344 103 23MB
English Pages 648 Year 2009
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
GEOINFORMATICS FOR NATURAL RESOURCE MANAGEMENT
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in rendering legal, medical or any other professional services.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
GEOINFORMATICS FOR NATURAL RESOURCE MANAGEMENT
P. K. JOSHI, P. PANI, S. N. MOHAPATRA AND
T. P. SINGH
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
EDITORS
Nova Science Publishers, Inc. New York
Copyright © 2009 by Nova Science Publishers, Inc. All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. For permission to use material from this book please contact us: Telephone 631-231-7269; Fax 631-231-8175 Web Site: http://www.novapublishers.com NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers’ use of, or reliance upon, this material. Any parts of this book based on government reports are so indicated and copyright is claimed for those parts to the extent applicable to compilations of such works. Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication. This publication is designed to provide accurate and authoritative information with regard to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA Geoinformatics for natural resource management / editors, P.K. Joshi ... [et al]. p. cm. Includes index. ISBN 978-1-60876-713-7 (E-Book) 1. Natural resources--Management. 2. Natural resources--Geographic information systems. I. Joshi, P. K. HC85.G46 2009 333.70285--dc22 2008047771
Published by Nova Science Publishers, Inc. Ô New York
CONTENTS Preface
ix
About The Editors
xi
Chapter 1 Chapter 2
In-Situ and Remote Sensing Analysis of Harmful Algal Blooms (Habs) Occurrences Associated With Ocean Environments in the South China Sea DanLing Tang, SuFen Wang, Yasuwo Fukuyo and Rhodora V. Azanza
Chapter 3
Monitoring Shallow Lakes in the Pampas Federico Dukatz, Rosana Ferrati, Claudia Marinelli, Rosana Cepeda and Graciela Canziani
Chapter 4
Applications Of Satellite Remote Sensing And GIS To Oceanography And Fisheries: Case Studies In The Western Iberia A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Chapter 5
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Geoinformatics for Natural Resource Management – An Introduction
Chapter 6
Chapter 7
Chapter 8
1
9
33
67
Determining Evapotranspiration and Assessing the Predictability of Vegetation Condition Using Satellite Remote Sensing Methods Chandana Gangodagamage
111
Geoinformatics for Water Accumulation Modelling - A Case Study from India Ashoke Basistha
131
Decision Support Systems Based on Automatic Water Balance Computation for Groundwater Management Planning - The Case of Israel’s Coastal Aquifer David G. Zeitoun and Abraham J. Melloul Enhancing Community Resilience Through Information Management: Flood Risk Mapping in Central Viet Nam Phong Tran and Rajib Shaw
151
165
vi Chapter 9
Chapter 10
Chapter 11
Chapter 12
Chapter 13
Chapter 14
Chapter 15
Chapter 16
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Chapter 17
Contents Systematic Assessment of Forest Cover Change and Forest Fragmentation in Indian Sub-Continent Using Multi-Scale Satellite Remote Sensing Inputs A Giriraj, P.K. Joshi, Shilpa Babar, M. Wegmann, C. Conrad, S. Sudhakar and C. Beierkuhnlein Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage: A Case Study From Bulanikdere, Turkey Fatih Sivrikaya, Günay Çakir, Sedat Keleş and Emin Zeki Başkent
185
215
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing and Modeling Ajit Govind and Jing Ming Chen
249
Geo-Information for Sustainable Development of Agriculture – Some Examples of North India N. R. Patel
291
Fuzzy Clustering Algorithms for Irrigated Area Classification From IRS LISS Data D. Nagesh Kumar
305
Remote Sensing and GIS Aided Land Degradation Assessment in the Greater Mekong Sub-Region Rajendra P. Shrestha and Kingshuk Roy
317
Remote Sensing Techniques for Geological Mapping and Exploration Vern Singhroy
333
Application of Soft Computing for Landslide and its Parametric Analysis T.N. Singh, Kripamoy Sarkar and Akshay Gulati
349
Land Subsidence Phenomena: Recent Developments in Terrestrial and Space-Borne Measurement Techniques and Modeling R.S. Chatterjee
383
Chapter 18
Impact of Informal Settlements on Urban Environment Sadhana Jain
411
Chapter 19
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis T. V. Ramachandra and Uttam Kumar
425
Chapter 20
Potential of the Information Technolgoy for the Public Participation in the Urban Planning Malgorazata Hanzl
Chapter 21
Exploring Alternate Futures with GIS: A Planning Support System Richard E. Klosterman
475 499
Chapter 22
Chapter 23
vii
Using SDI and Service Based Systems to Facilitate Natural Resource Management Ali Mansourian, Mohammad Taleai and Narges Babazadeh
513
Fundamentals for using Geographic Information Science to Measure the Effectiveness of Land Conservation Projects Robert Gilmore Pontius Jr, Shaily Menon, Joseph Duncan and Shalini Gupta
Chapter 24
Validation Of GIS-Performed Analysis Montserrat Gómez Delgado and Joaquín Bosque Sendra
Chapter 25
Digital Governance, Hotspot Geoinformatics, And Sensor Networks For Monitoring, Etiology, Early Warning, And Sustainable Management Ganapati Patil, Sanjay Pawde, Shashi Phoha, Vijay Singhal and Raj Zambre
Index
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Contents
539
559
573
607
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
PREFACE Agenda 21 supports sustainable development while safeguarding the Earth’s environment. This requires optimal management of natural resources which depends on the availability of reliable and timely information at the global, national, regional and local scales. One such technology, ‘Geoinformatics’, consisting of Remote Sensing (RS), Geographical Information System (GIS), and Global Positioning System (GPS) is source of reliable and timely information needed for natural resource management, environmental protection and addressing issues related to sustainable development. It offers a powerful tool, available for the last three decades, for resource assessment, mapping, monitoring, modeling, management etc. It is also capable to make use of recent developments in the digital integration of human reasoning, data and dynamic models. Many institutions and organizations are carrying out various research and operational applications of direct relevance particular to natural resource management. However, there are still limitations in understanding the underlying science and research elements, as there are larger questions of capacity building to use geoinformatics in natural resource management and associated sustainable development applications. These programmes also find gaps between the theoretical concepts and the operational utilization of these tools. This could be solved by providing wide range of applications and prospective potential of this technology to the students and research community in this area. This book contains chapters written by noted researchers and experts. The focus emerged with filling a gap in the available literature on the subject by bringing together the concepts, theories and experiences of the experts in this field. The authors in their contributions have reported their own research and synthesized work in the areas of natural resources such as land, water, oceans, forest, agriculture, urban ecosystems etc. The contents of the papers in this book reflect the experience, expertise, depth of understanding and vision of the authors in their respective fields. We are very grateful to them for their excellent contributions and enthusiastic cooperation. The views expressed in each paper are those of the authors alone. The editors have tried to bring them in a single thread through this book. We are also enriched with the valuable views of numerous experts, researchers and students in this field. We are sure that the publication will serve as a useful reference material to all concerned and interested users, in providing an updated status of knowledge on the subject. We trust that the contributions will be of interest to all. The editors are thankful to Nova Science Publishers for their patience and support of this effort, especially given the time it has taken us to complete it.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ABOUT THE EDITORS Dr. P.K. Joshi is Associate Professor with TERI University, New Delhi India. He holds M.Sc. in Environmental Sciences and M.Sc. Sustainable Development (~ Climate Change) followed by Ph.D. in Environmental Sciences (specialization in RS & GIS). His areas of research are spectral/temporal analysis of vegetation, landscape ecology, biodiversity characterization and resource utilization analysis. His peer recognitions include, Indian National Science Academy (INSA) Young Scientist Medal (2006), Gold-medal of Indian Academy of Environmental Science (IAES) (2006), Fellow of Indian Academy of Environmental Science (FIAES) (2004). He has published more than 75 papers in peerreviewed journals. He has supervised three Ph.D. theses.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Dr. P. Pani is Assistant Professor with Center for the Study of Regional Development, School of Social Sciences, Jawaharlal Nehru University, New Delhi India. She has undergone professional courses in RS and GIS from GDTA Toulouse France and Indian Institute of Remote Sensing Dehradun. She is receipt of 17th M.P. Young Scientist Award (2000). Her area of research is fluvial geomorphology, application of RS & GIS in geoenvironmental and geomorphic studies. She has published more than 10 papers in peer-reviewed journals. She is supervising four Ph.D. theses. Dr. S.N. Mohapatra is Reader with School of studies in Earth Sciences, Jiwaji University, Gwalior India. He holds M.Sc. in Geology, M.Sc. Tech Mineral Exploration, M.Phil. Geology followed by Ph.D. His current research is focused on the geoenvironmental studies of the coal mining areas, landslide prone areas, watershed development, lakes, and land resources evaluation. He has been awarded a certificate of merit by Institution of Engineers (India) (2001) for his research work. He is supervising six Ph.D. theses. He has supervised about 22 MSc Earth Science dissertations and 25 MSc projects in RS/GIS. He has also worked with Defence Terrain Research Laboratory (DRDO) Ministry of Defence. Dr. T.P. Singh is Assistant Professor with Symbiosis Institute of Geoinformatics (SIG), Symbiosis International University, Pune India. He holds M.Sc. in Environmental Sciences, MPhil Remote Sensing and GIS, MS Geoinformatics (University of Paris) followed by Ph.D. in Botany (specialization in RS & GIS). His area of research is application of geinformatics for natural resource management. He has supervised about 20 MSc theses and project works. He is specialized in environmental application of geospatial tools and LiDAR technology. He has published more than 20 papers in peer-reviewed journals.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 1
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
GEOINFORMATICS FOR NATURAL RESOURCE MANAGEMENT – AN INTRODUCTION A better knowledge of resources implies information about their potential, extension, composition and evolution, including their rate of transformation to other uses. There is need to obtain reliable data about resources at global, regional, national and micro-levels, which would help in planning resource management strategy for sustained yield so as to benefit society. Oflate it has been realized that we are not headed towards a sustainable future, but rather towards a variety of potential human and environmental disasters. Human beings have started recognizing that environmental problems are inseparable from those of human welfare and from the processes of economic development. Now, the world has been planning a transition from production oriented schemes to ecological sustainable schemes in every resource sector. So there is an imperative need to look again into the various factors to be considered towards a change for sustainable natural resource management. Agenda 21 of the Rio Earth Summit also identifies the necessity of integration of environmental and development issues while decision making on economic, social, fiscal, energy, agricultural, transportation, trade and other policies, with adequate public participation. It also emphasized that the increasing demand for any natural resource is creating competition and conflict that results in degradation, so an integrated approach to planning must examine all needs so that the most efficient trade-off can be made. Sustainable development of natural resources relies on maintaining the fragile balance between productivity functions and conservation practices through precise identification and systematic monitoring of problem areas in various resources and development sectors. Moreover, it calls for application of alternative resource management practices (such as alternative sources, crop rotation, reclamation of under utilized land/wasteland, energy efficient methods, use of bio-energy etc.) for the various resources. Optimal exploitation of these resources with proper enriching mechanism call for cutting across narrow confined sectoral approaches for taking a holistic view of the region. The foremost step is inventory of natural resources such as soil, water (oceans, rivers, ground water etc.), land use, land cover/natural vegetation, geology/geomorphology etc. through a tool which can arrest spatiotemporal dynamics and inter/intra conversions of these. The governing parameters like climatic factors (rainfall, temperature, humidity, wind speed etc.), socio-political governance (population, polices, practices etc.) and temporal drivers of change (affect of formal factors on resources) need to be collected, integrated, analyzed and represented. One of the most
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
2
P.K. Joshi, P Pani, S.N. Mohapatra et al.
optimal solutions to address these requirements and needs is integration of spatial coverage through remote sensing products, location specific factors through Geographical Information System (GIS), Global Positioning System (GPS), testing various algorithms with statistical tools, incorporation of information technology for better management, storage, retrieval, dissemination of this information bank for implementation, testing, evaluation and feedback mechanism. In totality, geoinformatics for natural resource management encompasses inventory, assessment, monitoring, analysis, integration, retrieval and decision making and communication of information for sustainable management of natural resources. Recent years have seen developments in the involvement of a broad range of stakeholders in many areas of natural resource management. The background to this has included better understanding of responsibilities towards importance of natural resources and the rapid changes in the tools (remote sensing, GPS, GIS, IT etc.) for more holistic and participatory approaches towards monitoring them. The remote sensing systems, right from the aerial photographs, satellite remote sensing (both in optical and microwave), LiDAR etc. have enhanced the capabilities to capture the details in wide range of spatial (from 0.62 m to 1 km), spectral (multi-spectral to hyper-spectral) and temporal (daily to monthly) resolutions. The GPS technology has now emerged with the sub meter accuracy and the facilities of real time mapping, data integration, analysis and presentation. The GIS has evolved from the database creation and spatial analysis to real-time time modeling and virtual mode of decision making. Multi-criteria decision analysis (MCDA) provides an umbrella approach that has been applied to a wide range of natural resource management situations. In combination with the increased focus on participatory resource management, the processes have come with an increased focus on the use of modeling techniques in addressing issues not just on the use of modeling, but also on the selection and adoption of modeling that is appropriate to the scale of management being undertaken. In context to this, natural resources modeling have expanded from the domain of the system expert to the stakeholders, spurring the development of various modeling tools aimed at making modeling more accessible to more people. Thus, the spatial decision support system (SDSS) offers some promise in terms of more adequately accommodating the inherent complexity of natural resource management, embracing ecological, biophysical, social, economical and political components and capturing the multitude of concerns, issues and objectives of stakeholders. The traditional natural resource management strategies were inherited with some of the obvious limitations when dealing with the complex issues. Some are (i) unrealistically presuming or aspiring to substitute analytical results and computations for judgment; (ii) to deemphasize the alternatives in favor of presumably objective feasible and optimal alternatives; (iii) misunderstanding and misrepresenting the drivers; and (iv) lack of framework beyond the typical utility and perceptions. In contrast to the limitations of traditional approaches; a number of workers have proposed application of geoinformatics for natural resource management. The geoinformatics is attributed with the key characteristics such as (i) search for alternatives, not necessarily optimal but which are acceptable on separate dimensions without requiring explicit trade-offs; (ii) reduced data demands with greater integration of social judgments; (iii) simplicity, transparency, efficiency and effectiveness of modeling tools; (iv) taking human beings and their perspectives as an active subject; (v) evaluation of bottom up and down approaches; and (vi) accounting degree of uncertainty guided by attempts to keep options open and resolving conflicts. Initial efforts attempting such integration are indicating a very promising potential for this approach.
Geoinformatics for Natural Resource Management – An Introduction
3
Given the work that is going on around the world towards spatial and temporal monitoring and assessment systems for natural resources management, particularly in the areas of integrated dynamic modeling with strong spatial data capabilities, it is probable that better understanding of these tools are required to meet these necessitates. Some of the broad areas where geoinformatics is being used are summarized in table 1. Table 1: Broad applications of geoinformatics for natural resource management Broad Areas Agriculture
Disaster Environment
Forest
Geology Land use Marine Soil
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Terrain
Monitoring Mixed-crop discrimination and inventory Large area crop inventory Crop stress detection monitoring Flood damage assessment Flood relief measures Environmental Impact Assessment (EIA) Monitoring sitting applications Types and species discrimination Forest stock mapping Biomass estimation Geological mapping Geomorphologic mapping Land use/Land cover mapping Change detection Coastal area mapping Oceanogrpahy Soil mapping Erosion prone mapping Cadastral mapping
Topography/ Surface
Contours Models Slope/aspect Visualization perspective
Urban
Town and city mapping Facility mapping Surface water monitoring Ground water targeting Command area management
Water Resources
Assessment Acreage forecasting Suitability analysis Understanding responses climate change Flood modeling Scenario generation Predictive modeling Scenario generation
of
Biodiversity characterization C and C stock assessment Land atmosphere interaction Landslide zonation Earthquake Markov change analysis Cellular automata Sea-surface temperature Chlorophyll estimation Soil erosion modelling Facility management (AM/FM) Photogrammetry Topographic modeling Risk assessment Generation of Information system Heat island/Oasis affect Infrastructure development Drought and risk modeling Runoff modeling Responses to climate change
The above examples are not the extensive areas but some of the most common monitoring and assessment applications possible with the geospatial tools for natural resource management. Understanding and proper implementation of these tools require a priori knowledge of the underlying physical, biological, chemical processes and the possible modes to integrate them. This knowledge can be obtained only through co-operation and research among scientists from pertinent disciplines. Therefore, it is imperative to encourage and
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
4
P.K. Joshi, P Pani, S.N. Mohapatra et al.
establish open communication and future co-operation among scientific community using geoinformatics as a tool to study the distribution of natural resources and their management. In this book, we intend to cover the application of geoinformatics in a wide range of natural resource management that are concerned with land surface interactions and processes. The chapters included in this book encompass wide range of resources viz., environment, ecosystems (terrestrial/aquatic), land use/land cover, geology, forests/vegetation, agriculture, urban areas, fisheries, water etc. These studies have demonstrated case studies to explain the importance of geoinformatics for natural resource management and their linkages with the human society. The coverage is complete but have a very wide range of areas listed amount the threatened natural resources which need special attention both by the policy makers and academicians cum researchers. Water is an all important resource for the human and other life forms on this planet. In many of the systems human and physical processes have further limited the quality and supply of water. Algal blooms are one of the challenges that occur frequently in the range of aquatic ecosystems such as oceans, sea, rivers to the local water bodies; causing enormous economic and ecological losses. These also touch upon the standard of living for citizens but also preserve resources to protect natural biodiversity and environment. Apart from this, flow of people, commerce, crops, and life in distinctly different regions call for the water resource monitoring, assessment and management. In the very contribution to this book, Tang et al. analyze regional, seasonal and annual variations of harmful algal blooms records during the period from 1980 to 2003 in South China Sea. They provide a very intensive taxonomy of the algal species found in the region their distribution and prospective impacts. In the third chapter, Dukatz et al. propose a real time classification of water bodies and the inference of their trophic state through the various hydrological and climatic phases using optical remote sensing data. The water bodies that were selected to illustrate this chapter were eight shallow lakes located in the Province of Buenos Aires, Argentina. Santos et al. (chapter 4) provide a brief description of principles, technologies and methodologies of satellite remote sensing and marine GIS oceanography and fisheries research and support. Determining crop-water requirement for irrigation is in high priority at all scales of irrigation schemes. The most basic, but highly important questions that need to be addressed in irrigation are when to irrigate (temporal resolution), where to irrigate (spatial resolution) and how much of water is needed for irrigation. Gangodagamage (chapter 5) uses Moderate Resolution Imaging Spectroradiometer (MODIS) sensor derived Normalized Difference Vegetation Index (NDVI), Enhanced vegetation Index (EVI), and thermal imageries to extract the spatial and temporal variability of the vegetation health and thermal stress of crops in irrigated agriculture fields over a periods of two years at16-day temporal resolutions. Basistha (chapter 6) takes an example from central India to advocate potential of geoinformatics in estimating water accumulation in check dam sites. Zeitoun and Melloul (chapter 7) explain computer based DSS, GIS and hydrological modeling for implementation of water management and planning. The study uses statistical estimation of volume water balance and the mass balance inside the selected aquifer. The other very important aspect of water resources in abundance in a particular time and area that is generally referred to flood, which need to mapped, monitored and used to develop mitigation plan. Tran and Shaw (chapter 8) demonstrate a process of integrating local knowledge into disaster risk management and lessons learned from a participatory GIS to prepare detail flood risk maps for the commune planners, villages and other stakeholders in Central Viet Nam.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Geoinformatics for Natural Resource Management – An Introduction
5
Land use/ land cover change (LULCC) is regarded as the most important variable of global change affecting ecological systems with an impact on natural resources including soil and water quality, global climatic system and biodiversity, biogeochemical cycles, quantifying woody biomass and climate change feedbacks. Giriraj et al. (chapter 9) propose Rapid Ecological Alerts (RET’s) using temporal MODIS Vegetation Continuous Field (MODIS 44B-VCF) to monitor tree cover change in the Indian region over 6 years (2000 – 2005) period. Sivrikaya et al. (chapter 10) have taken another example from Turkey to demonstrate both spatial and temporal changes of land use/land cover and their effects on landscape pattern to release the implications for landscape planning and ecosystem management. This work also looks into explicit estimates of the carbon storage (aboveground and belowground) using inventory data and geospatial tools. The terrestrial carbon (C) cycle has a great role in influencing the climate. Hence, it is critical that we fully understand the scale-specific (spatial and temporal) complexities of the terrestrial C cycle. Govind and Chen (chapter 11) introduce various remote sensing approaches, both empirical and process based strategies that are currently used in C cycle research. Agriculture is the mainstay of the economy of Asian and developing countries and only sustainable agriculture is likely to provide the long-term benefits required to achieve development and poverty alleviation. Patel (chapter 12) discusses some major areas of sustainable development of agriculture with some key examples from North India. For the sustainable management of this, assessment at field level crop conditions and productivity throughout the growing season is a great challenge before researchers. Kumar (chapter 13) evaluates the use of fuzzy classification techniques using multi-temporal satellite imagery to classifying crops in southern part of India. Land degradation has been an important concern while talking about food security and environmental conservation. Although there have been attempts to produce baseline desertification and land degradation information at the global and regional level, there is a need for more precise large-scale data as the earlier attempts were based on expert judgments and of coarse-scale in nature. Shrestha and Roy (chapter 14) use vegetation cover, runoff, water use efficiency and soil loss as indicator parameters to assess land degradatation in parts Myanmar. The earth crust provides a substratum to living creatures on this planet. Apart from this it is the source of numerous mineral resources which are very important in day to day activities of human beings and its kind. Singhroy (chapter 15) explains advanced Earth Observation (EO) techniques that are currently being used in geological mapping and mineral exploration and geohazard monitoring. He has covered a wide range of avialable sensors and their comparitive account. The geological and geomorphological setup is ab important aspect while studying disasters such as landslide, earthquake, landsubsidence. Singh et al. (chapter 15) have focused on application of soft computing tool like Artificial Neural Network, Genetic Algorithm and Fuzzy Logic to asses the behavior of landslide parts of lesser Himalaya. Land subsidence is another very important geophysical phenomenon which is generally categorized under risk and hazard. Chatterjee (chapter 17) exaplains range of terrestrial and space-borne techniques available for measuring the land subsidence in different parts of India. Jain (chapter 18) touches a very sensitive and important issue of natural resource management while analyzing impact of informal settlements on urban environment. She emphasizes the importance of reliable and timely information about the locations, form and morphology of these areas for facilitating proper decision making and planning.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
6
P.K. Joshi, P Pani, S.N. Mohapatra et al.
Ramachandra and Kumar (chapter 19) with an example of Banglore city have analyzed urbanization pattern and trends using multi- temporal and multi-resolution remote sensing data. The study unravels the pattern of growth and its implication on local climate and also on the natural resources. In similar lines, Hanzl (chapter 20) tries to explain importance of information technology along with geospatial tools in urban planning with citizens’ participation. Klosterman (chapter 21) uses two examples from US to describe GIS based planning support system (PSS) called What if?TM to evaluate alternative scenarios for the suitability of land and alternative public polices. We are thankful to Ms Ann LeRoyer, Senior Editor, Lincoln Institute of Land Policy to kindly permit reprinting the the chapter. There are numerous resource management related issues and problems that represent immediate concerns for sustainable development of the countries. Due to diversity of natural resources, broad spectrum of governmental and private organizations from different political/administrative levels (local, state/provincial and national levels) are involved for decision-making and planning process of managing natural resources. The delivery of a sustainable natural resource management strategy relies on the collaborative efforts and coordination of involved organization, wherein spatial data is the most crucial one. Mansourian et al. (chapter 22) argue for the design and implementation of a spatial data infrastructure model to better resolving the challenges because of mult-source datasets. Development vis-à-vis conservation has always been a matte of discussion with the diversified stakeholders. Pontius et al. (chapter 23) present a general conceptual framework to assess the effectiveness of land conservation projects by using GIS and land change modeling to analyze development and conservation in the presence of leakage in part of the IndoMalayan realm. Modeling and decision making with geospatial tools are the recent trends of natural resource management. The most important part is the degree of uncertainty and the validation of the developed models before setting them up at operation mode. Delgado and Sendra (chapter 24) explain the process for a risk control procedure of the decision making based on an uncertainty analysis of the data and the model used. Taking the tools to real time application at operational level, Patil et al. (chapter 26) present a live example of application of geoinformation science for sustainable natural resource management and digital governance. Both the authors describes a three-echelon information systems design at regional/district/block levels; in the context of Jalagaon district of India, which has the world’s first Watershed Surveillance Research Institute named JALASRI. Natural resource management relies on the use of the potential tools, economic and production data, which are often available at different non-comparable scales. This causes some challenges in data interpretation and on the availability and cross compatibility of official data for detailed modeling and analysis. Integration of geoinformatics and modeling aspects offer the capacity to examine the effect of interacting natural systems. This book is an effort towards understanding these interactions at a broader scale.
Geoinformatics for Natural Resource Management – An Introduction
7
QUESTION BANK Expand the Following 1. 2. 3. 4. 5.
AM/FM DSS EO EVI GIS
6. 7. 8. 9. 10.
GPS IT LiDAR LULCC MCDA
11. 12. 13. 14. 15.
MODIS NDVI PSS RET SDSS
Fill in the Blanks 1. Agenda 21 of the Rio Earth Summit also identifies the necessity of integration of __________ and __________ issues. 2. Geoinformatics for natural resource management encompasses __________, __________, __________, __________, __________, __________, __________ and __________ of information for sustainable management of natural resources. 3. Now, the world has been planning a transition from __________ oriented schemes to __________ sustainable schemes in every resource sector. 4. Sustainable development of natural resources relies on maintaining the fragile balance between __________ functions and __________ practices. 5. Understanding and proper implementation of geospatial tools require __________ knowledge of the underlying processes.
Short Answer Questions
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. Describe the challenges of the natural resource management. 2. How a society is responding towards sustainable development? 3. What are the recent changes in technology applications towards natural resource inventory? 4. What are limitations of traditional strategies of resource management? 5. What are characteristics features of geoinformatics for natural resource management?
Long Answer Questions 1. What is Geoinformatics? How geoinformatics technology has changed the natural resource monitoring? 2. Give an account of various strategies of natural resource management. Explain the drawbacks of traditional strategy.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 2
IN-SITU AND REMOTE SENSING ANALYSIS OF HARMFUL ALGAL BLOOMS (HABS) OCCURRENCES ASSOCIATED WITH OCEAN ENVIRONMENTS IN THE SOUTH CHINA SEA DanLing Tang1, SuFen Wang1, Yasuwo Fukuyo2 and Rhodora V. Azanza3,* 1
Research Center of Remote Sensing on Marine Ecology and Environment (RSMEE), South China Sea Institute of Oceanology, Chinese Academy of Sciences, China 2 Asian Natural Environmental Science Center, Tokyo University, Japan 3 University of the Philippines, Diliman, Philippines
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ABSTRACT Harmful algal blooms (HABs) occur frequently in the South China Sea (SCS), causing enormous economic losses in aquaculture. This paper analyzed historical HAB records during the period from 1980 to 2003 in SCS. The data show that HABs-affected areas have expanded and the frequency of HABs varied during this period. The seasonal and annual variations, as well as causative algal species of HABs are different among the four regions. Areas with frequent HABs include the Pearl River Estuary (China), the Manila Bay (the Philippines), the Masinloc Bay (the Philippines), and the western coast of Sabah (Malaysia). HABs occurred frequently during March-May in the northern region of SCS, May-July in the eastern region, July in the western region, and year-round in the southern region. Among the species that cause HABs, Noctiluca scintillans dominated in the northern region, and Pyrodinium bahamense in the southern and eastern regions. Causative species also varied in different years for the entire SCS. Both P. bahamense and N. scintillans were the dominant species during 1980 to 2003. Some species not previously recorded formed blooms during 1991-2003, including Phaeocystis globosa, Scrippsiella trochoidea, Heterosigma akashiwo and Mesodinium rubrum. Variations in HABs are related to various regional conditions, such as a reversed monsoon wind in the entire SCS, river discharges in the northern area, upwelling in *
Email: [email protected]
10
DanLing Tang, SuFen Wang, Yasuwo Fukuyo et al. Vietnam coastal waters during southwest winds and near Malaysia coastal waters during northeast winds, and eutrophication from coastal aquaculture in the Pearl River estuary, Manila Bay and Masinloc Bay.
Key words: Algal blooms, aquaculture, China, GIS, remote sensing
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
INTRODUCTION Harmful Algal Blooms (HABs) have spread worldwide and become a global problem causing enormous economic loss and serious impacts on human health (Anderson, 1997a; b), HABs have been reported from Florida (Philips et al., 2004), British Columbia (Taylor and Haigh, 1993), Norwegian waters (Dahl and Tangen, 1993; Vila et al., 2001), northeast Atlantic (Edwards and Johns, 2006), and the upwelling regions off South Africa (Pitcher and Calder, 2000). Detrimental ecosystem effects associated with HABs range from pelagic and benthic community mortalities to fish/shellfish aquaculture mortalities, attributable to both biomass losses (i.e., due to low dissolved oxygen) and toxic effects for humans (Heil et al., 2005; Anderson et al., 1994). The worldwide increase in aquaculture is part of the problem, and has been blamed for pollution of the ecosystem (ICLARM, 1993; Stewart, 1997), especially in Asia (FAO/NACA, 1995). In a positive feedback cycle, the increasing frequency, intensity and geographic distribution of HABs poses a serious threat to the coastal fish/shellfish aquaculture and fisheries (Anderson, 1998). Intensive marine farming and rapid industrialization over the past two decades along the coast of the South China Sea have placed a heavy stress on the marine environment and greatly degraded marine ecosystems in the region. Massive mortality and devastating diseases as well as HABs in marine farms have already hampered the sustainable development of the marine culture industry (Qi et al., 2004). The understanding of historical HAB distribution and concentration in the South China Sea offers a basis for urgent management measures required to mitigate deteriorating coastal water quality and the adverse environmental impacts on aquaculture development (Chua et al., 1989; FAO/NACA, 1995). The littoral countries of the SCS have similar coastal ecosystems and access to common marine resources, such as coastal cultivation of oysters and shrimp, and deep-sea fishing for tuna and other migratory species. The sea plays an important role in the economies of the coastal nations, by providing food and employment for the increasing population. About half of the coastal population's protein intake comes from the sea (Roseberg, 1999). The problems of environmental pollution around the SCS are primarily due to population growth and urbanization in coastal cities, economic growth and increased material consumption, and highly polluting technologies for energy production and primary resource extraction. In the SCS, HABs have caused mass mortality of fishes and other ecosystem impacts (Qi et al., 2004; Xia and Wu, 1996; Tang et al., 2003) in the past 20 years. For example, in November 1997, HABs off Raoping (Guangdong province, China) led to an estimated economic loss more than US$ 8 million (Qi et al., 2001). During March to April 1998, HABs in the mouth of the Pearl River in Guangdong province and Hong Kong caused economic losses estimated at over US$45 million (Qian et al., 2000; Tang et al., 2003). In summer 1983, Pyrodinium bahamens bloomed in Magued Bay and the Samar Sea, Philippines, poisoning 700 people, killing 70 people, and causing economic losses up to US$500,000
In-Situ and Remote Sensing Analysis of Harmful Algal Blooms…
11
(Maclean, 1989). In 2002, a Prorocentrum minimum bloom in the northern Philippines led to economic losses of US$120,000 (Azanza et al., 2005). Local HAB studies in the SCS area have focused on algal species (Hodgkiss and Lu, 2004; Azanza et al., 2005; Masuda et al., 2001), environmental factors (Tang et al., 2003; Usup et al., 2002; Yoshida et al., 2000), and oceanic dynamics (Tang et al., 2004a; 2004b; 2006a) in association with upwelling that supports long term phytoplankton blooms (Tang et al., 2004b). Due to the extensive area of the SCS, many environmental factors, such as monsoon winds, weather, ocean circulation, upwelling and human activities play roles in HAB formation (Han et al., 1995; Tang et al., 2004a). The aims of this study were to analyze the spatial and temporal dynamics of HAB occurrences from 1980 to 2003, and identify potential risk factors of HABs in the SCS. The study addresses the need to improve our understanding about the distribution of HABs and the conditions of HABs outbreaks in the region, thus enhancing our predictive ability.
STUDY AREA
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The South China Sea (SCS) is the largest semi-closed sea in the western tropical Pacific Ocean (Fig. 1A). It is surrounded by Malaysia, Thailand, Vietnam, Brunei, Indonesia, Philippines and China. The coastal fringes of SCS are home to about 300 million people in 2000 (Fig. 1A) (Global Statistics, 2000). The SCS is a distinctive ecosystem, in which the combination of geology and climate produces a remarkable amount of biological diversity and immense genetic resources (Roseberg, 1999).
Figure 1. A. Study area –the South China Sea (SCS). Numbers in circles show human population (in million) of the local coastal regions. B. The number of monitoring stations and routine sampling frequency for HABs in the SCS from 1980 to 2003.
12
DanLing Tang, SuFen Wang, Yasuwo Fukuyo et al.
With a total area of 3.5 ×106 sq km, the SCS is the largest marginal sea in the world (Fig. 1A). Northeast monsoon winds occur from November to March and southwest monsoon winds from June to September, with transition periods in April-May and September-October (Mohsin and Ambak, 1996). A large anticyclonic circulation (Fig. 2A) appears during the southwest monsoon, and a cyclonic circulation (Fig. 2B) occurs during the northeast monsoon (modified from Hu et al., 2000 and Tang et al., 2004a). Four major (Pearl, Mekong, Han and Red Rivers) (Fig. 2B) and many small rivers discharge into the sea.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 2.: Upper-layer mean circulation patterns in the SCS and satellite images of Chl a concentration and SST for phytoplankton bloom events (Hu et al., 2000). Four rivers are identified by white arrows: HR (Han River); RR (Red River); PR (Pearl River); MR (Mekong River). A. Southwestern monsoon. a and b: satellite images showing upwelling and phytoplankton bloom in the coastal water of Vietnam (Tang et al., 2004a; 2004b); c and d: SST images for summer season (June to August) of 1998 and 1995-2004 in the SCS (Zhao and Tang, 2006). B. Northeastern monsoon. The numbers in squares show the annual river discharge (cu km). a: SeaWiFS images showing high Chl a concentration in the Taiwan Bank Upwelling (Tang et al., 2002); b: CZCS Chl a image showing a phytoplankton bloom in Luzon Strait ; c and d: Chl a and SST in a upwelling area near Sabah (Isoguchi and Kawamura, 2006).
We sorted the data for four SCS regions (Fig. 3) influenced by similar hydrographic conditions: A. North (southern China and northern Vietnam); B. East (mainly west of the Philippines); C. South (western Malaysia, Brunei and Palawan Island) and D. West (southern coast of Vietnam).
In-Situ and Remote Sensing Analysis of Harmful Algal Blooms…
13
Figure 3. (a) HAB distributions in the SCS from 1980 to 1990. (b) HAB distributions in the SCS from 1991 to 2001. Boxes show four study regions. Each red circle means one HAB occurrence. PR: Pearl River; GD: Guang Dong Province, RR: Red River Delta; BT: Binh Thuan Province; S: Sabah; Ma: Manila Bay; MC: Masinloc Bay.
DATA AND METHODS
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
HAB Data Collection and Analysis HAB data were collected and compiled from various sources, including government statistics reports of China, Malaysia, Vietnam, Brunei and the Philippines (96 websites), research journals (107 papers), conferences and other reports (80 reports), and newspapers. For the northern SCS, the collection did not include HABs occurring in Hong Kong, where the HAB outbreak is a regular occurrence and requires a separate study. For the western region, there were no HAB records prior to 1990. We analyzed the affected areas, seasonal frequency, and causative species of HABs, and changes for 1980 to 1990 versus 1991 to 2001. Routine monitoring of HABs has been carried out in many countries; the number of monitoring stations and routine sampling frequency in the present study was collected from those departments of China, Malaysia, Vietnam, Brunei and the Philippines that were in charge of routine monitoring on HABs (Fig. 1B).
Satellite Data Quickscat Sea Surface Wind Wind speed and direction over the ocean surface were retrieved from QuikScat measurements of backscattered power (Wentz et al., 2001). Monthly averaged QuickScat
14
DanLing Tang, SuFen Wang, Yasuwo Fukuyo et al.
wind vector images were produced by Remote Sensing Systems and sponsored by the NASA Ocean Vector Winds Science Team (http:// www.ssmi.com/qscat/qscat_ description.html). Avhrr Sea Surface Temperature (Sst) SST data from the Advanced Very-High Resolution Radiometer (AVHRR) Pathfinder Version 5 SST Project (spatial resolution 4km/month-daytime), a reanalysis of the AVHRR data stream, were obtained from the Physical Oceanography Distributed Active Archive Center (PODAAC), the Jet Propulsion Laboratory (JPL), and NASA (http://podaac.jpl.nasa.gov/sst). Seawifs-Derived Chl A SeaWiFS-derived Chl a images of 1×1 km2 spatial resolution were processed through the SeaWiFS Data Analysis System (SeaDAS) using the Ocean Color 4-band algorithm (OC4,4) (O'Reilly 1998). To investigate the spatial distribution of Chl a, we first processed images at 1×1 km2 spatial resolution, and then made monthly average Chl a images at 4×4 km2 (Tang et al., 2003b). Monthly satellite images of winds, SST and Chl a concentration for March, June, September and December were processed for 2001 to present general conditions for the SCS, because satellite data coverage are relatively good for this year (Tang et al., 2004a; 2006a). Some other images were processed to analyze special HAB occurrences.
Results
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Spatial distribution The occurrence of HABs increased and the affected area spread substantially from 1980 to 2001 (Fig. 3). HABs occurred frequently off the Pearl River, Sabah, Masinloc and in the Manila Bay from 1980-2001 (Fig. 3). Among the four regions the southern region (C) had the highest frequency, with 369 HABs observed (Fig. 4). The fewest HABs were observed in the western region (D in Fig. 4), but HABs were not officially reported in Vietnam prior to the early 1990s.
Figure 4. The total numbers of HAB in the four SCS regions of the SCS from 1980 to 2003.
In-Situ and Remote Sensing Analysis of Harmful Algal Blooms…
15
Occurrences of Habs During 1980 To 2003 In the northern region (A in Fig. 5), HABs occurred most frequently in 1990 (19), 1991 (24), and 1998 (18). In the eastern regions, HABs occurred almost every year. In the southern region, HABs also occurred almost every year, but most frequently from 1994 to 1998 (C in Fig. 5). A few HABs were observed in the western region, but with two peaks in 1999 (9 HABs) and 2002 (8 HABs) (D in Fig. 5). For the entire SCS area, HABs were frequent in 1991 (57), 1995 (55) and 1998 (67) and a high number occurred from 1994 to 1998 (> 40 y-1) (E in Fig. 5). About 206 HABs were observed during 1980-1990, and 499 during 1991-2003.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 5. Reports of HAB events in the SCS from 1980 to 2003.
Seasonality HABs have strong seasonal characteristics in the SCS area (Fig. 6 and 7). In the northern region, the highest frequency (33) was in April. The highest frequency (25) is in June in the eastern regions and in July (8) in western regions. In the southern region, HABs occurred relatively evenly year-round (20-40 y-1) (Fig. 6). For the entire SCS, from 1980 to 2003, March to May had the highest number of HAB records (>70 month-1) and September to November had the lowest (30-50 month-1) (E in Fig. 8). Very few HABs were recorded in the western region during September to February (0.05) (Fig. 10A and B). HAB occurrence decreases after 1998 (E in Fig. 6) though the frequency of monitoring continues to increase (Fig. 1b), and so the ratio (HAB occurrence /HAB monitoring frequency) decreases in recent years (Fig. 10B). The increase of monitoring frequency may contribute only a small part (Fig. 10 A) to the increase in HAB reports.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 10. A. Relation between HAB occurrence and monitoring frequency during 1980 to 2003. B. Ratio: Divide annual HAB occurrences by annual monitoring frequency.
SUMMARY The present study shows regional, seasonal and annual variations of HABs in the South China Sea. The causative algae varied among regions and periods. Key factors influencing HABs differ across the regions (Table 2), including wind, current, temperature (factors are not showed in Table 2). Eutrophication appears to be the key factor for HABs in coastal and bay waters, such as the Pearl River estuary (A in Fig. 11), Manila Bay and the northwest coast of Sabah (B and C in Fig. 11). Monsoon winds inducing regional upwelling may be one of the key reasons for HABs in some locations, such as coastal water along Vietnam in the southwest monsoon season (D in Fig. 11) and coastal water along Sabah in northestern monsoon (Upwelling in
26
DanLing Tang, SuFen Wang, Yasuwo Fukuyo et al.
Fig. 11). Meanwhile, strong wind can resuspend cysts above the sediment, potentially causing blooms in Manila Bay and west of Sabah. Coastal currents may shift HAB masses between Sabah and Palawan in the south region. Relatively high frequencies of HABs in the northern region and low frequencies in the western area in 1998 may be related to the specific environmental conditions associated with El Niño in the area.
Figure 11. Four examples of HABs (A, B, C, and D) with associated environmental factors in the SCS. Eu: Eutrophication; NE: Northeast wind; SW: Southwest wind; Up: Upwelling.
ACKNOWLEDGMENTS
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The present work was supported by following grants awarded to DanLing TANG: One Hundred Talents Programme of Chinese Academy of Sciences (CAS); The CAS/SAFEA International Partnership Program for Creative Research Teams.
REFERENCES Alexander, M.A., Blade, I., Newman, M., Lanzante, J.R., Lau, N.C. and Scott, J.D. (2002). The Atmospheric Bridge: the Influence of ENSO Teleconnections on Air-Sea Interaction Over the Global Oceans. Journal of Climate 15: 2205-2231. Anderson, D.M. (1997a). Bloom dynamics of toxic Alexandrium species in the northeastern United States. Limnol. Oceanogr 42: 1009-1022. Anderson, D.M. (1997b). Turning back the harmful red tide. Nature 388: 513-514.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In-Situ and Remote Sensing Analysis of Harmful Algal Blooms…
27
Anderson, D.M. (1998). Study of red tide monitoring and management in Hong Kong: literature review and background information. Technical Report No. 1, Hong Kong Agriculture and Fisheries Department: 120. Anderson, D.M., Kulis, D.M., Doucette, G.J., Gallager, J.C. and Balech, E. (1994). Biogeography of toxic dinoflagellates in the genus Alexandrium from the northeast United States and Canada as determined by morphology, bioluminescence, toxin composition, and mating compatibility. Marine Biology 120: 467-478. Azanza, R.V and Taylor, F.M. (2001). Are Pyrodinium blooms in the Southeast Asian region recurring and spreading? A review at the end of the millennium. Ambio 30: 356-364. Azanza, R. V., and I. U. Baula, 2005. Fish kills associated with Cochlodinium blooms in Palawan, the “last frontier” of the Philippines. Harmful Algae News 29: 13-14. Azanza, R.V., Fukuyo, Y., Yap, L.G. and Takayama, H. (2005). Prorocentrum minimum bloom and its possible link to a massive fish kill in Bolinao, Pangasinan, Northern Philippines. Harmful Algae 4: 519-524. Badylak, S and Phlips, E.J. (2004). Spatial and temporal patterns of phytoplankton composition in a subtropical coastal lagoon, the Indian River Lagoon, Florida, USA. Journal of Plankton Research 26: 1229-1247. Chua, T. E., Paw, J.N. and Guarin, F.Y. (1989). The environmental impact of aquaculture and the effects of pollution on coastal aquaculture development in Southeast Asia. Marine. Pollution Bulletin 20: 335-343. Corrales, R.A and Crisostomo, R. (1996). Variation of Pyrodinium cyst density in Manila Bay, Philippines. In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms. Intergovernmental Oceanographic Commission of UNESCO : 181-183. Dahl, E and Tangen, K. (1993). 25 years experience with Gyrodinium aureolum in Norwegian waters. In: Smayda, T.J., Shimizu Y. (eds.), Toxic phytoplankton blooms in the sea. Elsevier, Amsterdam: 15-22. Edwards, M and Johns, D.G. (2006). Regional climate change and harmful algal blooms in the northeast Atlantic. Limnology and Oceanography 51: 820-829. FAO/NACA. (1995). Report on a regional study and Workshop on the Environmental Assessment and Management of Aquaculture Development. FAO, Rome and Network of Aquaculture Cent. in Asia and the Pacific, Bangkok, Thailand. NACA Environment Aquaculture Development Series 1: 492. Garcés-Vargasa, J., Schneider, W., Abarca del, R.R., Rodney, M.R. and Zambrano, E. (2005). Inter-annual variability in the thermal structure of an oceanic time series station off Ecuador (1990–2003) associated with El Niño events. Deep-Sea Research I 52: 17891805. Global Statistics (2000). http://www.geohive.com/ Han, W.Y., (1995). Marine chemistry of the Nansha islands and South China Sea. China Ocean Press, Beijing. Heil, C.A., Glibert, P.M. and Fan, C.L. (2005). Prorocentrum minimum (Pavillard) Schiller: A review of a harmful algal bloom species of growing worldwide importance. Harmful Algae 4: 449-470. Hodgkiss, I.J and Ho, K.C. (1997). Are changes in N:P ratios in coastal waters the key to increased red tide blooms? Hydrobiologia 352: 141-147.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
28
DanLing Tang, SuFen Wang, Yasuwo Fukuyo et al.
Hodgkiss, I.J and Lu, S.H. (2004). The effect of nutrients and their ratios on phytoplankton abundance in Junk Bay, Hong Kong. Hydrobiologia 512: 215-229. Hu, J., Kawamura, H., Hong, H. and Qi, Y. (2000). A review on the currents in the South China Sea: Seasonal circulation, South China Sea Warm Current and Kuroshio intrusion. Journal of Oceanography 56: 607-624. ICLARM. 1993. Environment and Aquaculture in Developing Countries. International Center for Living Aquatic Resources Management, Manila, Philippines, Conf. Proc. 31: 359. Isoguchi, O and Kawamura, H. (2006). El Niño–related offshore phytoplankton bloom events around the Spratley Islands in the South China Sea. Geophysical Research Letters 32: L21603, doi:10.1029/2005GL024285, 2005. Ji, S. (2003). Noctiluca scintillans red tide. In: Wang, Y., Wang, S.S., Qi, Y.Z., Ji, S., Wu, Y.L., Zou, J.Z., Zhang, Z.X., Lin, Y.S., Lin, Y.T., Zhou, M.J., Hong, J.C., Xia, Z.W., Qian, H.L., Xu, N., Huang, C.J. and Liang, S. Red tide on the coasts of China. Science Press, Beijing. Kuo, N.J., Zheng, Q.A. and Ho, C.R. (2004). Response of Vietnam coastal upwelling to the 1997-1998 ENSO event observed by multisensor data. Remote Sensing of Environment 89: 106-115. Lim, P.T., Usup, G., Leaw, C.P. and Ogata, T. (2005). First report of Alexandrium taylori and Alexandrium peruvianum (Dinophyceae) in Malaysia waters. Harmful Algae 4: 391-400. Maclean, J.L. (1989). Indo-Pacific red tide 1985-1988. Marine Pollution Bulletin 20: 304310. Masuda, M., Abe, T., Kawaguchi, S. and Phang, S.M. (2001). Taxonomic Notes on Marine Algae from Malaysia. Ⅵ. Five Species of Ceramiales(Rhodophyceae). Botanica Marina 44: 467-477. Mohsin, A.K.M and Ambak, M.A. (1996). Marine fishes and fisheries of Malaysia and neighbouring countries. Universiti Pertanian Malaysia Press, Serdang. O'Reilly, J.E., Maritorena, S., Mitchell, B.G., Siegel, D.A., Carder, K.L., Garver, S.A., Kahru, M. and McClain, C. (1998). Ocean color chlorophyll algorithms for SeaWIFS. Journal of Geophysical Research 103(24): 937-953. Philips, E.J., Badylak, S., Youn S. and Kelley, K. (2004). The occurrence of potentially toxic dinoflagellates and diatoms in a subtropical lagoon, the Indian River Lagoon, Florida, USA. Harmful Algae 3: 39-49. Pitcher, G.C and Calder, D. (2000). Harmful algal blooms of the southern Benguela Current: A review and appraisal of monitoring from 1989 to 1997. South African Journal of Marine Science 22: 255-271. Qi, Y.Z., Hong, Y. Lu, S.H. and Chu, J.H. (1994). Outbreak of Chattonella marina red tide and its relations to ecological parameters in Dapeng bay, South China Sea. Oceanologia et Limnologia Sinica 25: 132-138 (in Chinese, with English abstract). Qi, Y.Z., Chen, J.F., Wang, Z.H., Xu, N., Wang, Y. and Shen, P.P. (2004). Some observations on harmful algal bloom (HAB) events along the coast of Guangdong, southern China in 1998. Hydrobiologia 512: 209-214. Qi, Y.Z., Shen, P.P. and Wang, Y. (2001). Taxonomy and lifecycle of genus Phaeocystis. Journal of Tropical and Subtropical Botany 9: 174-184 (in Chinese, with English abstract).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In-Situ and Remote Sensing Analysis of Harmful Algal Blooms…
29
Qian, H.L and Liang, S. (1999). Study on the red tide in the Pearl River Estuary and its near waters. Marine Environmental Science 18: 69-74 (in Chinese, with English abstract). Qian, H.L., Liang S. and Qi, Y.Z. (2000). Study of the characteristics and the causes of formation on the red tide in coastal water of Guangdong. Ecologic Science 19: 8-16 (in Chinese, with English abstract). Rosenberg, D., (1999). Environmental pollution around the South China Sea: Developing a regional response to a regional problem. Resource Management in Asia-Pacific Working Paper No. 20. Resource Management in Asia-Pacific Project, Division of Pacific and Asian History, Research School for Pacific and Asian Studies, The Australian National University Publishers, Canberra. Schoemann, V., Becquevort, S., Stefelsb, J., Rousseau, R. and Lancelot, C. (2005). Phaeocystis blooms in the global ocean and their controlling mechanisms: a review. Journal of Sea Research 53: 43-66. Shaw, P.T and Chao, S.Y. (1994). Surface circulation in the South China Sea. Deep Sea Research. Part I 41: 1663–1683. Stewart, J.E. (1997). Environmental impacts of aquaculture. World Aquaculture 28: 47-52. Tang, D.L., I.H. Ni, D.R. Kester and F.E. Muller-Karger, 1999. Remote sensing observation of winter phytoplankton blooms southwest of the Luzon Strait in the South China Sea. Marine Ecology Progress Series 191: 43-51. Tang, D.L., Di, B.P., Wei, G.F., Ni, I.H., Oh, I.S. and Wang, S.F.(2006a). Spatial, seasonal and species variations of harmful algal blooms in the South Yellow Sea and East China Sea. Hydrobiologia DOI: 10.1007/s10750-006-0108-1 Tang, D.L., Kawamura, H., Dien, T.V. and Lee, M.A. (2004a). Offshore phytoplankton biomass increase and its oceanographic causes in the South China Sea. Marine Ecology Progress Series 268: 31-41. Tang, D.L., Kawamura, H., Doan-Nhu, H. and Takahashi, W. (2004b). Remote sensing oceanography of a harmful algal bloom off the coast of southeastern Vietnam. J. of Geophysical Research (Ocean) 109: doi:10.1029/2003JC002045 (C3), C03014-C03014. Tang, D. L., Kester, D.R., Ni, I-H., Kawamura, H. and Hong, H.S. (2002). Upwelling in the Taiwan Strait during the summer monsoon detected by satellite and shipboard measurements. Remote Sensing of Environment 83(3): 457-471. Tang, D.L., Kawamura, H., Oh, I.S. and Baker, J. (2005). Satellite evidence of harmful algal blooms and related oceanographic features in the Bohai Sea during autumn 1998. Advances in Space Research 37: 681-689. Tang, D.L., H. Kawamura, P. Shi, W. Takahashi, L. Guan T. Shimada, F. Sakaida and O. Isoguchi, 2006b. Seasonal phytoplankton blooms associated with monsoonal influences and coastal environment in the sea areas either side of the Indochina Peninsula. Journal of Geophysical Research (Ocean) 111: doi: 10.1029/2005JG000050. Tang, D.L., Kester, D.R., Ni, I.H., Qi, Y.Z. and Kawamura, H. (2003a). In situ and satellite observation of a harmful algal bloom and water condition at the Pearl River estuary in late autumn 1998. Harmful Algae 2: 89-99. Tang, D.L. Kawamura, H. Lee, M.A. and Dien, T.V. (2003b). Seasonal and spatial distribution of chlorophyll a and water conditions in the Gulf of Tonkin, South China Sea. Remote Sensing of Environment 85(4): 475-483.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
30
DanLing Tang, SuFen Wang, Yasuwo Fukuyo et al.
