cooperation Arctic, 4, 23 climate change and, 34 Arctic Oscillation, 268 Argo, 45, 52 ARs. See Atmospheric rivers ASCAT, 43, 59 Ashok, K., 66, 67, 68f, 81 Asia. See also Monsoons ENSO atmospheric teleconnections in, 322–23 Asia‐Pacific Economic Cooperation (APEC), 235 Asymmetry, of ENSO, 153–67 of ENSO amplitude, 157–63, 158f, 160f, 161f of ENSO simple models, 143–44 evolution of, 163–66, 164f–66f, 167t Atlantic Meridional Mode (AMM), 320, 323–24 NA TCs and, 388 Atlantic meridonial overturning circulation (AMOC), 280–81 Atlantic Multidecadal Oscillation (AMO), 32 CP El Niño and, 78 ENSO low‐frequency modulation and, 183 NA TCs and, 388 Atlantic Multidecadal Variability (AMV) ENSO atmospheric teleconnections and, 323 ENSO low‐frequency modulation and, 183

Atlantic Niño, 76, 232, 249, 250, 254–56, 259–60 Atlantic Ocean. See also North Atlantic; South Atlantic ENSO atmospheric teleconnections in, 319–20 ENSO evolution asymmetry and, 166 ENSO low‐frequency modulation and, 183 ENSO remote forcing in, 254–57, 255f hurricanes in, 31 ATLAS. See Autonomous Temperature Line Acquisition System Atmosphere‐only GCM (AGCM), 205, 206, 210–11 Atmospheric bridge, 368 Atmospheric rivers (ARs), 367 Atmospheric teleconnections, 313f, 316f–18f in Atlantic Ocean, 319–20 ENSO and, 311–28 global warming and, 326–27 in Indian Ocean, 315, 319 in land, 320–26 in Maritime Continent, 320–21 in NA, 320 nonlinear, 314–15, 315f sea level extremes and, 411 in South Atlantic Ocean, 320 Atwood, A. R., 179 Australia BOM in, 76, 228, 230, 235 corals in, 418 ENSO atmospheric teleconnections in, 321–22 mangroves in, 422 MHWs and, 416 seagrass in, 421–22 sea level extremes and, 412 Autonomous Temperature Line Acquisition System (ATLAS), 43

El Niño Southern Oscillation in a Changing Climate, Geophysical Monograph 253, First Edition. Edited by Michael J. McPhaden, Agus Santoso, and Wenju Cai. © 2021 American Geophysical Union. Published 2021 by John Wiley & Sons, Inc. 491

492 INDEX

Available potential energy (APE), 72–73 El Niño and, 186 ENSO low‐frequency modulation and, 188f AVHRR. See Advanced Very High Resolution Radiometer Balmaseda, M. A., 54 Barnett, T., 228 Barnston, A. G., 234 Barrier layers, 216 Basher, R. E., 390 Battisti, D. S., 138 Bejarano, L., 124 Bell, R., 380f Bellenger, H. E., 206f, 477 Betts, R. A., 456 Bivalves, ENSO past variability and, 91, 94t Bjerknes, Jacob, 7–8, 42, 121, 122, 127, 138 Bjerknes feedback, 8, 29, 69 convective thresholds and, 207 ENSO amplitude asymmetry and, 161 ENSO models and, 206, 207 GHG and, 292, 298 in Indian Ocean, 251, 253 in La Niña, 10 in recharge/discharge oscillator model, 203 SST threshold and, 75 volcanoes and, 275, 277 WWBs and, 156 Bjerknes‐Wyrtki‐Jin Index (BWJ), RO and, 127–38, 128t, 130f, 132f, 133t, 135f, 136f, 137t, 138t BOM. See Bureau of Meteorology Bong, H., 101 Bonham, S. G., 99 Braganza, K., 272 Brewer‐Dobson circulation, 271 Bureau of Meteorology (BOM), in Australia, 76, 228, 230, 235 BWJ. See Bjerknes‐Wyrtki‐Jin Index C3S. See Copernicus Climate Change Service Cai, W., 77, 161, 260, 299, 394 California Cooperative Oceanic Fisheries Investigations (CalCOFI), 343 California Current Ecosystem (CCE), 436–38 California Undercurrent, 345

Camargo, S. J., 381, 382, 385, 392, 394 Cane, M. A., 127, 138, 177, 228–30, 479 Cane‐Zebiak model (CZ) ENSO and climate change and, 474 ENSO irregularity and, 155 ENSO low‐frequency modulation and, 177–78, 192 ENSO simple models, 122–27, 123f, 125f, 126f, 146–47 Capacitor effect, 166 Capotondi, A., 76 Carbon cycle, 3, 41 in El Niño, 454, 454f, 463–65, 464f, 465t ENSO and, 453–67 prediction of, 465–66, 466t, 467f GMST and, 190 in La Niña, 464–65 oceans and, 462–63 regional climate drivers of, 461–62 SST and, 455, 458f, 463 Carbon dioxide, 290, 293 atmosphere to ocean surface flow, 456–57, 456t atmospheric growth of, 455–56, 456f, 456t in El Niño, 459f ENSO low‐frequency modulation and, 175, 179 from fossil fuels, 22 global annual emissions of, 461, 461f GMST and, 22, 191f wildfires and, 459–61, 461f Carpenter, T. H., 9, 71 Carré, M., 100 Catch per unit effort (CPUE), 442, 443f Caves, ENSO past variability and, 93, 95t CCE. See California Current Ecosystem CCSM4, 276, 277 CDA. See Coupled data assimilation Central and eastern north Pacific (CEP), TCs in, 383–86, 387f Central equatorial Pacific (CEP), 159–60 Central North Pacific, fisheries in, 435–36 Central Pacific (CP), 10–11 El Niño in, 12, 14, 52, 58, 66, 67f, 69 AMO and, 78 Atlantic Ocean and, 254

ENSO atmospheric teleconnections and, 312, 315–16, 319, 320, 323, 324, 327 ENSO CZ and, 124 ENSO diversity and, 78–79 ENSO low‐frequency modulation and, 175 ENSO oceanic teleconnections and, 341, 344 ENSO remote forcing and, 257–58 fisheries and, 430 GHG and, 299 global warming and, 77 HCS and, 430 La Niña and, 76 NA TCs and, 388 NPSG and, 439 PDO and, 78 PMM in, 76 predictability of, 76–78 SSTA in, 67, 71, 74, 80f, 207–8 SST in, 75, 76–77 TCs and, 381f, 383 thermocline in, 75 WWBs in, 75, 76 ENSO in, 29, 30–31 climate change and, 474 CEP. See Central and eastern north Pacific; Central equatorial Pacific CERA‐20C, 56, 57 CERA‐SAT, 56 CESM. See Community Earth System Model CGCMs. See Coupled general circulation models Chan, J. C. L., 379, 380, 382, 383 Chan, K. T. F., 383 Chand, S. S., 391, 394 Chatterjee, A., 464 Chen, C., 77, 79, 215 Chen, D., 396 Chen, G., 383 Chen, H.‐C., 142 Chen, M., 167 Chiang, J. C. H., 364 Chiodi, A. M., 82 Clarke, A. J., 254 Clement, A. C., 290 Climate change, 33–34. See also Global warming ENSO and, 12–15 ISV and, 474–77, 475f–77f paleoclimate reconstruction and, 473–81, 480f

INDEX  493

ENSO diversity and, 77–78 GHG and, 290–93, 291f mangroves and, 422 Maritime Continent and, 298 sea level extremes and, 410 TCs and, 392–94, 395f volcanoes and, 268–71, 269f, 270f Climate Diagnostics Center, of NOAA, 15 Climate Forecast System, of NCEP, 257 Climate Model Intercomparison Project phase 3, 77, 78–79, 202 emergent constraints on, 214 ENSO amplitude asymmetry and, 161 GHG and, 290, 293, 297 understanding from, 205, 206, 206f volcanoes and, 274 Climate Model Intercomparison Project phase 5 (CMIP5), 102, 202 for ECT, 210 on El Niño, 209f emergent constraints on, 214–15, 215f ENSO amplitude asymmetry and, 161, 163 ENSO and climate change and, 474 GHG and, 290, 293, 297, 299, 301, 302 for SST, 211 understanding from, 205, 206f volcanoes and, 271, 274–75, 277f, 282 Climate Model Intercomparison Project phase 6 (CMIP6), 479 Climate Prediction Center (CPC), of NOAA, 228–29 Climate system atmosphere in, 22 ENSO in, 21–34 climate change from, 33–34 impacts of, 34 ice sheets in, 22–23 land in, 22 oceans in, 22 solar irradiance in, 22 SST in, 23 teleconnections in, 31–33, 32f tropical Pacific in, 23–25, 26f Cloud albedo ECT and, 212

volcanoes and, 277 CM2.1, 179 CMIP3. See Climate Model Intercomparison Project phase 3 CMIP5. See Climate Model Intercomparison Project phase 5 CMIP6. See Climate Model Intercomparison Project phase 6 CMIPs. See Coupled model intercomparison projects CNRM‐CM5, 79 CMT. See Convective momentum transport Cobb, K. M., 100 Colbert, A. J., 388 Cold tongue, 24, 25f. See also Equatorial cold tongue easterly winds and, 78 ENSO CZ and, 127 ENSO diversity and, 78 ENSO low‐frequency modulation and, 183 in interglacials, 99 Walker Circulation and, 77 Cold Tongue Index, 17 Cold tongue mode (CTM), 78 Collins, J. M., 385 Comboul, M., 91, 96 Community Earth System Model (CESM), 102, 278–79 ENSO and climate change and, 474 iCESM, 98, 99–100 Compo, G. P., 57 Convective momentum transport (CMT), 217 Convective thresholds, 202 Bjerknes feedback and, 207 ENSO amplitude and, 140 of SST, 82 in EP, 74 Cook, K. H., 325 Copernicus Climate Change Service (C3S), 230 Corals ENSO past variability and, 88–91, 90f, 94t frequency entrainment hypothesis and, 100 La Niña and, 96 ocean extremes and, 416–19, 419f, 420f Coupled data assimilation (CDA), 56

