Techniques for disaster risk management and mitigation 9781119359173, 1119359171, 9781119359203, 1119359201

Citation preview

Techniques for Disaster Risk Management and Mitigation

Techniques for Disaster Risk Management and Mitigation Edited by Prashant K. Srivastava Sudhir Kumar Singh U. C. Mohanty Tad Murty

This edition first published 2020 © 2020 John Wiley & Sons, Inc. 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. The right of Prashant K. Srivastava, Sudhir Kumar Singh, U. C. Mohanty, and Tad Murty to be identified as the authors of the editorial material in this work has been asserted in accordance with law. Registered Office John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA Editorial Office Boschstr. 12, 69469 Weinheim, Germany For details of our global editorial offices, customer services, and more information about Wiley products, visit us at www.wiley.com. Wiley also publishes its books in a variety of electronic formats and by print‐on‐demand. Some content that appears in standard print versions of this book may not be available in other formats. 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. 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. 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. Library of Congress Cataloging‐in‐Publication data is available. Hardback ISBN: 9781119359180 Cover Design: Wiley Cover Images: © AC Rider/Shutterstock, © Peter J. Wilson/Shutterstock, © Naypong Studio/Shutterstock, © Lucy Brown–loca4motion/Shutterstock Set in 10/12 and Times New Roman by SPi Global, Pondicherry, India Printed in the United States of America 10 9 8 7 6 5 4 3 2 1

Late Prof. Tad Murty

Finally, after several ups and downs, I am about to submit this book with help of coeditors. I had been writing the acknowledgement back and forth, wherein I expressed immense gratitude toward my parents, beloved wife, and kids. However, I felt that was not enough. I started this book with esteemed Professor Tad Murty who passed away in 2018. He was an inspiration, being humble, helpful, and persistent even in difficult times. This book would be incomplete without mentioning his dedication and perseverance towards his work. It is noteworthy that Professor Murty was an IndianCanadian oceanographer and an expert on tsunamis. He was the former president of the Tsunami Society. He was an adjunct professor in the Department of Civil Engineering and Earth Sciences at the University of Ottawa. Professor Murty held a PhD degree in oceanography and meteorology from the University of Chicago. He was coeditor of Springer's journal, Natural Hazards, a renowned journal in the field. He took part in a review of the 2007 Intergovernmental Panel on Climate Change. Professor Murty characterized

himself as a global warming skeptic. In a 17 August 2006 interview, he stated, “I started with a firm belief about global warming, until I started working on it myself....I switched to the other side in the early 1990s when Fisheries and Oceans Canada asked me to prepare a position paper and I started to look into the problem seriously.” He mentioned that “when natural disasters strike, there is more loss of life and more loss of materials in the developing world. Is it because there are more people here? Or, is it because the developing world is not as prepared as the developed world?”. Hence, advanced techniques are needed to combat natural disaster; then we planned and started this book. With his unfortunate death, I really miss having scientific discussions with him; and even more now as the book is completed, and I wish he could have been here with us. May God rest his soul in peace; he will be forever in our hearts. —Prashant K. Srivastava

CONTENTS Contributors����������������������������������������������������������������������������������������������������������������������������������������������������������ix Preface����������������������������������������������������������������������������������������������������������������������������������������������������������������xiii

Section I: Introduction 1. Concepts and Methodologies of Environmental Hazards and Disasters Nicolas R. Dalezios, George P. Petropoulos, and Ioannis N. Faraslis�����������������������������������������������������������������3 2. Indigenous Knowledge for Disaster Solutions in the Hilly State of Mizoram, Northeast India Kewat Sanjay Kumar, Awadhesh Kumar, Vinod Prasad Khanduri, and Sudhir Kumar Singh�����������������������������23 3. Urban Risk and Resilience to Climate Change and Natural Hazards: A Perspective from Million‐Plus Cities on the Indian Subcontinent Amit Kumar, Diksha, A. C. Pandey, and M. L. Khan����������������������������������������������������������������������������������������33 4. The Contribution of Earth Observation in Disaster Prediction, Management, and Mitigation: A Holistic View Varsha Pandey, Prashant K. Srivastava, and George P. Petropoulos����������������������������������������������������������������47

Section II: Atmospheric Hazards and Disasters 5. Tropical Cyclones Over the North Indian Ocean in Changing Climate R. Bhatla, Raveena Raj, R. K. Mall, and Shivani...............................................................................................65 6. Simulation of Intensity and Track of Tropical Cyclones Over the Arabian Sea Using the Weather Research and Forecast (WRF) Modeling System with Different Initial Conditions (ICs) Sushil Kumar, Ashish Routray, Prabhjot Singh Chawla, and Shilpi Kalra...........................................................77 7. Development of a Soft Computing Model from the Reanalyzed Atmospheric Data to  Detect Severe Weather Conditions Devajyoti Dutta, Ashish Routray, and Prashant K. Srivastava...........................................................................85 8. Lightning, the Global Electric Circuit, and Climate N. Jeni Victor, Sagarika Chandra, and Devendraa Siingh.................................................................................93 9. An Exploration of the Panther Mountain Crater Impact Using Spatial Data and  GIS Spatial Correlation Analysis Techniques Sawyer Reid Stippa, Konstantinos P. Ferentinos, and George P. Petropoulos..................................................111

Section III: Land Hazards and Disasters 10. Satellite Radar Interferometry Processing and Elevation Change Analysis for  Geoenvironmental Hazard Assessment Sergey Stankevich, Iryna Piestova, Anna Kozlova, Olga Titarenko, and Sudhir Kumar Singh­­������������������������127 11. Assessing the Use of Sentinel‐2 in Burnt Area Cartography: Findings from a Case Study in Spain Craig Amos, Konstantinos P. Ferentinos, George P. Petropoulos, and Prashant K. Srivastava..........................141 vii

viii CONTENTS

12. Assimilating SEVIRI Satellite Observation into the Name‐III Dispersion Model to  Improve Volcanic Ash Forecast Prajakta Patil, I. M. Watson, Shona Mackie, Prashant K. Srivastava, Tanvir Islam, and Sourabh Sakhare...................................................................................................................................151 13. Geoinformation Technology for Drought Assessment Arnab Kundu, D. M. Denis, N. R. Patel, R. K. Mall, and Dipanwita Dutta......................................................171 14. Introduction to Landslides H. K. Pandey................................................................................................................................................181 15. Probabilistic Landslide Hazard Assessment using Statistical Information Value (SIV) and GIS Techniques: A Case Study of Himachal Pradesh, India Ankit Sharma, Ujjwal Sur, Prafull Singh, Praveen Kumar Rai, and Prashant K. Srivastava...............................197 16. One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India Dhruvesh P. Patel, Prashant K. Srivastava, Sudhir Kumar Singh, Cristina Prieto, and Dawei Han....................209

Section IV: Ocean Hazards and Disasters 17. Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal Prasad K. Bhaskaran, A. D. Rao, and Tad Murty...........................................................................................239 18. Space‐Based Measurement of Rainfall Over India and Nearby Oceans Using Remote Sensing Application Anoop Kumar Mishra and Kishan Singh Rawat�������������������������������������������������������������������������������������������295 19. Modeling Tsunami Attenuation and Impacts on Coastal Communities S. Piché, I. Nistor, and T. Murty�������������������������������������������������������������������������������������������������������������������309 Index������������������������������������������������������������������������������������������������������������������������������������������������������������������325

CONTRIBUTORS Craig Amos Department of Geography and Earth Sciences University of Aberystwyth Wales, United Kingdom

Dipanwita Dutta Department of Remote Sensing and GIS Vidyasagar University West Bengal, India

Prasad K. Bhaskaran Department of Ocean Engineering and Naval Architecture Indian Institute of Technology Kharagpur West Bengal, India

Ioannis N. Faraslis University of Thessaly Department of Planning and Regional Development Pedion Areos Volos, Greece

R. Bhatla Department of Geophysics Banaras Hindu University Varanasi, Uttar Pradesh, India; DST-Mahamana Centre of Excellence in Climate Change Research Institute of Environment and Sustainable Development Banaras Hindu University Varanasi, Uttar Pradesh, India

Konstantinos P. Ferentinos Hellenic Agricultural Organization “Demeter” Soil & Water Resources Institute Department of Agricultural Engineering Athens, Greece Dawei Han Department of Civil Engineering University of Bristol Bristol, UK

Sagarika Chandra Indian Institute of Tropical Meteorology Pune, India

Tanvir Islam NASA Jet Propulsion Laboratory Pasadena, California, USA

Prabhjot Singh Chawla Gautam Buddha University Greater Noida, India

Shilpi Kalra Gautam Buddha University Greater Noida, India

Nicolas R. Dalezios University of Thessaly Department of Civil Engineering Pedion Areos Volos, Greece

M. L. Khan Department of Botany Dr. Harisingh Gour Vishwavidyalaya Sagar Madhya Pradesh, India

D. M. Denis Department of Irrigation and Drainage Engineering Sam Higginbottom University of Agriculture, Technology and Sciences Uttar Pradesh, India

Vinod Prasad Khanduri Department of Forestry Uttarakhand University of Horticulture and Forestry Ranichauri, Uttarakhand, India

Diksha Department of Geoinformatics Central University of Jharkhand Ranchi, India

Anna Kozlova Scientific Centre for Aerospace Research of the Earth National Academy of Sciences of Ukraine Kiev, Ukraine

Devajyoti Dutta National Centre for Medium Range Weather Forecasting (NCMRWF) Ministry of Earth Sciences Uttar Pradesh, India

Amit Kumar Department of Geoinformatics Central University of Jharkhand Ranchi, India ix

x CONTRIBUTORS

Awadhesh Kumar Department of HAMP Mizoram University Mizoram, India

H. K. Pandey Department of Civil Engineering Motilal Nehru National Institute of Technology Allahabad, India

Kewat Sanjay Kumar Department of Forestry Mizoram University Mizoram, India

Varsha Pandey Institute of Environment and Sustainable Development and DST-Mahamana Center for Excellence in Climate Change Research Banaras Hindu University Uttar Pradesh, India

Sushil Kumar Gautam Buddha University Greater Noida, India Arnab Kundu DST‐Mahamana Centre of Excellence in Climate Change Research Banaras Hindu University Varanasi, Uttar Pradesh, India; Institute of Environment and Sustainable Development Banaras Hindu University Varanasi, Uttar Pradesh, India

Dhruvesh P. Patel Department of Civil Engineering School of Technology PDPU, Gujarat, India N. R. Patel Department of Agriculture and Soil Indian Institute of Remote Sensing (ISRO) Uttarakhand, India

Shona Mackie Department of Earth Sciences University of Bristol Bristol, United Kingdom

Prajakta Patil Department of Earth Sciences University of Bristol Bristol, United Kingdom

R. K. Mall DST‐Mahamana Centre of Excellence in Climate Change Research Banaras Hindu University Varanasi, Uttar Pradesh, India; Institute of Environment and Sustainable Development Banaras Hindu University Varanasi, Uttar Pradesh, India

George P. Petropoulos School of Mineral & Resources Engineering Technical University of Crete Kounoupidiana Campus Crete, Greece; Department of Soil & Water Resources Institute of Industrial & Forage Crops Hellenic Agricultural Organization “Demeter” (former NAGREF) Directorate General of Agricultural Research, Larisa, Greece

Anoop Kumar Mishra Center for Remote Sensing and Geoinformatics Sathyabama University Chennai, Tamil Nadu, India Tad Murty Department of Civil Engineering University of Ottawa Ontario, Canada I. Nistor Department of Civil Engineering University of Ottawa Ontario, Canada A. C. Pandey Department of Geoinformatics Central University of Jharkhand Ranchi, India

S. Piché Department of Civil Engineering University of Ottawa Ontario, Canada Iryna Piestova Scientific Centre for Aerospace Research of the Earth National Academy of Sciences of Ukraine Kiev, Ukraine Cristina Prieto Environmental Hydraulics Institute Universidad de Cantabria Parque Científico y Tecnológico de Cantabria Santander, Spain

CONTRIBUTORS  xi

Praveen Kumar Rai Amity Institute of Geo‐Informatics and Remote Sensing Amity University Noida, India; Department of Geography Institute of Science Banaras Hindu University Uttar Pradesh, India Raveena Raj Environmental Science Department of Botany Banaras Hindu University Varanasi, Uttar Pradesh, India A. D. Rao Centre for Atmospheric Sciences Indian Institute of Technology Delhi New Delhi, India Kishan Singh Rawat Center for Remote Sensing and Geoinformatics Sathyabama University Chennai, Tamil Nadu, India Ashish Routray National Centre for Medium Range Weather Forecasting (NCMRWF) Ministry of Earth Sciences Uttar Pradesh, India Sourabh Sakhare Indian Institute of Surveying & Mapping Survey of India Training Institute Hyderabad, India Ankit Sharma Amity Institute of Geo‐Informatics and Remote Sensing Amity University Noida, India Shivani Environmental Science Department of Botany Banaras Hindu University Varanasi, Uttar Pradesh, India Devendraa Siingh Indian Institute of Tropical Meteorology Pune, India

Prafull Singh Amity Institute of Geo‐Informatics and Remote Sensing Amity University Noida, India Sudhir Kumar Singh K. Banerjee Centre of Atmospheric and Ocean Studies University of Allahabad Allahabad, India Prashant K. Srivastava Institute of Environment and Sustainable Development and DST-Mahamana Center for Excellence in Climate Change Research Banaras Hindu University Uttar Pradesh, India Sergey Stankevich Scientific Centre for Aerospace Research of the Earth National Academy of Sciences of Ukraine Kiev, Ukraine Sawyer Reid Stippa Department of Geography and Earth Sciences University of Aberystwyth Wales, United Kingdom Ujjwal Sur Amity Institute of Geo‐Informatics and Remote Sensing Amity University Noida, India Olga Titarenko Scientific Centre for Aerospace Research of the Earth National Academy of Sciences of Ukraine Kiev, Ukraine N. Jeni Victor Indian Institute of Tropical Meteorology Pune, India I. M. Watson Department of Earth Sciences University of Bristol Bristol, United Kingdom

PREFACE We are often told our Universe began with a Big Bang, a disaster that made the stars and galaxies we see today. And, ever since, life on earth has evolved and flourished, surviving a series of unexpected bangs and calamities of one or another type. Nowadays, every place on earth is vulnerable to some kind of disaster, whether natural or human induced. Not only are developing and third world countries with less infrastructure and facilities in danger, but leading developed economies too face severe negative impacts due to disasters in terms of human and capital losses, mostly due to our lack of understanding of the processes involved in contributing to the severity of a disaster. With the most stated reason for the increase in the number and frequency of these disasters being the recent episodes of climate change and variability in earth’s history, many researchers have directed their studies towards developing more advanced and sophisticated early warning systems and techniques for precise prediction and forecasting of disaster. In this context, this book highlights state‐of‐the‐art new approaches, various modelling aspects, the role of field observations and management strategies, and efficient use of infrastructure in combating disasters. It addresses the interests of a wide spectrum of readers with a common interest in geospatial science, geology, water resource management, database management, planning and policy making, and resource management. The chapters in book focus mostly on emphasizing the investigation and identification of disasters through advanced computational techniques in conjunction with Geographic Information Systems (GIS) and Earth observation data sets for better management, adaptation, and mitigation of natural disasters. The book is divided into four sections. Section I focuses on a general introduction to the disaster management and mitigation, with an overview on the different types of disaster and the importance of the existing traditional technologies mostly widely used for natural disasters, emphasizing the relevance of indigenous approaches in disaster management. The section also underlines the importance of community‐based techniques in disaster management, postdisaster management, and developing mitigation plans. Section II contains chapters presenting detailed studies on atmospheric hazards and disasters, with some studies focusing on extreme weather events such clouds burst and tropical cyclones. To highlight the

advancement in modern technologies for disaster management on land surfaces, Section III presents the role of modern technologies for disaster management and mitigation in cases such as drought and landslides. The section contains articles focusing on the role of earth observing techniques, database management through cloud management, and emergency preparedness using Global Positioning Systems (GPS). The next section, Section IV, illustrates the application and capability of satellite and mesoscale modelling for better understanding and management of oceanic disasters disasters and hazards. Section I, opens with an introductory chapter (Dalezios et al.) on concepts and methodologies of environmental hazards and disasters, providing the basics concepts of disasters in different fields. Chapter 2 (Kumar et al.), on indigenous knowledge for disasters solution in hilly states, discusses the role of local or indigenous people’s knowledge towards understanding and developing mitigation plans for disasters in hilly areas. Chapter 3 (Diksha et  al.) presents an overview of the risk of disasters in urban areas and the relationship with climate change. The chapter provides a perspective from those cities on the Indian subcontinent with more than million inhabitants. The last chapter of this section (Pandey et al.), on the role of earth observing techniques in disaster prediction, management, and mitigation, provides a brief description of remote sensing and GIS techniques in disaster monitoring. Section II of the book focuses on atmospheric hazards and related disasters. Chapter 5 (Bhatla and team) provides detailed accounts of tropical cyclones over the North Indian Ocean in changing climate, while Chapter 6 (Kumar and team) provides a detailed analysis of the simulation of the intensity and track of tropical cyclones over the Arabian Sea using the WRF modelling system. In Chapter 7 (Dutta et al.) a soft computing model developed using reanalyzed atmospheric data to detect severe weather conditions is described. Chapter 8 (Victor et al.) covers lightning, the global electric circuit, and the ­relationship with the climate. Chapter  9 (Stippa et  al.) provides an exploration of the Panther Mountain crater impact using spatial data and GIS spatial correlation analysis techniques. Section III of the book focuses on land hazards and disasters; it highlights disasters on land such as drought,

xiii

xiv Preface

landslides, volcanic eruption, and forest fires. The section starts with Chapter 10 (Stankevich and team) exploring satellite radar interferometry processing and elevation change analysis for geo‐environmental hazard assessment and continues with Chapter 11 (Amos et al.) documenting the use of Sentinel‐2 in burnt area cartography and the findings from a case study in Spain. Chapter 12 (Patil and team) provides an assessment of the Name‐III dispersion model after assimilating the SEVIRI satellite observation for volcanic ash forecast. Chapter 13 (Kundu et al.) describes geo‐information technology for drought assessment using satellite and geospatial techniques. Chapter  14 (Pandey et  al.) provides an introduction to the causes and control of landslides, while Chapter  15 (Sharma and team) reviews probabilistic landslide hazard assessment using Statistical Information Value (SIV) and GIS techniques. Chapter 16 (Patel et al.) introduces 1D hydrodynamic modelling for flood risk assessment, to simulate and understand flooding risk in coastal areas. The last section of the book contains chapters discussing oceanic hazards and disasters, with Chapter  17 (Bhaskaran et  al.) covering tropical cyclone induced

storm surges and wind‐waves in the Bay of Bengal, Chapter  18 (Mishra and Rawat) discussing space‐based measurement of rainfall over India and nearby oceans using remote sensing applications, and Chapter 19 (Piche et al.) detailing the modelling of tsunami attenuation and the impact on coastal communities. We believe that this book will be beneficial for people with a common interest in disaster management and ­mitigation. The variety of techniques outlined in this book, such as geospatial techniques, remote sensing and applications, emergency preparedness, policy making, and other diverse topics, in the earth, environmental, and hydrological sciences fields will provide readers with updated knowledge. We hope that this book will be beneficial for academics, scientists, environmentalists, meteorologists, environmental consultants, and computing experts working in the area of disaster risk management and mitigation. Prashant K. Srivastava Sudhir Kumar Singh U. C. Mohanty Tad Murty

Section I Introduction

1 Concepts and Methodologies of Environmental Hazards and Disasters Nicolas R. Dalezios1, George P. Petropoulos2, and Ioannis N. Faraslis3 ABSTRACT Natural disasters have significant impact on several sectors of the economy, including agriculture. Moreover, under climate uncertainty, the role of several sectors of the economy, such as agriculture, as a provider of environmental and ecosystem services, is expected to further gain importance. Indeed, increasing climate variability and climate change lead to increases in climate extremes. The objective of this review is to present concepts and methodologies of environmental hazards and extremes that affect agriculture and agroecosystems, based on remote sensing data and methods, since this is a field gaining in potential and reliability. In this chapter, the relationship between environmental hazards and agriculture is presented; this is followed by concepts and quantitative methodologies of environmental hazards affecting agriculture, namely hydrometeorological hazards (floods and excess rain, droughts, hail, desertification) and biophysical hazards (frost, heat waves, wild fires). The emphasis is on concepts and the three stages of hazard development: forecasting‐nowcasting (before), monitoring (during), and assessment (after). Examples and case studies are presented using recorded data sets, model simulations, and innovative methodologies. 1.1. ­INTRODUCTION

change and increasing climate variability, which have led to increases in climate extremes. During the 21st century, scientific projections, among others, point to changes in climate extremes, such as heat waves, heavy rainfall, and droughts, in many semiarid and arid regions around the world. Specifically, southern Europe and the entire Mediterranean basin are characterized as vulnerable regions due to the combined effect of temperature increases and reduced precipitation in areas already coping with water scarcity (Dalezios et al., 2018a; Srivastava et al., 2019). Agricultural production risks could become an issue in areas such as the entire Mediterranean basin, as mainly droughts and heat waves are likely to increase the incidence of crop failure. As yield variability increases, food supply is at increasing risk. Under a changing climate, the role of agriculture as a provider of environmental and ecosystem services is expected to gain further importance. Improvement of water use efficiency in dry regions is an example of agricultural management. Vulnerability of agriculture can be reduced through adaptation measures and tools to

Agriculture faces several current and future challenges, such as international competition and further liberalization of trade policy. Additionally, environmental hazards play a major role in agriculture; this has resulted in a gradual and significant increase of the economic cost associated with all environmental hazards. Needless to say, agricultural production is highly dependent on ­climate, and is adversely affected by anthropogenic c­ limate  University of Thessaly, Department of Civil Engineering, Pedion Areos, Volos, Greece 2  School of Mineral & Resources Engineering, Technical University of Crete, Kounoupidiana Campus, Crete, Greece; Department of Soil & Water Resources, Institute of Industrial & Forage Crops, Hellenic Agricultural Organization “Demeter” (former NAGREF), Directorate General of Agricultural Research, Larissa, Greece 3   University of Thessaly, Department of Planning and Regional Development, Pedion Areos, Volos, Greece 1

Techniques for Disaster Risk Management and Mitigation, First Edition. Edited by Prashant K. Srivastava, Sudhir Kumar Singh, U. C. Mohanty, and Tad Murty. © 2020 John Wiley & Sons, Inc. Published 2020 by John Wiley & Sons, Inc. 3

4  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

increase climate variability (EU, 2007). Some farming systems may adapt more readily to climate pressures due to an inherent resilience. Other systems may need interventions for adaptation. However, besides traditional knowledge and technologies, more sophisticated technologies seem to be required due to increasing climate ­variability and change. Seasonal to interannual climate forecasting is a new branch of climate science that promises to reduce vulnerability in agriculture. Improved seasonal forecasts are now being linked to decision making for cropping. The application of climate knowledge to improving risk management is expected to increase the resilience of farming systems. Environmental degradation is one of the major factors contributing to the vulnerability of agriculture because it directly magnifies the risk of environmental disasters. In order to ensure sustainability in agricultural production, a better understanding of the environmental hazards and disasters that impact agriculture is essential. A comprehensive assessment of impacts of natural disasters on agriculture requires a multidisciplinary, multisectoral, and integral approach involving several components and factors. Priority should be given to supporting applied research, since research is necessary to understand the physical and biological factors contributing to disasters. Community‐ wide awareness and capacity building on environmental hazards and disasters, mainly for farmers and stakeholders, should also be included in any research effort. Programs for improving prediction and early warning methods, as well as dissemination of warnings, should be expanded and intensified. Moreover, efforts are required to determine the impact of disasters on natural resources. Recent research findings suggest that variability of climate, if encompassing more intense and frequent ­ extremes, such as major large‐scale hazards like droughts, heat waves, or floods, results in the occurrence of natural disasters that are beyond our socioeconomic planning levels. It is estimated that about 65% of the global damage from natural disasters has a meteorological origin. Also, meteorological factors contribute to 87% of people affected by natural disasters and 85% of relevant deaths (UN/ISDR, 2015; EM‐DAT, 2009; WMO, 2004). This is expected to stretch regional response capabilities beyond their capacity and will require new adaptation and ­preparedness strategies (Salinger et  al., 2005). Disaster prevention and preparedness should become a priority, and rapid response capacities to climate change need to be accompanied by a strategy for disaster prevention. Nevertheless, each type of extreme event has its own particular climatic, cultural, and environmental setting, and mitigation activities must use these settings as a foundation of proactive management. There is significant complexity involved in homogenizing and issuing global or regional statistics for disasters affecting agriculture,

since this depends on the specific climatic zone and environment where the agricultural activity takes place, as well as the type, areal extent, and microclimatic and agronomic characteristics of the crops in that zone, including agroclimatic features. Nevertheless, international organizations, such as the Food and Agriculture Organization (FAO), the World Meteorological Organization (WMO), or the United Nations International Strategy for Disaster Reduction (UN/ISDR), issue statistics periodically that refer to environmental hazards and disasters that affect agriculture and agroecosystems. There is an urgent need to assess the forecasting skills for environmental hazards affecting agriculture in order to determine those where greater research is required. It is well known that lack of good forecast skill is a constraint to improve management, mitigation, and adaptation. A holistic and integrated approach to environmental risks has gradually explored the use of common methodologies, such as risk analysis, including risk assessment and management. Indeed, through risk analysis, there are efforts to develop preventive measures and hazard processes before the crisis. It should be stated that current natural disaster management is crisis driven. It is thus realized that there is an urgent research need for a more risk‐based management approach to natural disaster planning in agriculture, which would include a timely and user‐oriented early warning system (Dalezios, 2017). Agricultural risk zoning is also an essential component of natural disaster mitigation and preparedness strategies. GIS and remote sensing and, in general, geoinformatics are increasingly employed due to the complex nature of databases to facilitate strategic and tactical applications at the farm and policy levels. Therefore, additional research is required to incorporate GIS, remote sensing, global positioning systems (GPS), simulation models, and other computational techniques into an integrated multihazard risk management framework for sustainable agriculture that includes early warnings of natural disasters (Sivakumar et  al., 2005). There should also be more research attention to the potential impact of the increasing frequency and severity of extreme events associated with global change and appropriate mitigation strategies. In general, risk assessment methodologies include three stages, or sectors, such as forecasting and early warnings before the phenomenon occurs, monitoring during a natural disaster, and estimating damage after the end of a disaster. In addition, risk identification involves quantifying, monitoring, and event risk, statistical inference, and database development, which should include records and historical information on disasters and their impacts. Risk assessment also entails reviewing the risk of these events, that is, the probability of occurrence, as well as the magnitude–duration–frequency and area‐to‐risk ratio. Finally, the risk assessment includes an environmental

Concepts and Methodologies of Environmental Hazards and Disasters  5

impact assessment and cost–benefit analysis of the adaptation options for the development of countermeasures (Dalezios & Eslamian, 2016). The current scientific and technological capabilities of remote sensing cover all three areas of risk management. Remote sensing has gradually become an important tool for the quantification and detection of the spatial and temporal distribution and variability of several environmental variables at different scales. At the present time, the growing number and effectiveness of pertinent observation satellite systems present a wide range of new capabilities in assessing and monitoring several features of environmental variables. Moreover, remote sensing methods have also reached a significant level of accuracy and reliability over the last 40 years. Specifically, remotely sensed models are currently considered suitable for crop water use estimation at field as well as regional scales (Dalezios, 2014). Thus, remotely sensed forecasting‐­nowcasting, monitoring, and assessment of environmental hazards are becoming attractive, since these systems provide consistently available data with high ­resolution covering large areas. In this chapter, the major environmental hazards affecting agriculture are considered under increasing ­climate variability, namely hydrometeorological hazards (floods and excess rain, droughts, hail, desertification) and biophysical hazards (frost, heat waves, wild fires). The emphasis is placed on environmental hazards concepts and methodologies on the three stages of hazard development, namely forecasting‐nowcasting (before), monitoring (during), and assessment (after). Examples and case studies are presented using recorded data sets, model simulations, and innovative methodologies. in selected agricultural regions in southern Europe. 1.2. ­HYDROMETEOROLOGICAL HAZARDS IN AGRICULTURE In this section, hydrometeorological hazards affecting agriculture are considered, namely floods and excess rain, droughts, hail, and desertification. For each hazard, some concepts are presented, along with methodologies on the three stages of hazard development: forecasting‐nowcasting (before), monitoring (during), and assessing (after). 1.2.1. Floods and Excess Rain Floods can be devastating disasters that can affect anyone at almost any time (Ireland et al., 2015). Flooding has been one of the most costly disasters in terms of both human casualties and property throughout the last centuries. Hazards associated with flooding can be divided into primary hazards that occur due to contact with water, secondary effects that occur because of the flooding, such as disruption of services and health impacts, for example

famine and disease, and tertiary effects, such as changes in the position of river channels. The term hazard (or cause), which in this case is flood, may be defined as the potential threat to humans and their welfare (Smith, 2013). Hazards can include latent conditions that may represent future threats and can have different origins, such as natural hazards or those induced by human processes (UN/ISDR, 2005). 1.2.1.1. Flood Forecasting The prediction of flood events is of hydrological importance. As a prognosis, it is not only the estimation of the frequency of a hydrologic episode of a certain size, but also the forecast of the size and time of a flood peak. In order to reduce the risk due to flooding, three steps are considered for flood prediction. First, determination is conducted of the probability and frequency of high discharges of streams that cause flooding. Second, floods can be modeled and maps can be produced to determine the extent of possible flooding that may occur in the future. Third, since the main causes of flooding are abnormal amounts of rainfall and sudden melting of snow or ice, storms and snow levels can be monitored to provide short‐term flood prediction. Determining the timing and magnitude of floods is necessary for design flood purposes. In most cases, it is also necessary to classify the  flood flows according to the flood‐producing ­mechanisms, for example in flood‐frequency studies. The classification of flood flows in various physical types should provide a better and reliable estimate of the magnitude of design floods, which, in turn, is necessary for the design of hydrotechnical projects. Flood‐frequency analysis is used to predict design floods for sites along a river. The frequency of occurrence of floods of different magnitude can be estimated by a variety of methods depending on the availability of hydrometric data (Loukas et al., 2002). Under normal conditions, observed annual peak flow discharge data are used to calculate statistical information, which then constitute the basis to construct frequency distributions, which delineate the likelihood of various discharges as a function of recurrence interval or exceedance probability. The choice of a design flood magnitude with its assessed return period depends both on the expected life of the scheme and on the degree of protection required. However, in ungauged watersheds, the flood flow is estimated by various methods, which require the estimation of rainfall of particular critical duration and return period. This leads to the design storm concept, which is still the dominant design method in hydrological engineering. 1.2.1.2. Flood Monitoring Flood monitoring can be achieved through hydrological simulation. A hydrological model is an approximation of

6  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

the real hydrological system. The input data and the outputs are measurable hydrological parameters and the structure of the models is a set of equations, which relate the inputs to the outputs. Modeling efforts are considered in three levels of temporal and spatial scaling (Schultz & Engman, 2000). 1. Design of water supply systems. This type of ­modeling requires long‐term time series data records of hydrological variables with the minimum time step being the month. For instance, the conventional data source could be observed or generated runoff data. The employed hydrological model could be a transfer function model in convolution integral, which is a stochastic black‐box model based on the theory of linear systems. 2. Design of flood protection measures. This type of modeling requires data from numerous cases of short‐ term extreme hydrological events with a time step of days, hours, or even 10 min, for example, in the case of urban systems. The data source could be observed or extrapolated runoff data. Hydrological modeling could include a rainfall model in association with a rainfall‐runoff model of a distributed or lumped system type, for example, unit hydrograph. 3. Operation of water resource systems. This type of modeling requires short‐term or even real‐time data with a time step of 10 minutes, hours or even a day. The data source should be rainfall observed in real time, forecast of rainfall, as well as observed runoff. Additional data sources should be ground‐based weather radar and IR data from geostationary meteorological satellites. The hydrological modeling system to be used should include a rainfall model and a rainfall‐runoff model, preferably of distributed system type, in order to conduct now‐casting of extreme events in real time or semi-real time. 1.2.1.3. Assessment and Causes of Floods Extreme values ​​ of runoff, namely flood spikes, are ­typically formed abruptly after a corresponding sudden rainfall, and include massive bodies of water. There are several causes of flood peaks, where the most important (a)

of these are rainfall of high intensity and duration, melting of snow and ice, or the sudden destruction of water‐saving techniques. In general, runoff is subject to rainfall fluctuations in relation to the climate of a region. Apart from seasonal variation, permanent changes in runoff usually occur due to the following causes: urbanization, abandonment of the countryside and development of wild vegetation in the catchment area, natural disasters, forest burning and destruction, changes in land use, construction of large water storage, and large development projects. Variability in runoff spatiotemporal distribution may also be expected due to climate change, since runoff is the hydrological variable that is particularly sensitive to climate change or climate uncertainty. 1.2.1.4. Damages from Excess Rain Heavy rainfall has caused significant damages to ­agricultural production globally. The cost of crop damage over the next decades could increase dramatically. It is estimated that, due to climate change, the enhanced hydrological cycle is responsible for increasing the damage from recent floods. Determination of cause and effect between increases in rainfall and flood damage is difficult due to simultaneous changes in population growth, economic growth, and infrastructure, among others. However, recent data show that total annual rainfall and heavy rainfall have increased in many parts of the world during the last century, particularly during the last two decades (Milly et  al., 2002), often resulting in large harvest losses and other losses due to floods (Chagnon et al., 1997). For example, floods in the US Midwest in 1993 caused damage to farmers estimated at around $6–8 billion, about 50% of total flood losses (FEMA, 1995). Also, agricultural production was negatively affected by the 1997 US North Dakota flood, which caused a total loss of about one billion dollars. In addition, the 2001 Mississippi floods delayed planting. Heavy rainfall, leading to floods, could also cause soil erosion on farmland (Figure 1.1a, b). Excessive soil moisture is an important element of crop damage along with (b)

Figure 1.1  Heavy rainfall, leading to floods, could also cause soil erosion on farmland.

Concepts and Methodologies of Environmental Hazards and Disasters  7

extreme rainfall. In order to quantify the significance of the effects of excessive soil moisture on current and ­projected future crop production estimates, a simplified model could be used (Rosenzweig et  al., 2002). During the Mississippi floods of 1993, about 70% of total harvest losses occurred in mountainous areas due to the saturation of soils by heavy rainfall. Over the past 20 years, excessive soil moisture cost US farmers five times more than direct flood damage, according to crop insurance data.

Hail is a natural environmental hazard and can destroy an entire agricultural production within a few minutes. Since the need to combat the devastating effects of hail on agricultural production has always been urgent, attempts to tackle hail were made long before sufficient understanding and knowledge of the phenomenon were developed.

vertical winds can reach speeds over 176 km per hour. As a hailstone starts to form, it becomes too heavy for the top of the cloud. It starts to fall through the cloud ­gathering more supercooled droplets as it falls. If the hailstone is not heavy enough, the updraft pushes the hailstone back up into the upper recesses of the cloud. The hailstone once again gathers more droplets and grows even bigger. It then starts to fall again. This cycle continues until the hailstone is heavy enough to overcome the updraft. At this point, the hailstone leaves the cloud and falls onto the ground. This process creates hailstones from a few millimeters in diameter up to 15 cm that weight more than half a kilogram. Pea or golf‐ball‐sized hailstones are not uncommon in severe storms. Hail falls along paths, which are called hail swaths. These vary from a few square kilometers or less to large belts 16 km wide and 160 km long. Figure  1.2 illustrates the sequence of the downstream process recording the sequence of weather events leading to hail.

1.2.2.1. Hail Formation Hail forms inside thunderstorm cumulonimbus (Cb) clouds, mainly those with intense updrafts, high liquid water content, great vertical extent, large water droplets, and a good portion of the cloud layer below freezing temperature. Hail grows in the main updraft of a cloud, where most of the cloud is in the form of supercooled water. This is the water that remains liquid although its temperature is at or below 0 ° C. The growth rate is maximized at about ‐13 ° C, and becomes vanishingly small much below ‐30 ° C as supercooled water droplets become rare. Cumulonimbus clouds contain vast amounts of energy in the form of updrafts and downdrafts. These

1.2.2.2. Climatology of Hail Hail is most common in early summer at midlatitudes where surface temperatures are warm enough to promote the instability associated with strong thunderstorms, but the upper atmosphere is still cool enough to support ice. Hail is actually less common in the tropics despite a much higher frequency of thunderstorms than in the midlatitudes, because the atmosphere over the tropics tends to be warmer over a much greater depth. Also, entrainment of dry air into strong thunderstorms over continents can increase the frequency of hail by promoting evaporational cooling, which lowers the freezing level of thunderstorm clouds giving hail a larger volume to grow in. Hail is also

1.2.2. Hail

Synoptic

Meso

Scale

Scale

Convective cloud process

Prediction

Cloud initiation Storm environment

Hail initiation Hail growth Hail fail

Synoptic environment

Hail at ground

Figure 1.2  Precipitation procedure chain.

8  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

much more common along mountain ranges, because mountains force horizontal winds upward (known as orographic lifting), thereby intensifying the updrafts within thunderstorms and making hail more likely. In Greece, hail produced during the cold season is of small diameter and generally causes no damage to ­agriculture. However, during the warm season, hail is produced in continental parts of Greece mainly from convective storms formed due to intensive heating, which creates instability in the air. The terrain of Greece, with mountains, forces air masses to lift and therefore produce thunderstorms often accompanied by hail, which often damages agricultural production as conditions allow thunderstorms to grow enough and produce hailstones of large diameter locally. The annual distribution of hail days over Greece is shown in Figure 1.3. The figure clearly indicates that hail days are more often over western and eastern parts of Greece during the cold season, while ­during the warm season hail days are observed in central and northern continental parts of Greece. 1.2.2.3. Hail Damage Hail can cause serious damage, notably to automobiles, skylights, glass‐roofed structures, but most commonly to agriculture. Even small hailstones can destroy crops and slice plants to ribbons in a few seconds. Thus, although a hailstorm lasts only for a few minutes (2–8 min) and rarely more, the damage caused is of great importance. Efforts for mitigation of hail damage were started by humans long before science was able to explain hail formation and processes. In the fourteenth century, people in Europe attempted to ward off hail by ringing church bells and firing cannons. Hail cannons were e­specially famous

in  the wine‐producing regions of Europe during the nineteenth century, and modern ­versions of them are still used in parts of Italy. After World War II, scientists across the world experimented with cloud “seeding” as a means of reducing hail size. Cannons that fired silver iodide (AgI) into thunderclouds from the ground were also used. Hail suppression ­programs were used worldwide to reduce hail damage to crops. 1.2.2.4. Hail Suppression Protection from hail is classified into two major methods, namely active and passive protection. Active protection includes modern techniques applied before a hailstorm hits a region in order to reduce or suppress hail and passive protection includes actions taken by farmers so that hailstones do not hit the plants. Active protection methods are cloud‐seeding techniques from aircraft, hail cannons, and hail rockets. Hail cannons ignite a charge of acetylene gas in a specially designed blast chamber releasing an explosive pressure wave creating a cavitation effect, which disrupts the formation process of the hailstone embryo. Hail rockets are fired from the ground inside cumulonimbus clouds in order to influence hail clouds to prevent hail and to stimulate rainfall. Cloud seeding from aircraft includes seeding of thunderstorm clouds with silver iodide or other material (e.g., dry ice) in order to diminish hail production. Passive protection from hail is performed by applying a safety net over the ground and thus protecting crops or fruit trees. 1.2.2.5. Design of Hail Suppression in Greece The National Hail Suppression Program (NHSP) of Greece was established in 1984 and designed as an

1 WEST

0,9

CENTRAL NORTH

0,8

EAST SOUTH

Number of hail days

0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 JAN

FEB

MAR

APR

MAY

JUN

JUL AUG

SEP

OCT NOV

DEC

Figure 1.3  Number of observed hail days for different climatic zones of Greece.

Concepts and Methodologies of Environmental Hazards and Disasters  9

­ perational and, at the same time, as a research cloud‐ o seeding program. The hail‐suppression cloud‐seeding ­hypothesis of the NHSP is based on the cloud microphysical concept of beneficial competition theory. This ­hypothesis assumes a lack of natural ice nuclei in the storm environment and that the injection of artificial nuclei of AgI (silver iodide) will increase the total ice nuclei in thunderstorm clouds. Hence, the supercooled water available to each embryo is limited and the hailstones that are formed will be smaller and produce less damage if they reach the ground. Protected areas of the NHSP are shown in Figure 1.4. The seeding criteria in the NHSP require that seeding must be conducted on every storm reaching a radar intensity of 35 dBZ or greater, at altitudes between the ‐5 °C and ‐30 °C levels, while these storms are within the project area or within a 20 min upwind buffer of the project area. Cloud seeding is conducted using aircraft (Piper Cheyenne II) equipped with seeding racks containing both droppable and end‐burning silver iodide flares. 1.2.2.6. Quantitative Hail Forecast Indicative methods of forecasting hail, now‐casting, and storm tracking include predictive meteorology, numerical weather forecast, cloud models of one dimension (1‐D), two‐dimensions (2‐D), and three‐ dimensions (3‐D). Also highlighted is the use of weather

radar and satellite systems. However, for quantitative hail forecast, the most prevalent methodology is to effectively combine some or all of the above methods, which has been applied to the NHSP of Greece. An objective synoptic index combines hydrostatic instability theory and scale variables to estimate the potential switching range. The index is derived from statistical linear regression, where several independent variables are combined depending on their relationship with the dependent variable, which is the index. In particular, the index is called the Convective Day Category (CDC), and each day is defined as the maximum degree of switching intensity for an area in distinct classes. The classes include all types of switching, such as rainfall, storms, and hail of various sizes. In addition, 10 internationally known indicators of atmospheric instability have been selected and are utilized together with the CDC for optimal quantitative hail ­forecast (Dalezios & Papamanolis, 1991). 1.2.2.7. Hail Monitoring: Operational Part and Evaluation Weather surveillance is provided by two 5 cm C‐band weather radars. A routine 20 hr a day radar operation is conducted based on scheduled shifts, which is extended to 24 hr when continued storm activity threatens the project area. A network of hail pads and rain gauges operates in the project area to provide quantitative Rulgaria

Albania

Fyrom

A2 A1

LR X3

A3

100 km Aegen sea

N

Figure 1.4  Map of Greece showing the three protected agricultural areas of the NHSP.

10  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

­ easurements of hail falls. Hail‐pad data are primary m used for evaluation of the cloud seeding effect and also for an adequate sampling of hail falls and fine‐scale analysis of hail swaths. Figure  1.5a illustrates a schematic view of the cloud‐seeding operation. Figure 1.5b shows the operational framework of a hail suppression program. In particular, the start is done by quantification of hail (top left), followed by radar storm (bottom left), then cloud seeding by airplane (top right), and finally recording hail on the ground (bottom right). An integrated data processing and analysis approach is conducted in order to synthesize a vast amount of multisource data, such as radar, hail pads, aircraft, or compensation amounts, for the selection of all the appropriate parameters for the statistical evaluation. 1.2.3. Drought Drought is a natural phenomenon repeated over time on a regional scale. Drought is referred to as “nonevent,” as its basic cause is the lack of rainfall in a region over a period of time. In addition, drought is a natural hazard with a slow onset, often seen as a creeping phenomenon. It is difficult to determine the impact of drought, as it is a complex phenomenon that evolves gradually in each region. The effects of drought are very critical and costly, affecting more people than any other natural disaster globally (Keyantash & Dracup, 2002). Indeed, drought is seen as one of the most important natural hazards, with a significant impact on the environment, society, agriculture, and the economy. (a)

1.2.3.1. Types and Definitions of Drought It is known that there is no precise and universally accepted definition of drought, since there is a range of drought‐affected areas, there is a divergent spatial and temporal distribution and water demand for different uses. Definitions of drought must refer to area, use, or impacts. Indeed, droughts are regional in scope and each region has particular climatic characteristics. Considering drought as a risk, there is a tendency to classify droughts in different species or types. In the international literature, three functional definitions of drought have been established: (1) meteorological or climatic, (2) agricultural or agrometeorological, and (3) hydrological (Wilhite et  al., 2000). As a fourth type of drought, the socio‐ economic impact of drought can also be considered. With the exception of meteorological drought, other types of drought, such as agriculture and hydrology, emphasize the human or social aspects of drought, namely the interaction between the physical characteristics of meteorological drought and human activities that are dependent on precipitation (Keyantash & Dracup, 2002). Needless to say, the relationship between the ­different types of drought is complex. Here is a brief description of the types of drought. 1. Meteorological drought is a natural occurrence at a regional scale and, in general, characterized by a rainfall irregularity being lower than the average for some time and by prolonged and abnormal lack of humidity. 2. Agricultural drought refers to agricultural impacts, resulting from shortages in water availability for agricultural (b) CDC

Seeding

5 4 3 2 1 0 –1 –2 –3

Radar

Hail pad

Figure 1.5  (a) Cloud seeding operation; (b) operational framework of hail suppression program.

Concepts and Methodologies of Environmental Hazards and Disasters  11

use leading to crop failure and exists when soil moisture is exhausted so that crop yields are significantly reduced. 3. Hydrological drought is considered to be a rather long period during which the actual water supply, either surface or groundwater, is less than the minimum flow required for normal operation in a basin as a result of meteorological drought. 4. The socioeconomic impact of drought is defined as a loss from the average or expected income and can be measured by social and economic indicators (McVicar & Jupp, 1998). Indeed, the socioeconomic drought refers to the gap between supply and demand of economic goods resulting from the three other types of drought, such as water, food, raw materials, or transport. The types of drought along with the time sequence of the processes are shown in Figure 1.6. 1.2.3.2. Features of Drought Several features are considered for the assessment and monitoring of drought. Conventional data and remote sensing methods are used to determine the spatial and temporal variability of the various drought characteristics (Dalezios et al., 2012). Key features are described here. Quantification of drought is not an easy matter and can be considered by using indicators or indices. Indeed, drought indicators are variables, describing the characteristics of drought. Several indicators can also compose a single index, called a drought index. In addition, it is necessary to calculate various features of drought, such as severity, duration, periodicity, areal extent, start, and end of drought. Severity is defined according to its category: mild, moderate, strong, and extreme. The severity is usually determined through drought indicators. Periodicity is the reoccurrence of drought. Duration of a drought episode is defined as the period from beginning to end, usually in

months. Since drought is a complex phenomenon, estimating the start time and end time is a complex technical issue. Start is the beginning of a drought episode, identified by indications or markers, exceeding a threshold value. The end of a drought episode marks the end of drought, again based on the threshold values ​​of the indications or markers. Areal extent of drought is considered to be the spatial coverage of the phenomenon as quantified in classes with indications or indices. Monitoring drought development is vital for economically and environmentally sensitive areas and is a very important input into any drought preparedness and mitigation plan. Primary data for meteorological, agricultural, and hydrological droughts are climate v­ariables, such as temperature, rainfall, runoff, soil moisture, reservoir storage, groundwater levels, snow, and ­vegetation. Drought indicators provide ease of application and are extensively used in drought quantification (Keyantash & Dracup, 2002). 1.2.3.3. Drought Monitoring: Drought Early Warning System (DEWS) The DEWS focuses on monitoring drought conditions (Wilhite, 2009) on the basis of drought indicators. During the last decade, a web‐services‐based environment has been developed for integration of regional and continental drought monitors, for computation and display of spatially consistent drought indicators on a global scale, such as in situ Standardized Precipitation Index (SPI), satellite‐ based indices, and modeled soil moisture, and for drill‐ down capacity to regional, national, and local drought products. Each continental drought monitor is developed and functions according to the unique conditions of that continent. At the present time, there are several regional and continental drought monitor models, leading to a global drought monitor (GDM); these coordinate and

Precipitation deficit (meteorological drought)

Critical soil moisture deficit (soil moisture drought)

Critical streamflow and groundwater deficit (hydrological drought)

Evapotranspiration

Pre-event soil moisture, surface water,and/or groundwater storage

Figure 1.6  Simplified sketch of processes and drivers relevant for meteorological, soil moisture (agricultural), and hydrological droughts (from IPCC, 2012, SREX).

12  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

exchange information toward a global drought informa­ tion system (GDIS) (Dalezios et  al., 2018a). The four major regional‐continental models are (1) the North American drought monitor (NADM), which consists of the United States Drought Monitor (USDM), Canada and Mexico, (2) the European drought observatory (EDO) model, (3) the African drought monitor (ADM), and (4) the Australian drought monitor model. 1.2.3.4. Meteorological DEWS: RDI For illustrative purposes, a case study using an empirical model and leading to the drought early warning system (DEWS) based on the remotely sensed Reconnaissance Drought Index (RDI) (Dalezios et  al., 2012) is briefly presented. By plotting the cumulative monthly areal extent values of the extreme RDI drought class, that is, class 4 (Dalezios et al., 2012) with values lower than –2, for all the drought episodes, two figures are produced: Figure  1.7 for droughts of large areal extent and

Areal extent (number of pixels)

500

Figure 1.8 for droughts of small areal extent. In addition, curve fitting is conducted for each of these figures resulting in the following polynomials: equation (1.1) for droughts of large areal extent and equation (1.2) for droughts of small areal extent, both with high coefficient of determination. y

0.4771x3 9.7934 x 2

78.221x 36.078 R2

0.9676 (1.1)

y



0.4868x 2 3.3415x 4.78 R 2

0.9618 (1.2)

It is worth noticing that for droughts of large areal extent (Figure 1.7), drought starts during the first three months of the hydrological year, whereas for droughts of small areal extent (Figure  1.8), drought starts in spring (April). This finding signifies the possibility of using the fitted curves for monitoring and early‐warning drought

y = 0.4771x3 – 9.7934x2 + 78.221x – 36.078

450

R2 = 0.9676

400

1984 – 1985

350 300

1989 – 1990

250 200

1991 – 1992

150

Curve fitting

100 50 Sep

Jul

Aug

Jun

May

Apr

Mar

Feb

Jan

Dec

Oct

Nov

0

Figure 1.7  Cumulative large areal extent (number of pixels, 8 x 8 km2) of extreme drought (>2.0) during drought years based on remotely sensed RDI (from Dalezios et al., 2012).

Areal extent (number of pixels)

90

y = 0.4868x2 –3.3415x + 4.78

80

R2 = 0.9618

70

1987–1988

60

1992–1993

50 40

1996 –1997

30

1999 –2000

20

2000 – 2001

10

Curve fitting

0 Sep

Aug

Jul

Jun

May

Apr

Mar

Feb

Jan

Dec

Nov

Oct

–10

Figure 1.8  Cumulative small areal extent (number of pixels 8 x 8 km2) of extreme drought (>2.0) during drought years based on remotely sensed RDI (from Dalezios et al., 2012).

Concepts and Methodologies of Environmental Hazards and Disasters  13

assessment in a region. This finding justifies the use of the fitted curves of Figures  1.7 and 1.8, along with the corresponding equations (1.1) and (1.2), for drought prognostic assessment or DEWS.

Aridity Index: P/PET PET > P

1.2.4. Desertification

64%) as well as total population (31.16%) in 2015 (UNDP, 2014; Figure  3.1). This contributes significantly in the socioeconomic development of the country, and also leads to higher risk due to the increasing pressure of the population in the scenario of increasing hazards events. The cities with a million‐plus population (during 2015) in selected countries of the Indian subcontinent exhibited a considerable increase in urban population during the last few decades. The urban population in India has increased rapidly in the last four decades (1981–2011), where it has increased from 23.27% to 31.16%.The demographic figure indicates an enormous increase in urban population from 159 million in 1981 to 377.11 million in 2011 with varied growth rates in various cities (Census of India, 1981, 1991, 2001, 2011). The urban population of Nepal accounts for less than half (42 %) of the total population (2.86 million persons), where only two cities have a population of more than a million (Census of Nepal, 2011; UNDP, 2014). These cities with a million‐ plus population in Nepal share an insignificant proportion (5.6%) of the total population in 2015. The urban population in Bhutan is a significant proportion (38%) of the total population although there is no city having a million‐plus population (Census of Bhutan, 2011). 3.2. ­URBAN SYSTEM IN CHANGING CLIMATIC CONDITIONS Climate change and the growing cities influence each other from the macro to the micro scale with varied intensity. Urban centers are drivers of global warming and are affected by climate change through rapid land transformation, energy consumption, and greenhouse gas (GHG) emissions (Lankao et al., 2008; Anand & Seetharam, 2011). Apart from this, the impact of cities on the local environment is determined by the design of cities, urban structure, economic activities and development pattern, population growth and size, traditions and culture, variability in meteorological conditions, and so on (EEA, 2012; Hebbert et  al., 2011; Carter et  al., 2012). GHG e missions and land use transformation are the ­ most significant anthropogenic influences on climate (Mahmoud & Gan, 2018). The urban area is the major source of heat due to the complex topography and mass

URBAN RISK AND RESILIENCE TO CLIMATE CHANGE AND NATURAL HAZARDS  35

N 1,100 KM

Population (in thousands) < 1,600

Year 1,600 to 3,500

> 15,000

3,500 to 15,000

1985

2015

Physiographic zones Western Himalayas

Desert

South Deccan

Eastern plains

Eastern Ghats

North-East range

East Deccan

Eastern Himalayas

East coast

Western Ghats

Central highlands

North Deccan

Northern plains

West coast

Central Himalayas

Figure 3.1  Spatiotemporal distribution of cities with million‐plus population (2015) in part of the Indian subcontinent (India, Nepal, and Bhutan) representing urban growth patterns in major cities in varied physiographic regions during 1985–2015.

of buildings, replacement of pervious surfaces, and energy consumption (Gartland, 2008; Smith et al., 2009), inducing rise in land surface temperature in urban regions as compared with its suburbs and rural area (Aikawa et  al., 2008). The decrease / absence of green cover in urban areas decreases the evapotranspiration and hence the latent heat flux. Further, the radiative properties of the urban environment are found to be distinctly different and to absorb more radiation due to the nature of the urban canopy (Huong & Pathirana, 2013). These changes in surface heat budget provide the atmospheric conditions over urbanized areas with some unique characteristics compared with those of pristine or rural areas (Shepherd et  al., 2002; Pathirana et  al., 2014). These changes can have a significant impact on the local circulation and meteorological parameters and their association with precipitation (Aikawa et al., 2008).

The increase in impervious surfaces due to urbanization causes an increase in flooding frequency due to poor infiltration and high runoff (Huong & Pathirana, 2013). Floods are considered the greatest and most frequent natural hazard that threatens the world’s largest cities (UNISDR, 2018), and vulnerabilities in developing country contexts especially are compounded by historical socioeconomic inequalities and less durable infrastructures (Satterthwaite et al., 2010). Urban areas especially suffer from a comparatively high flood risk due to their high population number and density, multiple economic activities, and many infrastructure and property values (Pelling, 2003). The climate change impacts have been observed as increased incidences and intensity of extreme weather events on various levels of urbanization at the local to global scale, particularly cities in developing countries and coastal regions (Anand & Seetharam,

36  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

2011). Mostly the marginalized groups in low and middle income nations, with negligible contributions to climate change, are most at risk from major direct and indirect impacts of climate change (Satterthwaite et  al., 2007). While cities contribute toward the changing climate, climate change and global warming seriously affect city life, health, and the infrastructure (Anand & Seetharam, 2011). The impact of climate change is evident in the form of sea level rise, storm surge, intense rainfall events, water supply and quality, lack of sanitation, heat waves, cold waves, and so on (CED, 2010), that individually and collectively pose serious threats to urban infrastructure, quality of life, and existence of urban systems (World Bank & UDLG, 2010). The geographic locations (coastal zones) of a majority of urban centers are in the areas that make cities more vulnerable to adverse climate change events. The combined impact of rapid expansion of impervious surfaces, land use transformation, industrial processes, transportation, and high energy consumption induced anthropogenic carbon dioxide emissions particularly in the urban centers (Oke, 1981; Roth, 2002; Grimmond, 2007). The cumulative impact leads to higher air temperature in urban areas than the rural surroundings and is generally referred to as the urban heat island (UHI) effect (Santamouris, 2001; Seto & Kaufmann, 2009; Taniguchi, 2007). Studies report temperature elevations of around 2°C in the majority of cities, 5–7ºC commercial and high‐density residential areas (Bonan, 2002), 5°C in urban soil (Reiter 2006) with some exceptional cases of 8–10ºC (Givoni, 1998). The warming periods are becoming more frequent and longer as compared with previous (Lau & Nath, 2012; Meehl & Tebaldi, 2004), which interacts nonlinearly with UHIs to produce extremely high heat stress for urban ­ residents and increased morbidity and mortality (Shahmohamadi et  al., 2011; Li & Bou‐Zeid, 2014). The global climate change and heat loss from buildings contributing to changes of the temperature profiles at local and regional scales is difficult to discriminate (Gunawardhana et  al., 2011). The influence of increasing UHI induces intensive consumption of energy for cooling urban environments artificially that contributes to high GHG emissions (Shahmohamadi et  al., 2011). The cumulative impact contributes to global warming, which is reported to cause glacier melting and sea level rise. Approximately 360 million urban residents live in coastal areas less than 10 m above sea level and are vulnerable to frequent flooding, tsunami, and storm surges (Satterthwaite & Moser, 2008). Fifteen of the 20 megacities in the world are at risk from rising sea levels and coastal surges (Oliver-Smith, 2009). The IPCC predicts a rise in average sea level over the next 100  years ranging between 13–28 cm in a low scenario and 26–59 cm in a high scenario (IPCC, 2007).

3.3. ­URBAN HAZARDS AND RISK The development of human civilization is primarily associated with urban areas as these areas are typically more productive than rural areas and are centers of innovation, development, and growth (Cohen, 2006). The growth of cities is therefore not merely considered as an adverse footprint on natural surfaces. On the other hand, extensive urbanization or indiscriminate growth has led to serious land use problems that influence local to regional climatic variables (Kumar & Pandey, 2013, 2016; Chan, 2017). The significant changes in land surface and atmospheric composition due to anthropogenic influences increase the susceptibility of the cities to natural disasters (Oke, 1981). Most of the urban centers around the world having larger populations are at risk from the effects of climate change and natural disasters (Jayakody et al., 2016). Due to certain geographic locations, urban areas are already exposed to multiple hazards, including meteorological as well as geological, and hazards which are expected to aggravate due to a large population concentration and intensified anthropogenic activities (Wilbanks et  al., 2007). Rapid population growth together with rapid urbanization and frequent migration in d ­eveloping countries are closely linked to intense poverty of large urban populations living in hazard‐ prone areas or relatively high‐risk zones (UNISDR, 2016). This pushes an already stressed urban system beyond a threshold of s­ustainability (Wilbanks et  al., 2007; Chan, 2017). All megacities are exposed to long‐term to short/ immediate occurring natural hazards ranging from geological (earthquake ground shaking and mass movements) to meteorological (floods and storms), and extreme climatic events (heat and cold waves and wildfires), that necessitate adoption of different risk reduction strategies for varying conditions in megacities (UNPD, 2010). Asian population giants (China and India) are the worst affected countries by weather‐related disasters, as more than three billion (75%) of the population was affected by various disasters during 1995–2015 (CRED & UNISDR, 2015). Indian megacities such as Delhi, Hyderabad, Ahmedabad, Surat, and Indore, have settled and developed along river banks, while others like Chennai, Mumbai, Kolkata, Vishakhapatnam, Kochi, and Bhubaneshwar, urbanized as trade centers along coasts because of expedient site and situation, are prone to various kinds of hazards due to the diverse physiographic locations and meteorological variability (Pandve, 2010; Parikh et al., 2013). A majority of the ­cities experience multiple hazards like tsunami, storm surge, flood, sea level rise, earthquake, as well as high ­ surface temperature. Many of the Indian cities are at the stage of metropolis and megalopolis. The physical growth of cities in India is

URBAN RISK AND RESILIENCE TO CLIMATE CHANGE AND NATURAL HAZARDS  37

driven by increase in population and urbanization, that often occurs at the cost of productive lands, leading to more or less haphazard, unplanned, and unregulated urban growth (Reddy, 1996; Ramachandran, 1999). This has been widening the large gaps between necessary urban ecosystem demands and supply services (Mundu & Bhagat, 2008). The urban population in India has increased from 0.026 billion in 1901 (10.84% of global population) to 0.39 billion in 2011 (31.95%) indicating colossal pressure on the existing available resources (Census of India, 2011). Urban risk is a combined measure of the expected losses at varied scales due to hazard events of varied intensity occurring in an urban area over a specific time period (UNISDR, 2009). Cities are complex and interdependent systems, extremely vulnerable to risks both from natural and human‐made hazards (Godschalk, 2003). Most of these losses occurred at locations where vulnerable urban settlements were developed near known hazard areas, such as floodplains, earthquake fault zones, and hurricane prone shorelines (Godschalk, 2003). As per EEA (2009), the urban area is exposed to multiple hazards due to the intrinsic and complex nature of the urban environment and growth patterns influencing varied aspects of nature. Approximately, 2.4 million properties are at risk of flooding from rivers and the sea, with the significant influence in the urban areas (EEA, 2009). In Asia, 13% of the global urban population lives in low‐elevation coastal zones (LECZs) (McGranahan et  al., 2007) leaving important socioeconomic and biophysical resources to the threat of future climate change, due to changing climatic patterns and extreme meteorological events like floods (Jameson & Baud, 2016). The multihazard assessment exhibits that flood is a most prominent natural hazard in the Indian subcontinent, and the million‐plus cities in the eastern Indo-Gangetic Brahmaputra Plain (IGBP) region, followed by western coast regions that, were significantly affected by a high number of hazard events during 1980 to 2010 (Figures 3.2 and 3.3). The urban floods in cities like Mumbai, Chennai, and Srinagar have become regular, and increasingly devastating (Sengupta, 2016). Severe urban floods have been reported in Indian history among which Mumbai is worst affected. The extreme rainfall event of 994 mm on 26 July 2005 for Mumbai has been classified as “very heavy” (>200 mm/day as per the IMD’s criteria for classification of rainfall) and it indicated the perils of rapid development in highly concentrated urban areas (Rafiq et  al., 2016). The heavy monsoon rainfall during 2018 induced severe flooding in many regions in India, including the central highlands (Madhya Pradesh) and eastern IGB plains (Assam, UP, Uttrakhand), with the worst in the western coastal regions (Kerala) during July and August. The EM‐DAT based during decadal hazard frequency (1980s to 2010s) in million‐plus

population c­ ities indicates that the maximum flood events recorded in the 2010s (followed by 1990s, 2000s, and 1980s in different parts of the country) led to socioeconomic damage in different cities (Figures 3.2 and 3.3). The study indicates that the flood hazard is a higher risk in cities located in the eastern Indo‐Gangetic plain (IGP) (Guwahati, Darbhanga, Patna, Bardhhaman, Kolkata, etc.), parts of northern IGP (Gorakhpur, Varanasi), western and eastern coast  (Surat, Mumbai, Kannur, ­ Kottayam, Thiruvananthapuram, Guntur, etc.), and parts of western Ghats. The flood risk is moderate to low in the western and central Himalayan region (Srinagar, Jammu, Kathmandu, etc.), central highlands (Bhopal, Jabalpur, Jaipur, etc.), parts of the south Deccan plateau (Warangal, Hyderabad, etc.), western desert regions (Jodhpur, Ajmer, etc.), and parts of eastern Ghats (Coimbatore). However, flood is also one of the prominent hazards in the eastern Himalayan region as well as northeastern range, but due to insignificant population density and urban growth in these regions, the hazard risk was insignificantly low. Tsunami is another most destructive hazard caused by sudden movement on the ocean floor, which generally results from an undersea earthquake. In the Indian LECZ, Chennai is a rapidly growing city that is by now highly sensitized to floods due to the tsunami in 2004 and massive flooding in 2005 (Kennedy et  al., 2014). The decadal hazard frequency (EM‐DAT based) in million‐ plus cities from the 1980s to 2010s indicates that tsunami occurred once – in the year 2004 – in the last four decades (1980s–2010s) affecting the eastern coast of India and causing significant loss of life and property in cities located on the eastern coast (for example, Chennai, Vishakhapatnam, and Kakinada). It shows that the risk of tsunami hazard is higher in coastal plains as well as eastern and western ghats. Cyclone is another of the most destructive and frequent hazards in the region and the cities in the Indian subcontinent are largely affected by cyclonic events due to their geographic proximity to the Bay of Bengal, Arabian Sea, and Indian Ocean. Although, the frequency of cyclones in the Bay of Bengal is about five times higher than the Arabian Sea (IMD, 1979), the Indian east coast is more at risk through the north Indian Ocean basins, which induce one of the least intense cyclone/ hurricane basins in the world (Revi, 2008). Cyclones and storm surge have a devastating impact on large coastal urban centers, namely Mumbai, Chennai, and the million‐plus cities of Vishakapatnam, Surat, Bharuch, Bhavnagar, and Jamnagar, as well as causing critical hindrances in important ports such as Kandla (Kumar et  al., 2006; Taru, 2005). Cyclones Phailin (October 2013) and Hudhud (category 3 in October 2014) were the most severe cyclones that hit the regions in Odisha and Andhra Pradesh states and were followed by heavy rainfall and

38  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

N 1,000 KM

Cumulative frequency of hazard events Nil 5 Flood Nil 5 Drought Nil 5 Cyclone Earthquake Nil 5 Landslide Nil 5 6.5) in the

Indian cities located in Kangra (1905), Bihar (1934), Assam (1950, 1987), Uttarkashi (1991), Latur (1993), Chamoli (1999), Bhuj (2001), Kashmir (2005), Sikkim (2011), and others (Sorkhabi, 2006; www.ndma.gov.in). The earthquake, rainfall events, and rapid developmental activities influenced some of the major landslides in the Himalayan region that had serious casualties, deaths, and damage. Major landslide devastation was recorded in Assam (1991), Itanagar and Kalimpong (1993), Kashmir (1994), Jammu and Kullu (1995), Uttarkashi (2003), Joshimath‐Badrinath and Tehri dam (2004), and northern Sikkim (2012) (www.dnaindia.com). The decadal hazard frequency (EM‐DAT based) in million‐ plus cities during the 1980s to 2010s indicates that the maximum earthquake events were observed in the 2010s followed by the 2000s, 1990s, and 1980s. The earthquake hazard induced higher risk in cities located on the western coast (Ahmedabad, Vadodara, Surat, etc.),

Frequency of disaster events

25 20 15 10 5 0

0 Asansol Bilaspur Bokaro Cuttack Dhanbad Durg-Bhilainagar Jamshedpur Korba Raipur Ranchi Raurkela Ahmadnagar Akola Amravati Aurangabad Chandrapur Dhule Jalgaon Latur Malegaon Nagpur Nanded Waghala Parbhani Pune Anantapur Belgaum Bellary Bijapur Davangere Gulbarga Hubli-Dharwad Hyderabad Ichalakaranji Karimnagar Kolhapur Mysore Nizamabad Sangali Shimoga Solapur Tumkur Warangal

Frequency of disaster events Agra Aligarh Allahabad Amritsar Bareilly Bathinda Chandigarh Delhi Farrukhabad Firozabad Gorakhpur Hardwar Jalandhar Kanpur Lucknow Ludhiana Mathura Maunath Bhanjan Meerut Moradabad Muzaffarnagar Panipat Patiala Rampur Rohtak Roorkee Saharanpur Shahjahanpur Varanasi Yamunanagar Baharampur Barddhaman Begusarai Bhagalpur Bihar Sharif Darbhanga Durgapur English Bazar Gaya Guwahati Habra Kolkata Muzaffarpur Patna Purnia Siliguri

0

E. Coast

Bhubaneswar Brahmapur Chennai Guntur Kakinada Madurai Nellore Puducherry Rajahmundry Ranipet Thanjavur Thiruvananthapuram Tiruchirappalli Tirunelveli Vellore Vijayawada Visakhapatnam Ahmedabad Anand Bhavnagar Bhiwandi Cherthala Jamnagar Junagadh Kannur Kayamkulam Kochi Kollam Kottayam Kozhikode Malappuram Mangalore Mumbai Navsari Rajkot Surat Thrissur Vadodara Bangalore Coimbatore Dindigul Erode Hosur Kadapa Kurnool Palakkad Salem Tirupati Tiruppur Nashik

Frequency of disaster events

Frequency of disaster events

0

Flood W.Him.

Drought C. Him.

E Deccan

Cyclone

Earthquake

NE Range

Landslide

N Plains

N Deccan

W.Coast

C. Highlands

E Ghats

Jodhpur

Hisar

Bikaner

Ajmer

Ujjain

Udaipur

Satna

Sagar

Kota

Jhansi

Jaipur

Jabalpur

Indore

Gwalior

Dewas

Bhopal

Bhilwara

Alwar

Shillong

Imphal

Aizawl

Agartala

Pokhra

Kathmandu

Srinagar

Jammu

Dehradun

URBAN RISK AND RESILIENCE TO CLIMATE CHANGE AND NATURAL HAZARDS  39

25 20

15

10

5

Desert

25

20

15

10

5

E Plains

S Deccan

25

20

15

10

5

W. Ghat

Cities in different physiographic zones

Tsunami

Figure 3.3  Frequency of major disaster events affecting million‐plus cities located in varied physiographic regions in the Indian subcontinent from 1980 to 2010 (Physiographic region: W = Western; E = Eastern; N = Northern; S = Southern; C = Central; Him = Himalayas; G = Ghats).

40  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Himalayan region (Srinagar, Kathmandu, etc.), eastern IGBP (Kolkata, Guwahati, Siliguri, Darbhanga, etc.), and northern plains (Delhi, Kanpur, Lucknow, etc.). The risk is moderate to low in cities located in northeastern ranges (Shillong, Imphal, etc.) central Highlands (Jabalpur, Bhopal, Indore, etc.), northern Deccan plateau (Latur, Nagpur, etc.), and the east coast (Chennai, Vishakhapatnam, etc.). Landslides caused by earthquake, excessive rainfall, and instability of slope have been significantly affecting the mountainous region. The decadal hazard frequency (EM‐DAT based) in million‐plus cities during the 1980s to 2010s indicates that the maximum ­frequency of landslides events was in the 1990s followed by the 2000s, 1980s, and 2010s (Figures  3.2 and 3.3). Although the landslides occur mostly in mountainous regions having insignificant population growth, the impact of the landslide hazard has also been less significant in major urban regions. However, the landslides hazard induced higher risk in cities located in the western and eastern Himalayan region (Srinagar, Jammu, etc.) followed by the northeastern range (Agartala etc.), parts of the western coasts, and Ghats (Mumbai etc.). A significant part of India and its neighboring countries is influenced by the drought hazard‐risk as a result of extreme climate change conditions. The drought impact is primarily evident in the hydrological conditions, agricultural land, and productivity, and its influence is also observed in the urban milieu. The decadal hazard frequency (EM‐DAT based) in million‐plus cities during the 1980s to 2010s indicates that the maximum drought events and losses due to the effect of El Nino were during the 2000s followed by the 1980s, 1990s, and 2000s in different physiographic regions (Figures  3.2 and 3.3). ­ Drought has induced higher risk in cities located in the western deserts (Jodhpur, Ajmer, etc.), followed by the western coast (Ahmedabad, Rajkot, Vadodara, etc.), western Ghats (Nashik etc.), northern plains (Jaipur, Delhi, Lucknow, Kanpur, etc.), Deccan plateau (Aurangabad, Pune, Hyderabad, etc.), and central highlands (Jaipur, Gwalior, Indore, etc.). 3.4. ­URBAN RESILIENCE AND ADAPTATION The notion of resilience is increasingly gaining prominence in the literature on cities and climate change. Resilience is the amount of disturbance an urban system, community, or individual can absorb after the occurrence of any disaster, irrespective of its impact, frequency, or magnitude (Folke, 2006; Frantzeskaki, 2016). Although resilience has been explored in many complex socioecological systems, it has only recently been applied in the context of cities (Folke et al., 2002; Ernstson et al., 2010; Pelling, 2003; Boyd et  al., 2008; Sperling et  al., 2008; ECAWG, 2009). Enhancement of resilience is widely cited

as a key goal for both adaptation and mitigation efforts in cities and urban regions (Crichton, 2007; Muller, 2007; Revi, 2008; Sanchez‐Rodrıguez, 2009). Urban resilience exhibits coping capability with stresses and disturbances caused by external factors (Davic & Welsh, 2004). The domain of urban resilience studies is broadly classified into four categories, namely, (1) urban ecological resilience, (2) urban hazards and disaster risk reduction, (3) resilience of urban and regional economies, and (4) promotion of resilience through urban governance and institutions (Leichenko, 2011). An ideal resilient city may be referred to as having sustainable infrastructure, efficient governance, and capable dwellers to absorb external pressures or to adapt or transform in front of such pressures (Papa et  al., 2015). The inefficiencies of urban systems hinder their ability to adapt to climate change and affect the city resilience to climate change (Parikh et al., 2013). Urban planning, therefore, will need to work increasingly at urban and periurban scales as well as also regional scales while considering responsibility for the global connectivity and resource imprint of cities that influence the ability of cities to improve resilience and enable sustainability transitions (McPhearson et  al., 2015). The resilience indicates toward a specific city hazard action plan for the short term as well as long term in order to enhance the coping capability. Cities are homes to the future; they presently accommodate more than 50% of world population. The major challenge is to make cities more sustainable and resilient in terms of socioeconomic as well as environmental footprint and most important with respect to urban hazards (Frantzeskaki, 2016). Development of climate change adaptation as well as a disaster resilience framework is required for urban India in order to reduce the risk generated by climate change, natural hazards, and haphazard urban growth, so that it can transform existing cities to be more inclusive and productive and also reduce structural vulnerability (Revi, 2008). Therefore, it is increasingly essential to plan and design the cities with a focus on disaster resilience. Studies reported that 11 major Asian cities most likely to be affected by climate change include Dhaka (Bangladesh), Jakarta (Indonesia), Manila (Philippines), Kolkata (India), Phnom Penh (Cambodia), Ho Chi Minh city (Vietnam), Shanghai (China), Bangkok (Thailand), Hong Kong (China), Kuala Lumpur (Malaysia), and Singapore (UNPD 1993; Nicholls, 2003; Satterthwaite et  al., 2007; WWF, 2009). Thus, an adaptation strategy is indispensable to increase cities’ resilience to climate change. Adaptation to climatic variability includes measures to reduce vulnerability for immediate climatic impacts (with or without climate change). Individual adaptation (including reduction in the use of air conditioning, vehicles) can contribute to reduction in GHG emissions and

URBAN RISK AND RESILIENCE TO CLIMATE CHANGE AND NATURAL HAZARDS  41

make urban centers more livable (Lankao et al., 2008). Mitigation of greenhouse gas emissions or enhancement of greenhouse sinks will reduce future global warming and sea level rise (Nicholls, 2003). Recent analyses suggest that global mean sea level rise is almost independent of future emissions to 2050, and future emissions become most important in controlling sea level rise after 2100 (Church et al., 2001). The concept of rooftop greening, urban green spaces, and so on, which can incorporate carbon sink zones or prevent heat trapping from the surfaces, should be promoted. The vulnerability profile of the urban population due to climate change is mostly uneven (Bartlett et al., 2009). People with high adaptive capacity will be less vulnerable but people residing in informal settlements and low income are most vulnerable (Winchester & Szalachman, 2009). These populations have less adaptive capacity to deal with the impacts of climate change because of poor governance and infrastructure facilities (Revi, 2008). Recent research highlights an urgent need to improve our understanding and action on climate variability and adaptation in urban areas as an urgent priority, particularly where poverty levels and population growth rates are highest (Huq et al., 2007). While urban centers are the major hotspots for climate risks, they are also the hubs of development, and sources of innovations and policies, to reduce the emissions of GHGs and increase our capability to cope with climatic hazards. Carbon storage in urban areas has a contribution comparable to tropical forests, where carbon is found in soils, vegetation, landfills, and the structures and contents of buildings (Churkina et al., 2010). Urban and suburban areas tend to be net sources of carbon to the atmosphere, whereas exurban and rural areas tend to be net sinks (Zhao et al., 2011). Effects of urban growth patterns on carbon storage and emissions are dependent upon land use planning policies and implementation. Many cities have adopted land use plans with explicit carbon goals targeting reducing carbon emissions through reduction in transportation and energy use (Brown et  al., 2014). Thus, urban areas with the combination of increased vulnerabilities and increased opportunities can develop important strategies and resources for creating innovative adaptation and mitigation measures (Lankao et al., 2008). Flood needs consistent measures for evaluating the flood resilience of vulnerable urban agglomerations. Resistance measures include construction of dikes, embankments, restoration of natural wetlands, and flood plains, can contribute in effective implementation of flood defense plans and flood risk mitigation. Besides physical measures, spatial planning is indispensable, which includes adaptation to the flood hazard for e­ xisting or planned residential areas, infrastructure, and hydraulic structures and fields (Doroszkiewicz & Romanowicz, 2017). The settlements

in flood prone areas should be equipped with high‐rise towers with food‐medical facilities for inundation situations; haphazard construction also needs be restricted in flood plains to minimize the impact of flood hazard. The duration and magnitude of the ground shaking and building materials are vital parameters linked with the extent of loss and damage due to an earthquake event. Since, the magnitude of an earthquake is a natural phenomena, earthquake‐resistant construction considering slope stability and seismic retrofitting is required to be adopted effectively. Seismic retrofitting is another method of altering the existing building structures in order to make them more resistant to seismic activities (NIDM, 2007). The construction of infrastructure should also be avoided in highly seismic zones. The experiences and coping measures from previous emergency events, traditional skills, and local environmental knowledge play an important part in reducing the impact of hazards (Paul, 2009). The impact of cyclone Aila in Bangladesh impacted as severe loss of life collapse of mud house walls, which were not cyclone resistant, collapsed; this led to adaptation of walls and the building of houses using the sticks from Goran trees/ bamboo with mud layers to make the walls more durable. People also raised the homestead base so that their houses did not get inundated during floods (Sultana & Mallick, 2015). Apart from building structural resilience, n ­ onstructural measures must be developed by authorities about land use planning, m ­ onitoring of construction work, and controlling of settlements in hazard‐prone areas to avoid fatalities and loss of property (NIDM, 2007). Risk reduction strategies for disasters vary from one region or community to another but the ideas from each strategy can help in building new strategies for the respective regions (Sultana & Mallick, 2015). Every individual interacts with the surroundings differently and observes the adverse effects of hazardous events depending on their own coping capability (Lummen & Yamada, 2014). Structural resilience needs to be emphasized for disaster‐ affected communities or regions by empowering the disaster management authorities, regional planners, policy makers, and governing bodies to build a sustainable risk reduction plan. 3.5. ­CONCLUSION The risk in urban areas is often more prominent and intense than that in rural areas due to the fact that 54% of the total global population resides in urbanized landmass. Climate change is a global phenomenon that is driven by land use change, energy consumption, and various other anthropogenic activities inducing alterations in the atmospheric composition, hydrological cycle, and carbon cycle as well. This chapter has addressed the

42  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

multiple linkages between climate change, hazard events in urbanized surfaces, and the intrinsic relationships leading to the increase in exposure to risk of urban dwellers emphasizing the Indian scenario. The rapid changes in the land use pattern such as urbanization along with high population growth leads to problems like deforestation, increase in greenhouse gases, lack of open spaces, slums, traffic congestion, lack of impervious surfaces, and improper drainage. A majority of the CO2 emission is contributed by buildings, pavements, and other concrete surfaces, followed by emission from vehicles, which is all together growing exponentially in India. Climate change also poses serious threats to the urban environment inducing many natural and human‐made disasters. The increase in the impervious surfaces and improper drainage in urban centers increases the vulnerability and risk to urban flooding, which is frequent in Indian cities. Climate change together with sea level rise would lead to a series of disasters such as flash floods, cloudbursts, disease outbreaks, epidemics, property damage, dislocation, and death. High temperatures and a moisture‐laden atmosphere would lead to high humidity, increasing the frequency of vector‐borne diseases. The impact of drought is a major threat from climate change affecting a large population and economy. The urban heat island is one of the most important emerging risks to the urban centers caused by temperature variations in the core and surroundings of the city. The scenario is often more destructive in terms of urban hazard and risk in developing nations as the low‐ income and middle‐income nations lack basic amenities, proper planning strategies, technology, and, most important, sound economic systems to deal and cope with increasing hazard and risk. In the last four decades, the frequency of natural disasters recorded has increased almost threefold, from over 1,300 events in 1975–1984 to over 3,900 in 2005–2014. It poses particular threats to life, urban infrastructures, energy supplies, and transport connectivity to a large extent. The risk multiplies because of poor adaptation and resilience capability due to poverty, poor governance, and degraded infrastructure, which further increase the severity of disaster impact on communities and populations. Resilience must be adapted to the urban environment in order to make the region sustainable and more livable. Social mobilization awareness in synergy with an organizational framework must be improved in light of increased hazard and risk. Various geospecific solutions to multihazards need to be devised for individuals, in order to improve the resilience capacity of urban dwellers to cope with the changing climate conditions. Also, methods need to be devised to control the impact of climate change. Although c­ontrolled and managed urbanization with sustainable rationale can be

induced to yield positive effects in a longer time frame, construction of resilient cities that focuses on the frequency of hazard events in these cities as well as their physiographic region is needed. ACKNOWLEDGMENTS The authors gratefully acknowledge the Emergency Events Database (EM-DAT) launched by the Centre for Research on the Epidemiology of Disasters (CRED). ­REFERENCES Aguilera, F., Luis Valenzuela, M., & Botequilha‐Leitao, A. (2011). Landscape metrics in the analysis of urban land use patterns: A case study in a Spanish metropolitan area. Landscape and Urban Planning, 99 (3–4), 226–238. Aikawa, M., Hiraki, T., & Eiho, J., (2008). Change of atmospheric condition in an urbanized area of Japan from the viewpoint of rainfall intensity. Environmental Monitoring and Assessment, 148, 449–453; doi:10.1007/s10661‐008‐0174‐0. Anand, P., & Seetharam, K. (2012). Climate change and living cities: Global problems with local solutions. In B. Yuen and A. Kumssa (Eds.),Climate change and sustainable urban development in Africa and Asia; https://journals.co.za/ content/aref/3/2/EJC125224. Asian Development Bank (2015). Global increase in climate‐ related disasters. Working Paper. Manila, Philippines. Bartlett, S., Dodman, D., Hardoy, J., Satterthwaite, D., & Tacoli, C. (2009). Social aspects of climate change in urban areas low‐ and middle‐income nations. http://www.dbsa.org/ Vulindlela/Presentations/Session1_Dodman_old.pdf. Bonan, G. (2002). Ecological climatology. Cambridge University Press. Boyd, E., Osbahr, H., Ericksen, P., Tompkins, E., Lemos, M., & Miller, F. (2008). Resilience and “climatizing” development: Examples and policy implications. Development 2008, 51, 390–396. Brown, D. G., Polsky, C., Bolstad, P., Brody, S. D., Hulse, D., Kroh, R., et al. (2014). Land use and land cover change. In J. M. Melillo, T. Richmond, and G. W. Yohe, Eds. (pp. 318– 332), Climate change impacts in the United States: The third national climate assessment, U.S. Global Change Research Program,; doi:10.7930/ J05Q4T1Q; https://nca2014. globalchange.gov/report/sectors/ land‐use‐and‐land‐cover‐change. Carter, J. G., Connelly, A., Handley, J., & Lindley, S. (2012). European cities in a changing climate: Exploring climate change hazards, impacts and vulnerabilities. The University of Manchester, Manchester, UK. CCC (Committee on Climate Change) (2010). Building a low‐ carbon economy–the UK’s innovation challenge. http://www. theccc.org.uk. CED (2010). Impact of climate change on urban areas in India. http://base.d‐p‐h.info/fr/fiches/dph/fiche‐dph‐8632.html.