Tang, K.W., Jakobsen, H.H. and Visser, A.W. (2001). Phaeocystis globosa (Prymnesi ophyceae) and the planktonic food web: Feeding,growth, and trophic interactions among grazers. Limnol. Oceanogr 46: 1860–1870. Taylor, F.J.R and Haigh, R. (1993). The ecology of fish-killing blooms of the chloromonad flagellate Heterosigma in the Strait of Georgia and adjacent waters. In: Smayda, T.J., Shimizu, Y.(eds.), Toxic phytoplankton blooms in the sea. Elsevier, Amsterdam: 705710. Usup, G., Pin, L.C. , Ahmad, A. and Teen, L.P. (2002). Alexandrium (Dinophyceae) species in Malaysian waters. Harmful Algae 1: 265-275. Usup, G and Azanza, R.V. (1998). Physiology and bloom dynamics of the tropical Dinoflagellate Pyrodinium bahamense. In: Physiological Ecology of Harmful Algal Blooms. Anderson, D.M., Cembella, A.D., Hallegraeff., G.M (DES). NATO ASI series Vol. G41/ Spring- Verlag, Berlin : 81-94. Vila, M., Camp, J., Graces, E., Maso, M. and Delgado, M. (2001). High resolution spatiotemporal detection of potentially harmful dinoflagellates in confines waters of the NW Mediterranean. Journal of Plankton Research 23: 497-514. Wang, H.K., Huang, L.M., Huang, X.P., Song, X.Y., Wang, H.J., Wu, N.J. and Li, C. (2003). A red tide caused by Gyrodinium instriatum and its environmental characters in Zhujiang River estuary. Journal of Tropical Oceanography 22: 55-62 (in Chinese, with English abstract). Wentz, F.J., Smith, D.K., Mears, C.A. and Gentemann, C.L. (2001). Advanced algorithms for QuikScat and SeaWinds/AMSR, International Geoscience and Remote Sensing Symposium, NASA, Sydney, New South Wales, Australia. Xia, B.C and Wu, R.H. (1996). Analysis on red tide outbreak in Dapeng bay of South China Sea through phytoplankton communities change. Acta Scientiarum Naturalium Universitatis Sunyatseni 35: 260-264 (in Chinese, with English abstract). Xie, S.P., Xie, Q., Wang, D.X. and Liu, W.T. (2003). Summer upwelling in the South China Sea and its role in regional climate variations. Journal of Geophysical Research (Ocean) 108(C8), 3261, doi: 10.1029/2003JC001867. Xu, N., Qi, Y.Z., Chen, J.F., Huang, W.J., Lu, S.H. and Wang, Y. (2003). Analysis on the cause of Phaeocystis globosa Scherffel red tide. Acta Scientiae Circumstantiae 23: 113118 (in Chinese, with English abstract). Yñiquez, A. Z., Azanza, R.V., Dalem B. and Siringan, F. (2000). Dinoflagellate cyst record in sediment cores from two sites in Manila Bay, Philippines, with different degrees of toxic red tide influence. The 9th International Conference on Harmful Algal Blooms, Tasmania, Australia. Yoshida, M., Ogata, T., Thuoc, C.V., Matsuoka, K., Fukuyo, Y., Hoi, N.C. and M. Kodama, M. (2000). The first finding of toxic dinoflagellate Alexandrium minutum in Vietnam. Fisheries Science 66: 177-179. Zhu, G.H., Ning, X.R., Cai, Y.M., Liu, Z.L. and Liu, Z.G. (2003). Studies on species composition and abundance distribution of phytoplankton in the South China Sea. Acta Oceanologica Sinica 25: 8-23 (in Chinese, with English abstract). Zhao, H and Tang, D.L. (2006). Annual variation of phytoplankton distributions in the South China Sea in relation with 1998 El Niño. Journal of Geophysical Research 112: C02017, doi:10.1029/2006JC003536
In-Situ and Remote Sensing Analysis of Harmful Algal Blooms…
31
QUESTION BANK Expand the Following 1. 2. 3. 4.
AVHRR CZCS ENSO HAB
5. 6. 7. 8.
JPL NACA NOAA PODAAC
9. SeaWiFS 10. SST
Short Answer Questions 1. 2. 3. 4.
What are algal blooms? What is the importance of aquatic ecosystems to human kind? What are various sources of stresses on the aquatic ecosystems? What spectral resolutions in remote sensing data can provide a good inventory to algal blooms in aquatic ecosystem? 5. Which are the important environmental variables to understand the status of aquatic pollution?
Long Answer Questions
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. Give an account of algal blooms in South China Sea. What are the various challenges to monitor and manage them? 2. Formulate a strategy to monitor the algal blooms in the ocean. Please provide a flow chart to carry out this activity.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 3
MONITORING SHALLOW LAKES IN THE PAMPAS Federico Dukatz, Rosana Ferrati, Claudia Marinelli, Rosana Cepeda and Graciela Canziani* Multidisciplinary Institute on Ecosystems and Sustainable Development, Universidad Nacional del Centro de la Provincia de Buenos Aires, 7000 Tandil, Argentina
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ABSTRACT This work proposes an analysis of the optical characteristics of shallow lakes based on information acquired by the Landsat TM and ETM+ sensors. This allows for a real time classification of water bodies and the inference of their trophic state through the various hydrological and climatic phases. The methods are based on the differential contribution from dissolved and particulate suspended matter to light dispersion and absorption phenomena in water as reflected in different wave-lengths captured by the sensors. Different methods and techniques that allow the classification of shallow lakes as a function of their turbidity, which defines different trophic chains, are proposed. The developed system is capable of handling the large volume of generated data in an efficient way, and allows for a temporal comparison of particular features observed in the satellite images, offering the automation of recurrent tasks. As an additional computational tool, a system for visualizing data under the form of leveled graphs is also presented. These represent the temporal variations in the area of water bodies as well as their dynamics. Thus, the visual recognition of behavioral patterns is facilitated and the identification of critical zones for further detailed study can be readily made. A statistical exploratory analysis that allows a categorization of the water bodies from field data taking into account the probability distribution of the variables being analyzed is presented. Finally, different methods for the classification of water bodies using only data retrieved from satellite images are presented. One of them is a technique based on spectral unmixing and another is the construction of an artificial neural network. Both computational tools, developed with the capability of dealing with high concentrations of algal biomass, yield good results which have been statistically validated.
Keywords: Classification, lakes, Landsat, Pampas, pollution *
Email: [email protected]
34
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
INTRODUCTION The Province of Buenos Aires is part of the Pampean region of Argentina. It is located east of the central zone, on the Atlantic coast and covers an area of 307,571 sq km. The Pampean plains comprise totally flat areas that alternate with soft hills covered by grasslands and a very low mountain relief over a small portion of territory in the South of the Province (Soriano, 1992). The Province of Buenos Aires has 13 million inhabitants (40 % of the country’s population), and is the main exporter of primary and secondary production in the nation. Historically, agricultural activities have been developed in the region, and in the last period, there is a clear tendency towards crops production and monoculture. Some 60 % of its surface is exploited under this system and there are no provincial regulations regarding land use. This results in an increase of activities that ensure the highest economic gains as a function of the current international market prices. As expected, the loss of soil quality and bad water management are the mid and long term consequences that affect not only the regional economic yield but also the sustainability of the system. The scarce morphogenetic power of the Pampean relief, with very low slopes and natural obstruction to drainage, together with the processes of wind deflation that occurred during the quaternary period, and the geological shortness of the present humid climate (Frenguelli,1956), have generated a high number of lentic water bodies that are characteristic of the region’s hydrology. In the Province of Buenos Aires alone, over 10,000 permanent or temporary shallow lakes can be found (Dangavs, 1982). On the one hand, they form a system of wetlands that covers more that one million hectares, distributed in both permanent and temporary water bodies called “lagunas”, their surface varying in the range of 10 to 40,000 ha., following the alternation of humid and dry periods characteristic of the Pampean climate. The Pampean “lagunas” are defined as shallow flatland lakes, polymictic, eutrophic and having highly variable water permanence time and salinity, and a high biomass in their biotic communities (Quirós et al., 2002). These wetlands do not exhibit thermal stratification. Normally they are subject to periods of floods and droughts that change their volume, and both water and sediments are disturbed by winds. This generates a constant removal of sediments that releases nutrients and reduces oxygen in the water column. These shallow lakes have highly variable salinity and turnover rates, are naturally eutrophic and suffer from environmental stress which increases even more their nutrient contents. As a result of the environmental conditions, the measured chlorophyll values are between 45 and 1400 mg/cu m, while for dissolved matter values lay between 440 and 2300 mg/l, and 500 and 2880 mg/l for total solids. This variability and wide range of values results in different turbidity states that modify the optic characteristics of the water and conditions the trophic type and the biogeochemical fluxes in the shallow lakes (Scheffer, 2004). Within this generality, the shallow lakes exhibit stages of clear waters and of turbid waters depending on the phytoplanktonic biomass, the presence of macrophytes, and the concentration of suspended inorganic solids. Shallow lakes in the clear water stage are more transparent and hence light penetration is deeper which allows the development of submerged macrophytes and produces a reduction in algal biomass. In these shallow lakes there is a dominance of fish related to the bafon and submerged vegetation, which facilitate the activities of visual predators. Shallow lakes in the turbid water stage are characterized by a biomass increase in all the pelagic communities due to a higher availability of nutrients that
Monitoring Shallow Lakes in the Pampas
35
are taken by phytoplankton to the detriment of macrophytes. Moreover, different trophic states can be observed depending on the relative concentrations of suspended organic and inorganic matter in the water column. In the so called “green” shallow lakes, water turbidity is mainly caused by the presence of abundant phytoplanktonic biomass, while the “brown” shallow lakes are characterized by a higher concentration of inorganic suspended matter arising from the sediment as an effect of the wind or from the surrounding landscape due to land use in the basin. These water bodies generate economic activities related to commercial and sports fishing, wild bird and fur mammal hunting, and also vegetation harvesting (reed for crafts). As sediment deposits, they provide mining resources. As landscape resource, they are excellent locations for tourism and recreational activities. On the other hand, the importance of the hydrological functions of these shallow lakes is impossible to miss. As natural dams, they regulate water excesses and deficits, particularly at times of humid and dry cycles. Also, they mitigate flows, maintain a base surface flow and permit the charge and discharge of groundwater (Dangavs, 1982). Besides being regulators of floods and droughts, they are refuges of high species diversity, and provide attractive spaces for recreation, tourism, fishing and other water sports. Given the functional, economic and esthetical importance of Pampean shallow lakes, their potential as indicators of the proper land use in the basin, their high numbers and their distribution over such an extended territory, and the knowledge on specific processing of satellite data developed by diverse authors (Brivio et al., 2001; Pozdnyakov et al., 2003; Nelson et al., 2003; Chipman et al., 2004), we propose the use of remote sensors for the classification of water bodies that will allow, as a second step, to suggest alternative strategies for an adequate management and generate real time control tools and application measures. The objective here is to present and validate, using field data, techniques for the determination of the trophic state of shallow lakes that permit to obtain the classification of lentic water bodies located in the Pampean region of Argentina.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Experimental Site The “lagunas” or shallow lakes selected for this study are located in the districts of Juarez (San Antonio, La Salada, El Chifle and La Barrancosa), Balcarce (La Brava) and Laprida (El Paraiso, Quilla Lauquen, Del Estado), on the water divide of the large hydrographic basins of the Province of Buenos Aires (Salado river, South streams, South-East streams, and South channels, (Ringuelet, 1962)). A map of the area is shown in Map 1.
Materials The field measurements carried on in each of the shallow lakes were Secchi disk depth, pH, electric conductivity, and water temperature. From the samples, the laboratory analyses provided Ca2+, Mg2+, Na+, K+, Cl-,NO3-, NO2-, NH4+, F-, ,SO43-, CO32- and HCO3-, ion concentration, phosphorus in water and soil, Chlorophyll-a, dissolved solids, volatile solids,
36
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Map 1. Location of the studied shallow lakes.
total solids, total hardness, and zooplankton density. The National Commission for Space Activities (Comision Nacional de Actividades Espaciales, CONAE) provided sixteen satellite images (224/86, 225/86, and 226/86), ten from LandSat 5 TM and six from LandSat 7 ETM+. The delay between the dates of the field trips and the satellite images was kept bellow five days (See Appendix I). Most of the image processing in this study was performed using ERDAS IMAGE 8.5. Additionally, some specific software systems were developed for particular tasks. For instance, the visualization tool based on leveled graphs, not found in commercial image processing software, is an original development.
Methods The field sampling was carried on between December 2004 and August 2006. The number of trips to each shallow lake was: eight (8) for La Brava; twelve (12) for Del Estado, Quilla Lauquen, and El Paraiso; sixteen (16) for La Salada, El Chifle, and San Antonio; and seventeen (17) for La Barrancosa.
Monitoring Shallow Lakes in the Pampas
37
Box 1. Landsat Satellite The Landsat series were constructed and put in orbit by USA for the observation with high resolution of our planet’s surface. Landsat satellites travel on a circular orbit at an altitude of 705 Km. They are equipped with specific instruments that can capture multispectral information, that is in several bands of different wavelengths (see Table 1 and Fig. 1). Bands 1 to 3 cover the visible spectrum while Bands 4 to 7 falls within the range of infrared radiation. A commonly used sub-division scheme of infrared wavelengths is: • • • •
•
Near-infrared (NIR): 0.75-1.4 µm in wavelength, defined by the water absorption. Short-wavelength infrared (SWIR): 1.4-3 µm, water absorption increases significantly at 1,450 nm. Mid-wavelength infrared (MWIR) also called intermediate infrared (IIR): 3-8 µm. Long-wavelength infrared (LWIR): 8–15 µm. This is the thermal imaging region, in which sensors can obtain a completely passive picture of the outside world based on thermal emissions only and requiring no external light or thermal source such as the sun, moon or infrared illuminator. Far infrared (FIR): 15-1,000 µm.
Hence Band 4 is in the NIR region, Bands 5 and 7 are in the SWIR region, while Band 6 is LWIR. NIR and SWIR are sometimes called reflected ifrared while MWIR and LWIR are sometimes referred to as thermal infrared. For most of the band, Landsate images have 30m×30m cells (pixels). The approximate size of each scene is 170Km×183Km or, in other terms, 6920×5960 pixels. Hence, a 100 ha shallow lake would be represented by 111 pixels.
Table 1. LandSat-7 ETM+ spectral bands
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Band 1 2 3 4 5 6 7
Spectrum Visible Blue Visible Green Visible Red Near infrared (NIR) Short-wavelength infrared (SWIR) Thermal infrared (LWIR) Short-wavelength infrared (SWIR)
Figure 1. LandSat 7 (ETM+) spectral bands
Wave length (µ) 0.450 – 0.515 0.525 - 0.605 0.630 - 0.690 0.775 – 0.900 1.550 – 1.750 10.400 - 12.500 2.090 – 2.350
Resolution (m) 30 30 30 30 30 60 30
38
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
Spectral Signatures Depending on the combination of bands it is possible to observe different characteristics of the surface. Radiation from the Sun is caused to bounce off differently according to the characteristics of the objects. If we plot how each wavelength reflects on any specific material, we obtain a graph that is proper to it and is called its spectral signature. This is what makes possible the identification of different objects, in our case soil, vegetation, water, etc. Through field and lab measurements it is possible to obtain reflection, absorption and transmission percentages on various surfaces. In Figure 2, the spectral signatures of different land covers are shown. Vegetation reflects better visible green than visible blue and red bands. This is why vegetation looks green. However, vegetation also reflects a high percentage of near infrared radiation (NIR) and this characteristic permits to distinguish it from, say, a green roof.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 2. Spectral signature of different land covers
When light penetrates a water body, the water itself contributes to absorption and dispersion. Additionally, in shallow lakes the presence of suspended organic and inorganic particles as well as dissolved substances has an effect on their optic properties. All colored matter produces a differential absorption, which means that only certain wavelengths are absorbed. Dissolved and suspended organic substances in lakes that have turbid sediments and abundant humic compounds yield dark chestnut-colored waters, usually acid and with poor bacterial activity (Neiff, 2004). In most shallow lakes, light is predominantly absorbed by particulate matter such as phytoplankton cells and suspended sediment particles. Given that this absorbent matter is colored, light absorption is not even for all wavelengths and some colors will penetrate more deeply than others. The apparent color of water is a very valuable characteristic for the functional differentiation of water bodies in a region. As an example, Sioli (1975) has classified rivers and water bodies in the Amazon following the criteria of “white waters”, “black waters”, and “clear waters” as expressions for their terrestrial environments. Suspended and dissolved matters contribute differentially to dispersion and absorption of light (Kirk, 1994). For example, suspended clay particles produce greater dispersion while dissolved organic substances produce absorption only. Phytoplanktons contribute to both dispersion and absorption (Fig. 3). Total absorption and dispersion
Monitoring Shallow Lakes in the Pampas
39
coefficients in any given wave length are a result of the sum of the individual contribution of water, phytoplankton, suspended sediments and organic compounds (Prieur and Sathyendranath, 1981).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 3. Behavior of solar radiation in water
Radiometric Corrections Several different factors have an influence on satellite images, such as distance from Sun to Earth, and solar incidence angle among others (Fig. 4). Hence it is necessary to perform corrections on the images before being able to make inferences about the characteristics of shallow lakes. All sixteen images were radiometrically corrected using the conversion to radiance methods for Landsat 5 and 7 products proposed by USGS (NASA/USGS, 1998; USGS, 2006). Afterwards, an atmospheric correction (Rayleigh method) on the radiance was performed in order to obtain surface reflectance values free from molecular scattering that reach the sensors as a result of the interaction between electromagnetic radiation and the molecular components of the atmosphere. Masking After the images have been corrected and geo-referenced, it is necessary to separate the water bodies from the rest of the image so as to avoid the influence of neighboring pixels. One way to do it is to draw the contours manually. However, it is more convenient to do it automatically, taking advantage of the spectral characteristics of water. This technique for identifying objects in the image is called masking. There are different types of masks that can be applied, and it is very important to precisely define the one being used because the choice will have different implications in the analysis of the information. As seen in Figure 2, water reflectance values are very low for Band 4 as compared to other type of surfaces. The most
40
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 4. Factors that influence satellite images.
common masking technique for separating water in an image is that of threshold masking. The technique is based on the elimination of all pixels for which Band 4 takes values above a certain chosen threshold. This type of mask is frequently used because of its simplicity and precision. However, if the study area encompasses a large extension, the technique may be less precise because other landscape elements can exhibit a low reflectance value in Band 4, for instance the shadows in a rugged landscape. For this reason, the mask used here is based on the Tasseled Cap technique. The Tasseled Cap (TC) transformation was originally designed for Landsat TM images. The TC indices relate six TM bands (1-5 & 7) to measures of vegetation (greenness), soil (brightness) and the interrelationship of soil and canopy moisture (wetness). Each index corresponds to a linear transformation of six TM bands using a set of empirically derived coefficients. Afterwards, the information present in the 6 original bands is compressed into 3 TC transform bands. The Tasseled Cap transformation is very appropriate for regional applications where atmospheric correction is not possible (Huang et al., 2002). The concept behind the Tasseled Cap transformation is a very useful tool for the understanding of spectral data in few bands that are associated to physical characteristics of the scene (Crist and Cicone, 1984). It was originally conceived for understanding important phenomena occurring in the spectral space during the development of crops (Kauth and Thomas, 1976). For the mask here used, the bands corresponding to soil Wetness and Brightness were combined. Pixels for which the value in the Wetness band is highest and the value in the Brightness band is lowest are considered part of a water body. In such a way, most of the characteristics that could lead to confusion with water bodies using other methods can be eliminated. After the water bodies have been separated and individualized, it is possible to determine the spectral signature of each shallow lake. Then, comparing to field data, signatures are associated to particular characteristics. The spectral signature of any given shallow lake is obtained through averaging the spectral signature of every pixel associated to
Monitoring Shallow Lakes in the Pampas
41
it. Once the spectral signature is obtained, it can be used to determine the trophic characteristics of the water body through the use of different classification or categorization techniques, such as artificial neural networks or spectral unmixing, as will be seen later on.
Temporal and Spatial Evolution In Figure 5, two masked images of the same region at different dates are shown. It is clearly seen that the water covered area in the first image is larger than in the second. These images belong to a portion of a region that includes ‘El Paraiso’ shallow lake in Laprida district, in the Province of Buenos Aires. The coordinates of the lake are 37º 34' 03.8'' S and 60º 48' 03.0'' W. In order to capture the shallow lake, its basin and the area of influence, a rectangular region of some 500,000 hectares was selected. Table 2 shows the temporal evolution of the water covered area in this region. In order to obtain a sharper result, water bodies smaller than 10 hectares have not been taken into account. The table shows the image area, total water covered area and area of water bodies larger than 10 hectares as well as the respective percentage. It is interesting to note how variable the water covered area can be and how rapid the changes can be. Another example, an interest area that includes ‘La Barrancosa’ shallow lake in the Benito Juarez district, Province of Buenos Aires, has been selected. The coordinates of the shallow lake are 37º 21' S and 60º 7' W. In Figure 6, a sequence of masked images shows changes recorded from may 2000 until October 2005. Dates and type of sensor for each image are given in Table 3.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
09/09/2004
07/05/2005
Figure 5. Examples of water masks in the same region at different dates.
Recurrence Maps Another immediate application of this type of methodology is the generation of recurrence maps or, in other words, theme maps that show the probability of occurrence of a given phenomenon in a particular region. A region of 10,773 hectares was selected so as to comprise the shallow lake, its basin and the influence zone (Fig. 7). This type of analysis is highly important for identifying non permanent water bodies as well as flood zones and their frequency of flooding for risk assessment.
42
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al. Table 2. Water covered area in the given region Date
Area covered by Water Bodies >10 Ha
Total Covered Area
Total Image Area
2004/09/09
13742,28
2,74%
25791,57
5,15%
501185,88
2004/09/25
33974,10
6,78%
55202,40
11,01%
501185,88
2005/01/15
69698,70
13,91%
92875,50
18,53%
501185,88
2005/03/04
66561,30
13,28%
94870,80
18,93%
501185,88
2005/05/07
11019,33
2,20%
15496,02
3,09%
501185,88
2005/06/08
58282,20
11,63%
85416,30
17,04%
501185,88
2005/10/14
5790,69
1,16%
10150,83
2,03%
501185,88
2006/02/19
35163,90
7,02%
51919,20
10,36%
501185,88
2006/04/20
44085,60
8,80%
73654,20
14,70%
501185,88
Table 3. Dates and types of sensor corresponding to Figure 6
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Fecha 10/05/2000 02/11/2000 27/04/2001 14/04/2002 07/10/2002 16/05/2005 19/07/2005 07/10/2005 23/10/2005
Satellite - Sensor LandSat 7 - ETM LandSat 7 - ETM LandSat 7 - ETM LandSat 7 - ETM LandSat 7 - ETM LandSat 5 - TM LandSat 5 - TM LandSat 5 - TM LandSat 5 - TM
Visualization Through Leveled Graphs One useful way to represent the temporal and spatial evolution of shallow lakes is through leveled graphs. Given the fact that data can be too numerous for a direct interpretation, they can be represented with nodes (circles) and links (connecting lines). Each node represents a water body at a given date and can be linked to the same water body, or other water bodies, at different dates in a sequential manner. In the shown visualization (Fig. 8), the size of each node is proportional to the area of the water body it represents. An identification number can be assigned to each node, as well as any relevant data such as its area in terms of pixels. One pixel represents 900 sq m approximately. Also, data can be assigned to each link. One application could be to keep record of the area of intersection of water-covered surfaces occupied by a given shallow lake at two consecutive times expressed in terms of pixels and their respective percentages (Dukatz and Ferrati, 2007).
Monitoring Shallow Lakes in the Pampas
Masked Images
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
La Barrancosa
Original Image (2000/05/10)
Figure 6. Sequence of images showing changes in area of “La Barrancosa” (right) and neighboring (left) shallow lakes.
43
44
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
0 % - No Water 1-25 % 26-50 % 51-75 % 76-99 % 100 % Permanent Water
Figure 7. Recurrence Map of floods in the region surrounding “La Barrancosa” shallow lake
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The graphs shown in Figures 8.a and 8.b have been extracted from the small interest area shown in Figure 6. A minimal threshold area for taking into account a water body or flooded land in the graph was set to 10 ha. The first graph (Fig. 8.a) is a visualization of the variation in the area of ‘La Barrancosa’ shallow lake, which is a permanent water body and does not exhibit significant changes in its evolution. The second graph (Fig. 8.b) is a visualization of a semi-permanent water body located SW of ‘La Barrancosa’. Dotted lines indicate the absence of a significant water cover during the corresponding period (second image). Some time later (third image), the shallow lake reappears as two separate water bodies that become one at a later date, its area varying significantly at successive dates. The color assigned to each node may also convey meaningful information, such as trophic state, presence or absence of contaminants, presence or absence of any given species, abundance of fish, etc. Thus, the particular dynamics of individual water bodies can be easily visualized and relevant data included in a very simple representation scheme.
a) Figure 8. Continued on next page.
Monitoring Shallow Lakes in the Pampas
45
(b) Figure 8. Leveled graphs corresponding to (a) “La Barrancosa” shallow lake, and (b) a neighboring water body located SW of it, as seen in Figure 6
Classification of Water Bodies In order to perform a classification of the shallow lakes from field data following their turbidity, the values of chlorophyll and solids in the samples and the Secchi disk depth were used. The classification was done by defining the quantiles probability distribution (Rohatgi, 1976). Four categories were obtained: • • •
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
•
Class 1 formed by clear water shallow lakes. Class 2, grouping categories 2B, 2G and 2BG, representing intermediate turbidity values and having brown, green, and brown-green coloration respectively. Class 3, comprising shallow lakes in category 3G, exhibiting high values of chlorophyll. Class 4, formed by shallow lakes in category 3BG, exhibiting the highest values of chlorophyll and suspended solids.
The National Commission for Space Activities (CONAE) provided sixteen Landsat images covering the study area and which dates are coincident with the time of the field sampling. The details about the shallow lakes, the images, and the field work dates are reported in Appendix I. Once the water covered surfaces were isolated using the appropriate mask, their individual spectral signatures were retrieved, as shown in Figure 9. A total sample of thirty three spectral signatures was obtained. It is worth noting the patterns of ‘extreme’ spectral signatures. The spectral signature drawn in the thick solid blue line corresponds to ‘La Brava’ shallow lake, which is considered a clear water body and exhibits minimum values for chlorophyll and suspended solids as well as maximum depth for Secchi disk (Class 1). The spectral signature drawn in the thick solid red line corresponds to ‘Del Estado’ shallow lake, a turbid water body, in extreme condition of maximum concentration of chlorophyll and suspended solids and minimum depth values for Secchi disk (Class 4). The
46
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
wider range of differences in the samples appears in Band 2 (visible spectrum) and Band 4 (near infrared). 0.1
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0 1
2
3
4
Bands
5
6
LBR-02-02-05 LPE-02-02-05 DE-04-03-05 EP-04-03-05 QL-04-03-05 LB-04-03-05 LB-13-03-05 SA-13-03-05 LB-22-04-05 LS-22-04-05 ECH-22-04-05 SA-22-04-05 DE-080605 EP-080605 QL-08-06-05 DE-07-05-05 EP-07-05-05 QL-07-05-05 DE-12-09-05 EP 12-09-05 QL-12-09-05 LB-12-09-05 SA-13-09-05 LS-13-09-05 ECH-13-09-005 SA-15-10-05 LS-15-10-05 ECH-15-10-05 QL-14-10-05 EP-14-10-05 DE-14-10-05 EP-09-12-05 DE-09-12-05 DE-26-01-06 EP-26-01-06 QL-26-01-06 LB-26-01-06 LB 11-07-05 ECH 11-07-05 LS 11-07-05 SA-11-07-05 QL-11-07-05
Figure 9. Spectral signatures obtained from LandSat images for La Barrancosa, Quilla Lauquen, San Antonio, La Salada, El Chifle, Del Estado, La Brava, and El Paraíso shallow lakes throughout the period of sampling as reported in Appendix I.
Spectral Unmixing Given the high number of shallow lakes in the Pampas region of Argentina, it is very useful to have an instrument that allows determining their turbidity from data retrieved from remote sensors.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Box 2 Spectral Unmixing It is assumed that a píxel in a remotely sensed image represents a distinct ground cover material and can be uniquely assigned to a ground cover class. However, it is known that actually the spatial resolution and the heterogeneous nature of target surfaces make necessary to investigate the subpixel information when a píxel’s corresponding spectrum is composed of a mixture of multiple materials. Spectral mixture can be analyzed with linear spectral mixture analysis, which models each spectrum in a spectral dataset as a linear combination of a finite number of spectrally distinct signatures (end-members), with coefficients or fractional abundances between 0 and 1 and adding up to one (Adams et al., 1985; Smith et al., 1990; Bateson and Curtiss, 1996). Each constituent of ground cover materials can thus be defined as an end-member (its spectral signature), and it is clear that the election of the most representative end-members in the mixture is key to the desired result. Each pixel in the image is a point or vector in an ndimensional space, n depending on the number of available bands. When the bands of
Monitoring Shallow Lakes in the Pampas
47
LandSat 5 or LandSat 7 satelites are being used, the vector will have six components: P = (B1, B2, B3, B4, B5, B7). Similarity criteria for classifying vectors are difficult to find when working in dimensions higher than two. The objective of the spectral unmixing method is to perform an analysis at the subpixel scale, focusing on determining the percentage of each constituent of the ground cover. In order to accomplish this, its is necessary to select the end-members present in the image and set the reflectance value of each pixel as
r =M. f where r is the reflectance in the pixel, M is a matrix which columns are the signatures of the selected end-members, and f is a vector representing the fraction of each end-member corresponding to the analyzed pixel. In this way, the reflectance observed in a given pixel can be represented as a linear combination of the selected end-members. Hence, it is most important to know how end-members can be identified and selected. If M is a square matrix and its columns are linearly independent, then its inverse matrix exist and there will be a unique solution. Otherwise, the matrix is not invertible and a pseudoinverse shall be computed:
M * = ( MM T ) −1 M T which is a least squares fitted solution. It is usually impossible to have more than three or four linearly independent end-member vectors, so that pseudo-inverse matrices are normally used. A vector of percentages
f '= M *. r can be used for reconstructing the image (pixel by pixel) and obtain from it the reflectance values
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
r' = M . f ' In such a way, the error of the method can be estimated by the quadratic difference
e = (r '−r ) 2 . This classifier has been widely used for the determination of land cover and is currently being applied to water bodies (Svab et al., 2005; Tyler et al., 2006).
For this study, after having obtained the spectral signatures for each shallow lake (Fig. 10), the two extreme signatures corresponding to clear and to turbid shallow lakes were selected as end-members. After normalization, the scalar product of both vectors was computed in order to determine their linear independence. Taking into account that the end-
48
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
members are not truly orthogonal, vectors for which the scalar product is bellow 0.3 are accepted. The linear independence of the vector base for different groups of bands (all bands, bands 1, 2 and 4 only, bands 2 and 4 only) that adequately represent the observed values and fulfill the condition of the established minimum value of the scalar product was analyzed. In this case, the best base that can be found is that formed by bands 2 and 4, which is very reasonable as can be observed in Figure 10. Selecting both Band 2 and 4 as input data, a spectral unmixing method was applied. Given the spectral signatures characteristic of the objects to be classified, this method finds the proportion of each extreme class in the pixel being analyzed. The mathematical procedure takes as reference the two extreme cases of clear and turbid shallow lakes and then analyzes in terms of percentage of turbidity all the other shallow lakes under study. Then, a quantiles probability distribution is performed on these results. This allows obtaining four categories, numbered from 1 to 4 in increasing degree of turbidity. In this way the shallow lakes are classified following the computed percentage of turbidity using spectral unmixing. However, the results need to be validated. Thus, it becomes necessary to compare the results of our classification from image data (Table 4, column 4) to classification obtained from field data (Table 4, column 5). Table 4. Classification of shallow lakes following their turbidity from satellite image data (column 4) and from field data (column 5), indicating their state as Clear (C), Green (G) or, Brown (B), or Green-brown (GB) Shallow Lake
Del Estado
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
El Chifle
El Paraíso
La Barrancosa
Date
2005/03/04 2005/05/07 2005/06/08 2005/09/12 2005/10/14 2005/04/22 2005/07/11 2005/09/13 2005/10/15 2005/03/04 2005/05/07 2005/06/08 2005/09/12 2005/10/14 2005/03/04 2005/03/13 2005/04/22 2005/07/11 2005/09/12
% Turbidity from bands 2 and 4 0.792531095 0.877031962 0.924875654 0.832126721 0.866491819 0.58918169 0.628806266 0.637835148 0.422368775 0.739619393 0.754311456 0.781521881 0.717096949 0.743081415 0.61592054 0.559267154 0.550347904 0.545138169 0.570154341
Classification Class from image data
Class from field data
4 4 4 4 4 2 3 3 1 3 3 4 3 3 2 2 2 2 2
3 (3G) 4 (3GB) 4 (3GB) 4 (3GB) 4 (3GB) 2 (2GB) 3 (2GB) 3 (3G) 2 (2GB) 3 (3G) 3 (3G) 3 (3G) 3 (3G) 3 (3G) 2 (2B) 2 (2GB) 2 (2B) 2 (2GB) 2 (2GB)
Monitoring Shallow Lakes in the Pampas Shallow Lake
La Barrancosa La Brava La Salada
Quilla Lauquen San Antonio
Date % Turbidity from bands 2 and 4 0.521829991 0.44903381 0.324523594 0.50102486 0.481098697 0.512106876 0.414638768 0.541566687 0.486154099 0.549101813 0.501318055 0.517938528 0.482929504 0.324859024
2005/10/14 2005/01/26 2005/02/02 2005/04/22 2005/07/11 2005/09/13 2005/10/15 2005/06/08 2005/10/14 2005/03/13 2005/04/22 2005/07/11 2005/09/13 2005/10/15
49
Classification Class from image data
Class from field data
2 1 1 2 2 2 1 2 2 2 2 2 2 1
2 (2B) 2 (2G) 1 (C) 2 (2G) 2 (2G) 2 (2G) 2 (2G) 2 (2G) 1 (C) 2 (2B) 2 (2B) 2 (2GB) 2 (2B) 2 (2GB)
For the analysis of the possible concordance between the two categorizations, the Kappa index is computed using the contingency table (Table 5) (Landis and Koch, 1977). This table orders each shallow lake, in each date, following both classifications, the one obtained from satellite data and the one obtained from field data, counting in the diagonal the number of cases in which there is coincidence. It becomes necessary to establish a measure of concordance that takes into account the conditional and marginal distributions in Table 5. For the analysis of the possible concordance between the two categorizations, the Kappa index is computed using the contingency table (Table 5) (Landis and Koch, 1977). This table orders each shallow lake, in each date, following both classifications, the one obtained from satellite data and the one obtained from field data, counting in the diagonal the number of cases in which there is coincidence. It becomes necessary to establish a measure of concordance that takes into account the conditional and marginal distributions in Table 5.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 5. Double classification of shallow lakes: contingency table Categories
Categories from bands 2 and 4
from field data
1
2
3
4
Total
1
1
1
0
0
2
2
4
15
1
0
20
3
0
0
5
2
7
4
0
0
0
4
4
Total
5
16
6
6
33
50
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al. Box 3. Kappa Index
Given a contingency table XIxJ with I category rows and J category columns, (Agresti, 1990) where
π ij =
nij n
represents the joint probability corresponding to cell ij and
πi =
∑n
ij
j
n
its corresponding marginal probability i, the Kappa index is defined by:
κ=
∑π i
ii
− ∑π i ∑π i i
1 − ∑ π iπ i
i
i
When interpreting the value obtained for κ it is useful to have in mind the following scale: (0.20 Poor; 0.21 – 0.40 Weak; 0.41 – 0.60 Moderate; 0.61 – 0.80 Good; 0.81 – 1.00 Very Good)
In this case, the index is 0.619 which indicates a good agreement between the classifications obtained from bands 2 and 4 of the images and that from the field data.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Box 5. Types of solid matter identified in water Total dissolved solids - includes all solids present in a water sample filtered through a 0.45 mm filter. Their amount is determined by evaporating a known volume of the filtrate sample in a 105 ºC oven. Total suspended solids - includes all solids present in a sample that remain on a glass fiber filter. Their amount is determined by filtering a known volume of sample and placing the filter in a 105 ºC oven for 24 hours to evaporate the water. Fixed solids - solids that remain after firing a sample in a 550ºC muffle furnace. This can be performed on total, dissolved, or suspended samples to determine total fixed solids, fixed dissolved solids, or fixed suspended solids respectively. Volatile solids - solids that are removed by firing a sample in a 550ºC muffle furnace. This can be performed on total, dissolved, or suspended samples to determine total volatile solids, volatile dissolved solids, or volatile suspended solids respectively.
Monitoring Shallow Lakes in the Pampas
51
A Principal Components Analysis (PCA) permits exploring the degree of association between variables through the analysis of the structure of covariance. Moreover, it permits to reduce the number of dimensions of the problem (Johnson et al., 1992). If there is a suspicion that Band 2 and Band 4 are describing the same field variables chlorophyll, total solids, dissolved solids, fixed solids, Secchi depth, then the PCA should show it. In Table 6, the correlations between measured variables and the two top Principal Components are shown. It can be seen that the largest variation in the positive sense can be defined for Band 4, chlorophyll, total solids, fixed solids, and filtered solids, while in the negative sense are Band 2 and Secchi depth. The results obtained from the concordance analysis as well as those from the PCA permit testing the hypothesis that Bands 2 and 4 could Table 6. Results obtained from Principal Components Analysis (PCA) % reconstruction of each Variable
Comp1
Comp2
variable in the plane
BAND 2
- 0.69970
0.18315
52.31
BAND 4
0.73597
- 0.16958
57.04
Chlorophyll.
0.89717
- 0.22452
85.53
Total Solids
0.93425
0.20324
91.41
Fixed Solids
0.89679
0.21089
84.86
Filtered Solids
0.77242
0.48022
82.72
SECCHI Depth
- 0.79546
0.42759
81.55
explain the variation of observations regarding chlorophyll, solids and Secchi depth. This is a good motivation for performing a Canonical Correlation Analysis, designed for studying the association between two groups of variables (Mardia et al., 1979). The statistical significance of this Canonical Correlation Analysis can be seen in Table 7.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 7. Canonical Correlation Analysis and Redundancy of the two groups of variables (satellite image data and field data) Band
Field
Variables
Variables
Amount of variables
2
5
Extracted Variance
100%
63.54%
Total Redundancy
63.93%
51.16%
52
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al. Box 6. Redundancy
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The redundancy or ‘average amount’ of variation of one group of variables explained by the other group is important because in both cases it is above 50 %. If the group formed by the data from the two bands is taken as response, its variation is explained in 63.93 % by the group of field data (Ter Braak, 1994). If the group formed by field data is taken as response, its variations is explained in 51.16 % by the information recorded by Bands 2 and 4. The statistical techniques here used corroborate the proposed hypothesis with a significance of 0.01, which allows affirming that the set of Bands 2 and 4 significantly explain the presence of chlorophyll and suspended solids in the dates of the satellite images. This permits to obtain real time information on the trophic state of the shallow lakes.
Artificial Neural Network Another computational tool constructed for the automatic analysis of data retrieved from remote sensors is an Artificial Neural Network model (ANN). An ANN is an informationprocessing system that has certain performance characteristics in common with biological neural networks. Some computational models are intended to mimic the way in which the human brain learns from incorporating new information. These models are included in the field of computational intelligence or artificial intelligence. The former mimic the physical processes that occur in the brain while learning. Some examples of these are artificial neural networks, fuzzy logic, and genetic algorithms. The latter, artificial intelligence models, mimic the logical learning process of human beings. Some examples of these are expert systems, agents, and case-based reasoning. Hence, an Artificial Neural Network model imitates the physical process of learning in the human brain. The model is formed by artificial neurons that emulate biological neurons and the synaptic connections among them, regulating them through the process of problem solving. They are appropriate for dealing with a large set of variables and their non linearity is convenient for analyzing complex systems. Once the system of neurons has been trained, the network allows the processing of imprecise information, the generalization of known responses to new situations, and the prediction of outcomes. Mimicking the way in which any person can learn, the network needs to be “trained” with a sufficiently large number of examples in order to be able to make the appropriate inferences. Hence, it is given groups of input data together with the expected output data (Fig. 10). The links with the neurons located in the so called hidden neuron layer then take different weighs and are educated depending on the required output, thus modeling complex relationships among variables. The system requires ‘feedforward’ and ‘backpropagation’ processes to allow the network to get trained. The visualization of this stage is accomplished through error analysis. If the error becomes smaller and asymptotic, the network will be ready to receive new input data and predict output.
Monitoring Shallow Lakes in the Pampas
53
1 Band 2
2
Band 4
3 4 Input data
Hidden neuron layer
Output data / Prediction
Figure 10. Artificial Neural Network model used for data analysis
Box 7 Artificial Neural Network A Neural Network is characterized by the pattern of connections between the neurons (called its architecture), its method of determining the weights on the connections (called its training), and its activation function. The architecture of an ANN consists of nonlinear elements (neurons, nodes, or units) forming layers. Each neuron is connected to other neurons by means of directed links, each with an associated weight. Each neuron computes the weighted addition of its input values and generates an activation function which is a function of the inputs it has received. This activation function typically behaves as a step function. This means that when the variable takes a value bellow a given threshold, the signal is zero or negative, while a value above the threshold produces a signal close to one. The activation function may or may not be continuous; however differentiable functions are usually used because they allow the minimization of errors. A commonly used activation function is the logistic sigmoid function
y (x ) =
1 1 + exp(− k .x )
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
which has its range in the interval [0, 1] and is used to normalize the response of output nodes (neurons). This function behaves as a floodgate that can be open (1) or closed (0). Since it is a continuous function, it is possible that it may also be partially open. For convenience, parameter k is set to be 1. In such a case, the derivative function can be expressed as
dy = y (1 − y ) dx The models that use a sigmoid function as transference function (or activation function), exhibit good learning characteristics, as seen in their training relative to an error taken as evaluation criterion, but they may require more computational time to accomplish their training.
54
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Usually, it is convenient to visualize neurons as arranged in layers and behaving in the same manner. The two important factors that determine a neuron’s behavior are its activation function and the pattern of weights associated to its connections. The neurons that are located in a layer usually share the same activation function and the same connection pattern. This connection pattern means that every neuron in a given layer is connected to every neuron in the next layer. A multilayer network is an ANN with one or more layers of neurons (called hidden layers) between the input nodes and the output nodes. Multilayer networks can solve more complicated problems than single layer networks, but it may be more difficult to train them. The method for updating the weights (training) is an important and distinctive characteristic of neural networks. There are two types of training methods: supervised and unsupervised. The ANN is fed with an input data set and the corresponding output. The model “learns” from the example cases and assigns weight coefficients to input variables, thus modeling complex relationships among variables. When new inputs are given, the trained network is able to predict the output. The Perceptron learning rule provides the convergence to the correct weights through an iterative process, given certain conditions. The process consists of three phases called ‘feedforward’, error computation, and ‘backpropagation’. First the input is given and the calculations through the network yield the output classification, then the error in the classification is computed, and finally the weights in the network are updated. The parameters of the backpropagation algorithm are the learning rate and the number of epochs used in the training of the network. The learning rate represents the speed at which the network learns, while the number of epochs indicates how many times the examples are processed through the network during training. Neither a high learning rate nor a large number of epochs imply a better learning. A large number of epochs can lead to an overtraining of the network, i.e.: the network learns only for the examples and is unable to generalize. A high learning rate can lead to oscillations, i.e.: the error may oscillate around the minimum without converging.
The networks used in this study are Multilayer Perceptron Artificial Neural Networks. (Rosenblatt, 1962; Minsky and Papert, 1969). In each case, the training of the proposed network was performed with a Back Propagation algorithm which is a supervised learning procedure (Rumelhart et al., 1986a, 1986b; McClelland and Rumelhart, 1988). It utilizes a method of descent on the gradient for minimizing the global quadratic error of the output calculated by the network. For the error computation, the Maximum norm has been used. It is defined by the absolute value of the largest component of the difference between the network output vector and the expected result vector: x ∞ = max xk 1≤ k ≤ n
Classification of Shallow Lakes Using An ANN For the purpose of classifying the shallow lakes into the three proposed states, -clear, green, or brown, -an ANN was constructed. Following the classification of shallow lakes using remote sensors data, this ANN was educated using as input data Bands 2 and 4 of Landsat 5 and Landsat 7, and the classes obtained from field data as output values
Monitoring Shallow Lakes in the Pampas
55
(Appendix). The network was tested with different amounts of learning stages or epochs, different learning rates, and different numbers of neurons in the hidden layer. The best network architecture was obtained with three neurons in the hidden layer (Fig. 10). The best results were obtained with 1,000,000 epochs and a learning rate of 0.00001. The model runs yield the approximation shown in Figure 11, where the plots show real data and the output of the ANN model using 80% of data for training and 20% for predicting. The results were evaluated by comparing the classification output yielded by the neural network to the classification obtained from field data. As seen in Figure 11, the plots showing both real data and the output of the ANN model coincide. In all cases the learning error becomes null in less than 20,000 epochs. Hence the prediction error is zero, which means that the classification method is accurate with respect to the proposed whole numbers classification values (1, 2, 3, and 4).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 11. Classification using field data (black line): the ANN model output (gray line) obtained in the training and prediction stage using 80% of data for training and 20% data for predicting. The vertical line separates both sets. X-axis: shallow lakes, Y-axis: class.
Figure 12. Learning error for the classification ANN. The ANN yields a precise classification hence the predicting error becomes null. X-axis: epochs, Y-axis: maximum absolute error.
56
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
Concentration Of Total Solids And Chlorophyll Using An ANN Utilizing the computational tools already developed, it is possible to retrieve additional information from the satellite images. The objective is to determine the concentration of total solid matter and of chlorophyll, with an acceptable error margin, from satellite data. Thus two ANN were constructed and fed with the reflectance values retrieved from Bands 2 and 4 of Landsat 5 and Landsat 7 satellites. The networks were trained using 35 input data, 80% of which were used for learning and 20% for predicting. The Total Solids model ran over 1,000,000 epochs with a learning rate of 0.000001. The number of hidden neurons varied until a best configuration was attained. The results were plotted together with the measured total solids values obtained in the field, as shown in Figure 13.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 13. Concentration of total solids (×10-3 mg/l) using field data (black line) and the ANN model output (gray line) obtained in the training and prediction stage using 80% of data for training and 20% data for predicting. The vertical line separates both sets. X-axis: shallow lakes, Y-axis: total solids concentration [10−4 mg/l].
a)
b)
Figure 14. Learning (a) and predicting (b) errors for the concentration of total solids. X-axis: epochs, Yaxis: maximum absolute error.
Monitoring Shallow Lakes in the Pampas
57
Learning and predicting errors expressed as the maximum module of the absolute differences and divided by 10,000 (in this case) behave asymptotically (Fig. 14(a) and (b)). The learning error exhibits a decreasing tendency and yielded the results for Total Solids shown in Table 8, column 4. The predicting error exhibits a local minimum at 800,000 epochs, value at which the predictions were computed.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 8. Concentrations of Chlorophyll-a and Total Solids, measured in the field and computed from the satellite images, at given shallow lakes and dates.
Date 2005/02/02 2005/03/04 2005/03/04 2005/03/04 2005/03/13 2005/03/13 2005/04/22 2005/04/22 2005/04/22 2005/04/22 2005/05/07 2005/05/07 2005/06/08 2005/06/08 2005/06/08 2005/07/11 2005/07/11 2005/07/11 2005/07/11 2005/09/12 2005/09/12 2005/09/12 2005/09/13 2005/09/13 2005/09/13 2005/10/14 2005/10/14 2005/10/14 2005/10/14 2005/10/15 2005/10/15 2005/10/15 2006/01/26
Shallow Lake La Brava La Barrancosa El Paraíso Del Estado San Antonio La Barrancosa La Salada San Antonio La Barrancosa el Chifle El Paraíso Del Estado Quilla Lauquen El Paraíso Del Estado La Salada San Antonio La Barrancosa el Chifle La Barrancosa El Paraíso Del Estado La Salada San Antonio El Chifle Quilla Lauquen La Barrancosa El Paraíso Del Estado La Salada San Antonio El Chifle La Barrancosa
TS measured 507.0 1474.0 1498.0 1788.0 1296.0 1565.0 1184.0 1376.0 1636.0 1756.0 1706.0 2522.0 806.0 1686.0 2347.0 1251.0 1397.0 1629.0 1942.0 1624.0 1700.0 2653.0 1284.0 1418.0 1736.0 812.0 1659.0 1830.0 2881.0 1246.0 1396.0 1953.0 1327.0
TS computed 557.7 1562.9 1380.3 2040.4 1464.2 1482.0 1209.7 1275.0 1360.6 1354.4 1836.1 2259.1 1481.6 1887.3 2478.6 1508.0 1490.8 1538.5 1598.4 1341.7 1728.6 2505.0 1490.7 1459.8 1574.6 1040.2 1477.2 1774.3 2759.6 1449.2 1398.4 1411.6 1235.8
Chlorophyll measured 45.3 89.1 340.4 325.3 50.4 155.4 133.9 65.1 95.7 209.3 384.5 600.8 122.1 369.0 667.0 260.2 179.8 153.5 120.9 115.2 371.9 533.3 112.0 94.0 249.9 76.3 93.2 303.5 671.2 130.6 117.5 222.2 71.3
Chlorophyll computed 42.0 183.1 322.3 422.8 136.9 143.4 101.5 104.6 130.2 151.0 340.3 577.4 135.3 379.8 697.0 123.1 146.6 148.7 201.2 139.4 284.3 544.4 125.0 111.2 198.8 89.6 126.8 319.4 625.2 91.6 68.9 89.4 85.6
58
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
The second network was constructed to simulate and predict chlorophyll concentrations. Figure 15 shows the plot of chlorophyll concentration values measured in the field and computed with the help of the ANN. Learning and predicting errors are plotted and shown in Figure 16 (a) and (b). It can be seen that the predicting errors exhibits a tendency to increase with the number of epochs, which indicates that the network is not properly predicting the data. The results for computed chlorophyll concentrations are shown in Table 8, column 6, together with the values measured in the field, column 5.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 15. Concentration of chlorophyll-a (×10-3 mg/l) using field data (black line) and the ANN model output (gray line) obtained in the training and prediction stage using 80% of data for training and 20% data for predicting. The vertical line separates both sets. X-axis: shallow lakes, Y-axis: chlorophyll concentration [10−3 mg/m3].
a)
b)
Figure 16. Learning (a) and predicting (b) errors for the concentration of chlorophyll-a. X-axis: epochs, Y-axis: maximum absolute error.
Monitoring Shallow Lakes in the Pampas
59
Validation Of Results In order to validate the results, a dispersion diagram was done. It allows analyzing the behavior of measured TS and computed TS from available data, excluding the values that distort the sample. Figure 17 shows a positive tendency and a strong association between the two sets of variables. A simple linear regression analysis, taking as response variable the measured TS values, was performed with the purpose of quantifying the tendency. The regression analysis yields a determination coefficient R² = 0.9, and a fitted regression line Measured TS = 1.04×Computed TS (p < 0.0001).
TSmeasured
2999.70
2346.85
1694
1041.15
388.30 447.61 1053.13 1658.65 2264.17 2869.69
TS calculated
Figure 17. Dispersion diagram relating measured and calculated total solids
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The graph, confidence bands, and prediction are shown in Figure 18. The residuals analysis indicates independence, normality, and homogeneity in the variance, which permits to assert that the regression model is valid and that 90% of the variation of measured TS values can be explained through the computed TS values. However, in the regression analysis proposed for measured and computed chlorophyll concentration values, the necessary assumptions for accepting the fit were not met.
CONCLUSION Many works in remote sensing using techniques and methods for classifying continental water bodies and for determining chlorophyll and solids concentrations in water similar to those here presented have been published. They have provided insightful advancements in knowledge, but also in ecosystem management. The choice of the sensor, the availability of satellite images in each country, as well as particular characteristics of individual water bodies, brings up a series of limitations that need to be solved in each case.