Coupled general circulation models (CGCMs), 122, 202, 205–6, 217–18 EMICs and, 204 ENSO amplitude and, 140 ENSO CZ and, 127 ENSO model biases of, 212 RO and BWJ and, 129 volcanoes and, 274, 275 Coupled linear instability, 122–27, 123f, 125f, 126f Coupled model intercomparison projects (CMIPs), 202, 217. See also Climate Model Intercomparison Project phase 3; Climate Model Intercomparison Project phase 5; Climate Model Intercomparison Project phase 6 on carbon cycle, 466, 467f ENSO atmospheric teleconnections and, 327 ENSO oceanic teleconnections and, 350 for Indian Ocean, 253 PMIP and, 208 on volcanoes, 268 Cox, P. M., 456, 466 CP. See Central Pacific CPC. See Climate Prediction Center CPUE. See Catch per unit effort Crowley, T. J., 272 CSIRO Mk3L, 179 CTM. See Cold tongue mode Cut‐off lows, 367–68 Cyclones. See Tropical cyclones CZ. See Cane‐Zebiak model Data assimilation (DA), 96. See also Simple Ocean Data Assimilation CDA, 56 Dateline El Niño, 67 Decadal timescales for ENSO, 12 for ENSO prediction, 235–38 Dee, S., 102 Delage, F., 326 Delayed oscillator (DO), 127 in EMICs, 204 ENSO amplitude asymmetry and, 161–63 ENSO low‐frequency modulation and, 176–77 RO and, 137–38

494 INDEX

Deser, C., 162, 165, 349 Detrital, 92, 93 Dewitte, B., 161 Diabatic heating, 31, 312 Di Lorenzo, E., 416 DiNezio, P. N., 162, 165 Ding, H., 188–89, 212 Diversity. See ENSO diversity DO. See Delayed oscillator Drake Passage, 344 Drexler, J. Z., 422 Droughts, 3, 6f in Africa, 7 in El Niño, 29, 364, 364f, 436 ENSO and, 334 with heat waves, 368–69 tree rings and, 92 Du, Y., 380 EAC. See East Australian Current E and C indices, 81 Earth models of intermediate complexity (EMICs), 204–5, 217–18 Earth’s energy imbalance (EEI), 22–23 climate change and, 34 Earth system models (ESMs), 454, 466, 467f CESM, 102, 278–79, 474 iCESM, 98, 99–100 MIROC, 79, 274 NorESM, 274, 278, 279 East Australian Current (EAC), 338 Easterly winds cold tongue and, 78 surges, 10 Eastern Pacific (EP), 10–11 ENSO atmospheric teleconnections and, 324 SST in, convective thresholds in, 74 Eastern Pacific (EP), El Niño in, 12, 14, 29, 58, 66, 67f, 69 Atlantic Ocean and, 256 ENSO atmospheric teleconnections and, 312, 323, 327 in Atlantic Ocean, 320 in Indian Ocean, 319 ENSO diversity and, 78–79 ENSO low‐frequency modulation and, 175 ENSO oceanic teleconnections and, 344 fisheries and, 430, 431f GHG and, 299 HCS and, 430 NA TCs and, 388

NPSG and, 439 predictability of, 76–78 SSTA in, 67, 69–71, 74, 207–8 SST in, 74, 75, 76–77 TCs and, 383 thermocline in, 75 time series for, 71f westerly winds in, 69, 70f WWBs in, 75 ECMWF, 55, 56, 56f SEAS5, 236 SST and SSH and, 76 ECT. See Equatorial cold tongue Eddebbar, Y. A., 277 EEI. See Earth’s energy imbalance EEZ. See Exclusive economic zone Eisenman, I., 157 Ekman divergence, 290 Ekman feedback ENSO low‐frequency modulation and, 181 ENSO model biases of, 212 GHG and, 296 Ekman layer, MHWs and, 416 Ekman outflow, 23 CZ and, 123 Ekman pumping, ENSO oceanic teleconnections and, 341 Ekman suction, 212 Ekman theory, 46 Ekman transport, ENSO oceanic teleconnections and, 344 Ekman upwelling, RO and BWJ and, 129 El Chichón volcano, 9, 43, 228 ELI. See ENSO Longitude Index El Niño, 22, 361–62. See also Central Pacific; Eastern Pacific Aleutian Low in, 69 APE and, 186 Atlantic Ocean and, 254 hurricanes in, 31 canonical, 71 carbon cycle in, 454, 454f, 463–65, 464f, 465t carbon dioxide in, 459f CEP TCs and, 385, 385n11, 385n113 climate change and, 34 CMIP5 on, 209f CMT in, 217 coastal, 74 conditions for, 4f corals and, 416–17 droughts in, 29, 364, 364f, 436 with heat waves, 368–69

ENSO amplitude and, 154 asymmetry of, 157–63 ENSO asymmetry and, 143–44, 154 evolution of, 164–66 ENSO atmospheric teleconnections and, 312, 313f, 315–16, 323 in Indian Ocean, 319 ENSO low‐frequency modulation and, 173–76, 177, 181, 183 ENSO oceanic teleconnections and, 338, 341, 344, 345 ENSO phase locking and, 141–43 fisheries and, 430–32, 435–36 floods from, 365–68, 366f, 367f frequency of, 25–26 GHG and, 292, 298 global warming and, 58, 173, 190, 369–70 GMST and, 13 identified events of, 29 in Indian Ocean, 7, 368 ISV of, 364–65 kelps and, 421 La Niña into, 72 mangroves and, 422 name derivation of, 7 NA TCs and, 387–88 NDH in, 167 of 1982‐1983, 42–43, 71 prediction of, 228 of 1986‐1987, 204, 228 of 1997‐1998, 207, 235 carbon cycle in, 463–65 carbon dioxide in, 455–56 Niño‐3.4 Index for, 46f NIO TCs and, 390 NPSG and, 439 OTEs and, 413–16 PDO/IPO and, 186 phytoplankton and, 334 in Pliocene, 99 recent progress and challenges of, 10–12, 11f research oscillatory theory and, 51 seagrass in, 422 sea level extremes and, 410–12 SLP in, 26 SO and, 26 SOI and, 28f South Pacific TCs and, 391–92 SSTA in, 65, 66, 66f, 203, 215 in CMIP5, 214 paleoclimate and, 271 SST in, 3, 8, 9, 26–27, 52, 93, 274 of 2015‐2016, 45, 47f, 48f super, 29

INDEX  495

TCs and, 379, 381f, 382f teleconnections of, 362 temperature anomalies of, 50f thermocline in, 29, 30f TIWs in, 50 trade winds in, 8, 9, 15–16, 29 of 2015‐2016, 45, 46 of 2002‐2003, 65 of 2015‐2016, 45–50, 49f, 65, 238 carbon cycle in, 454, 465–66 drought in, 436 SST in, 45, 47f, 48f trade winds in, 45, 46 typical impacts of, 5f volcanoes and, 272, 274–75, 278–79, 282 Walker Circulation in, 249 westerly winds in, 74 WNP TCs and, 379–80 WWBs in, 51–52, 51f, 156–57, 207, 207f El Niño and the Southern Oscillation (ENSO). See also ENSO diversity amplitude of asymmetry of, 157–63, 158f, 160f, 161f El Niño and, 154 ENSO low‐frequency modulation and, 182 ENSO simple models for, 138–41, 212–13 asymmetry of, 153–67 of ENSO amplitude, 157–63, 158f, 160f, 161f of ENSO simple models, 143–44 evolution of, 163–66, 164f–66f, 167t atmospheric teleconnections of, 311–28, 313f, 316f–18f in Atlantic Ocean, 319–20 global warming and, 326–27 in Indian Ocean, 315, 319 in land, 320–26 in Maritime Continent, 320–21 in NA, 320 nonlinear, 314–15, 315f in South Atlantic Ocean, 320 CEP TCs and, 385 climate change and, 12–15, 33–34 ISV and, 474–77, 475f–77f paleoclimate reconstruction and, 473–81, 480f in climate system, 21–34 climate change from, 33–34 impacts of, 34