URBAN RISK AND RESILIENCE TO CLIMATE CHANGE AND NATURAL HAZARDS  43 Census of Bhutan (2011). National Statistics Bureau, Royal Government of Bhutan. Census of India (1981, 1991, 2001, 2011). Primary Census Abstract. Government of India. Census of Nepal (2011). National Population and Housing Census 2011. National Planning Commission Secretariat. Central Bureau of Statistics Kathmandu, Government of Nepal. Chan, N.W. (2017). Urbanization, climate change and cities: Challenges and opportunities for sustainable development. Keynote paper presented at the Asia‐Pacific Chemical, Biological & Environmental Engineering Society (APCBEES), International Conference, 9 January 2017, School of Humanities, Universiti Sains Malaysia, Penang, Malaysia. Church, J. A., Gregory, J. M., Huybrechts, P., Kuhn, M., Lambeck, K., Nhuan, M. T., et  al. (Eds.) (2001). Climate change 2001. The scientific basis (pp. 639–693). Cambridge: Cambridge University Press. Churkina, G., Brown, D., & Keoleian, G. (2010). Carbon stored in human settlements: The conterminous United States. Global Change Biology, 16, 135–143; doi:10.1111/j.1365 2486.2009.02002.x. Clark, A. L. (2009). Environmental challenges to urban planning: Fringe areas, ecological footprints and climate change. Proceedings of the Workshop of Governance, Socioeconomic and Environmental Issue, 16–18 September, Ho Chi, Minh City, Vietnam, 1–8. Cohen, B. (2006). Urbanization in developing countries: Current trends, future projections, and key challenges for ­sustainability. Technology in Society, 28, 63–80. CRED & UNISDR (2015). The Human Cost of Weather‐ Related Disasters 1995–2015, Geneva. Crichton, D. (2007). What can cities do to increase resilience? Philosophical Transactions of the Royal Society A: Mathematical Physical and Engineering Sciences, 365, 2731–2739. Davic, R. D., & Welsh, H. H. Jr. (2004). On the ecological roles of salamanders. Annual Review of Ecology, Evolution, and Systematic, 35, 405–434. Dickson E., Baker, J. L., Hooranweg, D., & Tiwari, A. (2012). Urban risk assessments: Understanding disaster and climate risk in cities. Washington, DC: World Bank. Diksha & Kumar, A. (2017). Analysing urban sprawl and land consumption patterns in major capital cities in the Himalayan region using geoinformatics. Applied Geography, 89, 112–123. Available at http://dx.doi.org/10.1016/j.apgeog.2017.10.010. Doroszkiewicz, J., & Romanowicz, R. J. (2017). Guidelines for the adaptation to floods in changing climate. Acta Geophysica, 65(4), 849–861. Duranton, G. (2008). Cities: Engines of growth and prosperity for developing countries? The International Bank for Reconstruction and Development. Washington, DC: World Bank. ECAWG (Economics of Climate Adaptation Working Group) (2009). Shaping climate resilient development: A framework for decision‐making. http://www.mckinsey.com/App_Media/ Images/Page_Images/Offices/SocialSector/PDF/ ECA_ Shaping_Climate%20Resilent_Development.pdf.

EEA (European Environment Agency) (2009). Water resources across Europe  –  confronting water scarcity and drought, EEA Report No. 2/2009, European Environment Agency, Copenhagen, Denmark. EEA (European Environment Agency) (2012). Climate change, impacts and vulnerability in Europe 2012. EEA Report No 12/2012, European Environment Agency, Copenhagen, Denmark. EM‐DAT: The Emergency Events Database, Université catholique de Louvain (UCL) ‐ CRED, D. Guha‐Sapir; www. emdat.be, Brussels, Belgium. Ernstson, H., van der Leeuw, S., Redman, C., Meffert, D., Davis, G., Alfsen, C., et  al. (2010). Urban transitions: On urban resilience and human‐dominated ecosystems. AMBIO: A Journal of the Human Environment, 10. doi.org/1007/ s13280‐010‐0081‐9. Folke, C. (2006). Resilience: The emergence of a perspective for social‐ecological systems analyses. Global Environmental Change, 16, 253–267 Folke, C., Carpenter, S., Elmqvist, T., Gunderson, L., Holling, C. S., Walker, B., et  al. (2002). Resilience and sustainable development: building adaptive capacity in a world of transformations. Scientific Background Paper on Resilience for the Process of the World Summit on Sustainable Development on Behalf of the Environmental Advisory Council to the Swedish Government. Frantzeskaki, N. (2016). Urban resilience: A concept for co‐­ creating cities of the future. Resilient Europe, Urbact, European Union. Fulton, W., Hollis, L. E., & Williamson, C. (2006). The shape of metropolitan growth: How policy tools affect growth patterns in Seattle and Orlando. Washington, DC: Brookings Institution. Gandhinagar Gunawardhana, L. N., Kazama, S., & Kawagoe, S. (2011). Impact of urbanization and climate change on aquifer thermal regimes. Water Resource Management, 25, 3247–3276; doi:10.1007/s11269‐011‐9854‐6. Gartland, L. (2008). Heat islands: Understanding and mitigating heat in urban areas. Washington, DC: Earthscan. Givoni, B. (1998). Climate considerations in building and urban design. Canada: John Wiley & Sons. Godschalk, D. R. (2003). Urban hazards mitigation: Creating resilient cities. Natural Hazards Review, 4 (3), 136–143. Grimm, N. B., Faeth, S. H., Golubiewski, N. E., Redman, C. L., Wu, J. G., Bai, X. M., & Briggs, J. M. (2008). Global change and the ecology of cities. Science, 319, 756–60. Grimmond, S. (2007). Urbanization and global environmental change: Local effects of urban warming. Geographical Journal, 173 (1), 83–88. GSDMA/TARU (2005), Gujarat Vulnerability and Risk Atlas. Hebbert, M., Jankovic, V., & Webb, B. (Eds.) (2011). City weathers: Meteorology and urban design, 1950–2010, Manchester Architecture Research Centre, Manchester. Retrieved from http://www.sed.manchester.ac.uk/architecture/ research/csud/workshop/2011CityWeathers.pdf. Huong, H. T. L., & Pathirana, A. (2013). Urbanization and climate change impacts on future urban flooding in Can Tho city, Vietnam. Hydrology and Earth System Sciences, 17, 379–394.

44  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Huq, S., Kovats, S., Reid, H., & Satterthwaite, D. (2007). Editorial: Reducing risks to cities from disasters and climate change. Environment and Urbanization, 19 (1), 3–15. IFRC (2014). India: Cyclone Hud‐Hud, Information Bulletin No. 2, International Federation of Red Cross and Red Crescent Societies;https://reliefweb.int/report/india/india‐cyclone‐hudhud‐ information‐bulletin‐n‐2. India Metrological Department (1979). Tracks of storms and depressions in the Bay of Bengal and the Arabian Sea, 1877 to 1970. New Delhi. IPCC (2007). Climate change 2007: Fourth assessment report of the intergovernmental panel on climate change. Cambridge, MA: Cambridge University Press. Jameson, S., & Baud, I. S. (2016). Varieties of knowledge for assembling an urban flood management governance configuration. Chennai, India. Habitat International, 54, 112–123. Jayakody, R. R. J. C., Amaratunga, D., & Haigh, R. (2016).The use of public open spaces for disaster resilient urban cities. In Proceedings of the 12th International Conference of the International Institute for Infrastructure Resilience and Reconstruction (IIIRR) (pp. 146–152). IIIRR. Kennedy, L., Varrel, A., Denis, E, Dupont, V., Dhanalakshmi, R., Roumeau, S., et al. (2014). Engaging with sustainability issues in metropolitan Chennai. Chance2Sustain City Report: Chennai. Chance2Sustain City Report Series; http://chance2 sustain.eu/66.0.html. Kumar, A. (2016). Urban footprints on environment: A geoinformatics approach. In M. K. Guar, C. B. Pandey, & R. K. Goyal (Eds.), Remote sensing for natural resources management and monitoring (pp. 339–348). India: Scientific Publishers. Kumar, A., & Pandey, A. C. (2013). Spatio‐temporal assessment of urban environmental conditions in Ranchi township using remote sensing and GIS techniques. International Journal of Urban Sciences, 17(1), 117–141. Kumar, A., & Pandey, A. C. (2016). Geoinformatics based groundwater potential assessment in hard rock terrain of Ranchi urban agglomeration, Jharkhand (India) using MCDM‐AHP techniques. Groundwater for Sustainable Development. Elsevier; doi: 10.1016/j.gsd.2016.05.001. Kumar, V. S., Pathak, K. C., Pednekar, P., Raju, N. S. N., & Gowthaman, R. (2006). Coastal processes along the Indian coastline. Current Science, 91 (4). Lankao, P. R., Nychka, D., & Tribbia, J. L. (2008). Development and greenhouse gas emissions deviate from “modernization“ and “convergence.” Climatic Change, 38, 17–29. Lau, N‐C., & Nath, M. J. (2012). A model study of heat waves over North America: Meteorological aspects and projections for the 21st century. Journal of Climate, 25, 4761–4784. Leichenko, R. (2011). Climate change and urban resilience. Current Opinion in Environmental Sustainability 2011, 3, 164–168. Li, D., & Bou‐Zeid, E. (2014). Quality and sensitivity of high‐ resolution numerical simulation of urban heat islands. Environmental Research Letters, 055001; doi:10.1088/1748‐93 26/9/5/055001.

Luck, M., & Wu, J. (2002). A gradient analysis of urban landscape pattern: a case study from the Phoenix metropolitan region, Arizona, USA. Landscape Ecology, 17 (4), 327–339. Lummen, N. S., & Yamada, F. (2014). Implementation of an integrated vulnerability and risk assessment model. Natural Hazards, 73(2), 1085–1117. Mahmoud, S. H., & Gan, T. Y. (2018). Impact of anthropogenic climate change and human activities on environment and ecosystem services in arid regions. Science of Total Environment, 15 (633), 1329–1344. Mata-Lima, H., Alvino-Borba, A., Pinheiro, A., Mata-Lima, A., & Almeida, J. (2013). Impacts of natural disasters on environmental and socio-economic systems: What makes the difference?, Ambiente & Sociedade, 16 (3), 45–64. Mundu, G. B., & Bhagat, R. B. (2012). Civic amenities in slums of Mumbai, Urban India, 32(1), 1–17. McGranahan, G., Balk, D., & Anderson, B. (2007). The rising tide: Assessing the risks of climate change and human settlements in low elevation coastal zones. Environment & Urbanization, 19 (1), 17–37. McPhearson, T., Andersson, E., Elmqvist, T., & Frantzeskaki, N. (2015). Resilience of and through ecosystem services. Ecosystem Services, 12, 152–156; doi:10.1016/j.ecoser. 2014.07.012. Meehl, G. A., & Tebaldi, C. (2004). More intense, more frequent, and longer lasting heat waves in the 21st century. Science, 305, 994–997. Muller M. (2007). Adapting to climate change. Environment and Urbanization, 19, 99. National Institute of Disaster Management (NIDM) (2007). Preventive and mitigation measures. East Asia Summit Earthquake Risk Reduction Centre (EAS‐ERR), New Delhi. Available at http://nidm.gov.in/easindia2014/err/pdf/earthquake/ Prev_mitigation_measures.pdf. Nicholls, R. J. (2003). Case study on sea level rise impacts. Prepared for OECD Workshop on the Benefits of Climate Policy: Improving Information for Policy Makers, Organization for Economic Cooperation and Development, Paris, 12–13 December 2002. NPCC (2009). New York City panel on climate change: Climate risk information; http://www.nyc.gov/html/om/pdf/2009/ NPCC_CRI.pdf. Oke, T. R. (1981). Canyon geometry and the nocturnal urban heat island: comparison of scale model and field observations. Journal of Climatology, 1 (3), 237–254. Oliver-Smith, A. (2009). Sea Level Rise and the Vulnerability of Coastal Peoples. UNU Institute for Environment and Human Security (UNU-EHS): Germany. No.7/2009. Pandve, H. T. (2010). Climate change and coastal mega cities of India. Indian Journal of Occupational and Environmental Medicine, 14(1), 22–23; doi: 10.4103/0019‐5278.64612. Papa, R., Galderisi, A., Majelloa, M. C. V., & Sarettaa, E. (2015). Smart and resilient cities. TeMA Journal of Land Use Mobility and Environment, 8(1). Parikh, J., Jindal, P., & Sandal, G. (2013). Climate resilient urban development, vulnerability profiles of 20 Indian cities. COE‐IRADe, New Delhi.

URBAN RISK AND RESILIENCE TO CLIMATE CHANGE AND NATURAL HAZARDS  45 Pathirana, A., Denekew H.B., Veerbek, W., Denekew, H., Zevenbergen, C., & Banda, A. T. (2014). Impact of urban growth-driven landuse change on microclimate and extreme precipitation — A sensitivity study. Atmospheric Research, 138, 59–72. Paul, B. K. (2009). Human injuries caused by Bangladesh’s cyclone Sidr: An empirical study. Natural Hazards, 54(2), 483–495. Pelling, M. (2003). The vulnerability of cities: Social resilience and natural disaster. London: Earthscan, 209. Peng, X., Chen, X., & Cheng, Y. (2000). Urbanization and its consequences. Demography, 2; http://www.eolss.net/Eolss‐ Sampleallchapter.aspx. Rafiq, F., Ahmed S., Ahmad S., & Khan Amir Ali. (2016). Urban floods in India. International Journal of Scientific & Engineering Research, 7(1), 721–734. Ramachandran, R. (1999). Urbanization and urban system in India, 8th ed. New York: Oxford University Press. Rawat, P. K., Tiwari, P. C., & Pant, C. C. (2011). Climate change accelerating hydrological hazards and risks in Himalaya: A case study through remote sensing and GIS modeling. International Journal of Geomatic and Geosciences, 1 (4), 678–699. Reddy, K.N. (1996). Urban redevelopment: A study of high rise buildings. Reiter, M. (2006). Vadose Zone Temperature Measurements at a Site in the Northern Albuquerque Basin Indicate GroundSurface Warming due to Urbanization. Environmental and Engineering Geoscience, 12 (4), 353–360 Revi, A. (2008). Climate change risk: An adaptation and mitigation agenda for Indian cities. Environment and Urbanization, 20, 207 Roth M. (2002). Effects of cities on local climates. In Proceedings of the Workshop of IGES/APN Mega‐City Project, Kitakyushu, Japan, January 2002. Sanchez‐Rodrıguez, R. (2009). Leaning to adapt to climate change in urban areas. A review of recent contributions. Current Opinion in Environmental Sustainability, 1, 201–206. Santamouris, M. (2001). Energy and climate in the urban built environment. London, UK: James & James. Satterthwaite, D., Huq, S., Pelling M., Reid H., & Lankao P. R. (2007). Adapting to climate change in urban areas. Human Settlements Group and the Climate Change Group at the International Institute for Environment and Development (IIED). Satterthwaite, M. (2010). The contribution of cities to global warming and their potential contributions to solutions. Environment and Urbanization Asia, 1 (1), 1–12. Satterthwaite, D., & Moser, C. (2008). Pro poor Climate Change Adaptation in the Urban Centres of Low and Middle Income Countries. Washington, DC: The World Bank. Sengupta, R. (2016). Why urban India floods: Indian cities grow at the cost of their wetlands. Down To Earth, 2016. Centre for Science and Environment. Seto, K. C., & Kaufmann, R. K. (2009). Urban growth in south China and impacts on local precipitation, Fifth Urban Research Symposium. Shahmohamadi, P., Che‐Ani, A. I., Maulud, K. N. A., Tawil, N. M., & Abdullah, N. A. G. (2011). The impact of anthropogenic heat on formation of urban heat island and energy

­consumption balance. Urban Studies Research, 9; https://doi. org/10.1155/2011/497524. Shepherd, J. M., Pierce, H., & Negri, A. J. (2002). Rainfall modification by major urban areas: Observations from space borne rain radar on the TRMM satellite, Journal of Applied Meteorology, 41, 689–701. Smith, C., Lindley S., & Levermore, G. (2009). Estimating spatial and temporal patterns of urban anthropogenic heat fluxes for UK cities: The case of Manchester. Theoretical and Applied Climatology, 98, 19–35. Sorkhabi, R. (2006). The great Himalayan earthquakes. The Himalayan Journal, 62/5. Sperling, F., Valdivia, C., Quiroz, R., Valdivia, R., Angulo, L., Seimon, A., & Noble, I. (2008). Transitioning to climate resilient development perspectives from communities in Peru, Environment Department Papers, Climate Change Series, World Bank, 115; http://siteresources.worldbank.org/INTCC/ Resources/WB‐EDP‐115‐Transitioning‐to Climate‐Resilient‐ Development‐Peru.pdf. Sultana, Z., & Mallick, B. (2015). Adaptation Strategies after Cyclone in Southwest Coastal Bangladesh, Pro Poor Policy Choices. American Journal of Rural Development, 3 (2), 24–33; doi: 10.12691/ajrd‐3‐2‐2. Sutton, P. C., Anderson, S. J., Elvidge, C. D., Tuttle, B. T., & Ghosh, T. (2009). Paving the planet: Impervious surface as proxy measure of the human ecological footprint. Progress in Physical Geography, 33, 51. Taniguchi, M. (2007). Combined effects of urbanization and global warming on subsurface temperatures in four Asian cities. Vadose Zone Journal, 6, 591–596. TARU (2005). Disaster Management Plan, Blueprint for the Hazira Area Development Committee, Gandhinagar. Thomas, V., Albert, J. R. G., & Perez, R. T. (2013). Climate‐ related disasters in Asia and the Pacific. ADB Economics Working Paper Series No. 358. Manila: Asian Development Bank. UNISDR (2009). Terminology on disaster risk Rreduction. Geneva: United Nations; https://www.unisdr.org/files/7817_ UNISDRTerminologyEnglish.pdf. UNISDR (2018). Economic losses, poverty & disasters 19982017; https:// www.unisdr.org/2016/iddr/IDDR2018_Economic%20 Losses.pdf. UNISDR (2016). UNISDR strategic framework 2016–2021. Geneva, Switzerland: United Nations Office for Disaster Risk Reduction; http://www.unisdr.org/files/51557_ strategicframework.pdf. UNPD (United Nations Population Division) (1993). UN: World urbanization prospectus: The 2001 revision. ESA/P/ WP.173, 2001. UNPD (United Nations Population Division) (2014). World urbanization prospects: The 2014 revision. Nairobi: Department of Economics and Social Affairs. United Nations Publication. Wilbanks, T., Lankao, P. R., Bao, M., Berkhout, F., Cairncross, S., Ceron, J., et al. (2007). Industry, settlement, and society, IPCC, WGII Fourth Assessment Report, Cambridge University, 357–391. Winchester, L., & Szalachman, R. (2009). The urban poor’s vulnerability to the impacts of climate change in Latin America

46  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION and the Caribbean: A policy agenda. Fifth Urban Research Symposium 2009. World Bank & UDLG (2010). Cities and climate change: an urgent agenda; http://siteresources.worldbank.org. WWF (2009). Mega‐stress for mega‐cities: A climate vulnerability ranking of major coastal cities in Asia. Deutschland: World Wide Fund for Nature.

Yong, R. N., Mulligan, C. N., & Fukue, M. (2007). Geoenvironmental sustainability. New York: CRC. Zhao, T., Horner, M. W., & Sulik, J. (2011). A geographic approach to sectoral carbon inventory: Examining the balance between consumption‐based emissions and land‐use carbon sequestration in Florida. Annals of the Association of American Geographers, 101, 752–763; doi:10.1080/00045608.2011.567936.

4 The Contribution of Earth Observation in Disaster Prediction, Management, and Mitigation: A Holistic View Varsha Pandey1, Prashant K. Srivastava1, and George P. Petropoulos2

ABSTRACT In the era of climate change, disasters are among the events of concern in the research field due to their devastating effects worldwide. Satellite‐based Earth observation (EO) is an established technology for mapping and monitoring spatial information about disasters at frequent intervals in all weathers at real time. Further, the information from past and present conditions gathered and stored by EO enables identifying the changes occurring in Earth’s cover, and helps in framing management and mitigation strategies. Chapter 3 already introduced various types of disasters and their management briefly, and in this chapter we review some commonly used sensors on board EO satellites and spatial analysis approaches with their application in disaster impact assessment and modeling. It provides a holistic view of EO techniques in disaster prediction, management, and mitigation. 4.1. ­INTRODUCTION

spatial analyses that enable retrieval, mapping, monitoring, simulation, and representation of events linked with the disaster more effectively and accurately in a short time without visiting the ground. The past and ­present conditions identified by the remotely sensed images help in quantifying the changes of various resources directly from atmosphere, on the Earth surface, and by proxy methods that act in the subsurface layers of Earth. EO includes both geostationary and polar orbiting satellites with different temporal and spatial resolution (Nirupama et al. 2002). The polar and geostationary satellites are the two complementary data sources that can provide best disaster management options. The polar‐ orbiting satellites projected at low altitude ( 5000

Figure 4.4 This topographical image was created using the Shuttle Radar Topography Mission (SRTM) on 13 June 2005, when a 7.8 earthquake struck northern Chile. This image shows the geology that develops this earthquake. The epicenter is represented by the plus sign in this image. The light gray portion in the east represents the higher elevation mountains, while highest peaks appear white. The abrupt elevation and folds change around the epicenter cause landslides (image courtesy Jesse Allen, EO, University of Maryland’s Global Land Cover Facility).

altimetry from microwave and LiDAR data are used to measure the sea surface heights (Tralli et al., 2005; Panet et al., 2007). 4.3.5. Volcanic Activity Volcanic activity span can be monitored and detected by various data types and processing methods. These data include morphological analysis of explosion, tephra chronology, pyroclastic flow, and lithological composition. The first step of volcanic disaster prediction, management, and mitigation is the mapping and monitoring of its distribution and type of volcanic deposits. The EOS has become operational in some of the phases of volcanic disaster management, especially in ash cloud monitoring (Figure 4.5). The major applications of Earth observation for volcanic hazards include the monitoring of the three‐dimensional spatial distribution of s­ eismicity, the characteristics of deformation of the volcanic edifices (including identification of fractures, faults, rifts, flank instabilities, and calderas), gas and flux characteristics, mapping volcanic landforms and deposits, and monitoring thermal features (nature, location, temperature, and heat flux).

For volcanic hazard risk assessment, mitigation, and response, the observation strategies involve high-­resolution geodetic measurements (using InSAR and LIDAR), topographic slope mapping (DEMs), and hyperspectral imaging (Hyperion and AVIRIS). These high‐resolution data help in reducing ambiguity for identification of magma chamber geometry from other outward similar structures. Estimation of change in topography plays a  significant role for predicting volcanic eruptions. Topography measurements like ground deformation can be achieved by synthetic aperture radar (SAR) interferometer data significantly. Volcanic heat in the form of lavas, fumaroles, and hot pyroclastic flows can be mapped using Thermal IR (TIR) imagery of high resolution. Heat change may be clearly detected using very high resolution IR imagery like IKONOS and Quickbird. Having strong signal difference between their channel 4 and 5, TIR bands of AVHRR on board the NOAA satellite can detect volcanic materials such as ash cloud. PAN stereo‐ pair imagery also serves the purpose of finding hazardous activity due to its 3-D capabilities of data acquisition with moderate resolution. The lower spatial resolution of AVHHR has been found inadequate to detect the dynamic pattern of volcanic geothermal activity (Fujii & Satake, 2007). However, there are varieties of sensors available that provide data adequate for debris mapping, with higher spatial resolution satellites, such as SPOT‐4, Landsat, ASTER, Quickbird, IRS‐1D, and IKONOS. The monitoring of volcanic activity using the above‐mentioned EO satellites can delineate the assessment and mitigation planning of volcanic hazards. Using EO imagery, various computer models are increasingly being used in volcanic hazard assessment for predicting potential areas of devastation. The prediction and therefore management and ­ mitigation of volcanic eruption are an unmet objective in natural disaster areas using EO techniques. 4.3.6. Cyclone/Hurricane The low‐pressure zones occurring throughout the globe, usually over coastal areas, are referred to as “tropical cyclone basins” (Kerle et  al., 2003). The third assessment report on global climate change issued jointly by the WMO and IPCC says there is an increase in tropical water t­ emperature of 0.2–0.5°C, which indicates the ­ intensity of tropical cyclones will increase in the future. In many coastal areas, tropical cyclones/hurricanes are very common, often with devastating impact. The destructive nature of cyclones is a great threat to coastal people and the environment compared with other natural hazards. Although it is impossible to check cyclones, their impact can be minimized by many techniques and data sets to gather information for generating essential management approaches using Earth observing satellites and data sets.

54  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Figure 4.5  The Klyuchevskoy is one of the world’s most restless active volcanoes located along the Pacific Ring of Fire and has been studied by geologist for centuries from the ground. Klyuchevskoy’s ash‐laden plume, streaming west from the volcano, was imaged by Operational Land Imager (OLI) on board Landsat 8 on 19 August 2017. The plume color is brown and clouds are white. This image is the up close view taken by OLI (image courtesy of Joshua Stevens, US Geological Survey).

Tropical cyclones mostly originate in the ocean and lack monitoring to large spatial extension, which is greatly minimized by the availability of satellite‐based remote sensing observation, especially the geostationary satellites that provide synoptic views in significant high temporal scale. The shape and dynamics of cloud patterns or structures used to characterize tropical cyclones can be derived from proxy variables identified by satellite observation in nearly real time using geostationary observation. The gradient in brightness temperatures recorded by satellite sensors used to measures the structural symmetry characterizes the cloud organization in tropical cyclones (Watson & Albritton, 2001). Apart from visible infrared range satellite imageries, the sensors include space‐based scatterometers, MW radiometers, SAR, and rain radars that provide significant detailed information about the presence of secondary flows in the region between rainbands in tropical cyclones and provide a penetrating view below the upper level of cirrus clouds (Piñeros et al., 2008). Hypotheses, on long‐ term trends in tropical cyclone activity and the nature and characteristics of associated factors, need ­continuous past data records, which rely on satellite data (Katsaros

et al., 2002). Numbers of data recorded by various spacecraft are the source of such data. However, data from different satellite sensors have variable parameters needing rigorous analytical process, which can be overcome by the geostationary satellite observations. The INSAT‐3D satellite launched in 2013 is the first Indian geostationary satellite exclusively designed for meteorological observations and monitoring of land and ocean surfaces for weather forecasting and disaster warning (ISRO). It is equipped with a visible near infrared to thermal infrared imager that generates an image of Earth’s surface in every 26 minutes providing continuous data on outgoing long‐wave radiation, quantitative precipitation estimation, sea surface temperature, snow cover, and cloud motion winds. The 19 channel sounder with 18 spectral bands in shortwave, middle infrared, and long infrared regions, and 1 in a visible region provides the continuous data on vertical profiles of temperature, humidity, and integrated ozone. The data relay transporter (DRT) payload is used to receive meteorological, hydrological, and oceanographic data from Automatic Weather Stations (AWS), Automatic Rain Gauges (ARG), and Agro Met Stations (AMS).

THE CONTRIBUTION OF EARTH OBSERVATION IN DISASTER PREDICTION, MANAGEMENT, AND MITIGATION  55

Combined data are used to monitor and predict extreme weather events such as tropical cyclones, thunderstorms, cloudbursts, and heat wave. The data from these sensors are used to define or predict the center position of the cyclone, and to model its intensity and track, radius of maximum winds, and storm surge via numerical models (Knapp & Kossin, 2007; Elsner et al., 2008). 4.3.7. Wildfires Wildfires act as a triggering factor that initiates changes in ecosystems, affecting structure and patterns of vegetation that are critical to their overall functions and processes. The large‐scale fire events also influence the atmospheric and surface processes altering energy ­balances (Deb et al., 2011). Hence, identification of the forest fire risk area is the first step to avoidance of ­disastrous and damaging incidents of forest hazards. EO ­provides a very well suited technology framework for mapping forest fires at a range of spatial and temporal scales required by conservation initiatives.

In early research, the aerial survey photographs and later the middle and thermal infrared sensors were used in identifying and monitoring forest fire events (Knorr et al., 2011; Petropoulos et al., 2011, 2014). For regional to global scale active forest fire m ­ apping, the high temporal (daily) and coarse spatial ­resolution (1 km) Advanced Very High Resolution Radiometer (AVHRR) sensor of the National Oceanic and Atmospheric Administration (NOAA) satellite have been used since 1978. The 3.8 μm channel of the AHVRR sensor sensible to the surface temperature and with high temporal frequency allows active fire detection (Chuvieco & Congalton, 1989). The coarse resolution data by MODIS and VIIRS available from 2000 and 2012, respectively, are also used in related studies. Alternatively, the integration of MODIS and Visible Infrared Imaging Radiometer Suite (VIIRS) data enable multiple revisits in a day and allow more accurate and reliable fire information portals development, for example, Fire Information for Resource Management System (FIRMS) (Earthdata.nasa.gov; Figure 4.6). The MODIS 4 μm and

Figure 4.6  This image was captured by the Visible Infrared Imaging Radiometer Suite (VIIRS) on board the Suomi NPP satellite on 4 April 2018. This natural color image shows the smoke streaming from a number of fires in Russia’s far eastern Amur province. The charred areas appear black and ice covered rivers are white (image courtesy of Jeff Schmaltz, NASA, LANCE/EOSDIS Rapid Response).

56  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

11 μm radiances are used to detect fire based on threshold  values, relative differences from the background, the surface temperature as it relates to the reflection by sunlight, where the ancillary information such as water vapor, clouds, smoke aerosols, vegetation type and condition, and burn scars enhance the accuracy (Matson et  al., 1987). The spectral information from MODIS is also used in deriving the burned areas and is highly useful in studying the change in post‐fire periods (Kaufman et al., 1998). Roy et al.(2005) used the k nearest neighbors (kNN) method including a number of input spatial layers such as MODIS, climate, vegetation, and topography in developing Canada’s National Forest Inventory (NFI) sampling program. Alonso‐Canas and Chuvieco (2014) took ENVISAT‐MERIS and MODIS active fire data when mapping global burned areas using a burned area (BA) algorithm, estimated between 3.6 and 3.8 million km2 during 2006 to 2008. The recent trend in fire ­detection for monitoring and assessment includes the integration of higher spatial and lower temporal resolution data sets, such as Landsat/ASTER with coarse spatial and low temporal resolution MODIS/ MERIS/ AVHRR data. Miller and Yool (2002) and Masek et al. (2008) used multitemporal Landsat Thematic Mapper and Enhanced Thematic Mapper plus (ETM+) and again integrated these multitemporal data sets with other GIS  layers to derive the forest fire and burn severity in regional studies. French et al. (2008), Boschetti et al. (2015), Hilker et al. (2009), and Xin et al. (2013) adopted fusion techniques for EO data via some fusion models such as Spatial Temporal Adaptive Algorithm for mapping Reflectance Change (STAARCH) and spatial and temporal adaptive reflectance fusion model (STARFM), and observed improved accuracy in fire mapping and prediction. Walker et al. (2014), Falkowski et al. (2005), and others used EO ASTER data to map and characterize the fire fuels using field based and gradient modeling approach, whereas Lasaponara and Lanorte (2007) used the ASTER and Landsat ETM + data to validate the GOES and MODIS derived active fire products. In addition to model‐based approaches, the spectral indices sensible to forest fires and post‐fire reflectance such as normalized burned ratio (NBR) index, differenced Normalized Burn Ratio (dNBR), Relativized Burn Ratio (RBR), and normalized difference vegetation index (NDVI) are also used in identification of forest fire and post‐fire severity (Petropoulos et al., 2010; Veraverbeke et al., 2010; Auynirundronkool et al., 2010; Parks et al., 2014). The rapid development of both remote sensors and geoinformation technology over the last decades has also opened up new opportunities to the EO community to utilized such data in fire risk detection and monitoring (Roy et al., 2006; Vadrevu, 2008). Yet, to our knowledge, only a small number of studies have been concerned so

far in performing comparisons of fire risk probability predictions from different modeling approaches and data fusion performances. 4.3.8. Tsunamis Tsunamis are characterized by the longer gravity waves with unique frequencies in the ocean, which are accomplished by elliptical water mass movement (Fraser et al., 2000). Submarine natural and human‐made events such as earthquake, landslide, volcanic eruption, and meteorite impact or their combination are causes of tsunamis. The latest tsunami disasters include the Sumatra‐Andaman earthquake (0:58:53 UTC, 26 December 2004; Figure 4.7) that caused the loss of more than 2 million people around the Indian Ocean, mostly in Indonesia (163,795), followed by Sri Lanka (35,399), India (16,389), and Thailand (8,345) (International Federation of Red Cross and Red Crescent Societies, 2005; Galletti et  al., 2007). The past decadal statistics show the devastating effects on coastal human and forest communities, with millions of lives lost (Guha-Sapir et al., 2012). Early prediction and identification of real‐time tsunami events are difficult due to poorly understood undersea earthquake mechanisms and lack of observational data sets. Most of the tsunami warning systems (TWS) use radar data in on‐shore regions. Moreover, local tsunamis can hit the shore a short time after the quake, while offshore the lower wavelengths constrain its easy identification. The unusual change in sea surface heights (SSH) and wave heights or their anomalies are mostly related to tsunami waves by detection through EO data. A number of satellites are equipped with radar altimetry, including Jason‐1, Jason‐2, TOPEX, Envisat, and GFO, record the SSH profiles mostly in transects in comparison to the continuous coverage. The basic methodology or working principle includes the anomaly in SSH and wave heights for the same satellites before and during the tsunami events, which thus removes the slowly varying ­features of sea level, and exhibits the transient signal (Fujii & Satake, 2007). The wind speed data from satellite altimetry allow integration with related climate phenomena to improve accuracy. Numbers of modeling approaches are also used in predicting tsunami events integrated with ­various related proxies and parameters, for example, the MOST (Method of Splitting Tsunamis) model and ocean‐ general‐circulation‐models (OGCM)(Smith et  al., 2005). In comparison with past studies, Titov and Synolakis (1998) used three‐dimensional seafloor displacements of the earthquake and OGCM, and validated results with the satellite‐derived altimetry data. Song et  al. (2005) used GPS inversion technique to relate with the initial conditions of tsunami events, whereas Hoechner et  al. (2008) utilized the sea surface roughness as an indicator.

THE CONTRIBUTION OF EARTH OBSERVATION IN DISASTER PREDICTION, MANAGEMENT, AND MITIGATION  57 (a)

Wave height (cm)

(b)

0

2

4

6

8

10 40 80 200 400 – 1738

Wave travel time (hours) 0

17

Figure 4.7  These maps represent the modeled maximum wave height (a) and travel time (b) of the 26 December 2004 Indian Ocean tsunami. The maximum wave height (occurring when the wave came ashore) received by the Sumatra coastline was over 10 m, whereas Sri Lanka and Thailand had waves over 4 m. Travel time of the wave varies from minutes (Sumatra) to hours (Somalia) (map courtesy of NOAA Pacific Marine Environmental Laboratory Tsunami Research Program).

The tsunami early warning system integrated with the GIS data inputs allows an effective and improved system to reduce devastating effects, to enable management planning, to mitigate the post‐disaster losses, and to change mapping, restoring, and so on. Multiple GIS layers, such as elevation, slope, population density, inundation area, inundation depth, connectivity networks, and facilitation centers, are used in an integrated system by the disaster management authorities (Godin et al., 2009).

4.3.9. Lightning Lightning produces shock waves (proagated rapidly in  an acoustic wave as thunder) and electromagnetic radiations (in a range of extremely low to X-ray frequency) by sudden release of intense electric energy and corresponding elevated heat in the vicinity. The optical emission is due to dissociation, excitation, and subsequent recombination of atmospheric constituents in response to sudden intense heating in the lightning channel

58  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

(Christian et  al., 1989). The lightning detection system has a wide variety of applications in scientific communities like atmospheric science, forest management, commercial shipping, and government sectors. Analysis of long‐term lightning events with seasonal and yearly data over the local, regional, and global scale is used to relate with climate change phenomena. Lightning information integrated with microwave data is used to estimate the convective rainfall, severe storms, flash flood, hail storms, tornados, and related weather. Due to lack of active satellites in space, early research (1970s) used the satellite photographs in atmospheric observation, aware of their many limitations in identifying active thunderstorms (Turman, 1979). Turman (1977) explained the working principles and instrument set up used in early satellite‐based lightning detection systems. It summarized the five principles used to detect lightning: (1) electrostatic field charges, (2) magnetostatic field charges, (3) electromagnetic radiation (known as sferics), (4) optical radiation, and (5) atmospheric pressure variation. In addition, the satellite-based system used high‐frequency radio receivers or optical detectors. Due to ionosphere absorption of low radio‐frequency (rf), high rf signals were used for detection; sometimes difficulties were incurred due to human and extraterrestrial activities. In comparison, optical sensors were affected by cloud attenuation and background signal from sunlight reflected from Earth; they were mostly used for their advantages to reduce background clutter, simplistic working principle, and smaller size. Some of the old EO satellite‐based lightning detection systems with optical detectors include OSO‐2 (Vorpahl et  al., 1970), OSO‐5 (Sparrow & Ney, 1971), Vela (Turman, 1977), Defense Meteorological Satellite Program (DMSP), PBE‐2 (Piggyback Experiment) (Turman, 1979), and with rf detectors include RAE‐1 (Herman et al., 1973), ARIEL 3 (Bent & Horner, 1969), and ISS‐b (Kotaki, 1983). Christian et  al.(1989) described the working efficiency and early results of the Lightning Mapper Sensor (LMS) mounted in aircraft and working over the United States and surrounding provinces. They highlighted the advantages of the LMS, which was later used in the geostationary GOES satellite, as higher spatial extent with improved resolution and detection efficiency active in day and night compared with the previous sensors, mostly limited to regional or local studies incurred from extensively low accuracy. 4.4. ­CONCLUSIONS EO technology offers wide applications for disaster detection, monitoring, modeling, and suitable geospatial techniques with the socioeconomic data in evaluation of their impact and risk assessment. The spatially explicit

geospatial data also provide disaster management plans and mitigation policies. The wide range of data with various characteristics enables identification of surface and subsurface ground conditions and offers minute and rapid analysis. Use of visible, infrared, and microwave remote sensing truly gathers the information of the variables that directly or indirectly link with various disasters. However, these data have some biasness and require ground truth observations for calibration and validation. Moreover, there are certain limitations with EO, as it does not provide information about the subsurface ground features beyond certain depths. The prior data and information on disaster events like earthquakes, tsunami, and volcanic eruptions are very limited, and need acute field data. However, the geospatial analysis of land surface condition integrated with various spatial and nonspatial data helps in efficient management and framing effective mitigation plans. The management and mitigation strategies help in reducing and recovering the damage by natural disasters. For several types of disasters, the Earth observation data and geospatial analysis techniques have become operational in the prior warning system. ­ACKNOWLEDGMENTS GPP’s contribution to this work was supported by NERC’s Newton Fund RCUK project Toward a Fire Early Warning System for Indonesia (ToFEWSI). Prashant K. Srivastava would like to acknowledge SERB, Department of Science and Technology, for funding and support. ­REFERENCES Aleotti, P., & Chowdhury, R.(1999). Landslide hazard assessment: Summary review and new perspectives. Bulletin of Engineering Geology and the Environment, 58, 21–44. Amarnath, G. (2014). An algorithm for rapid flood inundation mapping from optical data using a reflectance differencing technique. Journal of Flood Risk Management, 7, 239–250. Asghar, S., Alahakoon, D., & Churilov, L. (2006). A comprehensive conceptual model for disaster management. Journal of Humanitarian Assistance, 1360, 1–15. Auynirundronkool, K., Chen, N., Peng, C., Yang, C., Gong, J., & Silapathong, C. (2012). Flood detection and mapping of the Thailand Central plain using RADARSAT and MODIS under a sensor web environment. International Journal of Applied Earth Observation and Geoinformation, 14, 245–255. Bent, R., & Horner, F. (1969). Measurement of terrestrial radio noise. Proceedings of the Royal Society A, 311(1507); doi. org/10.1098/rspa.1969.0133. Boschetti, L., Roy, D. P., Justice, C. O., & Humber, M. L. (2015). MODIS‐Landsat fusion for large area 30 m burned area mapping. Remote Sensing of Environment, 161, 27–42.

THE CONTRIBUTION OF EARTH OBSERVATION IN DISASTER PREDICTION, MANAGEMENT, AND MITIGATION  59 Büchele, B., Kreibich, H., Kron, A., Thieken, A., Ihringer, J., Oberle, P., et  al. (2006). Flood‐risk mapping: contributions towards an enhanced assessment of extreme events and associated risks. Natural Hazards and Earth System Science, 6, 485–503. Carrara, A., Cardinali, M., Guzzetti, F., & Reichenbach, P. (1995). GIS technology in mapping landslide hazard. Geographical information systems in assessing natural hazards (pp. 135–175). Springer. Carrara, A., Guzzetti, F., Cardinali, M., & Reichenbach, P. (1999). Use of GIS technology in the prediction and monitoring of landslide hazard. Natural Hazards, 20, 117–135. Christian, H. J., Blakeslee, R. J., & Goodman, S. J. (1989). The detection of lightning from geostationary orbit. Journal of Geophysical Research: Atmospheres, 94, 13329–13337. Chuvieco, E., & Congalton, R. G. (1989). Application of remote sensing and geographic information systems to forest fire hazard mapping. Remote Sensing of Environment, 29, 147–159. Colesanti, C., & Wasowski, J. (2006). Investigating landslides with space‐borne Synthetic Aperture Radar (SAR) interferometry. Engineering Geology, 88, 173–199. Colson, D., Petropoulos, G. P., & Ferentinos, K. P. (2018). Exploring the potential of sentinels‐1 & 2 of the Copernicus mission in support of rapid and cost‐effective wildfire assessment. international. Journal of Applied Earth Observation and Geoinformation, 73, 262–276. Cotecchia, V. (1978). Systematic reconnaissance mapping and registration of slope movements. Bulletin of the International Association of Engineering Geology‐Bulletin de l’Association Internationale de Géologie de l’Ingénieur, 17, 5–37. Deb, S., Kumar, P., Pal, P., & Joshi, P. (2011). Assimilation of INSAT data in the simulation of the recent tropical cyclone Aila. International Journal of Remote Sensing, 32, 5135–5155. Elsner, J. B., Kossin, J. P., & Jagger, T. H. (2008). The increasing intensity of the strongest tropical cyclones. Nature, 455, 92. Falkowski, M. J., Gessler, P. E., Morgan, P., Hudak, A.T., & Smith, A.M. (2005). Characterizing and mapping forest fire fuels using ASTER imagery and gradient modeling. Forest Ecology and Management, 217, 129–146. Fernández, T., Jiménez, J., Fernández, P., El Hamdouni, R., Cardenal, F., Delgado, et al. (2008). Automatic detection of landslide features with remote sensing techniques in the Betic Cordilleras (Granada, Southern Spain). The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 37 (B 8), 351–356. Ferraro, R.(2004). Satellite‐based climatologies related to the water cycle. Space 2004 Conference and Exhibition, 5862. Forkuo, E. K. (2011). Flood hazard mapping using Aster image data with GIS. International Journal of Geomatics and Geosciences, 1, 932–950. Fourniadis, I., Liu, J., & Mason, P. (2007). Landslide hazard assessment in the Three Gorges area, China, using ASTER imagery: Wushan–Badong. Geomorphology, 84, 126–144. Fraser, R., Li, Z., & Cihlar, J. (2000). Hotspot and NDVI differencing synergy (HANDS): A new technique for burned area mapping over boreal forest. Remote Sensing of Environment, 74, 362–376. French, N. H., Kasischke, E. S., Hall, R. J., Murphy, K. A., Verbyla, D. L., Hoy, E. E., & Allen, J. L. (2008). Using

Landsat data to assess fire and burn severity in the North American boreal forest region: an overview and summary of results. International Journal of Wildland Fire, 17, 443–462. Fujii, Y., & Satake, K. (2007). Tsunami source of the 2004 Sumatra–Andaman Earthquake inferred from tide gauge and satellite data. Bulletin of the Seismological Society of America, 97, S192–S207. Galletti, M., Krieger, G., Borner, T., Marquart, N., & Schultz‐ Stellenfleth, J. (2007). Concept design of a near‐space radar for tsunami detection. Geoscience and Remote Sensing Symposium, 2007. IGARSS 2007 (pp. 34–37). IEEE. Genzano, N., Aliano, C., Corrado, R., Filizzola, C., Lisi, M., Mazzeo, G., Paciello, R., et al. (2009). RST analysis of MSG‐ SEVIRI TIR radiances at the time of the Abruzzo 6 April 2009 Earthquake. Natural Hazards and Earth System Sciences, 9, 2073. Gillespie, T. W., Chu, J., Frankenberg, E., & Thomas, D. (2007). Assessment and prediction of natural hazards from satellite imagery. Progress in Physical Geography, 31, 459–470. Godin, O., Irisov, V., Leben, R., Hamlington, B., & Wick, G. (2009). Variations in sea surface roughness induced by the 2004 Sumatra‐Andaman tsunami. Natural Hazards and Earth System Science, 9. Godschalk, D., Beatley, T., Berke, P., Brower, D., & Kaiser, E. J. (1998). Natural hazard mitigation: Recasting disaster policy and planning. Island Press. Gu, Y., Brown, J. F., Verdin, J. P., & Wardlow, B. (2007). A five‐ year analysis of MODIS NDVI and NDWI for grassland drought assessment over the central Great Plains of the United States. Geophysical Research Letters, 34. Guha-Sapir, D., Vos, F., Below, R., Ponserre, S. (2012). Annual disaster statistical review 2011: the numbers and trends. Centre for Research on the Epidemiology of Disasters (CRED). Guo, H. (2010). Understanding global natural disasters and the role of Earth observation. International Journal of Digital Earth, 3, 221–230. Herman, J. R., Caruso, J. A., & Stone, R. G. (1973). Radio Astronomy Explorer (RAE): I. Observations of terrestrial radio noise. Planetary and Space Science, 21, 443–461. Hess, L. L., Melack, J. M., Filoso, S., & Wang, Y. (1995). Delineation of inundated area and vegetation along the Amazon floodplain with the SIR‐C synthetic aperture radar. IEEE Transactions on Geoscience and Remote Sensing, 33, 896–904. Hilker, T., Wulder, M. A., Coops, N. C., Linke, J., McDermid, G., Masek, J. G., et al. (2009). A new data fusion model for high spatial‐and temporal‐resolution mapping of forest disturbance based on Landsat and MODIS. Remote Sensing of Environment, 113, 1613–1627. Hoechner, A., Babeyko, A. Y., & Sobolev, S. V. (2008). Enhanced GPS inversion technique applied to the 2004 Sumatra Earthquake and tsunami. Geophysical Research Letters, 35. Hossain, F., & Anagnostou, E. N. (2004). Assessment of current passive‐microwave‐and infrared‐based satellite rainfall remote sensing for flood prediction. Journal of Geophysical Research: Atmospheres, 109. Huang, S., Li, J., & Xu, M. (2012). Water surface variations monitoring and flood hazard analysis in Dongting Lake area

60  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION using long‐term Terra/MODIS data time series. Natural Hazards, 62, 93–100. Ireland, G., & Petropoulos, G. P. (2015). Exploring the relationships between post‐fire vegetation regeneration dynamics, topography and burn severity: A case study from the Montane Cordillera Ecozones of Western Canada. Applied Geography, 56, 232–248. Ireland, G., Volpi, M., & Petropoulos, G. P. (2015). Examining the capability of supervised machine learning classifiers in extracting flooded areas from Landsat TM imagery: a case study from a Mediterranean flood. Remote Sensing, 7, 3372–3399. Irimescu, A., Stancalie, G., Craciunescu, V., Flueraru, C., & Anderson, E. (2009). The use of Remote Sensing and GIS techniques in flood monitoring and damage assessment: A study case in Romania. Threats to Global Water Security. Springer, 167–177. Jeyaseelan, A. (2003). Droughts and floods assessment and monitoring using remote sensing and GIS. In: M. V. K. Sivakumar and D. E. Hinsman (Eds.) Satellite remote sensing and GIS applications in agricultural meteorology (pp. 291–313). World Meteorological Organization, Geneva. Switzerland. Jónsson, S., Zebker, H., Segall, P., & Amelung, F. (2002). Fault slip distribution of the 1999 M w 7.1 Hector Mine, California, Earthquake, estimated from satellite radar and GPS measurements. Bulletin of the Seismological Society of America, 92, 1377–1389. Joyce, K. E., Belliss, S. E., Samsonov, S. V., McNeill, S. J., & Glassey, P. J. (2009). A review of the status of satellite remote sensing and image processing techniques for mapping natural hazards and disasters. Progress in Physical Geography, 33, 183–207. Karamesouti, M., Petropoulos, G. P., Papanikolaou, I. D., Kairis, O., & Kosmas, K. (2016). Erosion rate predictions from PESERA and RUSLE at a Mediterranean site before and after a wildfire: Comparison & implications. Geoderma, 261, 44–58. Katsaros, K. B., Vachon, P. W., Liu, W. T., & Black, P. G. (2002). Microwave remote sensing of tropical cyclones from space. Journal of Oceanography, 58, 137–151. Kaufman, Y. J., Justice, C. O., Flynn, L. P., Kendall, J. D., Prins, E. M., Giglio, L., Ward, D. E., et al. 1998). Potential global fire monitoring from EOS‐MODIS. Journal of Geophysical Research: Atmospheres, 103, 32215–32238. Kerle, N., Froger, J.‐L., Oppenheimer, C., & Vries, B. V. W. D. (2003). Remote sensing of the 1998 mudflow at Casita volcano, Nicaragua. International Journal of Remote Sensing, 24, 4791–4816. Khan, S. I., Hong, Y., Wang, J., Yilmaz, K. K., Gourley, J. J., Adler, R. F., Brakenridge, G. R., et al. (2011). Satellite remote sensing and hydrologic modeling for flood inundation ­mapping in Lake Victoria basin: Implications for hydrologic prediction in ungauged basins. IEEE Transactions on Geoscience and Remote Sensing, 49, 85–95. Knapp, K. R., & Kossin, J. P. (2007). New global tropical cyclone data set from ISCCP B1 geostationary satellite observations. Journal of Applied Remote Sensing, 1, 013505. Knorr, W., Pytharoulis, I., Petropoulos, G. P., & Gobron, N. (2011). Combined use of weather forecasting and satellite

remote sensing information for fire risk, fire and fire impact monitoring. Computational Ecology and Software, 1, 112. Kogan, F. (2002). World droughts in the new millennium from AVHRR‐based vegetation health indices. EOS, Transactions American Geophysical Union, 83, 557–563. Kogan, F. N. (2000). Contribution of remote sensing to drought early warning. In: D. A. Wilhite, M.V.K. Sivakumar & D. A. Wood (Eds.) Early warning systems for drought preparedness and drought management (pp. 75–87). World Meteorological Organization, Geneva. Kotaki, M. (1983). The global distribution of thunderstorm activity observed by the ionosphere sounding satellite (ISS‐b). Journal of.Atmospheric and Terrestrial Physics, 45, 833–847. Kussul, N. N., Shelestov, A. Y., Skakun, S. V., Li, G., & Kussul, O. M. (2012). The wide area grid testbed for flood monitoring using Earth observation data. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 5, 1746–1751. Lasaponara, R., & Lanorte, A. (2007). Remotely sensed characterization of forest fuel types by using satellite ASTER data. International Journal of Applied Earth Observation and Geoinformation, 9, 225–234. Lisi, M., Filizzola, C., Genzano, N., Grimaldi, C., Lacava, T., Marchese, F., Mazzeo, G., et  al. (2010). A study on the Abruzzo 6 April 2009 Earthquake by applying the RST approach to 15 years of AVHRR TIR observations. Natural Hazards and Earth System Sciences, 10, 395–406. Liu, C.‐C., Wu, A.‐M., Yen, S.‐Y., & Huang, C.‐H. (2009). Rapid locating of fire points from Formosat‐2 high spatial resolution imagery: example of the 2007 California wildfire. International Journal of Wildland Fire, 18, 415–422. Liu, J., Mason, P., Clerici, N., Chen, S., Davis, A., Miao, F., et al. (2004). Landslide hazard assessment in the Three Gorges area of the Yangtze river using ASTER imagery: Zigui– Badong. Geomorphology, 61, 171–187. Mandl, D., Sohlberg, R., Justice, C., Ungar, S., Ames, T., Frye, S., Chien, S., et al. (2008). A space‐based sensor web for disaster management. Geoscience and Remote Sensing Symposium, 2008. IGARSS 2008 (pp. V‐294–V‐297). IEEE. Markogianni, V., Kalivas, D., Petropoulos, G., & Dimitriou, E. (2018). An appraisal of the potential of Landsat 8 in estimating chlorophyll‐a, ammonium concentrations and other water quality indicators. Remote Sensing, 10, 1018. Masek, J. G., Huang, C., Wolfe, R., Cohen, W., Hall, F., Kutler, J., & Nelson, P. (2008). North American forest disturbance mapped from a decadal Landsat record. Remote Sensing of Environment, 112, 2914–2926. Massonnet, D., & Feigl, K. L. (1998). Radar interferometry and its application to changes in the Earth’s surface. Reviews of Geophysics, 36, 441–500. Matar, M. (2005). Space‐based facilities for environmental disaster monitoring, assessment and management. Recent Advances in Space Technologies, 2005. RAST 2005. Proceedings of 2nd International Conference (pp. 660–666). IEEE. Matson, M., Stephens, G., & Robinson, J. (1987). Fire detection using data from the NOAA‐N satellites. International Journal of Remote Sensing, 8, 961–970. McNeil, B. E., de Beurs, K. M., Eshleman, K. N., Foster, J. R., & Townsend, P. A. (2007). Maintenance of ecosystem nitrogen limitation by ephemeral forest disturbance: An

THE CONTRIBUTION OF EARTH OBSERVATION IN DISASTER PREDICTION, MANAGEMENT, AND MITIGATION  61 assessment using MODIS, Hyperion, and Landsat ETM+. Geophysical Research Letters, 34. Menzel, W. P., Holt, F. C., Schmit, T. J., Aune, R. M., Schreiner, A. J., Wade, G. S., & Gray, D. G. (1998). Application of GOES‐8/9 soundings to weather forecasting and nowcasting. Bulletin of the American Meteorological Society, 79, 2059–2077. Miller, J. D., & Yool, S. R. (2002). Mapping forest post‐fire canopy consumption in several overstory types using multi‐ temporal Landsat TM and ETM data. Remote Sensing of Environment, 82, 481–496. Mishra, A. K., & Singh, V. P. (2010). A review of drought ­concepts. Journal of Hydrology, 391, 202–216. Nirupama & S. S. P. (2002). Role of Remote Sensing in Disaster Management. Water Resources Research Report. https:// ir.lib.uwo.ca/wrrr/3. Ouzounov, D., Liu, D., Chunli, K., Cervone, G., Kafatos, M., & Taylor, P. (2007). Outgoing long wave radiation variability from IR satellite data prior to major Earthquakes. Tectonophysics, 431, 211–220. Panet, I., Mikhailov, V., Diament, M., Pollitz, F., King, G., De Viron, O., Holschneider, M., et  al. (2007). Coseismic and post‐seismic signatures of the Sumatra 2004 December and 2005 March Earthquakes in GRACE satellite gravity. Geophysical Journal International, 171, 177–190. Parks, S. A., Dillon, G. K., & Miller, C. (2014). A new metric for quantifying burn severity: The relativized burn ratio. Remote Sensing, 6, 1827–1844. Petropoulos, G., Knorr, W., Scholze, M., Boschetti, L., & Karantounias, G. (2010). Combining ASTER multispectral imagery analysis and support vector machines for rapid and cost‐effective post‐fire assessment: a case study from the Greek wildland fires of 2007. Natural Hazards and Earth System Sciences, 10, 305–317. Petropoulos, G. P., Griffiths, H. M., & Kalivas, D. P. (2014). Quantifying spatial and temporal vegetation recovery dynamics following a wildfire event in a Mediterranean landscape using EO data and GIS. Applied Geography, 50, 120–131. Petropoulos, G. P., Kontoes, C., & Keramitsoglou, I. (2011). Burnt area delineation from a uni‐temporal perspective based on Landsat TM imagery classification using support vector machines. International Journal of Applied Earth Observation and Geoinformation, 13, 70–80. Petropoulos, G. P., Kontoes, C. C., & Keramitsoglou, I. (2012). Land cover mapping with emphasis to burnt area delineation using co‐ orbital ALI and Landsat TM imagery. International Journal of Applied Earth Observation and Geoinformation, 18, 344–355. Piñeros, M. F., Ritchie, E. A., & Tyo, J. S. (2008). Objective measures of tropical cyclone structure and intensity change from remotely sensed infrared image data. IEEE Transactions on Geoscience and Remote Sensing, 46, 3574–3580. Popescu, M. (1994). A suggested method for reporting landslide causes. Bulletin of the International Association of Engineering Geology‐Bulletin de l’Association Internationale de Géologie de l’Ingénieur, 50, 71–74. Poser, K., & Dransch, D. (2010). Volunteered geographic information for disaster management with application to rapid flood damage estimation. Geomatica, 64, 89–98. Pradhan, B., & Lee, S. (2010). Delineation of landslide hazard areas on Penang Island, Malaysia, by using frequency ratio,

logistic regression, and artificial neural network models. Environmental Earth Sciences, 60, 1037–1054. Pulvirenti, L., Pierdicca, N., Chini, M., & Guerriero, L. (2011). An algorithm for operational flood mapping from Synthetic Aperture Radar (SAR) data using fuzzy logic. Natural Hazards and Earth System Sciences, 11, 529. Radbruch‐Hall, D., & Varnes, D. (1976). Landslides: Cause and effect. Bulletin of the International Association of Engineering Geology‐Bulletin de l’Association Internationale de Géologie de l’Ingénieur, 13, 205–216. Roberts, D. A., Dennison, P. E., Gardner, M. E., Hetzel, Y., Ustin, S. L., & Lee, C. T. (2003). Evaluation of the potential of Hyperion for fire danger assessment by comparison to the Airborne Visible/Infrared Imaging Spectrometer. IEEE Transactions on Geoscience and Remote Sensing 41, 1297–1310. Roy, D. P., Boschetti, L., & Trigg, S. N. (2006). Remote sensing of fire severity: assessing the performance of the normalized burn ratio. Geoscience and Remote Sensing Letters, IEEE, 3, 112–116. Roy, D. P., Jin, Y., Lewis, P., & Justice, C. (2005). Prototyping a global algorithm for systematic fire‐affected area mapping using MODIS time series data. Remote Sensing of Environment, 97, 137–162. Saha, A., Gupta, R., & Arora, M. (2002). GIS‐based landslide hazard zonation in the Bhagirathi (Ganga) valley, Himalayas. International Journal of Remote Sensing, 23, 357–369. Saraf, A. K., Rawat, V., Banerjee, P., Choudhury, S., Panda, S. K., Dasgupta, S., & Das, J. (2008). Satellite detection of Earthquake thermal infrared precursors in Iran. Natural Hazards, 47, 119–135. Smith, W. H., Scharroo, R., Titov, V. V., Arcas, D., & Arbic, B. K. (2005). Satellite altimeters measure tsunami. Oceanography, 18, 11–13. Song, Y. T., Ji, C., Fu, L. L., Zlotnicki, V., Shum, C., Yi, Y., & Hjorleifsdottir, V. (2005). The 26 December 2004 tsunami source estimated from satellite radar altimetry and seismic waves. Geophysical Research Letters, 32. Sparrow, J., & Ney, E. (1971). Lightning observations by satellite. Nature, 232, 540. Stramondo, S., Bignami, C., Chini, M., Pierdicca, N., & Tertulliani, A. (2006). Satellite radar and optical remote sensing for Earthquake damage detection: Results from different case studies. International Journal of Remote Sensing, 27, 4433–4447. Sugisaki, R., Anno, H., Adachi, M., & Ui, H. (1980). Geochemical features of gases and rocks along active faults. Geochemical Journal, 14, 101–112. Sun, D., & Pinker, R.T. (2003). Estimation of land surface temperature from a Geostationary Operational Environmental Satellite (GOES‐8). Journal of Geophysical Research: Atmospheres, 108. Tarchi, D., Casagli, N., Moretti, S., Leva, D., & Sieber, A. J. (2003). Monitoring landslide displacements by using ground‐ based synthetic aperture radar interferometry: Application to the Ruinon landslide in the Italian Alps. Journal of Geophysical Research: Solid Earth, 108. Titov, V. V., & Synolakis, C. E. (1998). Numerical modeling of tidal wave runup. Journal of Waterway, Port, Coastal, and Ocean Engineering, 124, 157–171.

62  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Tralli, D. M., Blom, R. G., Zlotnicki, V., Donnellan, A., & Evans, D. L. (2005). Satellite remote sensing of Earthquake, volcano, flood, landslide and coastal inundation hazards. ISPRS Journal of Photogrammetry and Remote Sensing, 59, 185–198. Tramutoli, V., Cuomo, V., Filizzola, C., Pergola, N., & Pietrapertosa, C. (2005). Assessing the potential of thermal infrared satellite surveys for monitoring seismically active areas: The case of Kocaeli (Izmit) Earthquake, August 17, 1999. Remote Sensing of Environment, 96, 409–426. Tronin, A. (2000). Thermal IR satellite sensor data application for Earthquake research in China International Journal of Remote Sensing, 21, 3169–3177. Tronin, A. A., Hayakawa, M., & Molchanov, O. A. (2002). Thermal IR satellite data application for Earthquake research in Japan and China. Journal of Geodynamics, 33, 519–534. Tucker, C. J., & Choudhury, B. J. (1987). Satellite remote sensing of drought conditions. Remote Sensing of Environment, 23, 243–251. Turman, B. (1977). Detection of lightning superbolts. Journal of Geophysical Research, 82, 2566–2568. Turman, B. (1979). Lightning Detection from Space: Satellite monitoring of lightning holds great promise as a means of studying and tracking thunderstorms on a worldwide scale. American Scientist, 67, 321–329. Vadrevu, K. P. (2008). Analysis of fire events and controlling factors in eastern India using spatial scan and multivariate statistics. Geografiska Annaler: Series A, Physical Geography, 90, 315–328. Veraverbeke, S., Lhermitte, S., Verstraeten, & W. W., Goossens, R. (2010). The temporal dimension of differenced Normalized Burn Ratio (dNBR) fire/burn severity studies: the case of the large 2007 Peloponnese wildfires in Greece. Remote Sensing of Environment, 114, 2548–2563.

Vorpahl, J., Sparrow, J., & Ney, E. (1970). Satellite observations of lightning. Science, 169, 860–862. Walker, J., De Beurs, K., & Wynne, R. (2014). Dryland vegetation phenology across an elevation gradient in Arizona, USA, investigated with fused MODIS and Landsat data. Remote Sensing of Environment, 144, 85–97. Wang, L., & Qu, J. J. (2007). NMDI: A normalized multi‐band drought index for monitoring soil and vegetation moisture with satellite remote sensing. Geophysical Research Letters, 34. Watson, R. T., Albritton, D. L. (2001). Climate Change 2001: Synthesis Report: Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press. Xin, Q., Olofsson, P., Zhu, Z., Tan, B., & Woodcock, C. E. (2013). Toward near real‐time monitoring of forest disturbance by fusion of MODIS and Landsat data. Remote Sensing of Environment, 135, 234–247. Yang, X., & Chen, L. (2010). Using multi‐temporal remote sensor imagery to detect Earthquake‐triggered landslides. International Journal of Applied Earth Observation and Geoinformation, 12, 487–495. Zebker, H. A., Rosen, P. A., Goldstein, R. M., Gabriel, A., & Werner, C. L. (1994). On the derivation of coseismic displacement fields using differential radar interferometry: The Landers Earthquake. Journal of Geophysical Research: Solid Earth, 99, 19617–19634. Zhang, W., Lin, J., Peng, J., & Lu, Q. (2010). Estimating Wenchuan Earthquake induced landslides based on remote sensing. International Journal of Remote Sensing, 31, 3495–3508. http://itic.iocunesco.org/index.php?option=com_content&view =article&id=1613&Itemid=2310 https://Earthdata.nasa.gov/Earth‐observation‐data/near‐real‐ time/firms

Section II Atmospheric Hazards and Disasters

5 Tropical Cyclones Over the North Indian Ocean in Changing Climate R. Bhatla1,2, Raveena Raj3, R.K. Mall2,4, and Shivani3

ABSTRACT The changing climate and its impact on extreme events like tropical cyclones is one of the great societal and scientific concerns. Whether or not the reason for variation in frequency of tropical cyclones is climate change impact is a persistent myth. In this chapter, analysis was carried out for monthly, yearly, interannual, decadal, and thirty yearly variation in frequencies of tropical cyclones for premonsoon, monsoon, and postmonsoon seasons over the Bay of Bengal (BoB) and Arabian Sea (AS). Here the direction of movement of these tropical storms (cyclonic storms and severe cyclonic storms, CS + SCS) before landfall is also observed in order to understand the areas that were most affected by these storms. Then the decadal frequencies of direction of movement are calculated and observed. The data in the present study were taken from NOAA (National Oceanic and Atmospheric Administration) India data for tropical cyclones (1931–1970), the India Meteorological Department (IMD) (1971–1980), and IMD Journal “Mausam” issues for tropical cyclones (1980–2012) for both the Bay of Bengal and Arabian Sea. The study reveals that the total systems of cyclonic storms formed in the Bay of Bengal over these 80 years exceed those of the Arabian Sea because it is relatively colder than the Bay of Bengal, so fewer systems are formed. The average annual frequency of tropical cyclones in the north Indian Ocean (Bay of Bengal and Arabian Sea) is about 5 (about 5–6% of global annual average) and about 80 cyclones form around the globe in the year. The frequency is more in the Bay of Bengal than in the Arabian Sea, the ratio being 4:1. The monthly frequency of tropical cyclones in the north Indian Ocean displays a bimodal characteristic with a primary peak in November and secondary peak in May. The months of May–June and October–November are known to produce cyclones of severe intensity. As most of the cyclones that originate in the Bay of Bengal dissipate on land, few reach the Arabian Sea. The frequency of tropical cyclones shows that there is a decreasing trend in recent decades compared with earlier decades. 5.1. ­INTRODUCTION

and clockwise in Southern Hemisphere. The Bay of Bengal, located north of the Indian Ocean, is responsible for the formation of the deadliest tropical cyclones in the world. The Arabian Sea is also located to the north of the Indian Ocean. Cyclones are very rare in the Arabian Sea, but it produces strong tropical cyclones. Cyclonic disturbances during the summer monsoon months (June to September) are responsible for flooding, which is caused by rainfall associated with the movement of monsoon depression and cyclonic storms. These ­disturbances occur very frequently during the summer monsoon, mostly originate from the Bay of Bengal and very rarely from the Arabian Sea. The cyclonic ­disturbances during monsoon months mostly form in the

Tropical cyclones are small cyclonic whirls having nearly circular isobars and very strong winds circulating in a counterclockwise direction in the Northern Hemisphere Department of Geophysics, Banaras Hindu University, Varanasi, Uttar Pradesh, India 2  DST-Mahamana Centre of Excellence in Climate Change Research, Institute of Environment and Sustainable Development, Banaras Hindu University, Varanasi, Uttar Pradesh, India 3  Department of Botany, Environmental Science, Banaras Hindu University, Varanasi, Uttar Pradesh, India 4  Institute of Environment and Sustainable Development, Banaras Hindu University, Varanasi, Uttar Pradesh, India 1 

Techniques for Disaster Risk Management and Mitigation, First Edition. Edited by Prashant K. Srivastava, Sudhir Kumar Singh, U. C. Mohanty, and Tad Murty. © 2020 John Wiley & Sons, Inc. Published 2020 by John Wiley & Sons, Inc. 65

66  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Indo‐Gangetic plain and its neighborhood after crossing the coast. Sometimes, after moving far into the interior of the country, these disturbances recurve and move north or northeast and break over foothills of the Himalayas under the influence of a westerly trough or western disturbances moving from west to east at higher latitudes. Associated with these cyclonic disturbances, both the Arabian Sea and Bay of Bengal branches of the summer monsoon currents strengthen considerably and cause heavy to very heavy rainfall over the region as these disturbances move through. It is observed that, in summer monsoon season, heavy rainfall generally occurs in the southwestern sector of these disturbances. Heavy rainfall also occurs in the northeastern sectors when they recurve from their westerly to northwesterly track toward north or northeast. The Arabian Sea’s coast is shared by India, Yemen, Oman, Iran, Sri Lanka, Maldives, and Somalia. Monsoons are characteristics of the Arabian Sea and responsible for the yearly cycling of its waters. In summer, strong winds blow from the southwest to the northeast, bringing rain to the Indian subcontinent. During the winter, the winds are milder and blow in the opposite direction, from northeast to the southwest. Every year, about 80–90 tropical cyclones form over the oceanic regions, out of which about 45 develop into severe cyclonic storms and cause damages over the landfall areas of the various coastal regions. In the Bay of Bengal and Arabian Sea, tropical cyclones usually occur from April to June and September to November. Climatic data of tropical cyclones are important for the preliminary forecasts of the movement, intensification, and coastal crossing of tropical cyclones. Formation and intensification are controlled by several atmospheric and oceanic factors, such as low‐level relative vortices and humidity, vertical wind shear, low‐ level convergence, upper‐level divergence, conditional instability, and sea surface temperature (SST) (Gray, 1968). The link between global warming and the frequency of tropical cyclones has been a topic of research of the past few years. Emanuel (1987) proposed that in a greenhouse-gas-warming climate the maximum potential intensity of tropical cyclones would increase. Knutson et al. (1998, 2001) and Knutson and Tuleya (1999, 2004) conducted hurricane model simulations with large‐scale thermodynamic conditions. Landsea et al. (1999) showed the recent observed increases in North Atlantic hurricane frequency and intensity are within the range of observed multidecadal variability. Landsea (2000) held that the tropical cyclones in various basins are not independent of one another because of a strong link with global phenomena. An increase in activity in one region may instead be accompanied by a decrease in tropical cyclone activity

in another basin. In the northwest Pacific basin, there was no significant relationship between typhoon activity parameters and the local SST warming (Chan & Liu, 2004). Recent studies on hurricane intensity trends have been discussed in detail by Emanuel (2005), Chan (2006), Klotzbach (2006), and Sriver and Huber (2007). Elsner et al. (2008) studied the increasing intensity of the strongest tropical cyclones. Yu and Wang (2009) indicated that the tropical cyclone season is likely to become longer in a warmer climate in the North Indian Ocean with increased cyclone intensity particularly in the month of May. Krishna (2009) discussed the decreasing trend in the tropical easterly jet (TEJ) reducing wind shear and increasing tropical cyclone formation in the north Indian Ocean. The possible effect of climate change on tropical cyclone is a change in the characteristics of the tropical cyclone, whether it changed or will change in a warming climate (Knutson et. al., 2010). Wang et. al. (2010) observed that the impact of the rising sea surface temperature on tropical cyclones is one of great societal and scientific concern. The increase in sea surface temperature and decrease in wind shear correspond to an increase in the cyclogenesis events and vice versa for the north Indian Ocean (Deo et al., 2011). Evan et al. (2011) studied the recent change in cyclone intensity over the Arabian Sea and its connection to a sixfold increase in the anthropogenic emission of aerosols leading to a weakening of the severe lower level and upper level winds that define the monsoonal circulation over the Arabian Sea. The objective of this chapter is to observe the frequencies of tropical cyclones that formed in the north Indian Ocean (Bay of Bengal and Arabian Sea) over an 80-year period (1931–2012). 5.2. ­DATA AND METHODOLOGY The monthly frequency of tropical cyclones for the years 1931–2012, for both the Bay of Bengal and Arabian Sea, was collected from NOAA (National Oceanic and Atmospheric Administration) India data for tropical cyclones (1931–1970), the India Meteorological Department (IMD) weekly/monthly weather reports (1971–1980), and IMD Journal “Mausam” issues for tropical cyclones (1980–2012). Data for the year 1956 was unavailable, so this year has been omitted. They were classified as cyclonic storms and severe cyclonic storms (CS + SCS) based on the criteria as per the IMD classification. The area of formation of these systems was also taken into account. Then the seasonal frequency of tropical cyclones (CS + SCS) was taken out for  the premonsoon (March‐April‐May), monsoon (June to September), postmonsoon (October‐ November‐December), and winter season (January‐

Tropical Cyclones Over the North Indian Ocean in Changing Climate  67

February). In the present study, more emphasis is on the premonsoon, monsoon, and postmonsoon seasons as compared with premonsoon and winter season. The total frequency (monthly and seasonal) for the year 1931–2012 was taken out to study the monthly and seasonal variation over these years. The following analysis has been carried out. 1. Interannual variation in frequencies, for monsoon and postmonsoon seasons, was calculated and the variations were observed. 2. The decadal frequencies of tropical cyclones (CS + SCS) were calculated and the trends in decadal frequency were observed for eight decades (1932–1942 to 2003–2012). 3. Thirty yearly frequencies were also taken out, since to derive on any conclusion for any meteorological data at least 30 years of data should be known. Then the trend over this data was observed. 4. The direction of movement of these tropical storms (CS + SCS) before landfall was observed in order to understand the areas that were most affected by these storms, for both Arabian Sea and Bay of Bengal. Then the decadal frequencies of direction of movement were calculated and observed. 5.3. ­RESULT AND ANALYSIS 5.3.1. Variations in Monthly and Seasonal Frequency of Tropical Cyclones Monthly and seasonal frequency variation of tropical cyclones over the Bay of Bengal can be seen in Figure  5.1. It shows that the frequencies of tropical cyclones forming over the Bay of Bengal were found to be least during February and March (as low as two from 1931–2012) and highest during the month of November, in which the total number of storms was 82. During summer monsoon months (June, July,

August, and September), not much fluctuation was observed. During premonsoon months (March, April, and May), maximum frequency was observed in the month of May. The maximum numbers of systems were formed during postmonsoon months (October, November, and December), with November having the maximum cyclonic system formation. The total number of cyclonic systems formed in the Arabian Sea was less when compared with those formed in the Bay of Bengal. As can be seen in Figure  5.2, in the monthly frequencies of tropical cyclones formed in the Arabian Sea, no cyclonic system was observed during February and March over 80 years. During the premonsoon season, maximum frequency was observed in the month of May, most of which were of the severe type. During the monsoon season, most of the cyclones were formed in the month of June, and July, August, and September had a smaller number of cyclonic disturbances during these years. Like the Bay of Bengal, the maximum numbers of tropical storms were formed during the postmonsoon season, with maximum frequency in the month of November, and it again decreased in the month of December. Maximum numbers of storms were formed in the month of November. Figure  5.3 shows the increasing trends in seasonal frequencies in tropical cyclones over the Bay of Bengal these 80 years. The frequency of tropical cyclones, as can be seen in Figure 5.3, from 1931 to 2012, is least during the winter season (January and February). Very rarely did storms occur during this period and these too were of low intensity. In these 80 years, only about seven storms formed during this season. From Figure  5.4, it can be observed that the frequency of tropical cyclones over the Arabian Sea during the winter season is only 1 in 80 years. The increasing trend is observed in seasonal frequency. There is sharp increase in frequency during the postmonsoon season (45) as compared with monsoon season (25).