60
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
3333.31
ST measured
2510.87
1688.44
866.00
43.56 447.61 1053.13 1658.65 2264.17 2869.69
STcalculated
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 18. Regression analysis of measured and calculated total solids with confidence and prediction bands
The present chapter is devoted to the study of the dynamics of shallow water bodies. However, much of the methods and techniques here presented can be used for other type of studies, for instance forests, land use for agriculture, or crops dynamics. The importance of an adequate visualization of results is worth noting. The study of particular ecosystems can generate a large volume of data which are impossible to read or understand straightforwardly. Hence, it becomes necessary to find alternative forms of data visualization that can synthesize key information in an user-friendly way. One example is the leveled graphs format presented here. Another important aspect for the use of satellite data is the need to have an adequate knowledge of statistical tools. A thorough understanding of univariate and multivariate techniques helps avoiding wrong or misleading results. Besides, it is necessary to know when it is possible to use a particular technique because in life sciences problems, such as those in environment and ecology, many traditional statistical methods are not easy to apply. Quite recently, some alternative methodologies have been developed with the purpose of solving these types of problems. The current bibliography on these methods is quite abundant. However, some relevant works are those of Anderson and Robinson (2001), Anderson and Ter Braak (2003), Manly (1997), and Pillar and Orloci (1996). The results obtained for the classification of shallow lakes using the method of spectral unmixing on the one hand, and artificial neural network models on the other, have been compared and statistically validated using field data. After validation, both methods can be used to gain knowledge on the trophic state of the numerous shallow lakes in this region in real time from Bands 2 and 4 of the LandSat TM images. Pampean shallow lakes are mostly eutrophic and hypereutrophic. They challenge the usual methodologies because of the influence that this character has on the optic characteristics that are captured by the satellite sensors. Much of the published research is based on LandSat imagery, and usually the
Monitoring Shallow Lakes in the Pampas
61
methods have been developed for water bodies that exhibit chlorophyll-a concentrations bellow 100 mg/m3. (Svab et al., 2005; Liu et al., 2003). However, that figure is widely surpassed by the concentrations found in the shallow lakes of the Pampean region: 45 to 1400 mg/m3. Future work shall be oriented to developing new methods for the determination of chlorophyll concentrations in eutrophic and hypereutrophic water bodies with greater precision.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REFERENCES Adams, J. B., Smith, M. O., and Johnson, P. E. (1985). Spectral mixture modeling: a new analysis of rock and soil types at the Viking Lander I site. J. Geophys. Res. 91:80988112. Agresti, A. (1990). Categorical Data Analysis. John Wiley, New York. 560pgs. Anderson, M. and Robinson, J. (2001). Permutations test for lineal models. Aust. N. Z: J Stat. 43(1): 75-88. Anderson, M. and Ter Braak C. J. (2003). Permutations test for multi-factorial analysis of variance. J. Statist. Comput. Simul. 73(2): 85-113. Bateson, A., and Curtiss, B. (1996). A method for manual endmember selection and spectral unmixing. Remote Sensing of Environment. 55: 229-243. Brivio, P. A., Giardino, C., and Zilioli, E. (2001). Determination of chlorophyll concentration changes in Lake Garda using an image-based radiative transfer code for LandSat TM images. International Journal of Remote Sensing 22(2-3): 487-502 Chipman, J.W., Lillesand, T.M., Schmaltz, J.E., Leale, J.E., and Goldmann, R.A. (2004). Mapping lake water clarity with LandSat images in Winconsin, USA. Special issue on Remote Sensing and Resources Management in Nearshore and Inland Waters. Canadian Journal of Remote Sensing 30(1):1-7 Crist, E.P., and Cicone, R.C. (1984). A physically-based transformation of Thematic Mapper data: the TM Tasseled Cap. IEEE Trans. Geosciences and Remote Sensing 22: 256-263. Dangavs, N.V. (1982). Los ambientes lagunares de la Provincia de Buenos Aires. Documento relativo a su conocimiento y manejo. CICPBA 43-49 Dukatz, F., and Ferrari, R. (2007). Sistematización del análisis y clasificación de cuerpos de agua según su permanencia mediante sensores remotos. In: JProceedings of the Internacional Congreso on Development, Environment and Natural Resources: Multilevel and Multi-scale Sustainability (Eds. Feyen, L.F. Aguirre, M. Morales R.), Cochabamba, Bolivia, Volume II:1102-1107 (D.L. 2-1-1269-07). Frenguelli, J. (1956). Rasgos generales de la hidrografía de la Provincia de Buenos Aires. LEMIT II(62): 1-19 Huang, C., Wylie, B., Homer, C., Yang, L., and Zylstra, G. (2002). Derivation of a Tasseled Cap transformation based on LandSat 7 ETM at-satellite reflectance. International Journal of Remote Sensing 23:1741-1748. Johnson, R.A., and Wichern, D. W. (1992). Applied Multivariate Statistical Analysis. Third edition. Prentice-Hall, Inc. New Jersey (USA), 642 pp. Kauth, R. J. and Thomas, G. S. (1976). The Tasseled Cap --a graphic description of the spectral-temporal development of agricultural crops as seen in LandSat. In: Proceedings
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
62
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
of the Symposium on Machine Processing of Remotely Sensed Data, June 29 -- July 1, 1976, (LARS, Purdue University), West Lafayette, Indiana, 41-51. Kirk, J.T.O. (1994). Light and Photosynthesis in aquatic ecosystems. 2nd edn. Cambridge University Press. Cambridge. 509 pp Landis, J.R., and Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics 33:159-174. Liu, Y., Islam, M. A., and Gao, J. (2003). Quantification of shallow water quality parameters by means of remote sensing. Progress in Physical Geography 27:24-43 Manly, B. F. J. (1997). Randomization, bootstrap and Monte Carlo methods in biology. 2nd Edition. Chapman & Hall, London, United Kingdom. Mardia, K., Kent, J., and Bibby, J. (1979). Multivariate Analysis. Academic Press. 521pp. McClelland, J. L. and Rumelhart, D. E. (1988). Explorations in Parallel Distributed Processing. Cambridge, MIT Press. Minsky, M.L., and Papert, S. A. (1969). Perceptrons, Expanded Edition. Cambridge, MIT Press. NASA/USGS LandSat 7 Program, (1998). LandSat 7 Science Data Users Handbook http://landsathandbook.gsfc.nasa.gov/handbook.html Neiff, J. J. (2004). El Iberá ...¿ en peligro? Fundación Vida Silvestre Argentina, ISBN 9509427-10-I. 89 pp. Nelson, S.A.C., Soranno, P.A., Cheruvelil, K.S., Batzli, S. A., and Skole, D. L. (2003). Regional assessment of lake water clarity using satellite remote sensing. Journal of Limnology 62(1): 27-32. Pillar, V., and Orloci, L. (1996). On randomization testing in vegetation science: multifactor comparisons of releve groups. Journal of vegetation science 7: 585-592. Prieur, L., and Sathyendranath, S. (1981). An optical classification of coastal and oceanic waters based on the specific spectral absorption curves of phytoplankton pigments, dissolved organic matter and other particular materials. Limnology and Oceanography 26: 671-689. Pozdnyakov, D., Korosov, A., Shuchman ,R., and Edson, R. (2003). Development of a tool for the assessment of water quality from visible satellite imagery taken over turbid inland waters (with Lake Michigan as an example). Geoscience and Remote Sensing Symposium IGARSS ‘03 2:746-748 Quirós, R., Resella, A., Boveri, M., Rosso, J., and Sosnovsky, A. (2002). Factores que afectan la estructura y funcionamiento de las lagunas pampeanas. Ecología Austral 12:175-185. Ringuelet, R. A. (1962). Ecología acuática continental. EUDEBA, Buenos Aires. 207pp. Rohatgi, V. K. (1976). An Introduction to Probability Theory and Mathematical Statistics. JohnWiley & Sons, New York. 673pp. Rosenblatt, F. (1962). Principles of Neurodynamics. Spartan, New York. Rumelhart, D.E., Hinton, G.E., and Willams, R.J. (1986a). Learning Internal representations by Error Propagation. In: Parallel Distributed Processing, Explorations in the Microestruture of Cognition (Eds. Rumelhart, D. E. and McClelland, J. L.); Vol. 1: Foundations. Cambridge, MIT Press. Rumelhart, D.E., Hinton, G.E., and Willams, R.J. (1986b). Learning Representations by Back-Propagation Error. Nature 323:533-536.
Monitoring Shallow Lakes in the Pampas
63
Svab E., Tyler A. N., Preston, T., Presing, M., and Balogh, K. (2005). Characterizing the spectral reflectance of algae in lake waters with high suspended sediment concentrations. International Journal of Remote Sensing 26(5): 919-928 Scheffer, M. (2004). Ecology of Shallow Lakes. Kluwer Academic Publishers. 357 pp. Sioli H. (1975). Tropical rivers as expressions for their terrestrial environments. In: Tropical Ecological Systems. Trends in terrestrial and aquatic research (Eds. Golley, F.B. and Medina, E.). Springer-Verlag, New York. 275-288 pp. Smith, M.O., Ustin, S.L., Adams, J.B., and Gillespie, A.R. (1990). Vegetation in deserts: I. A regional measure of abundance from multispectral images. Remote Sensing of Environment 31:1-26. Soriano A. (1992). Río de la Plata Grasslands. In: Ecosystems of the world 8ª. Natural Grasslands. Introduction and western hemisphere (Eds Coupland R.T.). Elsevier. New York, 367-407pp. Svab E., Tyler A.N., Preston T., Présing M., and Balogh K.V. (2005). Characterizing the spectral reflectance of algae in lake waters with high suspended sediment concentrations. International Journal of Remote Sensing 26(5): 919-928. Ter Braak, C. J. F. (1994). Canonical Community Ordination. Part I: Basic Theory and Linear Methods. Ecoscience 1(2): 127-140. Tyler, A.,N., Svab, E., Preston, T., Présing, M., and Kovács, W. A. (2006). Remote sensing of the water quality of shallow lakes: A mixture modelling approach to quantifying phytoplankton in walter characterized by high-suspended sediment. International Journal of Remote Sensing 27(8): 1521-1537. USGS (2006). Multi-Resolution Land Characteristics 2001 (MRLC2001) ImageProcessingProcedure(http://landcover.usgs.gov/pdf/image_preprocessing.pdf)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
APPENDIX I The water bodies that were selected to illustrate this chapter were eight shallow lakes located in the Province of Buenos Aires. The field sampling was carried on between December 2004 and August 2006. The number of trips to each shallow lake was: eight (8) for La Brava; twelve (12) for Del Estado, Quilla Lauquen, and El Paraiso; sixteen (16) for La Salada, El Chifle, and San Antonio; and seventeen (17) for La Barrancosa. The National Commission for Space Activities (Comision Nacional de Actividades Espaciales, CONAE) provided sixteen satellite images (224/86, 225/86, and 226/86), ten from LandSat 5 TM and six from LandSat 7 ETM+. The delay between the dates of the field trips and the satellite images was kept bellow five days as can be seen in the following table.
64
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al. Table Detail of the type of images and dates of field sampling
Images Path/Row, Satellite
Shallow Lake (area)
Field Sampling Date
La Barrancosa (181 ha)
2004/12/15
Date 226/86, Landsat 5 TM 2004/12/14
226/86, Landsat 5 TM 2005/03/04
La Barrancosa (172 ha) Del Estado (170 ha)
2005/02/28
El Paraíso (102 ha)
225/86, Landsat 5 TM
La Barrancosa (166 ha)
2005/03/13
San Antonio (139 ha)
2005/03/16
La Barrancosa (172 ha) 225/86, Landsat 5 TM
El Chifle (134 ha)
2005/04/22
La Salada (158 ha)
2005/04/28
San Antonio (182 ha) 226/86, Landsat 5 TM 2005/05/07 226/86, Landsat 5 TM 2005/06/08
Del Estado (162 ha) El Paraíso (91 ha)
2005/05/05
Quilla Lauquen (108 ha) Quilla Lauquen (101 ha) Del Estado (147 ha)
2005/06/06
El Paraíso (96 ha)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
La Barrancosa (164 ha) 225/86, Landsat 5 TM 2005/07/11
El Chifle (131 ha) La Salada (152 ha)
2005/07/07
San Antonio (174 ha) Quilla Lauquen (98 ha)
226/86, Landsat 7 ETM 2005/07/18
El Paraíso (107 ha)
2005/07/13
Monitoring Shallow Lakes in the Pampas Images Path/Row, Satellite
Shallow Lake (area)
Field Sampling Date
Date Del Estado (139 ha) 226/86, Landsat 5 TM
El Paraíso (97 ha)
2005/09/12
Quilla Lauquen (104 ha)
2005/09/07
La Barrancosa (167 ha) 225/86, Landsat 7 ETM 2005/09/13 225/86, Landsat 7 ETM 2005/10/15
San Antonio (175 ha) La Salada (188 ha)
2005/09/07
El Chifle (133 ha) San Antonio (171 ha) La Salada (151 ha)
2005/10/14
El Chifle (133 ha) La Barrancosa (167 ha)
226/86, Landsat 5 TM
Quilla Lauquen (102 ha)
2005/10/14
El Paraíso (96 ha)
2005/10/14
Del Estado (131 ha) La Barrancosa (158 ha) 225/86, Landsat 7 ETM 2005/12/02
Quilla Lauquen (94 ha) El Chifle (131 ha)
2005/12/02
San Antonio (163 ha)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
La Salada (153 ha) 226/86, Landsat 7 ETM
El Paraíso (83 ha)
2005/12/09
Del Estado (88 ha)
2005/12/02
La Barrancosa (140 ha) 226/86, Landsat 7 ETM
Quilla Lauquen (94 ha)
2006/01/26
El Paraíso (77 ha)
2006/01/30
Del Estado (83 ha) 224/86, Landsat 5 TM 2005/02/02
La Brava
2005/02/06
65
66
Federico Dukatz, Rosana Ferrati, Claudia Marinelli et al.
QUESTION BANK Fill in the Blanks 1. Artificial neural network uses__________ mathematical function. 2. Principal Components Analysis (PCA) permits exploring the degree of _____ between variables through the analysis of the structure of _____. 3. Reflectance from the vegetation is _______ in red band and _________ in NIR band. 4. Tasseled Cap transformation is to measure ______, ________ and _____ from the satellite data. 5. Turbid water reflects _______ than that of pure water in near infra red wavelength.
Small Answer Questions 1. Explain importance of radiometric correction. 2. What are spectral signatures? Explain the spectral signature of vegetation. 3. What is importance of accuracy assessment? What is importance of Kappa in accuracy assessment? 4. What is spectral unmixing? 5. What is the importance of Band 4 while assessing the water bodies?
Long Answer Questions
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. What is image masking? Why is this important? Elaborate criteria for generating masks for the study of (a) forested areas, (b) cultivated lands. Justify the selection of bands. 2. How is it possible to make Recurrence Maps with satellite images? 3. From the suggested literature, identify other parameters that could be estimated using Artificial Neural Networks. 4. Suggest a framework to monitor Lake Eutrophication using satellite remote sensing. What are the limitations of these tools while carrying out this assessment?
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 4
APPLICATIONS OF SATELLITE REMOTE SENSING AND GIS TO OCEANOGRAPHY AND FISHERIES: CASE STUDIES IN THE WESTERN IBERIA A. Miguel P. Santos*,1, Pedro B. Machado‡1 and Paulo Relvas†,2 1
Instituto Nacional de Recursos Biológicos (INRB) - IPIMAR, Av. Dr. Alfredo Magalhães Ramalho s/n, 1449-006 Lisboa, Portugal 2 Centro de Investigação Marinha e Ambiental (CIMA), FCMA-Universidade do Algarve,Campus de Gambelas, 8000-117 Faro, Portugal
ABSTRACT
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
This chapter focuses on the importance of remote sensing and GIS for oceanography and fisheries research and support. It provides a brief description of principles, technologies and methodologies of satellite oceanography (visible, infrared and microwaves) and marine GIS applications (with a short introduction to geostatistics). With the help of case studies in visible, infrared, microwave region; the chapter focuses on potential application of this tool. The chapter also provides details of GIS, geostatistical techniques and tools like Vessel monitoring system, SIG-IPMAR for the visualisation of data pertaining to fisheries.
Keywords: Fisheries, geostatistics, GIS, oceanography, remote sensing, satellites.
INTRODUCTION The water constitutes approximately 70% of the Earth, which is usually called the Blue Planet, being about 97% of it stored in the world oceans. Thus, the oceans are the main *
Email: [email protected] [email protected] † [email protected] ‡
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
68
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
element of the hydrosphere and an important component of the Earth climate system (Peixoto and Oort, 1992), as well as an important source and environment for many Human activities (e.g., transport, fisheries, oil exploration, and energy). However, the oceans are not easily accessible, are a ‘wild’ environment for Man and, due to their extension, difficult to observe, namely as a whole (in space) and synoptically (in time). Thus, appropriate technologies and methodologies that allow the observation of the ocean at the appropriate space and time scales are fundamental for their study and monitoring. In that sense, satellite remote sensing (RS) has these capabilities and it is a powerful tool for monitoring marine ecosystems, namely in a global scale. The main impact brought by ocean remote sensing was the capacity "for the first time" to sampling adequately ocean processes (Munk, 2000). In fact, most of the actual knowledge of the world oceanography was built based on remote sensing of the ocean surface. The synoptic view of large portions of the ocean, with relatively high resolution, significantly improves the knowledge acquired from more traditional data sources. The discrete nature of the direct observations does not allow contemporaneous data and fine resolution. The repetitive coverage of the satellite remote sensing data, at relatively frequent intervals, makes possible the detection and monitoring of the ocean state and behaviour. Along with the synoptic character, it is of particular interest for evaluation of larger scales phenomena and processes. Satellite imagery can also survey areas that may be physically or politically inaccessible or that are too large to cover with other methods. However, we should not be tempted to believe that remote sensing of the ocean will be the panacea for the observation and understanding of the ocean. Fedorov and Sklyarov (1981) wrote wisely in a paper included in the book ‘Oceanography from Space’ that space oceanography is not a new science "but may only be regarded as an essential supplement to the already working sophisticated arsenal of modern 'direct' or 'contact' means and methods of good old oceanography". In other words, good in situ observations are still needed. Despite the fact that many underwater phenomena have a surface discernible signature and satellite remote sensing could be used to study them, as for example the sea surface temperature patterns observed at surface which are representative of the upper layer (100-200 meters) for most of the regions or internal waves that manifest in the surface as slicks and roughness, we should remember that must of satellite sensors only measure the skin or the few centimetres/meters of the ocean surface. If our goal is a complete description of the interior of the ocean, 'traditional' in situ (e.g., ships observation, moorings and other profiling instruments) are still very important (Dickey et al., 2006). Thus, the way to better understand the ocean ecosystem should be the synergetic use of both in situ and satellite-derived data. Spatial management (or ocean zoning) has been advocated as a powerful tool to conserve biodiversity, restore habitats and manage fisheries (NRC, 2001; Crowder et al., 2006). Although the principle of management of spaces rather than species is not new on land, it has not been frequently implemented as a major marine management measure. An important issue that fisheries assessment and management should deal is the spatial and temporal structure in the distribution of fish stocks, which could affect the relationships between catch rate and abundance (Cooke, 1985). A good example in fisheries management that alludes to the statement "place matters" is the collapse of northern cod (Gadus morhua) off eastern Canada which are consider to be associated with spatio-temporal, changes in density and biomass (Hutchings, 1996). The acknowledgement of the importance of spatial processes in marine ecosystems imply the application of new technologies and methodologies (Cowen et al., 2007), as Geographical Information Systems (GIS) and RS, that allow their observation,
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 69 understanding and future application of this knowledge for a wise management, conservation and sustainable use of the ocean.
STATE-OF-THE-ART Satellite Remote Sensing
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Although satellite oceanography history is already almost four decades long, a wide variety of improved and innovative sensors have been launched in the last decade on board of satellites. The recognition of the potential of applying satellite RS to oceanography started in the 1960s with Earth photographs taken by the first astronauts. These photographs revealed an amazing amount of details in ocean structures. Figure 1 is an example of the kind of images that can be obtained by astronauts. Although these pictures had created enormous expectations among the scientific community, it is only in the end of the 1970s-beginning of the 1980s with the installation of the AVHRR (Advanced Very High Resolution Radiometer) on board the second generation of NOAA satellites that remotely-sensed satellite data started to play a major role in oceanography, namely in the determination of sea surface temperature. Since then, AVHRR has been carried on the meteorological satellites of the NOAA series until now, providing regular and continuous operational global observations of SST (Sea Surface Temperature) four times per day using two satellites for almost 30 years. This allow to produce consistent and reliable satellite-derived SST time series, namely for assess climate change (Emery et al., 1995; Kilpatrick et al., 2001). AVHRR is planned to continue to flow on future NOAA-EUMETSAT (European Organisation for the Exploitation of Meteorological Satellites) missions, namely the MetOp satellites series planned to be operational at least until 2015. The ATSR (Along Track Scanning Radiometer) and AATSR (Advanced Along Track Scanning Radiometer) are more recent instruments carried by the ESA's satellites (ERS-x and ENVISAT), which the main purpose is to make more accurate measurements of global SST (Stricker et al., 1995; Minnett, 1995a; b; Smith et al., 2001; Merchant et al., 2007).
Figure 1. - Space Shuttle photograph of the Oyashio Current in the Bering Sea on March 1992. In the image it is clear the detail and complexity of the features that are present in the ocean (Image courtesy of the Image Science and Analysis Laboratory, NASA Johnson Space Center "The Gateway to Astronaut Photography of Earth." http://earth.jsc.nasa.gov/sseop/efs/printinfo.pl?PHOTO=STS045-79N)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
70
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
The first satellite specifically designed for ocean applications was SeaSat-A launched in 1978 (Borne et al., 1979). Although it last only about 3 months due to spacecraft malfunction, its data clearly demonstrated the capabilities of microwave sensors for ocean research (e.g., circulation; surface winds, wave heights, tides, storm surges). After the demonstration of the potential of this kind of sensors for the monitoring of the global ocean, subsequent satellites missions carrying similar instruments were launched, such as GEOSAT that flown from 1985 to 1990 and subsequent follow-on missions, ESA's ERS series (1991-present) and ENVISAT (2002-present) satellites, the French-American TOPEX/Poseidon (1992-present) and Jason (2001-present), the Japanese ADEOS (Advanced Earth Observing Satellites, called Midori in Japanese) series (1996-1997 and 2002-2003), NASA's QuickSCAT (1999-present), and the Canadian RADARSAT (1995-present) (Pettersson et al., 1995; Heimbach and Hasselmann, 2000; Fu, 2001). One of the great advantages of the microwave sensors in relation to the visible and infrared ones is the capacity of provide an all-weather coverage of the ocean because they penetrates through clouds, rain and snow. Another important milestone in ocean remote sensing was the launched in 1978 of CZCS (Coastal Zone Color Scanner) flown on NIMBUS-7 satellite that allows an 8 years data set widely used in ocean colour (biology and optics) research. Despite of this immense potential for biogeochemical, as well as physical oceanography studies, the ocean colour community had to wait for about 10 years for a new ocean colour mission, namely the Japanese ADEOS OCTS (Ocean Color and Temperature Scanner on board ADEOS satellites) and POLDER (POlarization and Directionality of the Earth's Reflectances - French sensor on board ADEOS satellites) (1996-1997), the German MOS on board the Indian satellite IRS-P3 (1996-2004) and the US SeaWiFS flown on Orbview-2 satellite (1997-present). Presently there are 10 ocean colour sensors in operation and the continuity of colour measurements from space are guaranteed at least for the next decade with 6-7 more missions planned (Dickey et al., 2006; IOCCG, 2007). A good source of information about satellite ocean colour is the web site of the International Ocean-Colour Coordinating Group (IOCCG) at http://www.ioccg.org/ index.html. Salinity is an ocean state variable that partially controls sea water density and has important implications for example in climate, ocean circulation and ecology. However, their global monitoring remains poor because of the scarcity of observations in large portions of the ocean. The objective of the future ESA's SMOS (Soil Moisture and Ocean Salinity schedule for launch in 2008) and NASA/CONAE's [Comisión Nacional de Actividades Espaciales (Space Agency of Argentina)] Aquarius (2009) missions, is to measure surface salinity systematically at a global scale. Some more information about ocean remote sensing of salinity could be found in Lagerloef (2000; 2001), Le Vine et al. (2000) and Lagerloef and Schmitt (2006). Sensor calibration and data processing, a central problem in the remote sensing initial stages, has known a tremendous evolution. Consequently, the range of parameters observed from space, or derived from space observations, with positive results has broadened. The oceanographic scales resolved by remote sensing have decreased, and research on satellite observations of sub-mesoscale features is underway. Application of ocean RS that illustrated this evolution can be found in several reviews and books published during the last years, namely Ikeda and Dobson (1995), Nihoul et al. (1998), Halpern (2000), Cheney (2001), Fu (2001), Liu and Katsaros (2001), Liu and Wu (2001), McClain (2001), Minnett (2001), Parkinson (2001) and Plant (2001). To our knowledge the last review about the use of satellite
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 71 and airborne remote sensing methods in fisheries was published by Santos (2000), in which several references to previous reviews about the subject could be found. Nowadays, many countries use satellite remote sensing technology for operational fisheries forecasting services, namely Japan and the US (Santos, 2000). Platt et al. (2007) discuss the role of biological oceanography in fisheries management and the importance of satellite RS as a tool complementary of ship observation for studying ocean processes of relevance at appropriate time and space scales. Future missions planned until 2015 that include instruments for ocean monitoring are presented in Figure 2. ESA also plans to launch the Sentinel satellite series in 2011-2012 in the frame of the Global Monitoring for Environment and Security (GMES) programme that will monitor the marine environment (e.g., ocean circulation, sea-level, chlorophyll concentration, sediments, oil spills), sea ice and icebergs among other things (Atterna et al., 2007; Aguirre et al., 2007).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Geographic Information System Geographic Information Systems (GIS) appeared in the beginning of the 1960s after the pioneer work of Waldo Tobler and Peter Hagget in the fields of automated geography and spatial modelling. The Canadian GIS of 1964, considered by many as the original GIS, was coordinated by the geographer Roger Tomlinson and incorporated for the first time, attribute tables (thematic data) together with algorithms and functions for overlay operations (Burrough and McDonnell, 1998). Life-sciences is one of the areas with an extensive record on GIS applications, covering the fields of forest management, wildlife habitat, wild rivers preservation, wetland preservation, agricultural lands management, groundwater modelling, environmental impact and fisheries management. This is not surprising considering the benefits that a GIS has to offer, such as: a) visual display of the areas of interest and respective indicator, damage values; b) a set of spatial modelling techniques to conduct analysis and c) fast and efficient handling of large datasets. In addition, these systems have extended map update capabilities in digital format and allow the flow of spatially related information in a standardized format (Meaden and Do Chi, 1996). Oceanography and fisheries, two key research fields for the comprehension of the marine environment, have registered an increase in the number of GIS applications during the last decade. The following constitute some of the main examples of these applications: • • • • • • • • •
Recording of ecologic variability; Mapping of marine resources distribution and abundance; Data repository and analysis of seafloor surveys data; Monitoring of marine resources dynamics; Coastal management; Evaluation of marine pollution; Assess the variability of oceanographic processes; Enable the study of anthropogenic impacts on marine resources; and Multi-dimension visualization of marine environments.
72
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 2 - Present and future satellite oceanography missions (extracted from the Web site of the NOAA-NESDIS-Center for Satellite Applications and Research (STAR) at http://www.orbit.nesdis.noaa.gov/star/socdr_driverstrends_programs.php).
Despite the early emergence of a marine GIS, back in the 1960s (Coppock and Rhind, 1991 in Wright, 2000), GIS applications in oceanography and fisheries only started to have visibility in the 1980s, reflecting a growth of digital data in scientific institutions. Oceanographic case studies from these years more resembled RS methodological approaches than a true GIS application, aiming to find relationships between ocean currents/physical parameters and the distribution/abundance of marine resources. As for fisheries applications, initial systems were more focused in aquaculture problems such as the identification of suitable locations to increase farming profitability (Meaden, 1987). The main motivation factors to the development of GIS applications in fisheries during the 1980s were (Meaden and Kapetsky, 1991): • •
The importance of the spatial definition of fishing areas and related resources; Recognition of the dependence of high fish yields on physical and economic parameters; and
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 73 •
The existence of different interests competing in the spatial interface land/water.
GIS did not develop so fast in oceanography and fisheries as for earth sciences mainly because traditional applications and data models were designed for land-based situations. In fact, marine related data usually presents peculiarities that require different approaches and data models to be generated. Indicated below are some of the features/problems posed by this type of data in the scope of a GIS analysis: •
•
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
• • •
Multiple dimensionality of the data (e.g., natural phenomena attributes have commonly associated geographic coordinates, depth and time; vertical casts for salinity where sampling is frequent in the vertical direction but sparse in the horizontal); Dynamism (e.g., frequently, study objects correspond to dynamic forms such as ocean fronts or fish shoals); Inherent fuzziness of biological boundaries; Lack for spatial data structures that vary relative positions and values over time; and Need to combine distinct datums from land and sea.
The synergy of different types of marine data collected from multiple sampling platforms provide oceanographers and policy decision-makers with more information and insight than could be obtained by considering each type of data separately (Wright and Goodchild, 1997). This can be easily achieved within a GIS if regular two-dimension data sets are used, however, if these are enlarged with the addition of depth and time its analysis became more complex. In effect, the benefits of three dimensional modelling and volumetric visualization for interpreting multi-dimensional data depend upon the ability of the user to correctly interpret it and obtain meaningful results from it (Manley and Tallet, 1990). To enhance the capabilities of users in the analysis of multi-dimensional data is necessary that more threedimensional GIS tools became available in order to effectively understand the data and the environment it represents (Manley and Tallet, 1990). Further challenges to GIS in oceanography and fisheries include the handling of dynamic spatial data structures that are able to vary their relative positions and values over time, robust interpolation methods that account for data sparse in one-dimension as compared to the others and the development of adequate Information Technology tools (e.g., Oracle Spatial objects) for the representation of marine phenomena within a database management system. For instances, Breman (2002) and Wright (2002) present several applications of GIS in marine studies in subjects as broad as seafloor mapping, nautical charts, remotely operated vehicles, tracking marine mammals and fish, marine benthic habitat mapping, marine protected areas, fisheries, oil spill response, marine archaeology and marine policy. In Western Iberia waters, there are still few GIS applications to oceanography and fisheries, being most of them either land-based or associated with coastal (near-shore) or shore locations. Although existent marine applications were developed by several institutions, the majority were conceived recently by the Portuguese Laboratory for Fisheries and Sea Research (INRB-IPIMAR), and in the case study section we describe the main GIS applications to oceanography and fisheries developed for this region.
74
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Principles, Technologies and Methodologies Satellite Remote Sensing Remote sensing is the acquisition of information about an object or phenomenon without physical contact with it. In general, the term remote sensing is used to describe the measurement and analysis of information that is obtained with a sensor installed on board satellites and airplanes, although we can consider also as remote sensing the observations of vessel-mounted sonars and ecosounders. Thus, ocean remote sensing allows the measurement of oceanographic parameters and the knowledge of their spatial and temporal variability without physical contact with the sea. It is based on the measurement, processing and analysis of the electromagnetic radiation emitted by the ocean, or reflected by the sea surface from the incident solar radiation (passive remote sensing) or from emitting sources on board of the satellite (active remote sensing). Typically all the row of satellite remote sensing of the sea includes the sensor calibration, atmospheric correction, geo-location, sampling the ‘sea truth’ conditions, image processing and applications of the satellite remote sensing. Next we will detail each step. All sensors employed on ocean observing satellites use electromagnetic radiation to view the sea. Sensors are calibrated before launching, but no periodic recalibration can be executed. Some sensors use onboard reference targets, with constant optical characteristics, to detect instrumental drifts that can be corrected during data processing. Only certain wavelengths of the electromagnetic radiation are fully or partly transmitted through the atmosphere and the transmittance varies with the composition and state of the atmosphere. Several strategies can be applied to deal with the atmospheric effect, ranging from detailed modelling of the atmospheric conditions during data acquisition to simple calculations based solely on the image data. However, for satellite imagery in visible and infrared spectral bands, the cloud cover detection is the main difficulty. Analysis of the pixel reflectance by comparison with a preset threshold or with the surrounding pixels is used to map the cloud contamination. The geo-location consists in the attribution of a geographical location to each pixel. Often control points in the ground are used, but this is difficult far from coastal regions. In recent years the positional identification process has taken advantage of the GPS satellites signals. More sophisticated methods, based on Doppler effect for orbit determination, are used in radar-altimeter sensors due to the required accuracy. The choice of the strategy for the oceanographic sampling of the sea truth is very important, and is closely related to the spatial resolution of the remote sensing data when compared with the spatial variability of the measured parameter. The samples shall span in large spectrum of data values as possible, keeping in mind that the value measured in a point may not be representative of the average parameter within the whole pixel sampled by the satellite. Oceanographic remote sensing data are presented in several levels of processing that tend to be standardized. The level-0 contains the raw data as transmitted by radio-signal and received by the ground station. Level-1 adds calibration data, navigation data, and instrument and spacecraft details to the previous level. Level-2 product presents already oceanographic parameters derived from the application of algorithms (e.g. surface temperature, surface chlorophyll concentration, etc.), atmospherically corrected and geo-located in sensor coordinates. Level-3 includes oceanographic parameters sampled during a certain period and
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 75 interpolated on a geographical grid. Level-4 means images representing ocean variables averaged within each grid cell as a result of data analysis and/or modelling. These products can contain data merged from different sources, including in situ measurements, and the gaps where no observations are available filled by optimal interpolation schemes. Finally the multiple applications of the satellite remote sensing depend on the sensor and/or sensor configuration. For oceanographic purposes the important remote sensing devices are the infrared radiometers, visible wavelength sensors, passive microwave radiometers, active microwave sensing and satellite altimetry. Next we will briefly describe each device. Visible wavelength or ‘ocean colour’ sensors operate in the visible part of the electromagnetic spectrum, measuring solar radiation returned by the ocean in this band. The sunlight that penetrates the ocean surface is selectively absorbed, reflected and scattered by the suspended material in the upper layers. Thus, the ocean colour reflects the suspended matter in the top layer of the ocean, with depths of 50 meter or more in the case of colour radiance in the blue-green, in contrast with the infrared observations that sample only a surface film of the ocean. From the satellite remote sensing of the ocean colour it is possible to estimate, with variable accuracy, a set of biological and optical parameters and processes, including chlorophyll-a (phytoplankton) concentration, water-leaving radiation, diffuse attenuation coefficient, algal blooms and particulate matter as some of the most common. The most significant ocean colour sensors are the CZCS (Coastal Zone Color Scanner) that operated between 1978 and 1986, SeaWiFS (Sea-viewing Wide Field-of-view Sensor), since 1997, and MODIS (Moderate Resolution Imaging Spectroradiometer) on board of Terra satellite, launched in late 1999, and on board of Aqua satellite launched in mid 2002. Infrared scanning radiometers infer the sea surface temperature (SST) from near-infrared and infrared sensors measuring the electromagnetic radiation within the band 1-30 µm, emitted by the ocean surface. Processing is based on the fact that all surfaces emit radiation, the strength of which depends on the surface temperature. Stefan-Boltzman law states that the total emitted energy is proportional to the fourth power of the temperature. Sea surface temperature is difficult to interpret because the upper ocean (approx. top 10 meters) has a complex and variable vertical temperature structure that is related to ocean turbulence and airsea fluxes of heat, moisture and momentum. The thickness of the layer whose temperature is remotely sensed varies approximately between 10-20 µm depths. It is called skin SST and it measurement is subject to a large potential diurnal cycle including cool skin layer effects (especially at night under clear skies and low wind speed conditions) and warm layer effects in the daytime. In contrast, the measured in situ SST (called also bulk SST) corresponds to few centimetres or more, depending on surface waves. The SST measurements on buoys and ships may be anything between 0.5 and 3 m deep. The presence of surface films, like transient oil slicks, may also affect the difference between skin and bulk SST’s. The most important SST sensors are the Advanced Very High Resolution Radiometer (AVHRR) on NOAA near polar earth orbiting satellites since 1978, the ATSR (Along Track Scanning Radiometer) on board ESA satellites, MODIS and some others. Passive microwave radiometers operate at electromagnetic wavelengths 1.5–300 mm (frequency 1–200 GHz). They are not sensitive to scattering by the atmosphere or aerosols, haze, dust, or small water particles in clouds, due to the relatively long wavelength. So, the microwave sensors can survey in all weather conditions. This principle advantage is countered by the fact that thermal emission is very weak at these longer wavelengths. To overcome noise levels a large field of view must be received, that results in low spatial
76
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
resolution (25–150 km). These observations are used for studies of heat balance of the ocean. The emissivity of the sea at microwave frequencies varies with the dielectric properties of sea water (including salinity) and the surface roughness. Hence, the development of this technique in future may enable the measurements of surface salinity. Active microwave radiometers, with oblique viewing, measure the sea surface roughness based on the Bragg scattering. Oblique viewing of a smooth surface with active radar give virtually no return, but if the surface is rough significant backscatter occurs, which is perceived by the microwave scatterometer. The Synthetic Aperture Radar (SAR) is an active microwave device, based on the comprehensive analysis of contribution from individual points to the signal received when the sensor is at a particular point. The result has very high resolution. SAR images enable the analysis of small-scale and mesoscale eddies, internal waves, river plumes, oil slicks, ice packs, etc. Sea surface roughness is also the elemental parameter for the estimation of the wind over the ocean. The SeaWinds instrument on the QuikSCAT (Quick Scatterometer) satellite is a specialized microwave radar lofted in mid 1999 to measure the near surface wind speed and direction under all weather and cloud conditions over the global ocean. The derived operational product presently gives vector winds with 25 × 25 kilometers resolution. Remote sensing of the sea surface topography is carried out by satellite altimeters radars, which permanently transmit signals at high frequency toward the earth beneath them. The return time of the signal after reflection at the earth's surface is measured, and this yields the height of the satellite. From all sensors carried on satellites, the altimeter is the most dependent upon its orbit to achieve successful calibration and interpretation. Sea surface is rough rather than flat and each individual return signal is very noisy. The radar altimeter is able to remove the effect of the ocean waves by averaging many successive pulses. The most important are TOPEX/Poseidon, ERS and Jason satellites.
Geographic Information System One of the main advantages presented by GIS is the capacity to structurally integrate spatial data with non-spatial data on the same repository, allowing the formulation of queries that resulted in the generation of dynamic maps with new geographic data. Although no consensual definition of GIS exists in the literature (Chorley, 1987; Gaile and Willmott, 1989; Star and Estes, 1990; Clarke, 1995; Goodchild, 1997), a comprehensive but brief description of GIS would be: a computer-based system with software tools for input, storage, manipulation (analysis), and output of geographic information (Fig. 3). It is a system that provides an alternative way to access data based on geographic location, allowing the production of data queries based not only on attributes but also on spatial locations (Goodchild, 1992). Spatial data can be structured in two basic modes within a GIS: (i) vector and (ii) raster. In vector mode objects on a map are represented by points, lines or polygons with associated alphanumeric information in tabular form (Fig. 4). Points, besides being the structural units of lines and polygons can represent entities like harbours, fishing vessels or sampling locations. Lines represent entities such as isobaths or vessel tracks and polygons can be used to represent fishing grounds or marine protected areas. Raster was the first mode to appear in a GIS. It uses a much simple structure for the representation of objects as the whole map
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 77 Geographic layers
Import Export Publish Tabular data Create Edit Update
GIS
Query Display Custom
Report Report
Analyse
RDBMS
re sto
. Overlay . Extract …
Digital cartography
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 3 – Schematic representation of Geographic Information System capabilities with respective inputs and outputs. RDBMS – Relational Database Management System.
surface is composed by a grid of cells that forms a matrix with a given number of rows and columns (Meaden and Kapetsky, 1991). Rows and columns form cells (pixels in Remote Sensing) representing data elements, which have associated a digital or colour coding (Fig. 4). The size of the cells is often referred to as the resolution of the map. This mode is mostly used to represent remote sensing data such as sea surface temperature (SST) and Chlorophyll-a (Chl-a) concentration. Despite having more straightforward processing algorithms, raster is more prone to inaccuracies in the location and shape of objects to be represented. Data organization is an important aspect to consider prior to the development of a GIS, since the generation of the data structure model must encompass the type of the intended graphical representations. Other key issues that deserve attention when the construction of a GIS is concerned are: a) Data capture mechanisms (i.e., scanners, satellite, survey - using global positioning system, LIDAR); b) Cartographic projections (i.e., coordinate systems, datum); c) Creation of topology d) Data manipulation functions (i.e., generalization, extraction, aggregation, conversion), e) Data analysis functions (i.e., statistical analysis, classification, geostatistics, other interpolation techniques) and f) Display outputs (i.e., plotters, screen copy devices). GIS’s are by itself the expression of an interdisciplinary science, which rely on the knowledge of five academic disciplines (UNESCO, 1999): Geography, Computer Science, Mathematics, Statistics and Information Science. Together these branches of science form the pillars of an array of functionalities that enable spatially decision-making in numerous areas ranging from urban planning to oil exploration. In fact, a contemporary institution or business conducting serious work on spatially related data rely on the potential of GIS tools to improve its ability to solve knowledge discovery problems and provide customers with adequate location-based solutions.
78
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Figure 4. – Graphical representation of natural elements in vector and raster modes.
CASE STUDIES
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Visible Remote Sensing Mesoscale Features (Eddies, Filaments, Etc.) One of the most frequent applications of the ocean colour remote sensing is the analysis of phytoplankton concentrations. Since remote sensing covers large areas on a regular mode, it has proven to be a precious aid in the description of the spatial structure and temporal variability of the near surface phytoplankton concentrations in the Iberian region on a seasonal to inter-annual basis. For instance, summer coastal upwelling is clearly identified by the large pigment values over the continental shelf and upper slope. Less obvious is the continuous band of very high concentrations along most of the continental shelf seen in the ocean colour imagery during the winter, which almost vanishes during the transitions seasons (spring and autumn). Such enhanced primary production in winter may be attributed to the nutrient input from the river runoff (Peliz and Fiúza, 1999). The poleward current, a characteristic mesoscale feature that flows along the shelf break and upper slope off western Iberia from autumn till spring, is captured in the ocean colour imagery as a large intrusion of low pigment concentration progressing northward, limiting the pigment rich coastal waters over the shelf and the moderate pigment concentrations offshore. The frontal region between the rich costal waters and the oligotrophic offshore waters is populated during the summer by mesoscale features, such as filaments of upwelled water stretching seaward and detached eddies. The ocean colour images reveal the biological richness of these features, which increase the dispersion of the pigments into the relatively poor open ocean. The location of the filaments recurrently coincides with protrusions in the coastal morphology, such as prominent capes, and topographic features, such as submarine ridges. These well developed phytoplankton structures are generally related to moderate or
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 79 intense offshore transport, whereas the absence of filaments correspond to either weak offshore transport or coastal convergence (Sousa and Bricaud, 1992). Thus, filaments and jetlike features are found to be the main transport mechanism for the shelf-ocean exchange of phytoplankton and associated carbon fluxes. More recently attention has been drawn to the relatively high chlorophyll concentrations derived from ocean colour (SeaWiFS) images over the shelf and slope, stretching to large distances from the coast and associated with winter upwelling events. Such surface concentrations, high for wintertime (up to 3.5 mg m-3 against typical values of 4-5 mg m-3 in summer), are attributed to the retention of phytoplankton in a shallow buoyant layer of waters of inland origin, that spreads offshore under the influence of upwelling favourable winds (Ribeiro et al., 2005). An extreme winter event was identified through SeaWiFS derived chlorophyll concentration off southwest Iberia in February 2001 (Peliz et al., 2004). An extraordinary long filament transported coastal rich waters as far as 400 km offshore (Fig. 5). Costal fresh water river plumes, following a particularly intense period of rainfalls, provided buoyancy and nutrients for an extensive phytoplankton production that was dragged till far from the coast by the offshore eddy dynamics.
3 S Chl-a (mg/m )
38º N
SV
37º N
36º N
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
13º W
11º W
9º W
7º W
Figure 5 - SeaWiFS-derived chlorophyll-a distribution in SW Iberia in 12 February 2001. The southwestward filament is about 400 km long. S is Cape Sines and SV is Cape São Vincente (adapted from Santos et al., 2007).
Seawifs, the Western Iberian Buoyant Plume and the Survival of Sardine Larvae In the western Iberia, the transport/dispersal of sardine larvae is affected by factors such as the wind driven transport, the structure and circulation of the western Iberia Buoyant Plume-WIBP (Western Iberia Buoyant Plume), and the slope circulation associated with the Iberian Poleward Current (IPC) (Santos et al., 2004). The WIBP is a lens of water of 'low' salinity (< 35.8) fed by the winter discharges of several rivers of the NW coast of the Iberian Peninsula where phytoplankton productivity is enhanced by the availability of nutrients and stratification conditions (Ribeiro et al., 2005). For these reasons, the WIBP could be studied
80
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
using satellite-derived chlorophyll-a distributions (Ribeiro et al., 2005). The spatial patterns captured by in situ measurements during an oceanographic survey off Western Iberia and by SeaWiFS were very similar (Fig. 6). Furthermore, the locations of high chlorophyllconcentrations related with sub-surface maxima measured in situ with a fluorometer where also clearly seen in the SeaWiFS-derived distributions (Fig. 6). One of these maxima are related with a convergence zone located in the shelf break and it is interested to note that such frontal zone is not visible in thermal AVHRR imagery but present in SeaWiFS data for the same day. Thus, the use of ocean colour RS can be an important tool in the study of frontal structures of this nature and also for studying processes related to the variability of recruitment of sardine (Sardina pilchardus Walbaum, 1792) because the WIBP is a suitable environment for larval retention and survival (Santos et al., 2004; 2006a). The study of the evolution in time of the WIBP using sequential SeaWiFS images revealed that eastward frontal velocities could be of the order of 0.34 m s-1 (Ribeiro et al., 2005) and this is another interesting application of ocean colour RS that could be used to estimated the role of the WIBP in the drift of larval sardine during winter upwelling events off western Iberia (Santos et al., 2004).
9
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 6 - SeaWiFS-derived chlorophyll-a distribution of NW Iberia in 19 February 2000 (middle image). In the right plot is presented the in situ surface distribution of chlorophyll-a and in the left plot a cross-slope section obtained during an oceanographic survey contemporaneous to the satellite image. The white crosses in the satellite image indicate the location of the oceanographic stations and the red arrow the cross-slope section presented in the left plot. The shaded red areas in the left and right plots represent the locations of the maxima and other features seen in the satellite image (adapted from Ribeiro et al., 2005).
Infrared Remote Sensing Coastal Circulation Infrared imagery is the most popular remote sensing source used by coastal oceanographers. The conspicuous contrast of the sea surface temperature in most of the coastal regions, induced by the strong mesoscale activity, makes infrared sensors an ideal tool for monitoring and track mesoscale features there. In fact, the present recognition that the ocean circulation is dominated by meanders and eddies is itself a consequence of the satellite oceanography advent. Off western Iberia the first applications of infrared satellite imagery to oceanography came from the late seventies (Santos et al., 2006b). There, the coastal circulation is dominated
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 81 by mesoscale features, some of them associated to the upwelling regime that dominates the oceanography of the region during a substantial part of the year. The upwelling season is roughly defined from March to September (Wooster et al., 1976). The region is poorly sampled by in situ devices, in particular permanently moored instruments for long term observations and monitoring. Therefore, most of the present knowledge of the regional oceanography was built upon model simulations along with satellite imagery analysis, infrared imagery ahead. In situ observations have been limited to short periods (1 to 3 weeks) on board of research vessels, the majority dedicated to specific oceanographic features limited in space, rather than to the coastal circulation at the basin scale. Infrared satellite imagery, when transmitted to research vessels in almost real time, has proofed to be a powerful instrument for the guidance and optimization of many research cruises in the region. Some main results obtained via infrared remote sensing are summarized next. Box 1
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
What is upwelling? – The Ekman mechanism According to the Ekman theory the surface current lies 45º to the right (left) from the wind direction in the Northern (Southern) Hemisphere caused by the rotation of the Earth (Coriolis effect). However, that angle of deviation increase with depth but their amplitude decay exponentially until a depth where the frictional influence of the wind is null (Ekman depth) forming the so-called Ekman spiral in the layer directly influence by the frictional action of the wind (Ekman layer). The net movement of surface water in the integrated Ekman layer (Ekman transport) is about 90º to the right (left) of the wind direction in the Northern (Southern) Hemisphere. Thus, coastal upwelling occurs where Ekman transport moves surface waters away from the coast and is created by alongshore wind blowing across the ocean surface and pushing surface waters offshore (Ekman transport) perpendicular to the wind direction due to the Coriolis force. The displaced surface waters are replaced by upwelled cold and nutrient-rich subsurface waters that allowed the development of phytoplankton (primary producers) at the surface, which are the base of the food web. Thus, these ecosystems are some of the most productive of the world and maintained important populations of fish, marine mammals and seabirds. Examples of such systems are the four major Eastern Boundary Current Systems of the World Oceans: the Canary, California, Benguela and Humboldt Current systems. They account for about 50% of the total worldwide catches of marine species, while representing less than 3% of the ocean surface.
The thermal front that separates the coastal cold nutrient rich upwelled waters from the more oligotrophic offshore waters is much more complicated than a simple contorted border parallel to the coast. During the upwelling season the frontal region is populated with meanders and filaments. These are narrow contorted tongues of cooler upwelled water extending hundreds of kilometers seaward from the coastal zone. Filaments export a much larger mass along their principal axis than expected by the purely wind-driven Ekman circulation, being an important mechanism of exchange between coastal and open ocean waters, with obvious implications in the ecosystem functioning. The existence of such structures off western Iberia was first identified trough infrared remote sensing, and significant advances in the understanding of the filaments behaviour
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
82
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
were done, and still are, through remote sensing analysis. The number of filaments along the coast and the recurrent location of their roots in the major topographic features like prominent capes, their maximum extension and width, the development and decaying time scales, the frequency of occurrence and seasonal variability, all is presently known (Haynes et al., 1993). The perception that the offshore eddy field could drag and modulate the filaments offshore is also a result from infrared imagery (Peliz et al., 2004). The dynamics of the filament formation is closely related to the wind regime and possibly with the sea surface topography off the continental shelf, since offshore eddies result in sea level anomalies. The frequency of satellite passes (2 a day for NOAA-AVHRR) and the relatively sparse cloud cover over western Iberia during the upwelling season, make possible to follow the evolution of the coastal filaments and relate it with the forcing variables. Cape São Vicente, the culminating point of southwest Iberia, is the root of a major cold filament. Infrared imagery was fundamental for the definition of it recurrent pattern and a valuable tool in the understanding of the process of its growth and decay (Relvas and Barton, 2002). Recently, a research cruise guided by SST imagery transmitted on board in quasi real time, sampled several transects perpendicular to the filament axes. Thus, it was possible to infer the exchange between the nutrient rich coastal waters and the oligotrophic offshore waters promoted by the upwelling filament. The recent evolution of the remote sensing of the sea level anomalies and wind over the ocean trough microwave radars, along with the improvement in their accessibility, makes the study of filaments and other costal features highly remote sensing dependent in regions, like western Iberia, where the in situ sampling is poor. The upwelling regime has been extensively studied trough infrared imagery all over the world and Iberia is no exception. Remote sensing did allow the detection of narrow warm inshore countercurrents that flow poleward over the inner shelf, attached to the coast, whenever the upwelling favourable winds decay. Such countercurrents, about 10-20 km width, lay in a region traditionally not sampled because it is too close from the coast for oceanographers, but is too distant for coastal experts. They have been observed off the northern (Galicia) and central parts of the western Iberia (Sordo et al., 2001; Peliz et al., 2002), and off the southern Iberian coast (Relvas and Barton, 2002). Inner shelf countercurrents are also visible in other upwelling systems, such as the California system (Cudaback et al., 2005). Assuming that the temperature is a tracer of the near surface current, objective methods make possible the estimation of advective surface velocities from sequential infrared satellite images. Velocities of 0.1-0.3 m s-1 were estimated for the countercurrent off southwest Iberia, increasing when the continental shelf narrows. The occurrence of coastal countercurrents reduces the cross-shelf transport, characteristic of the upwelling regimes. This has ecological consequences by preventing the transport between the inner shelf and the outer shelf, so that alongshore dispersion and consequent retention of larval stages, phytoplankton and detritus prevail over the inner shelf. Satellite-derived sea surface temperature maps have been used to describe the upwelling patterns off western Iberia, owing to the clear thermal contrast between the cold, vertically mixed, upwelled waters typically found over the shelf, and the thermally stratified oceanic waters. Tacking advantage of this thermal contrast, automatic edge-detection methods have been developed to discern the location of such thermal fronts, which are known to be biologically actives. With the set up of satellite remote sensing observations of the ocean in a regular basis during the past decades, we can presently access to archives with relatively long
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 83
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
time series of images. The application of the front detection algorithms to this times series did permit the present knowledge of the recurrent location of upwelling fronts, with obvious benefit for fisheries research (Relvas et al., 2007). A long archive of satellite derived sea surface temperature is also a valuable data set for the generation of regional climatologies, useful for examining the ocean variability and climatic change assessments. The currently available time series of satellite data can adequately depict the known patterns of the climatology and seasonal variability of the western Iberia ecosystem and may be used for investigation of longer time-scale variability. The existing data set is already long enough to capture decadal changes that occurred in the ecosystem since mid 1980s. Analysis of the magnitude of the surface temperature gradient between coastal and offshore waters, an effective estimate of large-scale upwelling intensity, revealed a decadal scale shift of upwelling regime intensity from weak upwelling in 1980s to a stronger one in the 1990s. This regime shift may be associated to the abrupt change of the North Atlantic Oscilation (NAO) observed in the earlier 1990s (Santos et al., 2005).
Coastal Upwelling, Fronts, and Swordfish and Tuna Catch The widest application of satellite remote sensing to fisheries has been using infrared thermal data to detect frontal structures favourable for the aggregation of tuna (Laurs et al., 1984; Laurs, 1989), mainly because sea surface temperature is the oceanographic parameter that has been most successfully measured by satellite RS, is an indicator of important ocean processes (such as, coastal upwelling, fronts and eddies) and also because ocean colour and other sensors has not been available on an operational basis. Although the main mechanism of aggregation of tuna and tuna-like fish near ocean fronts are still in debate, it seems that behavioural mechanisms related with their feeding activity are the most accepted explanation (Fiedler and Bernard, 1987; Brill and Lutcavage, 2001). The fishing success of swordfish (Xiphias gladius Linnaeus, 1758), bigeye (Thunnus obesus Lowe, 1839) and albacore (Thunnus alalunga Bonnaterre, 1788) tuna off western Iberia in relation to frontal structures associated with the dynamics of coastal upwelling was studied by Santos et al. (2006b). Higher catches of swordfish were associated to the strong thermal fronts formed offshore between old upwelling waters and oceanic warmer waters during the relaxation of coastal upwelling. In Figure 7a; b there is not any thermal front in the fishing zone and therefore the fishing sets performed at that time led to low swordfish catches (0.7-1.8 swordfish/1000 hooks), however it is clearly seen in the northwest (upper left) corner of the satellite image of Fig. 7b that oceanic warmer waters were converging towards the coast corresponding to the beginning of the upwelling relaxation phase. After one week (Fig. 7c) a strong frontal structure developed in the fishing zone and separates old upwelled waters from open-ocean warmer waters advected shoreward meanwhile. This situation resulted in an increased catch about one order of magnitude greater (7.8 and 16.8 swd/1000 hooks) than one week before. At the same time, the highest catches of tuna observed during this study were located on the warm (offshore) side of the thermal fronts that constitute the edges of mushroom-like thermal structures (Fig. 8) that (often) constitute the extremity of upwelling filaments and are the surface manifestation of submesoscale anticyclonic/cyclonic eddy pairs.
84
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas (a)
(b)
(c)
1
Figure 7 – Longline fishing sets (white lines) off the Portuguese west coast superimposed on nearly contemporary NOAA-11 satellite-derived sea surface temperature (SST) distributions: (a) 16 August 1992 at 15:35 UTC; (b) 17 August 1992 at 15:23 UTC and (c) 23 August 1992 at 15:51 UTC. Swordfish (swd) catches per unit effort are: (1) 17-19 August 1992, 0.7-1.8 swd/1000 hooks; (2) 22 August 1992, 16.7 swd/1000 h.; and (3) 23 August 1992, 7.8 swd/1000 h. The SST scale is expressed in o C, with values increasing from purple to red; black represents clouds and land (adapted from Santos et al., 2006b).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Unfortunately, during this study (1990-1992) there is not any ocean colour sensor operational that allow any relationship with ocean productivity. However, Santos et al. (2006b) did not find any ‘preferred’ temperature range for these large pelagic fishes supporting the hypothesis of others that the reason why they aggregate near surface fronts are not determined by thermo-physiological mechanisms (Carey and Robison, 1981; Laurs, 1983; Podestá et al., 1993; Bigelow et al., 1999) but probably related to their behaviour in relation to feeding.
Figure 8 – Longline fishing set (white line) off the west coast of Portugal in 3 November 1992 superimposed on a NOAA-11 satellite-derived sea surface temperature (SST) distribution of November 2, 1992, at 05:18 UTC. Bigeye (bye) and albacore (alb) tuna catch per unit effort were 37.8 bye/1000 h. and 12.2 alb/1000 h., respectively. The SST scale is expressed in oC with values increasing from purple to red; black represents clouds and land (adapted from Santos et al., 2006b).