in CP, 29, 30–31 decadal timescales for, 12 in EP, 14, 29 equatorial dynamical processes of, 74–75 extreme climate impacts of, 362–70, 363f, 364f, 366f, 367f fisheries and, 334, 429–45 geographic distribution of, 16f GHG and, 13–14, 13f, 289–303 GMST in, 31 heat content and, 51–52, 51f historical background for, 7–10, 8f, 9f indices for, 15–17, 16f introduction to, 3–7 irregularity of, 154–57, 156f low‐frequency modulation of, 173–92, 188f ENSO model biases of, 214 external drivers of, 179–83, 180f, 182f GCM for, 176, 178–79 global warming and, 189–90 GMST and, 190, 191f PDO and, 179, 183–86, 185f perfect models for, 187, 189f prediction of, 187–89, 189f monsoons and, 370, 371f observations of, 41–59 data products for, 52–53 history of, 42–44, 44f in situ, 42–43, 44f oceanic and atmospheric reanalyses of, 53–57, 55f, 56f with satellites, 43–44, 52–53, 59, 181 variability of, 45–52, 46f–51f ocean extremes and, 409–23 corals and, 416–19, 419f, 420f kelps and, 419–21 mangroves and, 422 OTEs, 412–16, 414f, 415f seagrass and, 421–22 seal level extremes, 410–12, 411f shallow‐water marine ecosystems and, 416–22, 419f, 420f oceanic teleconnections and, 337–52 global warming and, 350–52, 351f in Indian Ocean, 347–49, 348f in Maritime Continent, 345–46 planetary waves and, 342–45, 346f in tropical Pacific, 338–42, 339f, 340f, 343f

past variability and paleoclimate reconstruction of, 87–104 Anthropocene and, 103 bivalves and, 91, 94t caves and, 93, 95t constraints on, 88–93 corals and, 88–91, 90f, 94t data citations for, 115t–18t in Holocene, 100–101 in interglacials, 99–100 lake sediments and, 92, 95t in last millennium, 101–3 in LGM, 99 marine archives of, 88–92, 94t marine sediments and, 91–92 Niño‐3.4 Index for, 89f in Pliocene, 98–99 quantitative approaches to, 93–98, 97f solar irradiance and, 101 terrestrial archives for, 92–93 tree rings and, 92–93, 95t volcanoes and, 101–2 prediction of, 227–41 of carbon cycle, 465–66, 466t, 467f decadal timescales in, 235–38 extreme climate impacts and, 370–71, 371f GHG and, 293–99, 294f, 295f history of, 227–31, 228f, 229f predictability in, 231–32 recent challenges of, 238–39, 239f, 240f skill of, 232–38, 234f, 236f, 237f recent progress and challenges of, 10–12, 11f remote forcing of, 249–61 in Atlantic Ocean, 254–57, 255f on Indian Ocean, 251–54, 252f, 253f Niño‐3.4 and, 259t, 260t PMM and, 257–59, 258f SOI and, 30 SSTA in, 121, 214, 298 asymmetry and, 144 CZ and, 122–23 phase locking and, 141–43 RO and BWJ and, 128 SST in, 10, 29–30, 293 RO and BWJ and, 135 SSTA and, 69, 79 TCs and, 377–97, 378f trade winds and, 10

496 INDEX

El Niño and the Southern Oscillation (ENSO) (cont’d) in tropical Pacific, 76, 290 oceanic teleconnections and, 338–42, 339f, 340f, 343f typical impacts of, 6f typical period of, 29 volcanoes and, 267–83 weather and, 334, 361–72 El Niño and the Southern Oscillation (ENSO), models for. See also specific models barrier layers in, 216 biases in, 209–13 GHG and, 299–301 challenges and opportunities of, 217–18 emergent constraints on, 214–15 ENSO diversity in, 208 evaluation of, 209–17, 211f, 215f external forcing on, 208–9, 213–14 harmonic oscillator, 203 hierarchy of, 202–5 history of, 201–2 improvement of, 215–17 simple, 121–47 for amplitude, 138–41 for amplitude, biases with, 212–13 for asymmetry, 143–44 coupled linear instability, 122–27, 123f, 125f, 126f for periodicity, 143, 144f for phase locking, 141–43, 142f for rectification, 144–46, 145f RO and BWJ, 127–38, 128t, 130f, 132f, 133t, 135f, 136f, 137t, 138t, 139f, 140f SSTA in, biases of, 213 TIWs in, 216–17 for understanding, 205–9, 206f, 207f, 209f volcanoes and, 273–81, 273t, 274f, 276f–77f, 279f, 280f WWBs in, 216 El Niño Modoki index (EMI), 81 TCs and, 381f EMICs. See Earth models of intermediate complexity Emile‐Geay, J., 96, 100, 101, 272, 275 Empirical normal modes, 203 Empirical orthogonal functions (EOFs), 69, 71–72, 411f of SSTA, 78 ENSO. See El Niño and the Southern Oscillation ENSO diversity, 65–82

CEP TCs and, 386 characteristics of, 67–74, 68t, 70f–73f climate change and, 77–78 in climate models, 78–78 in ENSO models, 208 global warming and, 208 indices of, 81–82 precursors and predictability of, 75–77 ENSO Longitude Index (ELI), 82 EOFs. See Empirical orthogonal functions EP. See Eastern Pacific EP‐CP index, 81 EP‐CP subsurface index method, 81 EPnew‐CPnew indices, 81 EPOCS. See Equatorial Pacific Ocean Climate Studies Equatorial cold tongue (ECT), 208 in CMIP5, 214 CMIP5 for, 210 ENSO low‐frequency modulation and, 183 in GCMs, 212 SSTA and, biases of, 213 SST and, 212, 214–15 Equatorial Pacific Ocean Climate Studies (EPOCS), 9–10 Equatorial Undercurrent (EUC), 338, 339–41, 342, 352 ERA‐20C, 57 ERS‐1 and 2, 43 ERSSTv5. See Extended Reconstructed SST dataset version 5 ESMs. See Earth system models EUC. See Equatorial Undercurrent Europe, ENSO atmospheric teleconnections in, 324–25 European Multi‐Model Seasonal‐to‐ Interannual Prediction project (EUROSIP), 230, 235 Ewel, K. C., 422 Exclusive economic zone (EEZ), 442 Experimental Long‐Lead Forecast Bulletin, TOGA and, 230 Extended Reconstructed SST dataset version 5 (ERSSTv5), of NOAA, 67, 67f Extreme climate impacts, of ENSO, 362–70, 363f, 364f, 366f, 367f prediction of, 370–71, 371f Fang, S.‐W., 258 Fedorov, A. V., 52, 124, 135, 394

Feely, R. A., 463 Fiedler, P., 437 Fisheries, 3, 41 CCE and, 436–38 in central North Pacific, 435–36 ENSO and, 334, 429–45 ENSO oceanic teleconnections and, 343 EP El Niño and, 430, 431f HCS and, 430–32 NPGO and, 438–40, 440f in NWP, 440–42, 441f in southwest Pacific, 442–44, 443f SST and, 14 of tuna, 432–35, 434f Floods, 3, 6f from El Niño, 365–68, 366f, 367f ENSO and, 334 fisheries and, 432 in La Niña, 29 from monsoons, 21 Floquet exponent analysis, 135 Flux adjustment/correction, 211 Food security, 3, 41 Ford, H. L., 99 Forecasting. See Prediction Fossil fuels, 13, 22, 34, 453, 458f, 465 Frappier, A., 93 Frauen, C., 314 Frequency entrainment, 100, 101 ENSO and climate change and, 473, 474, 478 LGM and, 99 Future projection (of ENSO), 214, 290, 301, 328, 394, 412 Gao, C., 269f, 272 General circulation models (GCMs), 78, 201–2. See also Coupled general circulation models AGCM, 205, 206, 210–11 on carbon cycle, 466 ECT in, 212 ENSO low‐frequency modulation and, 176, 178–79, 181, 183, 186, 188–89, 192 ENSO past variability and, 102, 103 global warming and, 369 for Indian Ocean, 100 OGCM, 205, 206, 207, 210–11 Geng, Licheng, 124 Geostrophic (as in inflow/transport/ current), 23, 26, 42, 75, 129, 296, 338, 348

INDEX  497

GFDL‐CM2.1, 77 ENSO low‐frequency modulation and, 177, 178f GHG and, 292 GFDL‐CM3, 79 GFDL‐ESM2M, 79 GFDL‐FLOR‐FA, 214 GHG. See Greenhouse gas Gill‐Matsuno response. See Matsuno‐ Gill response Gill model, CZ and, 122, 123 Glaciers LGM, 99 melting of, 4 climate change and, 34 volume of, 22 Global Carbon Budget, 461, 461f Global Carbon Project, 465 Global Forecast System, of NCEP, 57 Global mean surface temperature (GMST) carbon dioxide and, 22 climate change and, 34 El Niño and, 13 in ENSO, 31 ENSO low‐frequency modulation and, 190, 191f estimated changes to, 13f GHG and, 13–14, 13f PDO and, 33 RCPs and, 14f in super El Niño, 29 Global Ocean Observing System (GOOS), 42 Global Oceanographic Data Archaeology and Rescue Project (GODAR), 217–18 Global warming climate change and, 34 CP El Niño and, 77 El Niño and, 58, 173, 190, 369–70 ENSO amplitude asymmetry and, 163 ENSO atmospheric teleconnections and, 326–27 ENSO diversity and, 208 ENSO low‐frequency modulation and, 189–90 ENSO oceanic teleconnections and, 350–52, 351f GCMs and, 369 GHG and, 12–13, 291, 296, 302 Maritime Continent and, 291f PDO/IPO and, 186 GMST. See Global mean surface temperature