100 80 60 40 20 0 Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

–20

Figure 5.1  Monthly variation in frequency of tropical cyclones over the Bay of Bengal, 1931–2012.

68  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION 25 20 15 10 5 0 Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

Figure 5.2  Monthly frequency of tropical cyclones over the Arabian Sea, 1931–2012. 200 150 100 50 0 JF

–50

MAM

JJAS

OND

Figure 5.3  Variation in seasonal frequency of tropical cyclones over the Bay of Bengal, over 1931–2012. JF = January‐ February; MAM = March, April, May; JJAS = June, July, August, September; OND = October, November, December. 60 40 20 0 JF

MAM

JJAS

OND

Figure 5.4  Variation in seasonal frequency of tropical cyclones over the Arabian Sea, 1931–2012. JF = January‐ February; MAM = March, April, May; JJAS = June, July, August, September; OND = October, November, December.

5.3.2. Yearly Variations in Frequencies of Tropical Cyclones During the Premonsoon, Monsoon, and Postmonsoon Seasons For yearly variations in frequencies of tropical cyclones during various seasons, Figures 5.5–5.7 have been plotted. It can be seen from Figure 5.5 that there is consistenct in the annual frequency of tropical cyclone during premonsoon season over the Bay of Bengal in the period 1931–2012. Little fluctuation can be observed in Figure 5.5, where some years are devoid of any cyclones in the premonsoon season. Some years (1936, 1962, 1965, 1976, and 2009) had two but most years had only one tropical cyclone in the period 1931–2012 over this region in premonsoon season. The last cyclone that was formed in this season was in 2010. There were 53 cyclonic systems formed during the premonsoon season over the period 1931–2012 in which most are of very low intensity

systems having less frequency. From Figure 5.6, it can be seen that there has been a decreasing trend in the frequency of tropical cyclones during the monsoon season. A lot of fluctuations can be observed in Figure 5.6, where there was no tropical cyclone formation in some years, during this season. Maximum frequency was observed in 1940 (4). Most of the years witnessed only one cyclonic system formation during this period. The last tropical cyclone that was formed in this season was in the year 2005. From then no such system was formed until 2012. Only 73 cyclonic systems were formed during monsoon season from 1931 to 2012. Of these systems, most were of low intensity (simple cyclonic storms). The maximum number of storms were formed during the postmonsoon period (October, November, and December). About 172 storms formed in this season in these 80 years, out of which the majority were of the severe type. Figure  5.7 shows the variation in the frequency of tropical cyclones

Tropical Cyclones Over the North Indian Ocean in Changing Climate  69 March, April, May

3 2 1

2011

2008

2005

2002

1999

1996

1993

1990

1987

1984

1981

1978

1975

1972

1969

1966

1963

1960

1957

1954

1951

1948

1945

1942

1939

1936

1933

0

Figure 5.5  Annual variation in frequencies of tropical cyclones over the Bay of Bengal in the premonsoon season (March, April, May) 1931–2012. June, July, August, September

6 4 2

2012

2009

2006

2003

2000

1997

1994

1991

1988

1985

1982

1979

1976

1973

1970

1967

1964

1961

1958

1955

1952

1949

1946

1943

1940

1937

1934

1931

0

Figure 5.6  Annual variation in frequencies of tropical cyclones over the Bay of Bengal in monsoon season (June, July, August, September). 6

October, November, December

5 4 3 2 1

2012

2009

2006

2003

2000

1997

1994

1991

1988

1985

1982

1979

1976

1973

1970

1967

1964

1961

1958

1955

1952

1949

1946

1943

1940

1937

1934

1931

0

Figure 5.7  Annual variation in frequencies of tropical cyclones over the Bay of Bengal in postmonsoon season (October, November, December).

during the postmonsoon season. In this graph, a lot of fluctuation can be observed, where the trend first shows an increase and then a decrease in frequency in recent years. Maximum frequency that was observed was five (1966, 1968, and 1976). The overall variations of annual frequencies of tropical cyclone in postmonsoon season over the Bay of Bengal in the time period of 1931–2012 present a decreasing trend in this figure. Figure 5.8 shows that the trend in annual frequency of tropical cyclones in the premonsoon season over the Arabian Sea is almost constant and does not show much fluctuation in the period 1931–2012. Only in year 1974, the frequency was two (highest) and all other years have a frequency of one. Most of the years did not witness the formation of any cyclones. As seen in Figures  5.9 and

5.10, in the annual frequencies of tropical cyclones over the Arabian Sea, a lot of fluctuation was observed where some years show no cyclonic system formation at all. From both figures, it can be seen that the annual frequency of tropical cyclones in the Arabian Sea during both the monsoon and postmonsoon season shows a decreasing trend in recent years. There are large gaps in the occurrence of cyclonic systems seen in the Arabian Sea, especially in the monsoon season. There was no cyclonic disturbance during the years 1986–1991 in the Arabian Sea in both the monsoon and premonsoon seasons, which is the largest gap in the occurrence of tropical cyclones. The figures also show that the formation of tropical cyclones in midyear is more frequent than the beginning and recent years for both of the seasons.

70  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION March, April, May

3 2 1

2011

2008

2005

2002

1999

1996

1993

1990

1987

1984

1981

1978

1975

1972

1969

1966

1963

1960

1957

1954

1951

1948

1945

1942

1939

1936

1933

0

Figure 5.8 Annual variation in frequency of tropical cyclones during 1931–2012 over the Arabian Sea in ­premonsoon season (March, April, May). 3

June, July, August, September

2 1 2012

2009

2006

2003

2000

1997

1994

1991

1988

1985

1982

1979

1976

1973

1970

1967

1964

1961

1958

1955

1952

1949

1946

1943

1940

1937

1934

1931

0

Figure 5.9  Annual variation in frequency of tropical cyclones during 1931–2012 over the Arabian Sea in monsoon season (June, July, August, September). October, November, December

4

y = –0.003x + 1.4136 R2 = 0.0182

3 2 1

2012

2009

2006

2003

2000

1997

1994

1991

1988

1985

1982

1979

1976

1973

1970

1967

1964

1961

1958

1955

1952

1949

1946

1943

1940

1937

1934

1931

0

Figure 5.10 Annual frequency of tropical cyclones during 1931–2012 over the Arabian Sea in postmonsoon season (October, November, December).

5.3.3. Trends in Decadal Frequency of Tropical Storms in Different Seasonal Months The variation in decadal frequency of tropical cyclones during different seasons over the Bay of Bengal is shown in Table  5.1. The variation of frequency of tropical cyclones was almost constant during the winter season (JF) and does not show much fluctuation. In the premonsoon season (MAM), the frequency of tropical cyclones showed little fluctuation, but it was highest during the decade 1963–1972. After that, there was some decrease and again it increased in the last decade. The decadal frequencies of tropical cyclone showed a decreasing trend in recent decades. Minimum frequency for the monsoon and postmonsoon season was observed in the last decade (2003–2012). Minimum frequency for the premonsoon season was observed in the decade 1943–1952. Figure 5.11 presents the decadal frequencies of tropical cyclone in the Bay of Bengal. Over the eight decades, the overall frequency of cyclonic storms shows a decreasing trend. It

began decreasing after decades (1933–42) to (1953–1962) after which the frequency start to increase and reach its second highest level in the decade (1963–1972) then again started decreasing. Minimum frequency was observed in the recent two decades (1993–2002 and 2003–2012). It can also be seen from Table 5.1 that the maximum frequency of tropical cyclones is in premonsoon season which decreased in the first two decades (1933–1942 to 1943–1952) then again showed an increasing trend in the decades 1953–1962 to 1963–1972 after which it again decreased for the next three decades (1973–1982, 1983– 1992 to 1993–2002) followed by an increase in the last decade (2003–2012). Thus, the frequency shows an oscillating trend, first decreasing then increasing and again decreasing and increasing. During the monsoon season, the storms were more frequent in early decades (1931– 1942) and then the decreasing trend was observed over the Bay of Bengal. Only slight increase was observed in mid- decades (1963–1972 and 1973–1982) and again it decreased after 1973–1982. The frequency in the last

Tropical Cyclones Over the North Indian Ocean in Changing Climate  71 Table 5.1  Decadal Frequencies of Tropical Cyclones in Different Seasons Over the Bay of Bengal.

Table 5.2  Decadal Frequencies of Tropical Cyclones in Different Season Over the Arabian Sea.

Years

JF

MAM

JJAS

OND

TOTAL

1933–1942 1943–1952 1953–1962 1963–1972 1973–1982 1983–1992 1993–2002 2003–2012

1 1 0 1 1 1 1 1

8 3 6 11 7 4 4 8

21 14 8 10 11 4 2 1

23 18 15 30 28 24 17 13

53 36 29 52 47 33 24 23

60 50 40 30 20 10

2 01

2 –2 03 20

19

93

–2

00

2 –9

2 83 19

–8

2 73 19

–7

2 63 19

–6

2 53

–5 19

43 19

19

33

–4

2

0

Figure 5.11  Decadal frequencies of tropical cyclones over the Bay of Bengal.

decade, 2003–2012, was the minimum of these eight decades (which was as low as one) whereas during 1933– 1942 it was 21, which is a maximum for these eight decades. It can be observed that not only in monsoon season but also during postmonsoon season there is decreasing frequency of disturbances during the recent decades. The frequencies first show a decrease during the first three decades (1933–1942 to 1953–1962), then a sharp increase (about 50%) during 1963–1972, and subsequently a decreasing trend. The maximum number of cyclonic disturbance occurred during the decade 1963– 1972 and from that point there has been continuous reduction in numbers of such disturbances. In the last decade 2003–2012, there were 13 cyclonic disturbances. Thus the overall trend was decreasing. The decadal variation in frequency of tropical cyclones over the Arabian Sea is shown in Table 5.2. For the winter season (JF), only the decade 1933–1942 saw a tropical cyclone, while other decades are devoid of any cyclonic system. In the premonsoon season (MAM), the frequency of tropical cyclones shows little fluctuation, where it is highest during the decade 1973–1982 and 1993–2002 (4)

Years

JF

MAM

JJAS

OND

TOTAL

1933–1942 1943–1952 1953–1962 1963–1972 1973–1982 1983–1992 1993–2002 2003–2012

1 0 0 0 0 0 0 0

2 2 3 2 4 1 4 2

2 4 3 3 5 1 4 3

4 6 4 10 6 1 7 6

9 12 10 15 15 3 15 11

and lowest in 1983–1992 (1). In the table, decadal frequency of tropical cyclones shows a decreasing trend in recent decades. Minimum frequency for the monsoon and postmonsoon seasons was observed for 1983–1992 (1). The postmonsoon season has higher frequency than the premonsoon and monsoon seasons; its maximum frequency reach to 10, shown in decade 1963–1972. In monsoon season, maximum frequency was 5 in the decade 1973–1982. Figure 5.12 presents the decadal frequency of tropical cyclones (CS + SCS) in the Arabian Sea. It shows fluctuations, as it shows alternate increase and decrease in frequencies. Maximum frequency was observed during 1993–2002 and minimum during 1983–1992. The frequency gradually decreased from 1973–1982 to 1983– 1992, and then there was sharp increase in the frequency from 1983–1992 (3) to 1993–2002 (16). It again decreases in the last decade (2003–2012) to 10 in number. It can also be seen from Table  5.2 that the decadal frequencies of tropical cyclones in the premonsoon season over the Arabian Sea shows an alternate increasing and decreasing pattern. Maximum frequency was observed during the decades 1973–1982 and 1993–2002 (4). And least number of cyclones occurred during the decade 1983–1992 (1). The frequency during the last decade (2003–2012) was 2. During the monsoon season over the Arabian Sea, maximum numbers of tropical cyclones were formed during the decade 1973–82 (5) and the least number was observed in the decade 1983–1972 (1). There was a sharp fall in frequency between two decades. The frequency shows an alternate rise and fall during these eight decades. The decadal frequency of tropical cyclones in the postmonsoon season over the Arabian Sea is also shown in Table 5.2. It can be seen that the frequencies first show an increase and again decrease in recent decades. The trend of decadal frequency was observed to be fluctuating, and a minimum number of systems were formed during the decade 1983–1992 (1). The variation in the frequencies of these systems was almost similar in all the seasons, that is, premonsoon, monsoon, and postmonsoon season that showed alternate increase and decrease in the numbers in each decade. And all seasons (premonsoon, monsoon,

72  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

2 –2 03

20

19

93

–2

01

2 00

2 –9

2 83 19

73

–8

2 19

63

–7

2 19

19

53

–6

2 –5 43

19

19

33

–4

2

18 16 14 12 10 8 6 4 2 0

Figure 5.12  Decadal frequency of tropical cyclones over the Arabian Sea.

postmonsoon) have lowest frequency of tropical cyclone during the decade 1983–1992 (1). 5.3.4. Trends in Decadal Frequency of Tropical Cyclones During the Premonsoon, Monsoon, and Postmonsoon Seasons Seasonal trends of decadal frequency of tropical cyclones over the Bay of Bengal can be observed in Table  5.3. It can be seen that the frequency of severe cyclonic storms, in the premonsoon season, was more than the cyclonic storms. The frequencies of cyclonic storms and severe cyclonic storms showed alternate increasing and decreasing trends. The frequency of severe cyclonic storms first show a decrease in the first three decades (1933–1942 to 1953–1962), which again increased during 1963–1972 (from 2 in 1953–1962 to 7 in 1963–1972). It again showed a decreasing pattern and increased in last decade (from 3 in 1993–2002 to 6 in 2003–2012).The frequency of monsoonal cyclonic disturbances show decadal oscillations with their number fluctuating from as low as 1 for the decade 2003–2012 to as high as 21 during 1933–1942. The fre-

quency of monsoonal cyclonic disturbances in the north Indian Ocean has shown a significant decreasing trend during the twentieth century, registering about a 50% decrease from the beginning to the end of the century. The decreasing trend is more pronounced during recent decades. It has been observed that the frequencies of cyclonic storms and severe cyclonic storms first showed an increase during the middecades and then again showed a decreasing trend in recent decades. The frequency was more during 1963–1972 and 1973–1982 in both monsoon and postmonsoon seasons. In the postmonsoon season, the frequency of cyclonic storms increased in the last decade (from 4 in 1993–2002 to 8 in 2003–2012) and that of severe cyclonic storms decreased from 12 in 1993–2002 to 5 in 2003–2012. It is also observed that in the first three decades (1933–1942 to 1953–1962) of the postmonsoon season, the frequency of cyclonic storm was more than severe cyclonic storm, whereas from 1963–1972 to 1993–2002, the frequency of severe cyclonic storms was more, which again shows a decrease in last decade (2003–2012). Table  5.4 shows the decadal frequencies of cyclonic storms and severe cyclonic storms during the premonsoon, monsoon, and postmonsoon seasons over the Arabian Sea. It can be predicted from the table that most of the severe cyclonic storms were formed in the postmonsoon season. Maximum numbers of cyclonic storms were formed in the decade 1963–1972 during the postmonsoon season. In the premonsoon season the formation of severe cyclonic storms was more than that of the cyclonic storms. In the premonsoon season, the frequency of severe cyclonic storms was more in the last two decades (1993–2002 and 2003– 2012) when compared to the previous decades (1963– 1972 to 1983–1992). Frequency of cyclonic storms shows a decrease during the last decade (2003–2012) during all seasons, whereas severe cyclonic storm shows little increase as compared with those ­ during 1993–2002.

Table 5.3  Decadal Frequencies of Cyclonic Storms and Severe Cyclonic Storms During the Premonsoon, Monsoon, and Postmonsoon Seasons. MAM

JJAS

OND

Years

CS

SCS

TOTAL

CS

SCS

TOTAL

CS

SCS

TOTAL

1933–1942 1943–1952 1953–1962 1963–1972 1973–1982 1983–1992 1993–2002 2003–2012

2 1 4 3 3 0 1 2

6 2 2 7 5 4 3 6

8 3

16 12 7 2 5 4 1 1

5 2 1 8 6 0 1 0

21 14 8 10 11 4 2 1

14 13 8 8 14 10 4 8

9 5 7 22 14 14 13 5

23 18 15 30 28 24 17 13

10 8 4 4 8

Tropical Cyclones Over the North Indian Ocean in Changing Climate  73 Table 5.4  Decadal frequencies of Cyclonic Storms and Severe Cyclonic Storms During the Premonsoon, Monsoon, and Postmonsoon Seasons. MAM

JJAS

OND

Years

CS

SCS

TOTAL

CS

SCS

TOTAL

CS

SCS

TOTAL

1933–1942 1943–1952 1953–1962 1963–1972 1973–1982 1983–1992 1993–2002 2003–2012

1 1 0 1 3 0 2 0

2 1 3 1 1 1 2 2

3 2 3 2 4 1 4 2

2 2 0 3 2 0 3 1

0 2 3 0 3 1 1 2

2 4 3 3 5 1 4 3

1 4 1 7 1 1 3 2

3 2 3 3 5 0 4 4

4 6 4 10 6 1 7 6

5.3.5. Tricadal Variations in Frequencies of Tropical Cyclones Regarding the tricadal variation in frequencies of tropical cyclones over the Bay of Bengal and Arabian Sea, no such trend was observed during the winter season, as shown in Table 5.5. While the frequencies of cyclones showed a decreasing trend both during monsoon and postmonsoon period, an increase was observed in the premonsoon season in the recent 30 years, although the increase was not significant (from 23 in 1953–1982 to 26 in 1983–2012). During the recent 30 years, there was a decrease of 22 from 1953–1982 to 1983–2012 during monsoon season and decrease of 19 from 1953–1982 to 1983–2012 during postmonsoon season. From Table 5.6, a 30-year variation of frequency over the Arabian Sea is observed. It can be seen that the seasonal frequencies of tropical cyclones show a decreasing trend in the recent 30 years (1983–1992) as compared to the previous 30 years (1953–1982) in all the seasons. But the decrease was not as sharp as observed in the Bay of Bengal cyclonic storms. 5.3.6. Area of Formation of Severe Cyclonic Storms: North Bay (NB), South Bay (SB), and Central Bay (CB) Figure  5.13 shows that most of the severe cyclonic storms that formed in the Bay of Bengal were formed in the South Bay (99) and then in the Central Bay (30); the least number of severe cyclonic storms was formed in the North Bay (18). Especially in the month of December, all the systems were formed in the South Bay over the 80year period of 1931–2012 (Table  5.7). Most of the cyclonic storms were formed in the South Bay during premonsoon, postmonsoon, and winter seasons, whereas during monsoon season only six tropical cyclones (3 CS and 3 SCS) were formed, which was least of all the systems when compared with those formed in the North Bay

Table 5.5  Tricadal Variations in Frequencies of Tropical Cyclones. Years

JF

MAM

JJAS

OND

1953–1982 1983–2012

2 3

24 26

29 7

73 54

Table 5.6  Tricadal Frequencies of Tropical Cyclones (CS + SCS). Years

JF

MAM

JJAS

OND

1953–1982 1983–2012

0 0

9 7

11 8

20 15

SB NB CB

Figure 5.13 The frequencies of severe cyclonic storms formed over South Bay(SB), Central Bay(CB), and North Bay(NB).

and Central Bay. During the monsoon season, a maximum number of cyclones were formed in the North Bay, which showed the least formation of cyclonic systems in other seasons. The total number of cyclonic system formed in North Bay is 61 which is the lowest among all seasons. Central Bay has a total of 67 tropical cyclones, and South Bay is associated with the maximum number of cyclonic system in all seasons (178). A description for the area of formation of cyclonic storms and severe cyclonic storms for the Arabian Sea has not been included here because of insufficient data.

74  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

E/NE

0

S/SW 2

20

03

–2

01

00

–2

–9

2

N/NW

10

2

20

93

The direction of movement of cyclones formed over the Bay of Bengal can be observed in Table 5.8. It shows that most of the cyclones that formed over the Bay of Bengal moved north to northwest, and some also first move north to northwest then recurve to the east to northeast. Out of those cyclones that move north to northwest, many first go to the northwest and then move northerly before landfall. Only some of the systems that moved northwest were of the severe type and most were simply cyclonic storms. Whereas those that moved to the east/ northeast were mostly of the severe type that affected northeastern states. Some of the systems did not show much movement and showed a practically stationary behavior; they did not cause much damage. Apart from this, there were very few systems that moved south to southwest (very rare) and did not cause much damage. From Figure 5.14, it can be seen that, on decadal ­variation in the direction of movement of cyclonic storms (over the Bay of Bengal), those moving toward east–northeast showed less fluctuation and showed constant behavior, whereas those moving north–northwest first showed a sharp increase in middecade then again showed continuous decrease in number in the last three decades. The storms that move north or northwest show a decreasing trend in recent years. Arabian Sea cyclone directional movement before landfall is given in Table 5.9 and shows that the movement of the tropical cyclones for most is north to northwest and that some move east to northeast. The difference between the two is not much. The decadal trend of direction of movement of these systems also shows large fluctuations. The least number of cyclonic storms moved north‐ northwest or east‐northeast during the decade 1983–1992. Very few cyclonic systems moved south‐southwest. There are some systems that initially formed in the Bay of Bengal (as low pressure areas or depressions) and then

30

19

5.3.7. Direction of Movement of Cyclones (Before Landfall) Formed

40

2

There appears to be many fewer tropical cyclones when compared with the Bay of Bengal.

–8

6 69 13

83

6 58 20

0 0 0 3 0 0 1 3

19

12 3 8

16 10 12 29 21 10 11 11

2

34 3 13

35 25 17 21 27 21 13 7

–7

2 25 11

SW

73

1 14 2

E/NE

19

0 3 0

2

0 3 0

N/NW

1933–1942 1943–1952 1953–1962 1963–1972 1973–1982 1983–1992 1993–2002 2003–2012

63

SCS

19

CS

2

SCS

–6

CS

53

SCS

19

CS

2

SCS

–5

CS

Years

43

OND

–4

JJAS

19

North Bay (NB) South Bay (SB) Central Bay (CB)

MAM

33

Area of formation

JF

Table 5.8  The Direction of Movement of Tropical Cyclones in the Bay of Bengal.

19

Table 5.7  The Frequency of Cyclonic Storms (CS) and Severe Cyclonic Storms (SCS) Formed in North Bay, South Bay and Central Bay.

Figure 5.14 The frequency of tropical cyclones in the Bay of Bengal moving toward east/northeast, north/northwest, and south/southwest.

Table 5.9  Direction of Movement of Tropical Cyclones in the Arabian Sea. Years 1933–1942 1943–1952 1953–1962 1963–1972 1973–1982 1983–1992 1993–1902 2003–2012

W/NW

E/NE

SW

2 8 6 8 8 2 9 5

5 4 4 3 7 1 4 6

0 0 0 3 1 0 0 1

took westward movement with continuous increase in intensity and then following the same course emerged into the Arabian Sea before falling. Figure 5.15 shows the direction of movement of tropical cyclones in the Arabian Sea. It can be observed that, on decadal variation in the direction of movement of cyclonic storms, those moving west‐northwest show high fluctuation. First it increases in recent decades then constantly fluctuates in the middecades, after which it sharply decreases in last decades. Overall, the graph for storms moving west‐northwest presents an increasing trend. The storms that move east‐ northeast first show almost a constant trend until the middecades, then tend to increase and then decrease and

Tropical Cyclones Over the North Indian Ocean in Changing Climate  75 12 10 8

W/NW

6

E/NE

4

SW

2

Linear (SW)

19

33 19 –42 43 – 19 52 53 – 19 62 63 – 19 72 73 – 19 82 83 – 19 92 93 – 20 02 03 –1 2

0

Figure 5.15 The direction of movement of tropical cyclones over Arabian Sea.

again increase from the middecade to last decade. Only a few storms move south to southwest, and they can be seen only in the middecades and last decades. 5.4. ­CONCLUSIONS The system of cyclonic storms formed in the Bay of Bengal over this 80-year period total 306 and those in the Arabian Sea total 92. Thus, the frequency is less in the Arabian Sea than in the Bay of Bengal. The average annual frequency of tropical cyclones in the north Indian Ocean (Bay of Bengal and Arabian Sea) is about five (about 5–6% of global annual average) and about 80 cyclones form around the globe in the year. The frequency is more in the Bay of Bengal than in the Arabian Sea, the ratio being 4:1. The monthly frequency of tropical cyclones in the north Indian Ocean displays a bimodal characteristic with a primary peak in November and secondary peak in May. The months of May–June and October–November are known to p ­ roduce cyclones of severe intensity. Tropical cyclones developing during the monsoon months (July to September) are generally not so intense. The frequency of cyclones is less in the Arabian Sea than in the Bay of Bengal, maybe because cyclones that form over the Bay of Bengal are either those that develop in situ over the southeast Bay of Bengal and adjoining Andaman Sea or remnants of typhoons over the northwest Pacific that move across the South China Sea to the Indian seas. As the frequency of typhoons over the northwest Pacific is high (about 35% of the global annual average), the frequency in the Bay of Bengal also increases. The cyclones over the Arabian Sea originate over the southeast Arabian Sea (that includes the Lakshadweep area), and remnants of the Bay of Bengal cyclones move toward it. As most of the cyclones that originate in the Bay of Bengal dissipate on land, few reach the Arabian Sea. In addition, the Arabian Sea is relatively colder than the Bay of Bengal so fewer systems are formed compared with the latter. The frequency of tropical cyclones showed a decreasing trend in recent decades as compared

with earlier ones, the possible reason being the weakening of Hadley circulations, due to an increase in upper tropospheric temperatures (due to global warming), which is responsible for the increased sea surface temperature (SST). The tropical cyclones normally move west‐northwestward or northwestward. They may change their direction of movement toward the north, which is accompanied by the decrease in the speed of cyclonic storms. A larger fraction of such storms recurve toward the northeast at high speed. This trend is not inconsistent with recent climate model simulation that a doubling of CO2 may increase the frequency of the most intense cyclones (Webster et al., 2005). ­ACKNOWLEDGMENTS The authors wish to thank the India Meteorology Department (IMD) for providing weekly‐monthly weather reports and IMD Journal “Mausam” issues. The authors also express their sincere thanks to NOAA (National Oceanic and Atmospheric Administration) for providing the necessary data. ­REFERENCES Chan, J. C. L. (2006). Comments on changes in tropical cyclone number, duration, and intensity in warming environment. Science, 309, 1884–1846. Chan, J. C. L., & Liu, K. S. (2004). Global warming and western north Pacific typhoon activity from an observational ­perspective. Journal of Climate, 17, 4590–4602. Deo, A. A., Ganer, W. Y., & Nair, G. (2011). Tropical cyclone activity in global warming scenario. Nat. Hazard, 59, 771–786. Elsner, J. B., Kossin, J. P., & Jagge, T. H. (2008), The increasing intensity of strongest tropical cyclone. Nature, 455, 92. Emanuel, K. A. (1987). The dependence of hurricane intensity on a ­climate. Nature, 326, 483–485. Emanuel, K. A. (2005). Increasing destructiveness of tropical cyclones over the past 30 years. Nature, 436, 686–688. Evan et. al. (2011). Arabian Sea tropical cyclones intensified by emissions of black carbon and other aerosols. Nature, 479, 94. Gray, W. M. (1968). Global view of the origin of tropical disturbances and storms. Monthly Weather Review, 96, ­ 669–700. Klotzbach, P. J. (2006). Trends of global tropical cyclone activity over the past 20 years (1986–2005), Geophysical Research Letters, 33, L10805. Knutson T. R., McBride, J. L., Chan, J., et al. (2010). Tropical cyclone and climate change. Nature Geoscience, 3, 157–162. Knutson, T. R., & Tuleya, R. E. (2004). Impact of CO2 induced warming on simulated hurricane intensity and ­precipitation: Sensitivity to the choice of climate model and  convective parameterization. Journal of Climate, 17, 3477–3495. Knutson, T. R., Tuleya, R. E., & Kurihara, Y. (1998). The 365 recent increase in Atlantic hurricane activity: Causes and implications. Science, 366 (293), 474–478.

76  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Krishna, K. M. (2009). Intensifying tropical cyclone over north Indian Ocean during summer monsoon global warming. Global and Planetary Change, 65, 12–16. Landsea, C. W., Piekle, R. A. Jr., A. M. Mestas-Nuñez, A. M., & Knaff, J. A. (1999), Atlantic basin h ­ urricanes: Indices of climate changes, Climate Change, 42, 89–129. Landsea C. W. (2000). Climate variability of tropical cyclone: past, present and future. In R. A. Pielke Sr. & R. A Pielke Jr. (Eds.) Storms (pp. 220–241). New York: Routledge. Sriver, R. L., & Huber, M. (2007). Reply to comment by R. N. Maue & R. E. Hart on “Low frequency variability in globally

integrated tropical cyclone dissipation,” Geophysical Research Letters, 34, L11704. Wang B., Yang, Y., Ding, Q., Murakami, H., & Huang, F. (2010). Climate control of the global tropical storm days (1965–2008), Geophysical Research Letters, 3, L07704 Webster, P. J., Holland, G. J., Curry, J. A., & Chang, H‐R. (2005). Changes in Tropical Cyclone number, duration, and intensity in warming environment. Science, 309, 1844–1846.1575. Yu, J., & Wang, Y. (2009). Response of tropical cyclone potential intensity over north Indian Ocean to global warming. Geophysical Research Letters, 36, 1–9.

6 Simulation of Intensity and Track of Tropical Cyclones Over the Arabian Sea Using the Weather Research and Forecast (WRF) Modeling System with Different Initial Conditions (ICs) Sushil Kumar1, Ashish Routray2, Prabhjot Singh Chawla1, and Shilpi Kalra1

ABSTRACT An attempt is made to simulate the intensity and track of tropical cyclones (TCs) over the Arabian Sea (AS) using the Weather Research and Forecast (WRF) model with different initial conditions (ICs). For this purpose, two extremely severe cyclonic storms (ESCS), Chapala and Megh, are considered for evaluation of the performance of the WRF model in prediction of track and intensity of TCs on the basis of a total of 11 different initial ­conditions. The overall performance of the WRF model has been found reasonably good in predicting TCs over the basin. The model has realistically predicted the tracks of the TCs with most of the ICs at different stages of the storms. The WRF model is found to be more skillful for track prediction of TCs when initialized at the severe cyclonic stage (SCS) rather than at the cyclonic stage (CS) or lower. Therefore, the track errors are lesser with the advancement of model ICs in each TC case. The model is more capable to predict the landfall location than the landfall time of the storms. 6.1. ­INTRODUCTION

processes of cyclogenesis and the seasonal variations in climatology of genesis location and frequency of TCs through observational information. The prediction of track and intensity of a TC is a challenging task because its movement is highly variable (Raghvan & Sen Sarma, 2000). Srinivas et al. (2012) performed 11 real‐time simulations for the prediction of cyclone Jal using the high resolution advanced researched WRF (ARW) model over the Bay of Bengal (BoB) with different initial conditions (ICs). Mohanty et al. (2004) simulated the Orissa super cyclone (1999) using a mesoscale model for producing a five-day forecast of the storm for various parameters including sea level pressure, horizontal wind, and rainfall, and it is observed that the intensity of the storm is well matched on day‐1 and day‐2 as compared with day‐3. Rao and Prasad (2006) performed experiments using convection, planetary boundary layer, and explicit moisture process to study the sensitivity of cyclone track prediction. The result showed that the convective processes play an important role in cyclone track prediction. Davis et al. (2008) simulated hurricane Katrina on a numerical

A rapidly rotating storm system is known as a tropical cyclone (TC), which generates strong winds and a spiral arrangement of thunderstorms. It is one of the most devastating hazards to occur in nature and produces heavy to very heavy rainfall on coastal and nearby areas. Significant development of TC track prediction has been done by various models in the North Indian Ocean (NIO) (RSMC Report, 2008). The generation of TCs is attributed mainly to different natural factors like sea surface temperature (SST) greater than 26.5°C (Palmen, 1948), large Coriolis force (LCF), high low-level relative vorticity, convective instability, low horizontal wind shear, and so on. Gray (1968) observed that distinct Ekman or frictional veering winds are found in the subcloud layer of approximately 10° latitude. Gray (1975) attempted to study the physical Gautam Buddha University, Greater Noida, India National Centre for Medium Range Weather Forecasting (NCMRWF), Ministry of Earth Sciences, Uttar Pradesh, India 1  2 

Techniques for Disaster Risk Management and Mitigation, First Edition. Edited by Prashant K. Srivastava, Sudhir Kumar Singh, U. C. Mohanty, and Tad Murty. © 2020 John Wiley & Sons, Inc. Published 2020 by John Wiley & Sons, Inc. 77

78  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

model and observed the inner core structure and size of the vortex, which was sensitive to resolution and surface momentum exchange of the model. Pattanaik and Rao (2009) observed the real‐time forecast of TC Nargis with different ICs using the ARW model and noticed that it did not follow the natural westerly/northwesterly track. Osuri et  al. (2012) simulated the TCs over the North Indian Ocean (NIO) using the ARW model with six combinations of physical parameterization schemes and found that the better prediction of intensity, track, and rainfall with a combination of the YSU PBL scheme and KF convection schemes. Mohapatra et al. (2013) evaluated the intensity of operational TCs over the NIO and found that there is a need for further reduction in forecast errors for improved data availability, assimilation of more observational data in the numerical weather prediction (NWP) models. Mohapatra et  al. (2015) calculated the tropical cyclone landfall point forecast error (LPE) and landfall time forecast error (LTE) by comparing it with landfall forecast issued by the IMD during 2003–2014 over NIO. Various developments are going to improve the accuracy of numerical models for TC track simulation. The model’s physical parameterizations, which consist of cumulus convection, surface fluxes of heat, moisture, momentum, and vertical mixing in the planetary boundary layer, play a major role in simulation of the TC data by Kumar et al. (2015). Tropical cyclones are the subject of considerable attention among scientists, policy makers, and media. Particular devotion is toward the role of anthropogenic climate change in the patterns of cyclonic activities (Pielke, 2007). Increase in damage from extreme events due to greenhouse gas emissions is a potentially impacted concern (Mendelsohn et  al., 2012). In recent decades, natural disaster losses around the globe have increased significantly and the contribution of tropical cyclones in this trend is noteworthy. Better management to reduce the losses due to tropical cyclones and natural hazards is required. Climate change has had an impact on the trend of an increase in the intensity and frequency of tropical cyclones and enhances the natural disasters along coastal regions (Mendelsohn et  al., 2012). The TCs generate a storm surge, which is an unusual rise of sea level near the coastal region. It results in the sea water inundation of low lying regions of the coast. These disasters occur simultaneously, which is a meaningful challenge to natural disaster management. In recent years, the damage has increased due to hurricanes or tropical cyclones, because of the increase in population and wealth along the coastal area (Pielke, 2007). Future losses can be minimized with accurate forecasting of the TC well in advance and the development of proper dissemination systems. Understanding future risk should be improved with climate change projection scenarios and would reduce

damage to society. This chapter focuses on the application of a nonhydrostatic mesoscale model to simulate the physical and dynamic processes of TCs Chapala and Megh, formed in October–November 2015 over the Arabian Sea. 6.2. ­SYNOPTIC CONDITIONS ASSOCIATED WITH THE TROPICAL CYCLONES 6.2.1. ESCS Chapala In the morning of 28 October 2015, ESCS Chapala was observed on a low‐pressure area over the southeast Arabian Sea. It became a depression and deep depression (DD) on the same evening and moved northwestward near latitude 13°N and longitude 64.7°E. At 00 UTC of 29 October, the DD intensified into a cyclonic storm (CS) over the east‐central Arabian Sea. During the period 09 UTC of 29 October to 18 UTC of 29  October while moving west‐northwestward, it further intensified into a severe cyclonic storm (SCS), then into a very severe cyclonic storm (VSCS), and then into an ESCS. On 01–02 UTC of 3 November moving west‐northwestward, it crossed the Yemen coast to the southwest of Riyan as a VSCS at a wind speed of 65 knots near latitude 14.1°N and longitude 48.65°E. In the morning of 3 November while moving westward, it weakened into a SCS and into a CS by noon. It further weakened to a DD by midnight of 3 November. On 03 UTC of 4 November, the storm further weakened to a depression and marked as a low pressure area over Yemen near latitude 14.8°N and ­longitude 46.3°E (RMSC, 2015). 6.2.2. ESCS Megh It has been observed that a depression formed on 5 November at 00 UTC over the east‐central Arabian Sea from a low‐level circulation over Lakshadweep and neighborhood near latitude 14.1°N and longitude 66°E and moved westward/west‐southwestward. It became a cyclonic storm (CS) on 5 November at 12 UTC near latitude 14°N and longitude 64°E. It continued its movement in the west‐southwestward direction and further intensified into a SCS at 06 UTC on 7 November. Finally, it intensified into a VSCS at 15 UTC on 7 November near latitude 12.6°N and longitude 57.9°E and rapidly intensified into an ESCS at 03 UTC on 8 November near latitude 12.7° N and longitude 55.5°E (RMSC, 2015). In a short period interval of about 6 hr, it reached its peak intensity and weakened gradually into a VSCS at 00 UTC on 9 November near latitude 12.3°N and longitude 51.0°E. At 06 UTC of 9 November, while moving in a west‐northwestward direction, it weakened rapidly into a SCS at 21 UTC, into a CS at 03 UTC on

SIMULATION OF INTENSITY AND TRACK OF TROPICAL CYCLONES OVER ARABIAN SEA  79

10 November, and into a DD at 06 UTC of 10 November. At 03 UTC of 10 November, the TC moved northeastward and crossed the Yemen coast with a maximum ­surface wind (MSW) of 30 knots near latitude 13.4°N and longitude 46.1°E. At 15 UTC of 10 November, it weakened into a well‐marked low‐pressure area. 6.3. ­DESCRIPTION OF THE WEATHER RESEARCH AND FORECAST MODEL AND NUMERICAL EXPERIMENTS The Weather Research and Forecast (WRF) model (version 3.4) is configured with a single domain for two cyclones (Chapala and Megh) as given in Table  6.1. A single domain is set for both cyclones between 40°E and 78°E and between 5°N and 21°N over the Arabian Sea with 27 vertical levels and 9 km horizontal grid resolution with 469 x 208 grids. The initial and lateral boundary conditions for the WRF model were obtained from the National Centre for Environmental Prediction (NCEP), Final Analysis (FNL). The lateral boundary conditions were updated in six-hour intervals. The physics schemes used in this experimental study were the Yonsei University (Hong et  al., 2006) nonlocal diffusion scheme for PBL processes, Kain–Fritsch for cumulus convection (Kain & Fritsch, 1993), Purde Lin scheme (Lin et  al., 1983) for explicit moisture processes, five‐layer soil thermal diffusion model for surface processes, Rapid Radiation Transfer Model (RRTM) for longwave radiation (Mlawer et  al., 1997), and Dudhia (1989) scheme for shortwave radiation. The TCs Chapala and Megh both were initialized with six and five different ICs (24-hour interval) starting from 00 UTC of 28 October to 2 November and 00 UTC of 5 November to 09 November, respectively. The WRF model was integrated up to 12 UTC of 3 November for Chapala and 12 UTC on 10 November for Megh, respectively.

Table 6.1  WRF Model Configuration. Dynamics Horizontal resolution Domains of integration Time step Interval seconds Map projection Horizontal grid system Surface layer Initial/lateral boundary Cumulus scheme PBL scheme

Nonhydrostatic 9 km 40 E‐78 E 5 N‐21 N 50 s 21,600 s Mercator Arakawa C‐grid Thermal diffusion scheme Final Analysis (FNL) Kain–Fritsch YSU

6.4. ­RESULTS AND DISCUSSION 6.4.1. Intensity Prediction for Tropical Cyclones A total of 11 numerical simulations were carried out for the two tropical cyclones, that is, six initial conditions for ESCS Chapala and five initial conditions for ESCS Megh with 24-hour intervals. The time series of mean sea level pressure (MSLP) of Chapala and Megh are presented in Figures 6.1a and 6.2a, respectively. For the case of Chapala, IC 30 produced the lowest MSLP (950 hPa) on 12 UTC of 1 November and the experiment with IC 30 produced the highest MSLP (1,002 hPa) among all the series. It indicates that the experiment IC 07 produced the lowest MSLP (985 hPa) on 00 UTC of 9 November and the experiment with IC 05 produced the highest MSLP (1,007 hPa) among all the series in the case of Megh. It is observed from Figures 6.1a and 6.2a that the peak intensification occurred for the IC 30 and for IC 07, respectively, for Chapala and Megh. As per the IMD report of ESCS Chapala, the storm attained maximum atmospheric pressure drop between 09 UTC and 18 UTC of 30 October. Similarly, for the case of ESCS Megh, the storm attained maximum pressure drop between 03 UTC and 09 UTC of 8 November. It is noted that the experiments IC 29 and IC 07 for MSLP gave better comparison with IMD observations. The time variation of 10 m wind from different numerical experiments along with the IMD observations is presented in Figures  6.1b and 6.2b, for cyclones Chapala and Megh, respectively. IMD observation indicates that, in the case of Megh, the increase in 10 m surface wind occurred between 15 UTC and 18 UTC on 5 November and remained steady to 03 UTC on 7 November then increased to 09 UTC on 8 November to 49 m/s (IMD, 2015b). For Chapala, IMD estimated the increase in 10 m surface wind between 00 UTC of 29 October and 09 UTC of 30 October to 59 m/s and this peak intensity period became steady from 09 UTC to 21 UTC on 30 October then it decreased (IMD 2015a). The various numerical experiments of maximum winds indicate that the model underestimated the intensity for all the cases for both the cyclones. However, both the experiments show better delay intensification and weakening than the IMD observations. It is also noticed from the root mean square error (RMSE) of MSLP and 10 m wind from Figures 6.2c and 6.1c that it is less in the case of IC 07 for Megh. However, for the case of Chapala, the RMSE of MSLP and 10 m wind are less in the case of IC 29 and IC 01, respectively. The mean RMSE of 10 m wind and MSLP for all cases are significantly different for both cyclones. The combined mean absolute error in MSLP and 10 m surface wind for both the cyclones at different forecast length is given in Table 6.2. On the basis of 11 cases, it is

80  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION (a) 1020

MSLP (hPa)

1000 OBS

980

IC 28 IC 29

969

IC 30

940

IC 31

920

IC 02

IC 01

28

_0 0 28 _1 2 29 _0 0 29 _1 2 30 _0 0 30 _1 2 31 _0 0 31 _1 2 01 _0 0 01 _1 2 02 _0 0 02 _1 2 03 _0 0 03 _1 2

900

Time (dd_hr)

(b) 70

OBS

10 m Wind

60

IC 28

50

IC 29

40

IC 31

IC 30 IC 01

30

IC 02

20 10 0

28_0028_1229_0029_1230_0030_1231_0031_1201_0001_1202_0002_1230_0003_12

Time (dd_hr) 20 12.5782 18 11.1591 11.3438 10.5023 16

MSLP

14

14

10 m Wind

9.23701 10 7.5621

12

MSLP

12

10 16.378815.7941 8 15.1543 13.314 6 11.7898

8 18.0317 6

10mWind

(c)

4

4

2

2

10

0 28_00

29_00

30_00

31_00

01_00

02_00

I.C

Figure 6.1  Time evolution of (a) MSLP, (b) 10 m wind, and (c) RMSE of MSLP and 10 m wind for ESCS Chapala.

seen from Table 6.2 that the mean absolute error in MSLP is minimum initially and at 36 hr for 10 m surface wind. Wind field at the surface level 850 hPa is presented in Figure  6.3 (a–d) for both cyclones Chapala and Megh. Figure 6.3a depicts the wind profile for the IC 29 case at 00 UTC for Chapala, and it is observed that a strong

wind (60 m/s) is found around the center of the eye about 63°E and 13.8°N. Figure 6.3b shows the wind profile for the IC 29 case at landfall for Chapala, and it is observed that the wind velocity remains 60 m/s at 14.4°N and 52°E, and numerical landfall location is slightly different from observed landfall location. Figure 6.3c depicts the wind

SIMULATION OF INTENSITY AND TRACK OF TROPICAL CYCLONES OVER ARABIAN SEA  81 (a)

1020 1010

MSLP (hPa)

1000 990 980

OBS

970

IC 06

IC 05 IC 07

960

IC 08

950

IC 09

940 05_00 05_12 06_00 06_1207_00 07_12 08_00 08_1209_00 09_12 10_00 10_12

(b)

Time (dd_hr) 60

OBS IC 05

10 m Wind

50

IC 06 IC 07

40

IC 08 IC 09

30 20 10 0

05_00 05_12 06_00 06_12 07_00 07_12 08_00 08_12 09_00 09_12 10_00 10_12

(c)

Time (dd_hr) 25 17.32

20

MSLP 10 m Wind

18.01

20 18 16

15.49

MSLP

1.5 10

12

11.01 20.564

21.516

10 21.153 7.112

13.604

12.581

5

8

10 m Wind

14

6 4 2 0

0 5_00

6_00

7_00

8_00

9_00

Time (dd_hr)

Figure 6.2  Time evolution of (a) MSLP, (b) 10 m wind, and (c) RMSE of MSLP and 10 m wind for ESCS Megh.

profile for the IC 07 case at 00 UTC for Megh, and it is observed that a strong wind (40 m/s) is found at the center of the vortex about 56°E and 12°N over the Arabian Sea. Figure 6.3d depicts the wind profile for the IC 07 case at landfall, and it is observed that the vortex becomes weaker as wind velocity gets lower with 30 m/s as the cyclone reaches to the numerical landfall point at 13.7°N and 47°E, which is close to the observed landfall location. Figure 6.4a–d shows the mean sea level pressure for both cyclones Chapala and Megh. Figure 6.4a shows that at 00

UTC of IC 29, the lowest MSLP was 965 hPa at 63°E and 14°N at the center of the vortex. In Figure  6.4b when ESCS Chapala reached the landfall point, it got weak and MSLP reaches to 980 hPa. Figure  6.4c depicts the MSLP details for the IC 07 case at 00 UTC for Megh, and it is observed that the lowest atmospheric pressure was 1,007 hPa at 57°E and 13°N at the center of the vortex. In Figure  6.4d when cyclone Megh reaches the landfall point, the cyclone is weakened and pressure reaches to 1,000 hPa.

82  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Table 6.2  Mean Absolute Errors of MSLP (hPa) and 10 m Surface Wind (m/s) for Both Cyclones Chapala and Megh. Forecast length (hh) 00 12 24 36 48 60 72

MSLP(hPa)

10m Wind (m/s)

Number of verification

5.727 14.45 15.63 18.454 11.11 21.33 22.7

16.08 11.38 13.101 8.92 8.311 18.17 18.79

11 11 11 11 9 9 7

(b)

(a) 18N

18N

14N

14N

10N

10N

6N 45E

50E

55E

60E

65E

70E

75E 60

6N

(c)

45E

50E

55E

60E

65E

70E

75E 60

45E

50E

55E

60E

65E

70E

75E 60

(d)

18N

18N

14N

14N

10N

10N

6N 45E

50E

55E

60E

65E

70E

6N

75E 60

Figure 6.3  Wind field at 850 hPa for (a) Megh in the IC 07 case at 00 UTC, (b) Megh in IC 07 at landfall, (c) Chapala in the IC 29 case at 00 UTC, and (d) Chapala in IC 29 at landfall.

6.4.2. Track Prediction of Tropical Cyclones Figure  6.5 (a, b) depicts the tracks of Chapala and Megh from six and five ICs along with the IMD observations, respectively. From Figure 6.5a, we can see that case 1 follows movement along with the IMD observed track. Although in case 1, there is a big difference between the model and observed tracks due to the spin up time of the WRF model, after 24 hr the model is able to captured the  cyclonic event very well. Also, case 6 followed the observed track with a track error of less than 100 km. Similarly, in the case of ESCS Megh, it is seen from the Figure 6.5b that case 3 (IC 07) and case 5 more closely match the IMD observational track compared with other cases. At this point, we can say that as we minimize the forecast length, there is high probability of best track forecast by a numerical weather prediction model.

6.5. ­CONCLUSIONS This work is oriented toward the role of initial conditions in the simulation of intensity and track associated with tropical cyclones over the Arabian Sea by using the Weather Research and Forecasting model (WRF). We have considered two extremely severe cyclonic storms in the study and did the sensitivity experiments for different initial conditions. We have fixed the microphysics, planetary boundary layer, as well as cumulus schemes in the model throughout the study. We found that the ESCS Chapala track is well simulated in case 1 and in case 6, while in case 3 and case 5 it followed the observed track of ESCS Megh accurately. The WRF model reasonably well predicted the intensity of the TCs. The WRF model is found to be more skillful for track prediction of TCs when initialized at the severe cyclonic stage (SCS) rather

SIMULATION OF INTENSITY AND TRACK OF TROPICAL CYCLONES OVER ARABIAN SEA  83 20N

(a)

20N

18N

(b)

18N

16N

16N

14N

14N

12N

12N

10N

10N

8N

8N

6N

6N 45E

50E

55E

60E

65E

70E

75E

45E 980

965 970 975 980 985 990 995 1000 1005 1010 1015

20N

50E

(c)

20N

18N

18N

16N

16N

14N

14N

12N

12N

10N

10N

8N

8N

6N

985

55E 990

60E

65E

70E

75E

995 1000 1005 1010 1015 1020

(d)

6N 45E

50E

55E

60E

65E

70E

75E

45E

50E

55E

60E

65E

70E

75E

1000 1002 1004 1006 1008 1010 1012 1014 1016 1018 1020

1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018

Figure 6.4  Pressure level for (a) Megh in the IC 07 case at 00 UTC, (b) Megh in IC 07 at landfall, (c) Chapala in the IC 29 case at 00 UTC, and (d) Chapala in IC 29 at landfall. 24N

(a)

22N 20N 16N

Obs Case 1

18N

Case 2 Case 3

14N

Case 4 Case 5

12N

Case 6

10N 8N 45E 24N

48E

51E

54E

57E

60E

63E

66E

69E

72E

75E

78E

75E

78E

(b)

22N 20N 16N 18N

Obs Case 1

14N

Case 2 Case 3

12N

Case 4 Case 5

10N 8N 45E

48E

51E

54E

57E

60E

63E

66E

69E

72E

Figure 6.5  Track prediction of (a) Chapala for all six cases and (b) Megh for all five cases.

84  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

than at the cyclonic stage (CS) or lower. Therefore, the track errors are less with the advancement of model ICs in each TC case. The model is more able to predict the landfall location than the landfall time of the storms. ­ACKNOWLEDGMENTS The authors are thankful to the Department of Applied Mathematics, Gautam Buddha University, India, for providing the computing facility for the simulation of the cyclonic experiment. We thank NCEP NCAR for supplying FNL data as input in the WRF model. We appreciate the WRF model community for its support. ­REFERENCES Bhaskar Rao, D. V., & Hari Prasad, D. (2006). Numerical prediction of Odisha super‐tropical cyclone sensitivity to the parameterization of convection, boundary layer and explicit moisture processes. Mausam, 57 (1), 61–78. Davis, C., Wang, W., Chen, S. S., Chen, Y., Kristen, C., Mark, D., et al. (2008). Prediction of Landfalling Hurricanes with the advanced Hurricane WRF Model. Monthly Weather Review, 136, 1990–2005. Dudhia, J. (1989). Numerical study of convection observed during winter monsoon experiment using mesoscale two‐dimensional model. Journal of Atmospheric Science, 46, 3077–3107. Gray, W. M. (1968). Global view of the origin of tropical disturbance and storms. Monthly Weather Review, 96, 669–700. Gray, W. M. (1975). Tropical cyclone Genesis. Department of Atmos. Science. Paper No. 234, Colorado State University, Ft. Collins, CO. Hong, S.Y, Noh, Y., & Dudhia, J. (2006). A new vertical diffusion package with explicit treatment of entrainment processes. Monthly Weather Review, 134, 2318–2341. IMD (2015a). A report on extremely severe cyclonic storm “Chapala” over the Arabian Sea (28 October–4 November 2015). IMD (2015b). A report on extremely severe cyclonic storm “Megh” over the Arabian Sea (5–10 November 2015). Kanase, R., & Salvekar, P. S. (2014). Study of weak intensity cyclones over Bay of Bengal using WRF model. Atmospheric and Climate Sciences, 4, 534–548. Kain, J. S., & Fritsch, J. M. (1993). Convective parameterization for mesoscale models: The Kain–Fristch scheme. In K. A. Emanuel & D. J. Raymind (Eds.), The Representation of Cumulus Convection in Numerical Models (p. 246). American Metereological Society. Kumar, S., Routray, A., Tiwari, G., Chauhan, R., & Jain, I. (2016). Simulation of tropical cyclone Phailin using WRF modeling system. In M. Mohapatra, B. K. Bandyopadhyay,

& L. S. Rathore Tropical Cyclone Activity Over the North Indian Ocean, 291–300. Springer Lin, Y. L, Farley, R. D., & Orville, H. D. (1983). Bulk parameterization of the snow field in a cloud model. Journal of Climate and Applied Meteorology, 22, 1065–1092. Mendelsohn, R., Emanuel, K., Chonabayashi, S., & Bakkensen, L. (2012). The impact of climate change on global tropical cyclone damage. Nature Climate Change, 2, 205–209. Mlawer, E. J., Taubman, S. J, Brown, P. D, Iacona, M. J., & Clough, S. A. (1997). Radiative transfer for inhomogeneous atmosphere: RRTM, a validated correlated‐k model for the longwave. Journal of Geophysical Research, 102(D14), 16663–16682. Mohapatra, M., Bandyopadhyay, B. K., & Nayak, D. P. (2013). Evaluation of operational tropical cyclone intensity forecast over north Indian Ocean issued by India Meteorological Department. Natural Hazards, 68, 433–451. Mohapatra, M., Nayak, D. P., Sharma, M., Sharma, R. P., & Bandyopadhyay, B. K. (2015). Evaluation of official tropical cyclone landfall forecast issued by Indian Meteorological Department. Journal of Earth Science, 124, 861–874; doi: 10.1007/s12040‐015‐0581 Mohanty, U. C., Mandal, M., & Raman, S. (2004). Simulation of Orissa super‐cyclone (1999) using PSU/NCAR mesoscale model. Natural Hazards, 31, 373–390. Osuri, K. K., Mohanty, U. C., Routray, A., Kulkarni, M. A., & Mohapatra, M. (2012). Customization of WRF‐ARW model with physical parameterization schemes for the simulation of tropical cyclones over North Indian Ocean. Natural Hazards, 63, 1337–1359. Palmen, E. N. (1948). On the formation and structure of tropical hurricane. Geophysica, 3, 26–38. Pattanaik, D. R., & Rama Rao, Y. V. (2009). Track prediction of very severe cyclone “Nargis” using high resolution weather research forecasting (WRF) model. Journal of Earth Syst. Scence, 118, 309–329. Pielke, R. A. Jr. (2007). Future economic damage from tropical cyclones: sensitivities to societal and climate changes. Philosophical Transactions of the Royal Society A Mathematical, Physical and Engineering Sciences, 365(1860), 2717–2729. Raghavan, S., & Sen Sarma, A. K. (2000). Tropical cyclone impacts in India and neighborhood. In R. Pielke Jr., & R. Pielke Sr., Storms (pp. 339–356). London: Routledge. RSMC (2015). Report on cyclonic disturbances over north Indian ocean during 2015, No. ESSO/IMD/RSMC-Tropical Cyclones Report No. 01 (2016)/14. Srinivas, C. V., Yesubabu, V, Hariprasad, K. B. R. R., & Venkatraman, B. (2012). Real time prediction of a severe cyclone “Jal” over Bay of Bengal using a high Resolution mesoscale model WRF (ARW). Journal of Applied Meteorology and Climatology, 65, 331–357.

7 Development of a Soft Computing Model from the Reanalyzed Atmospheric Data to Detect Severe Weather Conditions Devajyoti Dutta1, Ashish Routray1, and Prashant K. Srivastava2

ABSTRACT In this chapter, we present an Artificial Neural Network (ANN)-based model to detect severe weather ­conditions by using the ERA‐I reanalyzed data set, in association with the measurements from the Tropical Rainfall Measuring Mission (TRMM) satellite over the northeastern part of India. Developing an ANN model is a ­nonparametric approach where the mapping of an input‐output data set is carried out without assuming a form of a functional relationship. For the present work, the following five environmental parameters are identified as inputs to the model: (1) Convective Available Potential Energy (CAPE), (2) Convective Inhibition (CIN), (3) vertical wind shear at lower heights (1–5 km), (4) vertical wind shear at higher heights (5–10 km), and (5) precipitable water at lower heights. The rain height (Echo Top Heights at 20 dBZ, ETH20dBZ) as a proxy for the intensity of the weather systems is considered from TRMM, Precipitation Radar (PR). An ANN is trained, by taking into account the five identified environmental parameters as input, and rain height parameter derived from the ETH20dBZ as output. The weight matrices are developed for this case. By using the feed forward network, the rain height is estimated for a given set of input data. The result is tested for a training and validation data set. For the training and validation data set, correlation coefficient for estimated and observed values is found to be around 0.73. On the basis of the estimated output, detected systems are classified in the following two categories: high‐intensity rain associated with lightning and hail (ETH20 dBZ, ≥ 18 km) and high‐intensity rain without lightning and hail (18 km > ETH20 dBZ ≥ 10 km). The probability of detection (POD) is found to be 85%. The results from the developed model are encouraging, and further work is in progress to fine tune the model. 7.1. ­INTRODUCTION

that the land gets saturated due to continuous heavy and steady rainfall for several hours to days. Heavy rain causes rapidly rising water along a stream or in a low‐lying area. A flood can occur anywhere when a large volume of rain falls within a short time. The rainfall intensity and duration can be considered as a key factor contributing to the flood. The heavy rainfall‐causing clouds fall under the categories of cumulus, cumulonimbus, and nimbostratus (Byers, 1959). The single cell of cumulonimbus takes on more importance when it serves as a building block of larger multicell thunderstorms (Houze, 1993). The thunderstorms often occur in large groups and complexes, in which individual thunderstorms and lines of thunderstorms are building blocks. These complexes are referred to as mesoscale convective systems (MCSs;

The tropical nature of climate in India is mainly responsible for many natural disasters. Among them are rain‐ related disasters, including droughts, floods, cyclones, and landslides. These pose a great threat to humans as well as to other living organisms. Flooding is one of the most costly natural disasters. The main conditions for flood are

National Centre for Medium Range Weather Forecasting (NCMRWF), Ministry of Earth Sciences, Uttar Pradesh, India 2  Institute of Environment and Sustainable Development and DST-Mahamana Center for Excellence in Climate Change Research, Banaras Hindu University, Uttar Pradesh, India 1 

Techniques for Disaster Risk Management and Mitigation, First Edition. Edited by Prashant K. Srivastava, Sudhir Kumar Singh, U. C. Mohanty, and Tad Murty. © 2020 John Wiley & Sons, Inc. Published 2020 by John Wiley & Sons, Inc. 85

86  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Houze, 1982). MCSs evolve over 3–6 hours and, at some stage, contain both the convective and stratiform precipitation regions (Houze, 1993). A MCS includes weather systems such as tropical cyclones, squall lines, lake‐effect snow events, polar lows, mesoscale convective complexes (MCCs), and monsoon depression (MD). It is shown that in the convective regimes, the hydrometeors are carried out up to higher heights with the help of vigorous updrafts. During the convective situations, updrafts up to 10 m/s and sometime even more are reported in the convective core (Balsley et al., 1988; Cifelli & Rutledge, 1994, 1998; May & Rajopadhya, 1999: Zipser & LeMone, 1980; Kumar et al., 2005). MCSs produce a large proportion of the Earth’s heavy rain and thus are important from a climatological point of view (Houze, 1993). These systems are an important link between atmospheric convection and large‐scale atmospheric circulation. Storm height is an important parameter to understanding the severity of the system. It is the height of the top of the rain column above the mean sea level. The strength of the convective system is measured by many parameters and one of them is the storm height. The vertical precipitation profiles reflect the dynamic and microphysical process in the clouds (Houze, 1981; Szoke et al., 1986; Hobbs, 1989; Liu & Takeda, 1989; Zipser & Lutz, 1994). By studying the maximum radar reflectivity profiles of convective cells in MCSs, Zipser and Lutz (1994) pointed out that the profile shape between land and oceanic cells shows significantly different in accordance to the different strength of the updrafts. The vertical profiles of reflectivity can be used to estimate the profiles of latent heating. Note that the latent heat release in deep convection accounts for the large fraction of the total diabetic heating budget (Riehl & Malkus, 1958), resulting in thermally forced dynamic response in the tropical atmosphere to convection (Hartmann et al., 1984; Mapes & Houze, 1995). For these reasons one of the primary goals of the Tropical Rainfall Measuring Mission (TRMM) and Global Precipitating Mission (GPM) is to characterize the four dimensional structure of latent heating in the tropics. Obtaining a direct measurements of latent heat release occurring in convection over the entire tropics is not possible, however remote sensing of cloudiness and precipitation structures over the tropics is possible using space‐borne measurements, such as from TRMM/GPM. The ground Doppler Weather Radar (DWR) has been used to detect severe storms for a long time. At present, India has 17 installed DWR. As DWR are sparsely located, alternative options are required to ascertain the severity of the storm. The reanalysis data of the atmosphere is available free of cost, and can be utilized to access the severity of the storm (Brooks et al., 2003). In this chapter, an attempt has been made to develop an Artificial Neural Network ANN-based technique to

estimate rain echo top height for 20 dBZ (ETH20dBZ) reflectivity thresholds with the help of environmental convective parameters, which can be used to detect the severity of the convective storms for disaster management. 7.2. ­METHODOLOGY This study was carried out using 2A25 and 1B11 of the TRMM data products during 1998 to 2014. The MCS are identified according to the criteria defined by Mohr and Zipser (1996). The atmospheric parameters from ERA‐I reanalyses are utilized in this study. 7.2.1. Observation from Precipitation Radar (PR) and Atmospheric Energetic Parameter 7.2.1.1. Rain Height Rain heights of MCSs are evaluated by using the reflectivity profiles as measured by Ku band radar of the TRMM in terms of echo top heights for ≥20 dBZ & ≥40dBZ. ETH(20 dBZ) is a measurement of the vertical extent of the precipitating system, which is a proxy for rain height, and ETH(40 dBZ) is the measure of the convective portion of the precipitating system. The following expression is utilized to estimate ETH(20 dBZ)



ETH20 dBZ

0.25*

i 80

Hi (7.1)

i 1

where Hi = 1, reflectivity ≥20 dBZ, Hi = 0, and reflectivity 100 km upper atmosphere

Thunder cloud

Lower atmosphere Fair weather return currents

100 km

Middle atmosphere Galactic cosmic rays (GCR)

Figure 8.1  Schematic of processes relevant to the GEC, extending from the Earth’s surface to the upper atmosphere, geospace and beyond. Thunderclouds and other electrified clouds serve as generators of current (light gray arrows) and maintain the Earth‐ionosphere leaky “capacitor” at a potential with respect to the ground. Currents return in the fair weather regions (gray arrows). Magnetic field lines can serve to electrically connect the Northern and Southern Hemispheres. Modification of electrical resistivity by ionizing solar energetic protons and galactic cosmic rays are two ways in which the solar system modulates the GEC; a third is through coupling with the magnetosphere. Magnetic fluxes due to auroral currents (dark gray arrows) produce “geomagnetically induced currents (GIC)” in the ground (from http://sisko.colorado.edu/FESD/) (Deierling et al., 2014).

transport schemes, aerosol modeling, cloud parameter­ ization, representation of terrestrial processes, ice sheet dynamics, oceanic dynamics, and so on (Steinhaeuser & Tsonis, 2013). Steinhaeuser and Tsonis (2013) com­ pared 23 climate models and commented that none of the models could predict the actual observations and, therefore, a  question arises on the ability of climate models to produce future projections in time and in space.

In this chapter, salient features of lightning discharges and the GEC and their impact on climate are briefly dis­ cussed. The details of lightning discharges are discussed in section 8.2; in addition, the impact of solar activity on convection and the role of cosmic rays, streamers, and leaders in lightning generation are also discussed. The distribution of lightning flashes on the land surface and ocean surface are also discussed. GEC and its relevance in the atmospheric discharges are discussed in section 8.3.

Lightning, the Global Electric Circuit, and Climate  95

Section  8.4 discusses the relevance of atmospheric dis­ charges in climate studies. As a conclusion, some unsolved problems are listed in section 8.5. 8.2. ­LIGHTNING DISCHARGES Lightning discharges are a manifestation of processes that generate, separate, and neutralize electrical charges in nature through thunderstorms. The charge generation pro­ cess includes both inductive and noninductive mechanisms (Saunders, 2008), whereas the charge separation process, which very sensitively depends on temperature and usually occurs between the altitude level where the temperature is about 0°C to the altitude level with a temperature of approximately ‐40°C, may be through the slow hydrody­ namic process (Saunders, 2008). The updraft in thunder­ storms drag small and light positively charged ice fragments, whereas heavy negatively charged hailstones predominantly fall downward due to gravity. Charge dis­ tribution in thunderstorms is very complex and may involve multilayers of charges (Lyons, 2006; Krehbiel et al., 2008; Siingh et al., 2011, 2012). The electrification process in thunderstorms depends on the existence of super cooled water, ice crystals, snow, hail, and soft hail (graupel), which may lie between the 0°C isotherm and the ‐40°C isotherm (Saunders, 2008) because ice may start melting above 0°C and all hydrome­ teors solidify below ‐40°C. This region can extend from 2 to 10  km altitude. Their presence in the mixed phase region is also governed by the large scale circulation of the atmosphere (Zipser & Lutz, 1994). The stronger updraft also enhances the collision between different par­ ticles, which may result in increased charge transfer bet­ ween particles, leading to rapid electrification. The pile up of charges of certain sign amounts to several tens of coulombs, resulting in a strong quasi‐electrostatic field inside and around the thunderstorm. Measurements of electric fields in thunderstorms and Monte Carlo calcu­ lations show that the field strength never reaches the value required for conventional breakdown of air, which is ~ 32 kV cm−3 at atmospheric pressure (Stolzenburg et al., 2007). Therefore, it was proposed that the thermal electron ava­ lanches (of mean thermal energy ~ several eV) produced by the the intense electric field of a thunderstorm may aid discharge phenomena. The breakdown mechanism operating at a lower break­ down field (~ 2.10 kV cm−1), called a runaway mecha­ nism, based on the relativistic electron avalanches, is also proposed (Gurevich et  al., 2009). A runaway electron ­produces a large amount of secondary low‐energy elec­ trons due to the neutral molecule ionization, which are accelerated to high energy in the presence of cloud electric fields. These energetic electrons in turn act as runaway electrons. As a  result, one may expect an exponentially

increasing a­valanche of runaway electrons (Colman et al., 2010), ultimately leading to the breakdown of air. The required threshold breakdown field for the runaway mechanism is one order of magnitude smaller than the conventional breakdown field. Electron acceleration during propagation of lightning streamers and stepped leaders may lead to breakdown and can be considered as an alternative to runaway mech­ anism (Chanrion & Neubert, 2010). Numerical simula­ tions showed that exponential growth of electric field in streamers can accelerate thermal electrons up to ~ 100 ke V (Celestin & Pasko, 2011), which may act as runaway electrons. Further, the electric field produced by stepped leaders can accelerate these energetic electrons up to the MeV energies. Atmospheric convection, an upward/downward move­ ment of air resulting from an imbalance between the vertical pressure gradient force and the gravitational force, may be due to the solar heating/energy input from other sources. The net force acting on the air parcel, called the buoyancy force, determines updraft velocity and hence nature and altitude location of thunderstorms. Thunderstorms with an updraft velocity 0.7) with surface tem­ perature and less correlated (CC ~ 0.19) with CAPE and OLR. Recently, Siingh et al. (2014) showed that the cor­ relation coefficient in different parts of India between lightning flashes and CAPE varied between 0.23 and 0.81. Good positive correlation with surface tempera­ ture is also reported. No correlation is found between lightning flashes and OLR. These results along with other studies (Weckwerth & Parsons, 2006) demonstrate that the development of convection is an interesting and complex problem, which is also controlled by orography of the region. The incident solar radiation affects heat input to the lower atmosphere, and the ultraviolet (UV) part of the radiation coupled with changes in the ozone concentration affects the heat budget in the stratosphere (Gray et al., 2005).

96  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Half of the temperature response in the tropical upper stratosphere is due to the solar irradiance change and half is due to the additional ozone feedback mechanism (Gray et  al., 2009). The heat input variations affect the east‐west tropical (Walker) circulation, which is inti­ mately connected with the north‐south tropical “Hadley” circulation (Meehl et  al., 2008). During the solar maximum period, Walker circulation is strengthened. The heating of the stratosphere leads to the “top‐down” influence (Shindell et  al., 2006) and may strengthen tropical convection with poleward shifted intertropical convergence zone (ITCZ) and south Pacific convergence zone (SPCZ) at solar maximum. Solar heating of the sea surface and dynamically coupled air–sea interaction lead to the “bottom‐up” mechanism, which also strengthens ITCZ and SPCZ circulation at solar maximum (Meehl et al., 2008). The response of the top‐down and the bot­ tom‐up mechanism is additive and produces an amplified sea surface temperature, precipitation, and cloud response in the tropical Pacific even for a relatively small solar forcing (Meehl et al., 2009). These circulations are cou­ pled with other circulations of the atmosphere (Gray et al., 2010). These large‐scale circulations play an active role in the development and intensification of convective processes, thereby changing distribution of lightning and precipitation. Earlier studies attempting to establish some relation between solar activity and thunderstorm activity showed both a positive correlation at some stations and negative one at other stations (Fritz, 1878; Brook, 1934; Myrbach, 1935; Sen, 1963; Kleymenova, 1967). Aniol (1952) ana­ lyzed mean thunderstorm frequency observed in South Germany in relation to sunspot numbers and reported negative correlation (R ~ – 0.55) for the years 1889–1993 and positive correlation (R ~ 0.74) for the period 1923– 1944. When the data of both periods were combined, the correlation was ~ – 0.02 (insignificant). Stringfellow (1974) reported positive correlation (R ~ 0.8) for 1930– 1973 in Britain, where earlier negative correlation was reported by Brooks (1934). Schlegel et al. (2001) analyzed data for the period 1992– 2000 over middle Europe and reported a significant corre­ lation between lightning frequency and Ap index and relative sunspot number on a yearly basis. They also reported significant negative correlation with cosmic ray fluxes. Girish and Eapen (2008) showed an inverse relation between sunspot cycle and thunderstorm/ lightning occur­ rence rates at Trivandrum, India, between 1853 and 2005. Siingh et al. (2013a) studied the effect of sunspot numbers, Ap index, cosmic rays, and F10.7 cm fluxes on the lightning activity in south Asia (8°–34° N, 60° –95° E) and Southeast Asia (8°–34° N, 95°–120° E) based on the data from 1998– 2010 and showed that the sunspot numbers/Ap index did not show any statistically significant results. However,

when F10.7 cm fluxes and cosmic ray fluxes were consid­ ered, lightning flashes showed negative correlation (corre­ lation coefficient was between ‐0.23 and ‐0.16). Neto et al. (2013) analyzed monthly thunder day data from seven cit­ ies in Brazil from 1951 to 2009 using wavelet analysis, and reported the 11-year periodicity in six cities with a pre­ dominant antiphase behavior with sunspot numbers. These results suggest that convective activity/thunder­ storm activity in relation to solar activity varies with the location of the observing station on the Earth and it may change its sign over the timescale of a few decades. The global variations of lightning activity from TRMM satellite data for the period 1995–2010 at a high grid res­ olution (0.5° × 0.5°) and low grid resolution (2.5° × 2.5°) were reported by Cecil et al. (2014). The flash distribution at high resolution is a bit noisy, maybe because of insuffi­ cient sampling rate for such a high resolution. The low resolution data over land (in the tropical zone) is much smoother and hence many details disappear. The global flash rate varies from ~ 35 flashes sec−1 in February (aus­ tral summer) to ~ 60 flashes sec−1 in August (boreal summer). The mean global flash rate is ~ 46 flashes sec−1. The peak monthly average flash rate (at 2.5° × 2.5° grid scale) is 18 flashes km−2 month−1 from early April to early May in the Brahmaputra valley of eastern India. The annual global lightning activity peaks in the summer hemisphere is in agreement with the seasonal migration of the ITCZ and the atmospheric circulation patterns (Cecil et al., 2014). Thunderstorms develop in subsidence and low‐level moisture conditions, and hence these are expected to develop in regions not far from the ITCZ. During the spring and fall, the distribution of lightning is fairly symmetric about the equator. Thunderstorm activity distribution is highly dependent on surface air temperature, and hence it is expected to be quite different during the El Nino phase (tropical land regions are warmed) as compared with the cool La Nina phase (Satori et al., 2009; Kulkarni & Siingh, 2016; Siingh et  al., 2017). Global thunderstorm/lightning activity with the El Nino southern oscillation (ENSO) perspective has been studied (Satori et al., 2009; Siingh et al., 2017). More lightning activity in the Pacific Ocean was observed in the cold La Nina phase as compared with the warm El Nino phase, with an opposite response in the coastal regions (Satori et  al., 2009; Siingh et  al., 2017). The moderate relative increase of lightning activity in the longitudinal range of Africa and Europe, more pronounced in the coastal regions of northwest Africa and eastern Mediterranean than equatorial Africa was reported ((Satori et al., 2009; Siingh et al., 2017)). The largest lightning response was observed in Southeast Asia. The greatest ENSO contrast occurred in the regions of s­ ynoptic scale subsidence and increased large‐scale subsidence and favored more lightning. At the interface of the Hadley

Lightning, the Global Electric Circuit, and Climate  97

cell and Ferrel cell, descending cool dry air creates tem­ perature inversion, which may act as an isolating lid on the planetary boundary layer, that helps in the buildup of wet bulb potential temperature and moist static energy in response to the shortwave radiation (Williams & Renno, 1993; Satori et al., 2009). This leads to CAPE enhance­ ment and hence initiation and strengthening of convection and thereby increased lightning activity. Thunderstorms derive their energy from CAPE, which is the upward integration of buoyancy force and depends on the vertical profile of the temperature difference bet­ ween a warmer rising air parcel within an updraft and a relatively cooler air far outside the updraft (Williams, 2005). Low buoyancy force leads to low altitude clouds and convection is referred to as shallow convection. Deep convection leads to towering clouds up to the tropopause leading to the possibility of enhanced lightning activity. However, CAPE is not the sole parameter to control light­ ning activity (Williams et al., 2002) as is evident from the fact that for the same value of CAPE over land surface and warm ocean water, maritime lightning activity is much reduced (Zipser, 1994). One of the reasons may be difference in conversion efficiency of CAPE to updraft kinetic energy. The large contrast of lightning distribution between the continent and ocean can be explained in terms of Bowen ratio (ratio of sensible heat to latent heat flux), which is high for the continent (Williams & Stanfill, 2002). Qie et al. (2003) analyzed lightning imaging sensor data over the Tibetan Plateau and reported that the Bowen ratio plays some role in lightning variation over seasons and plateau regions. They also showed in agreement with Williams et al. (1992) that lightning activity and monthly averaged CAPE are nonlinearly related. Flash per CAPE varied in different parts of the plateau between 6 and 18. In most parts of the plateau, the flash per CAPE is two to three times higher than that in Florida and Congo (Williams et al., 1992), although CAPE is low over the pla­ teau. Toumi and Qie (2004) proposed that the product of the Bowen ratio and CAPE could be a better measure of actual lightning on the plateau than CAPE or rainfall themselves. The involved sensible and latent heat fluxes play different roles. The latent heat flux is critical for rain­ fall amounts, but does not control deep convection. The sensible heat flux seems to modify the efficiency of light­ ning production for a given CAPE. Siingh et al. (2013a) showed a similar relation between lightning and convective rain over south Asia and Southeast Asia with correlation coefficient 0.68 and 0.81, respectively, and attributed it to the similar meteorolog­ ical factors having identical effects on lightning and pre­ cipitation and suggested that the convective processes in the two regions were similar. Liou and Kar (2010) found that the values of rain yield per lightning flash over Taiwan are different for inland and coastal stations. Also,

rain yields per flash are different for different seasons, which can be attributed to the cloud base height and CAPE. Larger cloud base height may lead to broader cloud with reduced entrainment so that more of the CAPE is effectively converted into vertical updraft and ice particle growth. Cloud base height is directly propor­ tional to the dew point depression of the surface and hence to the sensible heat flux. Siingh et al. (2014) showed that correlation coefficient between lightning flash and CAPE in different regions of India varied between 0.23 and 0.81, whereas the same between convective rain and CAPE lies between 0.68 and 0.86. However, lightning flashes are well correlated with surface temperature in all the considered regions, whereas the same is not true with the convective rain. These studies show that partly CAPE and surface temperature control the convective process and partly other factors such as sensible heat flux, total surface heat flux, orography of the region, thermody­ namic state of boundary layer, and so on. The required electric field for the conventional breakdown is much higher than the measured/simulated field in thunderstorms (Dwyer et al., 2006). As a result, an alternative runaway mechanism was proposed (Gurevich et  al., 1999), which required the presence of high‐energy charged particles such as cosmic rays and smaller electric fields (~ 2.16 kV cm−1) (Gurevich & Zybin, 2005). The high‐energy charged particles generate a considerable number of electron‐ion pairs producing an ionized domain, polarization of which in the presence of thundercloud electric field leads to a local enhancement of electric field at the edges of the domain and initiate discharge (Gurevich et al., 1999). The discharge is driven by low‐energy electrons, similar to self‐sustained labora­ tory discharges at atmospheric pressure initiated by sub­ nanosecond pulses of runaway electrons (Babich & Loiko, 2009). Based on simulation results, Babich et  al. (2012, p. 8.) reported, “it is very unlikely that the light­ ning discharge can be triggered by joint action of cosmic ray showers and relativistic runaway electron avalanches (RREAs) even in the presence of precipitation particles.” On the contrary, the RREAs seeded by low‐energy cosmic rays produce a plasma domain, at the edges of which the electric field of the cloud is enhanced above the breakdown field. In this mechanism, steady supply of seed electrons from the more common lower energy cosmic ray is required, instead of cosmic ray showers as proposed by Gurevich et  al. (1999). In support of their hypothesis, Gurevich et al. (2009) reported four events of lightning discharges while an extensive air shower (EAS) passed through a thunderstorm. However, it is not yet possible to identify the percentage of lightning flashes triggered by this process. Cosmic rays produce ion clusters in the lower atmosphere, which modify the vertical current and cause accumulation of space charges on the upper and

98  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

lower edges of the cloud (Tinsley, 2000; Nicoll, 2012). Space charges could influence microphysical processes such as droplet–droplet collision (Khain et  al., 2004), droplet–particle collision (Tripathi & Harrison, 2002), droplet formation (Harrison & Ambaum, 2008), and so on. In turn, these may influence cloud lifetime, cloud radiative properties, precipitation, and lightning activity. A number of studies have shown higher lightning activity over land surfaces than over ocean surfaces (Brooks, 1925; Orville & Henderson, 1986; Christian et  al., 2003; Kandalgaonkar et  al., 2005; Ranalkar & Chaudhari, 2009). This is attributed to the higher convection rate over land surfaces as compared with ocean surfaces for the same heat energy input. Satellite observations also reveal that the land–ocean contrast also exists in the diurnal variation. The diurnal variation of lightning activity may be linked to the diurnal varia­ tion of fair‐weather electric fields (Siingh et  al., 2007; Williams, 2009). Mach et  al. (2009, 2010) using aircraft observations of electrified clouds showed that land storms had greater flash rates than ocean storms but smaller mean conduction or Wilson currents. The peak flash rate occurs at different hours of the day for different continents, with the African continent having the largest peak flash rate at around 1500 hours UTC, but the diurnal peak in lightning activity occurs later because of the combined contributions from Africa, South America, and North America. The diurnal variations with local times for all the continents individually show peaks around 1600 hours local time (Bailey et  al., 2007). The land‐based storms have much larger mean flash rate (41 flashes per sec) and also larger variations between peaks and troughs than ocean lightning storms (mean 5 flashes per sec), which almost show a constant value. The global distribution of negative and positive cloud‐to‐ground (CG) discharges in the summer season (June to August, 2004) show that the negative CG discharges occur mainly in middle America, Africa, India, and Southeast Asia over both the land and oceanic regions (Sato et al., 2008). The occurrence over the oceanic region seems to be more dominant compared with that over the land region. For example, in middle America most of the negative CG dis­ charges occur over the north Pacific Ocean. The same feature can be seen in the northern and southern parts of the Indian Ocean. On the other hand, the global distribu­ tion of positive CG flashes in the same period shows that most of the positive CG discharges occur over the land region, especially in North America, equatorial and northwest parts of Africa, India, and Southeast Asia. The difference in lightning activity over land surface and ocean surface arises due to their differential response to solar radiation. The sensible heat flux over land sur­ faces is stronger than over ocean surfaces, and provides stronger updrafts required for more lightning (Williams

et  al., 2002, 2004). Further, larger concentrations of cloud condensation nuclei over land surfaces may cause more numerous and smaller cloud droplets, which sup­ press the coalescence process and provide more super­ cooled droplets in the mixed‐phase region where they participate in the charge generation processes (Williams & Stanfill, 2002). These mechanisms have been tested over islands of different areas (Williams et  al., 2004, Kumar & Kamra, 2010). Lightning discharges mostly occur in convective clouds, which have different features over land and oceans due to surface properties. Maritime have less supercooled water compared with the continental clouds (Black & Hallet, 1986). The updraft velocity is limited below the terminal fall velocity of the raindrops, whereas no such limit is observed in continental clouds (Zipser & Lutz, 1994). Contrary to this, Williams et  al. (2002) reported that over the Amazon basin (inhomoge­ neous terrain) the convective system resembles the real oceanic‐like tropical system. This suggests that surface properties are not the only cause of difference in convec­ tive processes over continents and oceans. Kumar and Kamra (2012) considered three sea regions (Arabian Sea, AS; Bay of Bengal, BB; and Chinese Sea, CS) and two land regions (peninsular India, PI, and Indo‐ China peninsula, ICP) and showed that the flash rates over the peninsular regions (PI and ICP) are 2.6–33 times those over the sea regions (AS, BB, CS). The flashes occur­ ring over oceans are more energetic than those occurring over peninsulas. Among the considered sea regions, flashes were found to be most frequent but least energetic in the BB and least frequent but most energetic in the AS region, which may be due to the warmer sea surface tem­ peratures with their average value remaining above the convective threshold in the BB. In another study, frequent lightning over the Gulf Stream in the Atlantic Ocean and the south Pacific Ocean near Australia were attributed to warm water and overlying cold air (Zipser, 1994). The family of high‐altitude lightning includes red sprites, blue starters, blue jets, gigantic jets, halos, and elves. Several groups across the world are working on them through ground‐based, airborne, and space‐based experiments and also theoretically. MacKenzie and Toynbee (1886) reported for the first time visual observa­ tion of a brief flash of light above a thunderstorm. After a gap of over 35 years, Wilson (1925) predicted the possi­ bility of electrical discharge in the mesosphere above an intense thunderstorm. Almost 70 years after the predic­ tion of Wilson, Vaughan & Vonnegut (1989) a­ nalyzed video recordings made by the space shuttle payload camera and showed many high‐altitude lightning dis­ charges called transient luminous events (TLEs) (Figure  8.2). The first ground‐based observation was reported by Franz et  al. (1990). Inspired by these early ­findings, two independent groups led by Waltor Lyons of

Lightning, the Global Electric Circuit, and Climate  99 EI. Density (cm–3)10–6

10–4

Temperature (°K)100

10–2

200

100

300

102

104

400

500

106 600

100 Elve Ionosphere

Altitude (km)

Temp.