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 85
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Microwaves Remote Sensing Altimetry, Sea-Level Height, Surface Winds and Ocean Circulation The coastal transition zone off western Iberia, defined as the region where coastal waters interact with the open ocean, is populated by active eddies that play an important role on the cross-shelf transport, with obvious biogeochemical consequences. Recent research suggests that such eddies can drag the coastal waters offshore and modulate the thermal front that separates the cold upwelled waters over the shelf from the offshore waters. For instance, the offshore eddy field may play a fundamental role in the cold upwelling filaments formation and development (Sanchez, 2005). Most of the present knowledge of the interaction process between the offshore eddy field and the coastal waters was built upon the analysis of the ocean topography acquired through microwave radar altimeters on board of satellites. Although the large errors associated with the detection of sea level anomalies trough satellite altimetry over shallow waters, off the continental shelf, in deeper waters, eddies leave a clear anomaly imprint in the sea level field. The analysis of the positive and negative sea level anomalies (anticyclonic and cyclonic eddies, respectively), along with infrared SST and/or visible imagery, contribute to clarify the dynamical processes associated to the formation and development of coastal features, such as meanders and filaments. The cooperative analysis of satellite altimetry data and in situ measurements, such as lagrangian drifters or data from cruises or moorings, is a powerful procedure to describe the near surface ocean circulation. Altimetry, along with surface drifters, was successfully used to infer the surface circulation in the eastern North Atlantic (Martins et al., 2002). Is satellite remote sensing limited to the sampling of the ocean surface, or can it tell us some aspects of the subsurface structure? The Mediterranean outflow trough the Strait of Gibraltar, along the southern frontier of the Iberian Peninsula, offers a unique feature to test such hypothesis. The Mediterranean water forms a relatively warm and salty tongue that extends westward from the Iberian Peninsula into the North Atlantic. It equilibrium depth is centered at about 1000 meters, with few hundred meters of thickness, except for a less conspicuous shallow core that flows at 400-600 meters depth along the southern Iberia continental slope. The classical view of the spreading of the Mediterranean water in the Atlantic changed dramatically in the early nineties with the discovery of meddies. Initially, little effort was put into the satellite detection of meddies, because it was supposed that such structures would leave no signal at the surface. However, some studies start to associate surface features with the presence of the Mediterranean water and meddies (Stammer et al., 1991; Pingree and Le Cann, 1993). Nowadays there is a reasonable confidence that some surface eddy structures visible in sea surface temperature satellite imagery off western and southern Iberia represent the signal of underlying meddy structures. Also, the altimetry data reveal positive sea level anomalies, consistent with the clockwise meddy rotation, of about a tenth of meter directly above a meddy (Oliveira et al., 2000). Nevertheless, it is difficult to separate the surface signals of meddies from the intrinsic upper layer dynamics and the tracking of the meddy pathways through remote sensing methods is still distant. Also the dynamics of the propagation of the meddy signal to the surface is still unclear.
86
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas Box 2 Meddies
Meddies are clockwise rotating eddies of Mediterranean water, with typical diameters less than 100 km and about 200 meters of thickness, which are able to transport water of Mediterranean origin over thousands of kilometres with very little mixing. They are a key factor for the understanding of the salt and heat transport in the Atlantic.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
A differently designed active microwave sensor, with oblique viewing, conceived to sample the sea surface roughness and estimate the wind over the ocean, is transported on board the QuikScat satellite. The wind stress is the main forcing mechanism in the dynamics of upwelling systems, like western Iberia. Traditionally the coastal meteorological stations are the source of the wind data in the upwelling systems studies. Typically they measure reliable wind velocities every hour or three hours, but in sparse locations and on land, subject to orographic effects. Remote sensed wind data are trustworthy, they cover a large area over the ocean and have daily passes over the eastern North Atlantic. They represent a powerful tool in upwelling systems because they allow the computation of wind stress curl over the ocean and, within a certain extent, the evaluation of the wind field inhomogeneity. Off southwest Iberia remote sensing winds retrieved from the QuikScat microwave scatterometer, with 25 km resolution, were compared with the winds measured in land-based coastal stations. Results did show that coastal winds largely fail on representing the offshore wind field. The exception is the station located in the prominent and exposed Cape São Vicente, the southwest tip of the Iberian Peninsula. The wind measured in other stations, located along relatively straight coastline segments, is representative of the wind pattern over a limited area of the coastal ocean only (Sanchez et al., 2007a). The combined analysis of satellite winds and sea surface temperatures, through canonical correlation analysis, did reveal the influence of the wind pattern on the establishment of characteristic patterns of the sea surface temperature at several time scales. The spatial scales of the analysis vary from a vast region centered in the Iberian Peninsula (0º-20ºW; 30º-50ºN), till the local scale around Cape São Vicente (5.5º-10.5ºW; 35.5º-39.0ºN). Due to the satellite remote sensing nature of the data, that covers the studied area in a continuous way, it was possible to conclude about the relationship of the wind pattern, and consequently of the sea surface pattern, to the North Atlantic Oscillation (NAO) (Sanchez et al., 2007b).
SAR, Sea Surface Roughness, Submesoscale Phenomena And Primary Production Synthetic Aperture Radar (SAR) is an active microwave sensor designed to observe the sea surface roughness. It performs well in all weather conditions. The first applications of SAR images off the Iberian Peninsula come from the late seventies, during the short SEASAT mission. The sea surface roughness pattern shown by a SAR image covering the coastal region from 40º to 41º N was interpreted as the surface manifestation of a large quantity of internal wave trains, propagating towards the shore (Alpers, 1985). Internal waves were first sampled from space over the western Iberia shelf through colour photography taken by the US astronauts during the Apollo-Soyuz mission in 1975 (Apel, 1979).
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 87 Box 3 Internal waves
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Internal waves are waves that propagate within the body of a stratified fluid. During the summer, the surface layer of the ocean (30-50 m) can be up to 10ºC warmer than the water bellow. The pycnocline formed provide the ideal interface along which such waves can travel. Large scale internal waves with tidal periods are found on the Iberian shelf, as well as in many others shelf edge regions, and are referred as internal tides. Shorter period internal waves, including packets of internal “solitary” waves, are also frequently observed at the west coast of the Iberian Peninsula. Direct observational evidence of the propagation of internal waves off western Iberia with shipmounted thermistor chains reveal the appearance of thermocline depressions at different points of the shelf, with vertical displacements as large as 45 meters (Jeans and Sherwin, 2001). Internal tidal waves may increase primary production in the upper pycnocline by increasing the average light intensity experienced by phytoplankton near the pycnocline. Light intensity decreases exponentially with depth (Beer’s law). A neutrally buoyant or slowly sinking phytoplankton cell, undergoing vertical displacements by internal tides, is exposed to a greater average light intensity than that at its average depth. Assuming that near the pycnocline the photosynthesis is proportional to the total daily irradiance, because of the linear response to the dim light conditions near the euphotic zone, then the vertical motion of internal waves will introduce a significantly enhancement in primary production (Lande and Yentsch, 1988). Shortperiod internal waves are also important biological factors because of their impact on the development and transport of plankton. Many species are dependent on environmental cross-shore transport mechanisms to reach adult habitats, since they are not autonomous in their early stages and cannot control their cross-shore position. Non-linear internal waves may produce a net transport of in-water particles (phytoplankton, zooplankton and even small fish), which in the upper layer is usually in the same direction as the internal wave propagation (when the pycnocline displacements are of depression type, as it is the case off the western Iberian coast). Typical distances reached by such horizontal transport are of the order of several kilometres (Lamb, 1997). This mechanism is particularly effective when the transport by internal waves cooperates with the wind drift and plankton swimming (Shanks, 1995). Internal tidal bores have also been identified as an important mechanism of nutrient supply to the near-shore. Larval accumulation can occur at the leading edge of internal tidal bores, causing aggregation of organisms in slicks (Pineda, 1999). Some of these organisms have the ability to remain at the leading edge of the internal motion by swimming against the dowelling currents or because they are sufficiently buoyant, and when the disturbance propagates all the way to the shore, the larvae would be effectively transported onshore, reach the adult habitat, and have an opportunity to complete their life cycle.
A prompt detection of internal waves through remote sensing is important since they are believed to impact the primary production over the shelf, thus in coastal regions, as it will be discussed in the next paragraph. Satellite SAR high resolution images can also provide a functional manner to predict the location of internal waves. A good agreement between predicted and observed locations of internal wave packets was already achieved in an experiment carried out off the northern part of the west coast of the Iberian Peninsula (da Silva et al., 1997). Internal tidal waves may leave a distinct signature in ocean colour remotely sensed images too (da Silva et al., 2002). Bands of enhanced levels of near-surface chlorophyll in the central region of the Bay of Biscay were associated with the uplifting of a subsurface
88
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
chlorophyll maximum by passing internal tide waves travelling away from the shelf break. Those bands were 30-50 km width and could be detected by the SeaWiFS sensor, with 1 km resolution. Almost contemporaneous SAR overpasses, with resolution of 25 m, did confirm the assumption. Because of its large coverage, independence of the day-night cycle and all-weather capability, space-borne SAR has been proven to be a useful tool for detection of surface films in the ocean, in particular oil spill monitoring and tracking. Research following the Prestige accident off northwest Iberia did show that radar imagery along with other information, such as wind data or in situ observations, integrated into a Geographic Information System (GIS) database, provide a powerful instrument to study the spatial distribution and the evolution of the oil slicks (Palenzuela et al., 2006).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Identification of Fishing Trips and Grounds Using GIS Tools and Automated Satellite Positioning Data Fisheries tend to exhibit a spatial component and hardly cover the full extent of the target species distribution, operating only in specific areas where higher yields can be obtained (Bordalo-Machado, 2006). In this sense, it is of utmost importance to understand how is the behaviour of the fishing fleets in space and time to correctly evaluate the impact of the commercial exploitation on fish stocks. Fishing activity can be inferred from skipper logbooks and interviews but this information is most of the times, considered as anecdotal due to the biases associated with skippers’ declarations. The best independent source for commercial fisheries data presently available are the satellite automated positioning systems, commonly referred to as Vessel Monitoring Systems (VMS), which record vessels’ geographic position, time, speed and course in real time. Nevertheless, the interpretation of the information is far from being simple, as no indication exists whatsoever on the type of operation (e.g. navigation and fishing) being conducted by the vessel. Besides, the overall map of VMS data points usually reduces to a confusing and thick plot of data points yielding very little information (Bordalo-Machado and Figueiredo, 2007). In order to infer particular operations carried out by vessels at sea and estimate fishing effort, several authors had to develop software procedures to identify vessel’s operations at sea. This section presents two examples of GIS application that use VMS data to characterize fishing activity in Portuguese mainland waters. The Portuguese VMS entitled MONICAP, is based on Inmarsat-C satellite communications between the vessels and a control centre. It is composed by an individual continuous monitoring equipment - ‘Blue Box’, installed on board each vessel which transmits the vessel’s position, heading and speed of the vessel through a landing earth station, directly to the Fishery Control and Surveillance Centre (FCSC) of the Portuguese Directorate General for Fisheries. At the FCSC, extensive record datasets are produced for each vessel and stored in a relational database.
Trawl Fisheries A Portuguese crustacean trawl fleet operates off the southwest and south coasts of Portugal, between 200 and 750 metres. The target species of this important Portuguese fishery are the Norway lobster (Nephrops norvegicus Linnaeus, 1758), the Deepwater rose shrimp
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 89 (Parapenaeus longirostris Lucas, 1846) and the red shrimp (Aristeus antennatus Risso, 1816). During the EU Study Project 99/059 Geocrust, MONICAP data from 1998 and 1999 was used, for the first time, to estimate and map fishing effort of the south Portuguese crustacean trawlers (Afonso-Dias et al., 2002). In this project coordinated by the University of Algarve, GIS-based software was developed using Visual Basic 6.0 programming language. ADODB (Database Abstraction Library for PHP (Hypertext preprocessor) and Python) and MapObjects 2.0 Pro were used to communicate with a MSAccess 2000 database and the maps produced. The manipulation of the data was conducted in an oriented-purpose application with a set of menus to perform different tasks and a map having a set of themes (or layers) to display the results. The developed software, Geocrust 1.0, contained seven modules with the following objectives (Afonso-Dias et al., 2004):
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. Preliminary Analysis: to map and edit the original MONICAP data for a single vessel or group of vessels, for different periods of time; 2. Trawl Speed Analysis: to analyse the distribution of vessel speed; 3. Fishing Trips Definition (Fig. 9): to identify fishing trips performed by a vessel and determine their limits; 4. Trawl Hauls Definition: to identify the trawl hauls within each fishing trip and determine their limits; 5. Cartography: to map Fishing Effort and Landings per Unit Effort (LPUE) 6. Statistics: to analyse and extract data from the database for further use in other computer packages. This module was developed, essentially, to provide tools for exploratory data analysis; 7. Simulator: to recreate the activity of selected crustacean vessels during a selected period of time. LPUE estimation and mapping was made possible by combining VMS data with official fisheries landings data. The GIS application further allowed the visual inspection of MONICAP data, making it easier for the user to detect and correct the errors. Using Geocrust 1.0, it was possible to cover half of the total fishing trips performed by the Portuguese crustacean trawl fleet in 1998 and 1999. This corresponded to 80% of valid fishing trips (with trawl hauls totally defined). Another important outcome of this project included the detection of areas of overlapping fishing activities between vessels in the same fishing grounds by mapping fishing effort together with LPUE estimates. The great usefulness of the application is the combined on-screen display of the data, which includes a graph of vessel’s speed along time together with its geographical representation. More recently, new automated procedures that do not require the intervention of the user to classify and delimit fishing operations were added to a new version of the application (Geocrust 2.0), namely Artificial Neural Network routines.
Longline Fisheries In the Portuguese continental slope, a deep-water longline fleet has been fishing at depths greater than 800 m for more than 20 years. The fishery is conducted by a traditional, family type fishing community who strives to keep the fish catches profitable with little changes to
90
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
the fleet artisanal features. The main target species is black scabbardfish (Aphanopus carbo Lowe, 1839) with occasional by-catch of deep-water sharks.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 9 - GeoCrust 1.0 application – Module 3 (Fishing Trip Definition). A four days track for a selected vessel is represented. The correspondent speed graph is also showed. The red triangles represent landings (sales in the fish market); the green and yellow triangles represent the beginning and the end of one fishing trip, respectively. The beginning (red cross) of a new fishing trip is being defined (green form) (adapted from Afonso-Dias et al., 2004).
Aiming to characterize the fishing operations of this fleet, a database software package containing automatic classification routines was developed by the Portuguese National Fisheries and Sea Research Institute (IPIMAR) using a short MONICAP dataset from the activity of two vessels in the period 2000-2001 (Bordalo-Machado et al., 2007). The dataset comprised information on date/time, longitude/latitude coordinates (WGS 84) and vessel’s speed (in knots). Data points exhibited time lags of approximately 10 minutes. Data processing occurred within an Oracle 9i database where the main tasks were: a) categorize vessel movement into fish and non-fishing operations; b) identify vessel trips and fishing grounds and c) calculate fishing effort in number of hours. Prior to these tasks, individual VMS data records were initially assigned to a status attribute according to vessel’s geographic position and speed. Five types of status were defined: 10 – Motionless in the fishing harbour area; 20 – Moving inside the fishing harbour area; 40 – In transit over the open sea (not fishing); 41 – Hauling the gear (vessel speed: 0 – 1 knot); 42 – Setting the gear in the fishing ground (vessel speed: 2 - 5 knot). The results obtained by the software package were very promising considering that almost 90% of the fishing trips held by the two vessels in the study period were identified. However, using a new and enlarged VMS dataset (nine vessels) for the period 2000-2004, the set of procedures did not proved so well in the
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 91 identification of fishing trips and classification of fishing operations. The reason was the large temporal discontinuities showed in the new dataset. Instead of a regular recording interval of 10 minutes, some consecutive records exhibited a time difference of more than 45 minutes. To overcome the problems found, a new set of software procedures were developed by incorporating new rules to record classification and assignment to individual fishing trips (Bordalo-Machado and Figueiredo, 2007). As before, the programming language used was PL/SQL but this time Oracle SPATIAL tools were also employed to generate graphic representation of VMS data points and fishing areas that could later be visualized in ArcGIS trough an ArcSDE connection. The overall data processing workflow, showing the several packages containing software procedures is illustrated in Fig. 10.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 10 – General workflow of the longline VMS data processing (adapted from Bordalo-Machado and Figueiredo, 2007).
The new developed procedures enabled the identification of 2,642 fishing trips from a total of 3,456 reported daily landings. Package AGEO enabled the detection of potential fishing grounds of the vessels by visualization in ArcGIS (Fig. 11) and made possible the uncovering of wrong classifications of fishing operations by the package RECOVMS. In these situations, polygons exhibit unusual geometric shapes and area values (e.g., 0.5 sq km or 78 sq km). The software procedures were developed with enough flexibility to improve existent conditions of fishing trip identification and validation by the incorporation of better classification data mining rules. Nonetheless, this task should be accomplished by researchers with sufficient knowledge on the fishery under study to guarantee that the new rules are adequate to the modus operandi of the vessels at sea.
92
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Figure 11 - Example of fishing areas generated by package AGEO. Areas encircled represent possible fishing grounds of the fleet vessels. Spatial references are not included due to data confidentiality. (adapted from Bordalo-Machado and Figueiredo, 2007).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Demersal Fisheries Research Using GIS and Geostatistical Techniques To estimate fish distribution and biomass it is important to have samples from research surveys as they provide independent information on the number and weight of fish that exist on a specific area or period (Jardim and Ribeiro, 2007). Marine populations usually display aggregated and patchy rather than random distribution patterns. This means that the probability of obtaining high yields is greater near fish concentrations areas, reflecting an existent but usually unknown spatial structure. The class of methods that has been more used in the literature to assess the spatial structure of marine resource populations is Geostatistics. These methods provide a good alternative to classical statistical methods to deal with spatially auto-correlated data, since the error terms of the samples are not stochastically independent (Bordalo-Machado, 2002).
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 93
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Box 4 Geostatistical analysis The most common way to begin with a geostatistical analysis is by considering the hypothesis that sample values (in the present case, fish catches) correspond to individual realizations x1, x2, …, xn of random variables RV (x1), RV (x2), …, RV (xn). If the fish abundance or biomass is spatially autocorrelated then we can say that the RVs are part of a random function Z(x). The characteristics of the model can then be inferred from the sample data set, however, some stationarity conditions must be assumed first. For Z(x) to be stationary within the study area its multivariate cumulative distribution function must be invariant under translation, meaning that any two vectors of Z(x) have the same multivariate cumulative distribution function, no matter the translation vector considered (Goovaerts, 1997). For simplicity the decision of stationarity is usually limited to the first two moments of Z(x). In this sense, Z(x) is stationary of order two when E{Z(x)} exists and is invariant in the study area and the covariance C(h) exists and depends only on a given distance vector h. As a consequence, the difference in Z(x) values will only depend on the geographic distance (h) separating two locations x. Sometimes there is not a constant mean in the study area and Z(x) does not have a finite variance. In this case is sufficient to consider that the increments [Z(x) – Z(x+h)] are stationary of order two, the so-called intrinsic hypothesis (Goovaerts, 1997). More recently, Bez (2002) suggested the use of the transitive approach, which requires fewer conditions that the former hypothesis and is appropriate for data sets with numerous zeros and global estimation variance in random stratified samplings. Stationary is a property of the random function that is assumed by the researcher and cannot be proven or refuted from the data. Nevertheless, exploratory analysis of the experimental values may provide some clues about it (Bordalo-Machado, 2002). After the establishment of stationary conditions for the random function model, the structure of the spatial phenomenon in the study area is then investigated. During this process, the mean and covariance of Z(x) are inferred throughout the estimation of the existing spatial correlation function parameters. Two functions have been commonly used in the literature: the covariogram and the variogram. These functions are used to find differences in the spatial continuity of the variable of interest along several directions, giving an indication of how the variance changes between points as the distance between them increases (Petitgas et al., 2003). The determination of a global variogram or covariogram model is a necessary step in order to proceed to the estimation of the variable in unsampled locations of the study area. For estimation, a family of generalized least-squares regression algorithms, usually known as Kriging, is commonly used. Kriging algorithms are unbiased linear interpolators with minimum variance (Petitgas et al., 2003) and exact interpolators as data values are honoured at their locations (Goovaerts, 1997). Local estimation at a point x0 is obtained by the interpolation of spatial correlation function (e.g. variogram) values at available data points in the surrounding neighbourhood selected. To evaluate the local uncertainty of the variable of interest in the study area, it is commonly used the indicator approach. In this method, the model of uncertainty is considered to be conditional to the local information available (Goovaerts, 1997) and indicator variables can be defined for different thresholds according with ancillary information on the object of study. For instance, if we would like to know how many of our observation values lie below a threshold of 0.5 we would define our indicator variable (I0.5) as follows:
⎧1 < 0.5 I 0.5 = ⎨ . ⎩0 ≥ 0.5
94
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
When all the observations are transformed to either 0 or 1, Kriging is conducted to obtain interpolated surfaces with values in the range 0 – 1. In the areas where values are closer to 1 the uncertainty of our variable to present values above or higher a given threshold (depending on the definition) is very small. This analysis can be very useful to improve existent sampling programmes.
A geostatistical approach was followed by IPIMAR to map the spatial distribution of the biomass from four deep-water crustacean species: Norway lobster, Red shrimp, Giant Red shrimp (Aristaeomorpha foliacea Risso, 1827) and Scarlet shrimp (Aristaeopsis edwardsiana Johnson, 1867), captured in the south slope of mainland Portugal in the period 2000-2002. The yield calculated as kg/h was used as a biomass index. The local uncertainty of the biomass was evaluated by conducting an Indicator approach. Estimation results were evaluated by the cross-validation test (Goovaerts, 1997). Samples were obtained during IPIMAR trawl surveys carried out in the spring/summer seasons of 2000, 2001 and 2002, covering depths between 200 and 900 m. Geostatistical analysis was conducted in S-PLUS 2000 (module Spatial Stats) and mapping in ArcGIS 8.2. The biomass interpolated surfaces produced for the three years enabled the location of particular areas where major changes have occurred for each of the species and spot their main concentration areas, as observed for Red shrimp in 2002 (Fig. 12). Interpolated surfaces obtained with indicator kriging values, allowed the identification of a potential fishing closure area for Norway lobster, which do not overlap with areas where higher yields for the other three deep-water crustaceans are expected.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
A Bathymetric Digital Chart for Trawl Surveys One of the main factors that can improve the sampling of marine resources in research trawl surveys is the knowledge of the sea-floor topography. To effectively plan trawl surveys and optimize the available time it is of paramount importance to have information on the location of features such as accentuated relieves or deep-water canyons. Only then, it is possible to avoid risks associated with fishing gear damage or loss and prevent the need to conduct additional exploratory transects on the sampling sites. During the 1980s, IPIMAR has initiated a research programme to describe and map the main features of the deep-water slope where commercial important deep-water resources occur. In the scope of these studies, IPIMAR conducted a geostatistical study in 2001 to analyse the spatial variability of the seafloor depth in a slope area of about 665 sq km in the central Portuguese coast (Machado and Sousa, 2003). Sampling occurred during IPIMAR's deep-water surveys held at the continental slope in 1995, 1997, 2000 and 2001, covering depths between 390 and 2,800 m. The data analysed were obtained from nine sounding transects and 14 fishing hauls. The information collected included the geographic coordinates (longitude/ latitude WGS84), depth (m), date/hour and substrate type (e.g., sand, mud).
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 95
kg / h
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 12 - Estimates of Red shrimp biomass for the south region of mainland Portugal obtained by ordinary kriging for 2002. (adapted from Bordalo-Machado et al., 2004).
Seafloor depth was considered as the variable of interest and its spatial continuity analysed in four distinct directions: N-S, E-W, SW-NE and NW-SW. After obtaining the global variogram, which describes the main spatial features of the variable in the study area, three variants of Kriging algorithms were used to estimate seafloor depth in unknown locations and to produce a bathymetric digital chart: Ordinary Kriging (OK), Universal Kriging (UK) and Kriging of the Residuals (KR). To evaluate the uncertainty of the estimates, a sequential gaussian simulation (100 realizations) was conducted on a regular grid of 82,180 points at a 100 m distance. For more information on the Sequential Gaussian Simulation see Goovaerts (1997). Data processing and analysis were conducted in S-PLUS 2000 (module Spatial Stats) and GeoMS software. GIS Mapping was carried out in MapInfo 5.5. A geometric anisotrophy model based on the directions N-S and E-W was selected to describe the spatial variability of seafloor depth in the study area. OK yielded was chosen to generate the bathymetric digital chart because it presented low variability of the estimation error (Table 1) and requires less parameters to be implemented. Table 1. Estimation error descriptive statistics of the three interpolators used: OK – Ordinary Kriging; UK – Universal Kriging and KR – Kriging of the Residuals
Mean Median Std. Deviation
OK 1.711 1.447 109.537
UK 1.141 1.479 110.714
KR 1.260 1.466 110.660
96
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In the digital chart it is possible to observe the effects of the anisotropy model (Fig. 13). There is a direction of higher spatial continuity of the values along the N-S axis, while in the perpendicular direction, seafloor depth values show much more variability. Having the digital chart inside a GIS it was possible to perform data queries to select adequate areas for trawling at specific depth ranges, for instance, between 400 and 800 m. The uncertainty chart produced will be useful to plan future sounding transects to collect more data in the locations where the uncertainty of the seafloor depth estimates was higher.
Figure 13 - Bathymetric digital chart based on Ordinary Kriging seafloor depth estimates. Profundidade = Depth.
SIG-IPIMAR – A Public Access Application for the Visualization of Oceanographic and Fisheries Data The recognition of GIS capabilities to combine different data types in common analysis scenarios and the increase of fisheries georeferenced data in public institutions, has led IPIMAR to promote the development of a GIS-based application to integrate information with spatial reference produced during the course of fisheries research studies and with interest to the general public. The application, entitled SIG-IPIMAR, has been recently developed along
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 97 the period 2006 - 2007 and comprises a system that enables the exploration and analysis of the geographic information produced by IPIMAR, as well as its dissemination to the general public trough the Internet (Fig. 14). It uses ESRI technology (ArcGIS Server, ArcSDE) and it is based on an Oracle 9i database.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 14 - General layout of the application SIG-IPIMAR.
SIG-IPIMAR is structured in an internal and external component. The Internal Component (IC) enables access to the full spatio-temporal resolution of the data but is restricted to IPIMAR authorized users. The External Component (EC) allows the visualization and query of aggregated data by space and time, being directed towards a wide audience of users, namely fishermen, students, teachers or environmental officers. Although SI has been created with five major themes: Demersal surveys, Artisanal fisheries, Environment, Zooplankton and Tuna species, it can be further expanded to display more thematic data through the use of a backoffice to manage new and existent layers of information. A short resume of the existent themes is presented below: •
•
Demersal surveys: Mapping of different marine species’ catch data from IPIMAR research surveys held along mainland Portuguese waters. Biomass and abundance data can be visualized in an aggregated form (average, max./min., total catch by ICES rectangle – resolution 0.5º × 1º ) or by individual capture coordinates (only in IC). Existent filters allow users to select the data by year, month, survey ID (only IC) and haul (only IC). This information can be further overlaid on environmental data. Artisanal fisheries: Fleet data, including vessels’ technical characteristics and major species caught can be visualized by landing port. Information about individual port infrastructures is also presented. No manipulation of data is allowed in this theme.
98
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas •
•
•
Environment: Mapping of in situ and satellite remote sensing data from three environmental parameters (water temperature, chlorophyll-a concentration and salinity) in the Portuguese continental EEZ area. Depending on the data source (satellite and in situ), the information can be further filtered by year, month or depth. Zooplankton: Mapping of different marine species larvae catch data from IPIMAR research surveys held along mainland Portuguese waters. It has the same functionalities described in the Demersal surveys theme. Tuna species: Visualization of pre-processed maps of tuna species distribution and abundance in the NE Atlantic area. There are options to show them overlaid on several environmental parameters, such as SST, dissolved oxygen, distance to fronts among others. No manipulation of data is allowed in this theme.
In each of the themes it is possible to use common GIS tools to visualize or manipulate (except Artisanal fisheries and Tuna species themes) geographic information. These include zoom, pan, select, ruler, clean selection and identify tools. In addition, there is also export and print options available, which are restricted in the EC and enlarged in the IC. In addition, there is also export and print options available, which are restricted in the EC and enlarged in the IC. Export includes an option to save maps into ESRI Layer files (*.lyr), which allows users to deal with pre-processed maps in ArcGIS desktop software within different analysis frameworks
Query Fields (e.g., year, species)
GIS toolbar
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Brown circles represent Hake survey yields (no./h)
Chlorophyll-a scale
Species photo
Figure 15 - SIG-IPIMAR display of Demersal surveys theme showing an example of a map produced with European hake (Merluccius merluccius Linnaeus, 1758) catch data from IPIMAR's surveys overlaid on a satellite-derived distribution of chlorophyll-a concentration from AQUA-MODIS.
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 99 One of the main features offered by SIG-IPIMAR is the possibility to overlay different types of data in a common map display. This enables, for instance, the detection of associations between marine resources abundance and environmental data. In the case of Demersal surveys data it is possible to find relations between yields (in number/hour or weight/hour) and satellite-derived SST or Chl-a. In the example of Fig. 15, the great majority of European hake (Merluccius merluccius Linnaeus, 1758) catches lie in zones of intermediate concentration of Chl-a.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Inform@R Portal – A Virtual Centre to Integrate Marine Thematic Data from Several Research Institutions In the beginning of 2005, the Portuguese Government created a Task Group for the Extension of the Portuguese Continental Shelf (EMEPC), under the dependency of the Ministery for National Defence The main goal of EMEPC is the collection of the necessary scientific advice to prepare the submission documentation for the extension of the Portuguese Continental Shelf beyond 200 nautical miles (in accordance with the United Nations Convention on the Law of the Sea). The importance of the task group mission requires acquisition of high quality data, so a large investment in innovating R&D components was done in recent years, namely by the purchase of sophisticated multibeam echosounders, acoustic doppler current profilers and a Remote Operated Vehicle (ROV) to explore the sea basin to depths down to 6,000 m. Besides the strategic objective to reinforce Portugal's position in issues concerning ocean matters, the project also aims to integrate information and know-how from diverse areas such as hydrography, geology, geophysics and international law, and leave a legacy for future researchers to explore and benefit from the large area that corresponds to the Portuguese EEZ. As a part of the EMEPC activity, an integrated ocean data gathering, treatment and exploitation system entitled InforM@r was constructed. This system was built in accordance with the specifications of ISO 19100 regulations and uses internet services and ESRI GIS technology. While still in development, InforM@r is both used to publish EMEPC’s metadata information and analyse the acquired data necessary to the accomplishment of the continental shelf project. In the near future, it is hoped that this system can act as a virtual data centre comprising meta-information from various earth and marine science organisms and contribute to implement an effective monitoring and management of ocean resources. The InforM@r portal can be accessed through the following URL: http://www.emepc.gov.pt/ in/informar.html. Box 5 ISO 191xx Series The ISO 191xx series of geographic information standards are developed by the ISO Technical Committee 211 and establish a structured set of standards for information concerning objects or phenomena that are directly or indirectly associated with a location relative to the Earth. These standards specifies methods, tools and services for management of geographic information, including the definition, acquisition, analysis, access, presentation, and transfer of such data between different users, systems and locations.
100
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
SUMMARY Ocean remote sensing is already operational for about 30 years, namely regarding sea surface temperature, and was because of their advance that we have today a complete different view of the ocean. The ocean is not anymore a smooth environment but rather populated by an amazing number of features like filaments, eddies and meanders. Nowadays it is well established as an important and irreplaceable tool, for evaluating the oceanographic conditions, and foremost for monitor changes of the ocean background. Currently existing variables measured or estimated by remote sensing are as vast as sea surface temperature, phytoplankton biomass, wave height and direction, surface currents, surface winds, suspended sediments, etc. Its practical use in oceanography and fisheries is nowadays a question of building the correct products for each application and depended largely in the advances of new computer facilities that allowed the processing, analyse and disseminate of vast amount of data in a global scale. Fortunately the advances on Geographic Information Science have enabled applications to move one step forward, acting not only as repositories for collection and display of data but with capabilities to perform complex simulations and modelling of oceanographic phenomena. The growing number of satellites and satellite passes did allow the construction of derived products, merging data from different sources, including in situ data when available. Such products are generally designed to meet specific objectives. For instance, satellite-borne infrared sensors have an inability to cooperate with the cloud cover. However, the sea surface temperature product retrieved from blending infrared and microwave imagery largely decrease the cloud contamination, although it losses spatial resolution. Fisheries and oceanography research can largely benefit from these “tailor made” products. Although these techniques could be used to make fisheries operations more efficient, they should not be used in the perspective of improving fishing effort if there is the danger of overfishing but could be an important tool to help management decisions to adjust fishing effort to the conditions of the fishing grounds and/or of the stocks, assuring a sustainable exploration of the living marine resources. With the set up of satellite remote sensing observations of the ocean surface in a regular basis in the end of the 1970s, we can now access to archives with imagery time series of about thirty years long or more. Such archives, mainly with sea surface temperature and chlorophyll imagery, are long enough to detect variations of the surface oceanographic patterns and investigate their relation with environmental changes or regime shifts observed in fisheries. Satellite remote sensing technology is an inexpensive tool, but it will be costeffective if we close the gap between the raw remote sensing data collected and the information needed by users. A critical issue in producing valuable information from satellite remote sensing is processing. We can consider that it involves two steps - pre-processing and conversion of data into information. The first step consists in transform the raw data into accurately calibrated measurements of georeferenced physical parameters. This process has known a rapid development and the results obtained are robust. The second step is highly dependent on the information requirements of users. Two groups can be envisaged. One group uses the satellite remote sensing information as inputs in other applications, such as data assimilation schemes in numerical models. It is impossible to conceive modern operational modelling products, like ocean forecasts, without the
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 101 incorporation of satellite remote sensing data. Another group requires accuracy and timeliness of the data relevant for specific tasks and decisions. These applicants, seen as “end users”, look for easy and quick ways to display and interpret the data and have little knowledge of remote sensing technology or how it is employed to derive information. A large variety of products derived from remote sensing data are presently available on a world wide basis for a broad range of applications. Reliable products with a high degree of processing, merging data from different satellites and origins including in situ measurements from observation networks and some analysis already performed, are actually available at near real-time and easily accessible. This makes satellite remote sensing a fundamental tool for research in oceanography, fisheries and in all branches of marine sciences. Today it is widely accepted the need of an ecosystem-based approach to the management of living marine resources to allow the appropriate balance between conservation and use of biodiversity. Despite its implementation is still limited and research means are commonly insufficient, most modern GIS applications can already handle complex spatial structures with the ability to vary their relative positions and values over time. In this context the use of both remote sensing and GIS tools to evaluate ocean conditions, variations in ocean biology and integrate large data sets from different sources (e.g., in situ, remote sensing, socio-economics) at relatively low costs are of major importance for the future management of living marine resources. The number of future satellite missions envisages already and the possibility for data merging from different satellites and sensors will allow the development of a Global Ocean Observing System that is feasible, sensible, cost-effective and necessary for the correct management of ocean resources. Finally, the recognition that GIS technology should be adapted to a 3-D and even 4-D (time dimension) framework suitable for the study of the ocean and that development are already underway with success are promising directions into the future to develop a powerful tool for integrating multi-disciplinary information in ocean management and conservation.
ACKNOWLEDGMENTS
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
This study was partially funded by FEDER and the Portuguese Government under the SIGAP Project (22-05-01-FDR-00013). PBM was supported by scholarship FCTSFRH/BD/16037/2004. This is a contribution to the European Network of Excellency EUROCEANS (EC FP6 NoE 511106).
REFERENCES Afonso-Dias, M.A., Simões, J.M., Pinto, C., and Sousa, P. (2002). Use of Satellite GPS data to map effort and landings of the Portuguese crustacean fleet. In: Final report of the E.U. study project 99/059 GEOCRUST. pp 106. Afonso-Dias, M., Simões, J., and Pinto, C. (2004). A dedicated GIS to estimate and map fishing effort and landings for the Portuguese crustacean trawl fleet. In: Proceedings of the Second International Symposium on GIS/Spatial Analyses in Fishery and Aquatic
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
102
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Sciences (Eds. Nishida, T., Kailola, P.J. and Hollingworth, C.E. Fishery) GIS research Group, Saitama, Japan, pp. 323-340. Aguirre, M., Berruti, B., Bezy, J.-L., Drinkwater, M., Heliere, F., Klein, U., Mavrocordatos, C., Silvestrin, P., Greco, B., and Benveniste, J. (2007). Sentinel-3. The ocean and medium-resolution land mission for GMES operational services. ESA Bulletin 131: 2529. Atterna, E., Bargellini, Edwards, P., Levrini, G., Lokas, S., Moeller, L., Rosich-Tell, B., Secchi, P., Torres, R., Davidson, M., and Snoeij, P. (2007). Sentinel-1. The radar mission for GMES operational land and sea services. ESA Bulletin 131: 11-17. Alpers, W. (1985). Theory of radar imaging of internal waves. Nature 314: 245-247. Apel, J.R. (1979). Observations of internal wave surface signatures in ASTP photographs. Apollo-Soyuz Test Project, Summary Sci. Rep., NASA SP-412, 505-509. Bez, N. (2002). Global fish abundance estimation from regular sampling: the geostatistical transitive method. Canadian Journal of Fisheries and Aquatic Sciences 59: 1921-1931. Bigelow, K.A., Boggs, C.H., and He, X. (1999). Environmental effects on swordfish and blue shark catch rates in the US North Pacific longline fishery. Fisheries Oceanography 8(3): 178-198. Bordalo-Machado, P. (2002). Geostatistical analysis of seafloor depth in the Portuguese continental slope off Figueira da Foz – its application in GIS. MSc. Thesis in Geographic Information Systems, Technical Superior Institute, Technical University of Lisbon, pp. 79. (in Portuguese) Bordalo-Machado, P. (2006). Fishing effort analysis and its potential to evaluate stock size. Reviews in Fisheries Science 14: 431-455. Bordalo-Machado, P., and Figueiredo, I. (2007). Extraction and classification of longline fishing trips from vessel monitoring systems data with sequential recording gaps. In: The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science 34(XXX) (Ed. Stein, A.), ISPRS working group II-7, 5th International Symposium on Spatial Data Quality, Enschede, pp. 114–118. Bordalo-Machado, P., Figueiredo, I., Silva, C., and Leotte, F. (2004). Spatial and temporal changes in the distribution and abundance of four deep-water crustacean species in the southwestern and southern slopes of mainland Portugal. In: Proceedings of the 4th Assembleia Luso-Espanhola de Geodesia e Geofísica – SIG session, Fig. da Foz, Portugal, pp. 43-44. Bordalo-Machado, P., Sousa, R.T., and Matos, J. L. (2007). A software toolkit to analyse Vessel Monitoring System data from longline fisheries. In: Proceedings of the Third International Symposium on GIS/Spatial Analyses in Fishery and Aquatic Sciences, Shanghai, 2005. Ed. Nishida, T., Kailola, P.J. and Hollingworth, C.E. Fishery GIS Research Group, Saitama, Japan, pp. 259-274. Borne, G.H., Dunne, J., A. and Lame, D. B. (1979). SEASAT mission overview. Science 204: 1405-1406 Breman, J., Ed. (2002). Marine Geography. GIS for the Oceans and Seas. ESRI press, Redlands, USA. pp. 204. (ISBN-1-58948-045-7). Brill, R.W., and Lutcavage, M.E. (2001). Understanding environmental influences on movements and depth distributions of tunas and billfishes can significantly improve population assessments. American Fisheries Society Symposium 25: 179-198.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 103 Burrough, P. A., and McDonnell, R. A. (1998). Principles of Geographic Information Systems, 2nd Ed., Clarendon Press. pp. 346 (ISBN 0-19-854592-4). Carey, F.G., and Robison, B.H. (1981). Daily patterns in the activities of swordfish, Xiphias gladius, observed by acoustic telemetry. Fishery Bulletin US 79(2): 277-292. Cheney, R.E. (2001). Satellite altimetry. In: Encyclopedia of ocean sciences (Vol. 5). Ed. Steele, J.H., Turekian, K.K. and Thorpe, S.A., Academic Press, San Diego, USA. pp. 2504-2510. Chorley, R. (1987). Handling Geographic Information. Report of the Committee of Enquiry. DoE, HMSO, London. Clarke, K. C. (1995). Analytical and Computer Cartography, 2nd Edition Englewood Cliffs, New Jersey. Prentice-Hall. pp. 288 (ISBN-10: 0130334812). Cooke, J.G. (1985). On the relationship between catch per unit effort and whale abundance. Reports of the International Whaling Commission 35: 511-519. Cowen, R.K., Gawarkiewicz, G., Pineda, J., Thorrold, S.R., and Werner, F.E. (2007). Population connectivity in marine systems: an overview. Oceanography 20(3): 14-21. Crowder, L.B., Osherenko, G., Young, O.R., Airamé, S., Norse, E.A., Baron, N., Day, J.C., Douvere, F., Ehler, C.N., Halpern, B.S., Langdon, S.J., McLeod, K.L., Ogden, J.C., Peach, R.E., Rosenberg, A.A., and Wilson, J.A. (2006). Resolving mismatches in U.S. ocean governance. Science 313: 617-618. Cudaback, C.N., Washburn, L., and Dever, E. (2005). Subtidal inner-shelf circulation near Point Conception, California. Journal of Geophysical Research 110, C10007, doi:10.1029/2004JC002608. da Silva, J.C.B., Jeans, D.R.G., Robinson, I.S., and Sherwin, T.J. (1997). The application of near real-time ERS-1 SAR data to the prediction of the location of internal waves at sea. International Journal of Remote Sensing 18(16): 3507-3517. da Silva, J.C.B., New, A.L., Srokosz, M.A., and Smyth, T.J. (2002). On the observability of internal tidal waves in remotely-sensed ocean colour data. Geophysical Research Letters 29(12), 10.1029/2001GL013888. Dickey, T., Lewis, M., and Chang, G. (2006). Optical oceanography: Recent advances and future directions using global remote sensing and in situ observations. Reviews of Geophysics 44, RG1001, doi:10.1029/2003RG000148. Emery, W.J., Wick, G.A., and Schluessel, P. (1995). Skin and bulk sea surface temperatures: Satellite measurement and corrections. In: Oceanographic applications of remote sensing. Ed. Ikeda, M. and Dobson, F.W., CRC Press, Boca Raton, USA. pp. 145-165 Fedorov, K.N., and Sklyarov, V.E. (1981). Space oceanography: Hopes and realities. In: Oceanography from space. Ed. Gower, J.F.R., Plenum Press, New York, USA. pp. 35-47. Fiedler, P. C., and Bernard, H. J. (1987). Tuna aggregation and feeding near fronts observed in satellite imagery. Continental Shelf Research 7(8): 871-881. Fu, L.-L. (2001). Ocean circulation and variability from satellite altimetry (Chapter 3.3). In: Ocean circulation and climate. Observing and modelling the global ocean. Ed. Siedler, G., Church, J. and Gould, J., Academic Press, San Diego, USA. pp. 141-172. Gaile, G. L., and Willmott, C. J. (1989) Geography in America. Columbus: Merrill Publishing Co. pp. 840 Goodchild, M.F. (1992) Geographical information science. International Journal of Geographical Information Systems 6(1): 31-45.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
104
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Goodchild, M.F. (1997). What is Geographic Information Science?, NCGIA Core Curriculum in GIScience, http://www.ncgia.ucsb.edu/giscc/units/u002/u002.html, posted October 7, 1997 Goovaerts, P. (1997). Geostatistics for Natural Resources Evaluation. Oxford University Press. pp. 483 (ISBN 0-19-511538-4). Halpern, D., Ed. (2000). Satellites, oceanography and society. Elsevier Oceanography Series, 63, Elsevier, Amsterdam, The Netherlands. pp. 367. (ISBN-0-444-50501-6). Haynes, R., Barton, E., and Pilling, I. (1993). Development, persistence, and variability of upwelling filaments off the Atlantic coast of the Iberian Peninsula. Journal of Geophysical Research 98(C12): 22681–22692. Heimbach, P., and Hasselmann, K. (2000). Development and application of satellite retrievals of ocean wave spectra. In: Satellites, oceanography and society. Ed. Halpern, D., Elsevier Oceanography Series, 63, Elsevier, Amsterdam, The Netherlands. pp. 5-33. Hutchings, J.A. (1996). Spatial and temporal variation in the density of northern cod and review of hypotheses for the stock's collapse. Canadian Journal of Fisheries and Aquatic Sciences 53(5): 943-962. Ikeda, M., and Dobson, F.W., Eds. (1995). Oceanographic applications of remote sensing. CRC Press, Boca Raton, USA. pp. 492. (ISBN-0-8493-4525-1). IOCCG (2007). Ocean-Colour Data Merging. Gregg, W. (ed.), Reports of the International Ocean-Colour Coordinating Group, No. 6, IOCCG, Dartmouth, Canada. pp. 68 Jardim, E., and Ribeiro, P.J. Jr. (2007). Geostatistical assessment of sampling designs for Portuguese bottom trawl surveys. Fisheries Research 85: 239–247. Jeans, D.R.G., and Sherwin, T.J. (2001). The variability of strongly non-linear solitary internal waves observed during an upwelling season on the Portuguese Shelf. Continental Shelf Research, 21: 1855-1878. Kilpatrick, K.A., Podestá, G.P., and Evans, R. (2001). Overview of the NOAA/NASA advanced very high resolution radiometer Pathfinder algorithm for sea surface temperature and associated matchup database. Journal of Geophysical Research 106(C5): 9179-9197. Lagerloef, G. (2000). Recent progress toward satellite measurements of the global sea surface salinity field. In: Satellites, oceanography and society. Ed. Halpern, D., Elsevier Oceanography Series, 63, Elsevier, Amsterdam, The Netherlands. pp. 309-319. Lagerloef, G. (2001). Satellite measurements of salinity. In: Encyclopedia of ocean sciences (Vol. 5). Ed. Steele, J.H., Turekian, K.K. and Thorpe, S.A., Academic Press, San Diego, USA. pp. 2511-2516. Lagerloef, G., and Schnitt, R. (2006). Role of ocean salinity in climate and near-future satellite measurements. Eos Transactions of the American Geophysical Union 87(43): 466. Lamb, K.G. (1997). Particle transport by nonbreaking, solitary internal waves. Journal of Geophysical Research 102(C8): 18641-18660. Lande, R.. and Yentsch, C.S. (1988). Internal waves, primary production and the compensation depth of marine phytoplankton. Journal of Plankton Research 10(3): 565571. Laurs, R.M. (1983). The North Pacific albacore - an important visitor to California Current waters. CalCOFI Reports 24: 99-106.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 105 Laurs, R.M. (1989). Applications de la télédétection satellitaire du germon du Pacifique Nord, Thunnus alalunga (Bonnaterre). In: Télédétection satellitaire et Pêcheries Thonières. Ed. Le Gall, J.-Y., FAO Document Technique sur les Pêches, No. 302, FAO, Rome, Italy. pp. 87-97. Laurs, R.M., Fiedler, P.C., and Montgomery, D. R. (1984). Albacore tuna catch distribution relative to environmental features observed from satellites. Deep-Sea Research 31(9A): 1085-1099. Le Vine, D.M., Zaitzeff, J.B., D'Sa, E.J., Miller, J.L., Swift, C., and Goodberlet, M. (2000). Sea surface salinity: Towards an operational remote-sensing system. In: Satellites, oceanography and society. Ed. Halpern, D., Elsevier Oceanography Series, 63, Elsevier, Amsterdam, The Netherlands. pp. 321-335. Liu, A.K., and Wu, S.Y. (2001). Satellite remote sensing SAR. In: Encyclopedia of ocean sciences (Vol. 5). Ed. Steele, J.H., Turekian, K.K. and Thorpe, S.A., Academic Press, San Diego, USA. pp. 2563-2573. Liu, W.T., and Katsaros, K.B. (2001). Air-sea fluxes from satellite data (Chapter 3.4). In: Ocean circulation and climate. Observing and modelling the global ocean. Ed. Siedler, G., Church, J. and Gould, J., Academic Press, San Diego, USA. pp. 173-180. Machado, P.B., and Sousa, A.J. (2003). A geoestatística como ferramenta de estudo da batimetria na vertente continental portuguesa. Número temático de Cartografia e SIG, FINISTERRA, Revista Portuguesa de Geografia. Vol. XXXVIII, 76: 95-107 (in Portuguese). Manley, T.O., and Tallet, J.A. (1990). Volumetric visualization: An effective use of GIS technology in the field of oceanography. Oceanography, 3(1): 23-29. Martins, C.S., Hamann, M., and Fiúza, A.F.G. (2002). Surface circulation in the eastern North Atlantic, from drifters and altimetry. Journal of Geophysical Research 107(C12), 3217, doi: 10.1029/2000JC000345. McClain, C. (2001). Ocean color from satellites. In: Encyclopedia of ocean sciences (Vol. 4). Ed. Steele, J.H., Turekian, K.K. and Thorpe, S.A., Academic Press, San Diego, USA. pp. 1946-1959. Meaden, G.J. (1987). Where Should Trout Farms be in Britain? Fish farmer 10(2): 33-35. Meaden, G.J. and Do Chi, T. (1996). Geographical Information Systems – Application to Marine fisheries. FAO Fisheries Technical Paper, No 356. FAO, Rome. pp. 335. Meaden, G.J., Kapetsky, J.M. (1991). Geographical information systems and remote sensing in inland fisheries and aquaculture. FAO Fisheries Technical Paper, No. 318, FAO, Rome, Italy. pp. 262. Merchant, C.J., Llewellyn-Jones, D., Saunders, R.W., Rayner, N.A., Kent, E.C., Old, C.P., Berry, D., Birks, A.R., Blackmore, T., Corlett, G.K., Embury, O., Jay, V.L., Kennedy, J., Mutlow, C.T., Nightingale, T.J., O'Carroll, A.G., Pritchard, M.J., Remedios, J.J., and Tett, S. (2007). Deriving a sea surface temperature record suitable for climate change research from the along-track scanning radiometers. Advances in Space Research doi:10.1016/j.asr.2007.07.041. Minnett, P.J. (1995a). Sea surface temperature from the Along-Track Scanning Radiometer. In: Oceanographic applications of remote sensing. Ed. Ikeda, M. and Dobson, F.W., CRC Press, Boca Raton, USA. pp. 131-143.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
106
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Minnett, P.J. (1995b). The Along-Track Scanning Radiometer. In: Oceanographic applications of remote sensing. Ed. Ikeda, M. and Dobson, F.W., CRC Press, Boca Raton, USA. pp. 461-472. Minnett, P.J. (2001). Satellite remote sensing of sea surface temperatures. In: Encyclopedia of ocean sciences (Vol. 5). Ed. Steele, J.H., Turekian, K.K. and Thorpe, S.A., Academic Press, San Diego, USA. pp. 2552-2563. Munk, W. (2000). Oceanography before, and after, the advent of satellites. In: Satellites, oceanography and society. Ed. Halpern, D., Elsevier Oceanography Series, 63, Elsevier, Amsterdam, The Netherlands. pp. 1-4. Nihoul, J.C, Strub, P.T. and LaViollete, P.E. (1998). Remote sensing (Chapter 12). In: The Sea. The global coastal ocean. Processes and methods. Ed. Brink, K.H. and Robinson, A.R., John Wiley and Sons. Inc., New York, USA. pp. 329-357. NRC (2001). Marine protected areas: tools for sustaining ocean ecosystem. Committee on the Evaluation, Design, and Monitoring of Marine Reserves and Protected Areas in the United States, Ocean Studies Board, Commission on Geosciences, Environment, and Resources, National Research Council. National Academy Press, Washington, D.C., USA. pp. 272. (ISBN-0-309-07286-7). Oliveira, P.B., Serra, N., Fiúza, A.F.G., and Ambar, I. (2000). A study of meddies using simultaneous in situ and satellite observations. In: Satellites, oceanography and society. Ed. Halpern, D., Elsevier Oceanography Series, 63, Elsevier, Amsterdam, The Netherlands. pp. 125-148. Palenzuela, J.M.T., Vilas, L.G., and Cuadrado, M.S. (2006). Use of ASAR images to study the evolution of the Prestige oil spill off the Galician coast. International Journal of Remote Sensing 27(9-10): 1931-1950. Parkinson, C.L. (2001). Satellite passive microwave measurements of sea ice. In: Encyclopedia of ocean sciences (Vol. 5). Ed. Steele, J.H., Turekian, K.K. and Thorpe, S.A., Academic Press, San Diego, USA. pp. 2531-2539. Peixoto, J.P., and Oort, H. (1992). Physics of climate. American Institute of Physics, New York, USA. pp. 520. (ISBN-0-88318-712-4). Peliz, A., and Fiúza, A.F.G. (1999). Temporal and spatial variability of CZCS-derived phytoplankton pigment concentrations off the western Iberia Peninsula. International Journal of Remote Sensing 20(7): 1363-1403. Peliz, A., Rosa, T.L., Santos, A.M.P., and Pissarra, J.L. (2002). Fronts, jets, and counterflows in the Western Iberian upwelling system. Journal of Marine Systems 35: 61-77. Peliz, A., Santos, A.M.P., Oliveira, P., and Dubert, J. (2004). Extreme cross-shelf transport induced by eddy interactions southwest of Iberia in winter 2001. Geophysical Research Letters 31, L08301, doi:10.1029/2004GL019618. Petitgas, P., Poulard, J.C., and Biseau, A. (2003). Comparing commercial and research survey catch per unit of effort: megrim in the Celtic Sea. ICES Journal of Marine Science 60: 66–76. Pettersson, L.H., Samuel, P., and Drottning, Å. (1995). Radar altimeter sensors and methods. In: Oceanographic applications of remote sensing. Ed. Ikeda, M. and Dobson, F.W., CRC Press, Boca Raton, USA. pp. 427-441. Pineda, J. (1999). Circulation and larvae distribution in internal tidal bore warm fronts. Limnology and Oceanography 44(6): 1400-1414.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 107 Pingree, R.D., and Le Cann, B. (1993). A shallow meddy (a smeddy) from the secondary Mediterranean salinity maximum. Journal of Geophysical Research 98(C11): 2016920185. Plant, W.J. (2001). Satellite remote sensing microwave scatterometers. In: Encyclopedia of ocean sciences (Vol. 5). Ed. Steele, J.H., Turekian, K.K. and Thorpe, S.A., Academic Press, San Diego, USA. pp. 2539-2551. Platt, T., Sathyendranath, S., and Fuentes-Yaco, C. (2007). Biological oceanography and fisheries management: perspective after 10 years. ICES Journal of Marine Science 64(5): 863-869. Podestá, G.P., Browder, J.A., and Hoey, J.J. (1993). Exploring the association between swordfish catch rates and thermal fronts on U. S. longline grounds in the western North Atlantic. Continental Shelf Research 13(2-3): 253-277. Raheja, N. (2003). GIS-based Software Applications for Environmental Risk Management. In: Map India Conference 2003. pp. 9. Relvas, P., and Barton, E. D. (2002). Mesoscale patterns in the Cape São Vicente (Iberian Peninsula) upwelling region. Journal of Geophysical Research 107(C10), 3164, doi:10.1029/2000JC000456. Relvas, P., Barton, E.D., Dubert, J., Oliveira, P.B., Peliz, A., da Silva, J.C.B., and Santos, A.M.P. (2007). Physical Oceanography of the Western Iberia, Ecosystem: latest views and challenges. Progress in Oceanography 74(2–3): 149–173. Ribeiro, A.C., Peliz, A., and Santos, A.M.P. (2005). A study of the response of chlorophyll-a biomass to a winter upwelling event off Western Iberia using SeaWiFS and in situ data. Journal of Marine Systems 53: 87-107. Sánchez, R. F. (2005). The regional oceanography off Cape São Vicente: from the large-scale to the upwelling filament. Ph.D. thesis. Universidade do Algarve, Faro, Portugal. Sánchez, R.F., Relvas P., and Pires, H.O. (2007a). Comparisons of ocean scatterometer and anemometer winds off the southwestern Iberian Peninsula. Continental Shelf Research 27: 155-175, doi:10.1016/j.csr.2006.09.007. Sánchez, R.F., Relvas, P., and Delgado, M. (2007b). Coupled ocean wind and sea surface temperature patterns off the western Iberian Peninsula. Journal of Marine Systems, 68: 103-127, doi:10.1016/j.jmarsys.2006.11.003. Santos, A.M.P. (2000). Fisheries oceanography using satellite and airborne remote sensing methods: a review. Fisheries Research, 49(1): 1-20. Santos, A.M.P., Chícharo, A., Dos Santos, A., Moita, T., Oliveira, P.B., Peliz; A., and Ré, P. (2007). Physical-biological interactions in the life history of small pelagic fish in the Western Iberia Upwelling Ecosystem. Progress in Oceanography 74(2-3): 192-209, doi:10.1016/j.pocean.2007.04.008. Santos, A.M.P., Fiúza, A.F.G., and Laurs, R.M. (2006b). Influence of SST on catches of swordfish and tuna in the Portuguese domestic longline fishery. International Journal of Remote Sensing 27(15): 3131-3152. Santos, A.M.P., Kazmin, A.S., and Peliz, A. (2005). Decadal changes in the Canary upwelling system as revealed by satellite observations: their impact on productivity. Journal of Marine Research 63: 359-379. Santos, A.M.P., Peliz, A., Dubert, J., Oliveira, P.B., Angelico, M.M., and Ré, P. (2004). Impact of a winter upwelling event on the distribution and transport of sardine eggs and
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
108
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
larvae off western Iberia: a retention mechanism. Continental Shelf Research 24: 149165. Santos, A.M.P., Ré, P., Dos Santos, A., and Peliz, A. (2006a). Vertical distribution of the European sardine (Sardina pilchardus) larvae and its implications for their survival. Journal of Plankton Research 28(3): 1-10. Shanks, A.L. (1995). Oriented swimming by megalopae of several eastern North Pacific crab species and its potential role in their onshore migration. Journal of Experimental Marine Biology and Ecology, 186: 1-16. Smith, D.L., Delderfield, J., Drummond, D., Edwards, T., Mutlow, C.T., Read, P.D., and Toplis, G.M. (2001). Calibration of the AATSR instrument. Advances in Space Research 28(1): 31-39. Sordo, I., Barton, E.D., Cotos, J.M., and Pazos, Y. (2001). An inshore poleward current in the NW of the Iberian Peninsula detected from satellite images, and its relation with G. catenatum and D. acuminata blooms in the Galician Rias. Estuarine, Coastal and Shelf Science 53: 787-799. Sousa, F. M., and Bricaud, A. (1992). Satellite-derived phytoplankton pigment structures in the Portuguese upwelling area. Journal of Geophysical Research, 97(C7): 11343-11356. Stammer, D., Hinrichsen, H.H., and Kase, R.H. (1991). Can meddies be detected by satellite altimetry? Journal of Geophysical Research 96(C4): 7005-7014. Star, J., and Estes, J. (1990). Geographic Information Systems: An Introduction. PrenticeHall, New York, USA. pp. 665 (ISBN-10: 0133511235). Stricker N.C.M., Hahne, A., Smith, D.L., Delderfield, J., Oliver, M.B., and Edwards, T. (1995). ATSR-2: The evolution in its design from ERS-1 to ERS-2. ESA Bulletin 83: 3237. UNESCO (1999). Module A, Introduction to GIS. Training Module on the Applications of Geographic Information Systems (GIS) for On-line Governance and Accessing Public Domain Information. http://gea.zvne.fer.hr/module/module_a/module_a1.html, Last update: 15-07-1999. Wooster, W.S., Bakun, A., and McLain, D. R. (1976). The seasonal upwelling cycle along the eastern boundary of the North Atlantic. Journal of Marine Research 34(2): 131-141. Wright, D.J. (2000). Down to the Sea in Ships: The Emergence of Marine GIS. In: Marine and Coastal Geographical Information Systems (Research Monographs in Geographic Information Systems). Ed. Wright, D.J. and Bartlett, D.J. Taylor & Francis, London Wright, D.J., Ed. (2002). Undersea with GIS. ESRI press, Redlands, USA. pp. 253 + CDROM. (ISBN-1-58948-016-3) Wright, D.J., and Goodchild, M.F. (1997), Data from the deep: Implications for the GIS community. International Journal of Geographical Information Science 11(6): 523-528.