GODAR. See Global Oceanographic Data Archaeology and Rescue Project Goddard Institute for Space Studies Model E, 279 Godfrey, J. S., 338 GOOS. See Global Ocean Observing System GPP. See Gross primary productivity GRACE. See Gravity Recovery and Climate Experiment GraphEM, 97f, 98 Gravity Recovery and Climate Experiment (GRACE), 462, 463f Gray, W. M., 387 Greenhouse gas (GHG), 3, 77 climate change and, 34, 290–93, 291f ENSO and, 13–14, 13f, 289–303 ENSO atmospheric teleconnections and, 326–27 ENSO model biases and, 299–301 ENSO prediction and, 293–99, 294f, 295f from fossil fuels, 22 global warming and, 12–13 GMST and, 13–14, 13f SST and, 14–15, 298–99, 300f tropical Pacific and, 303f Greenland, ice sheets of, 22–23 climate change and, 34 Gross primary productivity (GPP), 454, 457–59 Ground‐truthing, for ENSO past variability, 103–4 Guirguis, K., 367 Ha, K.‐J., 253 Habitat loss, 4 HadCM2. See Hadley Centre coupled model version 2 HadGEM2‐CC, 79 HadGEM2‐ES, 79 Hadley Centre coupled model version 2 (HadCM2), 293 Hadley Circulation, 9f, 12, 22 ENSO atmospheric teleconnections and, 314 GHG and, 291–92 Halloran, P., 98–99 Halmahera Eddy (HE), 441f Halpern, D., 9 Ham, Y.‐G., 214, 255 Hamlington, B. D., 52 Handler, P., 268

Harmonic oscillatory models, 203 Harrison, D. E., 67, 82 Hayes, Stan, 43 HCS. See Humboldt Current System HE. See Halmahera Eddy Heat budget analysis, 75, 293 Heat content, ENSO and, 51–52, 51f Heat waves droughts with, 368–69 ENSO and, 334 MHWs, 413–16, 414f Hemer, M. A., 412 Hiatus (global warming), 12, 33, 189–90, 267, 292, 301, 347, 464–65 Highest Astronomical Tide, 410 Hill, K. L., 349 Hill, N. J., 443 Hirst, A. C., 138 Hobday, A., 415 Hoeke, R. K., 410 Holbrook, N. J., 349, 413 Holocene ENSO and climate change and, 474 ENSO low‐frequency modulation in, 176 ENSO past variability in, 100–101 Horel, J. D., 31 Hu, S., 52 Huffman, G. J., 24f Hughen, K. A., 100 Humboldt Current System (HCS), 430–32 Hurrell, J. W., 31, 479 Hurricanes. See also Tropical cyclones in Atlantic Ocean, 31 Hydrogen sulfide, from volcanoes, 268 Ice ages, 21 Ice‐albedo feedback, 23 Ice sheets. See also Antarctica; Arctic; Greenland in climate system, 22–23 iCESM. See Isotope‐enabled Community Earth System Model IFA. See individual foraminiferal analyses IGY. See International Geophysical Year Indian Ocean corals in, 418 El Niño in, 7, 368 ENSO atmospheric teleconnections in, 315, 319

498 INDEX

Indian Ocean (cont’d ) ENSO oceanic teleconnections in, 347–49, 348f ENSO prediction and, 238 ENSO remote forcing on, 251–54, 252f, 253f GCMs for, 100 mixed atmospheric‐oceanic teleconnections in, 350 NIO, TCs in, 389–90 south, TCs in, 392 SST in, 368 Walker Circulation in, 12 Indian Ocean Basin Mode (IOBM), 251–54 Indian Ocean basin warming (IOBW), 319 Indian Ocean Dipole (IOD), 251–54, 368 ENSO atmospheric teleconnections and, 315, 319 ENSO oceanic teleconnections and, 345–46 fisheries and, 435 MHWs and, 415–16 NIO TCs and, 389–90 OTEs and, 413 Indices of spatial shifts in atmospheric convection, 82 Individual foraminiferal analyses (IFA), 99, 100 Indonesian Throughflow (ITF), 23, 338–39, 345–49, 348f, 350–51 Industrial Revolution, 3–4, 13 Infrared (IR), with satellites, 52 INM_CM4, 79 Inoue, M., 228–30 Interdecadal Pacific Oscillation (IPO), 32 ENSO atmospheric teleconnections and, 327 ENSO low‐frequency modulation and, 184–86 GHG and, 292 PDO and, 33 TCs and, 379 Interglacials, 21 ENSO past variability in, 99–100 Intergovernmental Panel on Climate Change (IPCC), 12, 14 AR4 and AR5 of, 202 International Comprehensive Atmosphere Data Set, 52 International Geophysical Year (IGY), 7

International Research Institute for Climate and Society (IRI), 230 Intertropical Convergence Zone (ITCZ), 23–24, 25f Atlantic Ocean and, 255 in El Niño, of 2015‐2016, 45 ENSO atmospheric teleconnections and, 321, 324 in Atlantic Ocean, 320 ENSO diversity and, 78 ENSO low‐frequency modulation and, 173, 175, 183 ENSO models and, 206 biases of, 212 ENSO oceanic teleconnections and, 338 ENSO prediction and, 235 GHG and, 293, 297 SST and, 368 TCs and, 379 volcanoes and, 278–79, 282 Intraseasonal variability (ISV), 207, 207f of El Niño, 364–65 ENSO and climate change and, 474–77, 475f–77f IOBM. See Indian Ocean Basin Mode IOBW. See Indian Ocean basin warming IOD. See Indian Ocean Dipole IPCC. See Intergovernmental Panel on Climate Change IPO. See Interdecadal Pacific Oscillation IR. See Infrared IRI. See International Research Institute for Climate and Society Isotope‐enabled Community Earth System Model (iCESM), 98, 99–100 ISV. See Intraseasonal variability ITCZ. See Intertropical Convergence Zone ITF. See Indonesian Throughflow Jadhav, J., 66 Japan Agency for Marine Earth Science and Technology (JAMSTEC), 43, 58–59 Jet streams, 22 teleconnections and, 31 Jin, F.‐F., 124, 128, 129, 133, 138, 141, 143, 144, 155, 157, 158, 386, 394. See also Bjerknes‐Wyrtki‐ Jin Index

JN93, 128 Joanides, S., 99 Johnson, G. C., 338 Johnson, N. C., 82 Joint Typhoon Warning Center, U.S. (JTWC), 378f Jones, C. D., 456 JTWC. See Joint Typhoon Warning Center, U.S. Kao, H.‐Y., 66, 69–70 Karamperidou, C., 77, 477, 477f, 479 Karspeck, A. R., 177 KBF. See Kuroshio Bifurcation KE. See Kuroshio Extension Kelps, ocean extremes and, 419–21 Kelvin waves, 9 in Atlantic Ocean, 256 ENSO amplitude asymmetry and, 159, 162 ENSO atmospheric teleconnections and, 312 ENSO mixed atmospheric‐oceanic teleconnections and, 350 ENSO oceanic teleconnections and, 337, 338, 341–42, 343, 344, 346f in Indian Ocean, 253, 254, 368 MHWs and, 415, 416 RO and BWJ and, 127, 138 Rossby waves and, 50 sea level extremes and, 410–11 SSH and, 48f thermocline and, 42, 48f volcanoes and, 277–78 WWBs and, 156 Khodri, M., 208, 274–75, 277 Khon, V. C., 100–101 Kikuchi, K., 389 Kilduff, D. P., 440f Kim, H.‐M., 383 Kim, J.‐W., 165 Kim, S. T., 77, 257, 296 Kirtman, B. P., 207, 257 Klein, S. A., 368 Klotzbach, P. J., 388, 393 Knox, R., 9 Koren, G., 462 Kosaka, Y., 82 Kossin, J. P., 388, 393 Koutavas, A., 99 Kug, J.‐S., 66, 214 Kuleshov, Y., 392 Kuroshio Bifurcation (KBF), 441f, 442 Kuroshio Current, 441

INDEX  499

Kuroshio Extension (KE), 341, 441 fisheries and, 435 Lake sediments, ENSO past variability and, 92, 95t Land in climate system, 22 ENSO atmospheric teleconnections in, 320–26 temperature of, volcanoes and, 277–78 Landsea, C. W., 393 La Niña, 361 Atlantic Ocean and, 256 Bjerknes feedback in, 10 carbon cycle in, 464–65 climate change and, 34 conditions for, 4f corals and, 96, 417 CP El Niño and, 76 duration of, 26, 29 into El Niño, 72 ENSO amplitude and, asymmetry of, 157–63 ENSO asymmetry and, 143–44, 154 ENSO atmospheric teleconnections and, 312, 315, 323, 324, 327 in Indian Ocean, 319 ENSO evolution asymmetry and, 164–66 ENSO low‐frequency modulation and, 175, 177, 183 ENSO oceanic teleconnections and, 341, 342 ENSO phase locking and, 141–43 fisheries and, 432, 435 floods in, 29 GHG and, 298, 302 NA TCs and, 388 Niñgaloo Niño and, 14 Niño‐3.4 Index for, 46f NIO TCs and, 390 NPP and, 459 OTEs and, 413–16 PDO/IPO and, 186 recent progress and challenges of, 10–12, 11f seagrass in, 422 sea level extremes and, 410–12 SOI and, 28f South Pacific TCs and, 390–92 SSTA in, 66, 71, 215 GHG and, 297 SST in, 3, 14