50

Gigantic jet

Sprite

EI. Density (nighttime) Stratosphere

Blue jet EI. Density (daytime)

Cloud-to-ground lightning

0 0

100 Distance (km)

200

Figure 8.2  Transient luminous events (sprites, blue jets, gigantic jets, and elves) above the mesoscale convective system along with altitude are shown. The variation of daytime and nighttime electron density (El. density) and temperature are also shown (modified after Pasko, 2003).

Mission Research Corporation (Lyons, 1994) and Davis Sentman of the University of Alaska, Fairbanks (Sentman & Wescott, 1993) initiated field programs to study the new phenomenon. The interest in TLEs has subsequently grown worldwide and several hundred research papers and many review papers and books have been published. Starters, blue jets, and gigantic jets are not coincident with particular CG strokes (Suzuki et al., 2012). However, preceding CG discharges may create an electrical condition that promotes their formation (Krehbiel et al., 2008). Even IC discharges can create favorable conditions for their gen­ eration (Krehbiel et al., 2008). In some cases, the upward discharges were observed to begin as part of normal IC flashes (Liu et al., 2015). Sprites, halos, and elves are caused by intense CG discharges. Sprites and halos are the result of the excitation and ionization of air molecules due to collision with electrons accelerated by a quasi‐electrostatic field in the upper atmosphere established by CG discharges and their possible continuing current (Pasko et al., 1997; Li et al., 2009). Elves are caused by an electromagnetic pulse radiated from the return stroke current of the CG dis­ charges (Newsome & Inan, 2010). For more details on TLEs, see Siingh et al. (2015). The electric fields from charge imbalances in the upper half of a thundercloud stimulate the discharges, either from fields between charge regions of opposite polarity within the cloud (gigantic jets, GJ) or between a charge region and the screening layer at the cloud top (blue jets, BJ). In Figure 8.2, we can observe the different types of

jets. Singh et al. (2017), reported four GJs observed dur­ ing 2013 and 2014 for the first time over the Indian inland in the Indo‐Gangetic plain. The observational site is outside the city of Allahabad at 25.4°N, 81.9°E, a loca­ tion from which the cameras cover thunderstorm activity in the middle of the Indo‐Gangetic plain. The four GJs observed from our site are shown in Figure 8.3. The jets were observed during two thunderstorms with cloud altitude up to ~17 km with short life span. The two shown in the left-hand columns (GJ1 and GJ2) were recorded on 2 August 2013 at 18:23:12.02 UT and 18:36:48.94 UT, respectively, at a distance of ~430 km southwest of Allahabad. GJ1 appeared in only one image, apparently fully developed and extinguished within the 40 ms exposure time. GJ2 appeared fully developed in the second frame with the lower portion in the first frame. The two jets (GJ3 and GJ4) in the right-hand columns were captured on 7 September 2014 at 14:03:20.31 UT and 14:53:44.76 UT, respectively, at a distance of ~350 km south of Allahabad. The structure of GJ3 is remarkable, bending ~10 km toward the horizontal direction at ~50 km altitude before branching upward. Similarly, the blue jets were also observed during the thunderstorm over the eastern coast of India and Bay of Bengal. Chanrion et al. (2017), observed the pulsating blue jets over the east­ ern coast (Machlipatnam) and Bay of Bengal. Figure  8.4 shows eight consecutive images beginning with the cloud illuminated by lightning deeper inside the cloud. (a) The first pulse reaches a length of ~13.25 km

100  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Altitude (km)

80

60

40

20

Altitude (km)

80

60

40

20

–20

0 Width (km)

20 –20

0

20–20

Width (km)

0

Width (km)

20 –20

0

20

Width (km)

Figure 8.3  Gigantic jets records on 2 August 2013 (GJ1, GJ2) and 7 September 2014 (GJ3, GJ4) at a site (25.4°N, 81.9°E) near Allahabad, India. The grayscale is inverted, bright appearing dark. The altitude is estimated with an error of +/−1.2 km derived from an error estimate of GJ position of +/−5 km (Singh et al., 2017).

(a)

(b)

(c)

(d)

(f)

(g)

(h)

10 km (e)

Figure 8.4  The pulsating blue jet from the top of the northern cloud. Frame (a) is the first of the time sequence. It serves as a reference frame to illustrate the structure of the cloud. Frames (b)–(h) show the pulsating blue jet (Chanrion et al., 2017).

Lightning, the Global Electric Circuit, and Climate  101

(53 pixels), (b) it then fades, leaving faint emissions up to ~17.5 km (70 pixels). (c) Then a second pulse appears at the base of the original discharge, with faint emissions reaching even higher to 10.5–20 km (42–80 pixels). (d) These are disconnected from the activity at the root and may be the remains of the first pulse. (e) The discharge then fades, with blue emissions remaining at the top. (f) The discharge finally rebrightens a third time to 18.5 km. (g, h) It reaches its maximum extent of ~21.5 km (86 pixels) before fading completely. A lower limit of the discharge expansion velocity from frame 6 to be 5 × 105 m s–1 was estimated. These images are the first of pulsating dis­ charges at the tops of storm clouds. They show the aston­ ishing variety of forms that electrical activity can take as we continue to discover new varieties of discharges in and above thunderstorms. 8.3. ­LIGHTNING DISCHARGES AND THE GLOBAL ELECTRIC CIRCUIT The sources of the GEC in the lower atmosphere are t­hunderstorms, shower clouds, point discharge currents, ionization by radon isotopes, and galactic cosmic rays. The estimated contribution of thunderstorms and electri­ fied rain/shower clouds to the upward current varied as reported by different authors: 60% and 40%, respectively (Rycroft et  al., 2007); 80% and 20% (Odzimek et  al., 2010); and 90% and 10% (Mach et  al., 2011). Shower clouds generally transport negative charges to the ground on raindrops (Liu et al., 2010). Mareev et al. (2008), using numerical simulation, showed that 55–75% of the charges neutralized during lightning discharges are transferred to the ionosphere during typical CG flashes. In the case of IC flashes, the amount of charge transferred to the iono­ sphere is ~5–15%. Maggio et al. (2009) estimated charge transfer due to lightning transients to be ~35% to the ground during CG flashes, and during IC flashes upward charge transfer was ~12%. There are approximately 75% IC flashes and 25% CG flashes (Rakov & Uman, 2003). Mallios and Pasko (2012) discussed the charge flow to the ionosphere during all phases of thunderstorm evolution, and charge flow to the ground is only during CG dis­ charges and cloud dissipation phase. The discharges in the middle atmosphere, commonly called transient luminous events (TLEs), may also act as generators situated in the stratosphere and mesosphere, and affect the vertical conductivity distribution and charge transport (Rycroft & Odzimek, 2010). A single sprite may  lower the ionospheric potential by ~1V (Rycroft & Odzimek, 2009, 2010) and could have space charge of the order of mC (Li & Cummer, 2011). However, ELF radia­ tion produced by current flowing in the sprite is comparable to that radiated by the causative lightning discharges (Pasko  et  al., 1998; Li & Cummer, 2011; Rycroft &

Odzimek, 2010). Thus the contribution of a sprite as a DC generator is small but as an AC generator it is substantial. The current flow in the GEC may be affected by cosmic rays, which produce ionization in the lower atmosphere and modify conductivity and potential. Harrison and Usoskin (2010) demonstrated a positive relation between the ionospheric potential and neutron count rate at Climax, CO, based on observations from 1966 to 1972. The ionospheric potential was ~17% less at solar maximum than at solar minimum. Rycroft et  al. (2008) estimated ~6% less conductivity near solar maximum due to cosmic ray ion production. This may result in ~23% less fair‐weather current at solar maximum than at solar  minimum. Changes of the cloud cover at Vostok, Antarctica, were found to be associated with extreme increase of the vertical electric field there (Kniveton et  al., 2008). Harrison and Ambaum (2010) reported a median 10% decrease in cloud amount at the Lerwick observatory, Shetland Islands, Scotland, associated with 10% decrease in Climax neutron rate. Thus, monitoring charges in current density, vertical electric field, and potential by solar activity and associated cosmic ray changes may provide a potential mechanism to under­ stand the linkage between cosmic ray and cloud prop­ erties. The variation of sunspot number, Ap index, F10.7 solar flux, and cosmic rays, which define solar activity, is shown in Figure 8.5a for the period 1989–2010 (sunspot cycles 22 and 23) (Siingh et al. 2013a). The variation of lightning flash rates with sunspot num­ bers for the geographical regions of South Asia and Southeast Asia, R1 (8°N‐35°N, 60°E‐95°E) and R2 (8°N‐35°N, 95°E‐120°E) are shown in Figure 8.5b. Note that lightning flashes are available for the period 1998–2010. Lightning flashes show an almost an opposite trend to sun­ spot number variations. It was also noted that flash rates started increasing even when sunspot numbers remained almost maximum (during the period 2001–2003) and showed a secondary maxima during the period 2003–2006, when the sunspot number showed a decreasing trend. Increasing trend in lightning flashes suggested that strong and deeper convec­ tions were increasing over both the regions. Electrical properties of the GEC are also affected by the presence of aerosols of natural and anthropogenic origin (Markson, 2007; Siingh et al., 2013b). The presence of space charges in volcanic ash layers even after a month from eruption (Harrison et al., 2010), Saharan dust several kilo­ meters above the surface (Nicoll et  al., 2011), and dust devils (Farrell et al., 2004) show that the altitude distribu­ tion of aerosol particles significantly affects electrical parameters of the atmosphere. Zhou and Tinsley (2010) showed that aerosols can increase the global columnar resistance by as much as 60–90% and the largest effect comes from the continental boundary layers. The effect of clouds on the global columnar resistance is about 10%

102  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

2200 2100

240 Cosmic rays

200

2000

160

1900

120

1800 1700

80

Apindex

F10.7(solar flux data)

40 0

1600

Ap Index (nT) and F10.7, cm

Cosmic rays flux (Counts/minutes)

(a)

(b) 350 Region 1 150

300

Region 2 (Pereira felix et al., 2010)

250

125

200 Sunspot number

150

100

100 75

50

Sunspot number

Total lightning flashes (× 103)

175

0

50 1990 1992 1994 1996 1998 2000 Years

2002

2004

2006

2008 2010

Figure 8.5  (a) Yearly variations of cosmic ray flux, F10.7 cm (solar radio flux) and Ap index for the 22 and 23 solar cycles are shown. (b) Yearly variations of sunspot number and lightning flash counts for regions R1 and R2 are shown (Siingh et al., 2013a).

(Zhou & Tinsley, 2010), which is also indirectly corrobo­ rated with the measurements of vertical current and co-­located cloud cover with thin and thick overcast condi­ tions (Nicoll & Harrison, 2009a). Even the presence of tur­ bulence in the troposphere affects droplet charging by the vertical current (Tinsley, 2008). There are wide variations in the estimated contribution of different charging sources, which are based on isolated and short‐term measurements, although many of them are dynamically coupled. One needs long‐term simulta­ neous measurements of space charges, vertical current, conductivity, and electric fields at different latitudes, lon­ gitudes, and altitudes under quiet and disturbed solar conditions. Environment of measurement should be selected in such a way that the role of aerosols could be deciphered. This also requires much more effort to put on the related simulation studies. 8.4. ­ATMOSPHERIC DISCHARGES AND CLIMATE The general circulation of the atmosphere driven by the Hadley circulation between the equator and midlatitude

determines the location of subtropical and polar jets and strongly influences the climate (Price, 2006). Hadley circulation in combination with jets influences the location and quality of storms. The orography and distribution of land mass influence the distribution of lightning and thun­ derstorms (Christian et al., 2003; Siingh et al., 2013, 2014). Lightning activity is a sensitive indicator of surface tempera­ ture (Williams, 1992; Siingh et al., 2013a, 2014), upper tropo­ spheric water vapor (Price, 2000; Price & Asfur, 2006), cloud cover (Sato & Fukunishi, 2005), ice crystal size (Sherwood et  al., 2006), ice water content in thunderstorms (Petersen et al., 2005), CAPE and aerosol concentration (Siingh et al., 2014). These parameters also affect the state of the atmosphere and hence climate. The global distribution of atmospheric discharges (CG, IC, TLEs) driven by solar heating and also influenced by land‐ocean distribution on the planet follows the general circulation patterns of the atmosphere (Williams, 2005). Lightning discharges are dominated by land surface areas in the tropics (Williams & Stanfill, 2002; Christian et al., 2003; Siingh et  al., 2011) with Africa, South America, and Southeast Asia regions ranking from the most light­ ning and least rainfall region to least lightning and most

Lightning, the Global Electric Circuit, and Climate  103

rainfall region. These regions dominate the Walker circulation, whereas rainfall is ­zonally uniform in the upwelling portion of Hadley circulation. The global circulation is energized by the convective processes in the atmosphere. Lightning discharges need stronger and deeper convection, whereas rainfall requires moderate updraft with modest lifting. Both phenomena are associ­ ated with the microphysics and dynamics of thunder­ clouds, which in turn depend on the surface temperature, humidity, orography, and geographical location of the region (Williams, 2004, 2009; Carey & Buffalo, 2007; Price, 2009; Siingh et  al., 2011, 2013a, 2014). Small changes in surface temperature may result in larger change in thunderstorm and ­lightning activity (Williams, 2005, 2009) establishing a nonlinear link. The increasing greenhouse gases in the atmosphere may lead to a warming at local‐global level. The more (less) warming at the surface than the upper troposphere may lead to a more (less) unstable atmosphere (Price, 2013) and as a result one would expect more (less) convection and thunderstorms. There would not be any change if the surface and upper troposphere warm at the same rate, because there will not be any change in the sta­ bility of the atmosphere. In a warmer climate, CAPE increases (Del Genio et  al., 2007), which shows clear increase in lightning activity (Williams et al., 1992; Pawar et al., 2012; Siingh et al, 2013a, 2014). While studying the effect of enhanced CO2 (greenhouse gas), models show ~10% enhancements in lightning activity for every 1°C global warming (Grenfell et  al., 2003). At tropical stations, lightning activity may increase by ~25% for increase of surface temperature by 1 K above 296 K (Williams & Renno, 1993). Convection transports water vapor in the atmosphere, which leads to a greater ozone loss, decreases warming of the atmosphere, and enhances precipitation (Price & Asfur, 2006; Siingh et al., 2011). Climate models predict 10–20% enhancement in water vapor for every 1 K increase of temperature in that layer, although the Clausius–Clapeyron equation predicts ~6% enhancement for 1 K change (Price, 2000; Dotzek & Price, 2009). Contrary to this, during the severe drought period of 1997–1998 (decreased water vapor), the lightning activity in Indonesia increased by 60% (Hamid et  al., 2001; Yoshida et al., 2007). Results were explained by consid­ ering that in the El Nino dry period fewer thunderstorms develop, but they are much more explosive producing much more lightning activity (Price, 2009). Thus, the ENSO anomaly and its impact on lightning distribution affect general circulation and climate. The space charges formed at the top and bottom of clouds due to the flow of vertical current (Nicoll & Harrison, 2009a,b, 2010) attach to aerosol particles, cloud condensation nuclei (CCN), and ice forming nuclei

(Tinsley et al., 2001) and enhance electroscavenging and ice formation depending on the droplet size distribution or modify droplet evaporation (Harrison & Ambaum, 2008). The enhanced electroscavenging of larger CCN and aerosol particles may protect smaller CCN and aerosol particles from scavenging (Tinsley, 2004), and may lead to narrowing of the droplet size distribution. As a consequence of this, there may be both reduced precip­ itation and enhanced cloud lifetime. The two processes simultaneously compete and the net result depends on local temperature, aerosol environment, strength of vertical current, and cloud dynamics. The microphysical processes linking space charges at the cloud edge with the cloud properties are illustrated in Figure 8.6 (Rycroft et al., 2012). In electroscavenging, the collision efficiency of charged particles and liquid droplets is enhanced (Tinsley et  al., 2001) and formation of ice crystal is facilitated. The other impact of cloud charge is the formation of an additional droplet or an enhancement of droplet size by droplet–droplet coalescence (electroacti­ vation) (Harrison & Ambaum, 2008). Droplet–droplet collision efficiency is also enhanced by charging (Khain et al., 2004) leading to droplet growth by charged coales­ cence. These processes may occur simultaneously in a real cloud and therefore much more research activity is required by considering different droplet size distribution, concentration, and variation in the vertical current both under fair‐weather and disturbed conditions. Harrison and Ambaum (2008) reported reduction (by 0.3 W m−2) in longwave radiation underneath a stratocu­ mulus layer during a solar flare event and suggested it to be due to change in the height of cloud base caused by variation in cloud droplet charge as a result of variation in the vertical current. The vertical current also causes change in cloud cover/cloud properties (Harrison & Ambaum, 2010) as is evident from the observation of ~10% more broken cloud cover within one day of a Forbush decrease event (Harrision et al., 2008). The time delay of a few hours is in accordance with the fact that a vertical current effect on cloud microphysical process is expected to take a few hours (Zhou & Tinsley, 2007). 8.5. ­CONCLUSIONS AND RECOMMENDATION In this chapter, we have discussed some recent advances in lightning discharges, the GEC, and their relevance in ­climate studies. Further research activities in this field may include: 1. Convective process leading to lightning discharges and precipitation depends on surface temperature, CAPE, orography of place and its location, moisture, fea­ tures of boundary layer and wind profile, and so on. But details of contributions from each component are not well understood and quantified.

104  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION (a)

Charged collection +

Contact nucleation

Ice crystal

Electroscavenging + Electrofreezing

(b)

+

Activation

Droplet growth

Electroactivation

Droplet growth

Nucleation

Neutral aerosol

+

+ +

+ +

Charged aerosol

+

+ + + Charging

Nucleation

Electrocollection

(c)

Charged and uncharged droplets

–

+

Coalescence

Figure 8.6  Schematic diagram of various cloud microphysical processes, which may be affected by electrical charge: (a) electroscavenging, (b) electroactivation, and (c) electrocollection (after Rycroft et al., 2012).

2. Observations suggest that the spatial‐temporal dis­ tribution of lightning discharges is influenced by the Walker and Hadley circulations and El Nino/La Nina conditions of the atmosphere. Further work is required to include these effects in simulation studies. 3. Based on experimental‐theoretical studies, it has been proposed that cosmic rays may influence cloud life­ time, cloud radiative properties, precipitation, and light­ ning activity. Quantitative studies are not widely available and hence it may not be possible, with the current under­ standing, to identify the percentage of lightning triggered by cosmic rays. 4. The basic physics of cloud formation and involved thermodynamics is known but detailed cloud microphysics and the complex connection between climate and ecosystem are not fully understood, which is important and essential for a nonlinear system. Clouds being nonlinear systems (Tsonis, 2013) are very sensitive to the initial conditions and to changes in parameters. Therefore, further work is needed to improve our understanding of the dependency of climate system on clouds and the components. For example, the influence of space charge on cloud microphysics could be investigated using simultaneous high resolution space

charge measurements from the cloud top to the bottom, particle and droplet parameters, water and water vapor dis­ tribution, precipitation current, and so on. 5. The details of interaction of negative streamers with the ionization irregularities in the presence of an electric field and energization of electrons leading to runaway discharge mechanisms and its efficiency remains an unsolved problem. The role of background plasma and magnetic field also needs to be studied. 6. The characterization of electric field and vertical current is based on isolated measurements during short periods and therefore may not represent long‐term global character. Simultaneous measurements of space charge, vertical current, conductivity, and electric field under international collaboration (maybe in campaign mode) at different altitudes, latitudes, and longitudes under both quiet and disturbed conditions are required. 7. The charged particles from galactic, solar, and magne­ tospheric origin under different events and for varying amounts of water vapor and aerosols (of different types) influence global, regional, and local GEC parameters dif­ ferently. Their occurrence and influence varies significantly with latitudes, longitudes, and altitudes, and with phase in

Lightning, the Global Electric Circuit, and Climate  105

the solar cycle. Hence, systemic study involving a chain of stations and using multi‐instruments is required to develop and test physical models. 8. Lightning discharge activity acts as a thermometer for the surface temperature (major climate variable). Lightning and TLEs are linked to the production of NOx, which leads to the production of O3, a strong greenhouse gas. Also, lightning can ignite fire in temperate and boreal forests leading to emission of greenhouse gases and ­aerosols, and warming of the atmosphere. Further, thun­ derstorms transfer water vapor (a greenhouse gas) from the  boundary layer to the upper atmosphere. All these may act as a strong influence on climate. Quantitative assessment and its relation with lightning/thunderstorm characteristics are lacking. ­ACKNOWLEDGMENTS Indian Institute of Tropical Meteorology Pune is funded by Ministry of Earth Sciences (MoES), Government of India, New Delhi. ­REFERENCES Aniol, R. (1952). Schwankungen der Gewitterhaufigkeit in Suddeutschland. Meteorologische Rundschau, 3 (4), 55–56. Babich, L. P., & Loiko, T. V. (2009). Sub nano second pulses of runaway electrons generated in atmosphere by high‐voltage pulses of microsecond duration. Doklady Physics, 429, 479– 482; doi: 10.1134/S1028335809110019 Babich, L. P., Bochkov, E. I., Dwyer, J. R., & Kutsyk, I. M. (2012). Numerical simulations of local thundercloud field enhancements caused by runaway avalanches seeded by cosmic rays and their role in lightning initiation. Journal of Geophysical Research, 117, A09316; doi:10.1029/2012JA017799 Bailey, J. C., Blakeslee, R. J., Buechler, D. E., & Christian, H. J. (2007). Diurnal lightning distributions as observed by the Optical Transient Detector (OTD) and the Lightning Imaging Sensor (LIS). Proceedings of the Thirteenth International Conference on Atmospheric Electricity, Vol. II. Beijing, China, ICAE, 657–660. Black, R. A., & Hallett, J. (1986). Observations of the distribu­ tion of ice in hurricanes. Journal of Atmospheric Science, 43, 802–822; doi: http://dx.doi.org/10.1175/1520‐0469(1986)043 2.0.CO;2 Brooks, C. E. P. (1925). The distribution of thunderstorms over the globe. Geophysical Memo, 3(24): 147–164. Brooks, C. E. P. (1934). The variation of the annual frequency of thunderstorms in relation to sunspots. Quarterly Journal of the Royal Meteorological Society, 60, 153–165. Carey, L. D., & Buffalo, K. M. (2007). Environmental control of cloud‐to‐ground lightning polarity in severe storms. Monthly Weather Review, 135, 1327–1353. Cecil, D. J, Buechler, D. E., & Blakeslee, R. J. (2014). Gridded lightning climatology from TRMM‐LIS and OTD: Dataset description. Atmospheric Research, 135–136, 404–414; doi: 10.1016/j/atmosres.2012.06.028

Celestin, S., & Pasko, V. P. (2011). Energy and fluxes of thermal runaway electrons produced by exponential growth of streamers during the stepping of lightning leaders and in transient luminous events. Journal of Geophysical Research, 116, A03315; doi:10.1029/2010JA016260 Chanrion, O., & Neubert, T. (2010). Production of runaway elec­ trons by negative streamer discharges. Journal of Geophysical Research, 115, A00E32; doi:10.1029/2009JA014774 Chanrion, O., Neubert, T., Mogensen, A., Yair, Y., Stendel, M., Singh, R., & Siingh, D. (2017). Profuse activity of blue electrical discharges at the tops of thunderstorms. Geophysical Research Letters; doi: 10.1002/2016GL071311 Christian, H. J., Blakeslee, R. J., Boccippio, D. J., Boeck, W. J., Buechler, D. E., Driscoll, K. T, et al. (2003). Global frequency and distribution of lightning as observed from space by the optical transient detector. Journal of Geophysical Research, 108, 4005; doi: 10.1029/2002JD002347 Colman, J. J., Roussel‐Dupré, R., & Triplett, L. (2010). Temporally self‐similar electron distribution functions in atmospheric breakdown: The thermal runaway regime. Journal of Geophysical Research, 115, 1–17; doi:10.1029/2009JA014509 Deierling, W., & Petersen, W. A. (2008). Total lightning activity as an indicator of updraft characteristics. Journal of Geophysical Research, 113, D16210; doi:10.1029/2007JD009598 Deierling, W., Kalb, C., Mach, D., Liu, C., Peterson, M., & Blakeslee, R. J. (2014). On the variability of Wilson current by storm type and phase. XV International Conference on  Atmospheric Electricity, 15–22 June, Norman, OK USA. Del Genio, A. D., Mao‐Sung, Y., & Jonas, J. (2007). Will moist convection be stronger in a warmer climate? Geophysical Research Letters, 34, L16703; doi:10.1029/2007GL030525 Dotzek, N., & Price, C. (2009). Lightning characteristics in severe weather. In: H.D. Betz, U. Schumann, & P. Laroche (Eds.), Lightning principles instruments and applications (pp. 487–508). Berlin: Springer. Dwyer, J. R., Coleman, L. M., Lopez, R., Saleh, Z., Concha, D., Brown, M., & Rassoul, H. K. (2006). Runaway breakdown in the Jovian atmospheres. Geophysical Research Letters, 33, L22813; doi:10.1029/2006GL027633 Farrell, W. M., Smith, P.H., Delory, G. T., Hillard, G. B., Marshall, J. R., Catling, D., et  al. (2004). Electric and magnetic signatures of dust devils from the 2000–2001 MATADOR desert tests. Journal of Geophysical Research, 109, E03004. Franz, R.C., Nemzek, R.J., & Winckler, J.R. (1990). Television image of a large upward electrical discharge above a thunder­ storm. Science, 249, 48–51 Fritz, H. (1878). Die wichtigsten periodischen Erscheinungen der Meteorologie und Kosmologie. Natuurkundige Verhandelingen van de Hollandsche Maatschappij der Wetenschappen te Haarlem. Deel III, Haarlem. Girish, T. E., & Eapen, P. E. (2008). Geomagnetic and sunspot activity associations and ionospheric effects of lightning phenomena at Trivandrum near dip equator. Journal of ­ Atmospheric and Solar‐Terrestrial Physics, 70, 2222–2226. Gray, L. J., Beer, J., Geller, M., Haigh, J. D., Lockwood, M., Matthes, K., et al. (2010). Solar influences on climate. Reviews of Geophysics, 48, RG4001; doi:10.1029/2009RG000282

106  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Gray, L. J., Haigh, J. D., & Harrison, R. G. (2005). A review of the influence of solar changes on the Earth’s climate. Hadley Centre technical note 62, The UK Met Office. Gray, L. J., Rumbold, S., & Shine, K. P. (2009). Stratospheric temperature and radiative forcing response to 11‐year solar cycle changes in irradiance and ozone. Journal of Atmospheric Science, 66, 2402–2417. Grenfell, J.L., Shindell, D.T. & Grewe, V. (2003). Sensitivity studies of oxidative changes in the troposphere in 2100 using the GISS GCM. Atmosphere Chemistry Physics Discussion, 3, 1805–1842. Gurevich, A. V., & Zybin, K. P. (2005). Runaway breakdown and the mysteries of lightning. Physics Today, 58(5); doi:10.1063/1.1995746 Gurevich, A. V., Karashtin, A. N., Ryabov, V. A., Chubenko, A. P., & Shepetov, A. L. (2009). Nonlinear phenomena in iono­ sphere plasma. The influence of cosmic rays and the runaway electron breakdown on the thunderstorm discharges. Physics, Uspekhi, 179, 779 (in Russian). Gurevich, A. V., Zybin, K. P., & Roussel‐Dupre, R. A. (1999). Lightning initiation by simultaneous effect of runaway breakdown and cosmic ray showers. Physics Letters A, 254, 79–87; doi:10.1016/S0375‐9601(99)00091‐2 Hamid, E. Y., Kawasaki, Z., & Mardiana, R. (2001). Impact of the 1997–98 El Nino on lightning activity over Indonesia. Geophysical Research Letters, 28, 147–150. Hansen, J., Sato, M., Ruedy, R., Nazarenko, L., Lacis, A., Schmidt, G. A., et  al. (2005). Efficacy of climate forcing. Journal of  Geophysical Research, 110, D18104; doi:10.1029/2005JD005776 Harrison, R. G., & Ambaum, M. H. P. (2008). Enhancement of cloud formation by droplet charging. Proceedings of the Royal Society London A, 464, 2561–2573. Harrision, R.G., Aplin, K., & Rycroft, M.J. (2010). Atmospheric electric coupling between earthquake region and the ionosphere. Journal of Atmospheric Solar Terrestrial Physics, 12, 376–381. Harrison, R. G., & Ambaum, M. H. P. (2010). Observing Forbush decreases in cloud at Shetland. Journal of Atmospheric and Solar‐Terrestrial Physics, 72, 1408–1414. Harrison, R. G., & Usoskin, I. (2010). Solar modulation in sur­ face atmospheric electricity. Journal of Atmospheric and Solar‐Terrestrial Physics, 72, 176–182. Harrison, R. G., Aplin, K., & Rycroft, M. (2010). Atmospheric electricity coupling between earthquake regions and the ion­ osphere. Journal of Atmospheric and Solar‐Terrestrial Physics, 72 (5–6), 376–381. Kandalgaonkar, S. S., Tinmaker, M. I. R., Kulkarni, J. R., Nath, A. S., & Kulkarni, M. K. (2005). Spatio‐temporal var­ iability of lightning activity over the Indian region. Journal of Geophysical Research, 110; doi:10.1029/2004JD005631 Khain, A., Arkhipov, V., Pinsky, M., et al. (2004). Rain enhance­ ment and fog elimination by seeding with charged droplets. Part I: Theory and numerical simulations. Journal of Applied Meteorology, 43(10), 1513–1529. Kleymenova, E. P. (1967). On the variation of the thunderstorm activity in the solar cycle. Glav Upirav Gidromet Scuzb Met. Gidr., 8, 64–68. Kniveton, D. R., Tinsley, B. A., Burns, G. B., Bering, E. A., & Troshichev, O. A. (2008). Variations in global cloud cover and

the fair‐weather vertical electric field. Journal of Atmospheric and Solar‐Terrestrial Physics, 70,1633–1642. Krehbiel, P. R., Riousset, J. A., Pasko, V. P., Thomas, R. J., Rison, W., Stanley, M. A., et al. (2008). Upward electrical dis­ charges from thunderstorms. Nature Geoscience, 1(4), 233– 237. doi:10.1038/ngeo162 Kulkarni, M.N., & Siingh, D. (2016). The atmospheric electrical index for ENSO modoki: Is modoki one of the factors responsible for the warning trends slowdown? NatureScientific Reports, 6, 24009; doi:10.1038/serp2240091-10. Kumar, R. P., & Kamra, A. K. (2010). Lightning activity variations over three islands in a tropical monsoon region. Atmospheric Research, 98, 309–316. doi:10.1016/j.atmosres.2010.07.014. Kumar, R. P., & Kamra, A. K. (2012). The spatiotemporal vari­ ability of lightning activity in the Himalayan foothills. Journal of Geophysical Research, 117, D24201; doi:10.1029/2012JD018246 Li, C., Ebert, U., & Hundsdorfer, W. (2009). 3D hybrid compu­ tations for streamer discharges and production of runaway electrons. Journal of Physics D: Applied Physics, 42, 20200. Li, J., & Cummer, S. A. (2011). Estimation of electric charge in sprites from optical and radio observations. Journal of Geophysical Research, 116, A01301; doi:10.1029/2010JA015391 Liou, Y. A., & Kar, S. K. (2010). Study of cloud‐to‐ground lightning and precipitation and their seasonal and geographical characteristics over Taiwan. Atmospheric Research, 95, 115–122; doi:10.1016/j.atmosres.2009.08.016 Liu, C., Williams, E. R., Zipser, E. J., & Burns, G. (2010). Diurnal variations of global thunderstorms and electrified shower clouds and their contribution to the global electrical circuit. Journal of Atmospheric Science, 67, 309–323; doi:10.1175/2009JAS3248.1 Liu, N., et al. (2015). Upward electrical discharges observed above tropical depression Dorian. Nature Communications, 6, 5995. Lyons, W. A. (1994). Characteristics of luminous structures in the stratosphere above thunderstorms as imaged by low‐light video. Geophysical Research Letters, 21, 875. Lyons, W. A. (2006). The meteorology of transient luminous events: An introduction and overview. In M. Fullekrug et al., NATO Advanced Study Institute on Sprites, Elves and Intense Lightning Discharges (pp. 19–56). Dordrecht, The Netherlands: Springer. Mach, D. M., Blakeslee, R. J., & Bateman, M. G. (2011). Global electric circuit implications of combined aircraft storm electric current measurements and satellite‐based diurnal lightning statistics. Journal of Geophysical Research, 116, D05201; doi:10.1029/2010JD014462 Mach, D. M., Blakeslee, R. J., Bateman, M. G., & Bailey, J. C. (2009). Electric fields, conductivity, and estimated currents from aircraft overflights of electrified clouds. Journal of Geophysical Research, 114, D10204; doi:10.1029/2008JD011495 Mach, D. M., Blakeslee, R. J., Bateman, M. G., & Bailey, J. C. (2010). Comparisons of total currents based on storm loca­ tion, polarity, and flash rates derived from high altitude aircraft over flights. Journal of Geophysical Research, 115, D0320; doi:10.1029/2009JD012240 MacKenzie, & T., Toynbee, H. (1886). Meteorological phe­ nomena. Nature, 33, 26. Maggio, C. R., Marshall, T. C., & Stolzenburg, M. (2009). Estimations of charge transferred and energy released by

Lightning, the Global Electric Circuit, and Climate  107 lightning flashes. Journal of Geophysical Research, 114, D14203; doi:10.1029/2008JD011506 Mallios, S. A., & Pasko, V. P. (2012). Charge transfer to the ion­ osphere and to the ground during thunderstorms. Journal of Geophysical Research, 117, A08303; doi:10.1029/2011JA017061 Mareev, E. A., Yashunin, S. A., Davydenko, S. S., Marshall, T. C., Stolzenburg, M., & Maggio, C. R. (2008). On the role of transient currents in the global electric circuit. Geophysical Research Letters, 35, L15810; doi: 10.1029/2008GL034554 Markson, R. (2007). The global circuit intensity: Its measurement and variation over the last 50 years. Bulletin of the American Meteorological Society, 88, 223–241; doi: http:// dx.doi.org/10.1175/BAMS‐88‐2‐223 Meehl, G. A., Arblaster, J. M., Branstator, G., & van Loon, H. (2008). A coupled air‐sea response mechanism to solar ­forcing in the Pacific region. Journal of Climate, 21(12), 2883–2897. Meehl, G. A., Goddard, L., Murphy, J., Stouffer, R. J., Boer, G., Danabasoglu, G., et al., (2009). Decadal prediction. Bulletin of the American Meteorological Society, 90, 1467–1485; doi: http://dx.doi.org/10.1175/2009BAMS2778.1 Myrbach, O. (1935). Sonnenfleckenzyklus und Gewitterhäufigkeit in Wien Kremsmünster und Bayern. Met Ls, 52, 6–10. Neto, P. O., Pinto, I. R. C. A., & Pinto, O. Jr (2013). The rela­ tionship between thunderstorm and solar activity for Brazil from 1951 to 2009. Journal of Atmospheric and Solar‐ Terrestrial Physics, 98, 12–21; doi: 10.1016/j.jastp.2013.03.010 Newsome, R.T., & Inan, U.S. (2010). Free-running groundbased photometric array imaging of transient luminous events. Journal of Geophysical Research, 115, A00E41. http:// dx. doi.org/10.1029/2009JA014834. Nicoll, K. A. (2012). Measurements of atmospheric electricity aloft. Surveys in Geophysics, 33, 991–1057; doi 10.1007/ s10712‐012‐9188‐9 Nicoll, K. A., & Harrison, R. G. (2009a). Vertical current flow through extensive layer clouds. Journal of Atmospheric and Solar‐Terrestrial Physics, 71, 2040–2046. Nicoll, K. A., & Harrison, R. G. (2009b). A lightweight balloon‐ carried cloud charge sensor. Review of Scientific Instruments, 80, 014501. Nicoll, K. A., & Harrison, R. G. (2010). Experimental determi­ nation of layer cloud edge charging from cosmic ray ioniza­ tion. Geophysical Research Letters, 37, L13802; doi: org/10.1029/2010GL043605 Nicoll, K. A., Harrison, R. G., & Ulanowski, Z. (2011). Observations of Saharan dust layer electrification. Environmental Research Letters, 6, 014001. Odzimek, A., Lester, M., & Kubici, M. (2010). EGATEC: A new high‐resolution engineering model of the global atmo­ spheric electric circuit‐currents in the lower ionosphere. Journal of Geophysical. Research, 115, D18207, http://dx.doi. org/10.1029/JD013341. Orville, R. E., & Henderson, R. W. (1986). Global distribution of midnight lightning: September 1977 to August 1978. Monthly Weather Review, 114, 2640–2653; doi: org/10.1175/ 1520‐0493(1986)114 Pasko, V.P. (2003). Electrical jets. Nature, 423, 927–929. Pasko, V. P., Inan, U. S., & Bell, T. F. (1998). Spatial structures of sprites. Geophysical Research Letters, 25, 2123–2126.

Pasko, V. P., Inan, U. S., Bell, T. F., & Taranenko, Y. N. (1997). Sprites produced by quasi‐electrostatic heating and ioniza­ tion in the lower ionosphere. Journal of Geophysical Research., 102, 4529; doi: 10.1029/96JA03528. Pawar, S. D., Lal, D. M., & Murugavel, P. (2012). Lightning characteristics over central India during Indian summer monsoon. Atmospheric Research, 106, 44–49. Petersen, W. A., Christian, H. J., & Rutledge, S. A. (2005). TRMM observations of the global relationship between ice water content and lightning. Geophysical Research Letters, 32, L14819; doi:10.1029/2005GL023236 Price, C. (2000). Evidence for a link between global lightning activity and upper troposphere water vapour. Nature, 406, 290–293. Price, C. (2006). Global thunderstorm activity. In M. Fullekrug et  al. (Eds.), Sprites, elves and intense lightning discharges (pp.  85–99). NATO Sciences Series. Springer. Price, C. (2009). Will a drier climate result in more lightning? Atmospheric Research, 91, 479–484. Price, C. (2013). Lightning applications in weather and climate research. Surveys in Geophysics, 34, 755–767; doi: 10.1007/ s10712‐012‐9218‐7 Price, C., & Asfur, M. (2006). Can lightning observation be used as an indicator of upper tropospheric water vapour vari­ ability? Bulletin of the American Meteorological Society, 87, 291–298; doi: 10.1175/BAMS‐87‐3‐291 Price, C., & Federmesser, B. (2006). Lightning‐rainfall relation­ ships in Mediterranean winter thunderstorms. Geophysical Research Letters, 33, L07813; doi:10.1029/2005GL024794. Qie, X., Toumi, R., & Yuan, T. (2003). Lightning activities on the Tibetan Plateau as observed by the lightning imaging sen­ sor. Journal of Geophysical Research, 108, 4551; doi:10.1029/2002JD003304 Rakov, V. A., & Uman, M. A. (2003). Lightning: Physics and effects. Cambridge University Press. Ranalkar, M., & Chaudhari, H. (2009). Seasonal variation of light­ ning activity over the Indian subcontinent. Meteorology and Atmospheric Physics, 104, 125–134; doi:10.1007/s00703‐009‐00 Rohrer, F., & Berresheim, H. (2006). Strong correlation bet­ ween levels of tropospheric hydroxyl radicals and solar ultra­ violet radiation. Nature, 442, 184–187; doi:10.1038/ nature04924 Rycroft, M.J., Harrison, R.G., Nicoll, K.A., & Mareev, E.A., (2008). An overview of Earth’s global electric circuit and atmospheric conductivity. Space Science Reviews, 137, 83– 105. http://dx.doi.org/10.1007/11214-008-9368-6 Rycroft, M. J., Nicoll, K. A., Aplin, K. L., & Harrison, R. G., (2012). Recent advances in global electric circuit coupling bet­ ween the space environment and the troposphere. Journal of Atmospheric and Solar‐Terrestrial Physics, 90–91, 199–211. Rycroft, M. J., & Odzimek, A. (2009). The impact of lightning flashes and sprites on the Earth’s global electric circuit: An overview of recent modeling results. In N. B. Crosby, T.‐Y. Huang, & M. J. Rycroft (Eds.), Coupling of thunderstorms and lightning discharges to near‐earth space (pp. 124–135). American Institute of Physics Conference Proceedings, CP 1118. Rycroft, M. J., & Odzimek, A. (2010). Effects of lightning and sprites on the ionospheric potential, and threshold effects on sprite initiation, obtained using an analog model of the global

108  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION atmospheric electric circuit. Journal of Geophysical Research, 115, A00E37, http://dx.doi.org/10.1029/2009JA014758. Rycroft, M. J., Odzimek, A., Arnold, N. F., Fullekrug, M., Kulak, A., & Neubert, T. (2007). New model simulations of the global atmospheric electric circuit driven by thunder­ storms and electrified shower clouds: The role of lightning and sprites. Journal of Atmospheric Solar Terrestrial Physics, 69, 445–456. http://dx.doi.org/10.1016/j. jastp.2007.09.004. Sato, M., & Fukunishi, H. (2005). New evidence for a link bet­ ween lightning activity and tropical upper cloud coverage. Geophysical Research Letters, 32, L12807; doi: 10.1029/2005GL022865 Sato, M., Takahashi, Y., Yoshida, A., & Adachi, T. (2008). Global distribution of intense lightning discharges and their seasonal variations. Journal of Physics D: Applied Physics, 41, 234011; doi:10.1088/0022‐3727/41/23/234011 Satori, G., Williams, E., & Lemperger, I. (2009). Variability of global lightning activity on the ENSO time scale. Atmospheric Research, 91, 500–507. Saunders, C. (2008). Charge separation mechanisms in clouds. Space, Science Reviews, 137, 335–353; doi:10.1007/ s11214‐008‐9345‐0 Schlegel, K., Diendorfer, G., Thern, S., & Schmidt, M. (2001). Thunderstorms, lightning and solar activity: Middle Europe. Journal of Atmospheric and Solar‐Terrestrial Physics, 63, 1705–1713. Schumann, U., & Huntrieser, H. (2007). The global lightning‐ induced nitrogen oxides source. Atmospheric Chemistry and Physics, 7, 3823–3907. Sen, A. K. (1963). Integrated field intensity of atmospherics in relation to local thunderstorms. Journal of Atmospheric and Solar‐Terrestrial Physics, 25, 306–308. Sentman, D.D., & Wescott, E.M. (1993). Observations of upper atmospheric optical flashes recorded from an aircraft. Geophysical Research Letters, 20, 2857–2860. Sherwood, S. C., Phillips, V. T. J., & Wettlaufer, J. S. (2006). Small ice crystals and the climatology of lightning. Geophysical Research Letters, 33, L05804; doi:10.1029/2005GL025242. Shindell, D. T., Faluvegi, G., Unger, N., Aguilar, E., Schmidt, G. A., Koch, D. M., et al. (2006). Simulations of preindus­ trial, present‐day, and 2100 conditions in the NASA GISS composition and climate model G‐PUCCINI. Atmospheric Chemistry and Physics, 6, 4427–4459. Siingh, D., Buchunde, P. S., Gandhi, H., Singh, R., Singh, S., Patil, M. N., & Singh, R. P. (2015). Lightning and convective rain over Indian peninsula and Indo‐China peninsula. Advances in Space Research, 55, 1085–1103; doi: 10.1016/j. asr.2014.11.014. Siingh, D., Buchunde, P. S., Singh, R. P., Nath, A., Kumar, S., & Ghodpage, R. N. (2014). Lightning and convective rain study at different parts of India. Atmospheric Research, 137, 35–48; doi.10.1016/j.atmosres.2013.09.018. Siingh, D., Dharmaraj, T., Kumar, P. S., Singh, R., Kumar, S., Chinthalu, G. R., et  al. (2017). Variability of lightning, convective rain and solar activity study over South/ Southeast Asia during ENSO episode for the period of 1998‐2010. Journal of Indian Geophysical Union, 21 (5), 401–414. Siingh, D., Gopalakrishnan, V., Singh, R. P., Kamra, A. K., Singh, S., Pant, V., et  al. (2007). The atmospheric global

electric circuit: an overview. Atmospheric Research, 84, 91–110; doi:10.1016/j.atmosres.2006.05.005 Siingh, D., Kumar, P. R., Kulkarni, M. N., Singh, R. P., & Singh, A. K. (2013a). Lightning, convective rain and solar activity ‐ over the South/Southeast Asia. Atmospheric Research, 120– 121, 99–111; doi.org.10.1016/j.atmosres.2012.07.026 Siingh, D., Singh, R. P., Gopalakrishnan, V., Selvaraj, C., & Panneerselvam, C. (2013b). Fair‐weather atmospheric electricity study at Maitri (Antarctica). Earth, Planets, and Space, 65, 1541–1553. Siingh, D., Singh, R. P., Singh, A. K., Kulkarni, M. N., Gautam, A. S., & Singh, A. K. (2011). Solar activity, lightning and cli­ mate. Surveys in Geophysics, 32, 659–703; doi: 10.1007/ s10712‐011‐9127‐1 Siingh, D., Singh, R. P., Singh, A. K., Kumar, S., Kulkarni, M. N., & Singh, A. K. (2012). Discharges in the stratosphere and mesosphere. Space Science Reviews, 169, 73–121; doi 10.1007/ s11214‐012‐9906‐0 Singh, R., Maurya, A. K., Chanrion, O., Neubert, T., Cummer, S. A., Mlynarczyk, J., et al. (2017). Assessment of Unusual Gigantic Jets observed during the Monsoon season: First observations from Indian subcontinent. Scientific Reports, 7, 16436; doi:10.1038/s41598‐017‐16696‐5 Steinhaeuser, K., & Tsonis, A. A. (2013). A climate model inter­ comparison at the dynamics level. Climate Dynamics; doi10.1007/s00382‐013‐1761‐5 Stolzenburg, M., Marshall, T. C., Rust, W. D., Bruning, E., MacGorman, D. R., & Hamlin, T. (2007). Electric field values observed near lightning flash initiations. Geophysical Research Letters, 34, L04804; doi:10.1029/2006GL028777. Stringfellow, M. F. (1974). Lightning incidence in Britain and the solar cycle. Nature, 249, 332–333. Suzuki, T., Hayakawa, M., Hobara, Y., & Kusunoki, K. (2012). First detection of summer blue jets and starters over Northern Kanto area of Japan: Lightning activity. Journal of Geophysical Research, 117, A07307. http://dx.doi.org/10.1029/2011 JA017366 Tinsley, B. A. (2000). Influence of solar wind on the global electric circuit, and inferred effects on cloud microphysics, temperature and dynamics in the troposphere. Space Science Review, 94, 231–258. Tinsley, B. A. (2004). Scavenging of condensation nuclei in clouds: dependence of sign ofelectro‐scavenging effect on droplet and CCN sizes. Proceedings of the International Conference on Clouds and Precipitation, 248. IAMAS, Bologna, 18–23. Tinsley, B. A. (2008). The global atmospheric electric circuit and its effects on cloud microphysics. Reports on Progress in Physics, 71, 066801; doi:10.1088/0034‐4885/71/6/06 Tinsley, B. A., Rohrbaugh, R. P., & Hei, M. (2001). Electro‐ scavenging in clouds withbroad droplets size distributions and weak electrification. Atmospheric Research, 115, 59–60. Toumi, R., & Qie, X. (2004). Seasonal variation of lightning on the Tibetan Plateau: A spring anomaly. Geophysical Research Letters, 31, L04115; doi:10.1029/2003GL0189 Tripathi, S. N., & Harrison, R. G (2002). Enhancement of contact nucleation by scavenging of charged aerosol particles. Atmospheric Research, 62, 57–70. Tsonis, A. A. (2013). Geoengineering carries unknown conse­ quences. Physics Today, 66, 8–9.

Lightning, the Global Electric Circuit, and Climate  109 Vaughan, O. H. Jr., & Vonnegut, B. (1989). Recent observations of lightning discharges from the top of a thundercloud into the air above. Journal of Geophysical Research, 95, 13179–13182. Weckwerth, T. M., & Parsons, D. B. (2006). A review of convection initiation and motivation for IHOP‐2002. Monthly Weather Review, 134, 5–22; doi: 10.117 5/MWR3067.1 Williams, E. R. (1985). Large‐scale charge separation in thun­ derclouds. Journal of Geophysical Research, 90, 6013–6025. Williams, E. R. (1992). The Schumann resonance: A global thermometer. Science, 256, 1184–1187; doi:10.1126/ science.256.5060.1184. Williams, E. R. (2005). Lightning and climate: A review. Atmospheric Research, 76, 272–287; doi:10.1016/j. atmosres.2004.11.014 Williams, E. R. (2009). The global electrical circuit: A review. Atmospheric Research, 91, 140–152. Williams, E. R., & Renno, N. (1993). An analysis of the conditional instability of the tropical atmosphere. Monthly Weather Review, 121, 21–36. Williams, E. R., & Stanfill, S. (2002). The physical origin of the land‐ocean contrast in lightning activity. Comptes Rendus Physique, 3, 1277–1292; doi:10.1016/S1631‐0705(02)01407‐X Williams, E. R., Chan, T., & Boccippio, D. (2004). Islands as miniature continents: Another look at the land‐ocean light­ ning contrast. Journal of Geophysical Research, 109, D16206; doi:10.1029/2003JD003833 Williams, E. R., Rosenfeld, D., Madden, N., Gerlach, J., Gears, N., Atkinson, L., et  al., (2002). Contrasting convective regions over the Amazon: implications for cloud electrifica­

tion. Journal of Geophysical Research, 107(20), 8082; doi:101029/2001JD000380 Williams, E. R., Geotis, S. G., Renno, N., Rutledge, S. A., Rasmussen, E., & Rickenback, T. (1992). A radar and electrical study of tropical “hot towers.” Journal of Atmospheric Science, 49, 1386–1395. Wilson, C. T. R. (1925). The electric field of a thunderstorm and some of its effects. Proceedings of the Royal Society London, 37, 32D. Wyatt, M. G., Kravtsov, S., & Tsonis, A. A. (2012). Atlantic multi decadal oscillation and northern hemisphere’s climate variability. Climate Dynamics, 38, 929–949; doi: 10.1007/ s00382‐011‐1071‐8 Yoshida, S., Morimoto, T., Ushio, T., & Kawasaki, Z. (2007). ENSO and convective activities in Southeast Asia and west­ ern Pacific. Geophysical Research Letters, 34, L21806; doi:10.1029/2007GL030758 Zipser, E. J. (1994). Deep cumulonimbus cloud system in the tropics with and without lightning. Monthly Weather Review, 122, 1837–1851; doi:10.1175/1520‐0493(1994)122\1837:DCC SIT2.0.CO;2 Zipser, E. J., & Lutz, K. R. (1994). The vertical profile of radar reflectivity of convective cells: A strong indicator of storm intensity and lightning probability? Monthly Weather Review, 122(8), 1751–1759. Zhou, L. & Tinsley, B.A. (2007). Production of space charge at the boundaries of layer clouds. Journal of Geophysical Research, 112, D11203. http://dx.doi.org/10.1029/2006JD007998. Zhou, L., & Tinsley, B. A. (2010). Global circuit model with clouds. Journal of Atmospheric Science, 67, 1143–1156.

9 An Exploration of the Panther Mountain Crater Impact Using Spatial Data and GIS Spatial Correlation Analysis Techniques Sawyer Reid Stippa1, Konstantinos P. Ferentinos2, and George P. Petropoulos3

ABSTRACT Identification and mapping of hypervelocity impact crater (HIC) sites require significant effort on ground truthing data collection and local instrument‐driven research. The recent advancements in Earth observation (EO) technology and geographical information systems (GIS) have increased our ability to study HICs. With EO imagery and relevant spatial data now readily available online at no cost, GIS and remote sensing provide a very attractive option in investigating the Earth’s surface. In this framework, our study addresses the use of GIS and EO techniques by looking at a possible impact crater in upstate New York, United States. The Panther Mountain crater is thought to have been created by a meteor impact over 300,000 years ago during the Devonian or Mississippian geologic periods. Using freely available data from previous research, this study aimed at mapping land cover and geologic data and analyzing their correlation at Panther Mountain and it surrounding area. Findings of the study have showed encouraging results. A correlation between Panther Mountain’s bedrock geology and vegetation was reported to be higher than the coefficient of the surrounding area. Similarly, the correlation between Panther Mountain’s surficial geology type and vegetation was significantly lower than that of the other region. The significant difference in correlations between the two regions supports the Panther Mountain impact site. All in all, the present study also produced encouraging results as regards to the use of GIS in identifying potential hypervelocity crater sites. 9.1. ­INTRODUCTION The use of geographic information systems (GIS) to study the Earth’s surface has had significant impact on our knowledge of the planet’s past and current geomorphic

 Department of Geography and Earth Sciences, University of Aberystwyth, Wales, United Kingdom 2 Hellenic Agricultural Organization “Demeter”, Soil & Water Resources Institute, Department of Agricultural Engineering, Athens, Greece 3  School of Mineral & Resources Engineering, Technical University of Crete, Kounoupidiana Campus, Crete, Greece; Department of Soil & Water Resources, Institute of Industrial & Forage Crops, Hellenic Agricultural Organization “Demeter” (former NAGREF), Directorate General of Agricultural Research, Larisa, Greece 1

history. In much the same way, our improvements upon GIS techniques have allowed for further study of our planetary neighbors, both locally and galactically. However, much information is yet to be discovered about our own planet. A surface as actively changing (geophysically, geochemically, etc.) as the Earth’s, makes studying it all the more challenging. Much like studies done on the Martian surface to explain its surficial history, similar geospatial techniques can be applied to study the Earth’s history. GIS and Earth observation (EO) play a substantial role for the identification of possible terrestrial impact structures, for mapping target‐rock lithologies and deciphering the structural style of known craters (Zumsprekel & Bischoff, 2005). Terrestrial impact craters are known to exist in many places on our planet’s surface. Many of them have  been located and studied using aerial and satellite

Techniques for Disaster Risk Management and Mitigation, First Edition. Edited by Prashant K. Srivastava, Sudhir Kumar Singh, U. C. Mohanty, and Tad Murty. © 2020 John Wiley & Sons, Inc. Published 2020 by John Wiley & Sons, Inc. 111

112  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

imagery. The Earth Impact Database, run by the Planetary and Space Science Centre (PASSC), maintains records of known impact structures all over the world. It has been speculated that Panther Mountain crater in upstate New York is a buried terrestrial impact site, however, it has not yet been officially acknowledged as such by the PASSC. It does, however, belong to the list of suspected Earth impact sites (SEIS) as a possible terrestrial crater (Rajmon, 2007). Panther Mountain is part of the greater Catskill Park and is currently maintained and protected by New York State’s Forest Reserve. The rosette drainage pattern around the mountain is unlike the dendritic shape the rest of the range exhibits. Geologist George Chadwick believed this shape to be caused by rising gas under the surface (Guterl, 2000). In the 1940s the Dome Gas Company drilled 1.8 km into the mountain only to find minor amounts of natural gas. Not until the 1970s was greater effort put into studying the unknown source of the mountain’s shape. The late Dr. Yngvar Isachsen, a geologist who was part of the New York Geological Survey, believed the circular structure was caused by a buried impact crater. He did several studies of the area where he examined the compact fracture pattern of the sedimentary rock. Isachsen was also able to show a gravitational pull at Panther Mountain different from the other surrounding peaks. However, he was unable to gain support for the discovery of iron spherules, which i­ndicate a high possibility of meteor impact. In 1999, a group of Canadian specialists were able to confirm the existence of shock lamellae in tiny quartz crystals, which could only have resulted from a large impact (Guterl, 2000). Isachsen’s research continued until his death in 2001. Unfortunately, there has been no further research done on Panther Mountain since. The existing methods of identifying terrestrial impact craters rely on different criteria than those used in identifying craters on the Moon or Mars, where surface structures are more capable of withstanding changing environmental conditions. To determine the existence of terrestrial craters, the identification of microscopic and macroscopic features, such as shatter cones, is considered more evidential than simply viewing circular structures. This remains the case because of the complexity of the Earth’s surface, where surficial features may be caused by other geomorphic processes. The identification methods can be tedious, however, and often require funding and resources (such as drilling equipment and gravity measuring devices). With the onset of more affordable and detailed remote sensing imagery, the possibility of extracting information from potential terrestrial impact sites (and the Earth’s surface as a whole) can become more accessible to the general public and lead to a greater understanding of our planet’s geomorphic history.

Since the Panther Mountain crater site was first detected in the 1940s, there has been a minimal amount of scientific evidence suggesting its location as an impact site. To the authors’ knowledge, no attempts have been made to study the area with available GIS data sets. Consequently, there is a lack of definitive remotely sensed data providing evidence for a meteorite impact. Significant amounts of research involved locating sources of detailed spatial information, such as surficial and bedrock mineralogy maps. Land cover and geologic mapping have been a proven success in the GIS community. Numerous studies have used different GIS software packages in analyzing vegetation and mineralogy. More specifically, there have been many studies examining the relationship between soil type and vegetation. A high correlation was found to exist between forest types and soil types in a study published by Gerdol et al. (1985). Another study performed on a mixed‐ oak forest found correlations to exist between vegetation diversity and soil physicochemical properties and soil enzyme activities (Rodríguez‐Loinaz et al., 2008). A study by Chen et  al. (2007) in northwestern China found a positive correlation between soil organic matter and total vegetation cover. It was found that “biodiversity was closely related to soil nutrients” in Mu Su Sandland, China (Yang et al., 2007) and in Huanjiang County, China, there was a positive correlation between soil bacteria and vegetation (He et  al., 2008). Another study conducted in southwest China showed a ­correlation between soil texture and water content with vegetation (Xu et al., 2008). There have also been many studies examining the relationship between bedrock and vegetation, either directly or through soil interaction. One study examined the relationship between vegetation diversity and soil as a result of the geologic history of the area (Pastor et al., 1982). Puy and Moat (1996) used GIS analysis to study correlation between vegetation and bedrock and found that the primary vegetation is very strongly influenced by the type of rock on which it occurs. Williard et  al. (2005), in a study concerning forest type in the Appalachian Mountains, found that the white ash occurrence, sugar maple and eastern hemlock were related closely to type of bedrock geology type. Another report ­examined the relationship between sedimentary bedrock and with certain mineral deposits and found that soil element concentration varied with bedrock chemistry (Neff et al., 2006). Several studies addressed bedrock fissures that could have an impact on vegetation. Aich and Gross (2008) indicated that the bedrock‐fracturing had a strong influence on vegetation growth, specifically where bedrock consists of the majority of the land cover and where the top soil layer is thin. A similar study also indicated that bedrock fracturing provided positive linear growth for vegetation (Stothoff et al., 1999). A study done at the

AN EXPLORATION OF THE PANTHER MOUNTAIN CRATER IMPACT USING SPATIAL DATA AND GIS SPATIAL  113

University of Massachusetts, Amherst, by Searcy et  al. (2003) revealed significant differences in vegetation distribution with respect to bedrock. Several of the previous studies described bedrock as having a significant relationship with surface vegetation. With complex‐crater formation having a significant impact on underlying bedrock, it is possible that the ­bedrock beneath Panther Mountain affects vegetation differently than the bedrock of the surrounding area. It has also been shown in several of the studies that soil type and vegetation can be correlated. It can therefore be presumed that soil interaction with vegetation as a result of bedrock redistribution under Panther Mountain may provide evidence for an impact crater. The purpose of this study is to provide additional evidence supporting the argument that Panther Mountain, New York, is the location of a terrestrial meteor impact crater, using GIS data. This is achieved by examining the differentiation of correlations between vegetation and geology at and around the mountain. It is necessary to determine a correlation between variables both inside and outside the crater location. Therefore, it is the goal of this study to differentiate spatial relationships between vegetation and geology. This is performed by (1) using spatial information on land cover to map vegetation types

inside and outside the proposed crater site, (2) using geology classifications to map mineralogy and soil types inside and outside the proposed crater site, and (3) examining correlations between geology and vegetation for both inside and outside the proposed crater site. 9.2. ­MATERIALS 9.2.1. Study Area Panther Mountain is located in the town of Shandaken, Ulster County, in upstate New York, USA. The exact location of the site is at latitude/longitude 42.056389, ‐74.395. Panther Mountain itself has a maximum elevation of 1,134 m (base elevation at 834 m above sea level) and the proposed crater site has a diameter of 10 km (Figure 9.1). The mountain is circumvented by the Esopus and Woodland Creeks, which outline the unique rosette circular pattern of the site. The mountain is surrounded on all sides by high ridges, thereby contributing to the belief that the site is known to be a complex crater formation rather than just simple crater type (Isachsen et al., 1994; Isachsen, 1998; Donofrio, 2010). The surrounding area (outside of the crater site) is home to the remaining Catskill peaks, which make up the

Figure 9.1  Location of the Catskill Mountains and the greater Allegheny Plateau in New York State. The study area of Panther Mountain is highlighted.

114  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

extended region known as the Catskill Mountains. Areas bordering the crater site were used for classification comparisons. This primarily includes regions within the Catskill Mountains. The Catskill Mountains were formed over 300 million years ago during significant erosion and sedimentary redistribution of the Acadian Mountains to the east. It is during this time that some have predicted the meteor impact to have formed the current crater site. Extended periods of sedimentary distribution, erosion, and possible gravitational uplift led to the enlargement of Panther Mountain and the subsequent complex crater we see today. Based on the review of well cuttings, the estimated percentages of rock types that make up the impact site consist of 50% sub greywacke, 40% red silty shale, and 10% pebbly conglomerate (Isachsen et al., 1994). Much of the land cover at Panther Mountain is protected forest reserve, consisting of the typical Catskill Mountain vegetation. This includes the traditional hardwood forest of birch, beech, and maple and boreal forests of red spruce and balsam fur in the higher elevations (Isachsen et  al., 1994). There exists a small amount of exposed rock at two points in the area as well as very little human‐made structures. There are also several creeks in the area with varying amounts of seasonal flow. Little agriculture or urban environment exists around the mountain. The New York State Forest Preserve prohibits further land development 9.2.2. Data Sets The area covered by the available data included the study site as well as a significant portion of the southern Catskill Park. All spatial data were downloaded in Google Maps .kmz file format. The land cover map data were provided by a previous study, which used multitemporal Landsat imagery to map vegetation in the Catskill Park (Driese et al., 2004). The surficial mineralogy and bedrock mineralogy maps were provided by the New York State Museum, New York State Education Department, and created by the New York State Geological Survey. The soil map data were provided by the United States Department of Agriculture, Natural Resource Conservation Services. All .kmz files were downloaded from the website http:// www.huttonstreet.org/MapData.html, which had previously compiled the necessary map data for this study site. The land cover map, created by Driese et  al. (2004), contains 10 classifications of the significant vegetation types as well as nonvegetation. This includes primarily broadleaf woodland, however evergreen and deciduous forest also exist in the region. Two of the classifications (human built‐up and nonforest) are the only exception to the woodland vegetation. According to the study, there was only a 46% accuracy using traditional accuracy measurements; however, there was an increased map ­

accuracy of 71% using fuzzy accuracy assessment measurements. Furthermore, an increase in accuracy ­ to  84% and 90% was noted when the class list dropped to four and three classes, respectively. The surficial geology map has seven classes. The descriptions of each class type are presented in Table 9.1. The bedrock geology map includes only five classes. Each of the five classified units consists of a sequence of rock type (lithology), the primary being the most significant. Table  9.2 illustrates each class. The soil map contained many more classifications than the other data sets, 52  classes in all. All .kmz files came georectified to the World Geodetic System (WGS) 1984 projection. Similarly, each individual .kmz file was matched to overlap the same geographic area, the southern Catskill Park, during the data acquiring. It is important to note, however, that the data only cover the area within the Catskill Park boundary and do not include private property. In order to process the spatial data, it was necessary to be able to review, convert, and manipulate the data.

Table 9.1  Description of the Surficial Geology Classifications. Surficial geology type

Description

Recent alluvium Lacustrine silt and clay Outwash sand and gravel Peat muck swamp deposits Bedrock Kame deposits Glacial till

A mixture of clay, silt, sand, and gravel deposited by flowing rivers Silt and clay deposits left within lake regions Sand and gravel deposited in glacial moraines by meltwater Partially decayed organic material deposited in wetland areas Solid rock underlying surface deposits A ridge, hill, or mound of stratified drift deposited by glacial meltwater Unsorted material deposited directly by glacial ice

Table 9.2  Description of the Bedrock Geology Classifications. Geologic class

Lithology

Geologic period

Oneonta formation Lower Walton formation Upper Walton formation Slide Mountain formation Honesdale formation

Shale‐sandstone‐ conglomerate Shale‐sandstone‐ conglomerate Shale‐sandstone‐ conglomerate Sandstone‐shale‐ conglomerate Sandstone‐shale

Middle‐Upper Devonian Upper Devonian Upper Devonian Upper Devonian Upper Devonian

AN EXPLORATION OF THE PANTHER MOUNTAIN CRATER IMPACT USING SPATIAL DATA AND GIS SPATIAL  115

Four different software packages were used at different processing stages conducted in this study. Google Earth was used for the initial review of the downloaded (.kmz) files from the source website. FW tools version 2.4.7 was used as a platform for the GDAL/OGR module in order to convert the .kmz files to ArcGIS shapefiles (.shp). ArcGIS Suite 9.3 was the primary software used in this study. Several toolsets were used within the ArcGIS toolbox to process the map data files. For the final processing of the data, two separate programs were used: (1) A statistical tool within a GIS software (ArcGIS) was used to determine a correlation between the four variables (land cover, bedrock geology, surficial

geology, and soil types) and (2) a statistical analysis toolset within a spreadsheet software (Microsoft Excel 2011) was used to produce correlation statistics. Both statistical methods were used in order to determine the more efficient process. 9.3. ­METHODOLOGY Figure 9.2 gives a schematic representation of the general structure of the methodology, which was followed in this study, including data preprocessing, processing, and postprocessing, as they are described in the following subsections.

Preprocessing FW tools (ogr) conversion

.kmz files

.shp files

ArcGIS

Clip

Merge

Calculate geometry

Join

.txt files

Filter

Macro conversion

Excel

Method 1 Excel Method 2 ArcGIS

Feature to raster

Processing Band collection statistics

PCC analysis Results

Postprocessing

.txt files

Excel

Fisher’s Z transformation

Error assessment

Figure 9.2  Schematic summary of data processing methodology.

116  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

9.3.1. Data Preprocessing

9.3.2. Data Processing

The preprocessing of the spatial data involved several steps, with a primary goal to create suitable tables and images for statistical analysis. These steps, presented in the order with which they were performed, were: 1. Data Conversion. It was necessary to convert the .kmz files into a recognizable format for ArcGIS. This was done using the OGR Simple Feature Library command line tool ogr2ogr, within FW Tools command shell. 2. Merge. Following the conversion to shapefiles, several toolsets within ArcMAP were used to preprocess the data. The merge tool was used to create single layer shapefiles for the bedrock, surficial, and soil shapefiles. The conversion process had created individual layers for each of the classifications. Each of the shapefiles was then changed to show only the unique values within their attributes, thereby revealing their respective class values. 3. Clip. In order to differentiate the Panther Mountain site from the rest of the data, the clip function was applied to each of the shapefiles. 4. Feature to Raster Conversion. In order to run the statistical analysis within ArcGIS, it was necessary to convert each of clipped vector files into raster images. This was done using the feature to raster conversion tool within arc toolbox. Subsequently, eight new raster images were created, four for each region (land cover, bedrock geology, surficial geology, and soil type). The original vector files continued to be used for the rest of the preprocessing. 5. Calculate Geometry. Coordinate data were added to the land cover layer’s attribute table using the calculate geometry tool. Latitude and longitude were calculated for the centroid location of each shape within the layer. The new attribute table was then used to create a point shapefile (import xy data) whereby each point contained land cover type and grid location. 6. Join. The land cover point‐shapefile’s attribute table was joined with each of the three geology classification layers. The result was six new point‐shapefiles where each point contained geometric location, land cover type, and one of the three variables (soil type, surficial geology, or bedrock geology). The data for the six tables went through several processing steps. Several database techniques were used to arrange, convert, and remove data to prepare them for statistical analysis. In order to run the statistical analysis, the description for each land cover and geology type needed to be converted to a numerical value. Representative maps of the entire study area produced during the preprocessing stages are shown in Figure 9.3.

Pearson’s correlation coefficient (r) was used as the ­rimary statistical interpretation of the correlations p ­between variables. It was assumed that a linear relationship existed between land cover and one of the other variables (bedrock geology, surficial geology, or soil type). However, it was also possible that the other variables might exhibit promising correlations and therefore were also analyzed as such. The r coefficient was calculated using land cover with each of the other variables. There were six combinations between the data variables that were used to determine a correlation: (1) land cover and bedrock geology for Panther Mountain; (2) land cover and surficial geology for Panther Mountain; (3) land cover and soil type for Panther Mountain; (4) land cover and bedrock geology for the surrounding area; (5) land cover and surficial geology for surrounding area; and (6) land cover and soil type for the surrounding area. 9.3.3. Data Postprocessing Statistical significance was calculated at a 95% confidence interval level for both of the statistical methods used. Pearson’s correlation coefficients require a slightly different technique for evaluating amounts of error. Pearson’s correlation (r) is not normally distributed; therefore it was necessary to use Fisher’s (z) transformation. Once converted to Fisher’s (z) it was then possible to determine the upper and lower margins of error. These values were then converted back to (r) values. The following equations were used to calculate the error values: Conversion to Fisher’s z:



z

1 1 r ln 2 1 r

artanh r (9.1)

where, z is the Fisher’s (z) transformation, and r is the correlation coefficient. Standard error (SE) was calculated for each of the scenarios by: SE



1 N 3

(9.2)

where, N is the paired sample size. The upper (Zu) and lower (ZL) limits are given by the following equations, where 1.96 is the Z‐score for 95% confidence:

Zu

z

1.96 SE (9.3)

AN EXPLORATION OF THE PANTHER MOUNTAIN CRATER IMPACT USING SPATIAL DATA AND GIS SPATIAL  117 74°33′0″W

74°33′0″W

74°27′0″W

74°24′0″W

74°21′0″W

74°18′0″W

N

(a)

74°33′0″W

74°30′0″W

74°27′0″W

74°24′0″W

74°21′0″W

(b)

74°18′0″W

N

42°9′0″1

42°6′0″N 42°6′0″1

42°3′0″N

42°3′0″1

42°0′0″N

42°0′0″1

41°57′0″N

41°57′0

41°54′0″N

41°54′0

Soil types LCF LEE

SEB

ARD ARF

LOC

SaB

AI

SaC

BP

Lm MO

Ba

MTB

TkA

Be

ORC ORD

TkB

Co CgB

OdB

TuB

HSF

TuC

HXE

OgB OiC

He

Pa

VAD

HgA

Pt

W

HgB

RXC RXE

WLB

HgC HgD

RXF

Wa

HwD

Ra RvB

AA

41°51′0″N

41°48′0″N

41°45′0″N

42°9′0″N

LcD

SGB

41°51′0

Surficial geology types al

Su

k

TkC

lsc

41°48′0

og pm

VAB

r

WOB

0 1 2

4

6

t

8 Kilometers

N

(c)

0 12

4

6

8 Kilometers

41°45′0

N

(d)

42°6′0″1

42°6′0″N 42°3′0″1

42°3′0″N 42°0′0″1

42°0′0″N 41°57′0

41°57′0″N 41°54′0

41°54′0″N

Land cover types

41°51′0″N

Bedrock types Dgo Dsw Dwh

41°48′0″N

Dws Dww

0 1 2

41°45′0″N 74°33′0″W

74°30′0″W

74°270″W

74°24′0″W

4

74°21′0″W

6

8 Kilometers 74°18′0″W

41°51′0

Beech dominated Birch dominated Hemlock/pine Human built up Maple dominated Mixed evgn/decid Non-forest Oak dominated Other deciduous Spruce/fir 74°33′0″W

74°30′0″W

74°27′0″W

41°48′0

41°45′0

0 12 74°24′0″W

4

74°21′0″W

6

8 Kilometers 74°18′0″W

Figure 9.3  Maps of the study area for (a) soil type classification (52 classes); (b) surficial geology classification (7 classes); (c) bedrock geology classification (5 classes); (d) land cover classification (10 classes).