Applications of Satellite Remote Sensing and GIS to Oceanography and Fisheries… 109
QUESTION BANK True or False 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.
Spatial processes are not important in marine ecosystems. Earth photographs taken by astronauts could be used to study the ocean. AVHRR is the most used sensor in ocean remote sensing. AVHRR is mostly used to estimate chlorophyll-a concentration. CZCS, ATSR and SeaWiFS are the most important ocean colour sensors. Visible wavelength is the same as ocean colour. Infrared wavelength is use to estimate sea surface temperature. Infrared wavelength penetrates 10-20 m deep in the ocean. Visible and infrared sensor could not penetrate through clouds, rain and snow. Salinity could not be measured by remote sensing. Surface winds could be measured by remote sensing. Level-1 processing contains the raw data. It is not possible to conduct attribute queries in a GIS. Raster layers cannot be transformed into a vector structure. Vector and raster layers can be overlayed in a single GIS session.
Short Answer Questions 1. 2. 3. 4. 5. 6. 7.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
8. 9. 10. 11. 12. 13. 14. 15.
What is ocean remote sensing? Describe the main steps in ocean remote sensing. Why is geo-location difficult in the open ocean? What are the main limitations of the use of visible and infrared satellite imagery to study the ocean? Why are infrared sensor data suitable for study coastal upwelling? What is the great advantage of the microwave sensors in relation to the visible and infrared ones? Is satellite remote sensing limited to the sampling of the ocean surface, or can it be used to study subsurface structures? Name 3 biological or optical parameters that could be estimate by ocean colour remote sensing. Name some ocean features that could be analysed by SAR images. Explain why the Western Iberia Buoyant Plume (WIBP) could be study using ocean colour remote sensing. Explain why ocean colour remote sensing could be important in explain the distribution of tuna and tuna-like species (e.g., swordfish). How can internal tidal waves increase primary production in the surface layers? How different is a GIS from a relational database? Explain in a short paragraph (no more than 8 lines) the main advantages of using GIS tools in oceanography and fisheries applications. Describe the major characteristics of kriging interpolation algorithms.
110
A. Miguel P. Santos, Pedro B. Machado and Paulo Relvas
Long Answer Questions
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. What is the spatial data structure of the chlorophyll concentration information used in SIG-IPIMAR? Explain why? 2. How can raster surfaces be generated from vector data? 3. What is the most appropriate data structure for representing vessel tracks in a GIS? 4. How can a GIS help the analysis of marine resources' abundance trends? 5. What are the advantages of geostatistics in the analysis of living resources distribution and abundance?
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 5
DETERMINING EVAPOTRANSPIRATION AND ASSESSING THE PREDICTABILITY OF VEGETATION CONDITION USING SATELLITE REMOTE SENSING METHODS Chandana Gangodagamage1 Department of Civil Engineering, University of Minnesota, Minneapolis, USA
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ABSTRACT In this paper, we investigate the predictability of vegetation health of irrigated areas using state-of-the-art satellite remote sensing methods. We use the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor derived Normalized Difference Vegetation Index (NDVI), Enhanced vegetation Index (EVI), and thermal imageries to extract the spatial and temporal variability of the vegetation health and thermal stress of crops in irrigated agriculture fields over a periods of two years at16-day temporal resolutions. First, several intricacies in determining the crop water requirements by estimating evapotranspiration using traditional and satellite based methods at irrigation fields are presented. Second, we consider the following three questions: (i) does the satellite based Evapotranspiration estimation methods provide a distinct advantage against the traditional ET estimation methods, (ii) How different is the ET values estimated from traditional methods and satellite based techniques and (iii) we used MODIS NDVI, EVI, and thermal images to determined vegetation health and crop-water- stress (crop thermal stress) and demonstrate the applicability of thermal images in predicting the vegetation health before hand the NDVI and EVI image materialized it. Ultimately, we provide the predictability time of the thermal images and its applicability in irrigation schemes using irrigation data collected from Ridibandi-Ela irrigation scheme in Sri Lanka.
Keywords: MODIS NDVI, EVI, Evapotranspiration, surface energy balance, crop water stress, thermal stress, predictability 1
Email: [email protected]
112
Chandana Gangodagamage
INTRODUCTION Determining crop-water requirement for irrigation is in high priority in all scales of irrigation schemes. The most basic, but highly important questions needs to be addressed in irrigation are when to irrigate (temporal resolution), where to irrigate (spatial resolution) and how much of water is needed for irrigation. However, answering these three questions is not so explicit, because of the spatial and temporal complexity and nonlinearity involve in microclimate and hydrologic fluxes (surface fluxes such as evapotranspiration), and vegetation processes. Traditional methods based on crop water balance techniques have proven to be unreliable and highly subjective in estimating hydrologic fluxes and microclimate accurately due to measurement resolution are unable to capture the dynamics between land surface and the atmosphere. However, proper estimates for spatial and temporal variability of the surface fluxes are important to achieving higher water efficiency. Reliably assessing the energy fluxes is important to determine the surface-atmosphere interactions. Surface energy balance models have been successfully applied to estimate the energy flux exchange between the ground surface and the atmosphere near the ground surface. These approaches require solving the energy balance equation at the ground surface (see Eq.1). The net radiation to the surface (Rn), the sensible heat flux (H), and the ground heat flux (G) are the main components of the surface energy balance at the surface. The latent heat flux LE= Λe, where the parameter Λ is the latent heat of vaporization of water (J/Kg) and e is the vapor flux density ( Kg/m2/s), is an important component of the energy balance. The latent heat flux is used in the phase change from a liquid to a gas, mainly from the soil and plants, which expends to evaporate and traspirate a 60% of the precipitation received at surface level. Both, the sum of evaporation and plant transpiration is known as evapotranspiration, which can be calculated as a residual of the difference between net radiation at the surface and losses due to the ground heat flux and sensible heat flux (Assefa M. Melesse and Vijay Nangia, 2005):
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
LE = Rn − G − H
(1)
Where, LE is Latent heat flux or moisture flux in to the atmosphere (W/m2), Rn is Net Radiation at the surface (W/m2), G is Ground heat flux (W/m2) and H is Sensible heat Flux (W/m2). Several surface-based methods accurately estimate surface fluxes at a point location. However, due to the difficulty of capturing the climatological data for every point locations limits the applicability of such methods to larger spatial extents. In those methods the fluxes at unknown locations are predicted. These predictions are highly subjective and depend on how accurately the interpolation methods can capture the spatial variability in a given region. In most methods the latent heat flux is frequently determined using Penman-Monteith equation (see Eq. 2).This equation provides relationship between the land-atmosphere dynamics and vegetation processes:
Determining Evapotranspiration and Assessing the Predictability …
Δ ( Rn − G ) + ρ a c p ( LE =
e s − ea ) ra
r Δ + γ (1 + s ) ra
113
(2)
Where, Rn, G, and H are same quantities as defined in Eq. 1, (es-ea) is the vapor pressure deficit between air and crop surface, ρa is the mean air density at constant pressure, cp is the specific heat of the air, Δ represents the slope of the saturation vapor pressure temperature relationship, γ is the psychrometric constant, and rs and ra are the bulk surface and aerodynamic resistance, respectively. As formulated above, Penman-Monteith approach provides straight forward methods to estimate the latent heat flux (evapotranspiration) from the vegetation. Most of the parameters can be obtained from weather data for a specific point location. Application of this method depends on the number of climatological measurements obtained in an irrigation fields. The other locations are estimated using spatial interpolation methods (Gangodagamage et al., 2007). The main objective of applying remote sensing satellite data is to obtain the spatial variability of the surface energy fluxes at ground level. Satellite images use measurements obtained at different spectral windows of the electromagnetic spectrum to characterize the landscape reflectance and emission at its grid resolutions. The measurements of reflected solar radiations at the surface provide information about the signatures of earth surface objects exposed to that particulate spectral window. This allows quantifying the land surface albedo, vegetation conditions, and thermal properties over wide range of spatial and temporal scales. Surface energy fluxes are related to surface temperature, vegetation condition, and surface emissivities of the land surface at ground level. Estimation of each term in Eq. 1 using remote sensing methods requires high quality data set with good spatial resolutions with combination of a few ground based data. The net radiation term in Eq. 1 can be calculated using the data from incoming (short wave radiation) and out-going (long wave radiation) radiation and associated surface albedo and emissivity and surface temperature (see Eq. 3). The short wave and long wave radiations data is necessary for this calculation.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Rn = (1 − α ) Rswr + εRlwr − εσT04
(3)
Where, Rn is the same quantity as defined earlier, Rswr and Rlwr are the short wave and long wave radiations, respectively, α is the albedo, ε is the emissivity of the surface, σ is the Stefan-Bolzmann constant, and To is the surface temperature The ground heat flux is estimated using surface temperature, albedo, and NDVI dataset. The sensible heat flux is estimated as a function of the temperature gradient above the surface, the wind speed, and the surface roughness. The aerodynamic transfer of heat flux to air, H, is predicted as a function of air density, the specific heat capacity of the air, the aerodynamic surface temperature(Tad), the reference height air temperature (Tr), and the aerodynamic resistance to the sensible heat transport between the surface and the reference height (rah) (see Eq. 4):
114
Chandana Gangodagamage
h = ρc p
Tad − Tr rah
(4)
The application of Eq. 4 is hindered due to the difficulty of estimating Tad from satellite data due to atmospheric contamination errors in sensor calibrations. Hence, the radiometric surface temperature Ts measured by the satellites deviate from the real aerodynamic temperature that drives the heat transfer process. The Surface Energy balance for Land surface (SEBAL) method (Bastiaanssen, 2000) introduced following formulation to avoid such practical intricacy.
H = ρc p
dT rah
(5)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
SEBAL assumed that there is a liner relationship between dT and the satellite measured radiometric surface temperature Ts such that dT = a × Ts + b . SEBAL uses the hot and cold pixel method to determined the empirical values of “a” and “b”. The hot pixel in satellite images are assumed to be experienced very little evapotranspiration where as cold pixels are assumed to be experienced maximum evapotranspiration. Based on these assumptions, a cold pixel is used to represent well vegetated and well watered crop and a hot pixel is used to represent a relatively dry land with low vegetation density (Senay et al., 2007). The difference of the average hot and cold temperature values of three or four pixel area are used to quantify the dT in Eq. 5. This procedure allows quantifying the actual evapotranspiration and per-pixel crop water requirement in spatial and temporal dimension in an irrigation filed based on satellite measurement in conjunction with minimum ground based data. Considering the surface energy balance at the ground level, it can be shown that evapotranspiration is inversely proportional to the crop surface temperature. However, this idea is not new (Tanner, 1963), but was popularized in early 80’s (Jackson et al., 1981). When plants are subjected to water stress, that causes partial stomatal closure, thus controls the evapotranspiration processes so as crops surface temperature start to rise against the surrounding air temperature. Hence, the difference in crop surface temperature and ambient air temperature (Tc-Ta) can be used as an indicator of crop water stress (Idso et al., 1977; Jackson et al., 1977). However this indicator assumed that other parameters such as net radiation, vapor pressure difference, and wind velocity that largely will be represented in the temperature difference. Then Eq. 4 can be represented using this temperature difference i.e., Tc can be approximated with aerodynamic temperature Tar and reference height temperature Tr can be approximated using Ta. This allows one to write temperature difference give is Eq.5 as dT = (Tc − Ta ) . In this paper we review the surface energy balance methods at a point locations using Penmanth-Monteith method. Then we use MODIS 500m NDVI, EVI, and thermal dataset (1 km) to estimate the evapotranspiration and the crop water stress. We also used irrigation data to estimate the water condition at irrigation field scale. Then compute the vegetation health and crop-water-stress using MODIS 16-day time series data in order to demonstrate the applicability of thermal images in predicting the vegetation health before hand the NDVI and
Determining Evapotranspiration and Assessing the Predictability …
115
EVI image assess the vegetation health. Ultimately we provide the predictability time of the thermal images and its applicability in irrigation schemes using irrigation data collected from Ridibandi-Ela irrigation scheme in Sri Lanka. Further, we also compare improvement of accuracy in point based measurements vs. satellite based measurement for the irrigation field.
STUDY AREA AND DETAILS OF THE IRRIGATION SCHEME
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Our experimental study area, Ridibandi-Ela irrigation scheme is located between latitudes of 7.7º N and 7.8º N and longitudes of 80.0º E and 80.2º E in in Kurunegala district, North western province of Sri Lanka. The total area of the irrigation filed is around 2400 hectares (ha). The complete command area is irrigated using Magalla tanks located North-East of the irrigation scheme (see Fig. 1, Fig. 2). The Aerial coverage of the Magalla tank is approximately 261 ha and its perimeter is around 11.5 km. The tank area is varying from time to time based the water availability. The design capacity of the reservoir is 6000 acre-ft.
Figure 1. The location of the Ridibandi-Ela irrigation scheme. The scheme is located in North-West of Sri Lanka latitudes of 7.7o to 7.8o and longitudes of 80o to 80.2o (source: Survey department of Sri Lanka and IWMI)
The Ridibandi-Ela irrigation system provides water for an area about 2400 ha. The reservoir is usually filled up to 7.6 Mm3 during the Maha season and up to 9.2 Mm3 during the Yala season (Yala and Maha are the major irrigation seasons). Besides the irrigating from the main reservoir, the irrigation scheme is also utilize some small tanks and agro-wells. The area is cultivated by 2577 households. Agriculture is the main activity in the region.
116
Chandana Gangodagamage
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Officially, around 3000 farmers have been living in the scheme. Hydrologically, the Ridibandi-Ela irrigation scheme is located in Daduaru-Oya basin which extends approximately 2600 sq km area to the East. This basin is the sixth largest river basin in Sri Lanka. The magnitude of the irrigation scheme is approximately one percent of the total area of the Daduaru-Oya river basin.
Figure 2. A map of a Ridibandi-Ela irrigation scheme. The yellow color thick line shows the boundary of the irrigation scheme The blue color polygon shows the irrigation tank that supply the water for the scheme. The Red color boundary shows the different tributary basin and blue color line shows the drainage network of the irrigation scheme
Determining Evapotranspiration and Assessing the Predictability …
117
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
During the Maha season (November to March) a 99 % of the area is used for paddy cultivation. But in Yala season (April to October), only a 90 % of the area is used for paddy cultivation due to water shortage. The Ridibandi-Ela irrigation scheme is located in dry zone of the Sri Lanka. The reservoir of the irrigation scheme receive the water from the DadaruOya main stream, where river flows vary with the North-East and the South-West monsoon rains, but has flow through the year. The temperature of the area is varies from 25º C to 30º C through out the year. The annual average rainfall received for last twenty year is around 1200 mm. Figure 3 shows the 8-days average rainfall from the year 1995 to 2002.The highest rainfall to the irrigation field receives during the Yala season.
Figure 3. Map of the irrigation scheme. The back color thick line shows it boundaries of ten farmer organizations. The shaded area represents the irrigated extent of the scheme. The names of the each farmer organizations are indicated.
118
Chandana Gangodagamage
Figure 4. Blue color line shows the eight-day composite daily rainfall for the period of 1995 to 2002. Green color thick horizontal bas shows the Yala season and Bleu color horizontal bar shows the Maha season. Lite blue color thick line shows the South-West monsoon and dark-blue color thick line shows North-East monsoon period.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
MATERIALS AND METHODS An old irrigation paper map was scanned and geo-referenced and a digital map for the irrigation scheme was created. Several Landsat-7 images and Advanced Space-borne Thermal Emission and Reflection Radiometer (ASTER) data (See Table 1) set were used to update the map information and demarcate the boundaries and sub-divisions of ten farmer organizations. The Shuttle Radar Terrain Mission (SRTM) 90m resolution data was used to prepare a Digital Elevation model (DEM) for the irrigation scheme. The Irrigation water supply data was collected for one complete irrigation season extending from November, 2001 to April, 2002. The crop water requirement was calculated using Penman-Monteith method (See Eq. 3) for total periods of two-year starting from Januaty, 2001 till December, 2002. Relative water supply (see Eq. 6), the ratio between the water supply (canal water supply, precipitation, ground water) at the irrigation field to the crop water demand considering all losses (infiltration, deep percolation) was calculated for the total periods.
Re lativewater sup ply =
Totalwater sup ply Cropwaterdemand
(6)
Determining Evapotranspiration and Assessing the Predictability …
119
Table 1. Data set characteristics. MODIS, Landsat-ETM (Enhanced Thematic Mapper), ASTER, SRTM data sets used are tabulated. The platform name, Sensor name, the product name, scale factor, and data acquisition period is demonstrated Platform
Sensor
TERRA
MODIS
TERRA
MODIS
TERRA Landsat 7 TERRA SRTM
MODIS ETM ASTER SRTM
Product Name /spatialtemporal resolution Vegetation 250/500 m 16 day MODIS land surface Temperature 1 km 8 day composite Raw Band Data LEVEL-1G 141/055 LEVEL-1B 90 m DEM
Scale Factor 10000
Duration 01/01/2000 to 31/12/2002 01/01/2000 to 31/12/2002
1 km NR NR NR
01/01/2000 to 31/12/2002 23/01/2000 27/09/2001 to 13/01/2001
The relative water supply was plotted against the crop water demand for a period of one irrigation season (January 2002 to April 2002) (See Fig. 6). Relative water supply also was plotted against the NDVI and EVI and shown in Fig. 6. 90
NDVI, EVI, (X 100)
80 70 60 50 40
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
30 0
10
20
30
40 50 time * eight−days
60
70
80
90
Figure 5. Time series measurements of the NDVI (o), EVI (Δ) for the periods of two years (01 January, 2001 to 22nd December, 2002) at 8-day temporal resolution. The NDVI and EVI values are multiplied from 100 and indicated in the figure. The NDVI values are always higher that the EVI values and both the NDVI and EVI follows the same trend.
120
Chandana Gangodagamage
700
relative water supply (RWS)
600 500 400 300 200 100 0 0
5
10 8−days
15
20
Figure 6. The relative water supply (RWS %) for the irrigation season of Nov-2001 to March-2002. The X- axis indicates the days scaled using a scaling factor 8. At the beginning of the irrigation season irrigation water supply for the field is 6 time higher than the crop water requirement. After 30-40 days to 120 days the irrigation water supply maintained approximately 1.8 time higher than the crop water requirement. 100 NDVI EVI
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
NDVI, EVI, (X 100)
90 80 70 60 50 40 30 0
100
200
300 400 500 relative water supply
600
700
Figure 7. The circles (o) and crosses (X) indicated the change of the NDVI and the EVI values with relative water supply. Both the NDVI and the EVI values increase linearly with relative water supply until two time of the relative water supply. After it exceeds the tow times of the relative water supply the NDVI and the EVI values remain constant. The sensitivity of the NDVI index is higher than the EVI index in response to relative water supply.
Determining Evapotranspiration and Assessing the Predictability …
121
Data Set Characteristics The main data sets for this study were prepared from the MODIS sensor of Terra satellite. The MODIS NDVI and EVI data (MODIS VI product) use reflectance measurements of the near infrared-NIR (841-876nm), Red (620-670nm), and blue (459-479nm) bands of visible range of the electro-magnetic spectrum. The MODIS VIs data products contain the NDVI and enhanced vegetation index (EVI) data sets at 250-m, 500-m, and 1-km spatial resolutions. The NDVI index is the normalized ratio of the NIR and the Red bands as given in below equation:
NDVI =
Where,
ρ NIR − ρ RED ρ NIR + ρ RED
(7)
ρ NIR is reflectance in NIR and ρ RED is reflectance in RED wavelength
100 90
NDVI*100
80 70 60 50
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
40 30 30
35
40
45
50 EVI*100
55
60
65
70
Figure 8. A cross-plot of NDVI and EVI time series data for Ridibandi-Ela irrigation scheme for the periods for the same period as in the Fig.5. The dash-line shows the 1:1 line between NDVI and EVI and spread of all data point above this line indicates high sensitivity of the EVI against NDVI.
The strength of the NDVI is, because it has been expressed as a normalized ratio, it cancels out the effect of multiplicative noises (cloud shadows, illumination differences, atmospheric attenuation, certain topographic variations etc.) presence in individual bands
122
Chandana Gangodagamage
(Huete et al., 2002). The main disadvantage of the NDVI is the inherent non-linearity which is generally displays in all ratio-based indices. The NDVI exhibits a saturated (asymptotic) signal over high biomass conditions and is very sensitive to the atmosphere and canopy background variations. The NDVI is chlorophyll sensitive and responds mostly to red band reflectance where as and the EVI is more the NIR sensitive and responsive to canopy structural variation including Leaf Area Index (LAI), canopy type and canopy architecture. The major difference between the NDVI and EVI is that EVI uses blue band reflectance (see Eq. 6) to enhance the sensitivity in high biomass areas where the NDVI exhibits a saturated signal.
EVI =
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Where,
ρ NIR
ρ NIR − ρ RED ×G + C1 × ρ RED − C 2 × ρ BLUE + L
(7)
ρ NIR is reflectance in NIR, ρ RED is reflectance in RED wavelength, ρ BLUE is
reflectance in BLUE wavelength, C1 is atmosphere resistance in RED reflectance (C1=6), C2 is atmosphere resistance in NIR reflectance (C2=7.5), L is canopy background brightness correction faction (L-1) and G is Gain factor (G=2.5). In this study we used the standard 16-day 250-m both NDVI and EVI data. Readers can refer further details and can download the data at no cost from Land Processes Distributed Active Center (LPDAAC) located at the USGS center for Earth Resources Observation and Science (EROS) Center (http://LPDAAC.usgs.gov). The MODIS 8-day Land Surface Temperature (LST)/Emissivity data product (MODIS11A2) was used to obtain the thermal surface temperature measurements at 1-km spatial resolution. We obtain per-pixel temperature data, but resample the data to a 500 m resolution to coincide with the spatial resolution of NDVI/EVI data sets. The temperature is extracted is the average day-time temperature in degree Kelvin for 8-day composite period for two years (see Fig. 5). A 24 hour actual evapotranspiration was computed pixel-by-pixel at 500-m spatial intervals. We used the same approach documented in SEBAL (Bastiaanssen et al., 1998; Bastiaanssen, 2000) algorithm. The instantaneous net radiation, Rn, is derived based on Eq.6. The incoming net radiation is measured from the ground data and outgoing net radiation is derived from MODIS data products (Surface albedo- α, surface emissivity-ε, surface temperature- Ts). The MODISs LST/E products provide per-pixel emissivity values. The surface albedo values are derived from MOD43B3 albedo data products at 16-day, 1-km spatial resolution. We hypothesized that albedo values remains constant for a period of 16days and the data was interpolated for 1000-m to 500-m to be compatible with other datasets. The surface temperature was obtained from LST/E data products at 8-day temporal resolution and 1-km spatial resolution. The datasets were temporally interpolated to 16-day sampling intervals and spatially sampled at 500-m sampling intervals. The instantaneous sensible heat flux (H) was computed based on Eq. 5. We used the temperature derived from MODIS LST/E product to calculate the dT as given in Eq. 5. The dT was assumed to be agreed to a linear relationship with measured MODIS surface temperature and coefficient of this linear relationship was determined by choosing proper hot and cold pixels in each LST/E image. The aerodynamics resistance rah was calculated using
Determining Evapotranspiration and Assessing the Predictability …
123
Eq. 10. The rah measures the resistance to the turbulent transfer of water vapor upward and corresponding vertical transfer of sensible heat towards or from the canopy. A neutral atmospheric stability conditions were assumed in which case the wind profile, U(z), above the crop surface can be defined as a logarithmic function of Zm, the elevation above the surface:
U ( z) =
Z −d U• ln( m ) k Z om
(8)
where, d is the zero displacement height (vegetation bring zero reference top of the plants), Zom is called the roughness length (m) of the momentum transfer which depend on the nature of the surface, U • is the friction velocity (ms-1) due to the eddy momentum transfer and k is the von Karman constant (0.4). The transfer of water vapor and the heat flux in to the turbulent layer above the canopy is determined by following equation:
ln( ra =
Zm − d Z −d ) ln( h ) Z om Z oh k 2U z
(9)
Where, Zm, d, k, Uz, Zom as similar quantities as defined in Eq. 8, Zh is the height of air temperature and humidity measurement, Zoh is the roughness length relative to vapor and heat transfer. We assumed that z=Zm=Zh and Zc=Zom=Zoh which leads to following formulation for aerodynamic roughness:
ln( z − d ) 2 ] Zc rah = k 2U z
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
[
(10)
The roughness length Z and displacement height d are derived from the canopy height. The Uz is measured at Z=3m. The latent head flux is obtained from Eq. 1. The evaporative fraction (EF) is used convert the instantaneous flux values determined by the satellite overpass time to evapotranspiration rates as:
EF =
λE ( Rn − G )
(11)
The above equation determines the part of the available energy for evapotranspiration. The evaporative fraction, EF, approaches zero when there is no moisture left for evapotranspiration. The instantaneous evaporative fraction is assumed to be equal to the daily
124
Chandana Gangodagamage
actual evapotranspiration (E24) (Brutsaert and Chen, 1996; Shuttleworth et al., 1989)which can be calculated as:
λE 24 = EF × Rn 24
(12)
The evapotranspiration calculations from Penman-Monteith and the MODIS data are shown are in Fig. 9. The values of the evapotranspiration calculated from the MODIS sensor data always are relatively higher (0.35 mm/day higher) than the values computed from Penman-Monteith method.
Et− Satellite measurements
6.5
6.0
5.5
5.0
4.5
4.0
3.0
3.5
4.0
4.5 5.0 5.5 ET−Penman−Monteith
6.0
6.5
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 9. A cross-plot of the comparison of the ground (Penmanth-Monteith) and the satellite (MODIS) measurements. The dash-line shows the 1:1 correspondence line between Et derived from ground measurement and satellite measures. In average, the satellite measurements displays a higher Et values Et values computed from ground measurements.
Crop water stress index (CWSI) The difference between the canopy surface temperature (Tc) and the air temperature (Ta), Tc-Ta, has been long documented as a reliable indicator of crop water stress (Idso et al., 1977; Jackson et al., 1977). The Tc-Ta is related to vapor pressure deficit (VPD) for many crops (Idso et al., 1977) and has been shown to exhibit liner relationship (Ehrler, 1973; Idso et al., 1981). There are different formulation has been suggested to estimate the crop water stress index. However, there are two main baseline has been considered in most formulations: nonwater-stressed crop temperature indicated by fully watered crops and maximum stressed baseline exhibits by the non-transpiring crops in which case crop stomata is fully closed. Here
Determining Evapotranspiration and Assessing the Predictability …
125
we implement the crop water stress as the ratio between the canopy temperature (Tc) minus lower bound temperature (Tmin) exhibits by a fully transpiring crop and the temperature exhibits by the non-transpiring canopy(Tmax) closed stomata minus Tmin as (Jones, 1999):
CWSI =
TC − Tmin Tmax − Tmin
(13)
The Tmin and T max are equivalent to the Tbase and Tmax in the original formulation of CWSI (Idso et al., 1981). We used MODIS LST measurements to estimate the Canopy temperature Tc. T min was taken as the average temperature of the lowest temperature of the crops and Tmax was taken as the average highest temperature of the crops in irrigation field selected from satellite image. The CWSI was calculated from each MODIS LST images and the spatial average of the each irrigation field is chosen to consider temporal variability of the CWSI (see Fig. 9). Further, we assumed that the CWRS is equivalent to the crop thermal stress index (CTSI).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
PREDICTABILITY OF THE VEGETATION CONDITION In this section we explore the possibility of crop thermal stress indicator, derived from thermal band data, to statistically predict the vegetation condition. First, we hypothesized that (1) the time scale of the variability of NDVI and EVI are much higher than the CTST (we used crop water stress as a surrogate variable for the crop thermal stress). In order to prove this hypotheses we use following approach. We consider the temporal behavior of NDVI, EVI, and crops thermal stress and show that the temporal correlation length of NDVI and EVI are much higher than the correlation length of thermal stress index. We further assumed that the crop vigor is highly correlated with NDVI/EVI which in fact is not a effective indicator to demonstrate the water stress experienced by the plant for shorter time period (e.g., around 2 weeks time period). Hence we need a method to predict vegetation condition before NDVI/EVI images provides the vegetation condition in order to successfully use in decision making processes in irrigation water supply. We hypothesized that the thermal bands are capable of detecting the increased temperature due to the stomata closure and will be indicated as increase in crop thermal stress. We use the time series of 8-day MODIS data to predict the forthcoming vegetation condition and test these hypothesis using NDVI/EVI time series measurements. Consider a real value time series z(t), with zero mean, the covariance function γ ( k ) :
(
)(
γ (τ ) = z (t + τ ) − z (t ) z (t ) − z (t )
)
(14)
Where, z (t ) represent the mean of the time series z(t), the bracket . indicates the time averaging over τ at position z(t) and correlation function
ρ (τ ) is given as:
126
Chandana Gangodagamage
ρ (τ ) =
γ (τ ) γ (0)
(15)
We calculate a correlation time τc(t) (integral time scale) from the above time dependant correlation function. This function found to exhibit an exponential decay for some time interval for NDVI, EVI, and CTSI as shown in Fig. 10. We use approximately the first 0.3 × ρ(τ) crossing width to determine the correlation time τc for each variable. The correlation time of NDVI and EVI time series show higher value than the correlation time of CTSI series (τc =8days). Further, the correlation time for the NDVI shows higher correlation that the EVI values (τc =16days and τc =24days, respectively).
1 NDVI EVI CWSI
0.9 0.8 Autocorrelation
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
2
4 6 time difference / 8 days
8
10
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 10. The autocorrelation of the NDVI (o) and the EVI (ϒ) time series. The X-axis represents the time difference between NDVI/EVI images scaled by 1/8 day scale factor. Both NDVI and EVI decays exponentially, but the NDVI shows more long-range dependence than the EVI values.
We consider the time series between the NDVI and the CWSI. The peak crosscorrelation between NDVI and CWSI (see Fig-11) occurs at the lag time of 8-day. This suggests that the crop condition exhibits by the NDVI can be predicted 8-days prior to the NDVI image acquisition. Then we consider the time series between the EVI and CWSI. The peak correlation exhibits at the lag time of 24-days. This suggests that CWSI data can be used to predict the vegetation condition 24-days prior to the EVI images.
Determining Evapotranspiration and Assessing the Predictability …
127
0.5 NDVI−CWSI EVI−CWSI
0.4
Crosscorrelation
0.3 0.2 0.1 0 −0.1 −0.2 0
2
4 6 time difference / 8 days
8
10
Figure 11. The crosscorrelation between NDVI-CWSI and EVI-CWSI. The solid-line shows the crosscorrelation between NDVI and EVI time series. The peak correlation occurs at the lag time of 8days. The dash-line shows the crosscorrelation between EVI and CWSI time series. The peak correlation occurs at the lag time of 24-days.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
DISCUSSION AND CONCLUSION The goal of this work is to examine the satellite derived evapotranspiration against the evapotranspiration derived from traditional method like Penman-Monteith approach. We further hypothesized that the CWSI time series derived from thermal data can be used to predict the vegetation condition exhibits by the NDVI and the EVI time series and then estimate the predictability strength of the MODIS thermal satellite measurements. Some clear results have emerged from the analysis. First, we estimated values of the evapotranspiration from the satellite measurements which exhibit a higher value that the values estimated by the Penman-Monteith methods. This difference may be due the satellite based methods can capture the per-pixel evapotranspiration by considering spatial variability of all relevant parameters such as temperature, net radiation, etc. Further, irrigation water supply for the agriculture field is much higher than the crop water requirement. Both the NDVI and the EVI measurements clearly, positively, respond the irrigation water supply to the irrigation field and reach a saturation level beyond which both NDVI and EVI remain constant while RWS increases further. However, the NDVI responds more quickly than the EVI measurements (see Fig. 7) is characterized by the higher slope of the graph of the NDVI vs RWS than the EVI vs RWS . This fact is further agreeing with the autocorrelation analysis we performed on the NDVI and the EVI time series (see Fig. 10). The EVI time series exhibits higher correlation length with itself than the NDVI time series. Hence, we can argue
128
Chandana Gangodagamage
that the NDVI is more sensitive to the environmental factors such as crop water stress and other microclimate perturbations. Second, our analysis of the predictability of the thermal data depicted here an interesting result. The interesting question is the information provided by the NDVI and EVI data at satellite overpass time t can be obtained from the thermal stress index derived at time t-τ where τ is the lag time obtained from cross correlation analysis. The statistics show that the maximum correlation between vegetation data products (NDVI, EVI) and thermal data products occurs at τ, where the ρ(τ)= ρmax (τEVI) value for EVI is higher (τEVI=16-24 days) than the ρ(τ)= ρmax (τNDVI) value of the NDVI (τNDVI=8-16 days). This suggests that thermal image has considerable potential in successfully implementing in decision making processes of irrigation water management. Another interesting question is whether the documented statistical delay (τ) provides enough time frame to processes the thermal data before NDVI, EVI images provide information about vegetation condition. However at present the thermal images are available at 8-day temporal resolution, which further provide substantial time frame to processes the data and get the information into management. Both of these questions are the subject of future investigations.
ACKNOWLEDGMENTS This work has been supported by the International Water management Institute, Head Quarters, Colombo, Sri Lanka.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REFERENCES Assefa M. Melesse and Nangia, V. (2005). Estimation of spatially distributed surface energy fluxes using remotely-sensed data for agricultural fields. Hydrol. Process. 19: 2653-2670. Bastiaanssen, W. G. M. (2000). SEBAL-based sensible and latent heat fluxes in the irrigated Gediz Basin, Turkey. Journal of Hydrology 229: 87-100. Bastiaanssen, W.G.M., Menenti, M., Feddes, R.A. and Holtslag, A.A.M. (1998). A remote sensing surface energy balance algorithm for land (SEBAL). 1. Formulation. Journal of Hydrology 212-213: 198-212. Brutsaert, W. and Chen, D. (1996). Diurnal Variation of Surface Fluxes During Thorough Drying (or Severe Drought) of Natural Prairie. Water Resour. Res., 32: 2013-2019. Ehrler, W. L. (1973), Cotton Leaf Temperatures as Related to Soil Water Depletion and Meteorological Factors. Agron J. 65: 404-409. Gangodagamage, C., Zhou, C.X. and Lin, H.S. (2007). Autocorrelation: Spatial. In: Shekhar and Xiong (Editors-in-Chief) Encyclopedia of GIS. Huete, A., Didan, K., Miura, T., Rodriguez, E.P., Gao, X. and Ferreira, L.G. (2002). Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sensing of Environment 83: 195-213. Idso, S.B., Jackson, R.D. and Reginato, R.J. (1977). Remote-Sensing of Crop Yields. Science 196:19-25.
Determining Evapotranspiration and Assessing the Predictability …
129
Idso, S.B., Jackson, R.D., Pinter, P.J., Reginato, R.J. and Hatfield, J.L. (1981). Normalizing the stress-degree-day parameter for environmental variability. Agricultural Meteorology 24: 45-55. Jackson, R.D., Reginato, R.J. and Idso, S.B. (1977). Wheat Canopy Temperature: A Practical Tool for Evaluating Water Requirements. Water Resour. Res. 13: 651-656. Jackson, R.D., Idso, S.B., Reginato, R.J. and Pinter Jr, P.J. (1981). Canopy temperature as a crop water stress indicator. Water Resour. Res. 17: 1133-1138. Jones, H.G. (1999), Use of infrared thermometry for estimation of stomatal conductance as a possible aid to irrigation scheduling. Agricultural and Forest Meteorology 95: 139-149. Senay, B.G., Budde, M., Verdin, P.J. and Assefa, M.M. (2007). A Coupled Remote Sensing and Simplified Surface Energy Balance Approach to Estimate Actual Evapotranspirationfrom Irrigated Fields. Sensors 7: 979-1000. Shuttleworth, W.J., Gurney, R.J., Hsu, A.Y. and Ormsby, J.P. (1989). FIFE:The variation in energy partitioning at surface flux sites, remote sensing and large-scale processes. Tanner, C. B. (1963). Plant Temperatures. Agron J 55: 210-211.
QUESTION BANK Expand the Following 1. 2. 3. 4.
ASTER CTSI CWSI ETM
5. 6. 7. 8.
EVI MODIS NDVI SEBAL
9. SRTM 10. TERRA
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Short Answer Questions 1. Define the term Evapotranspiration? List various methods in remote sensing to assess it. 2. Explain the term crop thermal stress index. How satellite remote sensing data could be used for this? 3. Express the importance of red, near infra red and thermal wavelength while assessing the vegetation stress. 4. What do you understand by the term predictability of the vegetation condition? 5. What is crop water stress index? 6. What is latent heat flux? Express the methods to evaluate it using satellite remote sensing.
Long Answer Questions 1.
Explain the framework for assessment of vegetation stress.
130
Chandana Gangodagamage
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
2.
What is MODIS? Explain various parameters which could be retrieved from MODIS data/
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 6
GEOINFORMATICS FOR WATER ACCUMULATION MODELLING - A CASE STUDY FROM INDIA Ashoke Basistha* Department of Hydrology, Indian Institute of Technology Roorkee, Roorkee 247 667 India
ABSTRACT
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Assessment of cumulative inflow volume at a point that can be techno-economically harnessed by a check dam is important from the viewpoint of a planner interested in increasing agricultural output through irrigation. This not only provides information on required storage volume but also on area under a particular crop that can be irrigated with the stored water. The study involves Soil Conservation Service Curve Number (SCS-CN) based inflow estimation in eleven selected check dam catchments of Karso watershed of district Hazaribagh, Jharkhand, India. Available storage volume and contributing catchment area at check dam point was estimated in Geographical Information System (GIS) environment with the help of Digital Elevation Model (DEM). Land use map prepared by classifying Landsat TM image was used for calculating CN values. The results show that on an average, water accumulated in each check dam can be used for cultivating paddy in approximately ten percent of the area of contributing catchments.
Keywords: CN, flow accumulation modelling, GIS, Jharkhand, Karso, SCS
INTRODUCTION For an agrarian economy like India, water availability is critical issue. Timely availability of water in adequate quantity is important for crop growth rates. With a goal to move towards sustainable development, small water harvesting structures with local use assume more importance than construction of few large dams and a wide network of distribution system. Also, it saves large initial expenditure and huge water loss incurred in transportation to long
132
Ashoke Basistha
distances. Deciding the proper location and capacity of check dams require estimation of water inflow. Often observed discharge data for such remote locations is not available, particularly so for developing countries, leaving no options other than depending on results from modelling of inflow from rainfall data. Estimation of cumulative inflow into the check dam can lead directly to the assessment of crop area that can come under irrigation coverage from that check dam. Curve Number (CN) model developed by United States Soil Conservation Service (SCS) is popular with researchers due to its simplicity and limited data requirement. It has been applied to derive the annual runoff potential in semiarid area of Rajasthan (Gupta et al., 1997). Such analysis using Remote Sensing (RS) and GIS can facilitate the planning process before expensive field investigations are taken up. The present work attempts to apply the technology on a watershed which is already under conservation practices, to serve as a demonstration.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Conceptual Background Estimating runoff from rainfall measurements is very much dependent on the time scale. For short durations (hours) the complex inter-relationship between rainfall and runoff is not easily defined, but as the time period lengthens, the connection becomes simpler until, on an annual basis, a straight line correlation may be obtained. Similarly, the size of the area being considered also affects the relationship. For very small areas of homogeneous nature the derivation of the relationship could be fairly simple; for very large drainage basins on a national or even international scale and for long time periods, difference in catchment effects are smoothened out giving relatively simple rainfall-runoff relationships. However, for short time periods, great complexities occur when spasmodic rainfall is unevenly distributed over an area of varied topography, geological composition and land use. In the intermediate scale of area and time, other physical and hydrological factors, such as evaporation, infiltration and groundwater flow are very significant and thus any direct relationship between rainfall alone and runoff is not easily determined. A plethora of models (e.g. TOPmodel, SWAT (Soil and Water Assessment Tool), HYSIM (Hydrological Simulation Model), GSFB (Generalized Surface inFiltration Baseflow; 8 parameters), IHACRES (6 parameters), LASCAM (Large Scale Catchment Model; 22 parameters), to name a very few) of all degrees of complexity have been developed over a period of more than a century to address this primary issue of computing runoff from rainfall (and other related parameters, as applicable), as observed records of runoff are limited compared to records of rainfall. The hydrological models vary in description of the components of the hydrological cycle, model architecture and structure, degree of complexity of inputs, number of parameters to be determined, time interval used in simulation, error and risk analyses, and output generated. Depending on the degree of representation of the real world systems, rainfall runoff models can be classified into three broad types (i) Distributed physically based models (which are based on the complex law of physics expressed as systems of non-linear partial differential equations), (ii) Systems based (Black/Grey box models which make little or no attempt to simulate individual constituent hydrologic processes and rely heavily on systems theory developed in other branches of engineering *
Email: [email protected]
Geoinformatics for water accumulation modelling - A case study from India
133
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
science; the essence being the empirical estimation of transfer functions which interrelate in the time domain the input (i.e., rainfall) to output (i.e., discharge), and (iii) Quasi-physical conceptual models which occupy an intermediate position between two types of models in terms of complexity, disaggregation and data requirements. Most of the computer based models, which are very popular these days, have a number of parameters, usually use a short time interval, produce hydrographs as well as water yield and provide continuous simulation (Mishra and Singh, 2004). Despite their comprehensive structure, many of these models have not yet become standard tools in hydrological practice in developing countries, such as India, Pakistan, Nepal, and other countries of Asia as well as African countries. The reason is twofold. First, most basins in these countries are ungauged and there is little hydrological data available. Second, these models contain too many parameters, which are difficult to estimate in practice and vary from basin to basin. Although some of these models have been applied to ungauged basins, the fact is that they are not easy for practical applications. Furthermore, when these models are compared on the same basin, they are found to vary widely in their performance (Mishra and Singh, 2004). This makes selection of a particular model very important. Basically it depends upon climatic and physiographic characteristics of the basin, length of records of different types of input data, quality of field data – both in time and space, availability and capability of computing facility, the possible need for transporting model parameters from smaller catchments to larger catchments, and the ability of the model to be conveniently updated on the basis of current hydro meteorological conditions etc. For the present case study, the hydrologic soil Cover Complex Number (CN) method developed by the United States Soil Conservation Service (SCS), as modified later on by the Soil Conservation Division, Ministry of Agriculture for Indian conditions (Handbook of Hydrology, 1972) was used. SCS-CN is one of the most accepted methods for computing the volume of surface runoff for a given rainfall event from small agricultural, forest and urban watersheds, which has now been renamed as Natural Resource Conservation Service Curve Number (NRCS-CN) method. It is a simple conceptual model, easy to understand, and useful for ungauged watersheds. It accounts for the major runoff producing watershed characteristics, viz., soil type, land use/treatment, surface condition, and antecedent moisture conditions. This was chosen as only daily rainfall data – and neither rainfall intensity nor rate of infiltration was available.
Previous Experiences The Soil Conservation Service Curve Number (SCS-CN) method has been a topic of much discussion, especially in the last three decades mainly due to its simplicity and stability, limited data requirement and amenability to easy coupling with GIS environment. However, due to spatial and temporal variability of rainfall, quality of measured rainfall-runoff data, and the variability of antecedent rainfall and the associated soil moisture amount the SCS-CN method has sufficient room for variability (Mishra et al., 2006). In its original form, it suffers from the limitations such as unaccounted rainfall intensity (which is an important source of variability in the methodology), lack of clear guidance as to how to vary antecedent moisture condition (AMC), the discrete unrealistic relation between CN and AMC, the fixing of the initial abstraction ratio at 0.2 (or 0.3, suggested for Indian conditions) precluding a
134
Ashoke Basistha
regionalisation based on geologic and climatic setting, the method’s varying accuracy for different biomes and the absence of an explicit provision for spatial scale effects. Studies generally describe the CN behaviour as unpredictable, owing to the effect of storm duration or rainfall intensity and, to some extent, the empirical relationship between initatil abstraction ( I a ) and potential maximum retention (S) (Jain et al., 2006). Also, the three AMC levels permit unreasonable sudden jumps in curve numbers (CN), which result in corresponding jumps in estimated runoff. Besides this, the constant initial abstraction coefficient in the SCSCN methodology, which in reality, depends largely on climatic conditions, is perhaps the most ambiguous assumption and requires considerable refinement. These are perhaps the reasons that the past research endeavours suggested a need for further improvement, overhauling, or replacement of the method (Sahu et al., 2007). The works of Hjelmfelt (1991), Ponce and Hawkins (1996), Mishra et al. (2004; 2006a; 2006b; 2007), Mishra and Singh (1999; 2004a; 2004b), Jain et al. (2006a; 2006b), Sahu et al. (2007) among many others are worth mentioning in this regard. But, being widely understood and accepted for what it is a conceptual model supported with empirical data to estimate direct runoff volume from infrequent storm rainfall depth, lumped to circumvent the often cumbersome description of spatial and temporal variability of infiltration and other losses, this is a method of choice by practicing engineers and hydrologists for soil and water conservation planning and design (Ponce and Hawkins, 1996). Further, it satisfies the requirements of developing countries and therefore, it is no surprise to have become popular, despite the lack of sophistication (Mishra and Singh, 2004).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Experimental Site The Indo-German Bilateral Project ‘Watershed Management’ is implementing participatory watershed programmes in 9 watersheds throughout India jointly through government departments and non-governmental organizations since 1989 (Honore and Pandey, 2002). Karso watershed in Hazaribagh district, Jharkhand, India is one of them where monitoring for rainfall, runoff and sediment yield is done under the project (Mishra et al., 2006). The Karso watershed (present experimental site) is a sub-watershed of river Barakar, a major tributary of river Damodar. It is situated in the high lands of Chhotanagpur Plateau in the district of Hazaribagh, Bihar (latitudes 24o 12' 30'' N and 24o 17' 00'' N and longitudes 85o 23' 45'' E and 85o 28' 00'' E). It covers an area of 27.44 sq km encompassing 16 villages in full and/or in part. The watershed is located at about a distance of 25 km towards the north of Hazaribag town and about 35 km towards the south of Tilayia reservoir covered in two 1: 25000 toposheets (SOI Index number 72H/7/SE and 72H/8/NE). The location map of the watershed is shown in Figure 1.
Geoinformatics for water accumulation modelling - A case study from India
24017’N 850
135
24017’N 850
Karso Watershed
24012’N 850
24012’N 850
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 1. Location of Karso Watershed
Physiographically the area varies from almost flat land to steep hills (the range of elevation being 385 m to 675m above the msl). The mean slope of the watershed is 9 %. The watershed can be divided into three main landscapes (i) the southern part which is hilly and undulating land - the source of most of the feeder streams; (ii) gently undulating and rolling uplands, that are dissected by narrow valleys and depressions; and (iii) valley land which is mostly confined along the main river Kolhuwatari Nadi which drains from south to north almost parallel to the Patna - Ranchi Road. The elevation zone map of the watershed is given in Figure 2. The shape of the watershed comes under the hydrologic category fan shaped. The main river Kolhuwatari Nadi is a 6th order stream joining with Mohuaghat Nadi just beyond the present outlet point considered and then flowing down to river Barakar as Nadhadwa Nadi or Barhi Nadi. The entire zone lies in the catchment of Tilayia Reservoir. The climate of the study area is subhumid tropical with an annual average rainfall of 124 cm, the minimum and maximum temperatures being 43° C and 2.4° C respectively. The principal rainfall (more than 80% of the annual total) occurs from June to September with occasional showers in December to January and heavy showers accompanied by thunders in May. This causes these months to be considerably humid. The annual average humidity varies from 66 to 77 percent. The maximum temperature occurs generally in April - May and the minimum temperature occurs in around January.