TCs and, 381f teleconnections of, 362 thermocline in, 29, 30f, 98–99 TIWs in, 50 tornados and, 72 trade winds in, 16, 29 typical impacts of, 5f volcanoes and, 273, 275, 278–81, 282 WMO on, 230 WNP TCs and, 380, 381 WWBs in, 75–76 Larkin, N. K., 67 Larson, S. M., 207, 257 Last glacial maximum (LGM), 99 Last Millennium Reanalysis (LMR), 96–98, 97f Lau, N.‐C., 379 LD. See Linear dynamic deterministic feedbacks LDH. See Linear dynamic heating Lee, S.‐K., 72 Lee, T., 52, 444 Leeuwin Current, 416 Lenssen, N., 370 Levine, A. F. Z., 144, 157 LGM. See Last glacial maximum L’Heureux, M. L., 30 Li, J., 92, 272, 380 Li, T., 165 Li, Y., 78 Li, Z., 389 Lian, T., 396 Lifamatola Passage, 347 LIM. See Linear inverse model Lim, H.‐G., 275, 278 Lin, I. I., 381n8 Linear dynamic deterministic feedbacks (LD), 136–37 Linear dynamic heating (LDH), 129 Linear instability theory, 122 Linear inverse model (LIM), 102–3, 203–4 ENSO low‐frequency modulation and, 177 ENSO prediction and, 236 Linear noise (LN), 136–37 Liu, Q. Y., 347 LLWBCs. See Low‐latitude western boundary currents LMR. See Last Millennium Reanalysis LN. See Linear noise Low‐frequency modulation, of ENSO, 173–92, 188f ENSO model biases of, 214

external drivers of, 179–83, 180f, 182f GCM for, 176, 178–79 global warming and, 189–90 GMST and, 190, 191f PDO and, 179, 183–86, 185f perfect models for, 187, 189f prediction of, 187–89, 189f Low‐latitude western boundary currents (LLWBCs), 338, 340, 342, 350 Lu, B., 129 Luo, X., 459 Lyon, B., 364 MADA. See Monsoon Area Drought Atlas Madden Julian Oscillation (MJO), 75, 202, 365–67 CEP TCs and, 386 ENSO asymmetry and, 144 ENSO models and, 216 TCs and, 379, 383 of NA, 388–89 of NIO, 390 WWBs and, 157 Magee, A. D., 391 Maher, N., 274 Maluku Channel, 347 Mangroves, 422 Mann, M. E., 275 Mantua, N., 416 Marine archives, of ENSO past variability, 88–92, 94t Marine heatwaves (MHWs), 413–16, 414f Marine sediments, ENSO past variability and, 91–92 Maritime Continent climate change and, 298 ENSO atmospheric teleconnections in, 320–21 ENSO oceanic teleconnections in, 345–46 global warming and, 291f Indian and Pacific Oceans and, 251 SST in, 159, 295f volcanoes and, 277 Market squid, 438 Mason, I. M., 385 Matsuno‐Gill response (Gill‐Matsuno response), 278, 314 ENSO atmospheric teleconnections and, 325 in Atlantic Ocean, 319 MHWs and, 416

500 INDEX

MC. See Mindanao Current McClatchie, S., 438 McGregor, S., 164, 207, 271, 275, 276, 277 McInnes, K. L., 410 McPhaden, M. J., 52, 78, 444 ME. See Mindanao Eddy Mean High Water Springs, 410 Mean sea level pressure (MSLP), 292 MEI. See Multivariate ENSO Index Mentaschi, L., 412 Meridional advective feedback, 74 Met Office, 465–66 Meynecke, J.‐O., 444 MHWs. See Marine heatwaves Mindanao Current (MC), 339, 350, 441f Mindanao Eddy (ME), 441f MIROC‐ESM, 79, 274 Mixed atmospheric‐oceanic teleconnections, 349–50 MJO. See Madden Julian Oscillation MME. See Multimodel ensemble Models, for ENSO. See also specific models barrier layers in, 216 biases in, 209–13 GHG and, 299–301 challenges and opportunities of, 217–18 emergent constraints on, 214–15 ENSO diversity in, 208 evaluation of, 209–17, 211f, 215f external forcing on, 208–9, 213–14 harmonic oscillator, 203 hierarchy of, 202–5 history of, 201–2 improvement of, 215–17 simple, 121–47 for amplitude, 138–41 for amplitude, biases with, 212–13 for asymmetry, 143–44 coupled linear instability, 122–27, 123f, 125f, 126f for periodicity, 143, 144f for phase locking, 141–43, 142f for rectification, 144–46, 145f RO and BWJ, 127–38, 128t, 130f, 132f, 133t, 135f, 136f, 137t, 138t, 139f, 140f SSTA in, biases of, 213 TIWs in, 216–17 for understanding, 205–9, 206f, 207f, 209f volcanoes and, 273–81, 273t, 274f, 276f–77f, 279f, 280f WWBs in, 216

MODerate Resolution Imaging Spectrometer (MODIS), 459 Modoki. See Central Pacific; El Niño Modoki index Monsoon Area Drought Atlas (MADA), 92 Monsoons, 25f ENSO and, 370, 371f ENSO atmospheric teleconnections and, 319, 322 floods from, 21 SO and, 7 volcanoes and, 269, 278 Moon, I.‐J., 393 MRI‐CGCM3, 79 MSLP. See Mean sea level pressure Mt. Pinatubo volcano, 268–69, 275 Multimodel ensemble (MME), 163. See also North American Multi‐Model Ensemble Multivariate ENSO Index (MEI), 17, 30 for SST, 58f Muñoz, A. G., 367, 371 NA. See North Atlantic NADA. See North American Drought Atlas NAM. See Northern Annular Mode Namias, Jerome, 8 NAO. See North Atlantic Oscillation National Centers for Environmental Prediction (NCEP), 55f, 56 Climate Forecast System of, 257 Global Forecast System of, 57 National Hurricane Center (NHC), 378f National Oceanic and Atmospheric Administration, U.S. (NOAA) AVHRR of, 43 Climate Diagnostics Center of, 15 CPC of, 228–29 EPOCS of, 9–10 ERSSTv5 of, 67, 67f on NA TCs, 388 OISST of, 52, 66f, 166f, 236f, 239f ONI of, 15, 28f TAO array at, 43 NBP. See Net biosphere productivity NCAR‐CCSM4, 79 NCAR‐CESM‐LE, 214 NCEP. See National Centers for Environmental Prediction NCT–NWP indices, 81 NDH. See Nonlinear dynamical heating

NDVI. See Normalized Difference Vegetation Index NEC. See North Equatorial Current NECC. See North Equatorial Countercurrent NEE. See Net ecosystem exchange Neelin, J. D., 143, 157, 364 NEP. See Net ecosystem productivity Neske, S., 207 Net biosphere productivity (NBP), 457 Net ecosystem exchange (NEE), 457 Net ecosystem productivity (NEP), 457 Net primary productivity (NPP), 457–58, 460f SST and, 459 New Guinea Coastal Undercurrent, 342, 350 New Guinea Counter Current (NGCC), 441f Newman, M., 78, 214, 238 Newtonian damping, volcanoes and, 277 New Zealand, ENSO atmospheric teleconnections in, 321–22 NGCC. See New Guinea Counter Current NHC. See National Hurricane Center Nicholls, N., 390, 392 Niñgaloo Niño, 14 Niño‐3, 69, 73f, 81 ENSO amplitude asymmetry and, 163 ENSO low‐frequency modulation and, 175, 188f RO and BWJ and, 129 SSTA in, 74, 210 thermocline and, 186 Niño‐3.4, 46f, 51 ENSO past variability and, 89f, 102 ENSO prediction and, 230 ENSO remote forcing and, 259t, 260t OTEs and, 413 PMIP3 and, 102f SSTA in, 66 SST in, 238 Niño‐4, 69, 73f, 81 SSTA in, 74, 238–39 SST in, 93 NIO. See North Indian Ocean NMME. See North American Multi‐Model Ensemble NOAA. See National Oceanic and Atmospheric Administration, U.S.