118  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

ZL



z

1.96 SE (9.4)

The intervals were converted back to (r) values by:



ru,l

e2 z 1 e2 z 1

tanh ru,l (9.5)

9.4.2. GIS Software Analysis

9.4. ­RESULTS The results of each of the two methods used to determine a correlation are presented in this section, followed by an error assessment, calculated for a 95% confidence interval. 9.4.1. Spreadsheet Software Analysis The six scenarios incorporated in the analysis based on spreadsheet program procedures included (1) land cover and soil type at Panther Mountain; (2) land cover and surficial geology at Panther Mountain; (3) land cover and bedrock geology at Panther Mountain; (4) land cover and soil type at the surrounding area; (5) land cover and surficial geology at the surrounding area; (6) land cover and bedrock geology at the surrounding area. The ­ correlations are shown in Table  9.3. A graph of the coefficients clearly shows the correlation differences between Panther Mountain and the surrounding region (Figure 9.4). The six correlation coefficients produced as a result of the Pearson’s correlation analysis highlighted several key points. It is evident that the largest coefficient is Panther Mountain’s bedrock geology and land cover scenario (–0.127), while the smallest coefficient belongs to the soil type and land cover scenario of the surrounding area (0.005). The surficial bedrock category held a positive correlation for both regions (Panther 0.049, Surrounding Area 0.074), while both the soil and bedrock scenarios for panther mountain were both negative (–0.100, –0.127, respectively). The bedrock scenario for the surrounding area was also negative (–0.031), albeit noticeably less than Panther Mountain’s coefficient. The change in magnitude between each scenario’s correlation was calculated to highlight the effects of the Table 9.3  The Results of the Correlation Analysis in Spreadsheet Software. Scenario Surficial geology Soil Bedrock geology Surficial geology Panther Soil Panther Bedrock geology Panther

crater impact at Panther Mountain (Table 9.4). Although the bedrock scenario for Panther Mountain had the ­largest standalone coefficient, the greatest change in correlation occurred in the soil scenario, which yielded a 10.52% difference. The bedrock scenario was close behind with 9.61% but there was little change between the surficial scenarios with only 2.55%.

Correlation coefficient (r) 0.074 0.005 −0.031 0.049 −0.100 −0.127

The correlation analysis conducted within the GIS software produced 12 correlation coefficients; however, only six were used to compare with the other correlation method (this was due to the fact that six scenarios ­conducted with GIS did not pertain to the objectives of the study. They did not use land cover as a variable (but are highlighted to show the potential use of other ­variables). A graph was also produced illustrating the relationships at Panther Mountain and the surrounding area for land cover and bedrock geology, land cover and surficial geology, and land cover and soil type (Figure 9.5). As depicted in the graph, all correlations were in the positive direction. The highest correlation was found ­between land cover and surficial geology at the surrounding area, with ~18%. The second largest correlation was between land cover and bedrock geology at Panther Mountain (~16%), and the third largest was land cover and soil at the surrounding area (~11%) (Table 9.5). The difference between correlations at Panther Mountain and the surrounding area were also more significant than the spreadsheet software method (Table 9.6). The largest change in correlation was found to be for the land cover and surficial geology scenario, with a total difference of ~13.5% (0.1354). There was a significant ~10.5% (0.1046) difference in correlations for the vegetation and bedrock geology scenario. The smallest correlation change was the vegetation and soil scenario with only a ~3% (0.0302) difference in correlation. 9.4.3. Error Assessment The error produced at the 95% confidence interval can be seen in Table  9.7, where the correlation values and total error intervals for each scenario are shown, as well as their percentage equivalents. 9.5. ­DISCUSSION 9.5.1. Spreadsheet Software Analysis Results The correlation analysis in the spreadsheet software proved to illustrate a significant difference between the two regions for the land cover and bedrock geology

AN EXPLORATION OF THE PANTHER MOUNTAIN CRATER IMPACT USING SPATIAL DATA AND GIS SPATIAL  119 Pearson's correlation coefficient for Panther Mountain and surrounding region (Excel) 0.1 Surr. area

Surficial, 0.074

Panther Mt.

Correlation coefficient (r)

0.05 Surficial, 0.0489 0

Soil, 0.005 Bedrock, –0.031

–0.05

–0.1

Soil, –0.100 Bedrock, –0.127

–0.15

Figure 9.4  Graph of the correlation analysis in spreadsheet software for each of the scenarios. There is a total of six correlation (r) results, three for Panther Mountain and three for the surrounding area.

Table 9.4  Difference Between Scenario Correlations in Spreadsheet Software. Scenario

Difference in (r)

Difference (%)

Soil Surficial Bedrock

0.105 0.026 0.096

10.52  2.55  9.61

s­ cenario, and the land cover and soil scenario. However, it is necessary to interpret the (r) values themselves to understand the correlation between the variables, at each region. Similarly, it is important to examine possible points of error and limitations set by the data. The correlations between surficial geology and land cover for both scenarios were the only significantly positive results.

Correlation coefficient for Panther Mountain and surrounding region (ArcGIS - Band collection statistics) 0.2 0.18

Correlation coefficient (r)

0.16

Surf, 0.181 Bed, 0.161

0.14

Soil, 0.142

0.12 0.1

Soil, 0.112

0.08 0.06 0.04 0.02 0

Bed, 0.058 Surf, 0.046 Surr. area Panther Mt.

Figure 9.5  Graph of the correlation analysis in GIS for each of the scenarios. There is a total of six correlation (r) results, three for Panther Mountain and three for the surrounding area.

120  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Table 9.5  The Results of the Correlation Analysis in GIS. Scenario

Correlation coefficient (r)

Surficial geology Soil Bedrock geology Surficial geology Panther Soil Panther Bedrock geology Panther

0.181 0.112 0.058 0.046 0.142 0.161

Table 9.6  Difference Between Scenario Correlations in GIS. Scenario

Difference in (r)

Difference (%)

Soil Surficial Bedrock

0.0302 0.1354 0.1036

 3.02 13.54 10.36

Although both correlations showed only a 2.6% difference, they accounted for a 7.4% and 4.9% correlation between surficial geology and land cover. This can be interpreted to show that surficial geology may have a link to vegetation type; however there are more variables to consider with this relationship. The second scenario where soil is compared with land cover for both regions, showed a much unexpected result. The almost nonexistent correlation between soil and land cover at the surrounding region is a puzzling outcome. It is a common understanding that soil type influences vegetation diversity. Looking at the soil map of the ­ ­surrounding region (Figure 9.3a), it is noticeable that the soil types increase in frequency toward the southern end of the Catskill Park. It is possible that the increase in the surrounding region’s sample size added to the discrepancy

between correlations. Furthermore, the Panther Mountain region yielded a negative (–10%) correlation between soil and vegetation. This suggests that soil has little to do with vegetation at the surrounding area, yet has a more significant negative association with vegetation at Panther Mountain. The last scenario showing bedrock and land cover correlation appeared to be the most promising in terms of individual correlations. The correlation between bedrock and land cover at the surrounding area was a low –3%, suggesting that bedrock has a minor relationship with ­ vegetation. However, the significant –13% correlation ­ between bedrock and vegetation at Panther Mountain ­ ­suggests that bedrock has an increased, negative impact on vegetation. This supports the argument that the Panther Mountain crater impact has influenced the bedrock geology in such a way as to distinguish it from the ­surrounding area. The bedrock in turn has impacted the distribution of vegetation differently than the other region. The coefficient values themselves show little correlation in their own right. All the values fall under a 13% correlation. This has little to show in terms of significant linear dependency. However, the aim of this study was not to determine linear dependency necessarily, but to show the difference in correlation between the two regions. The 11% difference between the soil scenario correlations is a more profound result (Table 9.4). Similarly, the almost 10% difference between the bedrock scenario correlations is also more significant than the correlation values themselves (Table  9.4). This amount of change between the two regions is cause to question the unforeseen differences between the Panther Mountain crater and the surrounding area, especially considering that

Table 9.7  The Results of the Error Assessment for All Scenarios (18 Total). Scenario Surficial surrounding (spreadsheet) Soil surrounding (spreadsheet) Bedrock surrounding (spreadsheet) Surficial Panther (spreadsheet) Soil Panther (spreadsheet) Bedrock Panther (spreadsheet) Surficial surrounding (GIS) Soil surrounding (GIS) Bedrock surrounding (GIS) Bedrock soil surrounding (GIS) Bedrock surficial surrounding (GIS) Surficial soil surrounding (GIS) Surficial Panther (GIS) Soil Panther (GIS) Bedrock Panther (GIS) Bedrock soil Panther (GIS) Bedrock surficial Panther (GIS) Surficial soil Panther (GIS)

PCC (r) 0.074 0.005 −0.031 0.049 −0.100 −0.127 0.181 0.112 0.058 −0.008 −0.300 0.023 0.046 0.142 0.161 0.103 0.059 −0.028

Total error (+/–) (95% conf.)

% (r)

% error (+/–)

0.010 0.010 0.010 0.019 0.019 0.019 0.014 0.015 0.013 0.009 0.009 0.009 0.009 0.009 0.009 0.015 0.015 0.015

7.41 0.52 −3.09 4.86 −10.00 −12.69 18.12 11.17 5.76 −0.78 −20.01 2.31 4.58 14.19 16.11 10.29 5.91 −2.84

0.99 1.04 0.99 1.93 1.92 1.90 1.42 1.45 1.32 0.90 0.86 0.90 0.90 0.88 0.87 1.45 1.46 1.47

AN EXPLORATION OF THE PANTHER MOUNTAIN CRATER IMPACT USING SPATIAL DATA AND GIS SPATIAL  121

both regions have formed under the same geologic processes. The implication of this difference in regional correlation may later contribute to a more efficient detection method for impact craters. 9.5.2. GIS Software Analysis Results The results of the correlation analysis within the GIS software were significantly different than those produced from the analysis conducted in the spreadsheet software. This is most likely attributed to a difference in technique. Nonetheless, it was evident that a distinct difference was found between Panther Mountain and the surrounding area correlations. A significant difference between the two approaches was that in the case of the GIS analysis, there were no negative correlations. This claims that there was a positive relationship between all the variables for each scenario. Unlike the spreadsheet software method, the largest correlation was found between land cover and surficial geology at the surrounding area (~18%), suggesting that surface geology has a more significant impact on vegetation away from the crater site. Looking at the surficial geology map (Figure 9.3b), it is evident that surface geology is more diverse at the surrounding region. The high land cover and bedrock correlation at Panther Mountain (~16%) also supports the idea that the ­underlying bedrock has had an impact on vegetation at the surface of the crater site. One of the most visible differences between methods is the high correlation for the land cover and soil scenario at both regions (for the GIS analysis). Although the difference in correlations between each region is minimal, it suggests that soil has a positive relationship with vegetation for both sites. This is an expected result, however (soil properties are known to correlate with vegetation), which validates the GIS method over the spreadsheet software method (which showed a negative correlation). A low correlation between land cover and surficial geology, but a high correlation between land cover and bedrock geology at Panther Mountain also provide further evidence toward the argument of the crater’s presence in the area. In reference to the surficial geology map (Figure 9.3b), it is evident that a majority of the surficial geology at Panther Mountain is actually bedrock. This is supportive of the idea that the thousands of years of erosion at Panther Mountain that exposed the crater outline also removed much of the surficial geology. Perhaps the most substantial piece of evidence for the Panther Mountain crater site is found in the differences between the scenario correlations (Table  9.6). The results showed an almost 14% difference in correlation for the land cover and surficial geology scenario. Similarly, there was a 10% change in correlation for the land cover and bedrock geology scenario. These percentages present a validation

for the location of a crater impact at Panther Mountain, provided the meteor disturbed the underlying bedrock on impact (in conjunction with complex crater creation). The correlations between bedrock and surface geology, bedrock geology and soil type, as well as surficial geology and soil type were also calculated. Although these values were not part of the objectives of the study, they still illustrate several key points. As seen in Table 9.8, there are two significant correlations to be noted. Apart from the correlations that are below 3% (r < ~2.8), the bedrock geology and soil correlation for Panther Mountain as well as the surficial geology and bedrock geology correlation for the surrounding area are greater than 10%. At the Panther Mountain impact site, bedrock geology is found to have an ~10% positive correlation with soil type. This is an encouraging result, as bedrock is the primary geologic layer beneath soil at Panther Mountain (as oppose to other surficial geology types). A higher percentage would have been expected, but a linear ­dependency is not necessarily the case between these two variables. The reverse is happening at the surrounding area, however. A –20% correlation exists between surficial geology and bedrock geology at this region. With the strongest correlation (albeit negative) between any of the variables, this result suggests bedrock geology has a significant negative relationship with surficial geology (or vice versa) outside the crater site. This supports the Panther Mountain impact site because it suggests that surficial geology has a greater distribution in the surrounding area, whereas inside the crater site, bedrock is the primary geology type at the surface. 9.5.3. Combined Evaluation Although the graphs of the correlations produced using the two methods appear to be significantly different, there is a similarity between them. The magnitudes of each scenario for both graphs are comparable. In the spreadsheet software analysis, the Panther Mountain correlations from largest to smallest are bedrock geology– soil–surficial geology. In the GIS analysis, the Panther Mountain correlations from largest to smallest are in the same order, bedrock geology–soil–surficial geology. In the surrounding area scenario, this case is almost true;

Table 9.8  GIS Analysis Correlation Results for the Six Scenarios That Did Not Include Land Cover As a Variable. Scenario Surficial Geology Panther Bedrock Geology Panther Surficial Geology surrounding Bedrock Geology surrounding

Bedrock Geology

Soil

 0.059 – −0.200 –

−0.028 0.103 0.023 −0.008

122  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

however the vegetation and soil correlation in this ­scenario is nearly nonexistent. There are several areas within this study that may have impacted the results of each method and are worth examining. The selection of the surrounding region using the clip tool may have influenced the low correlation between soil and vegetation. Although the Panther Mountain region and the surrounding area are both part of the southern Catskill Park, there may be areas within the park that have considerably different geology and soil types. Similarly, the entire extent of the Catskill Park was not used for this study, only the southern region. Would the data have been provided for the whole park, there is a ­possibility that the change in the correlation may have been different. The technique used for the spreadsheet software method of this study examined the relationship between point data. Although the original data were spatially distinctive, the spatial components were ignored for the final processing. Since the data were split between two regions, Panther Mountain and the surrounding area, it was not necessary to maintain locational accuracy but rather keep paired data points together. It is possible, however, to examine the spatial relationships between the data sets as a whole (as in the GIS software). GIS statistical analysis was used to examine correlations between variables based on spatial locations of an entire area. With this method, correlations were ­determined using several layers, each containing one of the four variables: (1) vegetation, (2) surficial geology, (3) bedrock geology, and (4) soil type. The results of this technique were visibly different, although a similar pattern was recognized. The differences between the correlations themselves can be seen in Table 9.9. There was not a significant difference between the ­correlations in a majority of the cases; however two of the cases showed substantial differences. Land cover and surficial geology and land cover and soil type at the ­surrounding area both showed over a 10% difference in correlations between the two methods. It is possible that this 10% discrepancy can be attributed to error in the processing of data. It is also possible that spatial characteristics (not used with the spreadsheet software method) have a greater impact on correlation than previously thought. Further study should address this result. Table 9.9  The Difference Between the Results From Each Method, Shown for Both Panther Mountain and the Surrounding Area. Scenario

Surrounding area diff.

Panther Mt. diff.

Surficial Soil Bedrock

0.107 0.107 0.027

0.003 0.042 0.034

The low correction coefficients for vegetation overall (r  < 20% for all land cover correlations), suggest that there are other factors affecting the distribution of ­vegetation. Although other variables were considered, it was the difference between correlations that was most important, not necessarily the extent of the correlation coefficients. Variables, such as topography, slope, aspect, and vicinity to water, all have effects on vegetation growth. Further studies may use these variables in conjunction with bedrock to show a greater correlation. The results show that error values are overall low, less than 2% error for all cases. However, the smaller ­correlations suffered the greatest error. In some cases, the error is greater than the correlation. This is perhaps due to the idea that as a correlation approaches zero (no correlation), error values are increasingly unreliable (considering that a positive correlation cannot become a negative correlation through error alone). It should also be noted that the GIS method scenarios have lower error percentages overall as compared with the spreadsheet software method. The strongest correlations, within the GIS scenario, also have some of the lowest error ­percentages. This significantly supports the results of the GIS method (and spatial correlation) over the spreadsheet software method. 9.6. ­CONCLUSIONS There were three primary objectives of this study, namely: (1) create a land cover map using classified GIS data for Panther Mountain and the surrounding area; (2) ­create mineralogy and soil type maps using classified GIS data for Panther Mountain and the surrounding area; and (3)  compare correlations between variables at Panther Mountain and the surrounding area. These objectives were fulfilled in order to provide evidence for the main goal of this work, which was to support the statement: “The differentiation of correlations between vegetation and geology at and around Panther Mountain, New York provides evidence for the location of a terrestrial meteor impact crater.” Additionally, although it was not the ­primary initiative of this study, it can be stated that the accuracy of the GIS method over the spreadsheet software method supports the idea that spatial location can have a significant role in correlation analysis. The difference in correlations between the coefficients of the two regions provides some evidence for the crater impact at Panther Mountain. Although these correlation values by themselves do not provide enough evidence of the existence of a crater impact, they do provide supporting evidence of an impact site. Coupled with other statistical analyses, perhaps the effectiveness of this statistical analysis may hold greater evidence for studying impact craters.

AN EXPLORATION OF THE PANTHER MOUNTAIN CRATER IMPACT USING SPATIAL DATA AND GIS SPATIAL  123

This study has addressed a new method for examining impact sites using correlation statistics. At this point, it is not possible to compare these results with other research methods for there has been no prior examination of terrestrial impact sites using a correlation comparison. Alternatively, this method can be applied to support current studies of terrestrial crater sites. Similarly, there is much that can be added to this method in order to improve accuracy and efficiency. What do the results of this study suggest about the implication of using GIS and statistics to provide ­evidence for terrestrial impact craters? GIS combined with statistical analysis is an efficient tool that can be used to show relationships between data types. This study examined how Pearson’s correlation coefficient, used in conjunction with GIS data, was able to illustrate distinctions in the correlation between land cover and soil type, land cover and surficial geology, and land cover and ­bedrock geology at the Panther Mountain crater and the surrounding region. It is by initiating alternative methods of analysis that the state of scientific research can be advanced to benefit all levels of the scientific community. ­ACKNOWLEDGMENTS The authors are grateful to the reviewers for their constructive feedback, which improved the manuscript. GPP’s contribution, has been supported by the FP7‐ People project ENViSIon‐EO (project reference number 334533) and the author gratefully acknowledges the financial support provided. ­REFERENCES Aich, S., & Gross, M. R. (2008). Geospatial analysis of the association between bedrock fractures and vegetation in an arid environment. International Journal of Remote Sensing, 29 (23), 6937–6955. Chen, B.‐M., Wang, G.‐X., Cheng, D.‐L., Deng, J.‐M., Peng, S.‐L., & An, F.‐B. (2007). Vegetation change and soil nutrient distribution along an oasis‐desert transitional zone in northwestern China. Journal of Integrative Plant Biology, 49 (11), 1537–1547. Donofrio, R. R. (2010). New York’s Panther Mountain impact crater: Enormous gas potential without hydraulic. Parwest Land Exploration, Inc. Oklahoma City, OK. Driese, K. L., Reiners, W. A., Lovett, G. M., & Simkin, S. M. (2004). A vegetation map for the Catskill Park, NY, derived from multi‐temporal Landsat imagery and GIS data. Northeastern Naturalist, 11(4), 421–442. Gerdol, R., Ferrari, C., & Piccoli, F. (1985). Correlation between soil characters and forest types: A study in multiple discriminant analysis. Vegetation, 60(1), 49–56.

Guterl, F. (2000). The Panther Mountain crater DISCOVER Magazine. Retrieved from discovermagazine.com/2000/aug/ featcrater. He, X.‐Y., Wang, K.‐L., Zhang, W., Chen, Z.‐H., Zhu, Y.‐G., & Chen, H.‐S. (2008). Positive correlation between soil bacterial metabolic and plant species diversity and bacterial and fungal diversity in a vegetation succession on Karst. Plant and Soil, 307(1–2), 123–134. Isachsen, Y. (1998). Metallic Spherules and a microtektite support the interpretation of a buried impact crater beneath Panther Mountain in the central Catskill Mountains, New York. Meteoritics & Planetary Science, 33. Isachsen, Y., Wright, S. F., & Revetta, F. A. (1994). The Panther Mountain circular feature possibly hides a buried impact crater. Northeastern Geology, 16(2), 123. Neff, A. J. C., Reynolds, R., Sanford, R. L., Fernandez, D., Lamothe, P., & Neff, J. C. (2006). Controls of bedrock geochemistry on soil and plant nutrients in southeastern Utah. Ecosystems, 9(6), 879–893. Pastor, J., Aber, J. D., Mcclaugherty, C. A., & Melillo, J. M. (1982). Geology, soils and vegetation of Blackhawk Island, Wisconsin. American Midland Naturalist, 108(2), 266–277. Puy, D. D. U., & Moat, J. (1996). A refined classification of the primary vegetation of Madagascar based on the underlying geology: Using GIS to map its distribution and assess it. Biogeographic de Madagascar, 205–218. Rajmon, D. (2007). Suspected Earth Impact Sites database. Houston, TX. Retrieved from http://impacts.rajmon.cz/ IDdata.html. Rodríguez‐Loinaz, G., Onaindia, M., Amezaga, I., Mijangos, I., & Garbisu, C. (2008). Relationship between vegetation diversity and soil functional diversity in native mixed‐oak forests. Soil Biology and Biochemistry, 40(1), 49–60. Searcy, K. B., Wilson, B. F., & Fownes, J. H. (2003). Influence of bedrock and aspect on soils and plant distribution in the Holyoke Range, Massachusetts. Journal of the Torrey Botanical Society, 158–169. Stothoff, S., Or, D., Groeneveld, D., & Jones, S. (1999). The effect of vegetation on infiltration in shallow soils underlain by fissured bedrock. Journal of Hydrology, 218(3–4), 169–190. Williard, K. W. J., Dewalle, D. R., & Edwards, P. J. (2005). Influence of bedrock geology and tree species composition on stream nitrate concentrations in mid‐Appalachian forested watersheds. Water, Air, and Soil Pollution, 160, 55–76; doi.org/10.1007/s11270‐005‐3649‐4 Xu, X.‐L., Ma, K.‐M., Fu, B.‐J., Song, C.‐J., & Liu, W. (2008). Relationships between vegetation and soil and topography in a dry warm river valley, SW China. Catena, 75(2), 138–145. Yang, X., Zhang, K., Hou, R., & Ci, L. (2007). Exclusion effects on vegetation characteristics and their correlation to soil factors in the semi‐arid rangeland of Mu Us Sandland, China. Frontiers of Biology in China, 2(2), 210–217. Zumsprekel, H., & Bischoff, L. (2005). Remote sensing and GIS analyses of the Strangways impact structure, Northern Territory. Australian Journal of Earth Sciences, 52(4–5), 621–630.

Section III Land Hazards and Disasters

10 Satellite Radar Interferometry Processing and Elevation Change Analysis for Geoenvironmental Hazard Assessment Sergey Stankevich1, Iryna Piestova1, Anna Kozlova1, Olga Titarenko1, and Sudhir Kumar Singh2

ABSTRACT The technique for potentially hazardous land surface displacement mapping using satellite radar interferometry processing and elevation change analysis is presented. The primary data source is Sentinel‐1 satellite synthetic aperture radar (SAR) imagery. The high‐precision measured locations of ground control points (GCP) are used for land surface displacement accuracy improvement. The developed technique was used for elevation change analysis within the Kryvyi Rih (Ukraine) urban area. This region is characterized by a high degree of geoenvironmental hazard. The final map of geoenvironmental hazard within the study area is provided as output of satellite radar interferometry data analysis. 10.1. ­INTRODUCTION

10.2. ­STUDY AREA

Human activity leads to an anthropogenic impact on the natural landscape. Building industrial and energy potential, concentration of population in urban areas, and environmental pollution is a challenge to natural and social systems. The frequency and scope of human‐ made and natural disasters significantly increased during the last few years. Environmental hazard maps delivering on demand are an urgent necessity now. This emphasizes the relevance of methods of development for geoenvironmental condition assessment to forecast, ­prevent, or mitigate the aftermath of disasters (Alcántara‐ Ayala & Goudie, 2010). Such problems are really important within the industrial regions of Ukraine. The Kryvyi Rih urban area is a typical one.

Kryvyi Rih is the center of the Kryvorizkyi Iron Ore Basin (Kryvbas). This is the most important source of raw materials for metallurgy in Ukraine. A powerful industrial area around the city was formed as the result of the development of the Kryvbas deposits. The Kryvyi Rih industrial agglomeration lies from north to south along the Saksagan and Ingulets Rivers up to 120 km in length. The population of Kryvyi Rih city is about 650,000. The population reaches a million people with the Kryvyi Rih agglomeration. The Kryvyi Rih industrial region plays a leading role in the economy of Ukraine. It has a strategic significance for economic independence and security of the state. Vigorous technogenic activity been undertaken for many years within the Kryvyi Rih urban area. Minerals are mined by the open‐pit method and the number of operating mines has increased. More than 35,000 hectares of land were disturbed due to mining processing. A powerful anthropogenic impact on the geological environment can be observed within the 585 km2 area of the Kryvyi Rih basin. This circumstance leads to changes in

 Scientific Centre for Aerospace Research of the Earth, National Academy of Sciences of Ukraine, Kiev, Ukraine 2  K. Banerjee Centre of Atmospheric and Ocean Studies, University of Allahabad, Allahabad, India 1

Techniques for Disaster Risk Management and Mitigation, First Edition. Edited by Prashant K. Srivastava, Sudhir Kumar Singh, U. C. Mohanty, and Tad Murty. © 2020 John Wiley & Sons, Inc. Published 2020 by John Wiley & Sons, Inc. 127

128  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

the regional landscape as well as in geological and hydrogeological conditions, namely: 1. Deep opencast mines occupy almost 40 km2, from which about 14 billion cubic meters of ore and rocks have been seized and moved. 2. Mine excavations occupy about 6 billion cubic meters. 3. Extensive underground cavities have produced a significantly weakened mass of rock inside a rock volume of 40–50 million m3; subsidence zones and dips have formed on part of the territory. 4. Tailings of iron ore concentrations spread over of 70  km2 area, more than 3 billion tons of crushed rock mass stored in dams of 35–90 m in height. 5. More than 9,000 hectares of land are waterlogged; closing and flooding of some mines violated the natural hydrogeological conditions. These drivers create an excessive human impact on the land surface and geological environment. Preliminary structural tectonic evaluations were ­conducted and geophysical data were obtained. Mining driftage, which is submeridional‐elongate shaped, coincides with the regional geological fault. This fault extends from the seismically active Vrancea zone, which is located across the Romanian and Ukrainian Carpathians (Kazakov et al., 2005). If a natural earthquake in Romania occurs, then the fault may be a wave‐channel and may provoke a local technogenic earthquake in Kryvbas. The Crimean Black Sea seismically active zone may also be able to create some seismic problems for the Kryvyi Rih. Ore mining was carried out at small depth (20–300 m) mainly (Berezovsky & Kolesnik, 2014). Therefore, areas of mining outlets pose a serious hazard to Kryvyi Rih city. Today the residential and industrial area of the city comes close to such sites directly. Cavities that are huge in total volume occur in the tectonically complicated area of the Kryvyi Rih. Even in a stable geological environment, this circumstance may result in seismotectonic processes activation and seismicity increasing (Pigulevsky et al., 2009). Human‐made stresses, caused by industrial explosions for mining, are superimposed on the local stress fields. In addition, global geodynamic processes deliver hazardous influence. The process of the oncoming flow of the African and Eurasian continental plates still continues. A powerful earthquake‐prone belt of the planet is formed as a result. This influence is a global driver within the study area according to the latest seismic and geophysical data. Global tectonic deformation processes change intensity cyclically. At present, they are in the activation phase. Observational data within the territory of the Kryvyi Rih basin substantiate this fact. The data of the United States Geological Survey (USGS) show an increase in the

number of destructive events such as dips and shifts of the land surface and natural and human‐made earthquakes, and so on. Earthquakes of magnitude 3–3.7 occurred each 8–10 years in Kryvyi Rih city during the last century. But four earthshocks with magnitude 3.2–4 were recorded over the past 10 years (last one was at 24 September 2016). The nature of these phenomena in the Kryvyi Rih territory has not been studied practically by instrument methods. Objective data on the seismic resistance of different sites inside the city are missing. There is not enough monitoring information to assess trends of exogeodynamic processes and their relation to seismic events in the region. 10.3. ­GEOENVIRONMENTAL MONITORING Endogenous and exogenous processes may initiate crust motion. Natural geoenvironment in the region is broken due to mining activity, cavities, and pumping of mine waters. Therefore, it can lead to disasters, such as soil destruction, landslides, sinkholes, and so on. Thus, efficient methods for in‐depth analysis of dangerous geological processes, and assessment of stress‐strain state of the rocks in the field of geomechanical and geological disturbances, become especially important. Large financial and time expenses are needed for ground‐based integrated monitoring system development over areas of geological hazard. Modern remote sensing methods provide data with the required spatial resolution and revisit time and these become the appropriate tools for required data acquisition. Airborne and satellite imagery provide quick assessment of geoecological environmental conditions, extract a high variety of land surface features, and reduce time and expenses. Comparison of different times and different sources for obtaining images (vehicles and machines) is possible. They need to have an unified scale, adjust and transform, used in similar spectral bands. Comparative ­analysis of two multitemporal images captures the changes in the violation of the geological environment in time and space. The main sphere of application of satellite data for the study of geology, a study of stratigraphy and lithologic‐ petrographic properties of the rock, are as follows. Mainly they are relevant to the following studies: 1. Structural‐tectonic study of the territory for which satellite images provide essentially new information then increase the depth of investigation 2. Satellite information used to search for mineral deposits 3. Geological mapping, creation of new kinds of geological and cartographic products 4. Study of geothermal and volcanic zones

SATELLITE RADAR INTERFEROMETRY PROCESSING AND ELEVATION CHANGE ANALYSIS  129

Satellite imagery is used for the study of tectonic zonation. A large amount of information about the geological environment can be obtained by satellite imagery interpretation. Plicate structures and faults of different orders, both regional and local, can be detected reliably. Line faults with/without displacements of structural blocks are recognized confidently. The study of the lithosphere’s deep structure using satellite imagery has great theoretical significance and makes it possible to see the important practical applications. Satellite imagery is useful for mineral prospecting. Intercorrelation between the structural‐tectonic interpretation outputs and the spatial distribution of mineral deposits exists. This technique helps to identify patterns of their allocation and association with known geological structures. Thus, deposit prediction possibility arises (Mikhaylov et al., 1993). The modern multispectral satellite imagery is analyzed and compared with the geological and cartographical data on the study area. This enables estimation and forecasting of the hazardous geological processes. Detection of geodynamic zones based on satellite imagery is very important in geoenvironmental studies. Geodynamic zones are identified as faulting dislocations in sedimentary cover because of recent tectogenesis. They are extracted by satellite imagery through linearly arranged elements of landscape (i.e., lineaments). Lineaments are areas of relative instability and high migration of substances. Some of them inherit the faults of the geologic basement. Certain hazardous geological processes, such as landslides, areal and linear erosion, suffusion and significant changes in the level of groundwater, are features of the geodynamic zones. The abovementioned processes are especially active within geodynamic nodes. The geological environment experiences permanent changes, which need continuous monitoring control and prediction. Image structure and texture patterns in different spectral bands contribute to territory under­ standing by landscape features and specific natural and human‐made components. As practice shows, the best results are achieved when satellite and ground‐based measurements are carried out both jointly and synchronously. The study outcomes are integrated and extrapolated in a geoinformation system and layered over topographic base map.

for environmental monitoring, including terrain or elevation data, soil moisture, depth of groundwater, ­ ­flooding areas, waterlogging, and others (Titarenko, 2014). An approach to rapid assessment of anthropogenic impact on the landscapes of the city of Kryvyi Rih is proposed based on remote sensing. Terrain defined the main indicator of remote anthropogenic impact. Recently the variety of tasks that use the method of satellite differential radar interferometry became wider. The radar interferometry technique engages the Earth’s surface imaging from satellites equipped with synthetic aperture radar (SAR) and it detects small land s­urface displacements with high accuracy.

10.4. ­RADAR INTERFEROMETRY

For processing and constructing an interferogram, the Sentinel‐1 interferometric synthetic aperture radar (InSAR) Single Look Complex (SLC) products must be taken in pairs with an interval of several months to determine the changes in the land surface. The difference between the two images of a pair will be 6–12 days approximately. Coregistration of the images that compose a pair into a joint stack is a principal step in interferogram generation.

Novel optical and radar imaging systems are a widely accepted source of accurate and operative data for Earth observation and monitoring. They are used to recognize soil types and analyze vegetation state and land degradation processes and pollution emissions. Radar surveys provide extensive supporting information

10.4.1. Data Collection The novel European Sentinel‐1 C‐band radar satellite system, operated by the European Space Agency (ESA), was selected as primary provider for data about the study area. SAR data also can be used for land deformation monitoring. The radar interferometry remote sensing technique combines two or more SAR images over the same area to detect phase changes occurring between acquisitions. The Sentinel‐1 mission has been especially designed as a two‐satellite constellation (Sentinel‐1A and Sentinel‐1B) for large‐scale area radar interferometry (Ferretti et al., 2015). Sentinel‐1A was launched on 3 April 2014, and Sentinel‐1B, on 25 April 2016. The identical satellite’s orbit has an altitude of almost 700 km. This configuration optimizes coverage, and provides global‐specified revisit time of six days. Moreover, the onboard SAR has dual‐ polarization capability (HH + HV or VV + VH), which can provide more ground surface information. The advantage of satellite radar as a remote sensing tool is that it can image the Earth’s surface through rain and cloud, and regardless of whether it is day or night. This is particularly useful for monitoring areas prone to long periods of darkness, such as the Arctic, or providing imagery for emergency response during extreme weather conditions. The open‐source software named Sentinel Application Platform (SNAP) provided by ESA was used for satellite radar interferometry data processing (Lazecky et al., 2017). 10.4.2. Data Processing

130  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

This process ensures that each ground target contributes to the same pixel in both images. Interferometric Wide (IW) SLC Sentinel‐1 data products in total contain three (single polarization) or six (dual polarization) images. Among them, one image per subswath and one per polarization band. The subswath images are generated by a series of bursts. For seamless integration of all burst data into a single image, the Deburst operator from the Sentinel‐1 toolbox menu could be applied. After that, the interferogram is formed and flattened by removing the topographic phase. Because of the 2π cyclic nature of the interferometric phase, the flattened interferogram produces an ambiguous measurement of the relative terrain altitude. A phase difference between two points on the flattened interferogram provides a measurement of the actual altitude variation. By means of adding the correct integer multiple of 2π to the interferometric fringes, phase unwrapping is carried out. For its proper implementation, the signal‐to‐noise ratio is increased by filtering the phase. In order to reduce the scope of processing, a subset around the selected study area is created for the pairs of images. The Statistical‐cost, Network‐flow Algorithm for Phase Unwrapping (SNAPHU) proposed by Chen and Zebker could be applied to Sentinel‐1 as well. This algorithm is available for Linux workstations only (Chen & Zebker, 2002). Because of topographical variations of a scene, the tilt of the satellite sensor’s sight axis, distances in SAR images are distorted. These distortions could be compensated for though terrain corrections so that the geometric representation of the image will be as close as possible to the real land surface. Terrain correction applied to the phase band enables geocoding the output product. A general dataflow diagram (Grandin, 2015) of the interferometric processing is shown in Figure  10.1. The same processes are applied to each pair of SAR images individually. The result of the processing each pair of images is terrain elevations maps. Then, elevation adjustments are performed by the Inverse Distance Weighting (IDW) method. It is executed using geodetic ground control points (O’Sullivan & Unwin, 2010). The total error in geodetic heights determined by radar interferometry comprises two components: systematic and accidental. The first one describes the deviation of reference ellipsoid within the area of interest; and the second one is caused by the noise of registration and processing of SAR images. The IDW method provides the equality between terrain elevations and geodetic heights within ground control points as well as inversely related elevations adjustment in the other points of radar image. The final result of the processing is the elevation change map, which is obtained as a difference between elevation images of input set.

10.5. ­MULTITEMPORAL ELEVATIONS CHANGE ANALYSIS One of the most important tasks for long‐term satellite monitoring is discovering and recognizing changes occurring in geosystems (Jat et al., 2008). The vast variety of methods for computer‐aided change detection within the monitored area can be generalized in three main groups: change detection in image pixels, change detection in land cover classes, and change detection in semantic space of land formations or processes (Coppin et  al., 2004). The first approach uses the multispectral images directly using numerous metrics and transformations, of which change‐vector analysis (CVA) is the most advanced (Chen et al., 2003). This method has a relatively simple implementation, while interpretation of its results involves significant difficulties. The second approach is based on a comparison not of the satellite images, but of the land cover classes detected from them. In most current studies, either areas of corresponding changes are just calculated (Antrop, 2004) or specific landscape metrics are implemented and estimated, such as the number of patches (NP), the patch density (PD), the largest patch index (LPI), and the ­contagion (Fichera et  al., 2012). However, this approach is the most promising for further improvement and operation. The third approach is significantly beyond the scope of remote sensing. It requires engaging deep knowledge about study objects from various fields of science. This approach also involves extensive ground‐based and laboratory research and sophisticated simulation. The procedure proposed here for long‐term elevations change mapping is implemented in two stages. In the first stage, a multitemporal elevation map is generated. After that, in the next stage, the hazard of elevations change is assessed. 10.5.1. Multitemporal Elevations Map In the first stage, a time series of the radar images for the region of interest is acquired. A time period between initial and final states recorded at the images should be significant or sufficient for some specific geological processes. The next required step is imagery preprocessing. It involves georeferencing, radiometric sigma0‐calibration, speckle filtration, azimuth‐range transforming, pixel‐to‐ pixel coregistration, and using the Deburst operator. The Sentinel‐1 InSAR SLC images were collected to figure out land‐surface elevation changes over the study area. The acquired time series consists of six images integrated in three pairs with a six‐month interval. Time gap between images coupled in pairs is 12 days approximately (Figure 10.2).

SATELLITE RADAR INTERFEROMETRY PROCESSING AND ELEVATION CHANGE ANALYSIS  131 Area of interest

SAR Image 1

SAR Image 2

Co-registration

Co-registration

Deburst

Deburst

Interferogram formation

Interferogram formation

Topographic phase removing

Topographic phase removing

Filtering

Filtering

Subsetting

Subsetting

Unwrapping

Unwrapping

Terrain correction

Terrain correction

Terrain elevations

Terrain elevations

Elevation adjustment

Elevation adjustment

Geocontrol points

Elevation change map

Figure 10.1  SAR image interferometric processing flow diagram.

2017

2016 05Feb

17Feb

01Jun

13 Jun

28 Nov

12 days

12 days

6 month

12 days

6 month

Figure 10.2  Sentinel‐1 (SLC) image acquisition timeline.

10 Dec

132  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION 33°10ʹ0″B

33°20ʹ0″B

33°30ʹ0″B

33°40ʹ0″B

48°10ʹ0″C

48°0ʹ0″C

47°50ʹ0″C

47°40ʹ0″C

0

2,5

5

10

15

20 km

Figure 10.3  Satellite radar interferometry study area.

The study area is situated around the Kryvyi Rih, a large city in the Dnipropetrovsk region of Ukraine. Its size is 25 × 58 km (Figure 10.3). To ensure subsequent ground‐truth correction of the achieved results, geodetic ground control points were also collected. As a result, two elevation change maps for each six‐ month interval were obtained. These maps display ­elevation difference calculated from adjacent image pairs by means of radar interferometry. The calculation procedure is described in section 10.4.2. Negative changes in terrain elevations were subdivided into four classes. A context unification of elevation classes’ composition for all time periods is the basic requirement for this stage. Contribution of each class

Table 10.1  The Percentage Contribution of Each Class. Class code

Elevation change, m

Percentage of total area (6.2016– 11.2016)

Percentage of totalarea,(11.2016– 2.2017)

1 2 3 4

0.9–1.5 0.6–0.9 0.3–0.6 0.15–0.3

0.04 0.14 0.61 1.45

0.92 0.79 2.69 8.52

estimated in percentage of total area is presented in Table 10.1. The resulting maps of elevation change classes within the study area are shown in Figure 10.4.

SATELLITE RADAR INTERFEROMETRY PROCESSING AND ELEVATION CHANGE ANALYSIS  133

Identified plots of subsidence are localized in areas of mining, open pits, and spoils. Multiple spots of land subsidence are detected mostly on roads leading from mines. Natural landslides along rivers are also recognized.

attempts have been made to develop a formal procedure for change values analysis and evaluation. They are based on multidimensional statistics and multivariate alteration detection (MAD) (Nori et al., 2008). The main disadvantage of such approaches is complete discrepancy between the classes semantic meaning, which, besides, can radically change when addressing different task missions. In the present case, for the evaluation of changes that have occurred, the expert knowledge of specialists in geosystem analysis and land use planning are involved. They were formalized and implemented into computer‐aided procedure through the analytic hierarchy process (AHP) by T. L. Saaty (Saaty, 2008). Based on the application

10.5.2. Elevations Change Hazard The second stage is the key milestone, most problematic for scientific justification and implementation. At this stage, the procedure involves detection, comparison, analysis, and evaluation of land cover change, which are performed for long‐term elevations change mapping. In order to identify significant changes in land cover, some (a) 33°10ʹ0ʺB

33°20ʹ0ʺB

33°30ʹ0ʺB

33°40ʹ0ʺB

48°10ʹ0ʺC

48°0ʹ0ʺC

47°50ʹ0ʺC

Study area 1 2 47°40ʹ0ʺC

3 4 0

2,5

5

10

15

20 km

Figure 10.4 Elevation change classes maps within the study area: (a) for 6.2016–11.2016 and (b) for 11.2016–2.2017.

134  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION (b) 33°10ʹ0ʺB

33°20ʹ0ʺB

33°30ʹ0ʺB

33°40ʹ0ʺB

48°10ʹ0ʺC

48°0ʹ0ʺC

47°50ʹ0ʺC

Study area 1 2 47°40ʹ0ʺC

3 4 0

2,5

5

10

15

20 km

Figure 10.4  (Continued)

content, an expert fulfils a pair‐wise comparison of all possible negative trends in terrain elevations change in the specified classes system. Then the expert submits ­priorities in the classic AHP nine‐point grading scale from equivalence up to extreme superiority and backward. The outcome is am × m size back‐symmetric matrix of both integers and multiplicative inversion fractions with unity diagonal, where m is a number of elevation change classes. The weight vector of classes w is calculated as the squared first eigenvector of the pair‐wise comparison matrix, and expert judgment consistency is estimated by consistency index (CI):

CI

m (10.1) m 1

max

where λmax is the maximum eigenvalue of the comparison matrix. In a simplified version of AHP, to compensate for slight expert inconsistency, a smoothing of the weighting factors by the CI value can be applied (Nogin, 2004). Nevertheless, the significance of elevation changes can be assessed by any evaluation method used for each of the possible pairs of before‐and‐after. It should be kept in mind that some change can have even opposite character, depending on the current issue that takes place. The procedure described above is implemented as a special software module (Table 10.2) for elevation changes mapping by radar interferometry time series. The module was developed in the open‐source SciLab environment for numerical computation (http://www.scilab.org/). As

SATELLITE RADAR INTERFEROMETRY PROCESSING AND ELEVATION CHANGE ANALYSIS  135 Table 10.2  Source Code of Software Module for Elevation Change Estimation.

// Elevation change expert estimation // SciLab v6.0.0 script © S.A. Stankevich et al., 1994-2017 clear();   function z=elchcl(x); // elevation change class if x500m 5.15–7.28 7.28–9 9–11.08 11.08–13.69 13.69–20.27 −0.78–3.54 3.54–5.01 5.01–6.39 6.39–8.13 8.13–12.66

Aspect

Roads proximity

Geomorphology

Lineament density

Geology

LULC

Thrust and fault proximity

TWI

SPI

Information value

Fij

Pij

Hj

Hjmax

Ij (SIV)

−1.37 −0.69 −0.25 0.37 1.17 0.35 0.38 0.54 1.06 1 0.18 ‐1.17 −1.26 −1.14 −1.98 −0.93 0.72 0.09 0.35 −0.3 0 0.39 0 −0.13 0.37 −0.51 −0.2 0 0.65 −0.8 −1.65 −1.24 0.21 0.98 0.4 0 0 −1.89 0 0.66 0.28 −1.2 1.26 0.63 0.62 0.38 −0.41 −1.32 −0.03 0.27 −0.51 −2 1.15 −0.26 −0.97 0.47 0.26 1.09

0.25 0.5 0.77 1.45 3.23 1.43 1.46 1.72 2.89 2.72 1.2 0.3 0.28 0.31 0.13 0.39 2.07 1.1 1.42 0.73 0 1.48 0 0.74 1.44 0.59 0.85 0 1.93 0.44 0.19 0.28 1.23 2.66 1.49 0 0 0.15 0 1.94 1.32 0.29 3.53 1.89 1.86 1.46 0.65 0.26 0.96 1.32 0.59 0.13 3.17 0.76 0.37 1.6 1.3 3

0.03 0.06 0.1 0.19 0.42 0.18 0.12 0.15 0.25 0.23 0.1 0.02 0.02 0.02 0.01 0.03 0.38 0.2 0.26 0.13 0 0.38 0 0.23 0.38 0.17 0.24 0 0.57 0.07 0.03 0.04 0.19 0.42 0.23 0 0 0.04 0 0.35 0.52 0.08 0.36 0.19 0.19 0.15 0.06 0.02 0.15 0.21 0.09 0.02 0.51 0.1 0.05 0.22 0.18 0.42

2.18

2.58

0.15

2.78

3.32

0.16

1.9

2

0.04

1.54

2.32

0.33

1.39

2

0.3

2.1

2.58

0.18

1.49

2.8

0.46

2.26

2.58

0.12

1.83

2.32

0.21

2.03

2.32

0.12

76°29ʹ0ʺE

76°34ʹ30ʺE

76°40ʹ0ʺE

76°45ʹ30ʺE

76°51ʹ0ʺE

31°16ʹ30ʺN

31°16ʹ30ʺN

31°22ʹ0ʺN

31°22ʹ0ʺN

76°23ʹ30ʺE

31°27ʹ30ʺN

PROBABILISTIC LANDSLIDE HAZARD ASSESSMENT USING SIV AND GIS TECHNIQUES  205

Legend Very low

76°23ʹ30ʺE

31°11ʹ0ʺN

31°11ʹ0ʺN

Low Moderate High Very high Testing Training 76°29ʹ0ʺE

76°34ʹ30ʺE

76°40ʹ0ʺE

76°45ʹ30ʺE

76°51ʹ0ʺE

Figure 15.4  Landslide susceptibility map generated using the SIV model.

True positive rate (sensitivity)

Area under curve = 0.7547

.75

.50 .25

slide prediction evaluated through the ROC curve (at 95% confidence level). The utmost vital factors triggering landslides in the present study area estimated for entropy values are the LULC class (1.74), which constitute mostly barren land and sparse vegetation. This was followed by other conditioning factors of TWI, lineament density, geomorphology, and slope. Hence, the SIV model can be adopted for landslide susceptibility in areas s­ imilar to the Bhanupali‐Beri area of Himachal Pradesh. ­REFERENCES

Figure 15.5  Validation of landslide susceptibility map through ROC curve.

t­ esting. Therefore, null hypothesis is rejected and alternate hypothesis is accepted (Sarkar et al., 2008) and both the models used in this study are statistically significant for landslide susceptibility assessment. 15.6. ­CONCLUSION The resultant landslide susceptibility map generated applying the SIV model shows that the areas with higher landslide susceptibility are mainly distributed along the south‐central and northeastern directions in the study area. The study also reflects that the Statistical Information Value method is an effective method of landslide susceptibility assessment and shows 75.47% accuracy in land-

Akgun, A. (2012). A comparison of landslide susceptibility maps produced by logistic regression, multi‐criteria decision, and likelihood ratio methods: a case study at İzmir, Turkey. Landslides, 9, 93–106; doi:10.1007/s10346‐011‐0283‐7 Bui, T. D., Pradhan, B., Lofman, O., & Dick, O. (2012). Landslide susceptibility assessment in the Hoa Binh Province of Vietnam: a comparison of the Levenberg‐Marquardt and Bayesian regularized neural networks. Geophysical. Journal of the Royal Astronomical Society, 171–172, 12–29. Bui, T. D., Pradhan, B., Revhaug, I., Nguyen, D. B., Pham, H. V., & Bui, Q. N. (2015a). A novel hybrid evidential belief function‐based fuzzy logic model in spatial prediction of rainfall‐induced shallow landslides in the Lang Son city area (Vietnam). Geomatics, Natural Hazards Risk, 6, 243–271; doi :10.1080/19475705.2013.843206 Bui, T. D., Tuan, T. A., Klempe, H., Pradhan, B., & Revhaug, I. (2015b). Spatial prediction models for shallow landslide ­hazards: a comparative assessment of the efficacy of support

206  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION vector machines, artificial neural networks, kernel logistic regression, and logistic model tree. Landslides; doi: 10.1007/ s10346‐015‐0557‐6 Beven, K. J., & Kirkby, M. J. (1979). A physically based, variable contributing area model of basin hydrology, Hydrological Sciences Bulletin, 24, 43–69. Bui, T. D., Tuan, T. A., Klempe, H., Pradhan, B., & Revhaug, I. (2016). Spatial prediction models for shallow landslide hazards: A comparative assessment of the efficacy of support vector machines, artificial neural networks, kernel logistic regression, and logistic model tree. Landslides, 13, 361–378. Catani, F., Lagomarsino, D., Segoni, S., & Tofani, V. (2013). Landslide susceptibility estimation by random forests technique: Sensitivity and scaling issues. Natural Hazards Earth Systems Science, 13, 2815–2831; www.nat‐hazards‐earth‐ syst‐sci.net/13/2815/2013/doi:10.5194/nhess‐13‐2815‐2013 Cerri, R. I., Reis, F. A., Gramani, M. F., Giordano, L. C., & Zaine, J. E. (2017). Landslides zonation hazard: relation between geological structures and landslides occurrence in hilly tropical regions of Brazil. Anais da Academia Brasileira de Ciências, 89(4), 2609–2623. Chen, W., Chai, H., Sun, X., Wang, Q., Ding, X., & Hong, H. (2016). A GIS‐based comparative study of frequency ratio, statistical index and weights‐of‐evidence models in landslide susceptibility mapping. Arabian Journal of Geoscience, 9, 1–16. Chen, X., Liu, C., Chang, Z., & Zhou, Q. (2015). The relationship between the slope angle and landslide size derived from limit equilibrium simulations. Geomorphology, 253; doi: 10.1016/j.geomorph.2015.01.036 Chung, C. F., & Fabbri, A.G. (2003). Validation of Spatial Prediction Models for Landslide Hazard Mapping. Natural Hazards, 30, 451–472. Constantin, M., Bednarik, M., Jurchescu, M. C., & Vlaicu, M. (2011). Landslide susceptibility assessment using the bivariate statistical analysis and the index of entropy in the Sibiciu Basin (Romania). Environmental Earth Sciences, 63, 397– 406. Das, I., Stein, A., Kerle, N., & Dadhwal, V. K. (2011a). Probabilistic landslide hazard assessment using homogeneous susceptible units (HSU) along a national highway corridor in the northern Himalayas, India. India. Landslides, 8, 293; https://doi.org/10.1007/s10346‐011‐0257‐9 Das, S. K., Biswal, R. K., Sivakugan, N., & Das, B. (2011). Classification of slopes and prediction of factor of safety using deferential evolution neural networks. Environmental Earth Science, 64, 201–210. Dehnavi, A., Aghdam, I. N., & Pradhan, B. (2015). A new hybrid model using step‐wise weight assessment ratio analysis (SWARA) technique and adaptive neuro‐fuzzy inference system (ANFIS) for regional landslide hazard assessment in Iran. Catena, 135, 122–148. Devkota, K., Regmi, A. D., Pourghasemi, H. R., & Pradhan, B. (2013). Landslide susceptibility mapping using certainty factor, index of entropy and logistic regression models in GIS and their comparison at Mugling‐Narayanghat road section in Nepal Himalaya. Natural Hazards, 65, 135–165; doi: 10.1007/s11069‐012‐0347‐6

Erener, A., Mutlu, A., & Düzgün, H. S. (2016). A comparative study for landslide susceptibility mapping using GIS‐based multi‐criteria decision analysis (MCDA), logistic regression (LR) and association rule mining (ARM). Engineering Geology, 203, 45–55. Goetz, J. N., Guthrie, R. H., & Brenning, A. (2011). Integrating physical and empirical landslide susceptibility models using generalized additive models. Geomorphology, 129, 376–386. Gorsevski, P. V., & Jankowski, P. (2010). An optimized solution of multicriteria evaluation analysis of landslide susceptibility using fuzzy sets and Kalman filter. Computers and Geosciences, 36(8), 1005–1020. Gorsevski, P. V., Brown, M. K., Panter, C., Simic, A., & Snyder, J. (2016). Landslide detection and susceptibility mapping using LiDAR and an artificial neural network approach: a case study in the Cuyahoga Valley National Park, Ohio. Landslides, 13 (3), 467–484. Guzzetti, F., Carrara, A., Cardinali, M., & Reichenbach, P. (1999). Landslide hazard evaluation: a review of current techniques and their application in a multi‐study, Central Italy. Geophysical Journal of the Royal Astronomical Society, 31, 181–216. Guzzetti, F., Reichenbach, P., Ardizzone, F., Cardinali, M., & Galli, M. (2006). Estimating the quality of landslide susceptibility models. Journal of Geomorphology, 81(1–2), 166–184. Guzzetti, F., Reichenbach, P., Cardinali, M., Galli, M., & Ardizzone, F. (2005). Probabilistic landslide hazard assessment at the basin scale. Journal of Geomorphology, 72(1–4), 272–299. Haoyuan, H., Wei, C., Chong, X., Ahmed, M., Youssef,. Pradhan, B., & Bui D.T. (2016). Rainfall‐induced landslide susceptibility assessment at the Chongren area (China) using frequency ratio, certainty factor, and index of entropy. Geocaroto International. Online at http:///www.tandfonline. com/loi/tgei20 HPSDMA (Himachal Pradesh State Disaster Management Authority) (2017). http://www.hpsdma.nic.in/ProfileOfState/ Landslide.html. Jaafari, A., Najafi, A., Pourghasemi, H., Rezaeian, J., & Sattarian, A. (2014). GIS‐based frequency ratio and index of entropy models for landslide susceptibility assessment in the Caspian forest, northern Iran. International Journal of Environmental Science and Technology, 11, 909–926. Kalantar, B., Pradhan, B., Naghibi, S. A., Motevalli, A., & Mansor, S. (2017). Assessment of the effects of training data  selection on the landslide susceptibility mapping: a comparison between support vector machine (SVM), logistic regression (LR) and artificial neural networks (ANN). Geomatics, Natural Hazards and Risk, Online. Accessed: 6 December 2017. Kwan, J. S. H., Chan, S. L., Cheuk, J. C. Y., & Koo, R. C. H. (2014). A case study on an open hillside landslide impacting on a flexible rock fall barrier at Jordan Valley, Hong Kong. Landslides, 11, 1037–1050; doi:10.1007/s10346‐013‐0461‐x Lee, S. (2005). Application of logistic regression model and its validation for landslide susceptibility mapping using GIS and remote sensing data. International Journal of Remote Sensing, 7, 1477–1491.

PROBABILISTIC LANDSLIDE HAZARD ASSESSMENT USING SIV AND GIS TECHNIQUES  207 Lee, S., & Min, K. (2001). Statistical analysis of landslide susceptibility at Yongin, Korea. Environmental Geology, 40, 1095–1113. Li, Y., Zhou, R., Zhao, G., Li, H., Su, D., Ding, H., et al. (2014). Tectonic uplift and landslides triggered by the Wenchuan earthquake and constraints on o ­ rogenic growth: a case study from Hongchun Gully, Longmen Mountains, Sichuan, China. Quaternary. International, 349, 142–152; doi:10.1016/j. quaint.2014.05.005 Ma, J., Tang, H., Hu, X., Bobet, A., Zhang, M., Zhu, T., et al. (2017). Identification of causal factors for the Majiagou landslide using modern data mining methods. Landslides, 14 (1), 311–322. Mathew, J., Jha, V. K., & Rawat, G. S. (2007). Weights of ­evidence modeling for landslide hazard zonation mapping of  Bhagirathi Valley, Uttarakhand. Current Science, 92 (5), 628–638. Mohammady, M., Pourghasemi, H. R., & Pradhan, B. (2012). Landslide susceptibility mapping at Golestan Province, Iran: a comparison between frequency ratio, Dempster–Shafer, and weights‐of evidence models. Journal of Asian Earth Sciences, 61, 221–236. Moosavi, V., & Niazi, Y. (2016). Development of hybrid wavelet packet‐statistical models (WP‐SM) for landslide susceptibility mapping. Landslides, 13, 97–114. Moore, I.D., & Wilson J.P. (1992). Length-slope factors for the Revised Universal Soil Loss Equation: Simplified method of estimation. Journal of Soil and Water Conservation, 47(5), 423–428. Onagh, M., Kumra, V. K., & Rai, P. K. (2012a). Application of Multiple Linear Regression Model in Landslide Susceptibility Zonation Mapping: The Case Study of Narmab Basin (Iran). International Journal of Geology, Earth and Environmental Sciences, 2 (2), 87–101. Onagh, M., Kumra, V. K., & Rai, P. K. (2012b). Landslide susceptibility mapping in a part of Uttarkashi district (India) by multiple linear regression method. international. Journal of Geology, Earth and Environmental Sciences, 2 (2), 102–120. Palenzuela, J. A., Marsella, M., Nardinocchi, C., Pérez, J. L., Fernández, T., Chacón, J., & Irigaray, C. (2015). Landslide detection and inventory by integrating LiDAR data in a GIS environment. Landslides, 12 (6), 1035–1050. Pham, B. T., Bui, D. T., Prakash, I., & Dholakia, M. B. (2017). Hybrid integration of Multilayer Perceptron Neural Networks and machine learning ensembles for landslide susceptibility assessment at Himalayan area (India) using GIS. Catena, 149, 52–63. Pourghasemi, H. R., Pradhan, B., & Gokceoglu, C. (2012). Application of fuzzy logic and analytical hierarchy process (AHP) to landslide susceptibility mapping at Haraz watershed, Iran. Natural Hazards, 63, 965–996. Pradhan B. (2013). A comparative study on the predictive ability of the decision tree, support vector machine and neuro-fuzzy models in landslide susceptibility mapping using GIS. Computers & Geosciences, 51, 350–365. Pradhan, B., & Lee, S. (2010). Landslide susceptibility assessment and factor effect analysis: back propagation artificial neural networks and their comparison with frequency ratio and bivariate logistic regression modelling. Environmental Modeling & Software, 25, 747–759; doi:10.1016/j.envsoft.2009.10.016

Pradhan, B., Seeni, M. I., & Kalantar, B. (2017). Performance evaluation and sensitivity analysis of expert-based, statistical, machine learning, and hybrid models for producing landslide susceptibility maps. In B. Pradhan (Ed.), Laser scanning applications in landslide assessment (pp. 193–232). Springer. Qianqian, B., Yumin, C., Susa, D., Qianjiao, W., Jiaxin, Y., & Jingyi, Z. (2017). An improved information value model based on gray clustering for landslide susceptibility ­mapping. International Journal of Geo‐Information, 6, 18. Rai, P. K., Mohan, K., & Kumra, V. K. (2014). Landslide hazard and its mapping using Remote Sensing & GIS techniques. Journal of Scientific Research, 58, 1–13. Raghuvanshi, T. K., Negassa, L., & Kala, P. M. (2015). GIS based grid overlay method versus modeling approach: A comparative study for Landslide Hazard Zonation (LHZ) in Meta Robi District of West Showa Zone in Ethiopia. Egyptian Journal of Remote Sensing and Space Sciences, 18, 235–250. Rohan, K., & Ambalagan, R. (2015). Landslide susceptibility zonation in part of Tehri reservoir region using frequency ratio, fuzzy logic and GIS. Journal of Earth System Science, 124 (2), 431–448. Sarkar, S., Kanungo, D. P., Patra, A. K., & Kumar, P. (2008). GIS based spatial data analysis for landslide susceptibility mapping. Journal of Mountain Science, 5(1), 52–62. Singh, P., & Sharma, A. (2014). Landslide Susceptibility Mapping of North Sikkim Himalaya using Geospatial Techniques. 15th ESRI User Conference, New Delhi, India. Singh, P., & Sharma, A. (2015). Probabilistic landslide susceptibility mapping using binary logistic regression model and geospatial techniques: A case study of Uttarakhand. 16th ESRI User Conference, New Delhi, India. Suzen, M.L. & Doyuran, V. (2004). Data driven bivariate landslide susceptibility assessment using Geographical ­ Information Systems: A method and application to Asarsuyu catchment, Turkey. Engineering Geology, 71, 303–321. Vahidnia, M. H., Alesheikh, A. A., Alimohammadi, A., & Hosseinali, F. (2010). A GIS‐based neuro‐fuzzy procedure for integrating knowledge and data in landslide susceptibility mapping. Computers &. Geosciences, 36, 1101–1114. Vernes, D. J. (1984). Landslide Hazard Zonation: A Review of Principles and Practice. International Association of Engineering Geology Commission on Landslides and Other Mass Movements on Slopes. International Association for Engineering Geology UNESCO, Natural Hazards Series no. 3. Wang, L.‐J., Guo, M., Sawada, K., Lin, J., & Zhang, J. (2016). A comparative study of landslide susceptibility maps using logistic regression, frequency ratio, decision tree, weights of evidence and artificial neural network. Geosciences Journal, 20, 117–136. Wilson, J. B. (2011). The twelve theories of co‐existence in plant communities: the doubtful, the important and the unexplored. Journal of Vegetation Science, 22, 184–195. Xu, C., Xu, X., Shen, L., Yao, Q., Tan, X., Kang, W., et al. (2016). Optimized volume models of earthquake‐triggered landslides. Scientific Reports, 6 (29797); doi:10.1038/ srep29797 Yalcin, A., Reis, S., Aydinoglu, A. C., & Yomralioglu, T. (2011). A GIS-based comparative study of frequency ratio, analytical

208  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION hierarchy process, bivariate statistics and logistics regression methods for landslide susceptibility mapping in Trabzon, NE Turkey; Catena, 85(3), 274–287. Youssef, A. M. (2015). Landslide susceptibility delineation in  the Ar‐Rayth area, Jizan, Kingdom of Saudi Arabia, using  analytical hierarchy process, frequency ratio, and

logistic regression models. Environmental Earth Sciences, 73, 8499–8518. Yilmaz I. (2009). Landslide susceptibility mapping using frequency ratio, logistic regression, artificial neural networks and their comparison: a case study from Kat landslides (Tokat-Turkey). Computers and Geosciences, 35.

16 One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India Dhruvesh P. Patel1, Prashant K. Srivastava2, Sudhir Kumar Singh3, Cristina Prieto4, and Dawei Han5

ABSTRACT A flood is a major disaster responsible for huge demolition and loss of properties and life due to heavy amounts of water released in a short span of time. The city of Surat in Gujarat, India, is repeatedly affected by floods among which the event of 2006 was a major one. It occurred due to a massive release of water from the Ukai dam downstream. Despite the great damage, there is little detailed analysis available. The present work is focused on the simulation of the discharge‐carrying capacity of the lower Tapi River responsible for major flooding in the city. Hydrodynamic modeling based on the 1-D Hydrologic Engineering Centers-River Analysis System (HEC‐RAS) was employed on 299 river cross-sections, including two line structures named the Kakrapar weir and Singanpur weir, and five major bridges across the Tapi River in Surat. Release of water from the Ukai dam is considered as the upstream boundary while the tidal level is considered as the downstream boundary for the 2006 flood event. The flow is simulated under unsteady flow conditions, calibrated from the year 1998 and validated for the year 2006. The simulated results indicated that the major sections situated in and around the city have less than 24,081 m3/s carrying capacity against 25,768 m3/s released from the Ukai dam in 2006. In the future, if water is released above the existing carrying capacity, similar catastrophic floods will happen. For flood mitigation and management, the sections with lower discharge carrying capacity must be expanded in order to minimize the losses from such disasters. 16.1. ­INTRODUCTION In a developing country like India, flood is a major disaster, which causes loss of life and properties. As a result, it decreases the economic growth of the country. Due to human intervention, such as construction of illegal houses, roads, and bridges in the floodplains and 1  Department of Civil Engineering, School of Technology, PDPU, Gujarat, India 2  Institute of Environment and Sustainable Development and DST-Mahamana Center for Excellence in Climate Change Research, Banaras Hindu University, Varanasi, India 3   K. Banerjee Centre of Atmospheric and Ocean Studies, University of Allahabad, Allahabad, India 4   Environmental Hydraulics Institute, Universidad de Cantabria, Parque Científico y Tecnológico de Cantabria, Santander, Spain 5  Department of Civil Engineering, University of Bristol, Bristol, UK

catchments, risk and loss of valuables are increased (Hassan et  al., 2006; Rahman et  al., 2010). Patel and Dholakia (2010a) and Patel and Srivastava (2013) discuss how floods have always been a major problem to human beings, as many settlements have grown up around rivers. Humans have been fighting floods in different ways for a long time, but while floods can not be fully controlled, their damages can be reduced by mitigation measures such as floodplain management (Parsa et  al., 2013). Timbadiya et al. (2014b) suggested that prediction of the river stage plays a vital role in structural and nonstructural measure of flood management. Horritt et al. (2007) have shown that hydrodynamic models that simulate the hydraulic behavior of river channels are effective tools in floodplain management. Accurate delineation of flood extent and depth within the floodplain are necessary for flood management and mitigation and to make accurate decisions regarding

Techniques for Disaster Risk Management and Mitigation, First Edition. Edited by Prashant K. Srivastava, Sudhir Kumar Singh, U. C. Mohanty, and Tad Murty. © 2020 John Wiley & Sons, Inc. Published 2020 by John Wiley & Sons, Inc. 209

210  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

construction and urban development (Noman et  al., 2003; Patel & Srivastava, 2013; Reza Ghanbarpour et al., 2011; Salimi et  al., 2008). Several researchers simulated accurate prediction of river flow and hydraulic behavior of river channels by a number of hydraulic models such as MIKE  11, 1-D HEC‐RAS, INFOWORK, FLOW‐2D, HSPF, UNET, WMS,LISFLOOD‐FP, TELEMAC‐2D, and RiverCAD (Ali et al., 2012; Bates et al., 2000; Bellos & Tsakiris, 2015; Castellarin et al., 2009; Horritt & Bates, 2002; Parsa et al., 2013; Reza Ghanbarpour et al., 2011; Timbadiya et al., 2011a; Tsakiris & Bellos, 2015). Out of all these models, the Hydrological Engineering Centers River Analysis System (HEC‐RAS) is widely used worldwide and is public domain software. HEC‐RAS was developed by the US Army Corps of Engineers, and allows one to perform one‐dimensional steady and unsteady river flow hydraulic calculations, sediment transport‐mobile bed modeling, and water temperature analysis (Brunner, 2008, 2010; Hicks & Peacock, 2005; Horritt & Bates, 2002). The HEC‐RAS model represents the terrain as a sequence of cross-sections and simulates flow to estimate the average velocity and water depth at each cross-section (Parsa et  al., 2013). Johnson et  al. (1999) used the HEC‐RAS model to predict and define desirable lands within 10 km of the Greybull River, Wyoming, USA, in the Bighorn basin. Tate and Maidment (1999) conducted a study to integrate HEC‐ RAS and Arcview software in order to study the privacy bed of Waller Creek in Austin, Texas. Sadeghi and Rad (2004) simulated flood risk in urban areas of Darabad, Tehran, by using the HEC‐RAS model with mapping of the cross-sections. They showed the ability of the model to simulate the water surface in different elevations. Nandalal (2009) reported a framework for flood modeling by deploying the HEC‐RAS model on the San Antonio River over 10,000 square kilometers. The flood water levels along the 79 km long Kalu River in Sri Lanka were simulated using the 1-D HEC‐RAS hydrodynamic model to reduce flood damages (Nandalal, 2009; Timbadiya et al., 2014b). Remo et al. (2012) have simulated a part of the Mississippi River by using the HEC‐RAS model, while Lee et  al. (2006) have determined the influence of bridge blockage and the Ba‐Tu overbank flow on the water stages in the Keelung River during Typhoon Nari using HEC‐RAS in an unsteady flow condition. In the case study of the Linggi River in Seremban Town, Malaysia, to calibrate and validate the results HEC‐2 (HEC‐RAS’s predecessor) was used and showed that the absolute error in predicted water surface levels was within 5% of the observed level (Said et  al., 2002). Villazón Gómez (2011) and Villiazon et al. (2009) found that HEC‐RAS presents results very close to other

­orldwide commercial modeling packages. The disw charge, river stage, and other hydraulic properties are interrelated and depend upon the characteristics of channel roughness and geometry (Timbadiya et  al., 2011b). Estimation of the channel roughness parameter is of key importance in the study of open‐channel flow, particularly in hydraulic modeling (Timbadiya et  al., 2011b). The channel roughness is not a constant parameter, and it varies along the river depending upon variation in channel characteristic along the flow, such as surface roughness, vegetation, channel irregularities, and channel alignment (Pappenberger et  al., 2005). Several researchers including Parhi et  al. (2012), Ramesh et  al. (2000), Timbadiya et  al. (2011b), and Wasantha Lal (1995) have calibrated channel roughness for different rivers for the development of the hydraulic model. Surat city in Gujarat experienced the devastating flood in 2006. It is estimated that the single flood event during 7–14 August 2006, in Surat and its twin city Hazira, resulted in 300 people dead and property damage worth INR 210 billion (Patel & Srivastava, 2013). Parmar and Rao (2002) and Patel and Dholakia (2010a) pointed out that reduction in the discharge carrying capacity and encroachment in the river floodplain are some of the parameters that have caused floods and thus significant loss of property as well as human lives in lower Tapi basin (LTB). Prediction of  accurate submergence, flood forecasting, flood management, and future protection from flood is highly dependent on the river stage as well as flow simulation during the flooding. Timbadiya et al. (2014a) carried out the work on 190 cross-sections of lower Tapi River with MIKE 11 and derived stage‐discharge curves for various sections. It has been observed that the existing literature lacks studies on maximum discharge carrying capacity of river sections, which results in the inability to establish a proper early warning level for a city like Surat. The information for danger and warning level can help the authorities to discharge excessive water from the dam up to the permissible limit of the cross-sections without spill. To fill this relevant knowledge gap, this study is focused on river hydraulic analysis and flow simulation of the Tapi River extending from the Ukai dam to the Arabian Sea using HEC‐RAS. Spatially located maps for the river sections with maximum carrying capacity around Surat city are generated after a rigorous sensitivity analysis and validation. This study provides strong supportive evidence of the potential value of HEC‐RAS for flood modeling. The  assessment of HEC‐RAS with respect to this particular aspect is an important step for successful and improved development of hydrological models, and thus can provide important assistance in building flood ­mitigation strategies.

One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India  211

16.2. ­STUDY AREA AND DATA USED

16.2.2. Lower Tapi River (LTR) The length of the river from the Ukai dam to the Arabian Sea is considered to be the lower Tapi River (LTR) (Figure 16.2), which is estimated as 122 km. The river reach is divided into 299 cross-sections. Detailed cross-sections of Tapi River, showing bed and bank RL (Reducing Level) at an average interval of 150 m to 200 m, were collected from the Surat Municipal Corporation (SMC) and Surat Irrigation Circle (SIC), Government of Gujarat, India, in AutoCAD format for just after the 2006 flood. The survey was carried out in two phases: first, from Singanpur weir to the Ukai dam for river sections L6A‐R6A to L201‐R201 with the chainage of 100.490 km; and, second, from Singanpur weir to the Arabian Sea for section LD1‐RD1 to LD85 to RD 85 with the chainage of 21.635 km. Symbols used for right bank sections are R or RD and for left bank sections L or LD. The lower Tapi River consists of two inline structures named the Kakrapar weir and Singanpur weir as well as five major bridges across the river at Surat city. Four major river gauge–discharge stations named

16.2.1. Tapi Basin and Surat District The Tapi basin is the northern most basin of the Deccan plateau and is situated between latitude 20–22°N and longitude 72–78°E, approximately (CWC 2000–2001; Patel & Srivastava, 2013). The Tapi basin is divided into three zones: upper Tapi basin (UTB), middle Tapi basin (MTB), and lower Tapi basin (LTB) (Patel & Dholakia, 2010a; Patel & Dholakia, 2010b; Patel et al., 2012b; Patel & Srivastava, 2013) (Figures 16.1 and 16.2). The river has a total length of 724 km, out of which the last section of 214 km is in Gujarat state. It meets the Arabian Sea in the Gulf of Cambay approximately at 19.2 km west of Surat city. The Tapi River covers an area of approximately 3,837 km2 in Gujarat state. Surat city is situated on the banks of the Tapi and 100 km downstream (D/S) of the Ukai dam (Patel & Srivastava, 2013), bounded by latitude 21°06″ to 21°15″ N and longitude 72°45″ to 72°54″ E (Figure  16.2). In the ­history of floods at Surat, the 2006 flood was reported as the most disastrous one (Patel & Srivastava, 2013). 70°0ʹE

80°0ʹE

90°0ʹE

73°0ʹ0ʺE

N W

74°0ʹ0ʺE

75°0ʹ0ʺE

76°0ʹ0ʺE

E

S

22°0ʹ0ʺN

40°0ʹN

78°0ʹ0ʺE

E

W

S

77°0ʹ0ʺE

Lower Tapi basin (LTB)

N

22°0ʹ0ʺN Ukai Dam

40°0ʹN 21°0ʹ0ʺN

21°0ʹ0ʺN

20°0ʹ0ʺN 20°0ʹ0ʺN

30°0ʹN

0 25 50

100 150 200 Kilometers

30°0ʹN 73°0ʹ0ʺE

74°0ʹ0ʺE

72°40ʹ0ʺE 72°50ʹ0ʺE

20°0ʹN

76°0ʹ0ʺE

73°10ʹ0ʺE

77°0ʹ0ʺE

73°20ʹ0ʺE

78°0ʹ0ʺE

73°30ʹ0ʺE

Lower Tapi River (LTR) and Surat city

N 21°30ʹ0ʺN

75°0ʹ0ʺE

73°0ʹ0ʺE

W

E 21°30ʹ0ʺN

S

20°0ʹN

21°20ʹ0ʺN

21°20ʹ0ʺN

10°0ʹN

10°0ʹN

21°10ʹ0ʺN

Surat city with its zone boundary 0

287.5 575

1,150

1,725

2,300 Kilometers

21°0ʹ0ʺN

21°0ʹ0ʺN

0

5 10

20

0°0ʹ

0°0ʹ

30

40 Kilometers

20°50ʹ0ʺN

70°0ʹE

80°0ʹE

90°0ʹE

20°50ʹ0ʺN 72°40ʹ0ʺE

72°50ʹ0ʺE

73°0ʹ0ʺE

73°10ʹ0ʺE

73°20ʹ0ʺE

Figure 16.1  Location of the Tapi basin, lower Tapi basin, and lower Tapi River with Surat city.

73°30ʹ0ʺE

212  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION 72°50′0″E

72°40′0″E

73°10′0″E

73°0′0″E

73°20′0″E

73°30′0″E

N W 21°30′0″N

E S

21°30′0″

116.5 Kapodra to Utrva Bridge Ukai Dam U/S boundary end 110.5 Amroli Bridge 21°20′0″N

91.5 Daboli-Jahagirpur Bridge

21°20′0″ Ghala

55.5 Swamivivekanand Bridge

Kakrapar weir Singanpur weir Hope Bridge

112.5 Railway Bridge

21°10′0″N

122.5 Nana Varaccha to Mota Varaccha 21°10′0″

58.5 Nehru (Hope) Bridge

Legend Gauge-discharge stations

51.5 Sardar Vallabhai Patel (SVP) Bridge 21°0′0″N

Tapi River River sections

21°0′0″N

Arabian Sea D/S boundary end 0

5

10

20

72°50′0″E

72°40′0″E

30

40 Kilometers

73°0′0″E

73°10′0″E

73°20′0″E

73°30′0″E

Figure 16.2  Lower Tapi River with inline structure, gauge‐discharge station, and bridges.