136
Ashoke Basistha
Elevation Zones 70%) are decreasing over time from 3.53% to 2.77%. On the other hand, the low percent tree cover (10% and 70%
2000 40.26 17.72 25.27 8.59 4.63 3.53
2001 41.09 17.31 25.03 8.52 4.59 3.47
2002 44.79 17.70 21.55 8.72 4.23 3.00
2003 45.51 17.66 21.50 8.33 4.14 2.96
2004 48.28 15.22 21.42 8.01 4.18 2.90
2005 49.37 14.48 21.29 8.17 3.93 2.77
State Level Analysis Here, we report state-wise temporal distribution of percent tree cover change from 2000 to 2005 (Table 2). North eastern states like Arunachal Pradesh, Manipur, Nagaland, Meghalaya and Mizoram showed high percentage of tree cover loss compared to the other states like Assam and Tripura (Table 2). For e.g. 70% percent tree cover class in Mizoram declined from 59.6% to 45.5%, Nagaland from 48.4% to 36% and Manipur from 47.4% to 33.3% and relative increment in the 50-70% class, as the high density tree cover are being cleared and converted to medium density tree cover class which is represented by simplified secondary stands or other forest types due to harvesting of selective species. Similarly, other tropical rain forests states like Karnataka, Kerala and Tamil Nadu that are part of Western Ghats also showed similar pattern of declining percent tree cover (Table 2). Forests in states like Bihar, Jharkhand, Chhattisgarh and Madhya Pradesh are mainly composed of tropical deciduous forests that reveals decline in percent tree cover (50-70% and >70% class) explaining the high anthropogenic pressure on forests in these areas. States bearing temperate forest like Jammu & Kashmir, Uttaranchal and Himachal Pradesh had lesser area of high percent tree cover (>70%) and predominantly the medium tree cover of 30-50% and 50-70% (Table 2).
194
A Giriraj, P.K. Joshi, Shilpa Babar et al.
Table 2. State wise temporal distribution of MODIS-VCF percent tree cover from 2000 to 2005. Percent tree cover was calculated only in the forest area using the GLCF mask
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
SNo.
States
1
Andhra Pradesh
2
Arunachal Pradesh
3
Assam
4
Bihar
5
Chattisgarh
6
Delhi
7
Goa
1
2
3
4
5
6
2.5 3.3 5.0 5.2 5.6 6.4 0.6 0.8 0.8 0.9
4.7 6.8 7.7 6.5 6.8 8.2 0.7 0.7 0.6 0.8
42.9 41.9 43.3 38.2 39.1 37.9 2.5 2.5 2.0 3.0
41.5 39.5 39.3 42.7 41.5 41.0 6.2 6.5 7.0 6.3
6.4 6.7 3.4 6.7 6.5 6.1 20.2 21.1 29.2 29.2
2.1 1.8 1.3 0.7 0.5 0.5 69.8 68.5 60.5 59.6
1.1 0.9 1.8 3.5
0.9 0.6 4.2 3.9
2.8 2.3 12.1 11.2
7.1 7.7 30.1 27.9
30.1 30.4 27.3 27.0
58.0 58.0 24.6 26.4
5.1 2.7 3.4 5.5 19.7 17.2 18.9 18.4 21.3 22.5 1.3 1.9 3.9 2.7 2.0 2.3 100.0 100.0 100.0 100.0 100.0 100.0 0.1 0.1 0.1 0.2 0.0 0.2
3.1 3.2 4.3 3.0 9.3 12.7 10.2 13.6 9.2 9.5 2.2 2.6 2.8 4.3 1.5 2.0 0.0 0.0 0.0 0.0 0.0 0.0 0.4 0.9 0.5 0.9 0.3 0.5
8.5 12.3 11.5 10.9 21.2 23.3 21.1 18.3 23.0 23.3 29.9 30.1 35.4 23.3 27.1 32.7 0.0 0.0 0.0 0.0 0.0 0.0 8.2 9.8 7.6 9.8 6.3 8.3
23.3 23.1 23.1 22.0 21.0 20.2 17.9 17.7 17.2 14.4 54.5 55.6 48.6 50.0 51.6 53.5 0.0 0.0 0.0 0.0 0.0 0.0 31.3 30.1 32.2 28.4 28.2 28.0
30.9 29.9 28.1 25.0 14.0 14.0 23.7 26.3 24.6 27.5 11.3 9.5 8.8 19.5 16.1 9.0 0.0 0.0 0.0 0.0 0.0 0.0 27.1 28.0 30.2 30.9 37.9 45.8
29.1 28.8 29.5 33.7 14.7 12.6 8.2 5.7 4.7 2.8 0.8 0.4 0.4 0.3 1.7 0.5 0.0 0.0 0.0 0.0 0.0 0.0 32.8 31.1 29.4 29.8 27.1 17.2
Systematic Assessment of Forest Cover Change and Forest Fragmentation…
195
Table 2. (Continued) SNo.
8
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
9
States
Gujarat
Haryana
10
Himachal Pradesh
11
Jammu & Kashmir
12
Jharkhand
13
Karnataka
14
Kerala
15
Madhya Pradesh
1
2
3
4
5
6
5.2 3.8 5.8 8.3 6.0 4.8 49.3
10.6 4.6 6.8 12.8 5.9 7.0 5.3
73.6 75.9 79.1 69.5 66.4 80.3 9.7
10.5 15.5 8.1 9.4 21.5 7.7 25.6
0.1 0.1 0.1 0.1 0.2 0.0 9.8
0.0 0.0 0.0 0.0 0.0 0.0 0.3
50.4 53.0
4.6 3.4
10.7 8.4
28.2 26.1
6.1 8.7
0.0 0.3
50.7
4.3
10.0
25.9
8.9
0.2
55.9 56.1 10.5 10.2 11.0 9.9 9.9 8.5 18.2 18.3 20.7 16.6 17.0 16.2 2.0 2.0 2.3 2.5 2.7 2.1 0.9 1.4 1.6 2.2 1.3 1.2 0.4 0.2 0.4 0.5 0.9 0.4 4.6 4.6 9.8 5.3 6.6 5.9
1.8 1.7 9.4 8.4 7.6 7.1 6.0 6.9 11.0 11.2 10.7 10.0 9.9 10.2 1.7 3.5 2.0 2.9 2.9 2.3 2.3 3.0 2.5 3.4 1.4 2.2 1.1 0.9 1.3 1.0 1.0 1.0 4.6 4.2 6.0 4.6 3.2 4.2
3.6 7.2 16.5 17.5 15.3 16.8 13.0 17.4 16.1 15.6 16.0 15.1 14.9 15.2 23.9 25.6 21.5 22.5 22.7 34.5 26.7 26.8 23.7 27.6 22.2 20.3 15.1 15.1 15.4 12.2 13.1 10.5 52.6 50.3 45.0 47.0 45.1 46.9
25.1 26.3 30.1 31.0 32.9 33.4 34.4 32.5 22.1 21.0 19.9 24.9 21.6 22.0 45.8 49.4 46.0 41.8 46.9 49.3 39.6 36.3 35.6 36.5 39.3 37.7 44.2 44.2 40.9 39.9 45.1 43.3 33.6 37.6 35.9 37.4 38.4 37.5
13.1 8.4 25.1 24.4 25.3 25.7 29.6 26.0 26.8 26.8 25.9 27.4 28.0 28.4 25.1 19.5 27.3 29.2 24.0 11.6 14.7 17.2 21.7 18.7 24.1 28.3 18.1 20.4 23.9 28.3 26.0 31.4 4.5 3.3 3.3 5.7 6.7 5.5
0.5 0.4 8.3 8.5 7.8 7.1 7.2 8.7 5.7 7.1 6.8 6.1 8.7 8.0 1.6 0.0 0.9 1.1 0.8 0.3 15.9 15.3 14.9 11.5 11.7 10.4 21.2 19.1 18.1 18.1 14.0 13.4 0.1 0.0 0.1 0.0 0.1 0.0
196
A Giriraj, P.K. Joshi, Shilpa Babar et al. Table 2. (Continued)
SNo.
States
16
Maharashtra
17
18
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
19
20
21
Manipur
Meghalaya
Mizoram
Nagaland
Orissa
1 1.1 1.4 2.1 2.5 2.3
2 2.7 2.6 2.5 5.6 3.0
3 40.0 41.1 48.0 34.6 37.1
4 48.8 50.8 43.6 47.7 48.7
5 6.5 3.8 3.4 9.3 7.9
6 0.9 0.4 0.4 0.3 1.0
2.2
3.5
34.3
53.7
5.8
0.5
0.9 1.3
1.3 1.5
4.8 5.8
17.2 19.3
28.4 26.6
47.4 45.5
1.3
1.5
5.4
15.2
33.5
43.2
1.8
1.3
5.2
15.6
35.8
40.4
1.2
1.3
5.0
17.3
36.9
38.3
0.8
1.2
5.2
15.5
43.9
33.3
0.4
1.4
16.3
43.5
19.3
19.0
0.4
1.7
17.4
44.2
20.6
15.8
0.5
1.2
12.5
38.6
33.1
14.1
0.4
1.1
12.3
39.1
34.2
12.9
0.4
1.2
14.4
40.0
32.0
11.9
0.5
1.0
13.3
37.5
36.7
11.0
0.1
0.3
3.2
13.4
23.3
59.6
0.1
0.4
3.9
14.9
26.0
54.7
0.1
0.2
2.6
10.8
32.8
53.6
0.1
0.3
3.3
16.0
30.2
50.1
0.1
0.3
3.6
18.3
30.5
47.2
0.1
0.2
4.4
14.1
35.7
45.5
0.6
0.9
6.0
20.9
23.3
48.4
0.6
1.1
6.3
17.2
30.0
44.8
0.8
1.1
5.7
16.3
31.9
44.2
0.7
0.8
5.6
17.1
35.6
40.3
0.4
0.6
4.2
16.0
42.0
36.8
0.5
0.7
5.9
16.3
40.5
36.0
0.9
2.1
25.1
48.8
19.5
3.7
1.1 2.1
3.1 2.6
25.2 33.7
51.0 43.2
18.2 14.7
1.3 3.8
1.5
3.2
22.9
45.5
25.1
1.8
1.4
2.1
25.8
45.7
22.0
3.0
1.3
2.3
30.1
48.0
16.2
2.1
Systematic Assessment of Forest Cover Change and Forest Fragmentation…
197
Table 2. (Continued) SNo.
22
23
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
24
States
Punjab
Rajasthan
Sikkim
25
Tamil Nadu
26
Tripura
27
Uttar Pradesh
Uttaranchal
1
2
3
4
5
6
54.6
9.6
19.0
13.8
2.9
0.0
62.1
6.2
16.5
14.0
1.2
0.0
71.3
3.8
11.4
13.1
0.4
0.0
51.6
11.0
20.7
14.6
2.1
0.0
66.8
5.5
12.0
13.6
1.9
0.1
70.7
3.4
9.9
14.2
1.8
0.0
11.8
12.2
72.4
3.6
0.0
0.0
10.0
7.3
72.8
9.9
0.0
0.0
18.9
17.6
63.2
0.3
0.0
0.0
10.7
8.2
73.8
7.4
0.0
0.0
12.9 10.6 1.6
4.5 6.1 3.1
64.5 70.7 7.5
18.0 12.5 8.3
0.0 0.0 49.2
0.0 0.0 30.3
1.0
2.4
9.4
13.1
46.7
27.4
1.8 0.9 0.9 0.6 2.1
3.4 1.6 1.4 1.4 10.2
8.9 8.8 9.0 8.9 34.6
17.6 20.4 19.2 27.2 34.5
45.6 47.3 49.6 43.2 10.4
22.7 21.0 21.3 18.7 8.2
1.5 8.2 7.4 9.4 10.3 0.5 0.6 2.3 1.1 0.8 0.9 28.7 23.1 25.7 28.4 26.6 27.2 6.8 7.4 8.5 6.2 6.9 6.7
9.5 13.5 12.3 9.7 8.2 2.1 2.7 1.9 2.6 2.8 2.3 12.4 16.4 14.4 12.5 11.0 9.7 8.2 8.2 6.5 7.1 5.1 4.7
41.5 38.4 36.8 29.4 28.5 15.7 14.2 11.5 13.0 13.4 13.1 19.1 20.6 21.4 20.7 22.0 22.2 14.8 15.3 14.0 15.8 13.9 12.6
31.3 26.2 29.0 35.2 32.6 41.0 44.2 36.6 36.2 40.6 40.0 16.9 18.1 16.7 19.8 17.5 17.5 27.5 27.8 26.5 27.2 28.4 28.7
10.1 8.9 9.9 12.2 17.2 27.9 25.5 38.0 38.9 35.6 39.0 20.2 16.9 16.0 16.5 17.2 18.6 34.0 33.8 37.2 36.5 38.4 40.6
6.1 4.9 4.6 4.1 3.1 12.7 12.8 9.8 8.3 6.8 4.7 2.6 5.0 5.7 2.1 5.7 4.7 8.7 7.5 7.4 7.1 7.3 6.8
198
A Giriraj, P.K. Joshi, Shilpa Babar et al. Table 2. (Continued)
SNo.
States
29
West Bengal
1 4.1 5.6 7.0 5.7 3.8 5.3
2 6.7 5.4 5.7 3.8 3.8 3.3
3 15.6 14.0 15.7 12.8 14.8 14.9
4 26.6 24.8 21.9 19.6 22.5 19.6
5 34.3 30.4 34.5 30.8 33.9 29.8
6 12.7 19.9 15.2 27.3 21.2 27.1
Category 1: < 5%; 2: > 5-10- 30-< 50%; 5: > 50-70%
Bio-Geographic Zone Analysis Of the major nine zones, Northeast and East Himalayas had high tree cover ratio (>70%) compare to the other biogeographical zones (Table 3). We observed a decline in tree cover of >70% class from 38 to 30% in the northeast and similar condition was observed in East Himalayas (66.4% to 56.1%). Western Ghats and northern Himalayan mountains also experienced similar decline in the >70% class and increment in the 50-70% tree cover class. East coast zones shown increment in the percent tree cover ( 5-10- 30-< 50%; 5: > 50-70%
Evaluating MODIS-VCF Products Prior detecting the tree cover change for entire study site, we evaluated the MODIS-VCF recent product with the field data gathered from biodiversity research program. We observed a reasonable relationship with an R2 of 0.83 from the four test areas (Fig. 4). Interestingly the detected percent tree cover with value 0 matched with the ground inventory identified as forest cleared site in the Eastern Ghats of Andhra Pradesh explaining the accuracy of the product. 100
MODIS-VCF tree canopy cover (%)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
80
60
Western Ghats Eastern Ghats Himalayas and Central India Northeast India
40
20
0 0
20
40 60 Field data tree crown cover (%)
80
100
Figure 4. Validation of MODIS-VCF tree canopy cover vs field data of crown cover from 60-sites, covering a wide range of tree cover density, where y=0.88x and R2 = 0.83. (See Hansen et al., 2002; 2003 for similar such relationship in detail).
Systematic Assessment of Forest Cover Change and Forest Fragmentation…
201
Detecting Forest Area Loss The results from the time-series VCF product using pixel based linear regression model is shown in Fig. 5. No-change tree cover pixels were eliminated from the current analysis. Further the change pixels were masked with GLC2000 to detect changes only within the forest areas. We characterize the high forest area loss (high abundant value within 5x5 km grid) based on high negative slope and high variance with low offset value. A total of 2391 pixels are identified as low forest area loss, 1956 pixels of medium forest area loss and 237 pixels are detected as high forest area loss. Pixels which are detected as high forest area loss can be treated as “critical areas of difference or Rapid Ecological Alerts (REA)” for further evaluation using high spatial satellite data to establish a detail monitoring system. For evaluating these outputs we have tested this in two study regions: Visakhapatnam district in the Eastern Ghats of Andhra Pradesh and Nagaland of Northeast region (Fig. 6 and 7).
A B
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
C
A
B
Figure 5. Classified map depicting the forest area loss using linear regression model with the threshold condition for the 5 × 5 km grid.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
202
A Giriraj, P.K. Joshi, Shilpa Babar et al.
Figure 6. Satellite image (MSS, TM, LISS-III) and classified map for the Visakhapatnam district, Andhra Pradesh showing the deforestation and degradation which was detected by the MODIS-VCF product as high forest area loss
Systematic Assessment of Forest Cover Change and Forest Fragmentation…
203
A
Figure. 7. Validated MODIS-VCF pixel based results having high forest area loss with the Landsat data for its accuracy. We observed a large scale deforestation and fragmentation in the northeast region (Nagaland) and similar such areas across the study site.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Further we analysed the critical forest area loss by combining the protected areas (PA) for monitoring the natural resources more efficiently. In the present analysis, we have included only the available GIS layers of national and international designations proposed by the UNEP world protected areas. Out of the 613 PAs, only 370 PAs having GIS database were used for the current analysis. Around 87 PAs have changes in forest areas from low to high. A detail table gives you the list of PA’s having changes in the forest cover and its count (Table 4). PAs like Mundanthurai Sanctuary, Dachigam National Park, Limber Sanctuary, Radhanagari Sanctuary, Shendurney Sanctuary and Shettihalli Sanctuary had high critical area of loss. PAs with low forest cover change and high count includes Khangchendzonga National Park, Kolleru Sanctuary, Gobindsagar Sanctuary and Simlipal Sanctuary (Table 4). Table 4. List of protected areas designated by UNEP, having changes in forest cover. PA layer was integrated with critical forest area map for monitoring the natural resources SNo
Protected area
1
Mundanthurai Sanctuary
Cateogry (critical forest loss) 3
Pixel sum
2
Dachigam National Park
3
1
3
Limber Sanctuary
3
1
4
Radhanagari Sanctuary
3
1
5
Shendurney Sanctuary
3
1
4
204
A Giriraj, P.K. Joshi, Shilpa Babar et al. Table 4. (continued)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
SNo
Protected area
6
Shettihalli Sanctuary
Cateogry (critical forest loss) 3
Pixel sum
7
Lado National Park
2
7
8
Dibru Sanctuary
2
6
9
Gumti Sanctuary
2
5
10
Nagarjunasagar National Park
2
5
11
Bhimbandh Sanctuary
2
4
12
Rajaji National Park
2
4
13
Laokhowa Sanctuary
2
3
14
Mehao Sanctuary
2
3
15
Pabha Sanctuary
2
3
16
Sonanadi Sanctuary
2
3
17
Keibul-Lamjao National Park
2
2
18
Kotgarh Sanctuary
2
2
19
Mahananda Sanctuary
2
2
20
Poba Sanctuary
2
2
21
Ramnabagan Sanctuary
2
2
22
Buxa Sanctuary
2
1
23
Dampa Sanctuary
2
1
24
Dudhwa National Park
2
1
25
Jaldapara Sanctuary
2
1
26
Katarniyaghat Sanctuary
2
1
27
Kaziranga National Park
2
1
28
Manas Sanctuary
2
1
29
Mouling National Park
2
1
30
Nakti Dam Sanctuary
2
1
31
Nellapattu Sanctuary
2
1
32
Puliebadze Sanctuary
2
1
33
Satkosia-Gorge Sanctuary
2
1
34
Semarsot Sanctuary
2
1
35
Yangoupokpi-Lokchao Sanctuary
2
1
36
Khangchendzonga National Park
1
32
37
Kolleru Sanctuary
1
26
38
Gobindsagar Sanctuary
1
11
39
Simlipal Sanctuary
1
10
1
Systematic Assessment of Forest Cover Change and Forest Fragmentation…
205
Table 4. (Continued)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
SNo
Protected area
Cateogry (critical forest loss) 1
Pixel sum
40
Dalma Sanctuary
7
41
Pong Dam Lake Sanctuary
1
7
42
Papikonda Sanctuary
1
6
43
Mount Abu Sanctuary
1
5
44
Singba Sanctuary
1
5
45
Fambong Lho Sanctuary
1
4
46
Great Indian Bustard (extension) Sanctuary
1
4
47
Simbalbara Sanctuary
1
4
48
Simlipal National Park
1
4
49
D' Ering Memorial Sanctuary
1
3
50
Dibang Sanctuary
1
3
51
Kistwar National Park
1
3
52
Naina Devi Sanctuary
1
3
53
Rupi Bhaba Sanctuary
1
3
54
Anamalai Sanctuary
1
2
55
Ashok Musk Deer Sanctuary
1
2
56
Bir Motibagh Sanctuary
1
2
57
Darjeeling PA1 (precise name unknown)
1
2
58
Hadgarh Sanctuary
1
2
59
Kaimur Sanctuary
1
2
60
Kugti Sanctuary
1
2
61
Megamalai Sanctuary
1
2
62
Nanda Devi National Park
1
2
63
Nandur-Madhameshwar Sanctuary
1
2
64
Palamau National Park
1
2
65
Tundah Sanctuary
1
2
66
Badalkhol Sanctuary
1
1
67
Baisipalli Sanctuary
1
1
68
Bannerghatta National Park
1
1
69
Barsey Rhododendron Sanctuary
1
1
70
Bir Bunerheri Sanctuary
1
1
71
Changthang Sanctuary
1
1
72
Hazaribagh Sanctuary
1
1
73
Indravati National Park
1
1
206
A Giriraj, P.K. Joshi, Shilpa Babar et al. Table 4. (Continued)
SNo
Protected area
Cateogry (critical forest loss) 1
Pixel sum
74
Kangerghati National Park
1
75
Kedarnath Sanctuary
1
1
76
Lawalong Sanctuary
1
1
77
Maenam Sanctuary
1
1
78
Nagarahole National Park
1
1
79
Orang Sanctuary
1
1
80
Pench National Park
1
1
81
Raksham Chitkul Sanctuary
1
1
82
Ranganathittu Sanctuary
1
1
83
Renuka Sanctuary
1
1
84
Sechu Tuan Nala Sanctuary
1
1
85
Sunabeda Sanctuary
1
1
86
Udanti Sanctuary
1
1
87
Valley of Flowers National Park
1
1
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Cateogry 1-Low forest area loss; 2-Medium forest area loss; 3-High forest area loss
Evaluating Critical Areas using High-Spatial Satellite Data This evaluation was carried out on a test site in forested taluk of Visakhapatnam district, Andhra Pradesh which is part of Eastern Ghats. In 1973, area under forest cover was about 3,019 sq km (55%) out of the total study area of 5,445 sq km. But there was a significant change in forest cover in three decades and the forest cover reduced to 34% (1,875 sq km). Major area loss was from open forest, which has shown a gradual decrease from 1,006 sq km to 779 sq km in 1990 and then to 465 sq km in 2003 (Fig. 6 and Table 5). The total loss of open forest was from 18.4% to 8.5%. Dense forest area was reduced from 1,651 sq km to 1,552 sq km during 1973-1990 and to 1,237 sq km in 2003. An unbiased set of testing samples has been selected from the reference field data for assessing the classification accuracy, which comes out to be 72.4% for MSS (1973), 72.0% for TM (1990) and 79.1% for LISS-III (2003) data respectively. The kappa indices for the 1973, 1990 and 2003 maps were 0.66, 0.67 and 0.74 respectively. Overall the users and producers accuracy was high which is sufficient for evaluation of land use and land cover changes. With the clear knowledge of forest cover change (Plate 1), rate of deforestation was computed and shown in Table 6. Rate of deforestation is showing increasing trend from 1990-2003 as compared to 1973-1990. During 1973-1990 the change is minimal with less than 1% except open forest (1.59). But during 1990-2003 the rate is very high for all the classes’ maximum being the grassland (6.23). Although forest conservation policies such as VSS through Joint Forest Management and biodiversity monitoring are operative, the trend of deforestation is similar.
Systematic Assessment of Forest Cover Change and Forest Fragmentation…
207
Table 5. Land-use and Land-cover for the Paderu and Chintapalle taluk in Visakhapatnam district of Northern Andhra Pradesh (Area in sq km) Land use class Dense Forest Open Forest Degraded Forest Sub-total Grassland Barren Water Agriculture Sub-total Total
1973 1651.32 1006.20 361.48 3019.00 253.22 43.05 59.00 2071.82 2426.82 5445.82
1990 1552.20 779.64 386.36 2718.20 216.78 215.71 39.88 2255.26 2727.63 5445.83
2003 1237.48 465.56 172.08 1875.12 90.59 180.05 40.87 3259.19 3570.70 5445.82
Plate 1. Pictorial representation of forest clearance in Visakhapatnam district. A. Google images showing the forest cleared site mainly on the slopes; B. Clearing the shrubs and underground by setting fires; C. Complete removal of land and ready for cropping (left to right).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 6. Rate of Deforestation for different land cover classes observed from 1973 to 2003 Land Cover Dense Forest Open Forest Degraded Forest Grassland
1973-1990 0.39 1.59 0.42* 0.97
1990-2003 1.62 3.68 5.78 6.23
1973-2003 0.96 2.57 2.47 3.43
*
Rate of increase
The forest area under fragmentation showed varied changes in different categories from 1973-2003 (Table 7). While the interior category (intact forest) decreased from 4,198.73 sq km (77%) to 1,639.19 sq km (30%), the other categories shows increasing trend (patch 767.8 to 1,873.3 sq km & perforated 217.8 to 756.9 sq km). There appear to be significant change in the edge category (i.e., increase in patch edge) with the area increasing 10-fold from 0.8 to 10.3% in three decades.
208
A Giriraj, P.K. Joshi, Shilpa Babar et al. Table 7. Fragmentation categories for the forest area using fragmentation model
Class Intact Patch Perforated Undetermined Transitional Edge
1973 77.1 14.1 4 2.3 1.7 0.8
1990 56.1 23.2 9.2 4.8 2.9 3.8
2003 30.1 34.4 13.9 7.1 4.2 10.3
Table 8a. District-wise forest cover assessment of Nagaland state using satellite data for the year 2005
District
Geographic Area
Dimapur Kohima Mokokchung Mon Phek Tuensang Wokha Zunheboto Total
758 3,283 1,615 1,786 2,026 4,228 1,628 1,255 16,579
Very Dense 0 29 1 11 65 120 1 9 236
Forest Cover ( sq km) Moderate Open Scrub Dense Forest 140 266 6 1,051 1,765 0 458 966 1 491 820 0 835 786 1 1,711 1,660 4 406 1042 1 510 576 0 5602 7881 13
Total Forest 406 2845 1425 1322 1686 3491 1449 1095 13719
Percent Area 53.56 86.66 88.24 74.02 83.22 82.57 89.00 87.25 82.75
Source: Forest Survey of India, SFR-2005
Table 8b. Forest cover change area matrix of Nagaland state in 2001 vs. 2005 2003 Assessment (Data of Dec. 2002 and Jan-Feb. 2003)
Total 2001
Very Dense
Moderate Dense
Open
Scrub
Nonforest
236
0
0
0
0
236
Moderate Dense Forest
0
5,602
0
0
258
5,860
Open forest
0
0
7,147
0
772
7,919
Scrub
0
0
0
13
218
231
Non-forest
0
0
734
0
1,599
2,333
Total 2005
236
5,602
7,881
13
2,847
16,579
Net change
0
-258
-38
-218
514
Very Dense Forest Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Forest Cover Assessment (Area sq km)
Source: Forest Survey of India, Assessment Report 2005
Systematic Assessment of Forest Cover Change and Forest Fragmentation…
209
Satellite image for the Nagaland state (Landsat-MSS 145/42 of 1974 and Landsat-ETM+ 135/42 of 2001) was used for validating the MODIS-VCF detected Rapid Ecological Alerts. We observed a large-scale deforestation for a temporal gap of 27 years (Fig. 7). Table 8a provides district wise forest cover area distribution for the Nagaland state. Nearly 46% of the total forest area is covered by two districts viz., Kohima and Tuensang. State contributes a total forest area of 13,719 sq km of which dense forest covers 5838 sq km (42.55%) and open forest covers 7881 sq km (57.45%). Forest cover change matrix was carried out between 2003 - 2005 using IRS satellite data. A total of forest loss 258 sq km observed in the moderate dense forest, 38 sq km in open forest and 218 sq km in scrub vegetation respectively (Table 8b).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
DISCUSSION The algorithms for change detection developed for the MODIS Vegetative Continuous Field (MODIS 44B-VCF) product require a time series of daily data composite into monthly data sets. The cases presented include the detection of deforested areas. However, the developed analogy is also true for any other kind of forest cover change and fragmentation because of external factors like forest fire, floods etc. To map the changes and fragmentation with much better accuracy following issues are need to be addressed (i) Misregistration errors in the composite may cause larger commission errors and/or omission errors; (ii) the look up table generated with Landsat, LISS or SPOT data required for detecting changes must be updated with MODIS data because of narrower spectral banks on the MODIS instrument and finally (iii) identifying the threshold method for identifying forest and non-forest classes need to be converged the strong ground truth exercise. Deforestation particularly in tropical countries, mainly in Asia is primarily due to the socio-economic factors and expansion of agricultural land related to the increase in population density. However, the relative importance of each of this varied for different regions (Bawa and Dayanandan, 1997). Other factors like infrastructure development, communication systems also increase the rate of deforestation. The technique presented here is capable to map these changes annually. In recent years, automated classification methods of satellite images have been improved. Their purpose is to reduce interpreter bias. Nevertheless, hardware, software, and specialized personnel are necessary. Regardless of method used to elaborate the coverage maps, the process must be consistent when establishing a monitoring system. As on date, MODIS data and Landsat datasets are available free of cost. Most of the time, these datasets are geometrically and radiometrically calibrated. The methodology presented here is promising for these applications and demonstrate how a suite of integrated results provide a more robust detection scheme than can be provided by a single algorithm. The primary limitations to large area monitoring with coarse resolution satellite systems includes the development of generic and robust methods, overcoming quality issues (commission and omission errors) and having a systematic ground level utilisation. On a methodological level, this study provides an integrated approach with multi-scale remote sensing data analysis of vegetation cover, useful in evaluating the forest dynamics and fragmentation in the region. We presented a comprehensive validation of analysis of MODIS data. We focused our analysis on the deforestation hotspots (North East Region – NER) of
210
A Giriraj, P.K. Joshi, Shilpa Babar et al.
Indian subcontinent. While this study has presented a new systematic approach to mapping information on deforestation, the proposed Rapid Ecological Alerts (RETs) need to be automated for every year.
CONCLUSIONS
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The study demonstrated a systematic assessment of forest cover change and fragmentation in Indian subcontinent using multi-scale satellite remote sensing inputs. The study has also added few dimensions by accounting the changes over past 6 years at state/province level, in different biogeographical zones and among the various identified and proposed protected areas. The validation at the local level has provided strengths to the assessment technique and developed analogy. However, more intensive local level assessment in different geographical climatic regions may further enhance the acceptability and utility of this work. In the last decades various attempts were taken by the forest departments to check the deforestation and carry out plantation activities in the region under afforestation programme (Anonymous, 2002d). Still there are evidences of loss of forest in some of the pockets (Srivastava et al., 2002; Joshi et al., 2008; Reddy et al., 2008) of Indian subcontinent. There is requirement of development of Rapid Ecological Alerts (REA) as suggested in the present study to check the shrinking forests areas. In addition to protecting the native flora within patches, ecological restorative efforts should be applied to critical patches, therefore, creating better connectivity between them. This will help not only to check the forest loss but also from losing the biodiversity of these forests. These kinds of studies are increasingly useful to understand the dynamics of forest for the sustainable management of forest ecosystems. The presented technique is capable to map these various kinds of changes. In recent years, automated classification methods of satellite images have been improved. Their purpose is to reduce interpreter bias. Nevertheless, hardware, software, and specialized personnel are necessary. Regardless of method used to elaborate the coverage maps, the process must be consistent when establishing a monitoring system. As on date, MODIS data and Landsat datasets are available free of cost. Most of the time, these datasets are geometrically and radiometrically calibrated. The methodology presented here is promising for these applications and demonstrate how a suite of integrated results provide a more robust detection scheme than can be provided by a single algorithm.
ACKNOWLEDGMENTS We are thankful DOS-DBT funded project on ‘Biodiversity Characterization at Landscape Level Project’ on the use of canopy cover datasets. We thank the Director and Deputy Director of National Remote Sensing Agency (NRSA), India for their constant encouragement and support. We are grateful to Dr. M.S.R. Murthy, Forestry and Ecology Division, NRSA for his suggestions and improvement.
Systematic Assessment of Forest Cover Change and Forest Fragmentation…
211
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REFERENCES Achard, F., Eva, H.D, Stibig, H., Mayaux, P., Gallego, J., Richards, T. and Malingreau, J.P. (2002). Determination of deforestation rates of the world’s humid tropical forests. Science 297 (5583): 999-1002. Anonymous. (2002a). Biodiversity Characterization at Landscape Level in North-East India using Satellite Remote Sensing and Geographic Information System. Indian Institute of Remote Sensing (NRSA), Dehradun. Anonymous. (2002b). Biodiversity Characterisation at Landscape Level in Western Himalayas India using Satellite Remote Sensing and Geographic Information System. Indian Institute of Remote Sensing (NRSA), Dehradun. Anonymous. (2002c). Biodiversity Characterization at Landscape Level in Western Ghats India using Satellite Remote Sensing and Geographic Information System. National Remote Sensing Agency (NRSA), Hyderabad. Anonymous (2002d). http://www.etagriculture.com/jan_feb2002/indo.html. Anonymous. (2007a). Biodiversity Characterization at Landscape level in Eastern Ghats using Satellite Remote Sensing and Geographic Information System. National Remote Sensing Agency, Hyderabad. Anonymous. (2007b). Biodiversity Characterization at Landscape level in Central India and West Bengal using Satellite Remote Sensing and Geographic Information System. National Remote Sensing Agency, Hyderabad. Bawa, K.S. and Dayanandan, S. (1997). Socioeconomic factors and tropical deforestation. Nature 386: 562–563. Census of India. (2001). Census of India, Govt. of India, New Delhi India. Champion, H.G., and Seth, S.K. (1968). A Revised Survey of Forest Types of India. New Delhi Government Publication, New Delhi. DeFries, R.S., Field, C.B., Fung, I., Justice, C.O., Los, S., Matson, P.A., Matthews, E., Mooney, H.A., Potter, C.S., Prentice, K., Sellers, P.J., Townshend, J.R.G., Tucker, C.J., Ustin, S.L. and Vitousek, P.M. (1995). Mapping the land surface for global atmospherebiosphere models: Toward continuous distributions of vegetation's functional properties. J. Geophys. Res., 100: 20867-20882 Forestry Survey of India [FSI]. (2005). State of Forest Report (SFR). Ministry of Environment and Forests. Govt. of India, Dehra Dun Geist, H.J. and Lambin, E.F. (2002). Proximate causes and underlying driving forces of tropical deforestation. BioScience 52(2): 143-50. Giglio, L., van der Werf, G.R., Randerson, J.T., Collatz, G.J, and Kasibhatla, P. (2006). Global estimation of burned area using MODIS active fire observations. Atmos. Chem. Phys. 6: 957-974. Goward, S.N., Markham, B., Dye, D.G., Dulaney, W. and Yang, J. (1991). Normalized difference vegetation index measurements from the advanced very high resolution radiometer. Remote Sensing of Environment 35: 257-277. GRASS, 2008 Geographic Resources Analysis Support System Software (GRASS). GRASS Development Team, ITC-irst, Trento, Italy. Hansen, J., R. Ruedy, Mki. Sato, and Lo, K. (2002). Global warming continues. Science 295, 275, doi:10.1126/science.295.5553.275c.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
212
A Giriraj, P.K. Joshi, Shilpa Babar et al.
Hansen M C, DeFries R S, Townshend J R G, Carroll M, Dimiceli C and Sohlberg R A 2003 Global per cent tree cover at a spatial resolution of 500 m: first results of the MODIS vegetation continuous fields algorithm. Earth Interact. 7:10 IGBP, 1998. International Geosphere Biosphere Programme (IGBP), Terrestrial Carbon Working Group (Steffan, W., Noble, I., Canadell, P., Apps, M.J., Schulze, E.-D., Jarvis, P.G., et al.), 1998. The terrestrial carbon cycle: implications for the Kyoto protocol. Science 280, 1393–1394. Joshi, P.K., M. Kumar, A. Paliwal, N. Midha, and P. P. Dash (2008). Assessing impact of industrialization in terms of LULC in a dry tropical region (Chhattisgarh), India using remote sensing data and GIS over a period of 30 years. Environmental Monitoring and Assessment DOI 10.1007/s10661-008-0211-z. Joshi, P.K., Roy, P.S., Singh, S., Agrawal, S. and Yadav, D. (2006). Vegetation cover mapping in India using multi-temporal IRS Wide Field Sensor (WiFS) data. Remote Sensing of Environment 103(2): 190-202. Kalacska, M., Sanchez-Azofeifa, G.A, Rivard, B., Calvo-Alvarado, J.C. and Quesada, M. (2008). Baseline assessment for environmental services payments from satellite imagery: A case study from Costa Rica and Mexico. Journal of Environmental Management 88 (2): 348-359. Kauffman, J. B., Cummings, D. L. and Ward, D. E. (1998). Fire in the Brazilian Amazon: 2. Biomass, nutrient pools and losses in cattle pastures. Oecologia 113: 415-427. Kiran Chand T.R., Badarinath K.V.S., Krishna Prasad V., Murthy M.S.R., Elvidge C.D. and Tuttle B.T. 2006. Monitoring forest fires over the Indian region using Defense Meteorological Satellite Program-Operational Linescan System nighttime satellite data. Remote Sensing of Environment 103 (2): 165-178. Mayaux, P., Holmgren, P., Achard, F., Eva, H.D., Stibig, H.J. and Branthomme, A. (2005). Tropical forest cover change in the 1990s and options for future monitoring Phil. Trans. R. Soc. B 360: 373–384. Puyravaud J. P. (2003). Standardizing the evaluation of the annual rate of deforestation. Forrest Ecology and Management 177: 593-596. Reddy, C. S., K. Ram Mohan Rao, Pattanaik, C. and Joshi, P. K. (2008). Assessment of Large-Scale Deforestation of Nawarangpur District, Orissa, India for its Sustainable Management: A Remote Sensing Based Study. Environmental Monitoring and Assessment (in press). Riitters, K., Wickham, J., O'Neill, R., Jones, B. and Smith, E. (2000). Global-scale patterns of forest fragmentation. Conservation Ecology 4(2): 3. Rodgers, W.A. and Panwar, H.S. (1988). Biogeographical classification of India. Wildlife Institute of India, Dehra Dun. Rosenqvist, A., Milne, A., Lucas, R., Imhoff, M. And Dobson. C. (2003). A review of remote sensing technology in support of the Kyoto Protocol. Environmental Science & Policy 6: 441-455. Roy, P.S., Joshi, P.K., Deepshikha Yadav, S. Agrawal, S. Singh, and C. Jegannathan (2006). Biome mapping in India using multi-temporal satellite data and other inputs. Ecological Modeling 197(1-2): 148-158. Sellers, P.J., Dickinson, R.E., Randall, D.A., Betts, A.K., Hall, F.G., Berry, J.A., Collatz, G.J., Dennings, A.S., Mooney, H.A., Nobre, C.A., Sato, N., Field, C.B, and Henderson-
Systematic Assessment of Forest Cover Change and Forest Fragmentation…
213
Sellers, A. (1997). Modeling the Exchanges of Energy, Water, and Carbon Between Continents and the Atmosphere. Science 275(5299):502-9 Srivastava, S.K., Singh, T.P., Singh, H., Kushawaha, S.P.S. and Roy, P.S. (2002). Assessment of large scale deforestation in Sonitpur district of Assam. Current Science 82 (12): 1484– 1497. Townshend, J.R.G. and Justice, C.O. 1988. Selecting the spatial resolution of satellite sensors required for global monitoring of land transformations. International Journal of Remote Sensing 9: 187-236. Townshend, J.R.G., Justice C.O., Skole, D.,, C.O., Malingreau, J.-P., Cihlar J., Teillet, P., Sadowski, F. and Ruttenberg, S. 1994. The 1 km resolution global data set: needs of the International Geosphere Biosphere Program. International Journal of Remote Sensing 15(17): 3417-3442. Zhan, X., DeFries, R., Townshend. J. R. G, Dimiceli, C., Hansen, M., Huang, C., and Sohlberg, R. (2000). The 250m global land cover change product from the Moderate Resolution Imaging Spectroradiometer of NASA’s Earth Observing System. International Journal of Remote Sensing 21(6&7): 1433–1460.
QUESTION BANK Expand the following 1. 2. 3. 4.
CDM GLC GLCF GRASS
5. 6. 7. 8.
IRS JI LISS RET
9. UNEP 10. VCF
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Short Answer Questions 1. What is use of regionally optimized model to map the vegetation cover? 2. Write specifications of MODIS data. What are its advantages over other satellite data? 3. What is classification? Enlist different algorithms of satellite data classification with special emphasis on maximum likelihood classifier. 4. What is change detection? How changes in land use land cover could be quantified? 5. What is forest fragmentation? What is the importance of quantification of forest fragmentation?
Long Answer Questions 1. What are different challenges to forest cover in your country? Explain some of the national initiatives to understand forest cover dynamics. 2. Explain different global forest cover assessment initiatives. What are the challenges of using satellite remote sensing data for forest cover mapping and monitoring?
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 10
SPATIOTEMPORAL DYNAMICS OF LAND USE/LAND COVER AND TIMBER CARBON STORAGE: A CASE STUDY FROM BULANIKDERE, TURKEY Fatih Sivrikaya1* Günay Çakir2†, Sedat Keleş1‡ and Emin Zeki Başkent1# 1
Faculty of Forestry, Karadeniz Technical University, 61080 Trabzon, Turkey, 2 Faculty of Forestry, Düzce University, 81620 Düzce, Turkey
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ABSTRACT Land use/ land cover change (LULCC) is regarded as the most important variable of global change affecting ecological systems with an impact on natural resources including soil and water quality, global climatic system and biodiversity. Quantification of landscape patterns through various landscape metrics can be used to describe whether the landscape pattern is changing and how, and this information may then be used to evaluate alternative ecosystem management activities and environmental conservation strategies. To analyze the spatial and temporal pattern of land use/land cover change in a forest planning unit, Bulanıkdere, northwestern corner of Turkey, was selected as a case study area. The main data used in this research are forest cover type maps at 1/25 000 scale for years 1985 and 2003, their inventory data, two Landsat satellite image acquired for the years 1987 (TM), and 2000 (ETM+). Landsat TM (1987) and Landsat ETM+ (2000). The spatial and temporal analysis of the forest structure was carried out in a systematic process. First of all, the stand type maps 1985 and 2003 were digitized, rectified and the spatial database was build with ArcGIS 8.3TM. Second, Landsat TM and ETM+ images of 1987 and 2000 were rectified and classified to create land use/ land cover type maps. Third, spatial and temporal land use/land cover changes as well as transition of cover types were analyzed with GIS. Fourth, the fragmentation of the areas was evaluated with *
Email: [email protected] [email protected] ‡ [email protected] # baş[email protected] †
216
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al. FRAGSTATSTM. Last, spatially explicit estimates of the carbon storage (aboveground and belowground) was determined and mapped with GIS according to inventory data for years 1985 and 2003. When both approaches were evaluated together, between 1985 and 2003 years, forest areas decreased from 7432.9 ha to 7351.5 ha according to stand type maps from management plans. On the contrary, between 1987 and 2000 years, forest areas increased from 7049,4 ha to 7456,7 ha according to land cover type maps from the classified images of Landsat TM and ETM+. During the whole study period, the average annual deforestation rate based on stand type map was 5 ha year-1, equivalent to 0.06% year-1 using the compound-interest-rate formula. Landscape analysis indicate that the forest landscape has been gradually fragmented and thus subjected to susceptibility of the area to further abrupt changes in the future. It was estimated that forest ecosystems in Bulanıkdere forest planning unit contain 304 264.3 tons of carbon above and below the ground according to the methodology used in this article in 1985. Though 263 895.9 tons of whole carbon storages in forest ecosystem are aboveground, the rest (40 368.4 tons) are belowground. However, it was estimated that forest ecosystems contain 510 030.1 tons of carbon with 443 032.5 tons aboveground and 66 997.6 tons of carbon belowground in 2003. As shown, carbon storages increased 205 765.8 tons during an 18 year period. There were apparent trends in the temporal structure of forest landscape, some of which may issue from unregulated management activities, social conflict and demographic movement. The study revealed that it is important to understand both spatial and temporal changes of land use/land cover and their effects on landscape pattern to release the implications for landscape planning and ecosystem management.
Key words: Land use/land cover, GIS, Landsat, Carbon storage, Biomass, Fragmentation
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
INTRODUCTION Increasing demand for intensive use of natural resources has mainly affected the land use pattern of globe. The amount and spatial distribution of land cover properties have changed depending on the rate, intensity and extent of land use activities. The capacity and capability of land cover to sustain the livelihood of all organisms on earth depend on the understanding of land cover/use dynamics over time. As such, the topic of Land Use/Land Cover Change (LULCC), especially forest cover change, has become a focal issue at local and global scales. LULCC is a dynamic, widespread and accelerating process, mainly driven by human and physical environments, economic condition, natural phenomena and anthropogenic activities, which in turn drive changes that would impact humans. Ecosystem dynamic in landscape structure provides an integrated approach and opportunity to understand the relationship between landscape pattern and changes in environmental conditions (URL-1 2008). In addition to simple area coverage, the structure of land use patches is an important characteristic for evaluating the processes and effects of land use change (Gautam et al., 2003). Ecosystem structure refers to the spatial characteristics of ecosystem patches including their size, shape, composition, and spatial arrangement. Ecosystem function refers to the ecological processes and relationships that exist within an ecosystem based on its structural characteristics. Forest dynamics, however, refers to the change of forest structure and its functions over time (Pickett and White, 1985; Levin et al., 1993; Başkent and Jordan, 1995a; 1995b; Meffe and Carroll, 1997; DiBari, 2007). Understanding these factors is critical to sustainable management of natural resources.
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage… 217 The causes of LULCC are generally divided into a few groups such as biophysical, economic and technological, demographic, institutional and cultural factors (Geist et al., 2006). On the other hand, LULCC have multiple impacts on natural resources including soil and water quality, global climatic system and biodiversity (Iida and Nakashizuka, 1995; Johnson et al., 1997; Chen et al., 2001; Dupouey et al., 2002; Upadhyay et al., 2005; Liu et al., 2006). Furthermore, LULCC related studies provide critical inputs to large-scale vegetation biomass and forest cover assessments that are key components of the carbon cycle. Future LULCC goals include very accurate biomass estimates to refine knowledge of carbon storage in vegetation, understanding regional land-use changes that affect biomass, and quantifying linkages and feedbacks between land-use and land-cover change, climate change, and other related human and environmental components (URL-1, 2008). Depending on management regimes, forests can play a vital role in either sequestering or releasing atmospheric carbon. While forests in most temperate forests are net carbon sinks, tropical forests account for about one third of the global carbon emissions (Sasaki and Yoshimoto, 2007). LULCCs are particularly severe in the tropics and have significant global impacts on forest use particularly biodiversity. Habitat fragmentation with forest loss has been recognized as a major threat to ecosystems worldwide (Armenteras et al., 2003; Laurance, 1999; Noss, 2001). These processes may have negative effects on biodiversity, by increasing isolation of habitats (Debinski and Holt, 2000), threatening endangered species, modifying species’ population dynamics (Watson et al., 2004), and increasing edge effects at the expense of interior habitat (Mace et al., 1998). Fragmentation may also have negative effects on species richness by reducing the probability of successful dispersal and establishment (Gigord et al., 1999) as well as by reducing the capacity of habitat patches to sustain a resident population (Iida and Nakashizuka, 1995). The ecological consequences of fragmentation may differ depending on the patterns or spatial configuration imposed on a landscape and how it varies both temporally and spatially (Ite and Adams, 1998; Armenteras et al., 2003). Therefore, an understanding of the relationship between landscape patterns and the ecological processes influencing distribution of species is required by resource managers to provide a basis for making appropriate land use decisions (Turner et al., 2001). There are various methods that can be used in the collection, analysis and presentation of natural resources data. The use of remote sensing (RS) and geographic information system (GIS) technologies can greatly facilitate the process. Satellite images and aerial photographs are useful for both visual assessments of natural resources dynamics occurring at a particular time and space as well as quantitative evaluation of land use/land cover changes overtime (Gautam et al., 2003). In explaining LULCC dynamics, remote sensing techniques using satellite imagery are effective methods to document the rates and patterns of change in forest ecosystems and can provide information to help resolve controversies about future management directions (Cohen et al., 1995). Satellite images provide a meaningful method for detecting landscape structure and its change over time (Çakır et al., 2007; Keleş et al., 2008; Başkent and Kadıoğulları, 2007; Kadıoğullari and Başkent, 2008; Jarvis, 1994). Existing forest cover type maps may be used for determining land use/land cover changes overtime (Sivrikaya et al., 2007a). Few studies have focused on the analysis of landscape quality change, i.e., temporal transitions of development stages, cover types and crown closures as indicators of stand quality (Sivrikaya et al., 2007a; Kadıoğulları, 2005), biomass and carbon storage (Sivrikaya et al., 2007a; Sivrikaya et al., 2007b).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
.
218
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
The objectives of this research are to: (i) map the distribution of land use/land cover class over a large area of Bulanıkdere forest; (ii) map landscape change over a 20-year period; (iii) describe spatial landscape patterns with varying management and disturbance regimes focusing on forest fragmentation; (iv) map biomass and carbon storage and; (v) consider the implications of the rates and patterns of change on conventional management activities that create future forest conditions.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REVIEW OF LITERATURE Applications of RS and GIS to illustrate spatial and temporal changes in land use/land cover over time have been reported by many investigators. The studies documented the spatial and temporal LULCC as well as the factors affecting the processes. There have been many studies attempting to document the change in land use and land cover as well as forest cover (Luque et al., 1994; Sachs et al., 1998; Bürgi, 1999; Rao and Pant, 2001; Evelyn and Camirand, 2003; Kennedy and Spies, 2004; Bray and Klepeis, 2005; Emch et al., 2005; Wakeel et al., 2005; Matsushita et al., 2006; Jansen et al., 2007; Haase et al., 2007; Feranec et al., 2007; Otto et al., 2007). On one hand, the effects of LULCC on carbon storage and sequestration (Evrendilek, 2004; Xu et al., 2007; Muukkonen and Heiskanen, 2007; Sharma and Rai 2007), on water quality and quantity (Werbist et al., 2005; Thanapakpawin et al., 2006; Siriwardena et al., 2006; Xian et al., 2007; Lathrop et al., 2007; Lin et al., 2007; Shi et al., 2007), on biodiversity (Turner et al., 2003; DeWalt et al., 2003; Cayuela et al., 2006; McIntyre and Lavorel, 2007; Lele et al., 2007; Jameson and Ramsay, 2007; Zhao et al., 2007; Pueyo and Alados, 2007), on soil erosion (Erskine et al., 2002; Ali, 2006; Szilassi et al., 2006) have been studied. On the other hand, there have been several studies focusing on the impacts of urbanization on LULCC (Negi et al., 1999; Bürgi et al., 2000; Zhang and Song, 2006; Xu et al., 2007; Abdullah and Nakagoshi, 2007; Weng, 2007), the impacts of grazing on LULCC (Zhao et al., 2007; McEvoy et al., 2007). There have been few studies attempting to document the change in landscape pattern for the Turkish forests over time. Köse and Başkent (1996; 2002) investigated the 40- year legacy of forest management plans in both Mediterranean and eastern Black Sea forests focusing on just the areal distribution of forest resources without the use of a spatial information technology. They showed that considerable changes occurred in planning parameters between two consecutive periods: the forest landbase and allowable cut levels decreased and the growing stock increased. Özdemir and Özkan (2003) documented the 10-year change in forest structure of Armutlu forest planning unit with Landsat images. Yıldırım et al. (2002) investigated 15-year change in land use/land cover in Gebze/Kocaeli. Tunay and Ateşoğlu (2004) documented the land use change between 1992 and 2000 in managed forests of Bartın with GIS and RS. Sivrikaya et al. (2007) investigated 33-year temporal and spatial change and fragmentation in Camili/Artvin. Sivrikaya et al. (2007) documented temporal biomass and carbon storage change in two different planning units. Başkent and Kadıoğulları (2007) investigated spatial and temporal change in İnegöl, Turkey. Kadıoğulları and Başkent (2007) documented temporal and spatial change in Gümüşhane, Turkey. A number of other related research initiatives are ongoing as well. NASA has just launched a new project to determine and monitor the carbon balance in Black Sea region.
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage… 219 Kilic et al. (2006) investigated the land use change between 1972 and 200 years in Hatay with Landsat images. A collaborative work with Istanbul Technical University, Russian Science Academy and Bristol University is ongoing to monitor urbanization and land use changes at the Marmara sea coast using both satellite and conventional field data. Very few of these research endeavors focused on a large-scale analysis of spatial forest structure over long time with information technologies.
Study Area The study area is in the Bulanıkdere forest planning unit, characterized by a flat terrain with an average slope of 12% and an altitude from 0 to 380 m above sea level. It is situated in the northwestern corner of Turkey (570000-585000 E, 4626000-4642000 N, UTM ED 50 Datum Zone 35N) (Fig. 1). Winters are mild and wet, and summers are relatively cool and dry. Mean annual temperature of the study area is 8-15 0C, and mean annual precipitation is 962 mm. Main soil types are sandy clay loam, clay loam and sandy loam (Stojchev et al., 1998). The sand dunes, sea water, lagoons (lakes), swamp, forest and riparian ecosystems are the major ecosystems of Bulanıkdere. In 1991, Nature Conservation Foundation (NCF) reported that the Bulanıkdere forest ecosystems are home for many birds and plants. Thus, the area was designated by NCF as one of the important bird (especially for black stork) and plant areas in Turkey. For example, there are 11 rare plant species like Logfia minima, Centaurea arenaria, Jurinea kilaea, and Trifolium bocconei which can be found in İğneada and Bulanıkdere Forest Planning Unit. Additionally, Aurinia uechtritziana, Salvinia natans, Silene sangaria, Trapa natans and Verbascum degenii, which are also listed in Bern Convention Categories (1979), have natural distribution in Bulanıkdere forests. The best example of ash-oak-black or common alder forest community types in Turkey, also dominating forest species in the Bulanıkdere lagoons, can be found here. In moist areas, black or common alder (Alnus glutinosa subsp. barbata) and ash (Fraxinus angustifolia subsp. oxycarpa) dominate while Quercus robur and other oak species (such as, Q. frainetto, Q. cerris var. cerris, Q. hartwissiana and Q. paetraea subsp. petraea) are the major tree species in relatively dry areas (Başkent et al., 2008)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Data Sources The main data used in this research are forest cover type maps at 1/25 000 scale for years 1985 and 2003 with forest inventory, two Landsat satellite image acquired for the years 1987 (TM) and 2000 (ETM+). A brief description of the satellite images used is given in Table 1. Landsat TM (1987) and Landsat ETM+ (2000) images were downloaded from the Global Land-Cover Facility site hosted by the University of Maryland (URL-2 2008). Landsat images were acquired during the dry season, which enables us to ensure that the images are completely cloud free, and also allow us to differentiate forest from non forest areas (mostly agriculture, settlements and dry grasslands at this time of the year), with a greater degree of
220
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 1. The location of the study area
accuracy (Nagendra et al., 2006; Voght et al. in press). The image was the highest quality, being both defect-free and cloud free. Summer and early fall image (May and July) data were used for this study, in order to separate the softwood and hardwood stands through leaf-on canopy differences rather than through the strong seasonal differences in reflection found during leaf-seasons. Topographic maps of 1:25000 scale produced by the General Command of Mapping of Turkey were used for rectification. Forest cover type maps, originally generated from both the stereo interpretation of infrared aerial photos with an average 1/16000 scale and ground measurements with 300 x 300 grids, were used for ‘ground-truth’ information required for classification and accuracy estimation of classified TM and ETM+ images. Table 1. Description of Satellite images Land Use/Land Cover Classes Softwood (SW) Hardwood (HW)
Description Forest areas with pure softwood trees Forest areas with pure hardwood trees
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage… 221 Table 1. Description of Satellite images (Continued) Land Use/Land Cover Classes Softwood-hardwood ( SW- HW) Hardwood-softwood ( HW -SW) Hardwood-hardwood ( HW- HW) Degraded forest (DF) Forest openings (FO) Agriculture-settlement (AS) Swamp (SM) Sandy area (SA) Water (W) Other
Description Mixed forest areas dominated by softwood trees Mixed forest areas dominated by hardwood trees Mixed forest areas with hardwood trees Degraded forest areas with estimated < 10% tree crown cover Treeless areas Agricultural lands and settlement areas Swamp areas Sandy areas Natural and artificial lake, River Forest nursery, Quarry, Forest depot
Methods
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The spatial and temporal analysis of the forest structure was carried out in a systematic process. First of all, the stand type maps 1985 and 2003 were digitized, rectified and the spatial database was build with ArcGIS 8.3TM. Second, Landsat TM and ETM+ images of 1987 and 2000 were rectified and classified to create land use/ land cover type maps. Third, spatial and temporal land use/land cover changes as well as transition of cover types were analyzed with GIS. Fourth, the fragmentation of the areas was evaluated with FRAGSTATSTM. Last, spatially explicit estimates of the carbon storage (aboveground and belowground) was determined and mapped with GIS to complete the analyses.