INDEX  501

Nonlinear dynamical heating (NDH), 137 in El Niño, 167 ENSO amplitude and, 140 ENSO amplitude asymmetry and, 157–58 ENSO asymmetry and, 144 NorESM. See Norwegian Earth System Model Normalized Difference Vegetation Index (NDVI), 459 NORPAX. See North Pacific Experiment North America, ENSO atmospheric teleconnections in, 323 North American Drought Atlas (NADA), 92 North American Multi‐Model Ensemble (NMME), 76, 235, 236 ENSO remote forcing and, 257 North Atlantic (NA) ENSO atmospheric teleconnections in, 320 SST in, 32–33 TCs in, 386–89, 387f North Atlantic Oscillation (NAO), 32–33 ENSO atmospheric teleconnections and, 320 remote forcing on, 250 North Equatorial Countercurrent (NECC), 23, 338, 441f North Equatorial Current (NEC), 23, 338, 340, 440, 441f ENSO oceanic teleconnections and, 341 fisheries and, 435 Northern Annular Mode (NAM), 32 North Indian Ocean (NIO), TCs in, 389–90 North Pacific mixed atmospheric‐oceanic teleconnections in, 349 sea level extremes in, 412 North Pacific Experiment (NORPAX), 8–10 North Pacific Gyre Oscillation (NPGO), 439–40, 440f North Pacific Index (NPI), 32 North Pacific Meridional Mode (NPMM), 79 in ENSO prediction, 231 PDO/IPO and, 186 North Pacific Oscillation (NPO)

ENSO remote forcing and, 257 PDO/IPO and, 186 trade winds and, 76 North Pacific Subtropical Gyre, 438–40 North Tropical Atlantic (NTA), 254–56 Northwest Pacific (NWP) ENSO atmospheric teleconnections in, 322 fisheries in, 440–42, 441f Norwegian Earth System Model (NorESM), 274, 278, 279 NPGO. See North Pacific Gyre Oscillation NPI. See North Pacific Index NPMM. See North Pacific Meridional Mode NPO. See North Pacific Oscillation NPP. See Net primary productivity NTA. See North Tropical Atlantic NWP. See Northwest Pacific OAFLUX, 52 O’Brien, J. J., 228–30 Observations, ENSO with satellites, 43–44, 52–53 for ENSO low‐frequency modulation, 181 in situ and, 44 of tropical Pacific, 59 in situ, 42–43, 44f Ocean extremes corals and, 416–19, 419f, 420f ENSO and, 409–23 kelps and, 419–21 mangroves and, 422 OTEs, 412–16, 414f, 415f seagrass and, 421–22 seal level extremes, 410–12, 411f shallow‐water marine ecosystems and, 416–22, 419f, 420f Ocean heat content (OHC), 58f TCs and, 378, 381, 383 Oceanic Niño Index (ONI), of NOAA, 15, 28f Oceanic teleconnections, 337–52 global warming and, 350–52, 351f in Indian Ocean, 347–49, 348f in Maritime Continent, 345–46 planetary waves and, 342–45, 346f in tropical Pacific, 338–42, 339f, 340f, 343f Ocean‐only GCM (OGCM), 205, 206, 207, 210–11

Ocean reanalyses (ORAs), 53–57, 55f, 56f Ocean Reanalyses Intercomparison Project, 54 Oceans. See also specific oceans carbon cycle and, 462–63 in climate system, 22 Ocean temperature extremes (OTEs), 412–16, 414f, 415f OGCM. See Ocean‐only GCM Ohba, M., 274, 277 OHC. See Ocean heat content OISST. See Optimum Interpolation SST Oliver, E. C. J., 388 OLR. See Outgoing Longwave Radiation Oman, L., 279 OMZs. See Oxygen minimum zones ONI. See Oceanic Niño Index Operational forecasts, 228 Optimum Interpolation SST (OISST), of NOAA, 52, 66f, 166f, 236f, 239f ORA‐20C, 57 ORAs. See Ocean reanalyses ORAS4, 58f OTEs. See Ocean temperature extremes Outgoing Longwave Radiation (OLR), 73–74, 82, 238 Oxygen minimum zones (OMZs), 345 Pacific Decadal Oscillation (PDO), 12, 32 CEP TCs and, 386 CP El Niño and, 78 El Niño and, 365 ENSO atmospheric teleconnections and, 323 ENSO low‐frequency modulation and, 179, 183–86, 185f fisheries and, 430, 435 IPO and, 33 NPGO and, 439–40, 440f NWP and, 441 OTEs and, 413 TCs and, 379, 383 Pacific Meridional Mode (PMM), 255 ENSO low‐frequency modulation and, 179 ENSO remote forcings and, 257–59, 258f SSTA and, 76 TCs and, 383

502 INDEX

Pacific North American (PNA) pattern, 31, 33 ENSO atmospheric teleconnections and, 313, 313f in Atlantic Ocean, 319 in NA, 320 Pacific Ocean. See specific regions Pacific Ocean Observing System, 59 Pacific South American (PSA) pattern, 313–14, 313f, 324 Page, S. E., 460 PAGES 2k Consortium, 97f, 98 Paleoclimate Modeling Intercomparison Project (PMIP), 99, 100 CMIP and, 208 Paleoclimate Modeling Intercomparison Project 3 (PMIP3), 102, 103 Holocene and, 100–101 volcanoes and, 271 Paleoclimate reconstruction, ENSO past variability and, 87–88 Anthropocene and, 103 bivalves and, 91, 94t caves and, 93, 95t climate change and, 473–81, 475f–77f, 480f constraints on, 88–93 corals and, 88–91, 90f, 94t data citations for, 115t–18t in Holocene, 100–101 in interglacials, 99–100 lake sediments and, 92, 95t in last millennium, 101–3 in LGM, 99 marine archives of, 88–92, 94t marine sediments and, 91–92 Niño‐3.4 Index for, 89f in Pliocene, 98–99 quantitative approaches to, 93–98, 97f solar irradiance and, 101 terrestrial archives for, 92–93 tree rings and, 92–93, 95t volcanoes and, 101–2, 271–73, 272t Palmer Drought Severity Index (PDSI), 364 Paris Agreement, 302 Park, J. H., 255 Passive microwave (PMW), 43 for SST, 52 Patricola, C. M., 82, 380f, 383 Pausata, F. S. R., 280, 282 PCs. See Principal components PDO. See Pacific Decadal Oscillation

PDSI. See Palmer Drought Severity Index Perfect models, for ENSO low‐ frequency modulation, 187, 189f Periodicity, ENSO simple models for, 143, 144f Phase locking, 141–43, 142f Philander, S. G. H., 124, 338 Physical nulls, for ENSO past variability, 104 Phytoplankton El Niño and, 334 ENSO amplitude asymmetry and, 159 PMIP3, Niño‐3.4 and, 102f Planetary waves, 342–45, 346f Plankton. See also Phytoplankton NPSG and, 439 Planton, Y., 207 Pliocene, ENSO past variability in, 98–99 PMIP. See Paleoclimate Modeling Intercomparison Project PMIP3. See Paleoclimate Modeling Intercomparison Project 3 PMM. See Pacific Meridional Mode PMW. See Passive microwave PNA. See Pacific North American pattern Porites, 98, 100, 419 Power, S. B., 302, 326 Prediction, of ENSO, 227–41 of carbon cycle, 465–66, 466t, 467f decadal timescales in, 235–38 extreme climate impacts and, 370–71, 371f GHG and, 293–99, 294f, 295f history of, 227–31, 228f, 229f predictability in, 231–32 recent challenges of, 238–39, 239f, 240f skill of, 232–38, 234f, 236f, 237f Predybaylo, E., 77, 275, 277, 278 Preethi, B., 325 Principal components (PCs) GHG and, 298–99 for SSTA, 69, 81 Process modeling, 296–99 for ENSO past variability, 104 Proxy system modeling (PSM), 104 PSA. See Pacific South American pattern PSM. See Proxy system modeling Putrasahan, D., 350 Pycnocline

ENSO oceanic teleconnections and, 338 EUC in, 341 LLWBCs and, 340 wind stress and, 350 Qu, T., 81–82 Quelccaya ice core, 93 QuikSCAT, 43 Quinn, William, 227 Rajagopalan, B., 479 Ramsay, H. A., 391 Rasmusson, E. M., 9, 71 Ratnam, J. V., 325 RCP4.5, 77 RCP8.5, 77 RCPs. See Representative concentration pathways Recharge/discharge oscillator model, 203 in EMICs, 204 Recharge oscillator (RO), 51, 122 BWJ and, 127–38, 128t, 130f, 132f, 133t, 135f, 136f, 137t, 138t DO and, 137–38 ENSO amplitude asymmetry and, 162–63 ENSO asymmetry and, 143–44 ENSO low‐frequency modulation and, 176 ENSO phase locking and, 143 Rectification, ENSO simple models for, 144–46, 145f Relative SST, 387, 387n15 Remote forcing, of ENSO, 249–61 in Atlantic Ocean, 254–57, 255f in Indian Ocean, 251–54, 252f, 253f Niño‐3.4 and, 259t, 260t PMM and, 257–59, 258f Ren, H.‐L., 76 Replication, for ENSO past variability, 104 Representative concentration pathways (RCPs), GMST and, 14f Respiration, 453–54, 457–59, 462, 464, 467 Reynolds, R. W., 7f, 11f, 48f, 52 Riascos, J. M., 422 Rickaby, R. E. M., 98–99 Rifai, S. W., 458 RO. See Recharge oscillator Rodrigues, R. R., 320 Rossby waves, 31