Kakrapar weir, Ghala, Singanpur weir, and Hope Bridge are situated on the LTR. The Ghala station is ­monitored by the Central Water Commission (CWC) and the rest are monitored by the SIC. The observed data ­collected from respective departments are used for calibration and validation purposes. 16.2.3. Inline Structures 16.2.3.1. Kakrapar Weir Kakrapar weir is situated 24.46 km D/S from the Ukai dam and is built for irrigation purposes for the nearby regions. It is an ungated ogee crested weir constructed in 1954. The structural parameters of the weir are the bed RL (33.3 m), crest RL (47.78 m), spillway approach height (48.78m), high flood level (HFL) (56.39 m), and top of dam RL (48.78 m). It is 621 m long with a catchment area of about 62,801 km2 (Patel & Srivastava, 2013). 16.2.3.2. Singanpur Weir Singanpur causeway cum weir is situated at 100.55 km away from Ukai dam and 76.09 km from Kakrapar weir. It

was constructed in 1995 for flood control purposes and is gated with an under sluice way, consisting of 16 spans each 6.20 m length. The structural parameters of the weir are the bed RL (0.4 m), top of crest RL (6 m), and FRL (5 m). The length of the bridge is 580 m, of which 98 m is the gated portion and the remaining is ungated. The length of the retaining wall is 80 m and 100 m for  the  left and upper sides, respectively (https://www.suratmunicipal.gov.in/Bridgecell/). The foundation of the bridge is open integrated and constructed by reinforced concrete abutments and piers. 16.2.4. Gauge and Discharge Station 16.2.4.1. Kakrapar Weir Kakrapar gauge station is situated at 24.46 km downstream of the Ukai dam and monitored by the Surat Irrigation Circle (SIC). The gauge station is located at ­latitude 21°16′ N and longitude 73°22′ E. The zero gauge level of Kakrapar station is 39.30 m as per Great Triangular Survey‐Reducing Level (GTS‐RL) while the maximum water level observed is 55.68 m for the corresponding discharge of 25,705 m3/s.

One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India  213

16.2.4.2. Ghala Ghala gauge‐discharge station is situated 64.37 km D/S of the Ukai dam and is monitored by the CWC. The station is located at latitude 21°17″ N and longitude 73° 01″ E. The zero gauge level of Ghala station is 1.87 m as per GTS‐RL, while maximum water level observed was 21.24 m for the corresponding discharge of 24,803 m3/s. 16.2.4.3. Singanpur Weir Singanpur gauge station is situated 100.55 km downstream of the Ukai dam and monitored by the SIC. The station is located at latitude 21°13″ N and longitude 72°48″ E. The zero gauge level of Ghala station is 0.4 m as per GTS‐RL while maximum water level observed was 15.42 m for the corresponding discharge of 24,649 m3/s. 16.2.4.4. Hope Bridge After discharging the water from Singanpur, it is gauged by the SIC at Hope Bridge. Hope Bridge is located at Surat‐Olpad‐Sahol Road, 103.328 km downstream of the Ukai dam, which is designed for a high flood level (HFL) (GTS‐RL +11.515 m) (Patel & Srivastava, 2013). Based on the gauged data at Hope Bridge, the safe and danger level for Surat city is decided. Before the 2006 flood event, the prefix warning level at Hope Bridge was 8.0 m for the corresponding discharges of 11,328 m3/s (400,000 ft3/s) while the maximum 12.5 m water level was observed with the corresponding discharges of 25,768 m3/s in 2006 flood.

16.2.5. Bridges Eight major bridges are situated across the Tapi River in and around Surat city (Figure  16.2). The details of bridge ID, bridge name, distance from the downstream end, number of piers, center line spacing, measured low cord, simulated flood peak, and discharge for the 2006 flood are presented in Table 16.1. 16.3. ­METHODOLOGIES This study is focused on the development of 1-D river hydrodynamic modeling of the lower Tapi River. The main steps included in the methodology are (1) 1-D HEC‐RAS model, (2) geometry and cross-sections, (3) bridges and inline structure, (4) boundary condition, and (5) calibration of HEC‐RAS for Manning’s roughness coefficient n. 16.3.1. 1-D HEC‐RAS Model HEC‐RAS is a one‐dimensional, water surface profiling application developed by the US Army Corps of Engineers (USACE) Hydraulic Engineering Center. HEC‐RAS is a hydraulic model that is composed of three 1-D hydraulic examination modules designed for (1) steady‐flow water surface profiles, (2) unsteady flow simulation, and (3) sediment transport, movable boundary computations (Lee et  al., 2006). In this study, the unsteady, gradually varied flow simulation function of HEC‐RAS is used, which depends on finite difference

Table 16.1  Bridge ID and Other Details.

ID 51.5

Bridge and inline structure name

Pier Simulated flood‐ Simulated flood‐ Year of Distance to No of centerline Bridge low peak stage for peak discharge construction D/S (m) piers spacing (m) chord (m) 2006 (m) for 2006 (m3/s) 17,798.19

14

50 * 14

11.400

12.42

24,368.50

18,561.82

24

25 * 25

13.900

14.08

24,372.67

18,954.01

12

11.515

13.46

24,374.22

January 2011

21,727.72

30

50.6 * 10 44.50 * 2 8.50 * 1 34.70 * 31

15.750

16.18

24,430.69

July 2010

30,964.71

15

16.250

16.69

24,529.13

112.5 Railway Bridge 116.5 Kapodra to Utrva Bridge

1915 May 2012

31,791.76 33,710.10

10 16

17.000 16.505

16.77 17.70

24,532.71 24,544.55

122.5 Nana Varaccha to Mota Varaccha Bridge

May 2001

36,337.93

13

15.000

17.72

24,565.07

55.5 58.5

Sardar Vallabhai 1991 Patel (SVP) Bridge Swami Vivekanand 1996 Bridge Nehru Bridge 1996

91.5

Daboli‐Jahagirpur Bridge 110.5 Amroli (New) Bridge

50.6 * 12 20 * 2 8*1 57 * 10 50 * 12 25 * 3 30 * 1 41.3 * 2 42 * 11

214  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

solutions of the Saint‐Venant equations  (16.1)–(16.2) (Timbadiya et al., 2011b). A t





Q t

Q2 / A x

gA

Q x

0 (16.1)

H x

gA So S f

0 (16.2)

Here A = cross‐sectional area normal to the flow; Q = discharge; g = acceleration due to gravity; H = elevation of the water surface above a specified datum, also called stage; So = bed slope; Sf = energy slope; t = temporal coordinate; and x = longitudinal coordinate (Timbadiya et al., 2011b). The method is based on the energy relationship that starts the calculations from one end of the range (supercritical flow at upstream to subcritical flow at downstream) and then continues the calculation from this section to the next one (Parsa et al., 2013). 16.3.2. Geometry and Cross-section For conducting flow simulation through HEC‐RAS, channel geometry, boundary conditions, and channel resistance are required. The river reach is 122 km and runs from the Ukai dam to the Arabian Sea and is segmented in the river cross-sections 150–200 m apart. The sections were surveyed by Chetan Engineers survey and mapping consultants in May 2007 and handed over to the SMC. The Tapi River steeply falls 29 m in between sections L 154‐R 154 and L 62‐R 62 (Kakrapar weir to Dhoran Pardi Village); beyond the section L 62‐R 62, the river falls gradually from 5–10 m near Kamrej, Kathor, Varacha, Amroli, and Singanpur Villages.

9

Observed tide height (m)

8

16.3.3. Bridges and Inline Structures Kakrapar and Singanpur weir inline structures and eight bridges located across the Tapi River are considered for further modeling (Table  16.1). These structures obstruct the flow passing through the river and hence the stage and discharge of the lower Tapi River is influenced by it. The Manning roughness coefficient ­varies through the action of the gates and piers. The simulated flow and Manning’s coefficient are cited in the next section. 16.3.4. Boundary Condition The boundary conditions are consistent with the nature of the river that requires predicting the flow characteristics along the range of the river. The boundary conditions represent the input and output states from the upstream range under the study. It is obvious that if the number of measurement stations in the range under study is increased, the accuracy of the results will be increased. In the present study, the release from the Ukai dam in 2006 (flood hydrograph) (Patel & Srivastava 2013) and tidal level in the sea (Figure 16.3) are considered for the upstream‐downstream boundary conditions for unsteady flow simulation of 1-D HEC‐RAS hydrodynamic modeling. 16.3.5. Calibration of HEC‐RAs for Manning’s Roughness Coefficient n The friction parameters have been considered as the form of Manning’s roughness coefficient (n). The data pertaining to the flood for the year 1998 have been used for calibration of Manning’s roughness coefficient n. The trial and error method is used for the 299 cross-sections to find the relation between the observed and simulated

Hazira inner

Hazira outer

7 6 5 4 3 2 1 0 1.8.2006 4.8.2006

7.8.2006 10.8.2006 13.8.2006 16.8.2006 19.8.2006 22.8.2006 25.8.2006 28.8.2006 31.8.2006

Time (hour)

Figure 16.3  Tidal cycle during the flood in 2006.

One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India  215 Table 16.2 Manning’s n Values and Root Mean Square Error (RMSE) (m) for the Floods of 1998 and 2006. Stream gauging station RMSE (m) Simulation

Year

Calibration

1998

Validation

2006

Manning’s n value

Kakrapar weir

Ghala

Singanpur weir

Nehru Bridge

n = 0.035 n = 0.032 n = 0.030 n = 0.025 n = 0.032 − 0.025

1.0902 1.0914 1.0914 1.1314 1.0823

1.8876 0.7960 1.0962 1.0987 –

2.4724 1.9089 2.0714 1.9286 –

2.5912 1.2851 1.8616 1.2034 1.5106

flows. For the 1998 flood, the flow hydrograph (the upstream boundary condition) and tidal level (the downstream boundary condition) are considered for flow ­simulation, including the Kakrapar and Singanpur weir inline structures and the Nehru Bridge, Swami Vivekanand Bridge, Sardar Vallabhai Bridge, and Railway Bridge. The calibrated results and RMSE in different Manning’s values are shown in Table 16.2. For the 2006 flood, the flow hydrograph (the upstream boundary condition) and tidal level (the downstream boundary condition) are considered for flow simulation including two inline structures and five major bridges constructed before the 2006 flood (Table 16.1). The validated results and RMSE for different Manning’s values are shown in Table 16.2. The root mean squared error (RMSE) is calculated for the gauge stations at Kakrapar, Ghala, Singanpur, and Nehru (Hope) Bridge for the flood in 1998 (Table 16.2, Figure 16.4 a, b, c, d). n



RMSE

i 1

X obs

X model

n

2

(16.3)

where Xobs is observed values, Xmodel is modeled values at time/place i, and n′ is the number of observations. From Table  16.2 and Figure  16.4 a, b, c, d, it can be quantified that the simulated results are closer to the observed stages for Manning’s roughness, 0.032 for Ukai to Kakrapar weir and 0.025 for Kakrapar to Arabian Sea. According to Manning’s n, further simulation of the flood in 2006 is carried out. 16.4. ­RESULTS To simulate the 2006 flood, the calibrated HEC‐RAS model was used. The flow hydrograph and tidal level have been considered for the upstream and downstream boundary conditions, including two inline structures and five major bridges. The detail profile of the simulated 2006 flood is shown in Figure  16.6. It is observed that 25,736 m3/s water was released on 9 August 2006 from the

Ukai dam. Initially, the water was retained through the Kakrapar inline structure and later it went further downstream. It was simulated that the discharged water fell nearly 60 m in elevation to 30 m downstream at the Kakrapar weir. After the Kakrapar weir, the bed of the River Tapi steeply falls by 29 m between sections L 154‐R 154 to L 62‐R 62 (Kakrapar weir to Dhoran Pardi Village) and beyond ­sections L 62‐R 62, the river falls gradually from 5 m to 10 m near Kamrej, Kathor, Varacha, Amroli, and Singanpur villages up to the Arabian Sea. The values obtained for the slope and velocity of 299 cross-sections of the River Tapi are plotted in Figure 16.5, and the power tread line is the best fit for the graph. It is observed that at 0.0002 slope value, the velocity is roughly 3 m/s while for the 0.0008 slope value, it is around 4.5 m/s. This clearly illustrates that when the slope increases, the velocity also increases and vice versa. Using these data, the general velocity equation is derived for the Tapi River and a value of R2 is achieved as 0.912, which shows a good correlation. The graph of sections versus maximum ­ discharge is plotted, where the maximum discharge is represented when the water surface touches the lowest bank RL. The graph shows that the maximum discharge carrying capacity is of section L179‐R179, 25,951 m3/s, whereas the lowest one is 5,341 m3/s for section RD30‐LD30 (Figure  16.6). The determined carrying capacities of these sections are then segregated and categorized as: Category A: Less than 8,500 m3/s (9.0 lakh cusecs) About 11.37% (34) of sections of the total (299) sections have a carrying capacity less than 8,500 m3/s, that is, 3.0 lakh cusecs (Table 16.3, Figure 16.6). The maximum

216  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

discharge carrying capacity of 5,341 m3/s is found at section RD30‐LD30. The sections RD30‐LD‐30 to RD‐33 to LD‐33 fall near the Nehru Bridge to the Sardar Vallabhai Bridge, the center portion of Surat city, having the discharge carrying capacity varying from 5,341 to 5,348 m3/s. It is observed that the central zone, west zone, and southwest zone were badly affected in the 2006 flood

(Patel & Srivastava, 2013). Most of the sections under these groups are near Hope Bridge, Mrutunjaya Mahadev Temple, Dayagi Gunj, and Palanpur Village and were severely affected in the 2006 flood (Table  16.3, Figure 16.6). About 20% (61) of the sections have carrying capacities between 8,500 m3/s and 12,740 m3/s and are shown in

(a) 25000 Simulated (N 0.035)

Simulated (N 0.025)

Simulated (N 0.030)

Kakrapar (observed)

20000

Simulated (N 0.022)

Discharge (m3/s)

15000

10000

5000

–5000

4/9/98*24 15/9/98*2 15/9/98*4 15/9/98*6 15/9/98*8 15/9/98*10 15/9/98*12 15/9/98*14 15/9/98*16 15/9/98*18 15/9/98*20 15/9/98*22 15/9/98*24 16/9/98*2 16/9/98*4 16/9/98*6 16/9/98*8 16/9/98*10 16/9/98*12 16/9/98*14 16/9/98*16 16/9/98*18 16/9/98*20 16/9/98*22 16/9/98*24 17/9/98*2 17/9/98*4 17/9/98*6 17/9/98*8 17/9/98*10 17/9/98*12 17/9/98*14 17/9/98*16 17/9/98*18 17/9/98*20 17/9/98*22 17/9/98*24

0

Time (hour)

(b) 25.00

Water level (m)

20.00

15.00

10.00

5.00

17/9/98*24

17/9/98*18

17/9/98*21

17/9/98*12

17/9/98*15

17/9/98*6

17/9/98*9

17/9/98*3

Simulated (N 0.025)

16/9/98*24

16/9/98*21

16/9/98*15

16/9/98*18

16/9/98*9

16/9/98*12

16/9/98*3

16/9/98*6

15/9/98*24

15/9/98*18

Simulated (N 0.030)

15/9/98*21

Simulated (N 0.022)

15/9/98*12

Simulated (N 0.035)

15/9/98*15

15/9/98*6

15/9/98*9

15/9/98*3

14/9/98*24

0.00

Ghala observed

Time (hour)

Figure 16.4  (a) Observed and simulated flood hydrograph at Kakrapar weir (1998), (b) observed and simulated stage at Ghala (1998), (c) observed and simulated stage at Singanpur weir (1998), (d) observed and simulated stage at Nehru (Hope) Bridge (1998).

One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India  217 (c) Singanpur (observed)

Simulated (N 0.035)

Simulated (N 0.032)

Simulated (N 0.030)

Simulated (N 0.025)

18.00 16.00

Water level (m)

14.00 12.00 10.00 8.00 6.00 4.00 2.00 17/9/98*24

17/9/98*18

17/9/98*21

17/9/98*15

17/9/98*9

17/9/98*12

17/9/98*6

17/9/98*3

16/9/98*21

16/9/98*24

16/9/98*15

16/9/98*18

16/9/98*9

16/9/98*12

16/9/98*3

16/9/98*6

15/9/98*21

15/9/98*24

15/9/98*18

15/9/98*15

15/9/98*9

15/9/98*12

15/9/98*6

15/9/98*3

14/9/98*24

0.00

Time (hour)

(d) Nehru (Hope) Bridge (observed)

Simulated (N 0.035)

Simulated (N 0.022)

Simulated (N 0.030)

Simulated (N 0.025)

18 16 14

Water level (m)

12 10 8 6 4 2

–2 –4

14/9/98*24 15/9/98*2 15/9/98*4 15/9/98*6 15/9/98*8 15/9/98*10 15/9/98*12 15/9/98*14 15/9/98*16 15/9/98*18 15/9/98*20 15/9/98*22 15/9/98*24 15/9/98*2 16/9/98*4 16/9/98*6 16/9/98*8 16/9/98*10 16/9/98*12 16/9/98*14 16/9/98*16 16/9/98*18 16/9/98*20 16/9/98*22 16/9/98*24 17/9/98*2 17/9/98*4 17/9/98*6 17/9/98*8 17/9/98*10 17/9/98*12 17/9/98*14 17/9/98*16 17/9/98*18 17/9/98*20 17/9/98*22 17/9/98*24

0

Time (hour)

Figure 16.4  (Continued)

Figure  16.6 and Table  16.4. The sections RD59-LD59, RD60-LD60, and RD 62-LD 62, which are located downstream of the Tapi River near Magdalla Bridge, have a water carrying capacity of 8,797 m3/s. It shows minimum ­discharge carrying capacity and falls surround-

ing the Nilam Society, Nandi Park Society, and SVNIT campus. The larget sections in this category are RD81‐ LD81, RD 83‐LD83, and RD 85‐LD85, which have the water carrying capacity of about 12,379 m3/s, located near CK Pithawalla College of Engineering, Dumas

218  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION 9 y = 62.37×0.352 R2 = 0.912

8 7

Velocity (m/s)

6 5 4 3 2 1 0 0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

0.0014

0.0016

0.0018

0.002

Slope (m/m)

Figure 16.5  Velocity‐slope curve lower Tapi River.

Table 16.3  Q Less Than 8,500 m3/s (>3.0 lakh cusecs). Sections RD30‐LD30 RD31‐LD31 RD32‐LD32 RD33‐LD33 RD35‐LD35 RD22‐LD22 RD38‐LD38 RD44‐LD44 RD82‐LD82 RD2‐LD2 RD79‐LD79 RD46‐LD46 RD51‐LD51 RD47‐LD47 RD41‐LD41 RD36‐LD36 RD49‐LD49 RD48‐LD48 RD78‐LD78 RD61‐LD61 RD58‐LD58 RD55‐LD55 RD54‐LD54 RD52‐LD52 RD77‐LD77 RD66‐LD66 RD65‐LD65

Q (m3/s)

Q (cusecs)

5,341.86 5,342.85 5,345.99 5,348.21 5,350.14 5,352.46 5,356.29 5,369.84 5,404.37 5,407.96 5,419.31 5,632.25 5,816.59 5,824.18 5,835.29 5,844.23 6,528.36 6,532.31 6,782.12 7,024.01 7,045.25 7,072.35 7,079.40 7,096.56 7,545.14 7,743.94 7,752.15

188,647.8 188,682.7 188,793.6 188,872.0 188,940.2 189,022.1 189,157.4 189,635.9 190,855.3 190,982.1 191,382.9 198,902.9 205,412.9 205,680.9 206,073.3 206,389.0 230,549.0 230,688.5 239,510.6 248,052.9 248,803.0 249,760.0 250,009.0 250,615.0 266,456.6 273,477.2 273,767.2

Sections

Q (m3/s)

Q (cusecs)

RD64‐LD64 RD63‐LD63 RD56‐LD56 RD53‐LD53 RD50‐LD50 RD40‐LD40 RD34‐LD34

7,760.54 7,769.86 7,840.26 7,871.68 7,899.69 7,971.27 8,009.25

274,063.5 274,392.6 276,878.8 277,988.4 278,977.6 281,505.4 282,846.7

Total: 11.37%

30000

25000

15000

Discharge (m3/s)

20000

10000

50000

RD83-LD83 RD75-LD75 RD67-LD67 RD59-LD59 RD51-LD51 RD43-LD43 RD35-LD35 RD27-LD27 RD19-LD19 RD11-LD11 RD3-LD3 L8A-R8A L13-R13 L21-R21 L27-R27 L34A-R34A L37-R37 L45-R45 L53-R53 L61-R61 L68-R68 L76-R76 L82-R82 L90-R90 L97A-R97A L105-R105 L111-R111 L118A-R118A L126-R126 L133-R133 L141-R141 L149-R149 L161-R161 L169-R169 L177-R177 L185-R185 L193-R193 L201-R201

0

Section (Ukai Dam to the Arabian Sea)

Figure 16.6  Simulated maximum discharge carrying capacity of sections from the Ukai Dam to the Arabian Sea. Table 16.4  Q Between 8,500 and 12,740 m3/s (3.0–4.5 lakh cusecs). Sections

Q (m3/s)

Q (cusecs)

Sections

Q (m3/s)

Q (cusecs)

Sections

Q (m3/s)

Q (cusecs)

RD76‐LD76 RD69‐LD69 RD62‐LD62 RD60‐LD60 RD59‐LD59 L188‐R188 RD45‐LD45 RD43‐LD43 RD42‐LD42 RD37‐LD37 L175‐R175 L167‐R167 L166‐R166 L163‐R163 L162‐R162 L160‐R160 L159‐R159 RD3‐LD3 L157‐R157 L150‐R150 L128‐R128 L125‐R125 RD26‐LD26 RD27‐LD27 RD29‐LD29 RD28‐LD28 RD25‐LD25

8,584.25 8,725.41 8,797.41 8,818.98 8,833.03 8,976.34 8,997.86 9,010.45 9,026.58 9,070.89 9,241.96 9,367.08 9,378.75 9,423.90 9,437.15 9,467.17 9,481.26 9,496.47 9,516.77 9,527.07 9,665.06 9,667.26 9,668.99 9,668.99 9,669.75 9,670.01 9,671.20

303,152.8 308,137.9 310,680.5 311,442.3 311,938.5 316,999.4 317,759.4 318,204.0 318,773.7 320,338.5 326,379.8 330,798.4 331,210.6 332,805.0 333,273.0 334,333.1 334,830.7 335,367.8 336,084.7 336,448.5 341,321.6 341,399.3 341,460.4 341,460.4 341,487.2 341,496.4 341,538.4

L118‐R118 RD24‐LD24 RD23‐LD23 L117‐R117 RD21‐LD21 RD19‐LD19 L126‐R126 RD1‐LD1 L129‐R129 L30‐R30 L27‐R27 L26‐R26 RD18‐LD18 L28‐R28 L22‐R22 L22A‐R22A L22B‐R22B RD39‐LD39 L39‐R39 L168‐R168 L169‐R169 L6A‐R6A L7A‐R7A RD73‐LD73 L11A‐R11A RD67‐LD67 RD57‐LD57

9,672.76 9,673.55 9,675.91 9,680.35 9,680.39 9,686.63 9,687.56 9,731.28 9,775.85 9,787.01 9,876.65 9,879.21 9,924.96 9,938.44 10,367.75 10,377.27 10,380.17 10,457.12 10,506.13 10,546.91 10,569.74 10,661.64 11,447.94 11,553.76 11,581.84 11,628.21 11,800.22

341,593.5 341,621.4 341,704.8 341,861.6 341,863.0 342,083.3 342,116.2 343,660.2 345,234.1 345,628.3 348,793.9 348,884.3 350,500.0 350,976.0 366,137.1 366,473.3 366,575.7 369,293.2 371,024.0 372,464.1 373,270.4 376,515.8 404,284.0 408,021.0 409,012.7 410,650.2 416,724.8

RD85‐LD85 RD83‐LD83 L156‐R156 RD81‐LD81 L121‐R121 LHD8‐LD80 L164‐R164

11,972.21 12,181.58 12,238.62 12,379.41 12,411.61 12,496.69 12,570.95

422,798.6 430,192.5 432,206.9 437,178.9 438,316 441,320.6 443,943.1

Total: 20.40%

220  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Village, Siddhnath Mahadev Temple, and Sultanbad Village at the mouth of the river. About 7% (22) of the sections have a carrying capacity range between 12,740 m3/s and 17,000 m3/s (Table 16.5, Figure 16.6). In this category, cross-section L176‐R176 has the minimum carrying capacity while L72‐R72 and L73‐L73 have the maximum water carrying capacity. The sections RD70‐LD70 to RD75‐ LD 75, near to ISCON mall, Rahul Raj Mall, Dumas Road can safely discharge water up to 15,077m3/s while section L73‐R73 has a maximum water carrying capacity up to 16,795m3/s upstream of Singanpur weir near Ghala Village (Figure 16.6). About 11% (33) of the sections have the capacity to lead the discharge from 17,000 m3/s to 21,230 m3/s. In this category, L174‐R174, L177‐R‐177, L178‐R178 are narrow, steep bed slope sections located upstream of Kakrapar weir with a discharge carrying capacity at about 19,911 m3/s (Table 16.6, Figure 16.6) and L10 A‐R 10A, L12‐R12, L17‐R17, and L18‐R18 are comparatively wider having moderate discharge carrying capacity of 18,429–20,340 m3/s. Most of these sections are located near the ONGC intake well, ESSAR intake well, river front road, and Ved Village. About 45% (135) of the sections can discharge water more than 21,230 m3/s (Figure 16.6). The major river sections from L52‐R52 to L135‐R135 have a discharge Table 16.5  Q Between 12,740 and  17,000 m3/s (4.5–6.0 lakh cusecs). Sections L176‐R176 RD68‐LD68 L170‐R170 L11‐R11 L184‐R184 L21‐R21 L23‐R23 L25‐R25 RD75‐LD75 RD74‐LD74 RD72‐LD72 RD71‐LD71 RD70‐LD70 L35C‐R35C L43‐R43 RD84‐LD84 L15‐R15 L165‐R165 L20‐R20 L29‐R29 L72‐R72 L73‐R73

Q (m3/s)

Q (cusecs)

13,073.16 13,308.70 13,838.96 13,995.73 14,153.42 14,576.98 14,740.85 14,773.64 14,991.55 15,014.28 15,048.29 15,062.47 15,077.68 15,126.88 15,499.60 15,622.64 15,695.47 15,736.54 16,004.53 16,402.22 16,748.73 16,795.32

461,678.6 469,996.7 488,722.9 494,259.2 499,828.0 514,786.0 520,573.1 521,731.1 529,426.6 530,229.3 531,430.4 531,931.1 532,468.3 534,205.8 547,368.4 551,713.5 554,285.5 555,735.9 565,200.0 579,244.4 591,481.4 593,126.7

Total: 7.35%

carrying capacity of 24,777–24,944 m3/s (Table  16.7, Figure 16.6), located from downstream of Kakrapar weir to Kathor‐Amboli Village. Only 4.88% (14) of the sections have carrying capacity more than 25,485 m3/s, that is, 9 lakh cusecs (Table 16.8, Figure 16.6). These sections, L189‐R189 to L201‐R201, are located just downstream of the Ukai dam. The results clearly indicate that most of the sections having discharge capacity less than 12,740 m3/s are located near Surat city (Figure  16.7). It is found that the river carrying capacity near Nehru (Hope) Bridge is reduced to 5,096 m3/s, which ultimately shows that any volume of water more than 5,096 m3/s is enough to flood the area. In the 2006 floods, the southwest zone and west zone were flooded badly and inundated up to 3–4 m in water (Patel & Srivastava, 2013). It is observed that the Tapi River sections between RD‐LD 22 to RD‐LD 30 and RD‐LD 44 to RD‐LD 47 at Surat can carry the maximum discharge of 5,096 –11,326 m3/s without causing significant damage to urban dwellings and infrastructure. 16.5. ­VALIDATION OF SIMULATED FLOW The calibrated model has been used to simulate the flood for the year 2006. The boundary conditions are taken as described in the results section. The comparison of observed and simulated hydrograph at Kakrapar weir for the values of n (0.035, 0.032, 0.030, 0.028) is shown in Figure 16.8a. It shows that the simulated flow hydrograph at Manning’s n 0.032 is in close agreement with the observed values. The root mean square error (RMSE) for Manning’s n 0.032 is 1.0823 as shown in Table 16.2. Due to the lack of data availability for Ghala and Singanpur, the model is further simulated for Nehru (Hope Bridge) for Manning’s n (0.035, 0.032, 0.030, 0.028, 0.025, and 0.022) (Figure 16.9a). It shows that the simulated stages at Manning’s n 0.025 are best matched with the observed values and the corresponding root mean square error (RMSE) is 1.5106 as shown in Table 16.2. Furthermore, the velocity‐slope curve, stage‐velocity curve, and rating curve for Kakrapar weir and Nehru Bridge have been prepared and shown in Figures 16.8– 16.9 b, c, d. The value of R2 for both stations shows the good correlation. The obtained values can be helpful for recommendation and prevention of flood at the lower Tapi River in the future. 16.6. ­SENSITIVITY OF STAGE RELATION WITH TIDAL CONDITION In 2006, tidal waves were observed at the mouth of the river during flooding so that it is necessary to check the back water flooding under tidal conditions. The model is

One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India  221 Table 16.6  Q Between 17,000 and 21,230 m3/s (6.0–7.5 lakh cusecs). Sections RD15‐LD15 L40‐R40 L12‐R12 L172‐R172 L137‐R137 L171‐R171 L17‐R17 RD17‐LD17 L124‐R124 RD16‐LD16 L155‐R155 L161‐R161 L35A‐R35A L173‐R173 L158‐R158 L18‐R18 L174‐R174 L177‐R177 L178‐R178 L109‐R109 RD14‐LD14 L8A‐R8A L10A‐R10A L35B‐R35B L36‐R36 RD20‐LD20 RD11‐LD11

Q (m3/s)

Q (cusecs)

Sections

Q (m3/s)

Q (cusecs)

17,896.09 18,411.97 18,429.76 18,550.79 18,560.08 18,593.45 18,707.36 19,040.03 19,050.48 19,054.11 19,086.53 19,262.55 19,467.25 19,516.64 19,732.21 19,831.24 19,880.13 19,911.02 19,921.21 20,004.39 20,061.33 20,289.07 20,340.11 20,481.04 20,488.52 20,761.25 20,863.31

632,000.4 650,218.7 650,847.0 655,121.1 655,449.2 656,627.7 660,650.4 672,398.7 672,767.7 672,895.9 674,040.8 680,257.0 687,485.9 689,230.1 696,843.0 700,340.2 702,066.8 703,157.7 703,517.5 706,455.0 708,465.9 716,508.5 718,311.0 723,287.9 723,552.1 733,183.5 736,787.8

RD10‐LD10 RD9‐LD9 L6‐R6 L7‐R7 L131‐R131 L9A‐R9A

20,873.86 20,883.14 20,944.45 20,969.03 20,984.23 21,044.49

737,160.4 737,488.1 739,653.3 740,521.3 741,058.1 743,186.2

Total: 11.03%

further run with and without the tidal boundary condition. The tide hourly data are collected from Gujarat Maritime Board (GMB), Gandhinagar, and interpolated as per the model requirement. The Hazira tidal level is considered for the downstream boundary condition (Figure  16.3). From the graph (Figure 16.10), it can be clearly seen that the simulated stages with tide and without tide match each other. In addition, the scatter plot shows a good correlation with R2 value 0.997 (Figure  16.11). It indicates no significant effect of tidal cycle on the simulated stages at Nehru Bridge during peak flow condition. Thus, for prediction of the flood levels in Surat city, the simulated stages derived at the Nehru gauging station for the year 2006 flood can be useful regardless of the timing of the tidal cycle (Timbadiya et al., 2014a). 16.7. ­RECOMMENDATIONS AND SUGGESTION FOR FLOOD PREVENTION The discharge carrying capacity of the lower Tapi River is moderate. The maximum discharge carrying capacity is of section L179‐R179 for a discharge of 25,951 m3/s whereas the least one is of section RD30‐LD30 with

a  discharge of 5,341 m3/s. According to the existing carrying capacity values, it is really a very challenging task to decide the amount of water discharge from the Ukai dam to downstream up and the safe levels. For smooth evacuation of flood conditions without causing severe damage, some salient recommendations are suggested. 1. The water spill from the river section is an important task for decision makers and for flood inundation. To find out the actual spill and scenario of the flood of 2006, the graph was prepared for the right‐left bank line, bed RL, and water surface elevation (WSE) corresponding to the release from the Ukai dam on 9 August 2006 (25,736.32 m3/s) (Figure 16.12 a, b). The section, L201‐ R201 to L186‐R186, falls immediately after the Ukai dam, is safe against 25,736.32 m3/s discharge, and could not spill water from the left or right bank RL. For section L171‐R171 to L155‐R155, the WS elevation falls 66.15 m to 50 m gradually, so water spills out from left and right banks. The Kakrapar weir leads the influence on WSE and affluxes raise the WSE. Downstream from Kakrapar weir to Varacha Village has sections L140‐R140 to L35‐ R35 and the river bed RL is 32 m to 5 m. It means that

Table 16.7  Q Between 21,230 and 25,485 m3/s (7.5–9.0 lakh cusecs). Sections L16‐R16 L110‐R110 L111‐R111 RD12‐LD12 RD8‐LD8 RD5‐LD5 L8‐R8 L24‐R24 L138‐R138 L47‐R47 L13‐R13 L31‐R31 L14‐R14 L35‐R35 L139‐R139 RD7‐LD7 RD6‐LD6 L19‐R19 L9‐R9 L140‐R140 RD13‐LD13 L34B‐R34B L110A‐R110A RD4‐LD4 L45‐R45 L104‐R104 L37‐R37

Q (m3/s)

Q (cusecs)

Sections

Q (m3/s)

Q (cusecs)

Sections

Q (m3/s)

Q (cusecs)

21,285.21 21,318.96 21,324.64 21,381.72 21,415.92 21,437.49 21,510.06 21,584.13 21,608.95 21,624.33 21,676.64 21,685.82 21,696.46 21,782.34 21,813.39 21,835.36 21,841.76 21,844.94 21,937.01 21,948.04 22,131.21 22,151.45 22,171.39 22,176.76 22,370.41 22,576.59 22,655.23

751,687.2 752,879.1 753,079.7 755,095.4 756,303.2 757,065.0 759,627.8 762,243.6 763,120.1 763,663.2 765,510.5 765,834.7 766,210.5 769,243.3 770,339.9 771,115.7 771,341.8 771,454.1 774,705.5 775,095.0 781,563.7 782,278.5 782,982.6 783,172.3 790,011.0 797,292.3 800,069.4

L136‐R136 L145‐R145 L141‐R141 L41‐R41 L146‐R146 L51‐R51 L36A‐R36A L32‐R32 L33‐R33 L147‐R147 L82‐R82 L38‐R38 L50‐R50 L44‐R44 L34A‐R34A L46‐R46 L34‐R34 L142‐R142 L144‐R144 L148‐R148 L49‐L49 L149‐R149 L10‐R10 L79‐R79 L97A‐R97A L42‐R42 L53‐R53

22,791.08 22,811.18 22,831.16 22,869.84 23,045.31 23,071.96 23,080.78 23,145.14 23,155.65 23,285.79 23,293.44 23,325.00 23,349.18 23,368.12 23,517.72 23,633.89 23,834.57 23,931.37 23,946.88 24,000.09 24,006.13 24,013.18 24,081.17 24,358.43 24,542.31 24,677.41 24,678.63

804,867.0 805,576.8 806,282.4 807,648.4 813,845.1 814,786.3 815,097.7 817,370.6 817,741.8 822,337.7 822,607.8 823,722.4 824,576.3 825,245.2 830,528.3 834,630.8 841,717.8 845,136.3 845,684.1 847,563.2 847,776.5 848,025.5 850,426.5 860,218.0 866,711.7 871,482.7 871,525.8

L48‐R48 L52‐R52 L54‐R54 L55‐R55 L56‐R56 L57‐R57 L58‐R58 L59‐R59 L60‐R60 L61‐R61 L62‐R62 L63‐R63 L64‐R64 L64A‐R64A L65‐R65 L66‐R66 L67‐R67 L68‐R68 L69‐R69 L70‐R70 L71‐R71 L74‐R74 L75‐R75 L76‐R76 L77‐R77 L78‐R78 L78A‐R78A

24,678.73 24,777.18 24,779.52 24,780.63 24,781.64 24,783.21 24,784.73 24,786.23 24,787.30 24,788.32 24,789.50 24,791.00 24,792.42 24,793.48 24,794.43 24,795.32 24,796.09 24,796.99 24,797.65 24,798.51 24,799.29 24,801.41 24,802.74 24,803.17 24,803.32 24,803.44 24,803.75

871,529.3 875,006.1 875,088.7 875,127.9 875,163.6 875,219.1 875,272.7 875,325.7 875,363.5 875,399.5 875,441.2 875,494.2 875,544.3 875,581.7 875,615.3 875,646.7 875,673.9 875,705.7 875,729.0 875,759.4 875,786.9 875,861.8 875,908.8 875,923.9 875,929.2 875,933.5 875,944.4 (Continued)

Sections L79A‐R79A L80‐R80 L81‐R81 L83‐R83 L84‐R84 L85‐R85 L86‐R86 L87‐R87 L88‐R88 L89‐R89 L90‐R90 L91‐R91 L92‐R92 L93‐R93 L94‐R94 L95‐R95 L96‐R96 L97‐R97 L98‐R98 L99‐R99 L100‐R100 L101‐R101 L102‐R102 L134‐R134 L103‐R103 L105‐R105 L106‐R106

Q (m3/s)

Q (cusecs)

Sections

Q (m3/s)

Q (cusecs)

24,804.24 24,804.36 24,804.45 24,871.11 24,873.20 24,874.60 24,876.27 24,877.35 24,878.93 24,881.34 24,883.40 24,885.03 24,886.46 24,888.22 24,890.19 24,891.83 24,893.37 24,895.70 24,898.50 24,901.14 24,903.20 24,904.69 24,906.14 24,907.34 24,908.30 24,911.71 24,912.86

875,961.7 875,966.0 875,969.2 878,323.2 878,397.1 878,446.5 878,505.5 878,543.6 878,599.4 878,684.5 878,757.3 878,814.8 878,865.3 878,927.5 878,997.1 879,055.0 879,109.4 879,191.6 879,290.5 879,383.8 879,456.5 879,509.1 879,560.3 879,602.7 879,636.6 879,757.0 879,797.7

L107‐R107 L112‐R112 L107A‐R107A L113‐R113 L108‐R108 L114‐R114 L115‐R115 L116‐R116 L143‐R143 L118A‐R118A L119‐R119 L120‐R120 L122‐R122 L123‐R123 L127‐R127 L130‐R130 L130A‐R130A L132‐R132 L133‐R133 L135‐R135 L180‐R180 L181‐R181 L182‐R182 L183‐R183 L185‐R185 L186‐R186 L187‐R187

24,914.16 24,914.99 24,915.23 24,915.85 24,916.29 24,916.94 24,917.71 24,918.76 24,920.20 24,921.16 24,921.80 24,923.07 24,924.81 24,926.00 24,931.54 24,938.51 24,939.43 24,941.57 24,942.53 24,944.06 25,216.83 25,240.62 25,248.89 25,273.46 25,352.56 25,370.39 25,419.30

879,843.6 879,872.9 879,881.3 879,903.2 879,918.8 879,941.7 879,968.9 880,006.0 880,056.9 880,090.8 880,113.4 880,158.2 880,219.7 880,261.7 880,457.3 880,703.5 880,736.0 880,811.5 880,845.4 880,899.5 890,532.4 891,372.5 891,664.6 892,532.2 895,325.7 895,955.3 897,682.6

Total: 45.15%

224  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Table 16.8  Q More Than 25,485 m3/s (>9.0 lakh cusecs ). Sections L189‐R189 L198‐R198 L190‐R190 L191‐R191 L195‐R195 L193‐R193 L194‐R194 L192‐R192 L199‐R199 L196‐R196 L200‐R200 L201‐R201 L197‐R197 L179‐R179

Q (m /s)

Q (cusecs)

25,493.91 25,543.18 25,602.37 25,619.16 25,629.48 25,688.68 25,693.00 25,719.74 25,731.67 25,732.88 25,733.62 25,736.32 25,743.12 25,951.32

900,317.4 902,057.4 904,147.7 904,740.6 905,105.1 907,195.7 907,348.3 908,292.6 908,713.9 908,756.7 908,782.8 908,878.1 909,118.3 916,470.9

3

Total: 4.88%

the river falls steeply by 29 m and has adequate discharge carrying capacity. The WSE is lower than the left and right bank RL and safe against spill. After section L 35‐R 35 (downstream of Ukai at 73,851 m–116,468 m), the

river bed falls gradually from 5 m to 10 m and thus indicates flat slope. Several sections, LD 60 to LD 78 near Mugdalla jetty, LD 21 to LD 36 near Nehru Bridge, Swami Vivekananda Bridge, and SVP Bridge are found to have very low bank RL. For example, section LD 26 at Nehru Bridge has left bank RL 10.47 m and right bank RL 5.51 m while the water surface RL, in the 2006 flood, observed was 12.50 m. These figures indicate that 2.03 m and 6.99 m water spilled from the left bank and right bank sections, respectively. This clearly pointed out that most of the bank RL was lower than the water surface RL at 25,736.32 m3/s during the 2006 flood. Water spilled from sections LD 20‐RD 20 to LD 47‐RD 47 near Adajan, LD 55‐RD 55 to LD 72‐RD 72 near Bhatp or village and other sections as shown in Figure 16.12 (a, b). It is recommended that sections having low RL need to be strengthened through appropriate bank and levees protection work. Patel and Srivastava (2013) found that encroachments named Pal Patiya, New Haveli Temple, Bhulka Vihar school, Hanuman Temple, and RTO building in and around right bank of river, in west zone, are undesirable and make the river narrower (Figure 16.13).

Figure 16.7  The maximum discharge carrying capacity of the Tapi River in and around Surat city.

(a) 30000

Discharge (m3/s)

25000 20000 15000 10000 5000 Simulated (N 0.035)

Simulated (N 0.032)

Simulated (N 0.028)

Simulated (N 0.030)

Observed

6/8/06*24 7/8/06*2 7/8/06*4 7/8/06*6 7/8/06*8 7/8/06*10 7/8/06*12 7/8/06*14 7/8/06*16 7/8/06*18 7/8/06*20 7/8/06*22 7/8/06*24 8/8/06*2 8/8/06*4 8/8/06*6 8/8/06*8 8/8/06*10 8/8/06*12 8/8/06*14 8/8/06*16 8/8/06*18 8/8/06*20 8/8/06*22 8/8/06*24 9/8/06*2 9/8/06*4 9/8/06*6 9/8/06*8 9/8/06*10 9/8/06*12 9/8/06*14 9/8/06*16 9/8/06*18 9/8/06*20 9/8/06*22 9/8/06*24

0

Time (hour)

(b) 18

y = 27.15x0.286 R2 = 0.975

16

Velocity in (m/m)

14 12 10 8 6 4 2 0

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

Slope (m/m)

(c) 18

y = 2E + 07x–3.89 R2 = 0.917

16

Velocity in (m/s)

14 12 10 8 6 4 2 0 0

10

20

30

40

50

60

70

Water stage in (m)

Figure 16.8  (a) Observed and simulated flood hydrograph at Kakrapar weir (2006), (b) velocity‐slope curve of Kakrapar weir, (c) velocity‐stage curve of Kakrapar weir, (d) rating curve of Kakrapar weir.

(d) 70

Legend W.S. Elev

W.S. Elev (m)

65

60

55

50

45

0

5000

10000

15000

20000

25000

30000

Q Total (m3/s)

Figure 16.8  (Continued) (a) 35 30

Water level (m)

25 20 15 10 5

Observed

Simulated (N 0.025)

Simulated (N 0.035)

Simulated (N 0.030)

Simulated (N 0.028)

Simulated (N 0.022)

Simulated (N 0.032)

6/8/06*24 7/8/06*2 7/8/06*4 7/8/06*6 7/8/06*8 7/8/06*10 7/8/06*12 7/8/06*14 7/8/06*16 7/8/06*18 7/8/06*20 7/8/06*22 7/8/06*24 8/8/06*2 8/8/06*4 8/8/06*6 8/8/06*8 8/8/06*10 8/8/06*12 8/8/06*14 8/8/06*16 8/8/06*18 8/8/06*20 8/8/06*22 8/8/06*24 9/8/06*2 9/8/06*4 9/8/06*6 9/8/06*8 9/8/06*10 9/8/06*12 9/8/06*14 9/8/06*16 9/8/06*18 9/8/06*20 9/8/06*22 9/8/06*24

0

Time (hour)

(b) 3.5

y = –1E+08x2 + 45019x–1.216 R2 = 0.955

Velocity in (m/s)

3 2.5 2 1.5 1 0.5 0 0

0.00005

0.0001

0.00015

0.0002

0.00025

0.0003

Slope (m/m)

Figure 16.9  (a) Observed and simulated stage at Nehru (Hope) bridge (2006), (b) velocity‐slope curve of Nehru (Hope) bridge, (c) velocity‐stage curve of Nehru (Hope) bridge, (d) rating curve of Nehru (Hope) bridge.

One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India  227 (c) y = –0.015x2 + 0.479x–0.688 R2 = 0.993

3.5 3

Velocity in (m/s)

2.5 2 1.5 1 0.5 0

0

2

4

6

8

10

12

14

16

Water stage in (m)

(d)

Legend

14

W.S. Elev

12 10 W.S. Elev (m)

8 6 4 2 0 –2 –4

0

5000

10000 Q Total

15000

20000

25000

(m3/s)

Figure 16.9  (Continued)

2. Until the 1994 monsoon, the warning level and danger level of Surat city were set at 10.18 m and 11.18 m, respectively. After the 1994 flood, new levels were defined, which were 8.5 m and 9.5 m, corresponding to the discharges of 11,328 m3/s and 13,027 m3/s respectively. During the 1998 flood, for the discharge of 11,328 m3/s, the gauge level was 8.0 m, which is below the warning level as decided earlier, even though the city was i­nundated up to 1 m. In the present study, it is clearly seen that the lowest bank RL near Nehru Bridge to Sardar Vallabhai Bridge is 5.0–6.0 m. The water level, rising above the said values will spill out from the sections and inundate the adjoining areas. Hence, it is strongly recommended to reset the warning and danger levels of the city to 5.5 m and 6.5 m, ­respectively (consid-

ering the riverfront at Adajan, Bapunagar Road, and surrounding areas of west zone). The lowest bank RL, rating curves, simulated flood peak stage, and simulated flood peak ­ discharge of different bridges are shown in Figures 16.14 and 16.15 (a–g) and should be utilized to fix the warning levels for the surrounding areas. 3. Other suggestions to prevent flooding at Surat city include effective operation of the Ukai reservoir and modification of its operation policy, development of mathematical model (1-D–2-D integrated) for the entire Tapi River channel network, hydrological data network upgrade, appropriate bank protection measures (i.e., levees), and appropriate storm water drainage arrangement for the low lying areas.

228  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Simulated stage without tidal level

Simulated stage with tidal level

Observed

16 14

Water level (m)

12 10 8 6 4 2

8/8/06*2 8/8/06*4 8/8/06*6 8/8/06*8 8/8/06*10 8/8/06*12 8/8/06*14 8/8/06*16 8/8/06*18 8/8/06*20 8/8/06*22 8/8/06*24 9/8/06*2 9/8/06*4 9/8/06*6 9/8/06*8 9/8/06*10 9/8/06*12 9/8/06*14 9/8/06*16 9/8/06*18 9/8/06*20 9/8/06*22 9/8/06*24

6/8/06*24 7/8/06*2 7/8/06*4 7/8/06*6 7/8/06*8 7/8/06*10 7/8/06*12 7/8/06*14 7/8/06*16 7/8/06*18 7/8/06*20 7/8/06*22 7/8/06*24

0

Time (hour)

Figure 16.10  Simulated Nehru bridge stage with tidal level, without tidal level, and observed.

Stage without tidal level (m)

16

y = 0.951x1.017 R2 = 0.997

14 12 10 8 6 4 2 0

0

2

4

6

8

10

12

14

16

18

Stage with tidal level (m)

Figure 16.11 Scatter plot with and without tidal level at Nehru bridge.

16.8. ­DISCUSSIONS HEC‐RAD 1-D hydrodynamic modeling has been ­ erformed under the following uncertainty and scarcity p of data: 1. The Varekhadi River and watersheds situated downstream of the Ukai dam are not considered for flood simulation (Patel et al., 2012a). The release from the Ukai dam and, at the same time, additional ­discharge from ­the

Varekahdi may lead to different results and need to be considered for future flood conditions. 2. Simulated and observed flows at Nehru Bridge are slightly different because the 1-D HEC‐RAS model creates a vertical wall‐type situation on either bank as the water level exceeds the bank level and the entire bank flow domain remains concentrated within both banks (Timbadiya et al., 2014b). Thus, it may not provide accurate results as the water spills from either bank of the river into the floodplain area (Timbadiya et  al., 2014b) and the flow starts behaving in a 2-D manner (Patel & Srivastava, 2013). Simulated results indicate that downstream of Kakrapar weir (L134‐R134, 31.968 km) to (L52‐R52, 77.527 km) near Kathor Village has adequate discharge carrying capacity with less chance of spill, where the model leads to precise results compared with those farther downstream. 3. While modeling bridge structures through HEC‐ RAS, it is observed that the multiple inline pilings for pier footers are impossible to create as actual (Figure 16.16a‐b). It affects the flow regulation and the Manning roughness coefficient n. 4. It was assumed that the LOB (Left Overbank) and ROB (Right Overbank) are the same as channel in downstream reach lengths in HEC‐RAS m ­ odeling, but actually it may differ and lead to the influence on  flow regulation as well as Manning’s roughness ­c oefficient n.

One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India  229 (a) Right bank line

River bed

Water surface elevation (9 Aug 2006) 80 70

50 40 30 20 10

Simulated water surface (m)

60

RD81 – LD81 RD75– LD75 RD69– LD69 RD63– LD63 RD57– LD57 RD51– LD51 RD45– LD45 RD39– LD39 RD33– LD33 RD27– LD27 RD21– LD21 RD15– LD15 RD9– LD9 RD3– LD3 L7A– R7A L10A– R10A L15– R15 L21– R21 L25– R25 L31– R31 L35– R35 L37– R37 L43– R43 L49– R49 L55– R55 L61– R61 L66– R66 L72– R72 L78– R78 L82– R82 L88– R88 L94– R94 L99– R99 L105– R105 L110– R110 L115– R115 L120– R120 L126– R126 L131– R131 L137– R137 L143– R143 L149– R149 L159– R159 L165– R165 L171– R171 L177– R177 L183– R183 L189– R189 L195– R195 L201– R201

0 –10 –20

River section (Ukai dam to Sea) (m)

(b) Left bank line

River bed

Water surface elevation (9 Aug 2006) 80 70

50 40 30 20 10

L201– R201

L173– R173 L180– R180 L187– R187 L194– R194

L166– R166

L141– R141

L148– R148 L159– R159

L128– R128 L134– R134

L115– R115 L121– R121

L97– R97 L103– R103 L109– R109

L83– R83 L90– R90

L64A– R64A L71– R71 L78– R78

L51– R51 L58– R58

L44– R44

L29– R29

L34B– R34B L37– R37

L17– R17 L22B– R22B

L7A– R7A L11– R11

RD4– LD4

RD11– LD11

RD32– LD32 RD25– LD25 RD18– LD18

RD46– LD46 RD39– LD39

RD81–LD81

RD74– LD74 RD67– LD67 RD60– LD60 RD53– LD53

0

Simulated water surface elevation (m)

60

–10 –20

River section (Ukai dam to Sea) (m)

Figure 16.12  (a) River bed line, right bank line, and water surface at 2006 flood; (b) river bed line, left bank line, and water surface at 2006 flood.

5. The lower Tapi River has an island (between L99‐ R99, 51.161 km and L93‐R93, 54.575 km) where the river is divided into two channels. In addition, the mouth of the river after LD79‐RD79 at 115.024 km is divided into two channels. This lack of segments would change the

­ redicted results. Only the major segment is considered p for flood simulation. 6. The sedimentation deposition, bank protection works (Gabion) after the 2006 flood, and illegal settlement along the right bank in the floodplain at west zone are not considered

230  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION (a)

(b)

Legend River River banks Settlement

Figure 16.13 (a) Toposheet overlay on Google maps, river banks and illegal settlements inside the banks, (b) photographs shows the settlements on the right banks of the Tapi River near the west zone.

Simulated flood peak stage in 2006

Maximum Q

Simulated flood peak discharge in 2006

20

27500

18

25000

16

22500 20000

Stage (m)

14

17500

12

15000 10 12500 8

10000

6

Simulated discharge (m3/s)

Lowest bank line RL

7500

4

5000

2

2500

0

0 51.5

55.5

58.5

91.5

110.5

112.5

116.5

122.5

Bride ID

Figure 16.14  Simulated stage and discharge at different bridges.

for modeling, and this may possibly lead to uncertainty in levee design and Manning’s roughness coefficient n. The Differential Global Positioning System (DGPS) could be helpful to identify the RL of the Gabion line (Joshi et  al., 2012) and illegal settlements. New certain danger and warning levels for Surat city could be proposed after this survey.

7. The storm/sewer water and the river water in the urban area were not considered in the current study because of the unavailability of data for modeling (Patel, 2011; Timbadiya et al., 2014a). 8. According to Timbadiya et al. (2014a), the observed water level and discharge at Ghala gauging station along

(a) 15

Legend W.S. Elev

W.S. Elev (m)

10

5

0

–5

–10

0

5000

10000

15000

20000

25000

Q total (m3/s)

(b) 15

Legend W.S. Elev

W.S. Elev (m)

10

5

0

–5

0

5000

10000

15000

20000

25000

Q total (m3/s)

(c) 20

Legend W.S. Elev

15

W.S. Elev (m)

10

5

0

–5

–10

0

5000

10000

15000

20000

25000

Q total (m3/s)

Figure 16.15  Rating curves of (a) Sardar Vallabhai Bridge, (b)Swami Vivekanand Bridge, (c) Daboli‐Jahagirpur Bridge, (d) Amroli (New) Bridge, (e) Railway Bridge, (f) Kapodra to Utrva Bridge, (g) Nana Varaccha to Mota Varaccha Bridge.

(d) 20

Legend W.S. Elev

W.S. Elev (m)

15

10

5

0

–5

0

5000

10000

15000

20000

25000

Q total (m3/s)

(e) 20

Legend W.S. Elev

W.S. Elev (m)

15

10

5

0

–5

0

5000

10000

15000 Q total (m3/s)

20000

25000

(f) 20

Legend W.S. Elev

15

W.S. Elev (m)

10

5

0

–5

–10

0

5000

10000

15000 Q total (m3/s)

Figure 16.15  (Continued)

20000

25000

One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India  233 (g)

20

Legend W.S. Elev

W.S. Elev (m)

15

10

5

0

0

5000

15000

10000 Q total

20000

25000

(m3/s)

Figure 16.15  (Continued)

(a)

(b)

Figure 16.16  (a) Nana Varaccha to Mota Varaccha Bridge, (b) Dabholi‐Jhagirpura Bridge consists of inline piling of pier footer.

the lower Tapi River seem to be less accurate because of extraction of sand near the gauging site. 16.9. ­CONCLUSIONS The present study addresses the discharge carrying capacity of lower Tapi River after the 2006 flood, which is one of the decisive factors for releasing the water from the Ukai dam to downstream. It is a life threatening decision and the predicted results will protect the 4.6 million

people settled at Surat city. This scenario addresses flood impact on human beings and the recommendation ­suggests flood resilience measures through the 1-D HAC‐ RAS hydrodynamic modeling. The research findings are summarized as follows: 1. River carrying capacity is an important decisive factor for flood mitigation and management practice. 1-D HEC‐ RAS hydrodynamic modeling is an applicable tool to derive the 1-D river hydraulic parameters like total discharge, water surface elevations, energy ­gradient elevation, energy gradient

234  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

slope, velocity, flow area, and Froude number at different channel sections, which help to improve the decision‐making system. 2. The model is calibrated for the 1998 flood and validated for 2006. Manning’s roughness coefficient for the lower Tapi River is 0.032 up to the Kakrapar weir and 0.025 downstream of the Kakrapar weir, which would be ­suitable for simulation of future flood in the lower Tapi River. 3. The discharge carrying capacity of lower Tapi river near Surat city is decreased significantly, the flow above 5,096 m3/s may possibly lead to floods in the low lying areas at Surat city. 4. The tidal cycle doesn’t have an influence on the flood stage at the Nehru Bridge gauging site. 5. The 1-D model could not predict the accurate stage when the water starts to spill from the section. A 2-D hydrodynamic model, including floodplain area, is required to improve the results. The results gained from this research will be very helpful to policy makers and hydrologists for developing integrated hydrodynamic (1-D, 2-D, 3-D) models, flood mitigation measures, and advanced flood warning systems to protect Surat city against flooding. ­ACKNOWLEDGMENTS The corresponding author (DPP) would like to express his sincere thanks to PDPU for providing a support to execute work. The authors would like to thank, Space Application Centre-Indian Space Research Organization (SAC-ISRO), Bhaskaracharya Institute for Space Applications and Geo-Informatics (BISAG), National Bureau of Soil Survey and Land Use Planning (NBSS & LUP), National Resources Information System, Survey of India (SOI), Central Water Commission (CWC), State Water Data Centre (SWDC), Survey of India (SoI), Irrigation Department for providing the necessary data, facilities, and support during the study period. ­REFERENCES Ali, H., Lee, T. S, Majid, M., & Hadi, M. (2012). Incorporation of GIS based program into hydraulic model for water level modeling on river basin. Journal of Water Resource and Protection, 4, 25–31. Bates, P., Stewart, M., Desitter, A., Anderson, M., Renaud, J. P., & Smith, J. (2000). Numerical simulation of floodplain hydrology. Water Resources Research, 36, 2517–2529 Bellos, V., & Tsakiris, G. (2015). Comparing various methods of building representation for 2D flood modelling in built‐up areas. Water Resources Management, 29(2), 379–397. Brunner, G. (2008). HEC‐RAS River Analysis System, Version 3.0. User’s manual CPD‐68, Hydrologic Engineering Center, Davis, CA.

Brunner, G. (2010). HEC‐RAS, River Analysis System Hydraulic Reference Manual (Version 4.1). Report CPD‐69 retrieved from US Army Corps of Engineers, Hydraulic Engineering Center. http://www hec usace army mil/software/hec‐ras/documents/HECRAS_4 1_Reference_ Manual.pdf. Castellarin, A., Di Baldassarre, G., Bates, P., & Brath, A. (2009). Optimal cross‐sectional spacing in Preissmann scheme 1D hydrodynamic models. Journal of Hydraulic Engineering, 135, 96–105 CWC (2000–2001). Water Year Book 2000–2001, Tapi basin, hydrological observation circle, Gandhinagar, Gujarat, India. Hassan, A., Ghani, A., & Abdullah, R. (2006). Development of flood risk map using GIS for sg. Selangor basin. Proceedings of the 6th International Conference on ASIA GIS, 9–10 Mar 2006, Johor, Malaysia. Hicks, F., & Peacock, T. (2005). Suitability of HEC‐RAS for flood forecasting. Canadian Water Resources Journal, 30, 159–174. Horritt, M., & Bates, P. (2002). Evaluation of 1D and 2D numerical models for predicting river flood inundation. Journal of Hydrology, 268, 87–99 Horritt, M., Di Baldassarre, G., Bates, P., & Brath, A. (2007). Comparing the performance of a 2‐D finite element and a 2‐D finite volume model of floodplain inundation using airborne SAR imagery. Hydrological Processes, 21, 2745–2759; https://www.suratmunicipal.gov.in/Bridgecell/. Johnson, G. D., Strickland, M. D., Buyok, J. P., Derby, C. E., & Young, D. P. (1999). Quantifying impacts to riparian wetlands associated with reduced flows along the Greybull River, Wyoming. Wetlands, 19, 71–77. Joshi, P. M., Sherasia, N. K., & Patel, D. P. (2012). Urban flood mapping by geospatial technique a case study of Surat City. IOSR Journal of Engineering, 2, 43–51. Lee, K. T., Ho, Y‐H., & Chyan, Y‐J. (2006). Bridge blockage and overbank flow simulations using HEC–RAS in the Keelung River during the 2001 Nari typhoon. Journal of Hydraulic Engineering, 132, 319–323. Nandalal, K. (2009). Use of a hydrodynamic model to forecast floods of Kalu River in Sri Lanka. Journal of Flood Risk Management, 2, 151–158. Noman, N. S., Nelson, E. J., & Zundel, A. K. (2003). Improved process for floodplain delineation from digital terrain models. Journal of Water Resources Planning and Management, 129, 427–436. Pappenberger, F., Beven, K., Horritt, M., & Blazkova, S. (2005). Uncertainty in the calibration of effective roughness parameters in HEC‐RAS using inundation and downstream level observations. Journal of Hydrology, 302, 46–69. Parhi, P. K., Sankhua, R., & Roy, G. (2012). Calibration of channel roughness for Mahanadi River (India) using HEC‐ RAS model. Journal of Water Resource and Protection, 4, 847. Parmar, B., & Rao, B. (2002). Flood control operation policy for ukai multi‐purpose reservoir. In ICOLD Symposium on Reservoir Management in Tropical and Sub‐tropical Regions, Brazil, 278. Parsa, A. S., Heydari, M., Sadeghian, M. S., & Moharrampour, M. (2013). Flood zoning simulation by HEC‐RAS model (Case Study: Johor River‐Kota Tinggi Region). Journal of River Engineering, x1 (1).

One-Dimensional Hydrodynamic Modeling of the River Tapi: The 2006 Flood, Surat, India  235 Patel, D. P. (2011). Flood assessment by integrated hydrological modeling with RS and GIS in water resources management. PhD Thesis, Gujarat University, Ahmedabad, Gujarat, India Patel, D. P., & Dholakia, M. B. (2010a). Feasible structural and nonstructural measures to minimize effect of flood in lower Tapi Basin. WSEAS Transactions on Fluid Mechanics, 3(5), 104–121. Patel, D. P., & Dholakia, M. B. (2010b). Identifying probable submergence area of Surat City using digital elevation model and geographical information system. World Applied Sciences Journal, 9, 461–466. Patel, D. P., & Srivastava, P. K. (2013). Flood hazards mitigation analysis using remote sensing and GIS: Correspondence with town planning scheme. Water Resources Management, 27, 2353–2368. Patel, D. P., Dholakia, M. B., Naresh, N., & Srivastava, P. K. (2012a). Water harvesting structure positioning by using geo‐ visualization concept and prioritization of mini‐watersheds through morphometric analysis in the Lower Tapi Basin. Journal of the Indian Society of Remote Sensing, 40, 299–312. Patel, D. P., Gajjar, C. A., & Srivastava, P. K. (2012b). Prioritization of Malesari mini‐watersheds through morphometric analysis: a remote sensing and GIS perspective. Environmental Earth Sciences, 69(8), 1–14. Rahman, M. M., Arya, D., Goel, N., & Dhamy, A. P. (2010). Design flow and stage computations in the Teesta River, Bangladesh, using frequency analysis and MIKE 11 modeling. Journal of Hydrologic Engineering, 16, 176–186. Ramesh, R., Datta, B., Bhallamudi, S. M., & Narayana, A. (2000). Optimal estimation of roughness in open‐channel flows. Journal of Hydraulic Engineering, 126, 299–303 Remo, J. W., Carlson, M., & Pinter, N. (2012). Hydraulic and flood‐loss modeling of levee, floodplain, and river management strategies, Middle Mississippi River, USA. Natural Hazards, 61, 551–575. Reza Ghanbarpour, M., Salimi, S., Saravi, M., & Zarei, M. (2011). Calibration of river hydraulic model combined with GIS analysis using ground‐based observation data. Research Journal of Applied Sciences, Engineering and Technology, 3, 456–463. Sadeghi, S., & Rad, R. J. (2004). Floodplain zoning in an

Iranian urban watershed using HEC‐RAS hydraulic model and ArcView GIS. Hydrology Journal, 27, 69–77. Said, S., Mohammed, T. A., Bardaie, M. Z., & Nor Basri, S. (2002). Hydraulic simulation of flood occurrences in a tropical river system: The case of Linggi River system Pertanika. Journal of Science & Technology, 10, 1–12. Salimi, S., Ghanbarpour, M. R., Solaimani, K., & Ahmadi, M. Z. (2008). Floodplain mapping using hydraulic simulation model in GIS. Journal of Applied Sciences, 8, 660–665. Tate, E. C., & Maidment, D. R. (1999). Floodplain mapping using HEC‐RAS and ArcView GIS. University of Texas at Austin. Timbadiya, P., Patel, P., & Porey, P. (2011a). HEC‐RAS based hydrodynamic model in prediction of stages of lower Tapi River ISH. Journal of Hydraulic Engineering, 17, 110–117. Timbadiya, P. V., Patel, P. L., & Porey, P. D. (2011b). Calibration of HEC‐RAS model on prediction of flood for lower Tapi River, India. Journal of Water Resource and Protection, 3, 805. Timbadiya, P., Patel, P., & Porey, P. (2014a). A 1D‐2D coupled hydrodynamic model for river flood prediction in a coastal urban floodplain. Journal of Hydrologic Engineering, 20. Timbadiya, P., Patel, P., & Porey, P. (2014b). One‐dimensional hydrodynamic modelling of flooding and stage hydrographs in the lower Tapi River in India. Current Science, 106, 708–716. Tsakiris, G., & Bellos, V. (2014). A numerical model for two‐ dimensional flood routing in complex terrains. Water Resources Management, 28, 1277–1291. Villazón Gómez, M. F. (2011). Modeling and conceptualization of hydrology and river hydraulics in flood conditions, for Belgian and Bolivian basins. PhD Thesis, KU Leuven, Belgium. Villiazon, M. F., Aranibar, C. R., & Willems, P. (2009). Hydrodynamics unsteady flow model applied to pirai river for flood events estimation in Santa Cruz‐Bolivia. 1st International Congress of Hydroclimatology, SENAMHI Cochabamba, Bolivia, 8–1. Wasantha Lal, A. (1995). Calibration of riverbed roughness. Journal of Hydraulic Engineering, 121, 664–671.

Section IV Ocean Hazards and Disasters

17 Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal Prasad K. Bhaskaran1, A. D. Rao2, and Tad Murty3

ABSTRACT The Bay of Bengal and the Gulf of Mexico are the two water bodies on the globe that are most prone to storm surges generated by tropical cyclones. In this chapter, a review has been made of the storm surge problem in the Bay of Bengal region located in the North Indian Ocean. Using the contemporary numerical models, not only storm surge elevations but also coastal inundations were computed for some recent cyclones in the Bay of Bengal. These models included the interactions between storm surges, tides, and wind waves. However, it should be noted that one of the challenging issues, which still remains unsolved to a large extent, is computing the interaction between storm surges and river flooding, and the contribution of this interaction to coastal flooding and inundation. A dramatic example of such an interaction was during the 29 October 1999 cyclone on the Odisha coast. 17.1. ­INTRODUCTION Storm surge and wind waves are the manifestation of surface winds blowing over the ocean surface, and turn out quite detrimental in coastal areas during tropical cyclone activity. Tropical cyclones form over the warm ocean surface and are widely recognized as being among the natural geohazards that can result in enormous loss of life, property, and damage to infrastructure during landfall and the postlandfall phase. During the landfall of a tropical cyclone, the worst affected areas are the low‐ lying coastal regions that directly bear the brunt resulting from abnormal rise in water levels due to extreme winds, storm surge, and wind‐wave activity. The total water level elevation near the coast is a combined effect ­resulting from the mutual nonlinear interaction between storm surges, astronomical tide, and wave‐induced Department of Ocean Engineering and Naval Architecture, Indian Institute of Technology Kharagpur, West Bengal, India 2  Centre for Atmospheric Sciences, Indian Institute of Technology Delhi, New Delhi, India 3  Department of Civil Engineering, University of Ottawa, Ottawa, Ontario, Canada 1 

setup/setdown. The worst possible scenario of extreme water level can occur when the storm surge coincides with the astronomical high water. In the hinterland regions, the major damage and devastation can result from extreme wind speed and coastal and inland floods due to torrential rainfall. About 80 tropical cyclones form over the global ocean basins annually, and about 5–6% of this total number form over the North Indian Ocean basin (Niyas et  al., 2009). The east coast of India that borders the Bay of Bengal is considered the most vulnerable and susceptible region in the world in the context of risk associated with tropical cyclones and extreme wind waves. The frequency of cyclones is much higher in the Bay of Bengal as compared with the Arabian Sea with a return period of about 4–5 in a year with landfall either in West Bengal, Odisha, Andhra Pradesh, or Tamil Nadu. Analysis of historical cyclone tracks clearly indicate that the State of Odisha located on the east coast of India receives the highest frequency of tropical cyclone landfall followed by Andhra Pradesh, West Bengal, and Tamil Nadu. A recent study on the assessment of historical cyclone tracks for  four decades in the Bay of Bengal clearly indicates a  rising trend in the energy metrics such as Power Dissipation

Techniques for Disaster Risk Management and Mitigation, First Edition. Edited by Prashant K. Srivastava, Sudhir Kumar Singh, U. C. Mohanty, and Tad Murty. © 2020 John Wiley & Sons, Inc. Published 2020 by John Wiley & Sons, Inc. 239

240  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Index (PDI) and the Accumulated Cyclone Energy (ACE) for tropical cyclones that form over the  Bay of Bengal region (Sahoo & Bhaskaran, 2015). The estimated PDI for tropical cyclones in the present decade is about six times higher as compared with the past over the Bay of Bengal basin, and that has direct implications on coastal vulnerability associated with storm surges and extreme wind waves over this region. Another recent study on coastal hydrodynamics using a coupled model for cyclone Hudhud in the Bay of Bengal (Murty et al., 2016) clearly indicates that the size of tropical cyclones that formed over this region has also increased during the present decade. Therefore in a changing ­climate the occurrence of high‐intensity tropical cyclones along with their increase in size has a direct implication on the vulnerability of coastal belts. It means that vast expanses of coastal regions are exposed to higher wind speeds, storm surge envelopes, differential coastal flooding scenarios, and impact from extreme wind waves. In their study, Murty et al. 2016 indicate that the existing parametric wind field formulation needs to be revisited and modified accordingly considering the overall radial distance in wind field envelope keeping in view the increased size of tropical cyclones over the Bay of Bengal region. A 3/5 power‐law was proposed (Murty et al., 2016) that takes care of the increased tropical cyclone size over this region. Despite the fact that the east coast of India is highly vulnerable to the impacts from tropical cyclone landfall, there is a growing concern and urgent demand among the scientific community at present to conduct a systematic and more focused study to address aspects of coastal vulnerability in a holistic view by considering the contributions from various environmental drivers leading to an overall assessment of risk and coastal vulnerability. Such studies have long‐term implications and beneficial value and therefore need to be planned in a holistic manner considering aspects of coastal, social, economic, and environmental vulnerability having wide socioeconomic implications. The study on tropical cyclone activity, storm surges, and associated extreme wind waves is a quite fascinating subject having many challenges that have unequivocally drawn the attention of the scientific community worldwide in order to provide better quality forecast in terms of cyclone track, intensity, and the probable landfall location to aid timely warnings for better emergency ­ operations and evacuation measures and efficient coastal zone management. The archives on historical cyclone track records signify that each cyclone track is unique in nature, thereby posing a real challenge to the atmospheric scientists and oceanographers worldwide to devise a reliable forecasting system to predict cyclone tracks, improve accuracy in tropical cyclone landfall, and estimate storm surge and associated coastal flooding and the role

of wave‐induced setup/setdown from extreme wind waves in the total water level elevation. Also, at present there is a growing necessity and pressing demand to improve the quality of numerical forecasts for the atmosphere and ocean due to high population density and rapid growth of urbanization, industrialization, and infrastructure development activities, which are progressing quite rapidly along the coastal belt. In the era of information technology and advancements in computational power, it is a need of the hour that demands accurate and r­eliable information on storm surges and extreme wind waves for appropriate action and timely warnings to the  coastal community. Hence, due to the complexity involved as well the beneficial aspects in terms of s­ocioeconomic implications, a detailed study is warranted along coastal areas of the Indian coast and that requires substantial strenuous research effort. Along with the  recent development and advancements in high performance computing (HPC) systems, it has now become possible to simulate very high‐resolution models for storm surges and wind waves with a reasonably high degree of accuracy. The ­importance of HPC systems in atmosphere and ocean modeling studies in terms of rapid computation is widely recognized in operational weather centers thereby aiding timely warnings and advisories during tropical cyclone events. In this context, some of the recent developments in numerical modeling include the implementation of state‐ of‐art hydrodynamic models such as the Advanced Circulation Model (ADCIRC) and Simulating Waves Nearshore (SWAN), and atmospheric models like Weather Research and Forecasting (WRF) in operational centers to obtain realistic estimates of storm surges and extreme wind waves for the affected regions during the impact of a tropical cyclone. At present, the role of an HPC system in computing power is quite evident and inevitable and allows dynamic coupling of atmosphere‐ ocean models to run ensemble predictions as well run in a real‐time mode providing realistic estimates of storm surge height, storm surge envelope, and associated wind‐ wave characteristics. In the Indian scenario, at present the coupled models (hydrodynamic model ADCIRC coupled with SWAN wave model) have proven efficacy in storm surge forecast. It involves a dynamic coupling between storm surge and wind waves through radiation stress and precisely accounts for the wave‐induced effect in the overall prediction of total water level elevation near the coast considering the combined mutual nonlinear interaction effects between storm surge, astronomical tides, and wind waves. Recent developments include a few case studies carried out using coupled as well stand‐alone models for recent very severe cyclone cases that had landfall along the east coast of India. The chapter provides an overview as well as discussions on the past studies carried

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  241

out on storm surges and extreme wind waves over the Bay of Bengal region and elucidates the recent developments carried out in this field. Though significant progress in storm surge, wind‐wave modeling, and developments in physical parameterization has been achieved in other ocean basins during the past few decades, there are gap areas that need introspection as well as require novel and innovative ideas in order to provide a reliable information and dissemination system that can save life and property during a tropical cyclone event. 17.2. ­METHODOLOGY The study first makes an analysis of the tropical cyclone activity over the North Indian Ocean basin covering various aspects on the annual frequency of cyclones, depressions, and severe and very severe cyclonic systems in the North Indian Ocean based on 121 years of data from the India Meteorological Department (IMD). The study also performs a trend analysis on tropical cyclone activity. Relevant studies on tropical cyclone‐induced storm surges over the Bay of Bengal basin are also discussed in detail. Thereafter, the progress and advancements made in storm surge modeling over the global ocean basins and in particular topical studies relevant to the Bay of Bengal basin are reported. The role of wind waves in operational sea‐state forecast and in particular their role during extreme weather events is highlighted. The progress in wind‐wave modeling studies both in context to global perspective as well in regional scale for the North Indian Ocean is discussed at length. Thereafter, the role of coupled models in an operational scenario is reported with special emphasis on wave–current interaction. The importance of coupled models for operational forecast and their efficacy in simulating realistic total water level elevations during tropical cyclone activity is highlighted. The role of continental shelf slope and width on the nonlinear interaction between storm surges, tides, and wind waves is discussed. Further, the results obtained from model simulations for four severe tropical cyclones (Thane, Aila, Phailin, and Hudhud) cases are discussed in detail. 17.3. ­TROPICAL CYCLONE ACTIVITY OVER THE NORTH INDIAN OCEAN Tropical cyclones generally form over the warm oceans and there are some favorable conditions that determine their formation and sustenance. The necessary conditions are sea surface temperature (SST) greater than 26°C, low magnitude of vertical wind shear, large low‐level vorticity, and higher midtroposphere relative humidity. It is well documented that the months May–June and October– ­ November are the seasons that produce cyclones of high

intensity. Over the Bay of Bengal region, the monthly frequency of tropical cyclone activity portrays a bimodal distribution, with the primary peak during November and a secondary peak during the month of May. It is seen that about 16% of tropical cyclones intensify into  severe cyclones, and about 7% further intensify into very severe cyclonic storms. The India Meteorological Department (IMD), the nodal weather agency under the Ministry of Earth Sciences, Government of India, has developed an E‐Atlas (Cyclone Warning Research Center, CWRC, India, 2011) that provides a concise picture of tropical cyclone activity over the North Indian Ocean basin. The E‐Atlas is a comprehensive collection of data, framed in a Graphical User Interface (GUI) based interface on all the weather disturbances that led to depressions, cyclones, and severe and very severe cyclones formation and dissipation over the North Indian Ocean region. The data period spans from 1891 until the present, covering a total period of 127 years. The historical track details are maintained by IMD and the Joint Typhoon Warning Center (JTWC) (https://www.usno. navy.mil/NOOC/nmfc-ph/RSS/jtwc/best_tracks/). The data source from JTWC is available for a period starting from 1945 onward, whereas the IMD has a data repository available for a longer duration. There are also other sources of data, such as the International Best Track Archive for Climate Stewardship (IBTrACS) from the World Meteorological Organization (WMO), which is maintained by NOAA National Centers for Environmental Information (https://www.ncdc.noaa.gov/ibtracs/). Data are provided on tropical cyclone best tracks with an objective to understand their distribution, intensity, and frequency over the global ocean basins (Knapp et al. 2010). There are several Regional Specialized Meteorological Centers (RSMCs) worldwide and other international centers that have contributed to the development of the IBTrACS global best track tropical cyclone data. The various agencies includes RSMC Miami, RSMC Honolulu, RSMC Tokyo, RSMC New Delhi, RSMC La Reunion, RSMC Nadi, RSMC Perth, RSMC Darwin, RSMC Brisbane, RSMC Wellington, China Meteorological Administration’s Shanghai Typhoon Institute (CMA/STI), Joint Typhoon Warning Center, NCDC DSI‐9635, NCDC DSI‐9636, UCAR ds824.1, and the Hong Kong Observatory (HKO). The RSMC New Delhi under IMD also contributes data on tropical cyclones for the Indian Ocean region to IBTrACS. There are some pioneering recent studies that resulted by using the IBTrACS v03r05 data (Knapp et al., 2010), such as the poleward shift in the maximum intensity of tropical cyclones (Kossin et al., 2014). In context of the Bay of Bengal region, a very recent detailed study by Sahoo and Bhaskaran (2017) resulted in the development of a comprehensive data set on tropical cyclone–induced storm surge and coastal inundation for the east coast of India. The annual distribution in the

242  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

frequency of depressions and cyclones in the North Indian Ocean region (Sahoo & Bhaskaran, 2017) for a period of 125 years (1891–2015) that best fitted with a third order polynomial representing their trend is shown in Figure  17.1. Their study (Sahoo & Bhaskaran, 2017) analyzed the past 125 years of tropical cyclone data available from IMD and mentions that a total of 1,405 cyclonic systems developed over the North Indian Ocean region (Figure 17.1a), which includes a total of 775 depressions, 332 cyclonic storms, and 298 severe cyclonic storms (CWRC, 2011). The classification is based on the maximum sustained wind speed as per the IMD norms available at http://imd. gov.in/section/nhac/termglossary.pdf and the Dvorak technique that used enhanced infrared and/or visible satellite imagery to quantify the intensity of the cyclonic system. The IMD classification or the “T” classification is used to estimate quantitatively the intensity of tropical cyclones based on the maximum sustained wind speed. For example, T1.0 is used to classify a Low Pressure System (wind speed 222 km h−1). As seen from Figure 17.1 for the 125 years of data, the postmonsoon season of October and November recorded a maximum of 238 and 204 events followed by the premonsoon season of June to August with a total count of 163, 156, and 181 events, respectively. It is interesting that the data reveal that during the period 1921– 1980, the frequencies were much higher (about 18 cyclones/year) as compared with the period from 1981 until the present (Sahoo & Bhaskaran, 2017). However, the trend in the present decade exhibits a higher frequency of very severe cyclonic storms (VSCS) as compared with the past (Figure 17.1b). Based on analysis of the 125 years of data, the annual probability of intensification in terms of percentage from depression to cyclonic storm was 44.8%, and from depression to severe cyclonic storm was 21.2%, and from cyclonic storm to severe cyclonic storm was 47.3%. The months of March‐April‐ May exhibited the highest probability of intensification (71.4%, 78%, and 69.9%, respectively) for depressions that eventually converted to cyclonic storms, and during the postmonsoon season October‐November‐December, the respective values were 50%, 67.6%, and 59.8% (Sahoo & Bhaskaran, 2017). The annual frequency of depressions, cyclones, and severe cycloni, and storms for the Bay of Bengal region is  shown in Figure  17.1c. From 1970 to the present, a decreasing trend is observed in the

tropical cyclone activity, and Figure 17.1d shows the total number of severe cyclonic storms in the Bay of Bengal. The statistics of tropical cyclone activity show that increased frequency of high intensity cyclones over the North Indian Ocean basin is a major concern for India and Bangladesh coastal regions. Singh et al. (2000, 2001) and Singh (2007) have also reported on the increasing trends in frequency of intense tropical cyclone activity over this region. The study by Srivastava et  al. (2000) focused on the low‐energetic cyclones and concluded that decreasing activity is noticed over the Bay of Bengal region in the last four decades. Other interesting studies by Wang et al. (2006) and Trigo (2006) advocate that an increased frequency of tropical cyclones can be expected in the head Bay of Bengal region as a consequence of northward shift in midlatitude storm tracks. A recent study by Kossin et  al. (2014) signifies that the recent year’s location of cyclogenesis has shifted due to global warming with a tendency of poleward shift. Their study (Kossin et  al., 2014) indicates that the poleward shift occurred at a rate of 53 and 62 km per decade in the Northern and Southern Hemispheres, respectively, however, there is an unclear trend in the shift of cyclogenesis for the North Indian Ocean basin. In the Indian context, there have been significant improvements in the operational forecasting of tropical cyclone track, intensity, landfall location, storm surge and coastal flooding, and extreme wind‐waves in recent years. The joint efforts from the operational weather centers like IMD and ESSO‐INCOIS (Indian National Centre for Ocean Information Services) under the Ministry of Earth Sciences, Government of India were quite instrumental in providing timely warnings and periodic bulletins through various modes of dissemination that resulted in a massive coastal evacuation effort during cyclone Phailin (2013). About 550,000 people from the coastal belts of Odisha and Andhra Pradesh States were evacuated to safer locations. 17.4. ­STUDIES ON TROPICAL CYCLONE–INDUCED STORM SURGES FOR THE BAY OF BENGAL One can find numerous studies in the literature that discuss the impact of tropical cyclone–induced disastrous storm surges in the Bay of Bengal. Some of the pioneering and notable studies include Murty and Flather (1994), Das (1994), Dube et  al. (1997), Madsen and Jakobsen (2004), Rao et al. (2007), and many others. Several factors that directly contribute to disastrous storm surge in the Bay of Bengal region are discussed in these studies. Most important, the convergence of the bay (funnel‐shaped), presence of wide continental shelf encompassing the deltaic environment in the head Bay of Bengal, densely populated low‐lying coastal belt, high tidal range, presence of

(a)

(b)

Yearly frequency of cyclones and depressions

Yearly frequency of very severe cyclonic storms

20

8

18

7

Frequency of very severe cyclonic storms

6

Poly. (frequency of very severe cyclonic storms)

16 14

y = –1E – 05x3 + 0.0575x2 – 111.87x + 72557

5

12 10

4

8

3

6

2

4

1

2 0 1891

(c) 18

Frequency of cyclones and depressions

1911

1931

1951

y = 6E – 07x3 – 0.005x2 + 13.098x – 10709

1971

1991

0 1891

2011

Yearly frequency of depressions, cyclones, and severe cyclonic storms in the Bay of Bengal

(d)

7 6

12

5

1971

1991

2011

y = –8E – 06x3 – 0.0466x2 – 90.557x + 58621

Annual

4

8

3

6

2

4 0

1951

Yearly frequency of severe cyclonic storms in the Bay of Bengal

14

2

1931

8

16

10

1911

Annual

y = 4E – 06x3 – 0.0222x2 + 46.678x – 32485

1891 1901 1911 1921 1931 1941 1951 1961 1971 1981 1991 2001 2011 2021

1 0 1891

1911

1931

1951

1971

1991

2011

Figure 17.1 Annual frequency of (a) cyclones and depressions, and (b) very severe cyclones in the north Indian Ocean; (c) depressions, cyclones, and severe cyclones, and (d) very severe cyclones in the Bay of Bengal (from Sahoo & Bhaskaran, 2017).