Geometric Correction of Landsat Images and Digitizing Stand Type Maps Subsets of satellite images were rectified using 1:25000 topographic maps with Universal Transverse Mercator (UTM) coordinate system (ED 50 datum). Registration was carried out using distinctive features such as road intersections and stream confluences that are also clearly visible in the image. A first-degree rotation scaling and translation transformation function and the nearest neighbor re-sampling method were applied. Re-sampling method uses the nearest pixel without any interpolation to create the warped image (Richards, 1994). A total of 32 ground points were used for registration of Landsat TM image with the rectification error of less than 1 pixel. The ETM+ images were registered to the already registered Landsat TM image through image-to-image registration technique. A very high level of accuracy in georeferencing of the images was possible because of the use of digital source as the reference data that allowed zooming to the nearest possible point location (Gautam et al., 2003). The forest stand type maps used in this research were scanned, saved in tiff format and then registered with 5–8m RMS error to the digital topographic maps in the same manner as to the ETM+ image. Rectified forest stand type maps were digitized with a 1/3000 to 1/5000 screen view scale with ArcGIS 8.3TM by a number of qualified GIS educated foresters.
222
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
Image Analysis Image processing, classification and spatial analysis were carried out using the ERDAS Imagine 8.6 TM image processing software and ArcGIS 8.3TM. We used supervised maximum likelihood classification method for the classification of all the images. Training sites were derived from the satellite images using reference maps/stand type map. To deliver the appropriate support size of each category, the required training set for each class was determine at least ten times the number of discriminating variables (e.g., wavebands) used in the classified map. Maximum likelihood classifier assigns a pixel to a particular class based upon the covariance information and a substantially superior performance is expected from this method compared to other approaches (Richards, 1994). We gathered ground reference data from nearly 78 ground truthing points as signatures for each satellite image. The points were sampled on the cover type (stand) maps of Bulanıkdere forest planning unit that are drafted from aerial photography with 1/150001/20000 scales and finalized from ground measurements by the State Forest Management Teams. The classified images were then checked to control the accuracy using ground data points that are not used in the classification process. Furthermore, these signatures were further controlled with image enhancement techniques such as transformed vegetation index and principle component analysis (PCA). For producing land use/ land cover maps for 1987 and 2000 and to investigate changes that occurred between these periods, the following ten forest cover classes were considered in image classification: (1) Softwood forest (SW); (2) Hardwood forest (HF); (3) Hardwood -hardwood forest (HW-HW); (4) Softwood-hardwood forest (SW-HW); (5) Degraded forest (DF); (6) Forest openings (FO); (7) Agriculture and settlements (AS); (8) Swamp; (9) Sandy areas and (10) Water (W). A brief description of each of the land use/ land cover classes is given in Table 2.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 2. Descriptions of Land Use/Land Cover Classes Satellite Type
Sensor
Bands
Pixel resolution (m)
Observation Date
Landsat 5
TM
7
28.5 × 28.5
11 May 1987
Landsat 7
ETM+
8
28.5 × 28.5
25 July 2000
Accuracy Assessment Accuracy assessment involves identifying a set of sample locations (ground verification points) that are visited in the field. Forest cover type maps are compared to the images for the same location by means of error or confusion matrices. Accuracy assessment was assessed for supervised classification of the 1987 and 2000 images using four measures of accuracy: overall accuracy, user’s accuracy, producer’s accuracy and Kappa coefficient. A total of 30 random points for each class was taken to determine the accuracy of the classification method. Accuracy was assessed in terms of errors of omission (producer’s accuracy) and commission (user’s accuracy) and Kappa coefficient. The overall map accuracy was calculated by dividing the total correct classified pixels, (major diagonal of the error matrix) by the total number of pixels in the error matrix. Overall accuracy did not take into account the proportion of agreement between datasets and it tends to overestimate classification accuracy (Congalton and Mead, 1983). Producer’s accuracy indicates the probability of reference pixel being correctly classified and it is a measure of omission error. User’s
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage… 223
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
accuracy is the probability of classified pixel actually represents that category on the ground. User’s accuracy is a measure of commission error (Jensen, 1996). The Kappa Coefficient measures the proportional improvement of classification over purely random assignment to classes. This accuracy measure attempts to control for a chance agreement by incorporating the off-diagonal elements as a product of the row and column of the error matrix (Cohen 1960). After accuracy assessment, all images were clumped, eliminated 2 × 1 pixels and vectorized in Erdas Imagine 8.6TM programme. These coverages were preprocessed to eliminate areas less than 0.18 ha (corresponding to 3 × 1 pixel) for faster analysis of spatial pattern.
Transition and Spatial Analysis Analyzing the changes in the amount of land use/land cover types, the temporal transitions among the cover types are also documented and evaluated to see the intertemporal dynamics using various parameters of both composition and configuration of forest resources. The transitions were evaluated using periodic results of management plans. Comparison of the change of each land use/land cover types from 1985 to 2003 was conducted by means of transition matrix from one land cover to the other. The transitions among major land cover types for the period were determined based on forest stand maps created as part of forest management plans. The land use/land cover polygon themes for years 1985 and 2003 were overlaid using ArcGIS 8.3TM and the area that changed from each of the classes to any of the other classes was computed. The spatial configuration of landscape structure is important as it has important implications to the design and management of natural resources (Başkent et al., 2000). The spatial dynamics of the forest landscape refers to the temporal change in the size, number, shape, adjacency and the proximity of patches in a landscape. The landscape pattern metrics involve the qualitative and quantitative measurements that expressed the characteristics of the landscape as a proxy measurement. We used a few metrics or measurements to quantify and spatially analyze the change in spatial structure as demonstrated by Başkent and Jordan (1995a; 1995b) and McGarigal and Marks (1995). Choices for appropriate landscape metrics are dependent upon the scale of analysis and objectives of the study (Abdullah and Nakagoshi, 2006; Turner and Gardner, 1991; Forman, 1995; Turner et al., 2003; Farina, 2000). Quantification and comparison of the spatial configuration of forest fragments were conducted based on the following landscape metrics selected after reviewing recent forest fragmentation studies (Echeverria et al., 2006; Cayuela et al., 2006; Franklin, 2001; Armenteras et al., 2003; Millington et al., 2003; Turner et al., 2003): (i) CA; (ii) NP; (iii) MPS; (iv) PERCLAND; (v) LPI; (vi) PD; (vii) PSCV; and (viii) AWMSI. Brief definitions for the selected metrics are listed in Table 3. These landscape pattern metrics represent qualitative aspects of land cover types, expressing the characteristics of the landscape as a whole (Abdullah and Nakagoshi, 2006). The selected metrics are used to characterize the spatial heterogeneity, fragmentation, complexity of patch shape, and connectivity for a given landscape. It is noted that number of patch (NP) and mean patch size (MPS) should be used complementarily since high NP and low MPS values reinforce an interpretation of fragmented landscape conditions (Matsushita et al., 2006; Leitao and Ahern, 2002). In order to study forest fragmentation processes, the land cover maps obtained from stand type maps (1985 and 2003) and Landsat images (1987 and
224
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
2000) were used to determine landscape spatial indices. These indices or spatial metrics were computed by FRAGSTATS (McGarigal et al., 1995) for each of the forest cover class. Table 3. Landscape metrics and their definitions Landscape metrics CA (ha) NP MPS (ha) PERCLAND (%) LPI (%) PD PSCV (%) AWMSI
Descriptions Class area; sum of areas of all patches belonging to a given class, in map units Number of patches Mean patch size; the average patch size within a particular class Percent of landscape Largest patch index; % of the landscape comprised by the largest patch Patch density; number of patches per 100 ha Patch size coefficient of variation Area-weighted mean shape index, the average perimeter-to-area ratio for a class, weighted by the size of its patches
Annual deforestation rates were calculated using the compound-interest-rate formula due to its explicit biological meaning (Puyravaud, 2003). This is,
P=
A 100 ln 2 t 2 − t1 A1
(1)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Where, P is percentage of forest loss per year, and A1 and A2 are the amount of forest cover at time t1 and t2, respectively.
Estimation of Biomass and Carbon In this study, both 1985 and 2003 forest inventory data were used to estimate forest timber biomass in the Bulanıkdere forest planning unit. Inventory data includes area and growing stock per hectare of each forest stand type. Volume and area data of each forest stand type were obtained from Bulanıkdere forest management plans carried out in 1985 and 2003 (Anonymous, 1985; Anonymous, 2003). In this paper, carbon storages of hardwood and softwood species were estimated separately. Biomass for each forest types was calculated using allometric equations from the literature (Asan et al., 2002; Yolasığmaz, 2004; Keleş and Başkent, 2006; Sivrikaya et al., 2007a). To estimate above ground biomass, timber volume of softwoods and hardwoods were multiplied by species-specific conversion factors. These conversion factors were 1.25 for hardwoods and 1.2 for softwoods (Asan et al., 2002). Equations that compute fresh-weight biomass were multiplied by species-specific conversion factors to yield dry-weight biomass. The conversion factors were 0.64 for hardwoods and 0.473 for softwoods (Asan et al., 2002). The root biomass was estimated according to the above-ground biomass. For this reason, the above-ground biomass was multiplied by predetermined root to shoot ratios. These ratios are 0.15 for hardwoods and 0.20 for softwoods (Asan et al., 2002). Total dry weight biomass of a tree was converted to total stored carbon by multiplying by 0.45 (Asan et al., 2002).
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage… 225
Mapping Carbon Storage The GIS presentation of above and below-ground carbon storage in Bulanıkdere was accomplished using the following GIS data; forest cover type maps for Bulanıkdere (for years 1985 and 2003). Forest cover type map of case study area was firstly digitized and processed using ArcGIS 8.3TM with a maximum root mean square (RMS) error under 10 m and spatial database established. Spatial database consists of stand type, crown closure, forest development stages and stand type area. Given the initial spatial database, the stand type volume was added to the database. Above and below-ground carbon storages were calculated using the GIS database including these stand type volumes and produced above and belowground carbon storage maps (m3/hectare) in 1985 and 2003 for Bulanıkdere by reclassifying a map of the forest cover type maps.
Results Accuracy Assessment The overall accuracy for 1987 TM and 2000 ETM+ were 87.92% and 92.33% respectively (Table 4 and Table 5) and the Kappa Coefficients for these maps were 0.8619 and 0.8987 respectively. Landsat TM image (1987) was classified for eight land classes successfully. According to the accuracy assessment results using supervised classification for 1987, forest openings had the lowest producer’s accuracy of 74.3% and degraded forest and hardwood forest had the user’s accuracy of 78.9%. Forest openings, degraded forest and hardwood forest weren’t classified successfully due to lower producer’s accuracy. However, this classification is generally acceptable due to a higher overall classification accuracy and Kappa statistics. Landsat ETM+ image (2000) was classified for ten land classes successfully. For the classification of 2000 land use/land cover map, the lowest values of the producer’s accuracy (78.9%) corresponded to sandy areas. Sandy areas weren’t classified successfully as indicated by the lower producer’s accuracy. However, this classification is generally acceptable due to a higher overall classification accuracy (92.33%) and Kappa statistics (0.8987). The reasons of misclassifications are more evident for the lower resolution TM and ETM+ satellite imagery (Cayuela et al., 2006).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 4. Confusion matrix for the Landsat TM (1987) image supervised classification Class
HW
HW SW-HW DF FO SM AS SA W Total Prod. Acc.
30 8
SWHW
DF
FO
SM
AS
SA
W
22 30
38 78.90
22 100.00
6 1 1
29 5 3 2
38 78.90
39 74.30
18 25
18 100.00
25 100.00
Overall Classification Accuracy: 87.92%, Kappa statistics: 0.8619
1 1 1 27 30 90.00
30 30 100.00
Total 30 30 30 30 30 30 30 30 240
User. Acc. 100.00 73.30 100.00 96,67 60.00 83,33 90,00 100.00
226
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
Table 5. Confusion matrix for the Landsat ETM+ (2000) image supervised classification Class
SW
HW
SW
24
2
HWHW
SWHW
HW
29
HW-HW
2
27
1
SW-HW
1
1
28
DF
1
FO
1
SM
DF
FO
SM
AS
SA
Total User. Acc.
2
2
30
80.00
1
30
96.66
30
90.00
29 27
1
2
2 27
1
AS
26
3 30
SA W Total Total Prod. Acc.
W
25
36 96.00
28 80.55
29 90.00
29 93.33 100.00
30 90.00
27 100.00
28 92.85
38
30
93.33
30
96.66
30
90.00
30
90.00
30
86.66
30
100.00
30
30
100.00
30
240
78.94 100.00
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Overall Classification Accuracy: 92.33%, Kappa Statistics: 0.8987
Figure 2. Land use/land cover type map of Bulanıkdere forest planning unit (1985–2003).
Temporal Change in Land Use/ Land Cover Types Landscape changes were mapped using both forest cover type/stand maps and classified Landsat images (Fig. 2 and Fig. 3). According to the digitized stand type maps from forest management plans between 1985 and 2003 years, the total forested areas decreased from 7432.9 ha (87.4% of the study area) to 7351.5 ha (86.4%) during a 18 year period (Table 6). There was a net decline of 81.4 ha (1.1%) forest areas as opposed to a net increase of 86.5 ha in agriculture and settlement areas. While there were not any softwood forests in 1985, there was 78.7 ha softwood forest in 2003. Hardwood forest decreased 323.3 ha, hardwood hardwood forest increased 278.5 ha, softwood-hardwood forest decreased 233.4 ha,
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage… 227 hardwood-softwood forest increased 51.5 ha, degraded forest increased 35.2 ha, swamp area increased 33.8 ha, forest openings increased 31.4 and water area decreased 1 1.2 ha. During the whole study period, the average annual deforestation rate based on stand type map was 5 ha per year, equivalent to 0.06% per year using the compound-interest-rate formula. As an overall change in Bulanıkdere forest planning unit, there was a net decrease of 81.4 ha in total forested areas. This could be attributed to the urbanization activities.
Figure 3. Land use/land cover type map of Bulanıkdere forest planning unit (1987–2000).
Table 6. Evolution of selected landscape variables in Bulanıkdere from 1985 to 2003 (stand type map) Year Land Use Class
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
SW HW
1985 ha
2003 %
ha
% 78.7 5137.3
0.9 60,4
5460.6
64.2
SW-HW HW-SW
395.0
4.6
161.6 51.5
1.9 0.6
HW-HW DF FO AS SM SA W Other Total
1294.1 221.6 61.6 371.3 316.6 159.3 98.8 127.4 8506.3
15.2 2.6 0.7 4.4 3.7 1.9 1.2 1.5 100.0
1572.6 256.8 93.0 457.8 350.4 141.6 87.6 117.4 8506.3
18.5 3.0 1.1 5.4 4.1 1.7 1.0 1.4 100.0
According to the classified Landsat images between 1987 and 2000 years, however, swamp and sandy areas decreased 499.3 ha and 125.7 ha, respectively while agriculture and settlement areas increased 211.2 ha (Table 7). Total non-forest areas decreased 4.8% (407.3 ha) causing the forested areas to increase with a similar rate of change. Within the major
228
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
increased forested areas; hardwood forest, softwood forest, softwood-hardwood forests and hardwood -hardwood forests increased substantially with 915.2 ha, 81.5 ha, 656.2 ha and 879.8 ha, respectively. When both approaches were evaluated together, between 1985 and 2003, forest areas decreased from 7432.9 ha to 7351.5 ha according to stand type maps from management plans. On the contrary, between 1987 and 2000 years, forest areas increased from 7049.4 ha to 7456.7 ha according to land cover type maps from the classified images of Lansat TM and ETM+. Forest areas based on stand type map decreased while they increased based on classified Landsat images. Both approaches presented the net increase in softwood forest, hardwood-hardwood forest and agriculture and settlement and a decrease in sandy areas accordingly. During the whole study period, the average annual forestation rate based on classified Landsat image was 31 ha year-1, equivalent to 0.43% year-1 using the compoundinterest-rate formula. The stand type maps were further analyzed to see any changes in forest structure. In terms of crown closure change, between 1985 and 2003 years, stands with the low crown closure (11–40%), increased about 16 ha (Table 8). However, stands with the medium crown closure (41–70%) decreased about 118 ha, the full crown closure (fully covered) decreased about 37 ha, and regenerated areas decreased about 10 ha. To a certain extent, changes in crown closure indicate the deterioration of forest structure as fully covered (full closure) areas decreased and changed in favor of less covered areas. Opening in forest structure (i.e., breakdown of crown closure) can be caused by inappropriate silvicultural prescriptions, harvesting activities, heavy commercial treatments, illegal cutting or natural breakdown or over utilization. However, there was no indication of natural breakdown in stand structure as a result of natural disturbances according to regular observations of the area. So, the likely cause would relate to uncontrolled stand treatments and/or illicit harvesting. The heavy forest industrialization and a lack of field or range foresters in the area are enough to support such conclusions. The implication can easily be reflected in the calculation of growing stock and then the quality of wood products. Table 7. Evolution of selected landscape variables in Bulanıkdere from 1987 to 2000 (Landsat images)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Year Land Use Class SW HW SW-HW HW-HW DF FO AS SM SA W Total
1987 ha
2000 %
ha
3916.5 277.9
46.0 3.3
1873.1 981.9 451.2 789.1 130.5 86.1 8506.3
22.0 11.6 5.3 9.3 1.5 1.0 100.0
81.5 4831.7 934.1 879.8 575.5 154.1 662.4 289.8 4.8 92.6 8506.3
% 1.0 56.8 11.0 10.3 6.7 1.8 7.8 3.4 0.1 1.1 100.0
Another parameter to analyze the change in the quality of forest structure is the development stages of forest ecosystems. According to stand type maps, the forests in 1985
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage… 229 were mostly clumped into a development stage class (regenerated) and b (young) development stage class with the areas of 501.1 ha and 5304.1 ha, respectively (Table 9). In 2003, the forest is generally concentrated into regenerated and young stages of development. However, young forest structure (a and b development stage) and mature forest structure (c and d development stage) decreased 90.4 ha and 50.2 ha respectively from 1985 to 2003. In other words, forest areas decreased and forest structure became more young stage. Forest structure generally clumped into young stage would be explained by the fact that there were almost regenerating activities in the previous years. Table 8. Changes in crown closure between 1985 and 2003 (stand type maps) Crown closure 0 (regenerated areas) 1 (low coverage. 11-40 %) 2 (medium coverage. 41-70 %) 3 (full coverage. 71-100 %) DF (sparsely distributed. 0-10 %) FO AS SM SA W Other Total
Area (ha) 1985 year 2003 year 9.6 16.3 288.4 170.8 6851.7 6814.6 221.6 256.8 61.6 93.0 371.3 457.8 316.6 350.4 159.3 141.6 98.8 87.6 127.4 117.4 8506.3 8506.3
Difference (+/-) -9.6 16.3 -117.6 -37.1 35.2 31.4 86.5 33.8 -17.7 -11.2 -10.0 0.0
Table 9. Change in development stage between 1985 and 2003 (stand type maps)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Development Stage a (regenerated 36 cm) DF FO AS SM SA W Other Total
Area (ha) 1985 year 2003 year 9.6 46.9 491.5 4765.2 5667.9 538.9 7.4 1109.6 1160.4 176.7 177.3 221.6 256.8 61.6 93.0 371.3 457.8 316.6 350.4 159.3 141.6 98.8 87.6 127.4 117.4 8506.3 8506.3
Difference (+/-) 37.3 -491.5 902.7 -538.9 1102.2 -1160.4 0.6 35.2 31.4 86.5 33.8 -17.7 -11.2 -10 0.0
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 10. The transition matrix of land use change from forest management plans in Bulanıkdere FPU between 1985 and 2003 2003 Land use Land Use Class
Total (ha)
SW
HW
SW-HW
HWSW
HW-HW
DF
FO
AS
SM
SA
W
Other
7.1
5.7
4.0
6.9
SW HW
5460.6
2.1
4640.7
3.3
2.0
598.4
114.8
32.9
42.7
SW-HW
395.0
73.0
86.6
150.1
48.2
13.5
15.7
1.0
2.0
895.5
13.7
4.2
66.4
9.8
19.5
81.5
12.8
10.1
1.3
12.1
3.5
9.9
28.0
8.1
13.5
28.1
4.5
14.1
298.4
1.5
6.3
0.2
4.9
1985 land use
HW-SW HW-HW
1294.1
DF
221.6
FO
61.6
AS
371.3
SM
316.6
3.5
4.9
4.3
298.4
2.4
3.1
SA
159.3
5.9
4.2
6.1
18.1
123.9
1.1
W
98.8
3.2
1.8
0.1
0.5
14.2
1.5
77.5
Other
127.4
0.5
0.9
3.2
16.6
19.2
8506.3
78.7
5137.3
1572.6
256.8
Total (ha)
287.0 2.9
0.2
83.9
8.2
161.6
1.3
51.5
93.0
457.8
1.8
1.6
14.1
0.1
4.5
87.0 350.4
141.6
87.6
117.4
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 11. The transition matrix of land use change from Landsat images in Bulanıkdere FPU between 1987 and 2000 Land Use Class HW
2003 Land use
Total (ha)
HW
SW
SW-HW
HW-HW DF
FO
AS
SM
SA
W
3916.5
2336.9
14.6
499.4
666.5
257.5
25.7
108.1
6.2
0.4
1.2
277.9
103.5
48.5
73.5
27.7
9.7
2.2
7.9
4.8
0.1
1873.1
1244.4
1.4
133.5
140.7
188.4
35.6
113.6
15.2
0.3
981.9
782.9
0.3
46.5
12.5
53.5
21.9
55.4
8.4
0.5
451.2
110.0
1.5
41.2
11.6
29.9
30.6
205.9
14.9
1.9
3.7
789.1
248.1
14.4
126.0
20.2
33.9
34.4
88.5
215.5
0.7
7.4
130.5
4.3
10.1
0.6
2.4
3.5
72.6
2.4
1.0
33.6
86.1
1.6
0.8
3.9
0.2
0.2
10.4
22.4
0.3
46.3
8506.3
4831.7
81.5
934.1
575.5
154.1
662.4
289.8
4.8
92.6
SW
1987 land use
SW-HW HWHW DF FO AS SM SA W Total (ha)
879.8
232
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
Transitions Among Land Use/Land Cover Types The magnitude and the direction of changes in landscape are the most important factors relating to landscape evolution (Antrop, 2000). Besides analyzing the changes in the amount of land use/land cover types, the temporal transitions among land use/land cover types are also documented and evaluated to see the intertemporal dynamics of landscape. Table 12. Changes in land use cover type from forest management plans in Bulanıkdere FPU between 1985 and 2003 Land Class
Use
Percent of Land Use Class in 1985 Unchanged in 2003 Lost to other Gained from other classes in 2003 classes in 2003
SW
78.7
78.7
HW
4640.7
819.9
496.6
-323.3
SW-HW
150.1
244.9
11.5
-233.4
51.5
51.5
HW-SW HW-HW
895.5
398.6
677.1
278.5
DF
81.5
140.1
175.3
35.2
28.0
33.6
65.0
31.4
298.4
72.9
159.4
86.5
298.4
18.2
52.0
33.8
123.9
35.4
17.7
-17.7
77.5
21.3
10.1
-11.2
87.0
40.4
30.4
-10.0
FO AS SM SA W Other
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Net gain/loss (Ha)
The transitions among major land cover types for the period were determined based on forest stand maps created as part of forest management plans and classified Landsat images (Table 10 and 11). A detail of losses and gains among the land cover classes over the study period is included in Table 12 and 13. A broad level analysis showed that about 269.6 ha (3.2% of the study area) forest areas changed into non forest while 188.2 ha (2.2% of the study area) non-forest areas changed into forest areas, with a net decline of 81.4 ha (1.1%) forest areas. 60.4 ha agriculture and settlements (16.3% of the agriculture and settlements) changed into forested area in 2003. On the contrary, 11.5 ha (5.2% of the degraded forest) degraded forest, 66.4 ha (1.2% of the hardwood forest) hardwood forest, 6.9 ha (1.7% of the softwood-hardwood forest) softwood-hardwood forest, 93.7 ha (7.2% of the hardwood hardwood forest) hardwood-hardwood forest, and 8.1 ha (13.1% of the forest openings) forest openings converted to non-forest areas. Among the forest cover groups, around 886 ha of non forest area and 5796 ha of the forest area in 1985 remained unchanged until 2003. Forest lost 1637.1 of its 1985 area to other classes and gained 1555.7 from other classes. These apparent dynamics of forest landscape generally issue from either mismanagement of the area or
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage… 233 uncontrolled forest protection. However, when the transitions were further investigated in detail, some interesting results can be revealed. For instance, 13.5 ha and 28.1 ha agriculture and settlements area changed into hardwood forest and hardwood-hardwood forest, respectively. In addition, 42.7 ha hardwood forest and 66.4 ha hardwood-hardwood forest converted to agriculture and settlements area. Table 13. Changes in land use cover type from Landsat images in Bulanıkdere FPU between 1985 and 2003 Percent of Land Use Class in 1987 Land Use Class
HW
Lost to other classes in 2000
Gained from other classes in 2000
2336.9
1579.6
2494.8
915.2
81.5
81.5
SW SW-HW
73.5
204.4
860.6 879.8
656.2 879.8
DF
188.4
1684.7
387.1
-1297.6
FO
21.9
960.0
132.2
-827.8
AS
205.9
245.3
456.5
211.2
SM
215.5
573.6
74.3
-499.3
SA
1.0
129.5
3.8
-125.7
W
46.3
39.8
46.3
6.5
HW-HW
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Net gain/loss (Ha)
Unchanged in 2000
According to Landsat images, major transition of land cover type between 1987 and 2000 relate to 110 ha hardwood forests created from agriculture and settlement. This is probably a result of afforestation of open areas into hardwood species. However, 108.1 ha hardwood forests were converted into agriculture and settlement, that would be explained by land clearing for settlements or social conflicts. This area is generally located near to urban settlements. 257.5 ha forest hardwood forest was converted to degraded forest due to mismanagement activities, uncontrolled forest protection and social pressure. As a result, forest landscape is fragmented and deteriorated. General analysis showed that there were visible trends in the temporal structure of forest landscape.
Spatial Pattern Analysis Landscape pattern indices or measurements provide a useful set of tool to explore inside variability, cross-site differences and changes over time. The simultaneous use of class-level and patch-level landscape pattern indices enabled assessment of the spatial configuration of forest cover and its relation to principal land cover types. We used patch, class and landscape level metrics to analyze forest fragmentation of study area with FRAGSTATSTM. The spatial analysis of the landscape pattern based on forest stand maps created as part of forest management plans indicated that the total number of patches increased from 124 to 237 between 1985 and 2003 years as all patch types were taken into account, indicating a highly
234
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
sensitive forest landscape for conservation. Fragmentation increased approximately two times as mean patch size (MPS) decreased from 68.6 ha to 35.9 ha (Table 14 and 15). This rapid decline in mean patch size was associated with a rapid increase in the patch density (PD) for the same period (1.46 to 2.79), and a substantial reduction in the size of the largest land use/land cover patch, from 47.0% of the total study area in 1985 to 40.1% in 2003. PSCV value increased from 550.4 to 659.1 and variant of patch size increased. Similarly, Area Weighted Shape Index value increased from 4.4 to 5.2 indicating a more irregular shape with more edgy environment. All these changes clearly indicate that the forest landscape has been gradually fragmented and thus subjected to susceptibility of the area to further abrupt changes in the future. The spatial structure of the landscape generated from the Landsat images between 1987 and 2000 was also quantified and analyzed to see the performance of the image classification procedure. According to land use analysis, the number of patches increased from 6298 to 12888 between 1987 and 2009 years. Mean patch size decreased from 1.4 ha to 0.7 ha same as largest patch index that is decreased from 25.43% to 20.98%. This rapid decline in mean patch size was associated with a rapid increase in the patch density (PD) and a substantial reduction in the size of the largest land use/land cover patch. These changes show that landscape fragmentation increased and the forest has become more susceptible to harsh disturbances. Other important indicator of the fragmentation is AWMSI, increased from 4.4 to 5.2 in the Bulanıkdere forest planning unit. These changes are significant and show that landscape fragmentation increased and forest has become more susceptible to harsh disturbances resulting in a hight probably of biodiversity loss. The number of patches with their sizes is a relatively good index of fragmentation (Southworth et al., 2004; Echeverria et al., 2006). In other words, one of the basic indications of forest fragmentation is the increase in number of smaller patches. In the Bulanıkdere FPU considerable changes were found in the distribution of forest patch sizes between time intervals (Table 16). The total number of forest fragments increased from 124 to 237 during the 18 year period based on stand type map. This corresponds to an annual increase in the number of fragments of 11% between 1985 and 2003. By 1985, 75% of the land use patch was concentrated in small patches between 0 and 20 ha; the remaining forest area occurred in isolated patches of larger than 20 ha. In 2003, 84% of the land use patch of landscape occurred in patches of less than 20 ha. According to classified Landsat images, 99.3% and 99.8% of the land use patch was concentrated in small patches between 0 and 20 ha in 1985 and 2003, respectively.
Spatial Distribution and Temporal Change of Carbon Storage It was estimated that forest ecosystems in Bulanıkdere forest planning unit contain 304 264.3 tons of carbon above and belowground according to the methodology used in this article in 1985. Though 263 895.9 tons of whole carbon storages in forest ecosystem are aboveground, the rest (40 368.4 tons) are belowground. However, it was estimated that forest ecosystems contain 510 030.1 tons of carbon with 443 032.5 tons aboveground and 66 997.6 tons of carbon belowground in 2003. As shown, carbon storages increased 205 765.8 tons over 18 years. Distributions of carbon storages of AFPU in the years of 1985 and 2003 are shown in Figure 4. Even though these changes represent a 1% decrease in forestation, total carbon storages have increased. This result can be explained by the fact that the area of stands with high growing stocks has increased.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 14. Selected landscape metrics of Bulanıkdere FPU (1985-2003 forest cover type map) Land Use SW HW SWHWHWDF FO AS SM SA W Other Landsca
1985 5460.6 395.0 1294.1 221.6 61.6 371.3 316.6 159.3 98.8 127.4 8506.3
CA 2003 78.7 5137.3 161.6 51.5 1572.6 256.8 93.0 457.8 350.4 141.6 87.6 117.4 8506.3
NP 1985 200 5 8 18 2 5 6 16 30 26 73 21 37 36 45 6 5 1 3 6 7 2 3 124 237
MPS 1985 2003 15.7 682.6 285.4 197.5 32.3 8.6 80.9 52.4 8.5 3.5 2.9 2.5 10.3 10.2 52.8 70.1 159.3 47.2 16.5 12.5 63.7 39.1 68.6 35.9
PERCLAND 1985 2003 0.9 64.2 60.4 4.6 1.9 0.6 15.2 18.5 2.6 3.0 0.7 1.1 4.4 5.4 3.7 4.1 1.9 1.7 1.2 1.0 1.5 1.4 100 100
LPI 1985 2003 0.5 47.0 40.1 2.5 1.4 0.2 4.8 5.9 1.1 0.5 0.1 0.2 1.1 1.4 1.8 2.0 1.9 1.3 0.6 0.5 1.3 1.2 47.0 40.1
PD 1985 200 0.06 0.09 0.21 0.02 0.06 0.07 0.19 0.35 0.31 0.86 0.25 0.43 0.42 0.53 0.07 0.06 0.01 0.04 0.07 0.08 0.02 0.04 1.46 2.79
PSCV 1985 2003 123.3 206.0 288.0 10.5 154.6 52.4 166.2 213.2 208.5 150.4 95.9 152.0 157.0 227.3 101.6 91.8 121.3 115.7 123.2 96.4 138.5 550.4 659.1
AWMSI 1985 2003 2.3 5.1 6.3 1.9 2.6 1.3 3.9 4.6 2.0 2.2 1.7 2.0 1.9 2.5 2.4 2.3 5.7 4.8 3.7 3.0 2.7 2.5 4.4 5.2
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 15. Selected landscape metrics of Bulanıkdere FPU (1987-2000 land cover type map of Landsat) Land Use SW HW SWHWDF FO AS SM SA W Landsca
1987 3916.5 277.9 187.1 981.9 451.2 789.1 130.5 86.1 8506.3
CA 2000 81.5 4831.7 934.1 879.8 575.5 154.1 662.4 289.8 4.8 92.6 8506.3
NP 1987 2000 95 989 2310 1000 3103 3245 1530 2244 506 501 659 935 1372 360 211 22 31 73 6298 1288
MPS 1987 2000 0.9 4.0 2.1 0.3 0.3 0.3 1.2 0.3 1.9 0.3 0.7 0.7 0.6 0.8 0.6 0.2 2.8 1.3 1.4 0.7
PERCLAND 1987 2000 1.0 46.0 56.8 3.3 11.0 10.3 22.0 6.7 11.6 1.8 5.3 7.8 9.3 3.4 1.5 0.1 1.0 1.1 100.0 100.0
LPI 1987 2000 0.23 25.4 20.98 0.21 0.53 0.39 1.88 0.06 3.72 0.10 0.67 1.41 3.04 1.52 0.59 0.01 0.48 0.41 25.4 20.98
PD 1987 2000 1.1 11.6 27.2 11.8 36.5 38.2 18.0 26.4 5.9 5.9 7.7 11.0 16.1 4.2 25 0.3 0.4 0.9 74.1 151.5
PSCV 1987 2000 293.7 1803.3 2123.2 321.8 383.3 307.3 704.4 156.1 768.9 223.9 409.2 805.5 1286.3 895.4 626.2 106.1 279.3 426.6 2160.2 2868.3
AWMSI 1987 2000 2.5 10.7 17.1 1.8 2.0 2.0 4.5 1.6 4.0 1.7 2.7 3.6 3.9 3.6 4.3 1.4 3.0 3.6 7.0 10.7
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage…237 Table 16. Variation of land use/land cover and number of patches over patch size classes Stand Type Map Patch
Area (ha)
Size (ha)
Landsat Images Number of
Area (ha)
patches
Number of patches
1985
2003
1985
2003
1987
2000
1987
2000
0-20
476.9
740.5
93
199
2780.7
4035.7
6253
12864
20-50
411.9
665.7
15
20
803.0
462.2
25
15
50-100
306.7
438.7
4
6
801.1
196.1
11
2
100-200
1049.9
872.7
7
7
560.6
366.8
4
3
200-500
1021.3
815.5
3
2
1397.8
575.2
4
2
>5000
5239.6
4973.2
2
3
2163.1
2870.3
1
2
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
DISCUSSION This study analyzed the temporal and spatial pattern of land use/land cover and carbon storage change in a forest planning unit in the northwestern corner of Turkey over 1985–2003 period using GIS and FRAGSTATSTM. We observed drastic changes in the spatial pattern of the forests during the 18-year period. As an overall change, there was a net decrease of 81.4 ha (1.1%) in total forested area compared to a net increase of 86.5 ha in agriculture and settlement areas because of heavy urbanization in the study area. This change could be explained by the following reasons. Among the general and main causes of deforestation are human population pressure and an increasing demand of land for agriculture and timber products from forests (Cayuela et al., 2006). In Bulanıkdere, human population increased from 3075 to 3135 (nearly 2.0%) inhabitants in the period 1985–2000. However, there is no important change in population of Bulanıkdere residences over 20 years and forest areas slightly decreased. During the period, the average annual deforestation rate based on stand type map was 5 ha year-1, equivalent to 0.06% year-1. The quantitative evidences of forest cover patterns showed that human activities have affected the forest cover type changes. According to the classified Landsat images between 1987 and 2000 years, total nonforest areas decreased 4.8% (407.3 ha) causing the forested areas to increase with a similar rate of change. During the whole study period, the average annual forestation rate based on classified Landsat image was 31 ha year-1, equivalent to 0.43% year-1 using the compoundinterest-rate formula. The results from two approaches may not be comparable to each other as the original date of data is not comparable. As such, throughout the paper, the results from both approaches were evaluated separately and not compared to each other. In comparing the results from both approaches, though, one would be cautious that the causes at different time would be different given the fact that the classification procedures are different. There was not softwood forest area in 1985 while there were 78.7 ha of area in 2003. 73 ha (93% of the
238
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
softwood forest) softwood-hardwood areas converted to softwood areas. This transition may be explained that hardwood trees dominated landscape due to harvesting activity.
Figure 4. Maps of carbon storage (above and belowground) of Bulanıkdere forest planning unit in a) 1985 b) 2003.
The most important indicators of forest ecosystem quality are development stages, cover types and crown closures. In terms of crown closure change, between 1985 and 2003, stands with the medium crown closure (41–70%) decreased about 118 ha, the full crown closure (fully covered) decreased about 37 ha. To a certain extent, changes in crown closure indicate the deterioration of forest structure as fully covered areas decreased and were changed in favor of less covered areas. According to development stage, young forest structure (a and b development stage) and mature forest structure (c and d development stage) decreased 90.4 ha and 50.2 ha respectively from 1985 to 2003. Forest structure became more young stage and forest structure generally clumped into young stage would be explained by the fact that there
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage…239 were almost regenerating activities in the previous years. These changes indicated that there were either intensive or repeated silvicultural activities such as pre-commercial thinning or illicit harvesting resulting in a less quality of forest structure in terms of timber production. Specifically, opening in forest structure (i.e., breakdown of crown closure) can be issued from heavy commercial treatments, illegal cutting or natural breakdown of forest structure. However, there was no indication of natural breakdown in stand structure as a result of natural disturbances according to regular observations of the area. So, the likely cause would relate to uncontrolled stand treatments and/or illicit harvesting. The heavy forest industrialization and a lack of field or range foresters in the area are enough to support such conclusions. The implication can easily be reflected in the calculation of growing stock and then the quality of wood products. The most important indicators of fragmentation are the number of patches and the increase in number of smaller patches (Abdullah and Nakagoshi, 2006; Echeverria et al., 2006; Southworth et al., 2004; Kammerbauer and Ardon, 1999). As well, biodiversity conservation is associated with both landscape configuration and composition and also forest management activities directly affect both components. Forest cover changes are the most common cause of loss biodiversity (Armsworth et al., 2004; Harris, 1984; Kilic et al., 2004). A variety of patch sizes, for example, is important for wildlife (Wei and Hoganson, 2005). Impacts of forest patch sizes on wildlife differ by wildlife species. Human activities generally cause patch sizes to decline, which is critical to some wildlife species (Bender et al., 1998). The composition and the configuration of forest resource changed considerably in Bulanıkdere. While the forest area decreased during 25 year period, number of patches, smaller patch size and patch density increased resulting in fragmentation of the landscape according to stand type map. Fragmentation also increased in landscape generated from the Landsat images. The trend signals a danger for the sustainability of forest resources and the resilience of the ecosystem mainly for biodiversity and to a less extend timber production. The possible reasons for increased fragmentation could relate the illicit harvesting, heavy commercial treatments, illegal cutting, uncontrolled stand treatments, illicit harvesting or road construction. The total road length in Bulanıkdere forest planning unit increased due to heavy transportation and harvesting activities. Forest area decreased approximately 1% causing to increase total carbon storages. This result can be explained by the fact that the area of stands with high growing stocks increased. However, the contradiction in increasing carbon storage in spite of decreasing forest area may issue from the data quality as it would be affected by the data gathering methods in 1985 and 2003. For example, the stand type maps in 1985 were produced with 1/16,000 scale color infrared aerial photos while the maps in 2004 were produced with High Resolution Satellite Image (IKONOS). As for the accuracy of the stand maps gathered from the aerial photo interpretation, since the study area were further ground sampled by 300m × 300m grids the predetermined cover types were checked on the ground and corrected accordingly. The crown closure, species mix and the development stages were measured directly on the ground for each sample plot. Thus, the results of the aerial interpretation were supported by the field measurements which helped improve greatly the accuracy of stand type maps. Growing stock and the biomass per hectare increases, the amount of carbon sequestered per hectare also increases. In conclusion, monitoring the dynamics of both composition and configuration of forest ecosystems as well as land use/land cover with carbon storage over time is increasingly
240
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
important in forest management, landscape ecology and biodiversity. Thus, it is extremely crucial that both composition and configuration of landscape be quantitatively evaluated to draw appropriate lessons for landscape management. Such studies contribute to sustainable forest management and biodiversity monitoring
SUMMARY
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Monitoring land use/land cover and carbon storage change and understanding the dynamics of forest ecosystems are increasingly important in management of forest resources. In this study, land use/land cover and carbon storage change in Bulanıkdere forest planning unit was investigated. Spatial and temporal changes in land use/land cover and carbon storage were analyzed using Geographical Information System (GIS) and FRAGSTATSTM. Land use/land cover and carbon storage changes were detected from a time series of satellite images of Landsat TM (1987), Landsat ETM+ (2000) and forest stand map obtained from forest management plan using Remote Sensing (RS) and GIS. The analysis showed that the total forested areas decreased from 7432.9 ha (87.4% of the study area) to 7351.5 ha (86.4%) during a 18 year period according to the digitized stand type maps from forest management plans between 1985 and 2003. According to the classified Landsat images between 1987 and 2000, total non-forest areas decreased 4.8% (407.3 ha) of landscape causing the forested areas to increase. During the period, the average annual deforestation rate based on stand type map was 5 ha year-1, equivalent to 0.06% year-1 using the compound-interest-rate formula. The spatial analysis of the landscape pattern based on forest stand maps created as part of forest management plans indicated that the total number of patches increased from 124 to 237 between 1985 and 2003 years as all patch types were taken into account, indicating a highly sensitive forest landscape for conservation. Carbon storages increased 205 765.8 tons during the 18 year period. There were apparent trends in the temporal structure of forest landscape, some of which may issue from unregulated management activities, social conflicts and demographic movements. The study revealed that it is important to understand both spatial and temporal changes of land use/land cover as well as carbon storage and their effects on landscape pattern to understand their implications to landscape planning and ecosystem management.
REFERENCES Abdullah, S.A. and Nakagoshi, N. (2006). Changes in landscape spatial pattern in the highly developing state of Selangor, peninsular Malaysia. Landscape and Urban Planning 3:263-275. Abdullah, S.A. and Nakagoshi, N. (2007). Forest fragmentation and its correlation to human land use change in the state of Selangor, peninsular Malaysia. Forest Ecology and Management 241: 39-48. Ali, A.M.S. (2006). Rice to shrimp: Land use/land cover changes and soil degradation in Southwestern Bangladesh. Land Use Policy 23: 421-435.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage…241 Anonymous, (1985). Istanbul Forest Regional Directorate, Demirköy Forest Enterprise, Management plan of Bulanıkdere forest planning unit. Between 1985–2004. Anonymous, (2003). Istanbul forest regional directorate, Demirköy Forest Enterprise, Management plan of Bulanıkdere forest planning unit. Between 2003–2012. Antrop M., (2000). Background concepts for integrated landscape analysis. Agriculture, Ecosystem and Environment 7: 17–28. Armenteras, D., Gast, F. and Villareal, H. (2003). Andean forest fragmentation and the representativeness of protected natural areas in the eastern Andes, Colombia. Biological Conservation 113: 245–256. Armsworth, P.R., Kendall, B.E. and Davis, F.W. (2004). An introduction to biodiversity concepts for environmental economists. Resource and Energy Economics 26: 115–136. Asan, Ü., Destan, S. and Özkan, U.Y. (2002). Istanbul korularının karbon depolama, oksijen üretimi ve toz tutma kapasitesinin kestirilmesi. Orman Amenajamanýnda Kavramsal Açılımlar ve Yeni Hedefler Sempozyumu. Bildiriler Kitabı. Istanbul-Turkey. pp 194– 202. Başkent, E.Z. and Jordan, J.A. (1995a). Characterizing spatial structure of forest landscapes: a hierarchical approach. Canadian Journal of Forest Research 25: 1830-1849. Başkent, E.Z. and Jordan, J.A. (1995b). Designing forest management to control spatial structure of landscapes. Landscape Urban Planning 34: 55-74. Başkent, E.Z. and Kadıogulları, A.I. (2007). Spatial and temporal dynamics of land use pattern in Turkey: A case study in İnegöl. Landscape and Urban Planning 81:341-353. Başkent, E.Z., Jordan, G.A. and Nurullah, A.M.M. (2000). Designing forest landscape (ecosystems) management. Forestry Chronicle 76(5): 739-742. Başkent, E.Z., Terzioğlu, S., and Başkaya, Ş. (2008). Developing and Implementing Multiple Use Forest Management Planning in Turkey, Environmental Management, in press. Bender, D. J., Contreras, T. A. and Fahrig, L. (1998). Habitat loss and population decline: A meta-analysis of the patch size effect. Ecology 79(2): 517–533. Bray, D.B., Klepeis, P. (2005). Deforestation, forest transitions, and institutions for sustainability in Southeastern Mexico, 1900-2000. Environment and History 11: 195-223. Bürgi, M., (1999). A case study of forest change in the Swiss lowlands. Landscape Ecology 14: 567-575. Bürgi, M., Russell, E.W.B., Motzkin, G. (2000). Effects of post settlement human activities on forest composition in the north-eastern Unites States: a comparative approach. Journal of Biogeography 27: 1123-1138. Cayuela, L., Rey-Benayas, J.M. and Echeverria, C. (2006). Clearance and fragmentation of tropical montane forests in the Highlands of Chiapas, Mexico (1975–2000). Forest Ecology and Management 226: 208–218 Çakır, G., Sivrikaya, F., Terzioğlu, S., Başkent, E.Z., Sönmez, T. and Yolasığmaz, H.A. (2007). Mapping Secondary Forest Succession by Using Geographic Information Systems (GIS) and Remote Sensing (RS): A Case Study from Bulanıkdere (KırklareliTurkey). TÜBİTAK Turkish Journal of Agriculture and Forestry 31: 71-81. Chen, L., Wang, J., Fu, B., & Qiu, Y. (2001). Land-use change in a small catchment of northern Loess Plateau, China. Agriculture, Ecosystems and Environment, 86: 163–172. Cohen, J. (1960). A coefficient of agreement of nominal scales. Educational and Psychological Measurements 20(1): 37–46.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
242
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
Cohen, W.B., Spies, T.A. and Fiorella, M. (1995). Estimating the age and structure of forest in a multi-ownership landscape in western Oregon, USA. Int. J. Remote Sens. 16(4): 721– 746. Congalton, R.G. and Mead, R.A. (1983). A quantitative method to test for consistency and correctness in photo interpretation. Photogrammetric Engineering and Remote Sensing 49: 69–71. Debinski, D. and Holt, R. (2000). A survey and overview of habitat fragmentation experiments. Conservation Biology 14: 342–355. DeWalt, S.J., Maliakal, S.K., Denslow, J.S. (2003). Changes in vegetation structure and composition along a tropical forest chronosequence: implications for wildlife. Forest Ecology and Management 182: 139-151. DiBari, J.N. (2007). Evaluation of five landscape-level metrics for measuring the effects of urbanization on landscape structure: the case of Tucson, Arizona, USA, Landscape and Urban Planning 79: 308–313 Dupouey, J.L. Dambrine, E., Laffite, J.D. and Moares, C. (2002). Irreversible impact of past land use on forest soils and biodiversity. Ecology 83: 2978-2984. Echeverria, C., Coomes, D., Salas, J., Rey-Benayas, J.M., Lara, A. and Newton, A. (2006). Rapid deforestation and fragmentation of Chilean Temperate Forests. Biological Conservation 130: 481-494. Emch, M., Quinn, J.W., Peterson, M., Alexander, M. (2005). Forest Cover Change in the Toledo District, Belize from 1975 to 1999: A remote sensing approach. The Professional Geographer 57: 256-267. Erskine, W.D., Mahmoudzadeh, A., Myers, C. (2002). Land use effects on sediment yields and soil loss rates in small basins of Triassic sandstone near Sydney, NSW, Australia. Catena 49: 271-287. Evelyn, O.B., Camirand, R. (2003). Forest Cover and deforestation in Jamaica: an analysis of forest cover estimates over time. International Forestry Review 5: 354-363. Evrendilek, F. (2004). An inventory based carbon budget for forest and woodland ecosystems of Turkey. Journal of Environmental Monitoring 6: 26-30. Farina, A. (2000). Landscape Ecology in Action. Kluwer Academic Publisher: Dordrecht, The Netherlands. Feranec, J., Hazeu, G., Christensen, S., Jaffrain, G., (2007). Corine land cover change detection in Europe (case studies of the Netherlands and Slovakia). Land Use Policy 24: 234-247. Forman, R.T.T. (1995). Some general principles of landscape and regional ecology. Landscape Ecology 10: 133–142. Forman, R.T.T. (1998). Land mosaics: The ecology of landscapes and regions. Cambridge, UK: Cambridge University Press. pp 652. Franklin, S. (2001). Remote Sensing for Sustainable Forest Management. Lewis Publishers. USA. Freemark, K., & Collins, B. (1992). Landscape ecology of birds breeding in temperate forest fragments. In J. M. Hagan & D. W. Johnson (Eds.), Ecology and conservation of neotropical migrant landbirds (pp. 443–454). Washington, DC: Smithsonian Institution Press.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage…243 Gautam, A.P., Webb, E.L., Shivakoti, G.P. and Zoebisch, M.A. (2003). Land use dynamics and landscape change pattern in a mountain watershed in Nepal. Agriculture, Ecosystems and Environment 99: 83–96. Geist, H., McConnell,W., Lambin, E.F., Moran, E., Alves D. and Rudel, T. (2006). Causes and trajectories of land-use/cover change. In: E.F. Lambin and H. Geist, Editors, Land-Use and Land-Cover Change, Springer, Berlin. pp 41–70. Gigord, L., Picot, F. and Shykoff, J. (1999). Effects of habitat fragmentation on Dombeya acutangula (Sterculiaceae), a native tree on La Re´union (Indian Ocean). Biological Conservation 88: 43–51. Haase, D., Walz, U., Neubert, M., Rosenberg, M. (2007). Changes to central European landscapes – Analyzing historical maps to approach current environmental issues , examples from Saxony, Central Germany. Land Use Policy 24: 248-263. Harris, L.D. (1984). The fragmented forest. Chicago: University of Chicago Press. Iida, S. and Nakashizuka, T. (1995). Forest fragmentation and its effect on species diversity in sub-urban coppice forests in Japan. Forest Ecology and Management 73: 197–210. Ite, U.E. and Adams, W.M. (1998). Forest conversion, conservation and forestry in Cross River State, Nigeria. Application Geography 18: 301–314. Jameson, J.S., Ramsay, P.M. (2007). Changes in high-altitude Ploylepis forest cover and quality in the Cordillera de Vilcanato, Peru, 1956-2005. Biological Conservation 138: 3846. Jansen, L.J.M., Bagnoli, M., Focacci, M. (2007). Analysis of land cover/use change dynamics in Manica Province in Mozambique in a period of transition (1990-2004). Forest Ecology and Management. Jarvis, C. (1994). Modeling forest ecosystem dynamics using multitemporal multispectral scanner (MSS) data. Advances in Space Research 14(3): 277–281. Jensen, J.R. (1996). Introductory digital image processing. Englewood Cliffs, NJ: PrenticeHall. pp 318. Johnson LB, Richards C, Host G, Arthur JW. (1997). Landscape influences on water chemistry in midwestern streams. Freshwater Biology. 37: 209–217. Kadıogulları, A.I. (2005). Determination of temporal dynamics of forest resources using remote sensing and geographical information system: case study in Gumushane and Inegol State Forest industries. Master Thesis, Karadeniz Technical University, pp 118. Kadıoğulları, A.İ. and Başkent, E.Z. (2008). Spatial and temporal Dynamics of land use pattern in Turkey: A case study in Gümüşhane. Environmental Monitoring and Assessment 138: 289-303 Kammerbauer, J. and Ardon, C. (1999). Land use dynamics and landscape change pattern in a typical watershed in the hillside region of central Honduras. Agriculture, Ecosystems & Environment 75: 93–100. Keleş, S. and Başkent, E.Z. (2006). Orman Ekosistemlerindeki Karbon Değişiminin Orman Amenajman Planlarına Yansıtılması: Kavramsal Çerçeve ve Bir Örnek Uygulama. Orman ve Av Dergisi 83(3): 9–16. Keleş, S., Sivrikaya, F., Çakır, G. and Köse, S. (2008). Urbanization and forest cover change in regional directorate of Trabzon forestry from 1975 to 2000 using landsat data. Environmental Monitoring and Assessment. (In Press). Kennedy, R.S.H. and Spies, T.A. (2004). Forest cover changes in The Oregon Cost Range Form 1939 To 1993. Forest Ecology and Management 200: 129-147.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
244
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
Kilic, S., Evrendilek, F., Berberoğlu, S. and Demirkesen, A.C. (2004). Environmental monitoring of land-use and land-cover changes in Amik Plain, Turkey. In: Geo-Imagery Bridging Continents, XXth ISPRS Congress, Istanbul, Turkey. Lathrop, R.G., Tulloch, D.L., Hatfield, C. (2007). Consequence of land use change in the New York – New Jersey Highlands, USA: Landscape indicators of forest and watershed integrity. Landscape and Urban Planning 79: 150-159. Laurance WF. (1999). Reflections on the tropical deforestation crisis. Biological Conservation 91: 109-117. Leitao, A.B. and Ahern, J. (2002). Applying landscape ecological concepts and metrics in sustainable landscape planning. Landscape and Urban Planning 59: 65–93. Lele, N., Joshi, P.K., Agrawal, S.P. (2007). Assessing forest fragmentation in northeastern region (NER) of India using landscape metrics. Ecological Indicators. Levin, S.A., Powell, T.M., Steele, J.H. (Eds.) (1993). Patch Dynamics. SpringerVerlag,Berlin Lin, Y.P., Hong, N.M., Wu, P.J., Wu, C.F., Verburg, P.H. (2007). Impacts of land use change scenarios on hydrology and land use patterns in the Wu-Tu watershed in northern Taiwan. Landscape and Urban Planning 80: 111-126. Liu, J., Liu, S., & Loveland, T. R. (2006). Temporal evolution of carbon budgets of the Appalachian forests in the U.S. from 1972 to 2000. Forest Ecology and Management 222: 191–201. Luque, S.S., Lathrop, R.G., Bognar, J.A. (1994). Temporal and spatial changes in an area of the New Jersey Barrens Landscape. Landscape Ecology 9: 287-300. Mace, G., Balmford, A. and Ginsberg, J.R. (1998). Conservation in a Changing World. Cambridge University Press, United Kingdom. Matsushita, B., Xu, M. and Fukushima, T. (2006). Characterizing the changes in landscape structure in the Lake Kasumigaura Basin, Japan using a high-quality GIS dataset. Landscape and Urban Planning 78(3): 241–250. McEvoy, P.M., Flexen, M., McAdam, J.H. (2007). The effects of livestock grazing on ground flora in broadleaf woodlands in Northern Ireland. Forest Ecology and Management McGarigal, K. and Marks, B.J. (1995). FRAGSTATS: spatial pattern analysis program for quantifying landscape structure. U.S. forest Service General Technical Report PNW 351. Science 260: 1905–1910. McIntyre, S., Lavorel, S. (2007). A conceptual model of land use effects on the structure and function of herbaceous vegetation. Agriculture, Ecosystems and Environment 119: 11-21. Meffe, G.K., Carroll, C.R. (1997). Principles of Conservation Biology. Sinauer Associates, Inc., Sunderland, MA. Millington, A.C., Velez-Liendo, X.M. and Bradley, A.V. (2003). Scale dependence in multitemporal mapping of forest fragmentation in Bolivia: implications for explaining temporal trends in landscape ecology and applications to biodiversity conservation. Journal of Photogrammetry and Remote Sensing 57: 289–299. Muukkonen, P., Heiskanen, J. (2007). Biomass estimation over a large area based on standwise forest inventory data and ASTER and MODIS satellite data: A possibility to verify carbon inventories. Remote Sensing of Environment 107: 617-624. Nagendra, H., Pareeth, S. and Ghate, R. (2006). People within parks—forest villages, landcover change and landscape fragmentation in the Tadoba Andhari Tiger Reserve, India, Applied Geography 26: 96–112.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage…245 Negi, A.K., Bhatt, B.P., Todaria, N.P. (1999). Local Population impacts on the forests of Garhwal Himalaya, India. The Environmentalist 19: 293-303. Noss, R.F. (2001). Forest fragmentation in the southern rocky mountains. Landscape Ecology 16: 371–372. Otto, R., Krüsi, B.O., Kienast, F. (2007). Degradation of an arid coastal landscape in relation to land use changes in Southern Tenerife (Canary Islands). Journal of Arid Environments 70: 527-539. Pickett, S.T.A., White, P.S. (Eds.) (1985). The Ecology of Natural Disturbance and Patch 522 Dynamics. Academic Press Inc., San Diego, CA. Pueyo, Y., Alados, C.L. (2007). Effects of fragmentation, abiotic factors and land use on vegetation recovery in a semi-arid Mediterranean area. Basic and Applied Ecology 8: 158-170. Puyravaud, J.P. (2003). Standardizing the calculation of the annual rate of deforestation. Forest Ecology and Management 177: 593–596. Rao, K.S., Pant, R. (2001). Land use dynamics and landscape change pattern in a typical micro watershed in the mid elevation zone of central Himalaya, India. Agriculture, Ecosystems and Environment 86: 113-123. Richards, J.A. (1994). Remote sensing digital image analysis: An introduction. Second, Revised and Enlarged Edition. Berlin Heidelberg New York: Springer. Sachs, D.L., Sollins, P., Cohen, W.B. (1998). Detecting landscape changes in the interior of British Columbia from 1975 to 1992 using satellite imagery. Canadian Journal of Forest Research 28: 23-36. Sasaki, N. and Yoshimoto, A. (2007). Forest Resource Management and Mathematical Modeling – FORMATH KOBE 2007 INTERNATIONAL, Vol. 7. Sharma, P., Rai, S.C. (2007). Carbon sequestration with land-use cover change in Himalayan watershed. Geoderma 139: 371-378. Shi, P.J., Yuan, Y., Zheng, J., Wang, J.A., Ge, Y., Qiu, G.Y. (2007). The effect of land use/cover change on surface runoff in Shenzhen region, China. Catena 69: 31-35. Siriwardena, L., Finlayson, B.L., McMahon, T.A. (2006). The impact of land use change on catchment hydrology in large catchments: The Comet River, Central Queensland, Australia. Journal of Hydrology 326: 199-214. Sivrikaya, F., Keleş, S. and Çakır, G. (2007a). Spatial distribution and temporal change of carbon storage in timber biomass of two different forest management units. Environmental Monitoring and Assessment 132: 429-438. Sivrikaya, F., Çakır, G., Kadıoğulları, A.İ., Keleş, S., Başkent, E.Z. and Terzioğlu, S. (2007b). Evaluating land use/land cover changes and fragmentation in the Camili forest planning unit of northeastern Turkey from 1972 to 2005. Land Degradation and Development 18: 383-396 Southworth, J., Munroe, D. and Nagendra, H. (2004). Land cover change and landscape fragmentation-comparing the utility of continuous and discrete analyses for a western Honduras region. Agriculture, Ecosystems and Environment 101: 185-205. Stojchev, G., Asan A. and Gucin, F. (1998). Some macrofungi species of European part of Turkey. Turkish. J. of Botany 22: 341-346. Szilassi, P., Jordan, G., van Rompaey, A., Csillag, G. (2006). Impacts of historical land use changes on erosion and agricultural soil properties in the Kali basin at the Lake Balaton, Hungary. Catena 68: 96-108.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
246
Fatih Sivrikaya Günay Çakir, Sedat Keleş et al.