INDEX  503

ENSO amplitude asymmetry and, 159 ENSO atmospheric teleconnections and, 312–13, 319 ENSO oceanic teleconnections and, 337, 341–42, 344 ENSO remote forcing and, 250, 257 in Indian Ocean, 368 Kelvin waves and, 50 MHWs and, 415, 416 OLR and, 73–74 PDO/IPO and, 186 RO and BWJ and, 138 sea level extremes and, 410 SSH and, 50 Roulston, M., 157 Rowell, D. P., 325 SACZ. See South Atlantic Convergence Zone Sadekov, A. Y., 99 SAM. See Southern Annular Mode Santoso, A., 77 SAOD. See Stratospheric aerosol optical depth Sardesmukh, P. D., 76 Sardine, 438 SASD. See South Atlantic Subtropical Dipole Satellites ENSO observations with, 43–44, 52–53 for ENSO low‐frequency modulation, 181 in situ and, 44 of tropical Pacific, 59 IR with, 52 Schouten, M. W., 350 SEA. See Superposed Epoch Analysis Seagrass, 421–22 Sea level anomalies (SLAs), 411f Sea level extremes, 410–12, 411f Sea level pressure (SLP) in El Niño, 26 ENSO atmospheric teleconnections and, 316 in NA, 320 MSLP, 292 SO and, 8f, 26 SOI and, 28f in tropical Pacific, 68 westerly winds and, 69 SEAPODYM, 433, 435 SEAS5, 55, 56f Seasonal footprinting mechanism (SFM), 257, 258

Sea surface height (SSH), 43 ECMWF and, 76 in El Niño, of 2015‐2016, 48f, 49f ENSO low‐frequency modulation and, 188f Kelvin waves and, 48f Rossby waves and, 50 thermocline and, 68, 186 volcanoes and, 274 Sea surface salinity (SSS), 43 in El Niño, of 2015‐2016, 45, 48f, 49f indices, 81–82 satellite observation of, 52 Sea surface temperature (SST), 24f anomalies in, 7f in Atlantic Ocean, 254, 256 atmospheric bridges and, 368 AVHRR for, 43 carbon cycle and, 455, 458f, 463 CEP TCs and, 385n112, 386 in climate system, 23 CMIP5 for, 211 Cold Tongue Index for, 17 convective thresholds of, 82 corals and, 91 in CP El Niño, 75, 76–77 ECMWF and, 76 ECT and, 212, 214–15 in El Niño, 3, 8, 9, 26–27, 52, 93, 274 of 2015‐2016, 45, 47f, 48f in EMICs, 204 ENSO and, 10, 11f, 29–30, 293 RO and BWJ and, 135 SSTA and, 69 ENSO atmospheric teleconnections and, 312, 314 in Atlantic Ocean, 319–20 ENSO CZ and, 146 ENSO evolution asymmetry and, 164 ENSO low‐frequency modulation and, 173, 174f, 176, 179, 183 ENSO oceanic teleconnections and, 344 ENSO prediction and, 231, 236–38 ENSO stochastic forcing and, 155, 156 in EP convective thresholds in, 74 El Niño in, 74, 75, 76–77 fisheries and, 14, 435, 436 GHG and, 14–15, 292, 298–99, 300f in Indian Ocean, 368 in La Niña, 3, 14

marine sediments and, 92 in Maritime Continent, 159, 295f in MEI, 30 MHWs and, 416 Multivariate ENSO Index for, 58f in NA, 32–33 TCs in, 388 in Niño‐3.4, 238 NORPAX and, 8 NPP and, 458–59 NPSG and, 439 OISST, 52, 66f, 166f, 236f, 239f OTEs and, 413 PMW for, 52 relative, 387, 387n15 remote forcing on, 250 SOI and, 16f in super El Niño, 29 TCs and, 378 teleconnections and, 31 threshold of, 75 TNI for, 69 trade winds and, 25 in tropical Pacific, 32f ENSO and climate change and, 474 volcanoes and, 267, 268, 269f, 273, 274, 275, 281t, 282 WES, 76, 78, 255, 257, 320 WWBs and, 157 Sea surface temperature anomaly (SSTA) in Atlantic Ocean, 254, 255 in CP El Niño, 67, 71, 74, 80f, 207–8 ECT and, biases of, 213 in El Niño, 65, 66, 66f, 203, 215 in CMIP5, 214 paleoclimate and, 271 in ENSO, 121, 214, 298 asymmetry and, 144 CZ and, 122–23 phase locking and, 141, 143 RO and BWJ and, 128 SST in, 69, 79 ENSO amplitude asymmetry and, 157–59, 161, 162, 163 ENSO CZ and, 124, 127, 146 ENSO evolution asymmetry and, 165 ENSO low‐frequency modulation and, 179, 180, 183, 187 ENSO model biases and, 213 EOFs of, 78 in EP El Niño, 67, 69–71, 74, 207–8 GMST and, 190, 191f

504 INDEX

Sea surface temperature anomaly (SSTA) (cont’d ) in La Niña, 66, 71, 215 GHG and, 297 in Niño‐3, 74, 210 in Niño‐4, 74, 238–39 PCs for, 69, 81 PMM and, 76, 257 in recharge/discharge oscillator model, 203 remote forcing on, 250 RO and BWJ and, 127, 131 SPMM and, 76 in tropical Pacific, 208 volcanoes and, 277, 279f WWV and, 75 ZA and, 74 SEC. See South Equatorial Current Sen Gupta, A. A., 350 SFM. See Seasonal footprinting mechanism SG. See Subgrid‐scale contributions Shallow‐water marine ecosystems, ocean extremes and, 416–22, 419f, 420f Sigl, M., 101 Simple Ocean Data Assimilation (SODA), 133, 135, 188f, 346 RO and BWJ and, 129, 131 Singh, A., 81 SINTEX‐F, 236 SLAs. See Sea level anomalies SLP. See Sea level pressure Smelly reef (taimasa), 418 Smith, T. M., 24f SMOS. See Soil Moisture Ocean Salinity SO. See Southern Oscillation Sobel, A. H., 364, 382, 394 SODA. See Simple Ocean Data Assimilation Soden, B. J., 388 SOI. See Southern Oscillation Index Soil Moisture Ocean Salinity (SMOS), 48f Solar irradiance in climate system, 22 ENSO past variability and, 101 Solberg, H., 122 South America, ENSO atmospheric teleconnections in, 323–24 South Atlantic Convergence Zone (SACZ), 324 South Atlantic Subtropical Dipole (SASD), 320

South Equatorial Current (SEC), 23, 338, 340 ENSO oceanic teleconnections and, 341 Southern Annular Mode (SAM), 32, 33 ENSO atmospheric teleconnections and, 313 MJO and, 367 Southern Oscillation (SO), 362. See also El Niño and the Southern Oscillation El Niño and, 26 monsoons and, 7 SLP with, 8f, 26 Southern Oscillation Index (SOI), 15, 16f, 26, 27f ENSO and, 30 ENSO oceanic teleconnections and, 344 ENSO prediction and, 230, 238 GHG and, 292 seagrass and, 421 SLP and, 28f tree rings and, 92 WNP TCs and, 379 South Indian Ocean, TCs in, 392 South Pacific, TCs in, 390–92, 391f South Pacific Convergence Zone (SPCZ), 23, 25f in El Niño, of 2015‐2016, 45 ENSO atmospheric teleconnections and, 321 ENSO model biases of, 212 GHG and, 297 sea level extremes and, 412 South Pacific TCs and, 390–91 South Pacific Meridional Mode (SPMM), 79 in ENSO prediction, 232 SSTA and, 76 South Pacific Ocean, mixed atmospheric‐oceanic teleconnections in, 349–50 Southwest Pacific, fisheries in, 442–44, 443f Spatio-temporal indices, 82 SPCZ. See South Pacific Convergence Zone Species extinctions, 4 SPMM. See South Pacific Meridional Mode Spring predictability barrier, 231–32, 480 Sprintall, J., 347 SSH. See Sea surface height

SSS. See Sea surface salinity SST. See Sea surface temperature SSTA. See Sea surface temperature anomaly Stammer, D., 53 State dependent noise, 11, 141, 145, 162, 176–77 Stefan‐Boltzmann Law, 22 Stein, K., 153, 155 Stepaniak, D. P., 21, 69 Stevenson, S., 102, 274, 275–76, 277 Stochastic forcing, ENSO irregularity and, 155–56 Storm surges sea level extremes and, 410–12 in TCs, 396, 436 Stratospheric aerosol optical depth (SAOD), 190 Stuecker, M. F., 322 Subgrid‐scale contributions (SG), 129, 137 Sulfur dioxide, from volcanoes, 268, 271 Sullivan, A., 68f, 69, 71f Sun. See Solar irradiance Sun, Y., 163 Sunspot cycle, 34 Super El Niños, 29 Superposed Epoch Analysis (SEA), 101 Surface air temperatures, 4 Surface atmospheric pressure, 7 Sverdrup balance, 129 Sverdrup equation, 23 Sverdrup transport, 212 Taimasa (smelly reef), 418 Takahashi K., 72f, 73f, 76, 161 Tam, C.‐Y., 383 Tanaka, K., 349 TAO. See Tropical Atmosphere‐Ocean TAO/TRITON, 43, 45, 49f, 50f, 52 Tardif, R., 98 Tartaglione, C. A., 388 TCs. See Tropical cyclones TCWV. See Total column water vapor TD. See Thermodynamic damping; Tropical depression TDH. See Thermodynamic heating Teleconnections. See also Atmospheric teleconnections; Oceanic teleconnections in climate system, 31–33, 32f of El Niño, 362 ENSO past variability and, 102 of La Niña, 362