244  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

numerous riverine systems, tidal creeks, mudflats, coastal geometry, complex geomorphic environment, and so on, results in the occurrence of disastrous storm surge in the Bay of Bengal as compared with other regions elsewhere in the world. The coastal inundation that results from storm surge during tropical cyclone landfall mainly depends on the storm surge height, vegetation characteristics prevalent over the affected regions, and onshore topography of the hinterland. The disastrous effects of this natural geohazard can be minimized to a large extent through reliable numerical model predictions that provide alerts and timely warnings to the coastal communities. Modeling the prevalent hydrodynamics along the coastal environment during tropical cyclone activity is a quite challenging task due to the complex nonlinear interaction mechanism between various environmental drivers such as tides, wind waves, currents, and storm surge. 17.4.1. Progress of Storm‐Surge Modeling in a Global Perspective Studies on storm‐surge modeling started during the late 1950s. During the past six decades of extensive research and efforts, there have been tremendous developments made. However, numerical modelers have been looking forward to robust, advanced techniques, and innovative ideas to understand and predict the potential variability in tropical cyclone–induced storm surges. A comprehensive overview on the various models adopted by operational centers globally is available in the studies by Murty (1984) and Sundermann and Lenz (1983). Flather (1976) and Flather and Proctor (1983) for the North Sea, Jelesnianski and Chen (1982) for the Gulf of Mexico and Atlantic coast, and Bode and Hardy (1997) for the European coast are some of the notable studies on storm surges. Prior studies on development of storm‐surge modeling started with statistical analysis based on archived storm records. Some of the pioneering efforts in this context that used empirical formulations include the studies by Conner et al. (1957), Donn (1958), Bretschneider (1959), Welander (1961), Miyazaki et  al. (1962), Harris (1963), and Jelesnianski (1965). Continued efforts and improvements in the empirical-based models resulted in the development of the SPLASH model (Jelesnianski, 1972), which estimates storm surge for a given bathymetry and approach angle of a tropical cyclone. Nomograms that were developed using this model gained popularity and were used for real‐time storm‐surge prediction. Thereafter, in 1976 the Federal Insurance Agency developed the FEMA TTSURGE (Federal Emergency Management Agency Tetra Tech SURGE), which was recommended by the National Academy of Sciences

(Camp Dresser & McKee, 1985). Subsequent developments and improvements in model parameterizations resulted in the development of the SLOSH (Sea, Lake and Overland Surges from Hurricanes) model in 1992 by the National Weather Service (Jelesnianski et al., 1992). It was a two‐dimensional, dynamic storm‐surge model that used a curvilinear polar coordinate grid structure for spatial discretization; it was extended to elliptical and hyperbolic grids thereafter. The National Hurricane Center, USA, uses the SLOSH model for real‐time forecasts of storm surges. The US Army Corps of Engineers (USACE) also developed a one‐dimensional numerical model called DYNLET (Amein & Kraus, 1991). Further research efforts have led to the development of three‐ dimensional and depth‐averaged numerical models. Parallel to the developments in the SLOSH model another model called ADCIRC (Advanced Circulation Model) was also developed. The ADCIRC model was a  joint collaborative effort between the USACE Engineering Research and Development Center, University of Notre Dame, and the University of North Carolina, USA. The present version of ADCIRC has the flexibility to run in two‐dimensional depth integrated (2DDI) and three‐dimensional (3D) modes. It is proven as one of the most robust and reliable models worldwide for storm surge and inundation studies, and is also used by the operational centers for real‐time forecasts. More details on the governing equations and technical details are available in Luettich et  al. (1992) and Luettich and Westerink (2004). The recent developments include the coupling of ADCIRC hydrodynamic model with SWAN (Simulating Waves Nearshore) wave model available in Dietrich (2010). There are many case studies performed and available in the studies by Hubbert et  al. (1991), Powell and Houston (1996), Powell et al. (1998), Houston et  al. (1999), Fleming et  al. (2008), Blain et  al. (2008), Westerink et al.(2008), Dietsche et al. (2007), Peng et al. (2004), Xie et al. (2004), and Cho et al. (2009). 17.4.2. Progress of Storm‐Surge Modeling in the Bay of Bengal Basin The storm‐surge problem for the Indian coast started with the development of empirical relations and the studies by Rao and Mazumdar (1966) led to generation of nomograms that represented the storm‐surge amplitude as a function of storm intensity and speed. Another study by Janardhan (1967) used empirical formulations considering the static wind setup and assuming a balance between wind stress and sea‐surface slope to estimate storm‐surge height at Sagar Islands located in the head Bay of Bengal. There were several other studies that relied on empirical models such as by Chaudhury and Ali (1974), Rao and Majumdar (1966), Qayyum (1983), and

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  245

Das et al. (1978). It was only during the early 1970s that numerical studies on storm surge were attempted. A ­pioneering study by Das (1972) led to the development of the first numerical model for storm‐surge prediction in the Bay of Bengal. Later, Das (1980) introduced nonlinear advective terms in the model equations and proposed that inclusion of tide‐surge interaction into the model physics advanced the arrival time of peak surge by about 2 hr. It was probably the study by Murty and Henry (1983) that developed for the first time a series of numerical models for tide and surge that used an irregular rectangular grid instead of a regular rectangular grid. Significant progress has been made in this subject and the study by Johns and Ali (1980) and Johns et  al. (1981) included the Ganges–Brahmaputra–Meghna River system using the depth integrated nonlinear equations of motion and continuity. The SPLASH model of Jelesnianski (1972) was later adopted by Ghosh (1977) for the east coast of India. In another study, Johns et al. (1981) used the full nonlinear depth‐averaged model of Jelesnianski (1976) to investigate storm‐surge activity for the 1977 cyclone Andhra. Literature review suggests that the most complex cyclone model used to model the Bay of Bengal storm surge (Jarrell et al. 1982) in the 1980s was based on the US National Weather Service for the standard project Hurricane (Murty et  al. 1986). In this study, 258 simulations were analyzed generated from a total of ­ eight  numerical storm‐surge models, five for the Sri Lanka/India/Bangladesh region, two for the Burma/ Thailand region, and one for the Andaman Islands region. Extensive studies on storm surge were carried out using a finite difference model for the Bay of Bengal region by Dube and Gaur (1995) also popularly known as IIT‐D storm‐surge model. An elaborate overview of finite difference models is available in Dube et al. (1997). Several case studies were performed using the IIT‐D model for the Indian coast. In addition, there are several studies reported on storm‐surge models by Rao et  al. (1997), Chittibabu (1999), Chittibabu et  al. (2000, 2002), Dube et al. (2000ba, 2000b, 2004), and Jain et al. (2006a, 2006b) for the Gujarat, Andhra Pradesh, Odisha, and Tamil Nadu coasts. Recent developments include the implementation of the ADCIRC model by Rao et  al. (2010) for the Kalpakkam coast located in Tamil Nadu State to evaluate extreme storm‐surge scenarios. The ADCIRC model uses a flexible finite element mesh that is capable to resolve the complex coastline geometry as well as sophisticated model physics to compute storm surge and inundation, and hence more advantageous as compared with the IIT‐D storm‐surge model. Though the IIT‐D model fared reasonably well in many case studies, it could be used only for storm‐surge computation, unlike ADCIRC which has capability for both storm surge and inunda-

tion. Also the inundation computation in ADCRIC uses a sophisticated drying and wetting algorithm. Studies by Bhaskaran et  al. (2013), Murty et  al. (2014, 2016), Gayathri et al. (2015), and Poulose et al. (2017) pioneered storm‐surge and coastal inundation modeling for the Indian coast using the coupled ADCIRC + SWAN model, which can handle both hydrodynamics and waves. The most recent studies on storm surge and coastal inundation using coupled ADCIRC + SWAN for various severe cyclonic storm surges along Indian coasts are by Bhaskaran et al. (2013) for cyclone Thane, Murty et al. (2014) for cyclone Phailin, Gayathri et  al. (2015) for cyclone Aila and Murty et al. (2016) for cyclone Hudhud. The studies could provide the total water level elevation at the coast during a cyclonic landfall episode. At present the coupled model is used by INCOIS (Indian National Centre for Ocean Information Services) for operational forecast of storm surge and inundation in the North Indian Ocean basin. 17.5. ­CHARACTERISTICS OF OCEAN WIND WAVES AND THEIR ROLE DURING EXTREME WEATHER EVENTS The air–sea interface is a boundary between the atmosphere and ocean that is quite dynamic in nature, and the exchange of momentum, heat, gas, and particles occurs across this boundary. The wind stress that acts over the near‐surface atmospheric boundary layer imparts momentum, thereby generating wind waves or surface gravity waves having wave periods ranging between 2 to 30 seconds. Study on the characteristics of wind waves such as their generation, propagation, and dissipation mechanisms have been a subject of immense interest for several decades having significant practical applications and economic importance. In the recent past, there has been significant research on the study of wind waves and their prediction due to increasing marine and offshore activities. A precise knowledge of the sea state and its prediction is very vital for various marine‐ related operations, efficient ship routing, strategic naval operations, port and harbor development activities, coastal zone management, and so on. Nevertheless, the scientific and engineering community has a profound interest in understanding the associated kinematics and dynamics of ocean wind waves for routine forecast and case‐based studies. The engineering community working in the related disciplines of ocean engineering, naval architecture, and civil and hydraulic engineering requires precise wave‐related information to design, operate, and manage structures or natural systems in the marine environment. Ocean waves also play a significant role in controlling coastal processes in the coastal and nearshore environments. As per the existing

246  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

knowledge, wind blowing over the ocean surface generates wavelets and the spectral components eventually develop over time extracting energy from the wind stress. Through nonlinear wave–wave interaction processes, the energy within a wave system gets redistributed thereby determining the overall wave energy at a particular location and time, and that can be conveniently expressed in the form of a wave spectrum. This is the present state of knowledge acquired despite several decades of ongoing research in the field of ocean wave modeling. The random nature of ocean waves and their complex interaction mechanism in terms of their kinematics and dynamics of wave evolution was a major challenge in the past. The fundamental and classical studies on water waves with valid assumptions and developments in mathematical formulations date back to the nineteenth century. Table  17.1 provides an overview on the major advances and developments in the field of ocean wind waves during the past few decades. The pioneering studies by Gelci et al. (1957) introduced the concept of energy balance equations to understand the phenomenon of wave evolution. Since then different categories such as first, second, and third generation wave models have evolved. At present, the third generation wave models are used for routine wave forecasting, and several advancements are noticed in the parameterization of physical processes in a wave forecasting system. At present, there has been a tremendous boost in computing power, information technology, data acquisition systems, satellite remote sensing, and an increasing number of in situ observational platforms. Broadly speaking, the wave models can be classified into phase‐averaging or phase‐resolving, wherein the phase‐averaged models are expressed in terms of energy balance with appropriate sources and sinks used to represent the relevant physical processes. Phase-resolving models are based on the governing equations of fluid mechanics formulated to obtain the free surface condition. However, the phase-averaging models have no prior restriction on the area to be modeled, whereas the phaseresolving models have an inherent limitation on the spatial dimension of the computational area. The various physical processes that are accounted for in phase‐averaged models include (1) wave generation by wind accounted due to momentum transfer from atmosphere to ocean, (2) refraction due to water depth, (3) shoaling due to shallow water depths, (4) diffraction due to obstacles, (5) reflection due to impact with solid obstacles, (6) bottom friction due to heterogeneity of bottom materials, (7) wave breaking effects when steepness exceeds a critical level, (8) nonlinear wave–waveinteraction due to quadruplets and triads resulting in wave energy redistribution, and (9) wave–current interaction effects. In deep waters, the physical processes can result from the combined effects

of wave generation by wind and quadruplet wave–wave interaction and dissipation due to white‐capping mechanisms. These deep‐water waves transform on reaching shallow waters due to dominant physical processes like refraction, bottom friction, depth‐induced breaking, triad wave–wave interaction, wave–current interaction, diffraction, and reflection (Holthuijsen, 2007). Hence, choosing an appropriate wave model for the desired task is very important considering the dominant physical processes relevant to the study area. 17.5.1. Progress of Wind‐Wave Modeling in a Global Perspective Broadly speaking, there have been significant improvements in the operational aspects of regional and global ocean wave forecasting systems routinely used for medium range forecasts of ocean state variables (Tonani et  al., 2015). The GODAE (Global Ocean Data Assimilation Experiment) project played a key role in the collaboration between national groups in the development of global  ocean forecasting systems (Smith, 2006). More information is available at the website https://www.godaeoceanview.org/ on the activities related to ocean analysis and forecasting. The GODAE team has partnerships from various countries like the UK, France, Norway, Italy, USA, Australia, Canada, Japan, Brazil, India, and China  (more details on future scope of activities by the  individual countries are available in Tonani et  al., 2015). The Marine  Modeling and Analysis branch of the  Environmental Modeling Center at the National Centers for Environmental Prediction (NCEP), USA, provides information on wave forecast using the NOAA WAVEWATCH III (NWW3) run four times a day providing the hindcast and forecast information. The model products are available in global and regional nested grids. The NWW3 model products for waves include  significant wave height, wind‐sea wave height, primary and secondary swell wave height, wind speed and  direction, peak wave period, wind‐sea period, and primary and secondary swell period. The basin‐scale products cover the Atlantic, Pacific, and Indian Ocean regions. The regional scale simulations cover the northeast and northwest Atlantic, east coast of the USA, northeast Pacific, waters of Alaska, and Australia–Indonesia areas. The localized version of NWW3 includes location‐ specific areas of the waters surrounding the USA. Readers can refer to the website https://polar.ncep.noaa. gov/waves/for more details. In addition, there are companies such as the Ocean Weather Inc. (at http://www. oceanweather.com/data/), which provides services to the coastal and ocean engineering community in the areas of marine meteorology, ocean waves and currents, ocean engineering, and statistics of environmental data.

Table 17.1  Research Advances in the Field of Ocean Surface Waves During the Past Few Decades. S.No.

Advances

1940s

1950s

1

Statistical theory

Theory of random Wave statistics & noise spectral developments

2

Nonlinear theory

Nonlinear theory of regular waves

Nonlinear theory of random waves

3

Experiments (laboratory and field measurements)

Basic studies and visual based observations

Observations from instruments

4

Air–sea interaction studies and wave projects Wave forecasting Sverdrup and techniques Munk

5

Source: Mitsuyasu (2002).

Sun glitter project SMB and PNJ wave forecasting methods

1960s

1970s

Similarity form Mathematical and work on developments in directional wave spectra – spectra nonlinear effects Computation of Wave instability dispersion and wave relation interaction studies Advances in field Studies on equilibrium – based planned ocean campaigns and experiments planned experiments JONSWAP field experiment First generation wave models

1980s

1990s

2000s

High frequency wave spectrum

Wave number – frequency spectra

Wave number – frequency spectra

Wave breaking computational works

Wave breaking and energy dissipation

Wave breaking and energy dissipation

Wave dynamics – Microwave use of satellite remote sensing observations

Ocean observing systems and satellite based platforms

HEXOS

Coupled atmosphere–ocean models Third generation wave models – ensemble modeling

Second generation Third generation wave models wave models (WAM)

SWADE, RASEX Third generation wave models with data assimilation (WAM, WW3)

248  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

17.5.2. Progress of Wind‐Wave Modeling in Context of Indian Seas The ocean state forecast information for the Indian subcontinent is quite vital, having diverse application and societal benefits amongst the user community. The forecasts have inherent economic advantages varying from traditional fishing to offshore‐related activities. Besides, with the numerous major and minor ports located along the Indian coastline, the environmental information on sea state related to wind waves, swell activity, currents, and tides is critical for efficient port operations. The movement of vessel traffic and operational activities inside a port requires prior knowledge of environmental factors that aids port operations. Offshore activities such as mooring operations and loading and off‐loading of liquid and gas products to facilities located in the hinterland also requires accurate, timely sea‐state information. Recreational activities at selected coastal locations also need appropriate information of sea state for smooth operations. Many applications in the ocean environment require precise information of sea state; some of them are optimum shipping routes, erection of marine systems, search and rescue operations in the sea, defense‐related activities, oil spills, and so on. During extreme weather events, sea‐state information is imperative for offshore oil platforms and planning and evacuation measures for the coastal population. The Earth System Science Organization (ESSO)–Indian National Centre for Ocean Information Services (INCOIS) established the Integrated Indian Ocean Forecasting System (INDOFOS) for medium‐range prediction of the surface and subsurface characteristics of the Indian Ocean. The predictions have a lead time of 5–7 days at present. The activities under the INDOFOS cover a broad gamut such as surface wave forecast c­ overing aspects of wave height, period, and direction for both wind waves and swells; sea‐surface currents; sea‐surface temperature; mixed layer depth; depth of the 20° isotherm; astronomical tides; wind speed and direction; and oil spill trajectory modeling. The forecast provided by ESSO–INCOIS is widely used by the user community such as fishermen, Indian Navy, Indian Coast Guard, shipping corporations, offshore oil and gas exploration companies, and the scientific community at large. The research activities under INDOFOS have expanded and, at present, location‐ specific forecasts are available for selected areas covering the Arabian Sea, Bay of Bengal, North Indian Ocean, South Indian Ocean, Red Sea, Persian Gulf, and South China Sea. Besides, detailed forecasts are also available for potential fishing zones, union territories, and island regions of India. The operational wave models and their resolutions used at ESSO–INCOIS include the MIKE‐21 SW (1° to 0.07°), NOAA WAVEWATCH III (1° to 0.05°),

and SWAN (0.002° × 0.002°). For general ocean circulation, the ROMS model configured at a resolution of 0.125° × 0.125° is used. The General NOAA Oil Spill Modeling Environment (GNOME) Oil Spill model simulates the oil spill trajectories. The forecasted winds obtained from atmospheric models from different meteorological agencies such as NCMRWF and ECMWF force the ocean models. To improve the nearshore and coastal forecasts, models configured using nested grids are used and run in high-performance computers (HPC). The services provided by INCOIS include location specific forecasts; forecast for coastal, deep sea, and island areas; port and harbor forecast; and web map services. In addition, emergency services by ESSO–INCOIS also cover oil spill advisories, search and rescue operations, and high wave alerts for coastal regions. The forecast services for port and harbor includes 75 locations along the Indian coast and island locations. These location‐specific forecasts are subdivided into two zones, one up to 20 km and the other ranging from 20 to 50 km. In addition, ocean wave forecasts are provided to neighboring countries like the Maldives. The advisories for oil spill cover a forecast period of three days updated at an interval of every 3–6 hours based on requirement. Similarly, the services for high wave alert to coastal regions cover a forecast of 1–2 days updated every three hours. The value added service covers information on the inland vessel limits in forecast mode for one day updated every three hours. Validation of these location‐specific forecasts in near real‐time is based on the availability of satellite passes over the Indian Ocean region. ESSO–INCOIS disseminates the  information in the vernacular by various modes such as e‐mail, mobile phone, TV, radio, and electronic display  boards to the stakeholders. For areas that have  no electricity supply, the dissemination mode is through  manual display boards. Readers can refer to the w ­ ebsite http://www.incois.gov.in/ portal/osf/osf.jsp for more details. The Meteorological and Oceanographic Satellite Data  Archival Centre (MOSDAC) under the Space Applications Centre (SAC), Indian Space Research Organization (ISRO) provides the forecast map of wind‐ waves using the WAM model every six hours extended to 120 hr. The WAM‐computed parameters such as wave height, period, and direction, swell height, and wind speed form an integral part of the forecast system covering the geographical domain extending from the zero meridian to 160°E longitude, and from 70°N to 70°S in the zonal direction. The classification of sea states from the WAM computed wave heights covers five broad categories: (1) slight, (2) moderate, (3) rough, (4) very rough, and (5) high. In addition to WAM forecasts, MOSDAC also provides wave forecasts from the SWAN (Simulating Waves Nearshore) model at a six-hour interval extended to 120 hr. The domain of SWAN runs covers 60°E–90°E

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  249

and 21°N–11°S. In addition to these wave forecasts, other products such as mixed layer depth (MLD), sea level anomaly (SLA), sea surface current, temperature, and salinity using Princeton Ocean Model (POM) also form activities of MOSDAC. 17.6. ­COUPLED WAVE‐HYDRODYNAMIC MODELS Prior studies have used wave and hydrodynamic models as separate entities to simulate the flow and wave conditions over a region; most of them are case‐based studies. Coupling of wave currents as a single modeling system has been long recognized and their interaction controls the momentum and energy exchange between the atmosphere and the ocean that needs to be better resolved. The coupling of these models can be achieved at various levels of complexity. One can find a complete review on wave–current interaction mechanisms in the study by Jonsson (1990) and more recently in Cavaleri et al. (2007). The effects from waves that are considered in the coupled modeling system are due to the radiation stress and Stokes drift. Another study by Babanin (2011) also shows that interaction of turbulence and bottom stress is also important. 17.6.1. Role of Wave–Current Interaction In a broad sense the wave–current interaction can be defined as the interaction mechanism between surface waves and the mean flow. The effect from currents that includes tidal and wind-driven currents, river currents, and so on, contributes to the mean flow. The process of wave–current interaction leads to transfer of energy thereby affecting both waves as well the mean flow. In shallow water depths, the propagation of wind waves is highly dependent on the bathymetric profile and coastal hydrodynamics. When waves encounter currents in tidal inlets, at river mouths, or nearshore zones, the wave dynamics will be affected based on the speed and direction of the interacting current. The waves affect the currents mainly through the exchange of momentum flux from waves to currents. In turn, the currents can also affect the waves in different ways. It can affect the effective wind, and the fetch that in turn affects the wave generation. The effect of depth refraction and current refraction can cause changes in the wave parameters. Strong currents can have a significant influence on wave propagation characteristics. In the presence of an opposing current, the wavelength will tend to shorten, thereby causing the group velocity to decrease. To maintain conservation of energy flux, the wave energy increases, resulting in a ­localized increase of wave height. In addition, opposing currents will refract the waves such that they are focused upon the area of strongest flow, which will cause a further

increase in wave heights. In a following current, the opposite occurs, where the wavelength increases, the group velocity increases, and the wave heights are reduced. Wave period will be longer in following currents and shorter in opposing currents. Thus, the Doppler shift plays an important role in affecting the wave characteristics. The modulation of absolute frequency by unsteady currents and modulation of intrinsic frequency by propagation over spatial gradients of current can also occur. Various empirical theories for wave–current interaction in the b ­ ottom boundary layer suggest that the friction coefficient experienced by waves in a current regime will be larger than in no current. This also applies to the effective current friction factor in the presence of waves. Another effect is the vertical wind shear on wave breaking (Wolf et al., 1988). The topic of wave–current interaction is found in the studies by Ardhuin et  al. (2009), Mellor (2003, 2011), Mellor et  al. (2008), Kumar et  al. (2012), and Zodiatis et  al. (2015). Bolanos et  al. (2011, 2014) advocated the importance of the wave–current interaction in a tidal dominated estuary and showed that inclusion of wave effects through radiation stress improved the velocity structure. The wave‐induced surface and bottom stress, and radiation stress are the mechanisms through which waves interact with currents. Surface waves may also affect currents in other ways, such as through the wave‐induced Stokes’ drift and the Coriolis wave stress (Huang, 1979; Jenkins, 1987). Wave‐induced wind stress increases the magnitude of currents both at the surface and near the seabed. On the other hand, wave‐induced bottom stress weakens the currents both at the sea surface and near the seabed. Near the sea bottom, there exist enhanced levels of turbulence due to wind–wave interaction (Grant & Madsen, 1979; Mathisen & Madsen, 1996, 1999). In particular, the short‐period oscillatory nature of wave orbital velocity leads to a thin boundary layer above the bottom. In this boundary layer, the fluid velocity changes from its free stream value to zero at the bottom, where no‐slip condition applies. The high shear velocity within the wave bottom boundary layer produces high levels of turbulence intensity and large bottom shear stress. In shallow coastal waters, the near‐bottom flow consists of waves and slowly varying currents. The strong turbulence intensity within the thin wave bottom boundary layer therefore can have an impact on the currents, especially in causing the currents to experience an increased bottom resistance in the presence of waves. Using the wave–current interaction model proposed by Grant and Madsen (1979), Ningsih et al. (2000) and Xie et al. (2001) have shown that the surface waves could significantly affect the currents by modifying the bottom drag coefficient. The waves affect the wind stress by increasing the  surface roughness length. The significance of wave–current interaction

250  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

depends mainly on the water depth. The current experiences an increased bottom resistance in the presence of waves in shallow waters with high wind‐wave activity. The wind waves modify the coastal circulation through enhancement of bottom stress. 17.6.2. Role of Coupled Models in Operational Forecast The ocean state is quite complex due to the mutual nonlinear interaction between the winds, currents, and waves. During extreme events like tropical cyclones, these interactions are significant as the energy associated is quite high. The nonlinear wave–current interaction mechanisms during an extreme event results in radiation stress. Radiation stress, a term coined by Longuet‐Higgins and Stewart (1964), causes the lowering (setdown) and raising (setup) of the mean water level that is induced by waves as they propagate into the nearshore regions. This is of  immense importance in operational prediction of  the  storm‐surge heights. Radiation stress can be defined as the depth‐integrated and phase‐averaged excess momentum flux caused by the presence of surface gravity waves exerted on the mean flow. The radiation stress describes the additional forcing due to the pre­ sence of waves that change the mean depth‐integrated horizontal momentum in the fluid layer. As a result, the varying radiation stress thereby induces changes in the mean surface elevation and the mean flow. In a practical sense, the nearshore waves induce currents through radiation stress, and resultant currents conversely affect the wave field, thus wave–current interaction always takes place to a greater or lesser extent. The radiation stress changes as a wave propagates through water of changing depth. Considering the wave–current interaction during cyclones, the radiation stress is modified by both changing water depth and external force. During a cyclonic event, there is a significant radiation stress being generated. It is clear that waves always interact with currents by means of radiation stress. The current field is formulated with depth‐averaged shallow water equations. The shallow water equations include the stress term, which incorporates the radiation stress. The wave‐induced stress thereby influences the mean water surface elevation and the depth‐averaged currents. This in turn can affect the wave characteristic that modifies the radiation stress ­generated by the waves. The energy of a surface wave is dependent on the mean water surface elevation. Therefore, the radiation stress from waves affects the current, and hence the mean water surface elevation and the depth‐ averaged velocity. These variations can affect the wave parameter, which again result in a modified radiation stress. Hence, the nonlinear interaction mechanism, focused on effect of radiation stress on currents and its countereffect on waves, can be explained.

In the nearshore areas, the effect of radiation stress also contributes to wave setup. As the waves approach breaking point, there will be a small progressive setdown of the mean water level below the still water level. This setdown is caused by an increase in the radiation stress owing to decreased water depth as the waves propagate toward the shore. The setdown is maximum just seaward of the breaking point. In the surf zone, there is a decrease in radiation stress as wave energy is dissipated. This effect is stronger than the radiation stress increase owing to continued decrease in the water depth. The result is a progressive increase or setup of mean water level above the still water level in the direction of the shore. The surf zone setup typically is significantly larger than the setdown that occurs seaward of the breaking point. The wave setup is of particular concern during storm events, when higher wind waves resulting from the storm can increase the mean sea level. Hence, the radiation stress plays a major role in coastal regions. Under certain conditions, it will become very important to take the interaction effects into consideration for an accurate prediction of nearshore waves and currents, and understand aspects related to resultant sediment transport and beach change. The storm events increase the water level and enhance the risk on the coastal structures. The mutual interaction between currents and waves through radiation stress thereby play an important role in the coastal environment. Hence, a proper understanding and quantification of the nonlinear interaction mechanism is crucial. To achieve reliable estimates of this mutual interaction mechanism, it is mandatory to have very high resolution spatial grids coupled to both wave and hydrodynamic models. 17.6.3. Effect of Continental Shelf on the Nonlinear Interaction Mechanism The basin characteristics, which include the coastline geometry, relief features of the bottom such as width and slope of the continental shelf, also play an important role in the overall development of storm surges. In the context of the Indian coastline, the west coast of India has a larger continental shelf area compared to the east coast. In general, the shelf width is about 60 km in Kerala State (off Kochi), and that gradually increases to about 330 km south of Gujarat (off Daman). The shelf break occurs along the entire coastline at water depths of about 130 m. Poulose et al. (2017) performed an idealized experiment representing the west coast of India to understand the role of the continental shelf on the nonlinear interaction mechanism between storm surges, tides, and wind waves. In this context, the tidal range also increases from south to north, and a maximum of about 11 m is attained over regions in the Gulf of Khambhat. Owing to high tidal

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  251

range in the northern regions, one can expect a higher degree of nonlinear interaction between storm surges, wind waves, and tides over this region during tropical cyclone activity. The bathymetry for the west coast of India highlighting the continental shelf region is shown in Figure 17.2a and the corresponding idealized bathymetry is shown in Figure 17.2b. The computational mesh for the study region was generated using the Surface Modeling System (SMS), which was maintained to 100 m near the coast and relaxed to 15 km at the open ocean boundary (Figure  17.3) located approximately 900 km away from the coast. Major tidal constituents such as S2, M2, K2, T2, N2, K1, O1, P1, and Q1 are provided as forcing at the open boundary obtained from the FES 2004 tide model. A minimum drag coefficient for bottom friction was specified as 0.005 with an explicit scheme in time discretization at a time step of one second. Numerical experiments were carried out using the ­computational mesh (Figure  17.3) considering 13 idealized cyclone tracks making landfall perpendicular to the coast at different locations separated by a horizontal distance of approximately 100 km. The Jelesnianski wind

formulation was used along each track considering a constant pressure drop of 40 mb with 30 km as the radius of maximum winds. The forward motion was maintained as 10 km h−1, which is the average speed of a tropical cyclone. Model simulations were performed to compute storm surge in stand‐alone mode as well in a coupled mode to investigate the nonlinear interaction between surge–wave, surge–tide, surge–tide–wave scenarios independently (Poulose et al., 2017). The results obtained in the stand‐alone mode of ADCIRC simulated storm surges for the 13 different locations (in Figure  17.3) are shown in Figure 17.4. The amplification of peak surge varied from about 1.5 m in the south to about 4.5 m in the north when the continental shelf width increased from 45 to 300 km, respectively. It indicates that the rate of amplification is about 12 cm for a linear increment of 10 km in the continental shelf width. The spatial distribution of currents showed that the magnitude was about 0.5 m s−1 at the shelf break and that increased while approaching the coast (Poulose et  al., 2017). The nonlinear interaction between various components is shown in Figure  17.5,

(a)

Gulf of Kutch GUJARAT Gulf of Khambhat Approx. 330 Km continental shelf width

Mahi River Narmada River Tapi River

MAHARASHTRA

Approx. 90 Km

Zuari River KARNATAKA

ARABIAN SEA

Approx. 60 Km KERALA

Figure 17.2  (a) Bathymetry for the west coast of India and (b) idealized bathymetry for the west coast of India.

252  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION (b) Depth (m)

330 km

Shelf depth: 3 – 130 m

5590 5383 5176 4969 4762 4555 4348 4142 3935 3728 3521 3314 3107 2900 2693 2486

Shelf break

2279 2072 1865 1658 1451 1245 1038 831

65 Km

624 417 210 3

Figure 17.2 (Continued)

and that varied with the phase of the tide and also was dependent on the tidal range at a particular location. Simulated results signify that the maximum interaction occurs during the low tide and it varied between 20 and 45%, while it was minimum during the flood tide (between 15 and 30%) from track 1 to 13. The interaction effect varied between 15 and 40% for both high and ebb tide conditions as the shelf width increased from south to north. The study brings to light that by removing the tidal effect from surge–tide–wave interaction, the surge levels are still modified with the nonlinear interaction and the differences noticed from surge‐wave case alone depict the fact that continental shelf width plays an important role in generating the nonlinearity in surge–tide–wave interaction. 17.7. ­STORM SURGE AND INUNDATION MODELING FOR CYCLONE THANE This section provides the methodology developed to model the storm surge and the inundation associated with severe cyclone Thane, which made landfall in Tamil Nadu located in the east coast of India. Figure  17.6 is  the flowchart that provides an overview on the

­ ethodology adopted to model the storm surge. The m inputs are  categorized into three sets: location specific inputs, meteorological, and oceanographic inputs. Features such as the topography and coastal orientation are the location‐specific inputs, whereas the oceanographic inputs include the bathymetry and tidal amplitudes. The meteorological input refers to the atmospheric condition during the cyclonic events. This section discusses the performance of the stand‐ alone ADCIRC model for the Thane event. Cyclone Thane made  its landfall between Cuddalore and Puducherry in the  morning hours of 30 December 2011 and was one of most severe cyclones the Tamil Nadu coast had ever experienced. The coastline of Tamil Nadu has a length of about 1,076 km constituting about 15% of the total coastal length of India. The coastline is bounded on the north by Pulicat Lake and the south by Kanyakumari, which stretches over 13 ­districts. The coastal orientation of Tamil Nadu is straight and narrow without many indentations except at  Vedaranyam. Fringing and patch reefs are present near Rameswaram. The Gulf of Mannar, Pitchavaram, Vedaranyam, and Point Calimere have well‐ developed ­mangrove systems. There are many industries along the  coastal region. There are two power plants at

Track 13 Track 12 Track 11 Track 10 Track 9 Track 8 Track 7 Coast line 66 hour

Track 6

Track 5 Track 4

6 hour

Track 3 Track 2 Track 1

Open boundary

Figure 17.3  Grid structure along with the synthetic cyclone tracks. 4.5 Surge 1

4

Surge 2 Surge 3

Elevation (m)

Surge 4

3.5

Surge 5 Surge 6

3

Surge 7 Surge 8

2.5

Surge 9 Surge 10

2

Surge 11 Surge 12 Surge 13 Landfall time

1.5 1 0.5 0 –0.5 30

35

40

45

50

55

Time (h)

Figure 17.4  Storm surge evolution along various locations on the west coast of India.

60

254  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION (a)

(b) 1.2

1.2 NLT (Track 1) NLT (Track 5) NLT (Track 9) NLT (Track 13)

0.8

Landfall time

0.4 Elevation (m)

0.4 Elevation (m)

0.8

0

NLT (Track 1) NLT (Track 5) NLT (Track 9) NLT (Track 13) Landfall time

0

–0.4

–0.4

–0.8

–0.8 High tide –1.2 30

35

40

45

50

55

–1.2 30

60

Low tide 35

40

50

55

60

Time (h)

Time (h)

(c)

(d) 1.2

1.2 NLT (Track 1) NLT (Track 5) NLT (Track 9) NLT (Track 13)

0.8

0.8

Landfall time

0.4 Elevation (m)

0.4 Elevation (m)

45

0

NLT (Track 1) NLT (Track 5) NLT (Track 13) NLT (Track 9) Landfall time

0

–0.4

–0.4

–0.8

–0.8

Midebb tide

Midflood tide –1.2 30

35

40

45

50

55

60

Time (h)

–1.2 30

35

40

45

50

55

60

Time (h)

Figure 17.5  Time variation of nonlinear term (NLT) for tracks 1, 5, 9, and 13 (in Figure 17.3) during (a) high tide, (b) low tide, (c) midflood tide, and (d) midebb tide.

Meterological inputs Location-specific inputs

Oceanographic inputs Dynamic storm model

Storm surge model equations

Ennore: the Ennore Thermal Power Plant with a production capacity of 200 MW and the North Chennai Thermal Power Plant with a production capacity of 600 MW. The cyclone‐induced peak surge and also the horizontal extent of inland flooding was computed using the ADCIRC model in a stand‐alone mode. The wetting and drying algorithm in the ADCIRC model enables the estimation of inland penetration of water from the storm surge. The model‐computed water level elevation and inundation re validated against the tide gauge observations and the postcyclone survey conducted by the ICMAM‐PD, Chennai.

Numerical solution

17.7.1. Details of Cyclone Thane Sea surface elevation coastal inundation

Figure 17.6  Flow chart for the stand‐alone modeling system.

Cyclone Thane, during 25–31 December 2011, was classified as the strongest tropical cyclone of 2011 in the North Indian Ocean region (Figure  17.3). The India Meteorological Department (IMD) classified Thane as

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  255

Bay of Bengal (BOB) 05 and the Joint Typhoon Warning Centre (JTWC) as 06B, that developed as a tropical disturbance in the west of Indonesia. Further, on 25 December 2011 the system was designated as a “Depression” as it continued to move in the northwest. On the next day it was named Thane and it continued its movement in the westward direction. After a slack period of almost three days, the system intensified into a very severe cyclonic storm on 28 December, and finally made landfall on 30 December on the north Tamil Nadu coast. It weakened rapidly and dissipated over the neighboring north Kerala in the morning of 31 December 2011. During the initial phase of Thane on 25 December, the wind speed near the center reached 65 kmh−1. On 28 December, the JTWC reported Thane as a Category‐1 Hurricane on the Saffir–Simpson wind scale. The maximum sustained wind was in the order of 120 km h−1. Thereafter, it continued to intensify with winds reaching about 165 km h−1 on 29 December, 2011. Figure 17.7 shows the satellite imagery of cyclone Thane as of 29 December 2011 (0735 Z). The devastation associated with this cyclone was quite severe. There are reports that Thane left at least 46 dead in Tamil Nadu and Puducherry (Punithavathi et al., 2012). The coastal belts of Cuddalore and Puducherry were the worst affected during this extreme event. The postcyclone survey conducted by the IMD (Medha & Sunitha Devi., 2015) reported the lowest observed mean sea level pressure as 969 mb recorded at Cuddalore, with a maximum estimated wind speed of 139 km h−1. At Puducherry the maximum recorded wind was about 125 km h−1 at the time of landfall. Gale wind speeds between 120 and 140 kmh−1 prevailed over north Tamil Nadu

Figure 17.7  Satellite imagery of cyclone Thane.

and the Puducherry coast (Medha et  al., 2015). Figure  17.8 shows the track of cyclone Thane and the affected areas on the Tamil Nadu coast. The IMD postcyclone survey (Medha et al., 2015) also reported that the associated storm surge was about 1 m, which inundated low‐lying areas of Cuddalore, Puducherry, and Villupuram districts at the time of landfall. The observations made by satellite and Doppler Weather Radar (DWR) reported the maximum intensity from 0300 UTC of 29 December 2011 until 0000 UTC of 30 December. 20 N

Track of very severe cyclonic storm, Thane based on 1200 UTC of 28 December 2011

Thane cyclone affected areas in Tamil Nadu 2011

15 N

D

DD

CS

31/12

31/00

30/12

VSCS VSCS VSCS CS CS CS CS 30/00 29/12 29/06 28/18 28/12 27/18 27/12 CS

SCS

CS 27/06 27/00 CS 26/18 CS

26/12 DD 26/06 DD DD 26/00 D

10 N

25/18 D 25/12

DATE/TIME: IN UTC IST=UTC+0530 HRS

Scale 0 30 60 120 Kilometers

Cone of uncertainty in track forecast D: Depression DD: Deep depression CS: Cyclonic storm SCS: Severe cyclonic storm VSCS:; Very severe cyclonic storm

70E

75E

80E

85E

Legend High Medium Low

90E 5 N

Figure 17.8  Track of cyclone Thane in the Bay of Bengal (left) and cyclone affected areas in Tamil Nadu (right).

256  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

The radar ­bulletins issued by DWR Chennai were discontinued beyond 0600 UTC on 30 December 2011, as the cyclonic system started weakening thereafter. 17.7.2. Data and Methodology In order to achieve the best possible topography of the study region, a blended product of the GEBCO and SRTM was used. The GEBCO (General Bathymetric Chart of the Oceans) is a global digital database generated through a combination of quality controlled depth soundings survey data blended with satellite‐derived gravity data. Even though the GEBCO data cover the topography for both oceans and land, the resolution is coarse and may not serve the purpose for realistic estimates of onshore inundation over the land area. Hence, for the study area, the high-resolution SRTM (Shuttle Radar Topography Mission) data are utilized for generation of the onshore topography. The high‐resolution SRTM data are blended with the GEBCO offshore topography leading to a hybrid topographic database for the study area (Figure  17.9). The GEBCO bathymetric data with a resolution of 30 arc second are the best available high‐resolution data for ocean, and they were blended with SRTM data having a horizontal resolution of 90 m for the onshore areas of Tamil Nadu. The SMS (Surface Modeling System) package is used to generate the finite element mesh for the study area. Figure 17.10a shows the finite element unstructured mesh

for the domain used in the present study. A enlarged section of Figure  17.10a near the coast is shown in Figure 17.10b; it clearly depicts the grid structure relaxing from the coastline toward both the inland and offshore boundaries. In order to study the inland inundation due to a storm surge, the onshore grid boundary is fixed at +10 m contour assuming that the inland penetration of seawater never exceeds this contour limit. It is reasonable as in most of the cases the +10 m contour is more than 5 km inland from the coast. In the offshore region, a resolution of about 20 km is used and the grid size refines to  100 m when approaching the nearshore region. The grid comprises 68,896 nodes and 136,809 elements. A rectangular open ocean boundary is used for the present study. The rectangular shape is computationally more intensive (due to larger node numbers) compared with a semicircular shape. The advantage is it avoids the problem of computational instability at corner node points separating the mainland from the offshore boundaries. The landward boundary of the grid is fixed from the coast (zero meter contour line) to the +10 m topographic elevation contour presuming that the surge would never exceed 10 m in this region. The flexible grid refines near the shoreline (zero contour line) and relaxes both in the onshore and offshore direction away from the shoreline. The wind field used to simulate coastal inundation for the Thane event is the dynamic Holland wind field model that utilizes the best track record files from the JTWC (Joint  Typhoon Warning Centre). The Holland model

Mesh module elevation 4450.0 3900.0 3350.0 2800.0 2250.0 1700.0 1150.0 600.0 50.0 –500.0

Figure 17.9  Blended topography of the study domain.

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  257 (a)

(b)

Pondicherry

10 m topo line

Coastline

Figure 17.10  (a) The study domain and the finite element grid structure, and (b) same as in (a) and zoomed near the coast. Mesh module wind stress or velocity (74) mag 4 08:24:00 45.0 40.0 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0

Figure 17.11  The Holland wind field over the study domain at time of landfall.

calculates the wind field and provides information on sea‐ level pressure distribution and gradient winds within a tropical cyclone. The wind speed in terms of surface stress is then specified to the ADCIRC model based on the relation proposed by Garratt (1977). The computed maximum wind speed was 40.3 m s−1, which is in close

accordance with the IMD report of 38.55 m s−1 (Figure 17.11). The ADCIRC model in the present study uses the spherical coordinate system, and simulations are executed from a cold start. The hybrid bottom frictional formulation is used in the present study. It has a definite

258  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION (a) Land fall time

Observed

12/30/2011 0.00

12/30/2011 12.00

12/30/2011 0.00

12/29/2011 12.00

12/29/2011 0.00

12/28/2011 12.00

12/28/2011 0.00

12/27/2011 0.00

12/27/2011 0.00

Predicted Model

12/26/2011 0.00

1 0.8 0.6 0.4 0.2 0 –0.0 –0.4 –0.6 –0.8

(b)

12/30/2011 12.00

12/30/2011 0.00

12/29/2011 12.00

12/29/2011 0.00

12/28/2011 12.00

12/28/2011 0.00

12/31/2011 0.00

Observed Model Predicted tide

Land fall time

12/27/2011 12.00

0.6 0.5 0.4 0.3 0.2 0.1 0 –0.1 –0.2 –0.3 –0.4 –0.5

Figure 17.12 The comparison of observed, predicted, and model‐simulated water levels at (a) Ennore and (b) Nagapattinam.

advantage over a constant and quadratic bottom frictional formulation especially for nearshore areas where bathymetric gradients are rapid. The hybrid formulation treats bottom stress as proportional to the local water depth, permitting realistic estimation of onshore inundation using the wetting and drying algorithm. The weighting factor that relates the relative contribution of primitive and wave portions in the GWCE (Generalized Wave Continuity Equation) is set to 0.01, which is a recommended value. The ramp function, which is the spinup time, is set as one day for a five-day total simulation length spanning 25–30 December 2011. The model time step and eddy viscosity were set to 2.0 s and 5.0 m2 s−1, respectively. For grid nodes in the open ocean boundary, six tidal constituents are prescribed: K1, M2, N2, O1, P1, and S2 in the ADCIRC model. The combination of these tidal constituents depicts the true tidal field that exists in the Bay of Bengal region. The amplitude and phase of these six tidal constituents at the open ocean boundary synchronizes with the model simulation start time (12 Z of 25 December 2011). The progressive tidal field propagates from the open boundary marching forward with time into the nearshore areas. The model simulations are validated with two tide gauge stations located near the landfall point, one located to the north and another to the south of the landfall point. The tide gauge station at Ennore (located north) is about 180 km from landfall point, whereas the one at Nagapattinam (located south) is about 100 km away. A

comparison of the observed water level to the predicted and model simulations indicates a good match (Figure 17.12). The predicted tide (Figure 17.12) for the two locations was performed using the SLPR2 (Sea Level Processing Software) model, and model simulations were made using the ADCIRC model. The observed water level is the combined effect of the sea level, tides, and meteorological residue. The contribution from the storm‐ surge component is extracted from the observed and model simulations by subtracting the predicted tide. Figure 17.13 shows the residual component (contribution of storm surge) at these two locations. 17.7.3. Results and Discussions Model computed total water level elevations show that the entire Tamil Nadu coast was affected by storm surges with varying magnitudes. The computed maximum water level elevation was about 1.2 m north of the landfall location and extends to the Mamallapuram coast ­ (Figure 17.14). The occurrence of peak surge is noticed toward the right side of the storm track with a spatial extent of about 200 km along the coast. The model results show a lower magnitude of about 0.7 m toward the southern parts of Tamil Nadu such as Nagapattinam. The simulated maximum water level elevation matches very well with the observations reported by IMD. The horizontal distance the seawater traveled crossing the zero meter contour line is treated as the extent of inland

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  259

12/31/2011 0.00

12/30/2011 12.00

12/30/2011 0.00

12/29/2011 12.00

12/29/2011 0.00

12/31/2011 0.00

12/30/2011 12.00

12/29/2011 12.00

12/29/2011 0.00

12/28/2011 12.00

12/28/2011 0.00

12/27/2011 12.00

12/27/2011 0.00

12/26/2011 12.00

12/30/2011 0.00

Obs_res Mod_res

Land fall time

12/28/2011 12.00

1 0.8 0.6 0.4 0.2 0 –0.2 –0.4 –0.6 –0.8 –1

Obs_res Mod_res

Land fall time

12/28/2011 0.00

(b)

1 0.8 0.6 0.4 0.2 0 –0.2 –0.4 –0.6 –0.8 –1

12/27/2011 12.00

(a)

Figure 17.13  The residual at (a) Ennore and (b) Nagapattinam. Max. water level elevation 1.16 1.02 0.88 Pulicat Lake 0.74 0.6 0.46 0.32 Chennai 0.18 0.04 –0.1 Puducherry Cuddalore

Nagapattinam

Figure 17.14  Maximum water surface elevation computed by ADCIRC and the magnified view near to Mamallapuram showing inundation.

inundation. The inland inundation is very much influenced by the inland topography. Even for small surge amplitudes, a larger inundation extent can be expected depending on the topography of the location. The study performed an extensive analysis on the extent of inland intrusion at 40 different locations along the coast estimated from the model simulations (Figure 17.15). The extent of onshore inundation varied from 2 to 349 m with an average inundation distance of 47.5 m. The maximum value of 349 m occurred at Cuddalore old

town, which had close proximity to the landfall of cyclone Thane. Table  17.2 provides more details on the inundation extent along with the respective distances from the coastline. The inland intrusion of seawater was greater in the south compared with the northern part of coastal Tamil Nadu. This is due to the low‐lying flat terrain features prominent over the south Tamil Nadu coast. The study signifies that the maximum inundation occurred toward the right of the storm track, which had a strong dependence on the beach slope. The beach slope and shoreline characteristics for the Tamil Nadu coast from the field measurements conducted by ICMAM‐PD are shown in Table 17.3. It is evident from Table 17.3 that Chennai has a very wide and flat beach with a slope of 1:190, whereas some locations in the south of Tamil Nadu such as Silver Beach south of Cuddalore have a slope of 1:220 (Figure 17.16a). The study also performs a comparison between the model‐ computed inland inundations against the observations collected from field campaign (Figure  17.16b). The model‐computed inundation matches fairly well with observations. It is found that for locations having mild slopes, such as the Marine Beach in Chennai, Periyakuppam in Kancheepuram, Veeranampattinam in Puducherry, and Silver Beach in Cuddalore, the model‐computed inundation is slightly lower as compared with the field measurements (Figure 17.16). This difference can be attributed to the GEBCO bathymetric data used in the present study. The study signifies that the overall performance of ADCIRC‐computed peak surge and coastal inundation is satisfactory.

260  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Figure 17.15  Estimated inundation along the selected 40 coastal locations.

17.8. ­STORM SURGE AND INUNDATION MODELING FOR CYCLONE AILA The head bay region located in the north Bay of Bengal is a low‐lying area and highly vulnerable to impact from tropical cyclones. This region features the world’s largest delta system, the Ganges–Brahmaputra delta, and severe cyclones such as the 1970 Bhola cyclone, the 1991 Bangladesh cyclone, and others, have had significant impact on this low‐lying deltaic environment. The coastal processes over this region are highly dynamic with time. The impact from a storm can lead to significant changes in the coastal geomorphology over this region, thereby making it more vulnerable to future disasters. The region is thickly populated and extends for about 350 km across the states of West Bengal in India and Bangladesh. Nearly more than one‐third of the total

population of Bangladesh (close to 50 million) resides in this region. The high population density leads to higher risk and economic loss from storm impact. Another notable feature of this deltaic environment is the presence of numerous tidal creeks and the river inlet systems. The tidal creeks and the riverine systems allow the free propagation of storm surges over a long distance upstream, resulting in the inundation of low‐lying inland areas. Another factor that contributes to the disastrous surge effect is the tidal amplitude of this region. Therefore, an inundation map for this region is highly necessary to understand the impact of storm surges. This section discusses the storm surge and inundation scenario for the severe cyclonic storm Aila that had landfall over the head Bay region. The ADCIRC model is employed to understand the extent of inland inundation that resulted from this cyclonic event. The model simulation was also

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  261 Table 17.2  Model Computed Inland Inundation Along the Tamil Nadu Coast From Cyclone Thane. S.No.

Location

1

Pulicat

2

Kattupalli Village

3

Minjur

4

Ennore

5

Tondiarpet

6

Chennai

7

Old Washermanpet

8

Parrys

9

Adyar

10

Muttukal

11

Covelong

12

Perur

13

Pattipulam

14

Mamallapuram

15

Kalpakkam

16

Thenpattinam

17

Odiyur

18

Thazhankadu

19

Kazhikupam

20

Kuilapalayam

21

Thazhamkuda

22

Cuddalore

23

Devanampattinam

24

Cuddalore Old Town

25

Sangoli Kuppam

Geographic coordinates

Distance from the coastline

Inland intrusion (in meters)

13° 24′36″ N; 80° 11′36″ E 13° 18′18″ N; 80° 19′30″ E 13° 16′12″ N; 80° 16′12″ E 13° 12′18″ N; 80° 18′54″ E 13° 07′30″ N; 80° 17′42″ E 13° 09′00″ N; 80° 15′00″ E 13° 06′20″ N; 80° 17′24″ E 13° 05′12″ N; 80° 17′06″ E 12° 59′36″ N; 80° 15′00″ E 12° 48′36″ N; 80° 13′48″ E 12° 47′54″ N; 80° 15′00″ E 12° 42′54″ N; 80° 13′30″ E 12° 40′30″ N; 80° 12′54″ E 12° 37′48″ N; 80° 11′24″ E 12° 30′54″ N; 80° 09′18″ E 12° 24′36″ N; 80° 06′36″ E 12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E

1.4 km

29.09

1.82 km

49.10

8.2 km

46.36

0.47 km

24.58

0.45 km

9.09

4.15 km

27.61

1.36 km

26.14

1.37 km

12.28

1.86 km

62.68

0.84 km

59.68

0.51 km

16.51

0.57 km

62.74

0.54 km

50.55

0.78 km

18.61

0.71 km

38.92

1.62 km

45.85

1.27 km

38.85

0.96 km

35.88

1.28 km

14.47

1.36 km

31.31

0.23 km

24.17

4.2 km

15.23

0.49 km

14.29

1.34 km

349.83

2.16 km

17.21 (Continued )

262  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Table 17.2 (Continued) S.No.

Location

26

Parangipettai

27

Poompuhar

28

Karaikal

29

Thethi Nagar

30

Nagore

31

Nagapattinam

32

North Poiganallur

33

South Poiganallur

34

Velankanni

35

Vettaikaraniruppu

36

Kovilpathu

37

Pushpavanam

38

Thopputhurai

39

Vedaranyam

40

Kodikarai

Geographic coordinates

Distance from the coastline

Inland intrusion (in meters)

12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E 12° 18′56″ N; 80° 01′48″ E 10° 48′28″ N; 79° 49′52″ E 10° 46′12″ N; 79° 49′48″ E 10° 46′12″ N; 79° 49′48″ E 10° 46′12″ N; 79° 49′48″ E 10° 46′12″ N; 79° 49′48″ E 10° 46′12″ N; 79° 49′48″ E 10° 46′12″ N; 79° 49′48″ E 10° 46′12″ N; 79° 49′48″ E 10° 46′12″ N; 79° 49′48″ E 10° 46′12″ N; 79° 49′48″ E 10° 46′12″ N; 79° 49′48″ E

2.2 km

163.55

1.48 km

14.91

1.61 km

11.60

0.24 km

1.89

1.29 km

47.58

2.36 km

21.05

0.94 km

68.37

0.74 km

79.50

1.11 km

60.37

2.42 km

48.54

1.41 km

32.78

1.96 km

75.88

2.13 km

25.87

2.11 km

64.31

1.02 km

60.88

Table 17.3  Beach Slope and Shoreline Characteristics for the Tamil Nadu Coast. S.No.

City

Distance from the landfall point

Beach slope

Shoreline characteristics

1 2 3 4

Pulicat (Thiruvallur) Near Marina beach (Chennai) Panayur (Thiruvallur) Nemil (Thiruvallur)

196 km north 154 km north 135 km north 944 km north

1:55 1:190 1:40 1:25

5 6 7 8

Kalpakkam (Kanchepuram) Periyakuppam (Kanchepuram) Alapakkkam (Kanchepuram) Chinamudaliyar kupam (Villipuram) Tantriyan kupam (Vilipuram) Veeranampattinam (Puducherry) Nallavadu (Puducherry) Narambai (Puducherry) South of Silver beach (Cuddalore) Rajapettai (Cuddalore) Chittripettai (Cuddalore)

85 km north 46 km north 30 km north 20 km north

1:15 1:135 1:25 1:10

Wide steep beach Very wide flat beach Steep beach with resorts Steep beach, fisherman villages and desalination plant Industrial township Fisherman village Sand dunes shrimp pond Rocky shore settlements near to the coast

14 km north 9 km north 4 km north Near landfall 4 km south 11 km south 20 km south

1:30 1:105 1:34 1:17 1:220 1:30 1:20

Steep beach with sea wall Long flat sandy beach, green belt Sandy beach settlements close to beach Steep beach Flat sandy beach with tourist establishments Steep beach, coconut tress Steep beach, village establishments

9 10 11 12 13 14 15

Chittripettai (Cuddalore)

Rajapettai (Cuddalore)

South of sliver Beach (Cuddalore)

Narambhai (Pondicherry)

Nallavadu (Pondicherry)

Veeranaampattinam (Pondicherry)

Alapakkam (Kancheepuram) Chinamudaliyar kuppam (Villupuram) Tantriyan kuppam (Villupuram)

Periyakuppam (Kancheepuram)

Kalpakkam (Kancheepuram)

Nemili (Thiruvalluru)

Panayur (Thiruvalluru)

250

Near marina beach (Chennai)

Pulicat (Thiruvalluru)

Onshore inundation distance (m)

(a)

(b)

Model computed

200 Field measurement

150

100

50

0

Coastal stations

Figure 17.16  (a) Beach slope along the Tamil Nadu coast, and (b) validation of ADCIRC computed inland inundation (in meters) with field‐based measurements conducted by ICMAM‐PD along the Tamil Nadu coast.

264  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

compared with the best possible data available for this extreme event. In addition, a few locations that were significantly affected by storm surge and inundation are also tabulated. 17.8.1. Details of Cyclone Aila A low‐pressure system that formed during the morning hours of 22 May 2009 over the southeastern region in the Bay of Bengal coincided with the leading edge of the advancing monsoon current. It intensified into a depression almost 30 hours later, and the India Meteorological Department (IMD) stated its geographical location centered around 16.5°N, 88.0°E at 0600 UTC on 23 May 2009. The IMD periodically monitored this system from satellite imagery, and Doppler Weather Radar (DWR) located at Kolkata monitored its trajectory, which approached toward the West Bengal coast. The system crossed the coast near Diamond Harbor (located at 21.5°N, 88.0°E) and further dissipated over the northern region of West Bengal after 0600 UTC on 26 May 2009. It attained the intensity of a Deep Depression with wind speed exceeding 14 m s−1 on 24 May 2009 (0300 UTC) transforming into a Cyclonic Storm (wind speed exceeding 17 m s−1) during 1200 UTC on the same day. The IMD named this system Aila. Thereafter, it intensified into a severe cyclonic storm (SCS) with wind speed exceeding 25 m s−1 at 0600 UTC on 25 May 2009 just before its landfall on the West Bengal coast. The system experienced rapid intensification while approaching the  coast, retaining its intensity level of SCS for more than 12 hours after the landfall. It is notable that Aila was the only cyclone in the past two decades to cross the West

Bengal coast during the premonsoon season. The IMD  report states that Aila resulted in a storm surge exceeding 2  m along Sundarbans on the Indian coast. The surge level was about 3 m for the Bangladesh region. The astronomical tide during the time of landfall ranged between 4 and 5 m, and the cumulative effects from storm surge resulted in a total water level elevation exceeding 4 m, which severely inundated the onshore regions. The trajectory of Aila followed a northward direction (Figure  17.17) toward West Bengal, which is the usual premonsoon ­ climatology track for cyclones that have cyclogenesis in the southeast Bay of Bengal. 17.8.2. Data and Methodology The topographic features of Sundarbans include low‐ lying alluvial flat plains, numerous river channels, and tidal inlet creeks. Accurate modeling of water level elevation and inundation demands proper specification of these tidal creeks, river drainage systems, and shallow mudflats. Therefore, a high‐resolution grid that essentially captures these topographic variations and complex coastline geometry is an essential prerequisite. High‐resolution bathymetric data from the GEBCO (offshore) blended with SRTM data (onshore) of 90 m grid resolution for the study. It is believed that the hybrid data set essentially handles better representation of the inland surge penetration and coastal inundation envelope. The study area extends from Paradip in Odisha State, India, up to Chittagong in Bangladesh. The topography of the study region is shown in Figure  17.18. The model domain comprises unstructured grid elements constructed using the grid generation tool Surface Modeling System (SMS). The resultant mesh

30N 28N 26N 24N 22N 20N 18N 16N 14N 12N 75E

78E

81E

84E

87E

90E

93E

96E

Figure 17.17  Track of cyclone Aila (IMD report).

99E

102E

105E

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  265 Mesh module elevation 3000.95 2747.05 2493.15 2239.25 1985.35 1731.45 1477.55 1223.65 969.75 715.85 461.95 208.05 –45.85

Figure 17.18  The topography of the study region (offshore from GEBCO and onshore from SRTM) and the finite element mesh generated for the study.

Figure 17.19  The enlarged view of mesh near to Sagar Island.

comprises 627,191 nodes and 317,589 triangular elements specified with a rectangular shape offshore boundary. This shape is preferred over a semicircular arc boundary so as to avoid computational instability at the corner nodes. The onshore model domain extends up to +10 m contour line over land and that provides a realistic scenario of the onshore inundation. The elevation of +10 m topography is a valid assumption, as in geographical space this contour line lies almost 4–5 km

inland. There is only a remote possibility that the waterline due to seawater intrusion reaches this distance onshore. The grid design is in accordance with the GEBCO bathymetry data having a resolution of about 20 km in the offshore region, refining to horizontal resolution of about 250 m in coastal and nearshore regions (Figure 17.19). The grid resolution of 250 m is sufficient to represent well a better picture of the peak storm surge near the coast. In addition, the high‐resolution grid in

266  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

the ADCIRC model. The ADCIRC computation provides the maximum water surface elevation during the entire episode of cyclone Aila. The estimate of inundation is obtained by defining a coastline boundary and determining the horizontal extent the water has moved inland. The coastline is set as zero meter elevation from the blended GEBCO + SRTM data set. The software ArcGIS is used to obtain the inundation envelope and to visualize the inundation along the coastal stretch. From the model result, the horizontal extent of inundation along various locations is determined to identify the high risk and most vulnerable areas. 17.8.3. Results and Discussions

Figure 17.20  The actual and modified track of cyclone Aila.

coastal areas essentially takes care of the numerous river systems and tidal creeks in the study area. The choice of grid resolution used in this study is supported by Blain et al. (1998) and Rao et al. (2009). The wind field for cyclone Aila was constructed using the Jelesnianski parametric formulation based on the best‐ track data from the Joint Typhoon Warning Centre (JTWC). The computed wind speed in terms of wind stress is provided to the ADCIRC model using the relation proposed by Garratt (1977). The actual track is close to the Hooghly River and the eye of the cyclone passed close to the river stretch. Hence, the wind field over the river domain was comparatively weak. In this study, a hypothetical experiment was performed in order to study the effect of strong winds in the river stretch, and t­herefore the actual track was shifted toward the left by a distance of approximately 50 km such that the radius of the maximum winds covers the domain of the river stretch (Figure 17.20). The storm‐surge model setup was executed using 60  days hot‐start open boundary tidal forcing as the initial condition covering a period of five days (from 22 May 2009 18:00 h until 26 May 2009 06:00 h) with a prescribed time step of 10 min. The hybrid bottom friction coefficient, which considers the local water depth, is used in model simulation. The hybrid bottom friction coefficient provides a better description as compared with linear and quadratic bottom friction formulations and is more accurate in shallow waters when alternate wetting and drying condition occurs. The computation of inundation used the wetting and drying algorithm in

Model simulations clearly indicate that the entire head Bay region including West Bengal and Bangladesh were severely affected by the impact of storm surge from cyclone Aila. The maximum surge height attained an amplitude of about 4 m along the regions of Dongajara and Sundarbans region. According to the media reports and the IMD documents, the storm surge reported was 3 m along the western parts of Bangladesh, which submerged several villages. It also mentions that the resultant storm surge over Sundarbans in West Bengal exceeded 2 m. The presence of numerous river drainage systems along with several tidal creeks in the head Bay region provides free allowances for storm‐surge propagation into the river systems. Model simulations show that surge propagated into most of the river systems. The surge amplitude in rivers, such as the Matla, Bidyadhari, and Garal, reached nearly 4 m. Model computation also indicates that surge propagated up to 40 km upstream in these rivers through the numerous tidal inlets; this resulted in inundation of banks far away from the coast. The worst affected areas with highest surge amplitude were the Sundarbans along the India and Bangladesh sector. Figure 17.21 shows the maximum storm surge observed along the Sundarbans region. The results clearly indicate that maximum surge occurred not only along the coast adjoining the mainland, but also inside the river systems. The black line with an arrow head (Figure 17.21) indicates the actual track of cyclone Aila. In order to investigate the interaction between tide and surge, a comparison of the tide and surge amplitude at nine different locations in Bangladesh coast was carried out. The markings in Figure 17.22 show the locations. The comparison signifies that at most locations the  storm tide and astronomical tide matched their phase   with varying amplitudes. Observational data are an essential prerequisite to verify numerical models. Unfortunately there are no observational data a­ vailable during the Aila event to compare with the numerical simulations. Therefore, the results presented in this ­ study are a hypothetical scenario on the development of storm‐surge height and associated coastal inundation.

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  267 Storm tide (m) 3.9 3.4 3.0 2.6 2.1 1.7 1.3 0.9 0.4 0.0

Figure 17.21  The peak storm surge observed along the Sundarbans region.

Figure 17.22  Representative locations along the Bangladesh coast used for the comparison of water levels.

The  difference in surge amplitude was evident as Aila approached its landfall location. For the locations Hiron Point, Tiger Point, and Basakhali, the difference between tide and storm tide was more than 1.25 m (Figure 17.23) as the cyclonic system approached toward the coast. The rise in storm surge is evident at all the eight stations during landfall, and found to be weaker when Aila was far away from the coast. For locations such as Kuakata, Char Chenga, Chittagong, and Sandwip located far off from Aila’s track, the surge amplitude was about 0.65 m. Another notable feature is the phase difference in resurgence time at most of the locations. The water levels were constantly higher for storm‐tide simulation almost four days prior to landfall at Char Jabar location. The relative water level heights were lower as compared with the remaining stations (Figure 17.23). At Char Jabar, the

water level e­ levation was about 0.6 m prior to landfall. The buildup of water level is evident increasing to about 1.0 m, as obtained from model s­ imulations ­during landfall. In particular, the station Char Jabar is located in a very shallow water depth environment and is sheltered by several shallow shoals and island barriers (Figure 17.22), unlike the case with the remaining stations. The shallow nature of this location along with island barriers itself provides a complex pattern of tidal propagation. The difference in water level compared with the remaining stations is due to the choice of the grid resolution. It deciphers the fact that there is a demand for high‐resolution grids in these areas to model shoals of smaller dimensions. The comparison of storm‐tide amplitude (Figure 17.24) indicates a significant phase difference in the storm‐surge characteristics. The locations Basakhali and Hiron Point

Water level (m)

3.5 Basakhali: tide 3 Basakhali: storm tide 2.5 2 1.5 1 0.5 0 5/2 5/2 5/2 5/2 5/2 5/2 5/2 5/2 –0.5 5/2 5/2 4/2 5/2 4/2 3/2 3/2 2/2 6/2 6/2 00 00 00 00 00 –1 00 00 00 0 09 91 90 90 91 90 91 91 91 0:0 2:0 –1.5 :00 :00 2:0 :00 2:0 2:0 2:0 0 0 0 0 0 0

(b)

Time (UTC)

(d)

(c) 2.5

Tiger point: tide Tiger point: storm tide

2.5 2 Water level (m)

3 Water level (m)

3.5 Hiron point: tide 3 Hiron point: storm tide 2.5 2 1.5 1 0.5 0 5/2 5/2 5/2 5/2 5/2 5/2 5/2 5/2 –0.5 5/2 5/2 4/2 5/2 4/2 6/2 3/2 6/2 3/2 2/2 00 00 00 00 00 00 00 –1 00 00 91 91 90 90 91 90 90 91 91 2:0 2:0 –1.5 :00 :00 2:0 :00 :00 2:0 2:0 0 0 0 0 0 Time (UTC)

Water level (m)

(a)

2 1.5 1 0.5 0

5/2 5/2 5/2 5/2 5/2 5/2 5 5/2 5/2 –0.5 /22/ 4/2 5/2 3/2 4/2 3/2 6/2 6/2 5/2 20 0 00 0 0 0 00 00 0 0 09 09 09 09 09 91 91 91 –1 9 0:0 0:0 12 12 0 0 2 2 2:0 :00 :00 :00 :00 :00 :00 0 0 0 Time (UTC)

Kuakata: tide Kuakata: storm tide

1.5 1 0.5 0 5/2 5/2 5/2 5/2 5/2 5/2 5/2 5/2 3/2 4/2 5/2 3/2 6/2 6/2 –0.5 5/22 4/2 5/2 00 /20 00 00 00 00 00 00 00 9 9 9 9 9 91 09 91 91 1 0 0 0 0 –1 2:0 :00 :00 :00 :00 2:0 2:0 2:0 12 0 :00 0 0 0 Time (UTC)

Figure 17.23  The comparison of tide and storm tide at selected locations.

(e)

(f) 1.2

Char chenga: tide Char chenga: storm tide

Water level (m)

1.5 1 0.5 0 –0.5

5/2 2/2

00 9

–1

5/2 3/2 12 :00

00 9

5/2 3 0:0 0

5/2 4

5/2

/20 /20 09 09 0:0 12 :00 0

4/2 00 9

Char jabbar: tide Char jabbar: storm tide

1 Water level (m)

2

5/2 12 :0

5/2

0

5/2 00

5/2

5/2 6/2

00 91 2:0 :00 0

90

5/2 00

90

6/2

:00

0.8 0.6 0.4 0.2 0 5 –0.2 /22/ 20

00

91 2:0

0

5/2 09

3/2

12 :0

5/2

00

0

90

3/2

5/2

00

:00

91 2:0

4/2

00

0

5/2

4/2

90

:00

5/2

00

0

5/2 90

5/2

:00

5/2 6

00

91 2:0

5/2

/20

09

0

6/2

0:0

00

91 2:0

0

0

(h) 2

Chittagong: tide Chittagong: storm tide

1.5 1 0.5 0 –0.5 –1

5/2

5/2 5/2 5/2 5/2 5/2 5/2 5/2 4/2 5/2 5/2 6/2 4/2 6/2 3/2 /20 0 0 00 00 00 0 00 0 0 0 09 09 09 90 90 91 91 9 91 0:0 0:0 12 12 : : 2 2:0 2:0 0 0 :00 :00 :00 0 0 0 0 0 0

2/2

5/2 3

Sandwip: tide Sandwip: storm tide

1.5 Water level (m)

2 Water level (m)

91 2:0

Time (UTC)

Time (UTC)

(g)

5/2

00

1 0.5 0 –0.5

5/2

2/2

00

–1

5/2 91 2:0

3/2

0

00

5/2 90

3/2

:00

5/2 00

91 2:0

4/2

0

00

5/2 90

5/2

4/2

:00

00

91 2:0

Time (UTC)

Time (UTC)

Figure 17.23 (Continued)

5/2

0

5/2

00

90

:00

5/2

00

5/2 91 2:0

6/2

0

5/2

00

90

6/2

:00

00

91 2:0

0

270  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION 3.5 3

Basakhali Hiron point Tiger point Kuakata

2.5

Storm tide (m)

2

Char chenga Char jabbar

1.5

Sandwip Chittagong

1 0.5 0 –0.5

5/2

5/2

5/2

2/2

3/2

3/2

00

91 2:0

90

0

–1

4/2

4/2

90

:00

0

5/2

5/2

5/2

00

00

91 2:0

:00

5/2

5/2

00

00

5/2

00

91 2:0

90

0

6/2

6/2

00

00

91 2:0

:00

5/2

5/2

00

90

91 2:0

:00

0

0

Time (UTC)

–1.5

Figure 17.24  Model computed amplitude and phase difference of the storm tide component at various locations. 1.6 1.4

Hiron point Basakhali

1.2

Tigerpoint Kuakata

Surge (m)

1

Char madras Char chenga

0.8

Char jabbar Sandwip

0.6

Chittagong

0.4 0.2 0 5/2

5/2

2/2

3/2

00

91 2:0

0

5/2

00

90

3/2

:00

5/2

5/2

4/2

00

91 2:0

0

00

4/2

90

5/2

00

:00

5/2

5/2

91 2:0

0

00

90

5/2

:00

5/2

5/2

6/2

00

91 2:0

0

00

6/2

90

:00

00

91 2:0

0

Time (UTC)

Figure 17.25  The surge alone component at different coastal locations from the time of storm genesis to dissipation.

(Figure  17.24) exhibited a good match in both surge amplitude and phase. In addition, the surge amplitude decreased as one progresses from west to east, proportional to the relative distance of respective locations from the track of cyclone Aila. The phase difference in surge characteristics (Figure 17.24) is evident as one progresses east from Aila’s track. During the landfall time, the surge amplitude was 2.75 m at Hiron  point and that reduces to 1.1 m at Char Chenga. Figure  17.25 provides the time series information on the potential storm surge at all the nine locations marked in Figure 17.22. The four stations located nearshore (Basakhali, Hiron Point, Tiger Point,

and Kuakata) e­xperienced higher surges, compared with the remaining five stations located within the riverine environment. The model could represent the time lag in occurrence of peak surge (Figure 17.24) at these four stations. The most devastating effect of storm surges is the inundation along the coast. The inundation volume depends mostly on the surge amplitude, but the slope of the beach is also a determinant of the inundation extent. The importance of beach slope on coastal inundation was discussed by Bhaskaran et  al. (2014). The complex ­ to­pography and geometry of the head Bay region is a real challenge in the overall estimation of inundation. Most

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  271

of this area comprises numerous river creeks and tidal systems. In order to determine the extent of inundation, the horizontal distance that the water moved inland from the coastline needs to be estimated, and to obtain the realistic estimates of inland inundation it is an essential pre‐requisite to represent accurately the coastal geometry of small island bodies. The most complex task is to define the coastline from where the inundation needs to be estimated. The study defined the coastline as the zero meter contour obtained from blended SRTM‐GEBCO data.

The results from this study indicate that the Sundarbans region experienced high inundation with seawater progressing up to half a kilometer inland. The inundation obtained from the model results shows that the entire coastal stretch including India and Bangladesh are inundated. The severe inundation was observed at the location of maximum surge. The low‐lying topography of this region and the propagation of storm surge through the rivers intensified the inundation scenario. Figure 17.26 shows the inundation of the Sundarbans region. The patches in the nearshore

(a)

(b)

Figure 17.26  (a) Model computed onshore inundation for the entire head Bay, (b) inundation envelope for the Sundarbans region marked as a rectangle in (a).

272  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION (a)

(b)

Figure 17.27 Comparison of inundation scenario (a) MODIS imagery of onshore inundation, (b) ADCIRC computed inundation for cyclone Aila.

regions and immediate vicinity seen in Figure 17.26a indicate indicate the scenario of inland inundation over Sundarbans and the nearby regions. The inundation over this region is quite high owing to low land elevation. Brakenridge et  al. (2013) investigated the flooding for cyclone Aila using satellite mapping techniques. Their study used eight MODIS images to obtain the flooding map of the Ganges–Brahmaputra delta. The inundation map as reported by Brakenridge et al. (2013) is shown in Figure 17.27a. The study also performed a comparison of inundation from the model computed result (Figure 17.27b) with the satellite imageries (shown in Figure 17.27a).

The patches in both the images indicate the inundation area. The MODIS observation indicates the flooding to be intense; however, the imagery is four days after Aila’s passage. The imagery also represents the flooding that resulted from heavy rainfall, which the current modeling study does not account for. The inundation extent for certain locations along the head Bay region is identified to assess the impact of inland inundation. Flooding occurred mostly along the eastern regions of West Bengal and the western parts of Bangladesh, mainly the Sundarbans area. The inundation over eastern regions of Bangladesh was very minimal.

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  273

> than 600 m 500 – 600 m 400 – 500 m 300 – 400 m 200 – 300 m 100 – 200 m

Figure 17.28  Storm‐surge affected areas and associated onshore inundation range.

The computed horizontal extent of inundation shows that locations cover mostly the regions in the Indian coast and Sundarbans region of Bangladesh. The Sundarbans area is a region of high biodiversity and of ecological importance, and for cyclone Aila it is the region that experienced the highest inundation extent. The places in the Sundarbans area named SNP (Sundarbans National Park Location) as human settlements are not available. The region is considered for analysis owing to its low‐ lying topography and  rich biodiversity. The average horizontal extent of the flooding in the identified locations reached nearly up to 350 m. For locations such as Bakkali, Bindapadmapur, Brojaballabpur, and Lothian Island, the inundation extent exceeded half a kilometer. The inundation extent over Sundarbans was higher ­compared with other locations. Figure  17.28 identifies the locations inundated between 200 and 700 m. It can be seen that most of the locations are inundated to an extent of 200–400 m range. The distance that water propagates inland depends not only on elevation of the land, but also on the vegetation characteristics of that region. An accurate estimate of inundation is always challenging as the data do not account for the type of vegetation and the friction it offers to the water flow. At some locations, the presence of buildings or some thick vegetation can divert the water to a region that is not vulnerable. Proper modeling of inundation requires accurate data of the coastal topography, the river systems, and the tidal amplitude in addition to storm‐surge characteristics.

17.9. ­COUPLED MODELING SYSTEM FOR CYCLONE PHAILIN Along the east coast of India, Odisha State experiences the highest impact of land‐falling tropical cyclones and associated storm surges. During the period 1897– 2007, it was reported by Jain et al. (2010) that about 70 cyclones struck the coast of Odisha. In addition to the highest number of land‐falling cyclones, the Coastal Vulnerability Index (CVI) for the Odisha coast attributed to other factors such as the rate of shoreline change, rate of sea‐level change, coastal slope, significant wave height, and tidal range also remains high. A study by Sreenivasa Kumar et  al. (2010) analyzed the vulnerability for this coast based on all the above mentioned parameters. The CVI based on these factors suggests a low vulnerability level along the coastal stretches of Ganjam, Chilka, and Southern Puri. The vulnerability is higher for a stretch of 107 km extending from northern Puri to Balasore. Another study by Rao (1968) also supports the fact of high vulnerability levels for the coastal stretch from Puri to Balasore in respect of storm surges. The 480 km coastline of Odisha spreads over six coastal districts that include Ganjam, Puri, Jagatsinghpur, Kendrapara, Bhadrak, and Balasore. In terms of cyclone intensity, the very severe cyclonic storm Phailin that made landfall over the Odisha coast during October 2013 was next to the 1999 super cyclone. Even though Phailin resulted in severe destruction during and

274  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

after landfall, the extent of destruction caused due to loss of life was averted due to better operational prediction skills and effective evacuation management measures. Timely warnings and alerts issued by the Indian Meteorological Department (IMD) and Indian National Centre for Ocean Information Services (INCOIS) helped the disaster management team to evacuate the human population along the coastal belt to safer locations. It was a major evacuation effort witnessed for the Indian coast and about 500,000 people were relocated to safer locations. Keeping in view the vulnerability and risk associated with the Odisha coast due to the high frequency of cyclonic storms, there is a need to develop a location‐specific operational forecasting system for such extreme events. This section demonstrates a coupled hydrodynamic‐wave modeling system to predict storm surge associated with cyclone Phailin. Extreme events like cyclones are always associated with strong winds. It drags the water toward the coast resulting in storm surge and associated inundation. In addition, the strong winds also result in extreme waves. Wave breaking effects along the nearshore regions will transfer the radiation stress to the underlying water column. Hence, the waves and the currents interact nonlinearly with each other by transferring momentum flux. High radiation stress produced during such extreme events can result in an increased water level elevation near the coast. The stand‐alone storm‐surge models fail to capture this interaction, and that can result in the modification of storm‐ surge amplitude. Hence, a coupled wave‐hydrodynamic model is a necessity to understand the mutual nonlinear interaction effects due to combined wave–current interaction mechanism. A methodology was developed to couple the hydrodynamic and wave model by including the radiation stress component. The tight coupling approach was adopted to achieve the best quality results. A tight coupled approach refers to the use of the same computational grid for both wave and hydrodynamic model. It reduces the error accounting due to the interpolation of data between two different grid systems. Moreover, the online coupling procedure could efficiently transfer the information between these two models in a prescribed time step. This implies that the calculation at each time step considers the information from both the hydrodynamic and wave model simultaneously. The study employed the coupled ADCIRC‐SWAN model for the simulation of storm surge associated with cyclone Phailin. Figure 17.29 provides the flow chart of the coupled modeling system used in this study. The SWAN model is driven by wind field, water level, and surface currents, all generated by the ADCIRC s­ imulation output. The ADCIRC model interpolates the spatial wind fields to the computational vertices, and  passes this information to the SWAN model. In a tight coupling mode of ADCIRC + SWAN, the ADCIRC model in the pre-

Start

Tight coupling and dynamic interaction between ADCIRC + SWAN

Initialization of MPI & arrays

Read the input parameters Message passing between models

Coupler

Compute wave parameters Computation of tides and associated storm surge

Compute radiation stress

End Parallel Unstructured SWAN model

Parallel ADCIRC model

Figure 17.29  Flow chart of the modeling system.

scribed coupling time interval uses the radiation stress from SWAN to extrapolate forward the wave forcing in time. After the completion of a coupling interval time step, the ADCIRC model passes on the wind velocity, water levels, currents, and roughness length information to the SWAN model. The gradient at each element is then projected to the vertices taking the area‐weighted average of gradients on elements adjacent to each vertex. The information on water levels and ambient currents thus computed by ADCIRC is shared with SWAN, thereby updating the time‐varying water depth and related wave processes such as wave propagation and depth‐induced breaking. In other words, the ADCIRC model is codriven by gradients of radiation stress (τs) computed from the ­ SWAN model, and mathematically expressed in the form: sx ,wave



sy,wave

Sxy

Sxx x Sxy

y S yy

x

y



where, Sxx, Sxy, and Syy are the wave radiation stresses.

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  275

Both the ADCIRC and SWAN models run on the same local submesh and cluster core. The two models march ahead with time and each forced with mutual exchange of information. The SWAN model uses the sweeping method to update wave information at the computational vertices, which makes it possible to handle larger time steps ­ compared with the ADCIRC model, which is limited by diffusion and Courant number time step, due to semiexplicit formulation, and also for implementing its wetting and drying algorithm. Therefore, at each coupling interval, ADCIRC runs first assuming that along nearshore and shallow coastal regions the wave properties are more dependent on circulation. 17.9.1. Details of Cyclone Phailin Cyclone Phailin was one of the strongest tropical cyclones, which affected the Odisha coast in October 2013. It was a very severe cyclonic storm that made landfall near Gopalpur with wind speed reaching up to 215 km h−1 and a central pressure of about 940 mb (Figure  17.30a). The system intensified into a deep depression while moving in the west‐northwestward direction on the morning of 9 October. On the evening of the same day, the system further strengthened into a cyclonic storm. Thereafter, the system rapidly intensified from a cyclonic storm to a very severe cyclonic storm (a)

during 10 and 11 October with an increase in wind speed from 83–204 km h−1. On 12 October, the maximum wind speed reached nearly 213 km h−1 and the system made its landfall with the same intensity. The system later weakened into a well‐marked low over southwest Bihar. Figure 17.30b shows the track of cyclone Phailin. The IMD report mentions that the system resulted in heavy rainfall over Odisha State leading to floods and the strong gale winds resulting in damage to the structures and buildings. A maximum storm surge of around 2–2.5 m above astronomical tide was reported along the low‐lying areas of Ganjam district. The storm resulted in a major evacuation effort by the National Disaster Management Authority for the coastal states of Andhra Pradesh and Odisha. 17.9.2. Data and Methodology The domain for the modeling study extends from Paradeep in the north to Puducherry in the south covering nearly 1,000 km of the coastline. The bathymetric data for the study domain were extracted from the General Bathymetric Charts of the Ocean (GEBCO) data downloaded from the British Oceanographic Data Centre (BODC). The surface water modeling system (SMS), software by Aquaveo, was used to generate the finite element mesh for the study region (Figure 17.31). The construction

(b) 10/14/2013 0000 UTC

10/13/2013 0600 UTC

Bhadrak

Paradeep

Ganjam Gopalpur

10/12/2013 1200 UTC

Visakhapattanam 10/11/2012 1200 UTC 10/11/2013 0000 UTC

Nellore Chennai Puducherry

Figure 17.30  The satellite imagery of cyclone Phailin.

10/10/2013 1200 UTC

10/10/2013 0000 UTC 10/9/2013 1200 UTC

10/9/2013 0000 UTC

10/8/2013 1200 UTC

276  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Bathymetry(m) 3954 0

Figure 17.31  The finite element mesh of the study region.

of mesh has a very fine resolution near to coastal region relaxing to coarser resolution in the offshore boundary. Along the coastal regions, a resolution less than 1 km was chosen and that gradually increased to about 30 km in the offshore region. This unstructured fine resolution mesh has the capability to resolve the sharp gradients in bathymetry along the nearshore regions. The rectangular boundary covers an offshore distance of approximately 750 km (Figure 17.31). A recent study by Bhaskaran et al. (2013) suggests that a high‐resolution flexible mesh in nearshore areas could essentially resolve the complex bathymetry, and thereby provide a better resolution for wave transformation. The criteria in fixing grid resolution of 1 km nearshore is justified based on the study supported by Rao et al. (2009), which highlights that a grid resolution of 1 km is sufficient and good enough for precise computation of storm‐surge height along the east coast of India. The capability of the unstructured grid over a rectangular domain was understood from earlier case studies of cyclones Thane (section  17.7) and Aila (section  17.8). Hence, for this coupled model run, the same design criteria were adopted. The computational time of model simulation is significantly affected by the number of nodal points in the unstructured mesh. A high‐resolution mesh demands more computational time and can serve as a limitation for operational studies. The grid structure used in the present

study is optimized considering the constraints in computational time and therefore can also be used effectively for operational purposes. The coupling of hydrodynamic and wave model takes into account the wave radiation stress and the corresponding wave setup in the water column. The stand‐alone version of the ADCIRC model fails to include the contribution from the waves. In order to understand the contribution from wave setup, a comparative study with and without wave setup was attempted. Hence, two model runs were made, one with a stand‐alone ADCIRC model, and the other with ADCIRC and SWAN coupled. Both the models were run for a period of 162 hours starting from 0600 UTC of 6 October 2013 up to the landfall and weakening of the cyclone. The models were forced in the open ocean boundary using the water level data obtained from the Le Provost tidal database using 13 tidal harmonic constituents: K1, M1, N2, O1, P1, S2, K2, L2, 2N2, MU2, NU2, Q1, and T2. The Jelesnianski parametric wind formulation was used to generate the cyclonic wind field. The Garratt (1977) formulation was used to provide necessary wind stress to the ADCIRC model. The bottom friction coefficient used in the coupled model was 0.0028 with 10 s as time step. The coupling interval between the models was determined based on the SWAN time step, as SWAN is an unconditionally stable model allowing higher time step. In this

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  277 Wind speed (m s–1) 61.7 54.9 48.1 41.3 34.5 27.7 20.9 14.1 7.2 0.4 Wind velocity (m s–1) 61.70 m s–1 0.44 m s–1

Figure 17.32  Phailin wind field generated using the Jelesnianski formulation.

study, the SWAN time step and coupling time interval was set to 600 s. The study by Bhaskaran et al. (2013) supports that the coupling time interval of 600 s performed fairly well in capturing the nonlinear interaction. The SWAN model was implemented in a nonstationary mode prescribed with 36 directional and 34 frequency bins. The logarithmic frequency bins ranged from 0.04 to 1.0 Hz, whereas the wave directions have an angular resolution of 10°. The quadruplet nonlinear wave–wave interaction was computed using the discrete interaction approximation (DIA) technique. The formulation of Madsen et al. (1989) was used to describe the bottom friction formulation in SWAN that permits spatially varying roughness length, and thereby the bottom friction coefficient. The value of 0.05 was used as the bottom roughness length scale. 17.9.3. Results and Discussions The coupled ADCIRC‐SWAN model estimates the wind‐wave characteristics in addition to the surface elevation and currents. The model simulation estimated a maximum wind speed of 60 m s−1. The report by IMD was in accordance with the model simulations. IMD reported a wind speed of 58 m s−1 during the landfall time with a pressure drop of 66 mb. Wind speed in excess of 25 m s−1 was experienced over the entire coastal stretch of Odisha. Even the northern districts of Andhra Pradesh experienced strong winds. The wind field for the study domain during the landfall time of cyclone Phailin is shown in Figure 17.32.

The first set of experiments was performed using ADCIRC in a stand‐alone mode including the effects of astronomical and meteorological forcing. In the second experiment, a coupled run was performed including the wave radiation stress from SWAN tightly coupled into the ADCIRC model. The model‐computed significant wave height and water level elevations are compared with the available field observations. During this extreme event, the observational data for waves were available only from the wave rider buoy off Gopalpur coast. The available observational data corresponds to the time when the storm intensified from a severe cyclonic storm to a very severe cyclonic storm. In case of water level elevation, the in situ data were available from the tide gauge located off Paradeep. These data were used to validate the model‐simulated water level elevations. Even though the wave characteristic at a location is significantly governed by the wind effects, a contribution from distant swells can also be expected. In the Bay of Bengal region, the presence of distant swells from a synoptic storm in the Southern Ocean was reported in studies by Nayak et  al. (2013) and Sandhya et  al. (2014). The influence from distant swells can significantly impact the locally‐generated wind waves. However, in the present simulation, the effects of distant swells are neglected in the study domain. The study signifies that the significant wave height simulated by the model is in close conjunction with the in situ buoy observation off Gopalpur coast (Figure 17.33). The gradual increase in wave height as the

278  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Significant wave height (m)

8 7 Without swells Observation

6 5 4 3 2 1 10

/10

10

/20

13

/10

6:0

0

10

/20

13

10

/10

12

/11

/20

:00

13

18

:00

10

/20

13

/11

10

0:0

/20

0

13

/11

6:0

0

10

/20

13

/11

12

10

/20

:00

13

10

/12

18

/12

/20

:00

13

0:0

0

10

/20

13

/12

6:0

/20

13

0

12

:00

Time (UTC)

Figure 17.33  Comparison of model computed significant wave height against buoy observations. Time of landfall

Residual water level (m)

3.4 2.9 2.4

Res-adc Res-adcswan

1.9 1.4 0.9 0.4 –0.1

10 /11 /2

10 /12 /2

01 30

10 /11 /2 :00

01 31 2:0

0

10 /13 /2

10 /12 /2

01 30

01 31 2:0

:00

0

01 30

10 /13 /2 :00

01 31 2:0

0

Time (UTC)

Figure 17.34  The water level at Ganjam for coupled and uncoupled run.

storm approaches close to the coast is evident from the model simulation as well in the observations. The coupled model computed water surface elevation at Ganjam for the simulation time period as compared against the stand‐alone ADCIRC water surface elevation (Figure  17.34). The IMD reported that the location Ganjam, situated toward the right side of the track, experienced severe storm surge. The residual water level from the stand‐alone ADCIRC run with only forcing from tides and wind stress estimates the residual water level up to 2.4 m during the landfall time. The effect of the wave radiation stress and the mutual interaction between the waves and current increased the water level at the landfall time by about 0.5 m. The coupled model showed a slightly higher water level throughout the simulation period with

a maximum of 2.9 m at the landfall time. These results indicate the fact that wave‐induced setup contributed about 23% to the peak storm surge. Unfortunately, there were no wave and water level observations available at Ganjam location to verify the model computation. From the network of observations, Paradeep was the only nearest available location to verify the water levels and is located approximately 220 km north of Gopalpur. The model‐computed surge residual for the Paradeep location is shown in Figure 17.35. The surge amplitude at Paradeep is reported to be about 0.8 m. The model results from the coupled run estimated surge amplitude of about 0.75 m. The figure clearly shows an increase in water level for the coupled model runs. It is clear from the figure that the surge residual from the coupled model run is closer to

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  279 Time of land fall

0.8 Residual (ADCIRC alone)

Residual water level (m)

0.7

Residual (coupled model)

0.6

Residual (observed)

0.5 0.4 0.3 0.2 0.1 0 –0.1

10 /10 /2

10 /10 /2

01 30

:00

10 /11 /2

01 31 2:0

0

10 /11 /2

01 30

:00

10 /12 /2

01 31 2:0

0

10 /12 /2

01 31 2:0

0

10 /13 /2

01 30

:00

10 /13 /2

01 31 2:0

0

01 30

:00

Time (UTC)

Figure 17.35  Model-computed storm surge for the Paradeep location.

the observation than the stand‐alone model run. During the time of landfall, the stand‐alone model run underestimated the surge level by 0.3 m, whereas for the coupled run the difference between observation and model is less than 0.08 m. The contribution from the wave radiation stress amplified the surge level by 0.25 m at Paradeep during the time of landfall. It is estimated that the wave‐ induced setup contributed to about 36% of the total water level elevation. A slight underestimation of the peak surge is observed even for the coupled model run. This marginal error can be related to the resolution of the coastal bathymetry as well as the meteorological forcing. The coupled model highlighted the contribution from the wave‐induced setup. The water level data at Ganjam and Paradeep indicate a contribution of about 23% and 36% from wave setup, respectively. Figure  17.36 shows the spatial distribution of storm surge with uncoupled and coupled runs. It is very clear from the surge distribution that a major portion of the Odisha coast is influenced by storm surge. The peak surge occurred toward the right side of the storm track. Figure  17.35a showed a peak surge of about 2.3 m at Ganjam using the uncoupled model run. For the coupled model run, the computed peak surge value attains up to 2.8 m. It is also noticed that the spatial alongshore extent of surge levels was higher in the coupled model run as compared with the ADCIRC simulation using a stand‐alone mode. Even the difference of 0.5 m is significant in context to the inland inundation extent. Higher surge amplitude and stronger waves could further push the extent of inundation. The role of wave setup in affecting the water level during extreme events is clear from the model simulations. The study thereby emphasizes the fact that realistic estimates of storm surge are possible when wave models are coupled with hydrodynamic models. It also needs to be noted that the contri-

(a)

Surge(m)

2.3 0

(b)

Surge(m)

2.8 0

Figure 17.36  Peak storm surge computed at Ganjam from (a) uncoupled and (b) coupled model runs.

bution of the wave setup is governed by several other factors like coastal geomorphology, translational speed of cyclones, and so on.

280  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Evolution of wave setup induced water levels along the coast from Gopalpur to Paradeep 0.6 10UTC of Oct12 12UTC of Oct12

Wave tup induced water level (m)

0.5

13UTC of Oct12 14UTC of Oct12

0.4

15UTC of Oct12 (Time of Iandfall) 16UTC of Oct12

0.3

0.2

0.1 0

50 Gopalpur

100

150

200

Distance (km)

250 Paradeep

Figure 17.37  Alongshore time evolution of wave induced setup between Gopalpur and Paradeep.

A detailed analysis on the time evolution of wave setup along a stretch of 250  km between Paradeep and Gopalpur is investigated to understand the nature of wave setup. Over the period of model simulation, the temporal modification of wave setup as a function of distance from the landfall point is discussed. Figure 17.37 shows the variations in water levels due to wave setup that extend from Gopalpur to Paradeep along the Odisha coast. It can be seen that the water level is much higher at Gopalpur compared with Paradeep, with a peak near Ganjam. Ganjam was the closest location to landfall and higher wave‐induced setup is observed at this location. The wave setup increased from 0.4 to 0.58 m at landfall time. Also at Paradeep, the time evolution of the wave set is evident. The wave setup‐induced water level increased from 0.2 to 0.35 m. Up to a distance of 140 km, the effect of wave setup is prominent, and beyond it the effect reduced considerably. Figure 17.38 provides the spatial distribution of wave‐induced setup along the coastal belt. The distribution of wave‐induced setup near the coast is evident from Vizianagaram to Bhubaneswar. From this figure the setdown process is also evident off the Tekkali coast extending far north off the Puri coast. This suggests that the effect of waves can either create a setup or setdown along the coast. Overall, the study highlighted the role of radiation stress in modifying the surge amplitude and, hence, the need of a coupled modeling systems to study the surge levels in a realistic manner.

17.10. ­COUPLED MODELING SYSTEM FOR CYCLONE HUDHUD The coupled model discussed in section 17.6 performed exceptionally well for cyclone Phailin that had landfall on the Odisha coast. However, the capability and short‐come of the coupled model can only be understood with more detailed studies. Also the wave‐induced setup could exhibit spatial variability depending upon the wind stress and the coastal characteristics. It results from radiation stress owing to the wave transformation along the coastal and nearshore waters. The wave transformation along this region is governed by the coastal bathymetry as well the geomorphologic features. Therefore, more location‐ specific studies along the coastal region are warranted for operational forecasts of such extreme events. Therefore, this section explores the performance of the coupled modeling system for cyclone Hudhud, which made landfall on the Andhra Pradesh coast. The coastal state of Andhra Pradesh is the second most vulnerable state after Odisha in regard to the number of tropical cyclones that make landfall. The coastal belt of Andhra Pradesh has major industrial establishments with a total of 10 ports, including the major and minor ones. In addition, coastal Andhra Pradesh is quite fertile with agricultural farmlands spread over several hectares and small‐scale industries. The two major river systems, the Godavari and Krishna, make the land highly fertile. In the context of cyclonic events, these river systems can

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  281

Wave setup induced water level(m) 0.6 0.5 0.4 0.3 0.2 0.1 0.0 –0.1 –0.2 –0.4

Figure 17.38  Spatial distribution of the wave induced setup along the coast.

play a role in modifying the surges. Moreover, the deltaic environment along the coastal stretch makes it highly vulnerable to inundation. The geomorphologic features of coastal Andhra Pradesh are highly diverse. The average shelf width over this region is about 43 km. One can find beach ridges, mudflats, mangrove forests, spits, lagoons, barriers, estuaries, and tidal inlets along the coast. In addition, at some locations rocky headland fringed cliffs, wave‐cut benches, and other erosional landforms are prominent. The recent cyclonic storm named Hudhud affected the entire coastal stretch from Visakhapatnam to Bheemunipatnam. Climate change is a major concern for the coastal community owing to the rise in sea level. This is a long‐ term change and there are certain other factors that indirectly influence the coastal community more often. The recent study by Sahoo and Bhaskaran (2016) indicates that there is an increase in the frequency of intense cyclonic storms in the Bay of Bengal basin. Emmanuel (2005) also pointed out that there is paradigm shift in the frequency of intense tropical cyclones and hurricanes. Detailed analysis on the effect of cyclone intensity in a

changing climate by Emmanuel (2005) proposed that the PDI (Power Dissipation Index) is directly linked to strength of tropical cyclones. Also, Sahoo and Bhaskaran (2016) indicated a manifold increase in PDI for the North Indian Ocean basin. The cyclonic systems that developed over the North Indian Ocean basin clearly show that the recent cyclonic storms are highly intense with a higher radius of maximum winds, and strong winds in the outer core. It brings to light that that earlier parametric wind formulations such as those studied by Jelesnianski and Taylor (1973) and Holland (1980), when applied for the present cyclonic systems, can underestimate wind speed as well as storm‐surge characteristics, and hence there is a need to revisit these formulations. The study by Murty et al. (2016) provides a detailed description of the need to modify the existing parametric wind formulations, and a modified formulation was proposed. For compact cyclones, the original Jelesnianski formulation was good enough. As mentioned, the cyclones in the present decade have a larger size, and therefore higher winds are spread over a larger area highlighting the necessity for a better wind formulation.

282  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

17.10.1. Details of Cyclone Hudhud A low‐pressure system that formed over Tenasserim coast adjoining the Andaman Sea in the early hours on 6 October 2014 upgraded to a depression on the next day, as documented by the India Meteorological Department (IMD). It intensified into a deep depression under favorable conditions moving in the west‐ northwestward direction. The Joint Typhoon Warning Centre (JTWC) and IMD issued timely advisories. On 8 October 2014, the system further intensified, and the IMD named it Hudhud. On entry to the southeast Bay of Bengal basin, the system continued to move in the west‐northwest direction. Figure  17.39 shows the satellite image of cyclone Hudhud at the time of landfall and the track details are shown in Figure 17.40. In the morning hours of 9 October, the system remained as SCS and further intensified into a very severe cyclonic storm (VSCS) in the following hours on 10 October. As per the report on 10 October by the JTWC, Hudhud was classified a Category‐1 tropical cyclone with advisory upgrading it to a Category‐2 later the same day. On 11 October 2014, the system underwent very rapid intensification reaching its peak intensity with a minimum central pressure of 950 mb, and with an average wind speed of 185 km h−1. The landfall occurred near Visakhapatnam in Andhra Pradesh during the noon

Figure 17.39  Satellite image of cyclone Hudhud at the time of landfall. Bangladesh

N

20°0′0″N

Myanmar India

Paradip Gopalpur

15°0′0″N

Srikakulam Visakhapatnam Kakinada

Nellore Chennai

Bay of Bengal

10°0′0″N

Puducherry

5°0′0″N

Sri lanka

80°0′0″E

85°0′0″E

90°0′0″E

95°0′0″E

Figure 17.40  The track of cyclone Hudhud based on the IMD report.

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  283

hours of 12 October 2014. The recorded maximum wind gust by the Cyclone Warning Centre in Visakhapatnam was about 260 km h−1. The Doppler Weather Radar (DWR) at Visakhapatnam reported the eye diameter as 66 km. Hudhud resulted in a trail of destruction after its landfall. It continued over land for quite some time and finally weakened into a low‐pressure system over eastern Uttar Pradesh before its final dissipation. As per the media reports, the catastrophe due to this extreme weather event resulted in a loss of more than 1,000 crores in Andhra Pradesh state. It was the first postmonsoon cyclone to cross Visakhapatnam since 1985, and interestingly landfall occurred on the same day as cyclone Phailin in 2013. The operational weather agency used numerical weather prediction (NWP) models, including dynamic statistical models during the cyclone Hudhud to predict genesis, track, and intensity. The operational forecasts were issued to the national and state level disaster authorities with hourly updates on its movement and intensity on the day of landfall for emergency preparedness. The IMD issued warning disseminations to local people in the affected states of India. The Indian National Centre for Ocean Information Services (INCOIS) at Hyderabad issued warnings through text messaging and Electronic Display Boards (EDB) to coastal populations, especially meant for the fishermen. The INCOIS bulletins also incorporated the cyclone warnings issued by the IMD.

17.10.2. Data and Methodology The GEBCO data provide a spatial resolution of 30 arc second, which is the best freely available data set for the ocean bathymetry. Using this data set (Figure17.41), the finite element mesh was generated using the Surface Modeling System (SMS). Instead of using a location‐ specific domain, as was used in previous studies, the entire Bay of Bengal is included in the present study domain. The finite element mesh is designed in such a way as to have fine resolution in the nearshore and coastal areas and a coarser resolution in the open ocean. The coastal resolution is less than 500 m, which gradually increases toward the open ocean relaxing to about 30 km. The resolution of the finite element mesh is in accordance to the criteria set by the previous studies by Bhaskaran et  al. (2013) and Rao et  al. (2009). The finite element mesh comprises 123,594 vertices and 235,952 elements (Figure 17.42). The best possible effort was made to optimize the grids in terms of the computational time. The present grid can be used for an operational scenario for any coastal state to simulate storm‐surge scenario associated with a cyclonic system for the east coast of India as it covers the entire Bay of Bengal. Moreover, the grid could resolve the complex coastal features and the wave transformation owing to the optimum resolution. Figure 17.43 shows the zoomed image of the coastal belt

Depth (m) 4500.0 4000.0 3500.0

PARADEEP

3000.0 2500.0 2000.0 1500.0

VISAKHAPATNAM

1000.0

KAKINADA

500.0 0.0

ENNORE CHENNAI

NAGAPAT TINAM

TUTICORN

Figure 17.41  The bathymetry of the study region.

284  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Figure 17.42  The finite element mesh of the study region.

that provides the location of the wave rider buoy and tide gauge measurement. The coupled ADCIRC‐SWAN model was run for a period of 120 hours with a ramp function of one day using the parallel High Performance Computing (HPC) system at INCOIS, Hyderabad, using 320 processors. The model was run with a bottom friction coefficient of

0.0028. This bottom friction coefficient is found suitable to capture the sandy bottom characteristics off the Andhra Pradesh coast. Moreover, the selected friction coefficient proved to be an optimum configuration with both the ADCIRC and the SWAN models (Murty et al., 2014). The ADCIRC model was forced with the tidal constituents from Le Provost tidal database along the open ocean boundary with a time step of about 10 s. The spectral distribution of the wave energy propagation and its evolution over space and time in the SWAN model is achieved using 36 directional bins and 35 frequency bins. The prescription of wave frequency used logarithmic frequency bins ranging from 0.04 to 1.0 Hz, with an angular resolution of 10°. The physical process for nonlinear wave–wave interaction activates using the ­ quadruplet Discrete Interaction Approximation (DIA) technique. The bottom friction formulation of Madsen et al. (1989) takes care of the bottom resistance for spatially varying roughness length in nearshore regions. This study used the Madsen formulation with 0.05 as the bottom roughness length scale. The source/sink functions used in the SWAN run for the wind input and white capping dissipation used the Komen et al. (1984) formulation. The coupling interval between the ADCIRC and SWAN models is set as 600 s. The ADCIRC model was forced using the modified parametric wind formulation of Jelesnianski and Taylor

Paradeep

Odisha Location of datawell wave rider buoy and tide gauge observation

Bhubaneshwar Berhampur

Andhra Pradesh Srikakulam Bheemunipatnam Visakhapatnam

Kakinada Yanam

Ongole

Figure 17.43  The zoomed image of the coastal region.

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  285

as discussed in Murty et  al. (2016). The existing Jelesnianski and Taylor wind formulation was used to calculate the maximum wind speed (V(r)) and expressed in the form: V r

Vr



2Rm r Rm2

r2

(17.1)

The modified version of the Jelesnianski formulation that considers increased cyclone size for the present decade cyclones (Murty et  al., 2016) that developed over the North Indian Ocean region can be expressed in the form: V r

Vr

2Rm r Rm2

r2

qr

(17.2)

In the above equations, V(r) is the value of the maximum wind speed and Rm is the radial distance from the storm center, where the maximum wind speed is concentrated. Murty et  al. (2016) proposed an optimum value of qr = 3/5 based on several numerical experiments. The modified formulation was also validated for several case studies. The study indicates that the proposed value of qr modified the radial wind profile from the center to the outer core of the cyclone. The study applied the modified formulation for five very severe cyclonic storms that developed over the Bay of Bengal basin.

The in situ observations during cyclone Hudhud are obtained from the directional wave rider buoy located off Gangavaram at a depth of 20 m, and the tide gauge at Visakhapatnam. The wave rider buoy measured the directional displacements from horizontal motion and vertical motion using the onboard compass. Two‐dimensional directional wave energy spectra representing the distribution of wave energy over different frequencies and directions were generated from the displacements. Finally, the wave parameters such as the significant wave height, maximum wave height, peak wave period, and mean wave period are derived from the wave spectrum. The wave rider buoy measures the wave height and wave periods ranging between 1.6 and 30 s with an accuracy of 0.5% of measured value. The data from these two locations obtained from the INCOIS were used to validate the ­coupled model performance. 17.10.3. Results and Discussions The modified and unmodified wind formulation resulted in an increase in the radial distance from the cyclone center. Strong outer core winds are observed for the modified wind formulation. Figure  17.44 clearly shows the effect of the modified wind formulation as a function of radial distance from the cyclone center. The wind stress computed using the modified Jelesnianski formulation indicated an increase in the area of the cyclonic storm. Figure  17.45a and 17.45b clearly demonstrates the difference in the extent of minimum wind speed of 5 m s−1

50 45 40 Modified

Wind speed (m s–1)

35

Unmodified

30 25 20 15 10 5 0

–100

0

100

200

300

400

500

Radial distance from the cyclone centre (km)

Figure 17.44  Comparison of the radial profile of the cyclonic wind speeds with the original and modified version of the Jelesnianski parametric wind model.

600

286  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION (a) Wind speed (m s–1) 24.0

Wind speed (m s–1) 30.0

21.9 19.8 17.7 15.6 13.4 11.3 9.2

27.2

09 October, 2014 (00 h)

24.4 21.7 18.9 16.1

Visakhapatnam Kakinada

13.3 10.6

7.1 5.0

24.00 m s 5.00 m s–1

35.0

Nellore

Wind velocity (m s–1) Chennai

Wind speed (m s–1) 46.0

31.7

41.4

10 October, 2014 (18 h)

18.3 15.0 Visakhapatnam

Kakinada

11.7 8.3 5.0

5.0

30.00 m s–1 5.00 m s–1

Wind speed (m s–1)

21.7

Visakhapatnam Kakinada

7.8

Nellore

Wind velocity (m s–1) Chennai –1

28.3 25.0

10 October, 2014 (00 h)

36.9

11 October, 2014 (06 h)

32.3 27.8 23.2 18.7 14.1

Visakhapatnam Kakinada

9.6

Nellore

Wind velocity (m s–1) Chennai 35.00 m s–1 5.00 m s–1

5.0

Nellore

Wind velocity (m s–1) Chennai 46.00 m s–1 5.00 m s–1

Figure 17.45  (a) Time series plot of the wind envelope from original Jelesnianski parametric wind formulation; (b) time series plot of the wind envelope from modified Jelesnianski parametric wind formulation.

with unmodified and modified winds. An increase in the radius of maximum wind is also evident. The modified formulation provided a better estimate of the wind field on a spatial scale. It is obvious from the wind field that the modified wind formulation will serve better for the simulation of storm surge and associated inundation. The maximum wind velocity computed by the model simulations is nearly 46 m s−1. As seen from the modified wind formulation, close to the landfall time, the entire Bay of Bengal experienced a minimum wind speed of nearly 5 m s−1. Figure  17.46 shows the model computed significant wave height for the study region. Higher significant wave heights are noticed on the right of the storm track in the nearshore region. In the coastal region, wave heights in excess of 8 m are observed north of Visakhapatnam, whereas along the southern side of the cyclone track the wave heights were less than 4 m. The right side of the storm track experiences stronger winds owing to the cyclonic circulation and hence the northern sector

e­xperienced higher wave height. The time evolution of the significant wave height at Gangavaram, south of Visakhapatnam, shows an increase in wave height as the cyclone approached its landfall. A comparison of wave height with in situ observation off Visakhapatnam exhibits a very good match (Figure 17.47). The study signifies that the coupled model has proven its efficacy in estimating realistically the hydrodynamic conditions for cyclone Hudhud. The surge residual computed from the coupled and stand‐alone model simulations for Bheemunipatnam location are shown in Figure  17.48. The surge residual from the coupled model takes into account the wave‐ induced setup and wind setup. The maximum computed storm surge at Bheemunipatnam was about 2.3 m. The difference in the surge residual between the stand‐alone and coupled run is about 0.5 m. The in situ observation of the water level was available only at Visakhapatnam to verify the model performance.

(b) Wind speed (m s–1)

Wind speed (m s–1)

24.0

30.0 27.2

21.9 19.8

24.4

09 October, 2014 (00 h)

17.7

21.7

15.6

18.9

13.4

16.1

Visakhapatnam

11.3

13.3

Kakinada

9.2

10.6

7.1 5.0

Wind velocity (m

Visakhapatnam Kakinada

7.8 5.0

Nellore

Nellore

Wind velocity (m s–1) Chennai

Chennai

s–1)

10 October, 2014 (00 h)

30.00 m s–1 5.00 m s–1

24.00 m s–1 5.00 m s–1

Wind speed (m s–1)

Wind speed (m s–1)

46.0

35.0

41.4

31.7

36.9

28.3

32.3

10 October, 2014 21.7 (18 h)

27.8

18.3

18.7

25.0

11 October, 2014 (06 h)

23.2

Visakhapatnam Kakinada 11.7 15.0

14.1

Visakhapatnam Kakinada

9.6

8.3

5.0

5.0

Nellore

Nellore

Wind velocity (m s–1) Chennai

Wind velocity (m s–1) Chennai

46.00 m s–1 5.00 m s–1

35.00 m s–1 5.00 m s–1

19°0ʹ0ʺN

Figure 17.45 (Continued) Gopalpur

Hs(m) 17.0 15.1

Palasa at Kasibugga

18°0ʹ0ʺN

13.2 9.4 7.6

Visakhapatnam

5.7

17°0ʹ0ʺN 16°0ʹ0ʺN 15°0ʹ0ʺN

Srikakulam Vijayanagaram

11.3

3.8 1.9

Kakinada

Bay of Bengal

0.0 Machilipattanam

Ongole

Nellore 80°0ʹ0ʺE

81°0ʹ0ʺE

82°0ʹ0ʺE

83°0ʹ0ʺE

84°0ʹ0ʺE

85°0ʹ0ʺE

86°0ʹ0ʺE

87°0ʹ0ʺE

Figure 17.46  Model‐computed significant wave height.

88°0ʹ0ʺE

89°0ʹ0ʺE

288  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

Significant wave height (m)

10 9 Wave rider buoy observation Significant wave height from coupled model

8 7 6 5 4 3 2 1 0

08

09

/10

/14

09

/10

12 :0

0

/14

/10

00

:00

/14

12 :0

13 10 11 11 12 12 13 /10 /10 /10 /10 /10 /10 /10 /14 /14 /14 /14 /14 /14 /14 12 0 1 1 0 1 0 2 0:0 2:0 2:0 0:0 0:0 0:0 :00 : 0 0 0 0 0 0 0 0

10 /10 /14 0

0

Time (UTC)

Figure 17.47  Validation of significant wave height with buoy observations off Visakhapatnam coast.

2.50

Near Bheemunipatnam

Surge (m)

2.00 1.50

ADCIRC+SWAN ADCIRC

1.00 0.50 0.00 10/10/2014 12:00

10/11/2014 00:00

10/11/2014 12:00

10/12/2014 00:00

10/12/2014 12:00

10/13/2014 00:00

10/13/2014 12:00

Time (UTC) 1.3

Near Visakhapatnam

1.1

Surge (m)

0.9

Observed ADCIRC

0.7

ADCIRC+SWAN

0.5 0.3 0.1 –0.110/11/2014 10/11/2014 10/11/2014 10/11/2014 10/12/2014 10/12/2014 10/12/2014 10/12/2014 10/12/2014 10/13/2014 10/13/2014 06:00 10:48 15:36 20:24 01:12 06:00 10:48 15:36 20:24 01:12 06:00 Time (UTC)

Figure 17.48  Time series plot of the surge residual (in meters) at Bheemunipatnam and Visakhapatnam.

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  289 0.50 Wave induced setup (m)

0.40 Visakhapatnam

0.30

Bheemunipatnam

0.20 0.10 0.00 10/10/2014 00:00 –0.10

10/10/2014 12:00

10/11/2014 00:00

10/11/2014 12:00

10/12/2014 00:00

10/12/2014 12:00

10/13/2014 00:00

10/13/2014 12:00

Time (UTC)

–0.20

Figure 17.49  Comparison of wave‐induced setup along Visakhapatnam and Bheemunipatnam. (a)

(b) 81° 0′ Slope (%)

0.6 – 0.9

Visakhapatnam

0.9 – 1.2 >1.2 Rajahmundry

Kakinada

17°

17°

0′

0′

15° 0′

Pulic at

15° 0′ Penner

81° 0′

50 km

83° 0′

Wave station (Wave ranges in meters)

Vulnerability rank 1 (Very low) 2 (Low) 3 (Moderate) 4 (High) 5 (Very high)

Kavali 0.99 Koduru 1.08

Lake

Lake

Penner

1 (Very low) 2 (Low) 3 (Moderate) 4 (High) 5 (Very high)

Pulic at

0′

Vulnerability rank

Vulnerability Level

15°

al ng

Be

i avar

ri va

Vijayawada

f yo Ba

Rajahmundry ake uL shn ler a Kol Vodalarevu Vijayawada 1.34 Gollapalem Machilipatnam 0.92 0.82 Chirala 0.27 Lankavanidibba Kottapatnam 0.42 0.91

Kri

God

Goda

r shn Kolle a

Kri

ke

a uL

Baruva 1.46 Amalapadu 1.06 Kalingapatnam 1.03 Koyyam 1.15 el Visakhapatnam ev 1.11 yL t i Pudimadaka l bi 1.03 ra Tetagunta lne u V 1.01 17° l ga Kakinada en 0′ B 0.21 f yo a B 83° 0′

Significant wave heights (m) 0.21 1.03 0.27 1.06 0.42 1.08 0.82 1.11 0.92 1.15 0.99 1.20 1.34 1.01

Bhimunipatnam

0.3 – 0.6

0′

81° 0′

Kalingapatnam

< 0.3

17°

Baruva

83° 0′

15° 0′

Durgarajpattanam 1.08 Pulicat 1.2 81° 0′

50 km

83° 0′

Figure 17.50  Coastal vulnerability rank based on (a) coastal slopes and (b) significant wave height during fair weather conditions for the Andhra Pradesh coast (Nageswara Rao et al., 2008).

A comparison of the surge residual computed from the stand‐alone and coupled model against observation showed a slight overestimation from the model runs. Also, the stand‐alone and the coupled model runs showed no significant difference in the surge amplitude. From the model simulations, it is found that Bheemunipatnam experienced a wave setup; whereas a setdown is observed at Visakhapatnam. As the cyclone approached landfall, an increase of wave setup along Bheemunipatnam is evident. Until the time of landfall, the wave setup is found to increase, and after the time of landfall the setdown is  noticed. This setdown could be attributed to the

­ redominant offshore winds at the location. On the other p hand, the wave setup remained almost invariant at Visakhapatnam, and that was followed by setdown during the landfall event (Figure 17.49). A detailed study on the vulnerability of the Andhra Pradesh coast, especially the stretch from Bheemunipatnam to Visakhapatnam, was carried by Nageswara Rao et al. (2008). The study used remote sensing techniques to assess sea level rise and coastal vulnerability. Figure 17.50 shows the vulnerability rank along this stretch based on the coastal slope and the significant wave height during fair weather conditions.

290  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

The study also makes an effort to compare the coastal vulnerability with the model computed wave setup. The study by Nageswara Rao et al. (2008) categorized Visakhapatnam as a region having very low to low v­ ulnerability, whereas the vulnerability rank was moderate or very high for the coastal stretch north of Visakhapatnam. However, the vulnerability for the region south of Vishakapatnam extending from Kakinada to Pulicat ranks the highest owing to its mildly sloping beaches. Also the vulnerability level considering the significant wave height is high between Visakhapatnam to Bheemunipatnam. The coastal areas south of Visakhapatnam especially south of Kakinada are highly vulnerable. The study signifies that the vulnerability for the narrow coastal stretch covering Visakhapatnam to Bheemunipatnam has a close resemblance with the observations seen in the wave‐ induced setup. It has a direct bearing on the coastal geomorphic features. The bottom features off Bheemunipatnam have Karstic pinnacled features both along the mid and shelf edge regions retarding wave propagation toward the near‐ shore areas, causing ­ piling up of water during extreme weather events. It is unlike the bottom dome‐shaped features observed off Visakhapatnam comprising reef structures with higher gradient in beach slopes. The wave‐induced setup has a direct bearing on beach slopes and bottom features, and the results obtained from this study will be of immense value to coastal zone authorities. 17.11. ­SUMMARY AND CONCLUSIONS The chapter elaborates on tropical cyclone-induced storm surges and wind waves in the Bay of Bengal region. It covers aspects on tropical cyclone activity over the North Indian Ocean region and lists various studies on storm surge for the Bay of Bengal covering progress in a global and regional perspective. In addition, relevant studies conducted on wind‐wave modeling and their role during extreme weather events in a global and regional scale have been discussed. The importance of coupled wave‐hydrodynamic models for operational needs is also highlighted. Further, the study discusses four recent severe cyclones, Thane, Aila, Phailin, and Hudhud, which had landfall along the east coast of India. Numerical simulations were carried out using the state‐of‐art ADCIRC and SWAN models for these extreme weather events. The study brings to light that coupled wave‐hydrodynamic models performed quite well in simulating these severe cyclones and therefore recommends their applicability in operational weather centers. Among the unsolved problems in storm‐ surge modeling is the interaction between storm surge and river flooding, as happened during the 29 October 1999 cyclone on the coast of Odisha. However, it should be noted that the interactions between storm surges, tides, and wind waves are reasonably well modeled.

­REFERENCES Amein, M., & Kraus., (1991). Dynlet1: Dynamic implicit numerical model of one dimensional tidal flow through inlets. US Army Corps of Engineers, Technical Report CERC‐91–10. Ardhuin, F., Chapron, B., & Collard, F., (2009). Observation of swell dissipation across oceans, Geophysical Research Letters, 36, L06607; doi:10.1029/2008GL037030 Babanin, A. V. (2011). Breaking and dissipation of ocean surface waves. Cambridge University Press. UK: Cambridge. Bhaskaran, P. K., Gayathri, R., Murty, P. L. N., Bonthu, S., & Sen, D. (2014). A numerical study of coastal inundation and its validation for Thane cyclone in the Bay of Bengal. Coastal Engineering, 83, 108–118. Bhaskaran, P. K., Nayak, S., Bonthu, S. R., Murty, P. N., & Sen, D. (2013). Performance and validation of a coupled parallel ADCIRC–SWAN model for THANE cyclone in the Bay of Bengal. Environmental Fluid Mechanics, 13(6), 601–623. Blain, C. A., & Rogers, W. E. (1998). Coastal tide prediction using the ADCIRC‐2DDI hydrodynamic finite element model: Model validation and sensitivity analyses in the Southern North Sea/English Channel (No. NRL/ FR/7322‐‐98‐9682). Naval Research Laboratory, Stennis Space Center, MS. Blain, C. A., Linzell, R. S., & Massey, T. C. (2008). MeshGUI: A mesh generation and editing toolset for the ADCIRC model (No. NRL/MR/7322‐‐08‐9083). Naval Research Laboratory, Stennis Space Center, Oceanography Division, MS Bode, L., & Hardy, T. A. (1997). Progress and recent developments in storm surge Modeling. Journal of Hydraulic Engineering, 123(4), 315–331. Bolanos, R., Brown, J. M., & Souza, A. J. (2014). Wave‐current interactions in a tide dominated estuary. Continental Shelf Research, 87, 109–123. Bolanos, R., Osuna, P., Wolf, J., Monabiu, J., & Sanchez‐Arcilla, A. (2011). Development of the POLCOMS‐WAM current‐ wave model. Ocean Modeling, 36, 102–115. Brakenridge, G. R., Syvitski, J. P. M., Overeem, I., Higgins, S. A., Kettner, A. J., Stewart‐Moore, J. A., & Westerhoff, R. (2013). Global mapping of storm surges and the assessment of coastal vulnerability. Natural Hazards, 66(3), 1295–1312. Bretschneider, C. L. (1959). Wave variability and wave spectra for wind‐generated gravity waves (No. TM‐118). Corps of Engineers, Beach Erosion Board, Washington, DC. Camp Dresser & Mckee, Inc. (1985). Study to revise and update FEMA’s storm surge model. Federal Emergency Management Agency,Washington, D.C. Cavaleri, L., Alves, J. H. G. M., Ardhuin, F., Babanin, A., Banner, M., Belibassakis, K., Benoit, M., et al. (2007). Wave modeling: The state of the art. Progress in Oceanography, 75(4), 603–674. Chaudhury, A.M., & Ali, A., (1974). Prediction of maximum surge heights associated with cyclones affecting Bangladesh: A preliminary investigation. Nuclear Science and Applications, 7. Chittibabu, P. (1999). Development of storm surge prediction models for the Bay of Bengal and the Arabian Sea. PhD Thesis, IIT Delhi, India, 262.

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  291 Chittibabu, P., Dube, S. K., Rao, A. D., Sinha, P. C., & Murty, T. S. (2000). Numerical simulation of extreme sea levels using location specific high resolution model for Gujarat coast of  India. Marine Geodesy, 23, 133–142; doi:10.1080/ 01490410050030698 Chittibabu, P., Dube, S. K., Rao, A. D., Sinha, P. C., & Murty, T. S. (2002). Numerical simulation of extreme sea levels for the Tamilnadu (India) and Sri Lanka coasts. Marine Geodesy, 25, 235–244; doi:10.1080/01490410290051554 Cho, B. H., Eum, H. M., Kim, H. S., Jeong, W. M., & Shim, J.  S. (2009). Wave simulations for Typhoon Maemi. The Fourteenth OMISAR Workshop on Ocean Models, 1–37. Conner, W. C., Kraft, R. H., & Harris, D. L. (1957). Empirical methods for forecasting the maximum storm tide due to hurricanes and other tropical storms. Monthly Weather Review, 85(4), 113–116. CWRC (2011). Cyclone eAtlas  –  IMD, Version 2.0/2011, Technical Note, Tracks of Cyclones and Depressions over North Indian Ocean (from 1891 onwards), Cyclone Warning and Research Centre, India Meteorological Department, Chennai. Das, P. K. (1972). Prediction model for storm surges in the Bay of Bengal. Nature, 239, 211–213. Das, P. K. (1994). Kalpathi Ramakrishna Ramanathan medal lecture, 1993: Prediction of storm surges in the Bay of Bengal. Proceedings‐Indian National Science Academy Part A, 60, 513–534. Das, P. K. (1980). Storm surges: A review of current problems. WMO workshop on storm surges, 10–15 November 1980, Rangoon, Burma. Das, P. K., Miyazaki, M., & Jelesnianski, C. P., (1978). Present techniques of storm surge prediction. WMO Report No. 500, Geneva. Dietrich, J. C. (2010). Development and application of coupled hurricane wave and surge models for southern Louisiana, Doctoral dissertation, University of Notre Dame, USA. Dietsche, D., Hagen, S. C., & Bacopoulos, P. (2007). Storm surge simulations for Hurricane Hugo (1989): On the significance of inundation areas. Journal of Waterway, Port, Coastal, and Ocean Engineering, 133(3), 183–191. Donn, W. L. (1958). An empirical basis for forecasting storm tides. Bulletin of the American Meteorological Society, 39, 640–647. Dube, S. K., & Gaur, V.K. (1995). Real time storm surge pre­ diction system for the Bay of Bengal. Current Science, 68,103–113. Dube, S. K., Rao, A. D., Sinha, P. C., Murty, T. S., & Bahulayan, N. (1997). Storm surge in the Bay of Bengal and Arabian Sea: The problem and its prediction. Mausam, 48(2), 283–304. Dube, S. K., Chittibabu, P., Rao, A. D., Sinha, P. C., & Murty, T. S. (2000a). Sea levels and coastal inundation due to tropical cyclones in Indian coastal regions of Andhra and Orissa. Marine Geodesy, 23, 65–73; doi:10.1080/01490410050030643 Dube, S. K., Rao, A. D., Sinha, P. C., & Chittibabu, P. (2000b). Storm surge modeling and prediction. In A. P. Dwivedi (Ed.), Mathematical analysis and applications (pp. 109–124). New Delhi: Narosa Publishing House.

Dube, S. K., Chittibabu, P., Sinha, P. C., Rao, A. D., & Murty, T. S. (2004). Numerical modeling of storm surges in the head Bay of Bengal using location specific model. Natural Hazards, 31, 437–453. Emanuel, K. (2005). Increasing destructiveness of tropical cyclones over the past 30 years. Nature, 436(7051), 686–688. Flather, R. A. (1976). Practical aspects of the use of numerical models for surge prediction. Rep. No. 30, Inst. Ocean Science, Bidston, UK. Flather, R. A., & Proctor, R., (1983). Prediction of North Sea storm surges using numerical models: Recent developments in UK. In J. Sundermann & W. Lenz (Eds.), North sea dynamics. Berlin, Heidelberg: Springer‐Verlag, 299–317. Fleming, F., Fulcher, C., Luettich, R., Estrade, B., Allen, G., & Winer, H., (2008). A real time storm surge forecasting system using ADCIRC. In M. Spaulding (Ed.), Estuarine and coastal modeling X (pp. 373–392). Reston, VA: American Society of Civil Engineers. Garratt, J. R. (1977). Review of drag coefficients over oceans and continents. Monthly Weather Review, 105(7), 915–929. Gayathri, R., Murty, P. L. N., Bhaskaran, P. K., & Srinivasa Kumar, T. (2015). A numerical study of hypothetical storm surge and coastal inundation for AILA cyclone in the Bay of Bengal. Environmental Fluid Mechanics; doi: 10.1007/s10652‐015‐9434‐z Gelci, R., Cazalé, H., & Vassal, J., (1957). Prévision de la houle, la méthode des densités spectroangulaires. Bulletin Inform. Comité Central Oceanography, d’étude Cotes, No. 9, 416. Ghosh, S. K. (1977). Prediction of storm surges on the East coast of India. Indian Journal of Meteorology, Hydrology and Geophysics, 28, 157–168. Grant, W. D., & Madsen, O. S., (1979). Combined wave and current interaction with a rough bottom. Journal of Geophysical Research, 84, 1797–1808. Harris, D. L. (1963). Characteristics of the hurricane storm surge. Technical Paper No. 48, Weather Bureau, US Department of Commerce, Washington, DC, 146. Holland, G. J. (1980). An analytic model of the wind and pressure profiles in hurricanes. Monthly Weather Review, 108(8), 1212–1218. Holthuijsen, L. H., (2007). Waves in oceanic and coastal waters. Cambridge, UK: Cambridge University Press. Houston, S. H., Shaffer, W. A., Powell, M. D., & Chen, J. (1999). Comparisons of HRD and SLOSH surface wind fields in hurricanes: Implications for storm surge modeling. Weather Forecast, 14, 671–686. Huang, N. E. (1979). On surface drift currents in the ocean. Journal of Fluid Mechanics, 91(01), 191–208. Hubbert, G. D., Holland, G.J., Leslie, L.M., & Manton, M.J., (1991). A real‐time system for forecasting tropical cyclone storm surges. Weather and Forecasting, 6, 86–97. Jain, I., Chittibabu, P., Agnihotri, N., Dube, S. K., Sinha, P. C., & Rao, A. D. (2006a). Simulation of storm surges along Myanmar coast using a location specific numerical model. Natural Hazards, 39,71–82; doi:10.1007/s11069‐005‐3176‐z Jain, I., Chittibabu, P., Agnihotri, N., Dube, S. K., Sinha, P. C., & Rao, A. D. (2006b). Numerical storm surge prediction model for Gujarat coast of India and adjoining Pakistan coast. Natural Hazards, 42, 67–73; doi:10.1007/s11069‐006‐9060‐7

292  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION Jain, I., Rao, A. D., Jitendra, V., & Dube, S. K. (2010). Computation of expected total water levels along the east coast of India. Journal of Coastal Research, 681–687. Janardhan, S. (1967). Storm induced sea level changes at Saugor Island situated in North Bay of Bengal. Indian Journal of Meteorology, Hydrology and Geophysics., 18, 205–212. Jarrell, J. D., Lewis, J. K., & Whitaker, R. E. (1982). Bay of Bengal: A system to evaluate storm surge threat, final report. Contact A.I.D/S.O.D/P.D.C.‐C0294, Science Applications Inc., Monterey, California, 66. Jelesnianski, C. P. (1965). A numerical calculation of storm tides induced by a tropical storm impinging on a continental self. Monthly Weather Review, 93, 343–358. Jelesnianski, C. P. (1972). SPLASH I: Landfall storms. NOAA Technical Memorandum NWS TDL‐46. Jelesnianski, C. P. (1976). A sheared coordinate system for storm surge equations of motion with a mildly curved coast. NOAA Tech. Memo, NWS/TDL‐61, Techniques Development Lab, Silver Spring, MD. Jelesnianski, C. P., & Chen, J. (1982). SLOSH (Sea, lake and overland surges from hurricanes). Techniques Development Laboratory, NOAA, Silver Spring, MD. Jelesnianski, C. P., & Taylor, A.D. (1973). NOAA Technical Memorandum. ERL, WMPO‐3, 33. Jelesnianski, C. P., Chen, J., & Shaffer, W. A. (1992). SLOSH: Sea, Lake, and Overland Surges from Hurricanes. US Dept. of Commerce, National Oceanic and Atmospheric Administration, National Weather Service, USA. Jenkins, A. D. (1987). Wind and wave induced currents in a rotating sea with depth‐varying eddy viscosity. Journal of Physical Oceanography, 17(7), 938–951. Johns, B., & Ali, A., (1980). The numerical modeling of storm surges in the Bay of Bengal. Quarterly Journal of the Royal Meteorological Society, 106, 1–18. Johns, B., Dube, S. K., Mohanty, U. C., & Sinha, P. C. (1981). Numerical simulation of the surge generated by the 1977 Andhra cyclone. Quarterly Journal of the Royal Meteorological Society, 107(454), 919–934. Jonsson, I. G. (1990). Wave‐current interactions. In B. Le Mehaute & D. M. Hanes, Ocean engineering science, the sea (9A, pp. 65–120). Wiley Interscience. Knapp, K. R., Kruk, M. C., Levinson, D. H., Diamond, H. J., & Neumann, C. J. (2010). The International Best Track Archive for Climate Stewardship (IBTrACS): Unifying tropical cyclone data. Bulletin of the American Meteorological Society, 91, 363–376. Komen, G. J., Hasselmann, S., & Hasselmann, K., (1984). On the existence of a fully developed wind‐sea spectrum. Journal of Physical Oceanography, 14, 1271–1285. 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. Kumar, N., Voulgaris, G., Warner, J. C., & Olabarrieta, M., (2012), Implementation of the vortex force formalism in the coupled ocean‐atmosphere‐wave‐sediment transport (COAWST) modeling system for inner shelf and surf zone applications. Ocean Modeling, 47, 65–95. Longuet‐Higgins, M. S., & Stewart, R. W. (1964). Radiation stresses in water waves; a physical discussion, with applications.

Deep Sea Research and Oceanographic Abstracts, 11(4), 529–562. Luettich, R. A., & Westerink, J. J. (2004). Formulation and numerical implementation of the 2D/3D ADCIRC finite element model version 44. XX, 74. Luettich, R. A. Jr., Westerink, J. J., & Scheffner, N.W. (1992). ADCIRC: an advanced three‐dimensional circulation model for shelves, coasts, and estuaries. Report 1: Theory and methodology of ADCIRC‐2DDI and ADCIRC‐3DL, dredging research program technical report DRP‐92‐6. US Army Engineers Waterways Experiment Station, Vicksburg. Madsen, H., & Jakobsen, F. (2004). Cyclone induced storm surge and flood forecasting in the northern Bay of Bengal. Coastal Engineering, 51(4), 277–296. Madsen, O. S., Poon, Y. K., & Graber, H. C. (1989). Spectral wave attenuation by bottom friction: theory. Coastal Engineering, 492–504. Mathisen, P. P., & Madsen, O. S. (1996). Waves and currents over a fixed rippled bed: 1. Bottom roughness experienced by waves in the presence and absence of currents. Journal of Geophysical Research: Oceans, 101(C7), 16533–16542. Mathisen, P. P., & Madsen, O. S. (1999). Waves and currents over a fixed rippled bed: 3. Bottom and apparent roughness for spectral waves and currents. Journal of Geophysical Research: Oceans, 104(C8), 18447–18461. Medha, K., & Sunitha Devi, S. (2015). Cyclones and depressions over the North Indian Ocean during 2014. Mausam, 66(3), 337–354. Mellor, G. L., (2003). The three dimensional, current and surface wave equations. Journal of Physical Oceanography, 33,1978–1989. Mellor, G. L. (2011). Wave radiation stress. Ocean Dynamics, 61, 563–568. Mellor, G. L., Donelan, M. A., & Oey, L‐Y., (2008). A surface wave model for coupling with numerical ocean circulation models. Journal of Atmospheric and Oceanic Technology, 35, 1785–1807. Mitsuyasu, H., (2002). A historical note on the study of ocean surface waves. Journal of Oceanography, 58, 109–120. Miyazaki, M., Ueno, T., & Unoki, S., (1962). Theoretical investigations of typhoon surges along the Japanese coast. Oceanography Magazine, 13(2), 103–117. Murty, P. L. N., Bhaskaran, P. K., Gayathri, R., Sahoo, B., Kumar, T. S., & SubbaReddy, B. (2016). Numerical study of coastal hydrodynamics using a coupled model for Hudhud cyclone in the Bay of Bengal. Estuarine, Coastal and Shelf Science, 183, 13–27. Murty, P. L. N., Sandhya, K. G., Bhaskaran, P. K., Jose, F., Gayathri, R., Balakrishnan Nair, T., et al. (2014). A coupled hydrodynamic Modeling system for PHAILIN cyclone in the Bay of Bengal. Coastal Engineering, 93, 71–81. Murty, T. S. (1984). Storm surges: Meteorological ocean tides (vol. 212). Ottawa: Department of Fisheries and Oceans. Murty, T. S., & Flather, R. A. (1994). Impact of storm surges in the Bay of Bengal. Journal of Coastal Research, 149–161. Murty, T.S., & Henry, R. F., (1983). Tides in the Bay of Bengal. Journal of Geophysical Research, 88(c‐10): 6069–6076; doi:10.1029/JC088iC10p06069

Tropical Cyclone–Induced Storm Surges and Wind Waves in the Bay of Bengal  293 Murty, T. S., Flather, R. A., & Henry, R. F. (1986). The storm surge problem in the Bay of Bengal. Progress in Oceanography, 16(4), 195–233. Nageswara Rao, K., Subraelu, P., Venkateswara Rao, T., Hema Malini, B., Ratheesh, R., Bhattacharya, S., et al. (2008). Sea‐level rise and coastal vulnerability: an assessment of Andhra Pradesh coast, India through remote sensing and GIS. Journal of Coastal Conservation, 12, 195–207. Nayak, S., Bhaskaran, P. K., Venkatesan, R., & Dasgupta, S. (2013). Modulation of local wind‐waves at Kalpakkam from remote forcing effects of Southern Ocean swells. Ocean Engineering, 64, 23–35. Ningsih, N. S., Yamashita, T., & Aouf, L. (2000). Three‐dimensional simulation of water circulation in the Java Sea: Influence of wind waves on surface and bottom stresses. Natural Hazards, 145–171. Niyas, N.T., Srivastava, A.K., & Hatwar, H.R. (2009). Variability and trend in the cyclonic storms over North Indian Ocean. IMD, Met. Monograph No. Cyclonic Warning -3/2009. Peng, M., Xie, L., & Pietrafesa, L. J. (2004). A numerical study of storm surge and inundation in the Croatan–Albemarle– Pamlico Estuary System. Estuarine, Coastal and Shelf Science, 59(1), 121–137. Poulose, J., Rao, A. D., & Bhaskaran, P. K., (2017). Role of continental shelf on non‐linear interaction of storm surges, tides and wind waves: An idealized study representing the west coast of India. Estuarine, Coastal and Shelf Science; http://dx.doi.org/10.1016/j.ecss.2017.06.007 Powell, M. D., & Houston, S. H. (1996). Hurricane Andrew’s landfall in south Florida. Part II: Surface wind fields and potential real‐time applications. Weather Forecasting, 11, 329–349. Powell, M. D., Houston, S. H., Amat, L. R., & Morisseau‐ Leroy, N. (1998). The HRD real‐time hurricane wind analysis system. Journal of Wind Engineering and Industrial Aerodynamics, 77/78, 53–64. Punithavathi, J., Tamilenthi, S., & Baskaran, R. (2012). A study of Thane cyclone and its impacts in Tamil Nadu, India using geographic information system. Archives of Applied Science Research, 4(1), 685–695. Qayyum, M. F. (1983). Prediction of storm surges for Bangladesh coast by empirical method: Results and discussions. WMO ESCAP 10th Session Panel on Tropical Cyclones, 22–29. Rao, A. D., Chittibabu, P., Murty, T. S., Dube, S. K., & Mohanty, U. C. (2007). Vulnerability from storm surges and cyclone wind fields on the coast of Andhra Pradesh, India. Natural Hazards, 41(3), 515–529. Rao, A.D., Jain, I., & Venkatesan, R., (2010). Estimation of extreme water levels due to cyclonic storms: a case study for Kalpakkam coast. International Journal of Ocean and Climate Systems, 1, 1–14. Rao, A. D., Jain, I., Murthy, M. R., Murty, T. S., & Dube, S. K. (2009). Impact of cyclonic wind field on interaction of surge‐ wave computations using finite‐element and finite‐difference models. Natural Hazards, 49(2), 225–239.

Rao, N. S. B. (1968). On some aspects of local and tropical storms in the Indian area (Doctoral dissertation, PhD Thesis, University of Jadavpur, India. Rao, N. S. B., & Mazumdar, S. (1966). A technique for forecasting storm waves. Indian Journal of Meteorology and Geophysics, 17, 333–346. Rao, Y. R., Chittibabu, P., Dube, S.K., Rao, A.D., & Sinha, P.C., (1997) Storm surge prediction and frequency analysis for Andhra coast of India. Mausam, 48, 555–566. Sahoo, B., & Bhaskaran, P.K. (2015). Synthesis of tropical cyclone tracks in a risk evaluation perspective for the east coast of India. International Conference on Water Resources, Coastal and Ocean Engineering, ICWRCOE’15. Aquatic Procedia, 4, 389–396. Sahoo, B., & Bhaskaran, P. K. (2016). Assessment on historical cyclone tracks in the Bay of Bengal, east coast of India. International. Journal of Climatology, 36(1), 95–109. Sahoo, B., & Bhaskaran, P. K., (2017). A Comprehensive Dataset for Tropical cyclone storm‐surge induced inundation for the East Coast of India. International Journal of Climatology; doi:10.1002/joc.5184 Sandhya, K. G., Nair, T. B., Bhaskaran, P. K., Sabique, L., Arun, N., & Jeykumar, K. (2014). Wave forecasting system for operational use and its validation at coastal Puducherry, east coast of India. Ocean Engineering, 80, 64–72. Singh, O. P. (2001). Long term trends in the frequency of monsoonal cyclonic disturbances over the North Indian Ocean. Mausam, 52(4), 655–658. Singh, O. P. (2007). Long‐term trends in the frequency of severe cyclones of Bay of Bengal: observations and simulations. Mausam, 58, 59–66. Singh, O. P., Ali Khan, T. M., & Md. Rahman, S. (2000). Changes in the frequency of tropical cyclones over the North Indian Ocean. Meteorology and Atmospheric Physics, 75(1–2), 11–20. Smith, N. (2006). Perspectives from the global ocean data assimilation experiment. In E. P. Chassignet & J. Verrorn (Eds.), Ocean weather forecasting, an integrated view of oceanography (pp. 1–17). Springer Netherlands. Srinivasa Kumar, T., Mahendra, R. S., Nayak, S., Radhakrishnan, K., & Sahu, K. C. (2010). Coastal vulnerability assessment for Orissa State, East coast of India. Journal of Coastal Research, 26(3), 523–534. Srivastava, A. K., Sinha Ray, K. C., & De, U.S. (2000). Trends in the frequency of cyclonic disturbances and their intensification over Indian seas. Mausam, 51, 113–118. Sundermann, J., & Lenz, W., (1983). North Sea dynamics. Berlin, Heidelberg: Springer‐Verlag. Tonani, M., Balmaseda, M., Bertino, L., Blockley E., Brassington, G., Davidson, F., et al. (2015). Status and future of global and regional ocean prediction systems. Journal of Operational Oceanography, 8(2), s201–s220; doi: 10.1080/ 1755876X.2015.1049892. Trigo, I. F. (2006). Climatology and Interannual Variability of Storm‐Tracks in the Euro Atlantic Sector: a Comparison between ERA‐40 and NCEP/NCAR Reanalysis. Climate Dynamics, 26(2–3), 127–143. Wang, X. L., Swail, V. R., & Zwiers, F. W., (2006). Climatology and changes of extratropical cyclone activity: comparison of

294  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION ERA‐40 with NCEP‐NCAR reanalysis for 1958‐2001. Journal of Climate, 19, 3145–3166. Welander, P. (1961). Numerical prediction of storm surges. Advances in Geophysics, 8, 315–379. Westerink, J. J., Luettich, R. A., Feyen, J. C., Atkinson, J. H., Dawson, C., Roberts, H. J., Powell, M.D., et  al. (2008). A basin‐to channel‐scale unstructured grid hurricane storm surge model applied to southern Louisiana. Monthly Weather Review, 136(3), 833–864. Wolf, J., Hubbert, K. P., & Flather, R. A. (1988). A feasibility  study for the development of a joint surge and wave  model. Birkenhead, Proudman Oceanographic Laboratory.

Xie, L., Pietrafesa, L. J., & Peng, M. (2004). An integrated storm surge and inundation Modeling system for lakes, ­estuaries and coastal ocean. Journal of Coastal Research, 20, 1209–1223. Xie, L., Wu, K., Pietrafesa, L., & Zhang, C. (2001). A numerical study of wave‐current interaction through surface and bottom stresses: Wind‐driven circulation in the South Atlantic Bight under uniform winds. Journal of Geophysical Research: Oceans, 106(C8), 16841–16855. Zodiatis, G., Galanis, G., Kallos, G., Nikolaidis, A., Christina Kalogeri., & Aristotelis Liakatas, S. S., (2015). The impact of sea surface currents in wave power potential modeling. Ocean Dynamics, 65, 1547–1565.

18 Space‐Based Measurement of Rainfall Over India and Nearby Oceans Using Remote Sensing Application Anoop Kumar Mishra and Kishan Singh Rawat

ABSTRACT Water is one of the most common substances in the Earth–atmosphere system. The source of all water in its most desirable state is rainfall. It is very crucial in determining the nature of our globe at a fundamental level and is essential for sustaining life by providing pure water over the land. The economy of India is dependent on ­agriculture, which in turn depends on rainfall. This chapter describes the various ways of monitoring rainfall. It describes the limitations of the conventional tools to measure rainfall. Spatial and temporal variability of rainfall over India is also discussed. The chapter gives an introduction to the principle of the satellite remote sensing to monitor rainfall. It discusses the limitations of available global rainfall products for their applicability over India. It also describes techniques to monitor rainfall over the Indian region and their validation with available ground truths. 18.1. ­INTRODUCTION Rainfall is one of the atmospheric parameters that has a direct effect on our life. While rain brings freshwater to support life on Earth, extreme rain events cause destruction and misery due to flash floods and droughts. The accurate estimation of rainfall at various temporal and spatial scales has many applications in meteorology, hydrology, and climate studies. Rainfall is a very difficult parameter to monitor due to its high spatial and temporal variability. Generally, rainfall estimation has been a clear‐ cut matter of setting out rain gauges. Conventional measurement of rainfall from a rain gauge is very accurate, but it gives point measurement. Moreover, rainfall measurement by rain gauges is greatly affected by various factors of topography, site, wind, and gauge design. Weiss and Wilson (1958) reported that rain gauge catch is subjective to the presence of nearby objects and structures that may act as shields and hence decrease the catch from what would have been measured in their absence. Center for Remote Sensing and Geoinformatics, Sathyabama University, Chennai, Tamil Nadu, India

Wind is also one of the factors contributing to errors in rain gauge observations. Monthly rainfall can be underestimated by about 12% by average winds as light as 5 m/s over a region where half the monthly rainfall comes at rates less than 1.8 mm/h. A very dense network of rain gauges has to be set up to measure rainfall accurately over an entire area, which is very expensive and is nearly unfeasible. Mishra (2013) studied a 50 km × 50 km region over the southern part of India and found that three or less rain gauges over the region are not sufficient to monitor daily accumulated rainfall accurately. The density of rain gauge networks is not adequate to monitor rainfall over the Indian region. The advent of weather radar offers an opportunity to monitor rainfall. However, in the case of weather radar too, for the measurements of rainfall, there are various limitations that restrict the use of such a system. These limitations include the lack of a proper relationship of backscattered microwave energy to drop size distribution, reflectivity versus rainfall relationship, beam filling problems, attenuation of the radar beam by intervening drops, absorption and reflection by the ground (anomalous propagation), and signal calibration. Moreover, due to its

Techniques for Disaster Risk Management and Mitigation, First Edition. Edited by Prashant K. Srivastava, Sudhir Kumar Singh, U. C. Mohanty, and Tad Murty. © 2020 John Wiley & Sons, Inc. Published 2020 by John Wiley & Sons, Inc. 295

296  TECHNIQUES FOR DISASTER RISK MANAGEMENT AND MITIGATION

procurement and maintenance expenses, it is less likely to be used in less developed countries like India. In developing countries like India, where rainfall is crucial to everyday life, an effective rainfall measuring technique is essential. Satellite remote sensing has the potential to overcome spatiotemporal inconsistencies of ground‐based rain gauge and radar measurements. It offers a unique opportunity to monitor rainfall over India and the nearby oceanic region. Geostationary satellites, limited to visible (VIS) and infrared (IR) wavelengths, are very essential for rainfall observations due to continuous monitoring of cloud systems. However, VIS observations of rain systems are restricted during the night. Rainfall measurements based on IR observations have large errors because IR radiances from cloud tops have only an indirect and weak relationship with surface rain. Microwave radiation, on the other hand, is able to penetrate through the clouds and thus provide a direct connection with the rain and cloud system. However, microwave measurements suffer from poor spatial and temporal samplings. A combination of highly accurate microwave measurements and continuous IR observations can be used to give accurate rainfall measurements. In the last decade, various satellite precipitation products have become widely available for users using IR, microwave, and merged observations. A majority of these data sets integrate different estimates of precipitation from different sensors and satellites into a precipitation product. These data sets include the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA) near real‐time product (Huffman et al., 2007), the Global Satellite Mapping of Precipitation (GSMaP) (Kubota et al., 2007), Climate Prediction Centre MORPHing (CMORPH) (Joyce et  al., 2004), Precipitation Estimation from Remote Sensing Information Using Artificial Neural Network (PERSIANN) (Sorooshian et al., 2000), and the Hydro‐Estimator (H‐E) (Scofield & Kuligowski, 2003). However, validation results over India show that these products have large errors (Mishra et  al., 2010; Mishra, 2013; Prakash et  al., 2014). The large errors over India may be attributed to the fact that global precipitation products are based on rainfall signatures derived for the globe and these signatures do not work for a topographically complex region like India due to the highly variable nature of precipitation. This chapter begins with a description of rainfall estimation using infrared observations over India and the nearby oceanic region. Next, use of microwave observations to study rainfall by development of regional rainfall signatures for India is described. In this context, the recently launched Indo‐French mission Meghatropiqes and Global Precipitation Mission (GPM) is discussed. A merged rainfall estimation technique using both IR and microwave is also described in this chapter. The last part of this chapter reports an increase in drought and flood

risks due to changing patterns of rainfall, which may be attributed to global warming. 18.2. ­RAINFALL ESTIMATION USING INFRARED OBSERVATIONS VIS and IR imagers from the geostationary weather satellite provide the rapid temporal update cycle needed to capture the growth and decay of precipitating clouds. Cloud observations in visible and infrared images obtained from satellites are often used to estimate rainfall rates over remote areas. Various empirical techniques for rainfall estimation were developed based on the relationship between the cloud brightness temperature and rainfall (Martin & Scherer, 1973). Using data from the Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE), Arkin (1979) found high correlations between areal coverage of cold cloud and 6‐hourly rainfall. Arkin (1979) reported that these correlations improved with increasing spatial and temporal averaging scales up to 2.5° and 24 hours. These results suggested a highly linear relationship between threshold cold cloud amount and climatic‐scale rainfall over tropical oceans. These results were used by Arkin and Meisner (1987) to propose the GOES Precipitation Index (GPI) as a rainfall estimation technique over tropical oceans. The GPI is calculated from the product of mean fractional coverage of clouds colder that 235K in a 2.5° × 2.5° box, the length of the averaging period in hours, and a constant rain rate of 3 mm/h. Durai et al. (2010) used the GPI technique to study rainfall during the southwest monsoon season of India using Kalpana data. They reported that the GPI technique is capable of capturing broad‐scale monsoon rainfall over India. Prakash et al. (2011) also studied large‐scale rainfall features using the GPI technique during the monsoon period of India. Validation results with ­a standard merged satellite product, namely the Global Precipitation Climatology Project (GPCP), and a rain gauge–based product, namely the Global Precipitation Climatology Centre (GPCC), showed that rainfall estimated from the technique is able to capture active and break spells of monsoons. However, it is found that rainfall is overestimated in dry conditions and underestimated in moist conditions because the falling raindrops evaporate before arriving at the surface under dry conditions and tend to grow in size under the moist conditions (Vicente et al., 1998). Thus, water vapor content below the clouds in the atmosphere may influence rain retrieval from the GPI technique. In order to refine the GPI technique from larger climatic scales to shorter scales for daily weather monitoring, Mishra et al. (2011) developed a technique, named Modified GPI (MGPI), to minimize these errors in rainfall estimation subject to the presence/absence of water vapor in the underlying

SPACE‐BASED MEASUREMENT OF RAINFALL OVER INDIA AND NEARBY OCEAN  297



RGPI

3F T (18.1)

where RGPI refers to rainfall from the GPI technique; F refers to fractional coverage of cloud top temperature