INDEX  505

Terrestrial archives, for ENSO past variability, 92–93, 95t Terrestrial water storage (TWS), 462, 463f Thermocline CEP TCs and, 385, 385n10 in CP El Niño, 75 in El Niño, 29, 30f in EMICs, 204 ENSO diversity and, 78 ENSO stochastic forcing and, 155 in EP El Niño, 75 GHG and, 291, 296 HCS and, 430 kelps and, 421 Kelvin waves and, 42, 48f in La Niña, 29, 30f, 98–99 Niño‐3 and, 186 RO and BWJ and, 129 SSH and, 68, 186 in western Pacific, 50 Thermocline feedback, 74, 75 ENSO CZ and, 124 ENSO model biases of, 212 Thermodynamic damping (TD), 129, 135 ENSO CZ and, 127 Thermodynamic heating (TDH), 129 Thirumalai, K., 99, 100 Thompson, D. M., 93 Tierney, J. E., 272 Timlin, M. S., 30, 57, 58f Timmermann, A., 30, 209, 275, 276, 277, 293, 477f Timor Passage, 347 Tippett, M. K., 234 TIWs. See Tropical instability waves TNI. See Trans Niño Index TOGA. See Tropical Ocean Global Atmosphere Tomczak, M., 338 Tornados, La Niña and, 72 Total column water vapor (TCWV), 23, 24f TPOS2020. See Tropical Pacific Observing System 2020 Trade wind charging (TWC), 76 Trade winds in El Niño, 8, 9, 15–16, 29 of 2015‐2016, 45, 46 ENSO and, 10 GHG and, 292, 293 in La Niña, 16, 29 MHWs and, 416 NPO and, 76 SST and, 25

in tropical Pacific, 23 Transition Zone Chlorophyll Front (TZCF), 435–36, 441 Trans Niño Index (TNI), 26–27, 31, 81 GHG and, 299 for SST, 69 Tree rings ENSO past variability and, 92–93, 95t volcanoes and, 102 Trenberth, K. E., 21, 28f, 31, 32, 69, 334, 479 Triangle Transocean Buoy Network (TRITON), 43, 59 TRMM. See Tropical Rainfall Measurement Mission Tropflux, 52 Tropical Atmosphere‐Ocean (TAO) array, 25f, 43 Tropical cyclones (TCs), 377n1 in CEP, 383–86, 387f climate change and, 392–94, 395f ENSO and, 377–97, 378f in NA, 386–89, 387f in NIO, 389–90 seagrass and, 421 sea level extremes and, 411–12 in South Indian Ocean, 392 in South Pacific, 390–92, 391f storm surges in, 396, 436 in WNP, 379–83, 380f–82f Tropical depression (TD), 377n1 Tropical instability waves (TIWs), 50 ENSO amplitude and, 140 ENSO amplitude asymmetry and, 158 ENSO asymmetry and, 144 in ENSO models, 216–17 ENSO oceanic teleconnections and, 340, 341 Tropical Ocean Global Atmosphere (TOGA), 10, 53, 201, 228 Experimental Long‐Lead Forecast Bulletin and, 230 Tropical Pacific in climate system, 23–25, 26f ENSO in, 76, 290 ENSO oceanic teleconnections in, 338–42, 339f, 340f, 343f GHG and, 303f satellite observations of, 59 sea level extremes in, 410 SLP in, 68 SSTA in, 208 SST in, 32f

ENSO and climate change and, 474 Tropical Pacific Observing System 2020 (TPOS2020), 217 Tropical Rainfall Measurement Mission (TRMM), 48f Tropical storms (TCs), 31, 377n1 Tropospheric Biennial Oscillation, 250 Tropospheric temperature mechanism, 312 Tudhope, A. W., 99, 100 Tuna, 432–35, 434f TWC. See Trade wind charging 20th‐Century Reanalysis Project, 57 TWS. See Terrestrial water storage Typhoons. See Tropical cyclones TZCF. See Transition Zone Chlorophyll Front Tziperman, E., 141 Van Gorder, S., 254 van Oldenborgh, G. J., 234 van Sebille, E., 347 Vertical advective feedback, 74 Vertical wind shear (VWS) CEP TCs and, 386 NIO TCs and, 389 TCs and, 378 Vimont, D. J., 250, 388 Volcanoes, 21 aerosols from, 9, 268–71, 279 Atlantic ocean and, 280–81 climate change and, 34–35, 268–71, 269f, 270f cloud albedo and, 277 El Chichón, 9, 43, 228 El Niño and, 272 ENSO and, 267–83 ENSO models and, 273–81, 273t, 274f, 276f–77f, 279f, 280f ENSO past variability and, 101–2 extratopical, 278–81 land temperature and, 277–78 La Niña and, 273 Maritime Continent and, 277 Mt. Pinatubo, 268–69, 275 paleoclimate reconstruction and, 271–73, 272t SST and, 267, 268, 269f, 273, 274, 275, 281t, 282 VolMIP, 283 VWS. See Vertical wind shear Wahiduzzaman, M., 390 WAIS. See West Antarctic Ice Sheet Walker, Gilbert, 7–8, 22, 227, 362

506 INDEX

Walker Circulation, 8, 9f cold tongue and, 77 in El Niño, 249 ENSO atmospheric teleconnections and, 312, 313f, 314, 319, 328 in Atlantic Ocean, 319 ENSO low‐frequency modulation and, 182 ENSO remote forcing and, 261 GHG and, 291 in Indian Ocean, 12 ITCZ and, 23 MHWs and, 415–16 NA TCs and, 387 Wallace, J. M., 31 Wang, B., 379, 382, 390 Wang, G., 302 Wang, S.‐Y., 76 Wara, M. W., 99 Warm Core Ring (WCR), 441f Warm water volume (WWV) in ENSO prediction, 231, 232 SSTA and, 75 Watanabe, T., 99 WCR. See Warm Core Ring WCRP. See World Climate Research Program Weather, 22 ENSO and, 334, 361–72 Wenzel, S., 466 WES. See Wind‐evaporation‐SST West Antarctic Ice Sheet (WAIS), 23 Westerly wind bursts (WWBs), 10, 202 in CP El Niño, 75, 76 in El Niño, 51–52, 51f, 156–57, 207, 207f of 2015‐2016, 46–48 in ENSO, 50 ENSO amplitude and, 140 ENSO amplitude asymmetry and, 160 ENSO asymmetry and, 144, 154 ENSO irregularity and, 156–57

in ENSO models, 216 in EP El Niño, 75 ISV of, 207, 207f in La Niña, 75–76 Westerly winds in El Niño, 74 in EP El Niño, 69, 70f SAM and, 33 SLP and, 69 Western Hemisphere Warm Pool (WHWP), 255, 255f, 256 Western North Pacific (WNPAC), 141 TCs in, 379–83, 380f–82f Western Pacific, thermocline in, 50 Western Pacific Warm Pool, 8, 9f, 50, 164, 277, 293, 415 WHWP. See Western Hemisphere Warm Pool Wildfires, 3, 6f carbon dioxide and, 459–61, 461f Williams, I. N., 82 Wind‐evaporation‐SST (WES), 76, 78, 255, 257, 320 Wind-evaporation-SST feedback, 76, 78, 232, 255, 257, 312, 320 Wind stress, 75–76 in CMIP5, 214 in EMICs, 204 ENSO amplitude asymmetry and, 159 ENSO CZ and, 127 ENSO low‐frequency modulation and, 183 ENSO model biases for, 210, 212 ENSO oceanic teleconnections and, 341 pycnocline and, 350 RO and BWJ and, 131–33 volcanoes and, 275 Wittenberg, A. T., 178, 187–88, 216 WMO. See World Meteorological Organization WNP. See Western North Pacific

WNPAC. See Western North Pacific Wolter, K., 30, 57, 58f World Climate Research Program (WCRP), 10 World Meteorological Organization (WMO), 42 on La Niña, 230 World Weather Watch (WWW), 42 Wu, G., 379 Wu, Y.‐K., 383 Wunsch, C., 479 WWBs. See Westerly wind bursts WWV. See Warm water volume WWW. See World Weather Watch Wyrtki, K., 9, 42, 43, 121, 122, 127, 138. See also Bjerknes‐Wyrtki‐ Jin Index Wyrtki frequency, 129 Xie, R., 124 Xie, S. P., 277 Yamagata, T., 341 Yee, 444 Yeh, S.‐W., 31, 299 Yip, C. K. M., 382 Yu, J.‐H., 380 Yu, J.‐Y., 66, 69–70, 77, 78, 81–82, 258 Yuan, J., 383 ZA. See Zonal advective feedback Zebiak, S. E., 127, 138, 177. See also Cane‐Zebiak model Zhang, H., 258–59 Zhang, S., 56 Zhang, T., 163 Zheng, X., 390 Zheng, Z.‐W., 381, 383 Zhou, X., 382 Zonal advective feedback (ZA), 74, 75, 129 ENSO model biases of, 212 Zonal strip models, in EMICs, 204