Thanapakpawin, P., Richey, J., Thomas, D., Rodda, S., Campbell, B., Logsdon, M. (2006). Effects of land use change on the hydrologic regime of the Mae Chaem river basin, NW Thailand. Journal of Hydrology 334: 215-230. Turner, M.G. and Gardner, R.H. (1991). Quantitative Methods in Landscape Ecology. Springer-Verlag, New York, USA. Turner, M.G., Gardner, R.H. and O’Neill, R. (2001). Landscape Ecology in Theory and Practice: Pattern and Process. Springer-Verlag, New York. Turner, M.G., Pearson, S.M., Bolstad, P. and Wear, D.N. (2003). Effects of land-cover change on spatial pattern of forest communities in the Southern Appalachian Mountains (USA). Landscape Ecology 13: 449-464. URL-1. (2008). http://www.usgcrp.gov/usgcrp/Library/ocp2008/ocp2008-hi-land.htm, 2008, A Report by the Climate Change Science Program (CCSP) and the Subcommittee on Global Change Research, A Supplement to the President's Fiscal Year 2008 Budget. URL-2. (2008). http://glcfapp.umiacs.umd.edu:8080/esdi/index.jsp Upadhyay, T. P., Sankhayan, P. L., & Solberg, B. (2005). A review of carbon sequestration dynamics in the Himalayan region as a function of land-use change and forest/soil degradation with special reference to Nepal. Agriculture, Ecosystem & Environment 105: 449–465. Xian, G., Crane, M., Su, J. (2007). An analysis of urban development and its environmental impact on the Tampa Bay watershed. Journal of Environmental Management 85: 965976. Xu, C., Liu, M., An, S., Chen, J.M., Yan, P. (2007). Assessing the impact of urbanization on regional ne primary productivity in Jiangyin County, China. Journal of Environmental Management 85: 597-606. Wei, Y. and Hoganson, H.M. (2005). Landscape impacts from valuing core area in national forest planning. Forest Ecology and Management 218: 89–106. Wakeel, A., Rao, K.S., Maikhuri, R.K., Saxena, K.G. (2005). Forest management and land use/cover changes in a typical micro watershed in the mid elevation zone of Central Himalaya, India. Forest Ecology and Management 213: 229-242. Watson J, Whittaker R, Dawson T. (2004). Habitat structure and proximity to forest edge affect the abundance and distribution of forest-dependent birds in tropical coastal forest of southern Madagascar. Biological Conservations 120: 311–327. Werbist, B., Putra, A.E.D., Budidarsono, S. (2005). Factors driving land use change: Effects on watershed functions in a coffee agroforestry system in Lampung, Sumatra. Agricultural Systems 85: 254-270. Weng, Y.C. (2007). Spatiotemporal changes of landscape pattern in response to urbanization. Landscape and Urban Planning 81: 341-353. Yolasığmaz, H.A. (2004). The concept and the implementation of forest ecosystem management (A case study of Artvin Planning Unit). PhD thesis, Karadeniz Technical University Faculty of Forestry. Trabzon. pp 185. Zhao, W.Y., Li, J.L., Qi, J.G. (2007). Changes in vegetation diversity and structure in response to heavy grazing pressure in the northern Tianshan Mountains, China. Journal of Arid Environments 68 465-479. Zhang, Y., Song, C. (2006). Impacts of afforestation, deforestation, and reforestation on forest cover in China from 1940 to 2003. Journal of Forestry, October-November, pp 383-387.
Spatiotemporal Dynamics of Land Use/Land Cover and Timber Carbon Storage…247
QUESTION BANK Expand the Following 1. 2. 3. 4. 5.
LULCC TM ETM NASA UTM
Short Answer Questions 1. 2. 3. 4.
Define the terms land use and Land cover. Define Landscape. What is landscape ecology? Define the term Biomass. How is it linked with carbon storage? What is landscape matrix? List some of the landscape matrices and their significance used for understand land use land cover dynamics? 5. What is transition probability? Why is it importance in understanding land use land cover dynamics?
Long Answer Questions
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. What is LULCC? Why is importance of studying LULCC? What are different methods to quality this? 2. What is the status of land use land cover change in your country? What are the different initiatives taken for understanding this phenomena?
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohaparta et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 11
MAPPING OF TERRESTRIAL CARBON SOURCES AND SINKS THROUGH REMOTE SENSING AND MODELING Ajit Govind* and Jing Ming Chen† Department of Geography and Program in Planning University of Toronto, Toronto, Ontario
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ABSTRACT The terrestrial carbon (C) cycle has a great role in influencing the climate. Hence, it is critical that we fully understand the scale-specific (spatial and temporal) complexities of the terrestrial C cycle. Remote sensing, in combination with modeling and precise ground measurements, is the only means to better understand the C cycle across a wide range of scales and help us arrive at global-scale conclusions. This chapter introduces various remote sensing approaches that are currently used in C cycle research. Various empirical and process-based strategies to map C indicators are discussed at the outset. Further, we examine the application of remote sensing techniques to deriving spatial datasets that aid in mapping C indicators. Here we emphasize different global-scale datasets that are currently available to remote sensing scientists and ecological modelers. Subsequently, we discuss a scheme to reconstruct the historical C-balance of an ecosystem using the InTEC model that primarily relies on remote sensing datasets. Here, we demonstrate how the current biomass and soil C pools result from the integrated effects of plant growth, stand age, climate change, atmospheric CO2 concentration, Ndeposition and disturbance. Since validation is essential in gaining confidence in remote sensing-based estimates, we discuss various approaches that measure the fluxes of C between the biosphere and the atmosphere. Because remote sensing alone is insufficient in explaining the complexity and non-linearity in the C cycle processes, it is essential to understand how different environmental factors influence plant physiology and biogeochemistry. To this end, we discuss the usefulness of a spatially explicit process model (BEPS-Terrainlab V2.0) that has a tight coupling between hydrological, ecophysiological and biogeochemical processes that runs within a remote sensing-driven modeling framework. * †
[email protected] [email protected]
250
Ajit Govind and Jing Ming Chen
Keywords: C-cycle, modeling, remote sensing, terrestrial, Terrainlab
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
INTRODUCTION Since the industrial revolution, the mean global CO2 concentration has risen from about 280 ppm to over 379 ppm (Keeling et al., 1995; IPCC, 2007). This rapid rise in atmospheric carbon dioxide concentrations occur due to the imbalances between the rates at which anthropogenic and natural sources emit CO2 and the rate at which the global C sinks remove CO2 from the atmosphere (Baldocchi et al., 2001). The first order effect of increased atmospheric CO2 is the increase in surface temperature of the Earth that subsequently leads to a climate change (Hansen et al., 1997; Frolking et al., 2004). It is plausible that a further increase in global average temperature of about 0.4°C over the next two decades and of 1.8 to 4.0°C by the end of this century (IPCC, 2007). Terrestrial ecosystems determine the atmospheric C balance through many mechanisms. In pristine terrestrial ecosystems, they are controlled mainly by photosynthesis and respiration. Other mechanisms that govern its C balance in these ecosystems include methane fluxes and lateral transport of dissolved organic carbon and spontaneous release of terrestrial C in the event of forest fires. Significant advances have recently been made in our understanding of the role of terrestrial ecosystems in the global C cycle by several approaches such as eddy covariance and tall tower measurements, inversion of atmospheric measurements, forest inventory, and simulation models. In the recent years, we have learnt that terrestrial ecosystems play a significant role in taking up atmospheric C as the net effects of photosynthesis and respiration, under conditions of negligible disturbance such as fire or biomass burning. Without this biosphere control, the amount of CO2 in the air would have risen more rapidly, and the global warming would have been greater than what we are experiencing in the last several decades. Accurate assessments of regional and global-scale changes in the terrestrial biosphere are essential to better understand the anthropogenic impacts on the global climate and its direct consequence on social, economic and geopolitical aspects. There are uncertainties regarding the dynamics of the terrestrial C cycle. The heterogeneous nature of terrestrial ecosystems presents a major challenge in our effort to improve regional and global C cycle estimation. Inadequate information on the C budget poses a great challenge in improving our understanding of the global climate change because there is a missing C sink. This issue has also become a major knowledge gap in formulating policies related to climate change, given the options of including sinks in national greenhouse gas inventory under the Kyoto protocol. Remote Sensing is the only technique that can monitor terrestrial processes across large ranges of spatial and temporal scales in addition to its great versatility in its applications. There are a number of ecosystem attributes that need to be monitored. However, dynamics of the vegetation cover are attached with greater importance due to the large human impacts on this terrestrial C reservoir which is mostly on the verge of conversion from a sink to source of C. Ever since the first polar orbiting meteorological sensors launched in the seventies, Advanced Very High Resolution Radiometer (AVHRR) and the series of high resolution Thematic Mappers, we have been served with multipurpose data for almost three decades. Currently, various sensors mounted on space borne platforms have been giving us much data
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
251
for quantitative description of Earth surface’s heterogeneities at various spatial, temporal, radiometric, spectral and angular resolutions. There has been a revolution in Earth science ever since sensors such as VEGETATION on board of SPOT-4; and MODIS, CERES, MOPITT, ASTER and MISR on board of Terra, etc., became operational. Contemporaneous advances in space research internationally have led to the development of a series of other satellites having sensors for specific applications. Remote sensing techniques have given us indispensable information based on which much improved understanding is made of terrestrial and oceanic C cycle processes. Currently, terrestrial C research is moving towards the creation of a terrestrial observing system that integrates field-based measurements, flux towers, remote sensing, and ecosystem modeling (Baldocchi et al., 1996; Running et al., 1999). Satellite remote sensing has been shown to be powerful tools for conducting local, regional, and global ecological modeling studies. This helps scientists to bridge the knowledge gaps that exist because of large spatial and temporal scales associated with the terrestrial processes, as remote sensing techniques can serve as the link over these gaps. In the recent years, the remote sensing of land surface, ecosystem modeling and ground measurements have become inseparable and are often synergistically linked to better understand the earth surface processes. This chapter introduces some of the active research areas in terrestrial C mapping using geoinformatics with special emphasis on remote sensing.
Some Basic Ecological Terms in Relation to Terrestrial Carbon Cycling Plants form the basis for any ecological niche. This is because vegetation is the primary producers from which mass and energy gets transformed to other living beings. The process of photosynthesis fixes the atmospheric C into the biosphere through ecophysiological processes that are controlled by a variety of environmental factors. Hence photosynthesis is the primary process that we should accurately simulate in order to understand the terrestrial C cycle. At stand or ecosystem scales, the C fixed through photosynthesis is called the gross primary productivity (GPP). A portion of C in GPP is consumed by autotrophic respiration, Ra. That is used for both maintenance, Rm and plant growth, Rg.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Ra = Rm + Rg
(1)
The portion of C that remains in the living biomass after autotrophic respiration is called net primary productivity (NPP), expressed in units of g C m2 yr−1 or t C ha−1 yr−1.
NPP = GPP − Ra
(2)
NPP could be estimated by a number of methods. Since ecosystem studies are difficult to generalize for large areas if we have only point measurements such as in forest inventories, we need to rely on measurement techniques that are temporally frequent and spatially broad, as well as economical and repeatable. As such, the only technique is regional scale modeling using remote sensing products as the inputs.
252
Ajit Govind and Jing Ming Chen
The net exchange of C between the biosphere and the atmosphere is mostly calculated as the difference between NPP and heterotrophic respiration, Rh. This quantity is termed as net ecosystem productivity (NEP).
NEP = NPP − Rh
(3)
Rh results from the decomposition of dead organic material in soils, the litter layer due to microbial and animal respiration. Conventionally, when NEP > 0, the land surface is considered as a C-sink, i.e., it absorbs more C than what is released to the atmosphere. Net biome productivity (NBP) is a term that is used when we consider the C losses due to disturbance at the biome level on the overall terrestrial C balance.
NBP = NEP − Disturbance
(4)
Disturbance losses can include forest fire, deforestation, insect defoliation, diseases, partial thinning etc.
Physics of Remote Sensing
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In order to fully appreciate remote sensing as a science to study the Earth system processes, it is important to understand the physical processes that operate behind this technique. The following is a brief recap of the basic physical principles behind remote sensing. Radiant energy from a source, in the form of photons, moves as a flux. This radiant flux (Φ) is defined as the rate at which the energy passes an area. If we express radiant flux per unit area, it is called as radiant flux density. Radiant flux density coming from an external source (e.g. sun) on the surface of a body is known as irradiance, which has a unit of Wm-2. If we consider that the energy is radiated through a given solid angle in space, we call it as radiant intensity which has a unit of WSr-1, where Steradians (Sr) is a unit of solid angle, a cone angle, in which the unit is a radians or 57o17′44″. There are 4 π Steradians on a complete sphere. In Figure 1a it can be seen that since the solid angle Ω=Area. Distance-2, we can derive an important radiation entity known as radiance. Radiance (L) is defined as the radiant flux per unit solid angle (
φ
Ω
) leaving a source having a projected surface area
(ACosθ). In other words, it is the radiant flux density per unit solid angle, Ω. It has a unit of WSr-1m-2. Radiance is the physical quantity that is captured by a sensor that is mounted on a spaceborne or air-borne vehicle such as the polar orbiting or geostationary satellites, aircraft, balloon or kite. Depending on the radiometric resolution of the sensor, the radiance is captured as digital numbers (DN). Normally, for most of the satellite sensors radiometric resolution exists in 256 grey shades that abstract the actual radiance into an 8 bit data format. This necessitates that prior to using remotely sensed data one needs to calculate the actual radiance by employing a linear equation which has the ‘Gain’ and the ‘Offset’ terms.
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
253
Figure 1a. Schematic representation of the concept of Radiance. (1b) Variation of the Earth- Sun distance in astronomical units for an annual time span. (1c) Solar spectral irradiance values for different bands of the ETM sensor. Details in figure 1b and 1c should be explicitly considered while preprocessing any remote sensing imagery using sensor specific values.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
However, since radiance values of a remote sensing imagery depend on a variety of environmental factors, it becomes difficult to analyze the inherent physical property of the reflecting surface based on radiance of the surface alone. To circumvent this and to reduce in between-scene variability, a normalization of the radiance in a particular band is carried out with solar irradiance in a particular band. This normalized product is known as reflectance. The reflectance of a surface is the characteristic feature of the surface alone. This is computed using the following formula.
ρ=
where,
π .Lλ .d 2
(5)
E sun ,λ .Cosθ s
ρ = Planetary reflectance (unitless), Lλ = Spectral radiance at the sensor's aperture 2
for a particular band, d =
Earth-Sun distance in astronomical units interpolated from
values listed in Figure 1c, E sun ,λ = Mean solar exo-atmospheric irradiances for a particular band (see Fig. 1c),
θ s = Solar zenith angle in degrees
254
Ajit Govind and Jing Ming Chen
The reader is advised to refer Landsat 7 Science Data User’s Handbook for more information. See Figure 1c for the band specific solar irradiance values that are used to calculate the reflectance of Landsat 7 imagery. Using equation 5, one can calculate the surface reflectance for each band using the corresponding irradiance and radiance values. This is a prerequisite step before we calculate spectral indices such as NDVI or RSR (Reduced Simple Ratio,RSR). Spectral indices are normally used to study vegetation because they compress the information that exists in various bands. These spectral indices aids in retrieving various biophysical properties of a land surface that are remotely sensed, provided that they are derived using a strict protocol that are in accordance with the underlying physics. The above mentioned operations assume that the reflecting surface emits radiance in all directions (hemisphere) with equal strength. Such surfaces are called Lambertian surfaces. For Lambertian surfaces, irradiance is π times its radiance. This is the reason why we use π in the numerator of equation 5 for calculating the reflectance. This π term integrates the radiance over a conceptual hemisphere. Natural surfaces such as forest canopies are generally non-Lambertian reflectors. Instead, they tend to display varying degrees of “anisotropy” or directionality (Fig. 2). This boils down to the fact that surface reflectance is not only a function of the spectral, spatial and polarizing properties of the target, but also a function of the direction from which the surface is illuminated and viewed (Fig. 3).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 2. Schematic representation of two types of reflecting surfaces. Lambertian surface where the directionanity is uniform in all directions, and Non lambertian surface where it is non uniform(anisotrophic).
The degree to which incident radiation on a surface is reflected anisotropically is determined by several factors such as the density and arrangement of objects on the surface, and hence the nature of the shadowing caused by these objects as a function of viewing and illumination zenith and azimuth angles as well as the intrinsic directionality of the reflectance, transmittance and absorption properties of the scattering materials. This directionality of surface reflectance can be exploited to retrieve information regarding the surface structure. Physical models that describe these processes are called as Radiative Transfer (RT) models.
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
255
Figure 3. Effects of anisotropic nature of a non-Lambertian surface
RT models can simulate reflectance spectra as a function of surface structural and biophysical characteristics as well as the zenith and azimuth angles of the illumination and sensor. The currently existing RT models vary with respect to various processes they describe at various scales (e.g. leaf biochemistry, multiple scattering, etc.). It can also invert reflectance spectra to derive various surface structural parameters. In this chapter, we often describe an RT model, the Four Scale model (Chen and Leblanc, 1997a). Currently, remote sensing in conjunction with RT models serve as invaluable tools to map various vegetation parameters that are otherwise impractical to retrieve using remote sensing data alone.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Remote Sensing as a Tool to Map Terrestrial Carbon Remote sensing data are currently available in a wide range of temporal and spatial resolutions. The information that these sensors capture in the optical region of the electromagnetic spectra are very useful for retrieving a large number of vegetation attributes. With the combined use of empirical or process models, it is possible to simulate various ecosystem processes that represent terrestrial C dynamics in a robust manner. However, the types of parameters that indicate the terrestrial C distribution depend on the sophistication of the model(s) that are used in conjunction with remote sensing. For example, it is easy to simulate GPP or NPP using simple models. However, the simulation of NEP or NBP requires rigorous description of ecosystem processes along with the representations of biogeochemical processes and feedback relationships, effects of disturbance and non disturbance factors in the recent past, age effects of the forest stands, forest succession, etc., need to be addressed in order to simulate parameters such as NEP and NBP.
256
Ajit Govind and Jing Ming Chen
Empirical Methods of Mapping NPP Monteith et al. (1976) first illustrated the positive linear relationship between NPP and Absorbed Photosynthetically Active Radiation (APAR). Solar radiation, the factor that determines vegetative growth, has Photosynthetically Active Radiation (PAR), that is essentially the visible region of the solar spectrum. The basis of this model is that photosynthetic fixation of C by leafs is proportional to absorbed visible quanta which are equated using a species specific parameter known as the light use efficiency denoted as ε. For this reason, these types of models are also referred to as epsilon models.
NPP = ε . APAR
(6)
where, NPP is in gC m-2 t-1 and APAR is in MJ m-2 t-1. The fraction of photosynthetically active radiation, fPAR is used to calculate the APAR as given below.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
APAR = fPAR × PAR
(7)
fPAR is normally expressed as a liner function of NDVI, which in reality is not always the case. Even though the light assimilation is non linear for individual leaves, it has been found to be linear for most canopies. Epsilon models are being widely used to simulate the regional scale NPP on account of its simplicity. Although it is simple, there could be many uncertainties in the large scale NPP estimates that are derived using this approach due to non-linearity existing in various biophysical processes. Many studies have demonstrated that significant scatter in the relationships between spectral vegetation indices and fPAR and therefore, it is likely that APAR and NPP need not necessarily relate linearly. This could be attributed either due to errors in the remote sensing based retrieval of fPAR or due to heavily abstracting the entire complexity of plant physiology in a single parameter, the light use efficiency. Errors in the retrieval of fPAR could be attributed to a number of factors that include poor characterization of the background contribution to scene reflectance and inadequate characterization of interception by non-photosynthetic elements of the canopy. In many of the studies, ε has been assumed to be a constant for a vegetation type or assume to vary only at various phonological stages. Measured values of ε are rare, especially for pristine vegetation types. In reality, variations in ε occur through out the growing season because of the dynamics of both environmental and plant physiological parameters that constrain ε. Similarly, annually averaged values of ε may vary on an inter-annual basis due to variations in climate and edaphic conditions, particularly for species that change the leaf area significantly. The global NPP and GPP products are derived using MODIS data (e.g. MOD15A2) using an epsilon model. There are many ways of deriving a temporally varying ε. One method is by constraining a maximum ε value by scalars of various environmental factors, resulting in a realistic and dynamic ε.
ε t = ε max .∏ f i
(8)
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
257
where, f i represents a particular scalar (0-1). There has been increasing recognition of the importance of respiration that is not directly related to APAR but is related to biomass, temperature and moisture.
Explicit Calculationof fPAR fPAR, the fraction of photosynthetically active radiation can be estimated based on empirical approaches that are linearly or non-linearly related to spectral indices such as NDVI. Global NPP or GPP products are calculated with the assumption that fPAR = NDVI (Running et al., 2004). Nowadays many groups calculate fPAR using a simple Beer's Law approach (Jarvis and Leverenz, 1983).
fPAR = 1 − e( LAI × − K )
(9)
However, these approaches are not precise because many of the biophysical and physiological relationships and feedback relationships are overlooked. An alternative and ideal strategy to calculate fPAR could be based on physical principles. One such algorithm was proposed by Chen, (1996b) based on earlier studies (Goward et al., 1994). For daily applications, it is warranted to calculate a ‘daily’ fPAR by temporally integrating the instantaneous fPARs. The instantaneous fPAR, denoted by fPARθ, can be calculated in the following manner.
fPAR θ = [1 − ρ 1θ ] − [1 − ρ 2 θ ]. exp
⎛ − G t (θ ) L e ⎞ ⎜ ⎟ ⎝ Cos θ ⎠
(10)
where, Gt(θ) is the projection coefficient for total PAR transmission, a canopy architectural parameter. This will be 0.5 if the angular distribution of elements (leaves) in the canopy is random. Le is the effective LAI, and θ is the solar zenith angle at a given instant. The cosine of the solar zenith angle can be calculated as given below.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
⎡ ( t − 12 ) π ⎤ cos θ = sin φ . sin δ + cos φ . cos δ . cos ⎢ ⎥ 12 ⎣ ⎦
(11)
where, φ is the latitude; δ is the solar declination; and t is the solar time in decimal hours.
δ=
[360.(DOY × 10)] 365
(12)
Since the instantaneous fPAR varies with the time of the day, we need to convert it into daily fPAR by using a temporal upscaling as a function of the diurnal variation in solar zenith angle as given by Chen (1996b). This facilitates it to be ideally related to vegetation indices derived from satellite platforms.
258
Ajit Govind and Jing Ming Chen π
fPARdaily =
2
fPARθ . cosθ .dθ ∫ θ min
π
2
∫ cosθ .dθ
(13)
θ min
π
2 1 = . ∫ fPARθ . cosθ .dθ 1 − sin θ min θ min
(14)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
where, θmin is the solar zenith angle at local noon and θ is the instantaneous solar zenith angle. fPARθ is the instantaneous fPAR. Figure 4 shows the variation of fPAR vis-à-vis the solar zenith angle during the diurnal cycle of a high latitude boreal ecosystem having a mean LAI value of 3.56.
Figure 4. Diurnal variation of fPAR for high latitude (49oN) forest ecosystem having an LAI (here 3.56) during a mid summer day. Note the variation of cosine of the solar zenith angle.
Other sophisticated methods of the calculation of fPAR based on radiative transfer modeling can be found in Myneni et al. (1997b)
Process Modeling of Photosynthesis Taking into account the fundamental plant physiological and biochemical processes of photosynthesis, dark respiration and allocation of photosynthates, process models can be developed. An example for a widely used leaf-level and instantaneous process model is the Farquhar's photosynthesis model (Farquhar et al., 1980).
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
259
Unlike an epsilon model which is highly empirical due to the site specificity of fPAR and ε, a process model could simulate C assimilation by plants under any meteorological conditions because of the involvement of the fundamental biophysics and plant physiology that govern the photosynthetic, dark respiration and C allocation processes, with minimal calibration. Hence, a process based model is superior over the empirical methods. However this model cannot be directly used for regional scale or at a daily time scale because this is a leaf level and instantaneous model. To circumvent this, there are various spatial upscaling strategies from leaf to canopy and temporal upscaling from instantaneous to daily or annual time steps that have been adopted by the scientific community. Spatial and temporal upscaling strategies are a prerequisite before using it for large scale applications in conjunction with remote sensing derived spatial inputs such as LAI. The following is a description of spatial and temporal upscaling strategy used in the current versions of the Boreal Ecosystem Productivity Simulator (BEPS) family of models (discussed later on). Temporal upscaling of the Leaf Scale Farquhar Photosynthesis model Farquhar et al. (1980) postulated that the process of photosynthesis for a C3 leaf at an instant can be approximated as the minimum of Rubisco-limited and light-limited gross photosynthesis rates.
A = min(Wc , W j ) − Rd
(15)
where, A is the net photosynthesis, Wc and W j are Rubisco-limited and light-limited gross photosynthesis rates in
Wc = Vm
Ci − Γ Ci + K
μmolm −2 s −1 respectively. Rd is the dark respiration. and
Wj = J
Ci − Γ 4.5Ci + 10.5Γ
where, Vm is the maximum carboxylation rate in
(16)
μmolm −2 s −1 ; Ci is the intercellular CO2
concentration; K is a function of enzyme kinetics J is the electron transport rate in μmolm s ; J is dependent on photosynthetic photon flux density (PPFD) absorbed by Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
−2
−1
the leaf. Chen et al. (1999a) developed an analytical solution for temporally upscaling equation 15 after equating it with A = g (Ci − Ca ) , another form of representing the process of photosynthesis (Leuning et al., 1990). Here, g is the conductance of CO2 through the pathway from the atmosphere outside of leaf boundary layer (μmol.m-2s1Pa-1) and Ci and
Ca are the intercellular and atmospheric CO2 concentrations, respectively. After accounting for the diurnal variation of g as function of several diurnally varying environmental controls, it is possible to yield two systems of equations as given below, which can be further be integrated using the extreme values of g as the boundary conditions.
260
Ajit Govind and Jing Ming Chen
Ac =
⎡(C a + K ) g + Vm − Rd ⎤ ⎢ ⎥dg 2( g n − g min ) g∫min ⎣⎢− ((C a + K ) g + Vm − Rd )2 − 4(Vm (C a − Γ) − (C a + K )Rd )g ⎦⎥ gn
η
(17) Aj =
⎤ ⎡(C a + 2.3Γ) g + 0.2 J − R d ⎥dg ⎢ 2 2( g n − g min ) g min ⎣⎢− ((C a + 2.3Γ ) g + 0.2 J − R d ) − 4(0.2 J (C a − Γ) − (C a + 2.3Γ )R d )g ⎦⎥
η
gn
∫
(18)
η (≈1.27) is the correction factor that is applied to account for the assumption that the diurnal pattern of g vary similar to the solar zenith angle, θ , which is a In these equations,
reasonable assumption because most of the environmental variables vary in accordance with θ . The reader is advised to refer to Chen et al. (1999b) for the complete derivation of this analytical solution for temporally upscaling the instantaneous Farquhar equation. This temporally integrated model is adequate for remote sensing applications that operate normally at a daily time step, simultaneously preserving the physiological meaning of photosynthesis as well as the mathematical consistency.
Spatial Upscaling of the Leaf Scale Farquhar Photosynthesis Model Although regional scale empirical modeling strategies are effective in capturing the trend of spatial distribution of terrestrial C, they are abstractions of the reality, especially in ecosystems that have canopies having complex architectures that affect their radiative transfer mechanisms. In order to spatially upscale a temporally upscaled leaf-level model to canopy scales, one needs to conceptualize variability in various biophysical factors that affect photosynthesis. Since light/radiation is the primary control on photosynthesis, there has been a convention to designate a canopy scale model based on the conceptualization of ‘leaf types’ that occur within a canopy. Some examples include ‘big leaf’, ‘Sunlit-shaded leaf’, ‘multi layer canopy’, etc. The simplest one is the ‘big leaf approach’, which assumes that the ecosystem (or the modeling unit, a pixel) is a big leaf. The abundance of various leaf types is quantified in terms of their respective Leaf Area Indices (e.g. sunlit leaf area index, LAI sun or
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
shaded leaf area index, LAI shade ) that can be used as weighting factors to upscale a leaf-level model to the canopy. Although a big-leaf model is easy to implement by simply using a canopy conductance g canopy ( g canopy = g s × LAI ) instead of stomatal conductance ( g s ) in a leaf level photosynthesis model, intuitively speaking, big leaf models do not capture the nonlinearities associated with the ecophysiological processes. Many studies including Chen et al. (1999) highlights the importance of upscaling canopy scale C cycling processes adopting a sunlit-shaded leaf scheme that uses a light regime specific g s . The temporally upscaled leaflevel model can be run with daily meteorological inputs and then spatially upscaled using the logic shown below.
A = [AsunLAIsun ] + [AshadeLAIshade]
(19)
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
261
Although the variations in the light regime within a canopy are accounted in the sunlitshaded schemes, physiological variability still exists within a given light regime. Govind et al. (2007a) has developed a ‘four leaf’ conceptualization by accounting soil water constraints within a given light regime. The modified spatial upscaling strategy by the simultaneous inclusion of light and root wetness as the environmental stress factors that affect stomatal dynamics and plant physiological process of photosynthesis is shown below.
[
]
A = [Asun, sat LAIsun.μ + Asun,unsatLAIsun (1 − μ)] + Ashade,sat LAIshade.μ + Ashade,unsat.LAIshade(1− μ)
Asun ,sat = f ( g s ,sun ,sat , φ ) Asun ,unsat = f ( g s ,sun ,sat , θ ) Ashade ,sat = f ( g s ,shade ,sat , φ )
(20)
Ashade ,unsat = f ( g s ,shade ,unsat , θ ) where, A is for a leaf within a given radiation and hydrologic regime, is calculated based on the function f which is the temporally integrated leaf-level Farquhar model. g s , sun , sat is the stomatal conductance of a sunlit leaf that corresponds to a saturated soil water condition ( φ ).
g s , sun ,unsat is the stomatal conductance of a sunlit leaf corresponding to an unsaturated soil water condition ( θ ). g s , shade , sat is the stomatal conductance of a shaded leaf corresponding to an saturated soil water condition, and g s , shade,unsat is the stomatal conductance of a shaded
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
leaf corresponding to an unsaturated soil water condition.
μ is the fraction of the root lying
in the unsaturated zone and calculated as a function of root geometry and water table position. More detailed strategies could be developed in similar lines by following a ‘nested approach’ with fractions of LAI and their respective stomatal conductance as the basic idea. Figure 5 shows various resource limitations or environmental stress regimes endured differentially by various vegetation parts. Here only the variations in light and water regimes are considered in a bimodal manner to develop a conceptualization for spatial upscaling. Another alternative for spatial upscaling is the multilayer strategy Bonan (1993) and Foley (1995). This scheme is less popular because it is not as effective in capturing the radiation gradients within a canopy as the sunlit-shaded leaf scheme. Nevertheless, multilayer models have the advantage of explicitly describing various foliage properties in different layers such as nitrogen content and photosynthetic capacity. It is intuitive that sunlit-shaded leaf and multilayer models, or combination of them, are more reliable than big-leaf models.
262
Ajit Govind and Jing Ming Chen
Figure 5. A schematic representation of various environmental stress /resource status within a tree that affects its net ecophysiological response.
LAI is the basic canopy structural parameter needed for any process-based canopy scale photosynthesis model. However, LAI alone is insufficient to describe the effect of canopy architecture on radiation absorption and distribution within a canopy. Vegetation has various levels of foliage organization. Herbaceous canopies (crops and grasses) generally have simple structures with leaves more or less randomly distributed in space, whereas foliage in forests is often organized in structures at various hierarchical levels, such as shoots, branches, whorls, tree crowns, and tree groups. The foliage clumping index Chen et al. (1997) is a parameter that describes foliage architecture, adequately. The use of a clumping index is critical to realistically describe radiation regimes and also for the calculation of sunlit or shaded LAI fractions (Chen et al., 1999) that facilitates spatial upscaling of leaf scale photosynthesis.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
LAI = Total Leaf Area Index ⎛ ⎡ 0.5.Ω.LAI ⎤ ⎞ ⎟ LAI sun = 2.Cosθ .⎜⎜1 − exp ⎢− Cosθ ⎥⎦ ⎟⎠ ⎣ ⎝ Sunlit LAI,
(21)
θ =Solar Zenith Angle, Ω is canopy clumping index.
LAI shade = LAI − LAI sun
(22)
As foliage clumping increases (low values for Ω ) for a given LAI, i.e., leaves are more aggregated as ‘clumps’ resulting in an increase in the shaded LAI and more diffuse radiation and vice versa. This canopy architecture-mediated change in the radiation quality affects
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
263
canopy scale photosynthesis significantly. As the clumping index can vary across a landscape, it is highly desirable to map the spatial distribution of this index using remote sensing data for large scale mapping of terrestrial C cycling using process models.
Remote Sensing as a Tool to Derive Spatially Explicit Datasets for Carbon Modeling
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Ecosystem-scale process or empirical models that simulate C, water, and energy exchange between terrestrial ecosystems and the atmosphere require many spatial data sets as inputs (Landsberg and Waring, 1997; Coops and Waring, 2001). At regional or global scales, the only possibility to derive these data sets in a cost effective and quick manner is through remote sensing. There are a number of surface parameters that can be estimated using remote sensing that are useful for C cycle modeling (Table1). Many of these parameters are estimated using a combination of information collected by a variety of sensors (data fusion) simulation modeling, photogrammetry, GIS and ground and laboratory measurements. Several research organizations and individual groups have been creating global, regional or national scale products which are now relatively easily accessible to a modeler who is interested in simulating the earth surface processes. The significance and retrieval strategy for some of these parameters are briefly reviewed in the following section. Each modeling group has their own modeling requirements which necessitate the development of these spatial products. Nowadays, since most ecosystem models only differ marginally, it is plausible to reuse these datasets in order to run any model with minimum alterations such as resampling or reprojecting or for the development of further datasets.
Land Cover Map Land cover data are critical for terrestrial C cycle modeling. Land-cover information is required to explicitly assign biophysical parameters that pertain to different plant factional types that determine the rates of various ecophysiological, biogeochemical and hydrological processes. Although land use and land cover and their changes have been mapped by many scientists involved in remote sensing at small scales, quite frequently, development of a regional and global scale land cover product needs a collaborative effort between several research groups around the world. Globally, the most updated land cover map is the GLC 2000. The development of this global product was contributed by many countries using the SPOT 4 VEGETATION data supported by the European Space Agency. For example, the North America part of the map was jointly produced by the Canada Centre for Remote Sensing and the US Geological Survey (Latifovic et al., 2004) using the method of classification by progressive generalization initially developed by Cihlar et al. (1998). Indian’s contribution was made by the Indian Institute of Remote Sensing, Dehradun. There are 35 cover types based on Natural Vegetation Classification Standard adopted by the US Federal Geographic Data Committee. These are matched with 23 cover types in GLC2000. For C modeling, these cover types are grouped into several functional types, such as deciduous forest, conifer, grassland, cropland, tundra, barren, etc., and C related parameter sets are assigned to these grouped cover types. Table 1 gives an idea about the various land cover specific parameters used for process modeling of terrestrial C cycle.
264
Ajit Govind and Jing Ming Chen
Leaf Area Index Terrestrial vegetation, being the primary producers, absorbs solar energy to convert atmospheric CO2 to biosphere biomass. However solar radiation is intercepted by plant parts that contain chlorophyll, the leaves. One of the promising applications of remote sensing as a tool to study the terrestrial C cycling process is through the estimation of leaf abundance that intercepts solar radiation on a landscape, in terms of Leaf Area Index (LAI). Leaves are the interfaces where the energy and mass exchanges occur between the atmosphere and the vegetated biosphere. LAI is a very important parameter for modeling many ecosystem processes. They include canopy radiative transfer, canopy hydrology (interception, canopy evaporation), surface energy balance, etc. Besides this, LAI is an important parameter required for any process based canopy photosynthesis model. This is because it is used as a weighting factor to upscale leaf-scale photosynthetic processes to the canopy scale, in process models such as BEPS on one hand, and fPAR calculation in epsilon models on the other. In big-leaf models, it can also be used to convert stomatal conductance ( g stomata ) to canopy conductance ( g canopy ) directly (as in big-leaf models) or for upscaling leaf-level photosynthesis to canopy photosynthesis using a sunlit shaded or the four-leaf scheme. LAI is also used as an upscaling parameter for the ecophysiological process of evapotranspiration using the Penman-Monteith approach (Liu et al., 2003). Since LAI controls ET, it could affect the local water balance which directly affects the C cycling mechanism through biogeochemical feedbacks that are essentially controlled by hydrology. LAI is defined as one half the total leaf area (all sided) per unit ground surface area (Chen and Black, 1992). This definition is suitable for both broadleaf and needle leaf forms and is the most accepted definition as of now in environmental biophysics (Jonckneere et al., 2004). LAI can be formulated as given below.
L=
(1 − α ).Le .γ Ωe
(23)
Where, L is the actual LAI; Le is the Effective LAI; (1 − α ) is the woody to total area ratio;
⎡γ ⎤ ⎥ is the clumping index denoted as Ω . Here, Ωe is the elemental clumping ⎣ Ωe ⎦
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
the term ⎢ index and
γ is the needle to shoot area.
Remote sensing can give satisfactory estimates (≈77 to 79% of the variability) of regional scale LAI, if a standard and strict protocol is followed along with the use of remote sensing imagery that is corrected for terrain, atmospheric and BRDF effects. The remaining discrepancy can be attributed to background reflectance and multiple scattering within canopies. The openness of the overstory and the spatial and temporal variations of the understory vegetation in forests pose special challenges to the retrieval of canopy LAI from remotely sensed data. Chen and Cihlar (1996) estimated the canopy LAI from the normalized difference vegetation index (NDVI) derived from TM images over the Boreal Ecosystem-Atmosphere Study (BOREAS) flux tower sites. The correlation between the retrieved LAI and the field-
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
265
measured LAI was small, with R2 of 0.5 in the late spring and 0.42 in midsummer examples of their results. These suggest that the effects of understory vegetation are important for the total reflectance of sparse canopies, typical in boreal forest ecosystems. We expect that similar background effects are plausible even in sparse forests such as savannah, monsoon forests, etc., where the overstory is thin. Therefore, vegetation indices used for retrieving the large scale LAI for pristine ecosystems should be carefully selected and should be calculated following a standard protocol. Simultaneous ground measurements of LAI and leaf optical properties should also be conducted using direct or indirect methods. Many ground based optical instruments measure canopy gap fraction based on radiation transmission through the canopy. Assuming a random distribution of leaves, the effective LAI (Le) can be calculated from the gap fraction by adopting Miller’s theorem (Miller and Norman, 1971) summarized in the following equation. π
2 ⎡ 1 ⎤ Le = 2. ∫ ln ⎢ ⎥. cosθ . sin θ .dθ P θ ( ) ⎣ ⎦ 0
(24)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
where P(θ) is the gap fraction at the view zenith angle θ. Although this equation was originally developed for the calculation of LAI, Chen et al (1991) regard this as the effective LAI because leaves in a plant are not randomly distributed instead they are “clumped”. In order to get the correct LAI, clumping factors have to be considered within the LAI definition as shown in equation 23. Landscape-scale LAI is estimated by relating remote sensingderived spectral signatures (which are normally compressed into indices such as NDVI, ISR, RSR, etc.) using ground measurements of LAI along designated transacts in order to develop site specific algorithms. After generating site specific algorithm(s), LAI could be inverted from a spectral index. Figure 6 pictorially demonstrates the procedure followed to derive a location specific LAI-NDVI algorithm that can be further used to map LAI. See the combined use of remote sensing, ground measurements and laboratory analysis. Various instruments are used to indirectly estimate leaf optical properties and LAI of forest and agro-ecosystems.
Figure 6. Schematic representation of LAI retrieval the protocol as suggested in Chen et al., 2003; Chen et al., 2006.
266
Ajit Govind and Jing Ming Chen
Different optical instruments can be used to indirectly measure effective LAI and other optical properties in order to derive land cover specific LAI-Spectral Index algorithm which can be inverted to calculate the spatial distribution of LAI (Fig. 6a) A scientist operating TRAC instrument which can measure elemental clumping index and LAI (Fig. 6b) LAI2000(LICOR) measure effective LAI (Fig. 6c) Calibration of LAI-2000, which is normally done in places where canopy is absent e.g. top of an eddy covariance tower (Fig. 6d) Digital hemispherical photography can provide a cost effective method of deriving effective LAI How ever it is error prone and tedious. (Fig. 6e) Close up of a digital camera with a hemispherical lens (Fig. 6f). An example of a digital hemispherical photograph of a boreal conifer stand having and LAI ~4.0. (Fig. 6g) Close up view of LAI-2000 Plant Canopy Analyser (LICOR) (Fig. 6h). Accupar ceptometer (Decagon Inc.), another instrument to measure LAI, mostly used in crop canopies. Here, one should keep in mind that spectral indices generated from reflectance values should be derived from radiance values after atmospheric, terrain and BRDF corrections. Atmospheric parameters (mainly ozone, water vapor, aerosol) can be obtained from MODIS 05 dataset or radiosonde measurements that can be used as inputs to a model such as 6S for atmospheric correction. Ground-based measurements of LAI were made in many forest ecosystems over large geographical areas, providing a solid foundation for LAI map validation over Canada (Chen, 1996; Chen et al., 1997; Chen et al., 1999). For conifer forests, it was shown that the reduced simple ratio (RSR, Brown et al., 2000 ) is most significantly correlated with ground based LAI as opposed to other spectral indices such as NDVI or SR. RSR is defined as
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
RSR =
ρ s − ρ s , min ⎤ ρn ⎡ ⎢1 − ⎥ ρ r ⎣ ρ s , max − ρ s , min ⎦
(25)
where ρn, ρr, and ρs are the reflectance in NIR, Red, and Shortwave infrared (SWIR) bands, respectively, and ρs,min and ρs,max are the minimum and maximum reflectance in the SWIR band, determined from the histogram of a given image. Fernandes et al. (2003) proposed an infrared simple ratio (ISR) between NIR and SWIR reflectances, showing similar sensitivities to LAI variability, however, further studies are needed to ascertain if ISR and RSR are too sensitive to the surface water conditions and are seriously affected by rainfall events before imaging. Regional scale LAI data are generated using land cover specific algorithms that relate remotely sensed signatures toground measurements. A detailed description of the standard protocol for conducting ground-based measurements of LAI and the associated leaf optical properties can be found in Chen et al. (2006). Although they deal with boreal forests, this procedure can be extrapolated to any ecosystem. The above mentioned strategy is suitable for modeling LAI at watershed scales, or if one has explicit ground-based measurements, for different biomes or at regional scales. However, at global scales and at fine temporal resolutions, if an LAI product is required, one has to consider the combined use of remotely sensed data and a RT model for the process based modeling of LAI dynamics. Based on a geometrical optical (GO) RT model, 4-Scale, developed by Chen and Leblanc (1997b), and the LAI algorithms previously derived for Canada-wide applications, Deng et al. (2006) produced a revised global scale algorithm for
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
267
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
deriving LAI. This algorithm makes use of red, near infrared and shortwave infrared bands from a satellite sensor. Global scenes of VGT data were acquired over large ranges of solar zenith and satellite view angles, and a BRDF-based approach was used for correcting the angular effects and standardization of the remote sensing data. While the conventional approach is to conduct BRDF normalization prior to the input of reflectance values into LAI algorithms, here, BRDF effects are explicitly considered in the algorithm itself. Further, using the4-Scale model, it solves for simulating the relationships between LAI and the spectral reflectance in various bands, by fitting key coefficients in the BRDF kernels, with Chebyshev polynomials of the second kind. Since vegetation structure is specific to land cover types, the simulations are explicitly made for different plant functional types. The global land cover classification for the year 2000 (GLC2000) dataset has been used in the global LAI mapping. More detail on the theoretical basis of the algorithms of this approach can be found in Deng et al. (2006). Figure 7 (see Fig. 10h for color image) compares the global LAI product created using this algorithm vis-à-vis the MODIS LAI product (Pisek and Chen, 2007). Note the comparison of the two products with that of ground measured LAI values at four Big Foot Project sites located in USA. There are several global LAI products that have been produced by other groups using MODIS data (Myneni et al., 1997a), POLDER (Lacaze et al., 2002a), MERIS (Baret et al., 2005), and VEGETATION and ASTER (Plummer et al., 2004). In addition to the differences in band width and response functions between the various sensors, LAI retrieval algorithms used to produce these products are also quite different, causing large discrepancies among the LAI products. There are also inconsistencies in LAI definition and measurement techniques and protocols. In order to produce RS-based LAI products, Chen et al. (2006) proposed a set of LAI measurement protocols.
Figure 7. Global LAI product derived using the algorithm developed by Deng et al. (2006) using the SPOT VGT data based on a geometrical optical model, 4-Scale, This product is compared with the MODIS LAI product for the same period and with the measurements taken four “Big Foot” project sites by Pisek and Chen (2007). Note the good agreement of the SPOT-VGT LAI product with the measurements.
268
Ajit Govind and Jing Ming Chen
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Clumping Index Nilson (1971) first hypothesized the use of a leaf dispersion parameter to quantify the effect of nonrandom leaf spatial distributions on radiation transmission through plant canopies. The leaf distribution in a plant canopy can either be more regular than random, or more clumped than random. Natural ecosystems generally have clumped distributions of leaves, such as groupings of leaves in shrubs and tree crowns, etc. (i.e. at various scales). Introducing a dispersion parameter in the radiative transfer mechanism, Chen (1996) called this parameter as the clumping index Ω. Further, Chen and Cihlar (1995) developed an optical instrument named TRAC (Tracing Radiation and Architecture of Canopies) to measure the stand scale clumping index based on a gap size distribution theory of Miller and Norman (1971). Since then, the measurement of clumping index at various scales has become an integral part of LAI mapping and also to parameterize process models to realistically describe radiative transfer within plant canopies. Over the years, many research groups have collected dataset on clumping index for various ecosystems in Canada and China at stand levels and have been using for watershed scale applications. Since canopy structures vary spatially, it is highly desirable to map this index at regional, national or global scales, for useful applications. Until recently, it was not possible to do this. But, ever since the availability of multi-angular POLDER data became available, this became pragmatic along with the use of a RT model such as 4-Scale. Chen and Leblanc (2001) first demonstrated that the variation of the magnitudes of reflectance from the Hotspot (a point on the principal plane where the illumination and observation directions coincide) to the Dark-spot (a point on the principal plane where the reflectance is minimum) is mostly determined by the degree of foliage organization (clumping). Through simulations using the 4-Scale model, Chen and Leblanc (1997) demonstrated that clumped canopies (e.g. conifer) create strong shadows in the forward viewing directions, reducing the dark spot reflectance considerably than that of less clumped canopies(e.g. grass or crops). They developed an angular index based on the hotspot and dark spot reflectance. The model simulations were later validated using airborne POLDER data ( Lacaze et al., 2002b) and space-borne POLDER data (Chen et al., 2003). Recently, using a large number of model simulations, Chen et al. (2005) reinforced that the normalized difference between hotspot and dark spot (NDHD) is strongly related to the clumping index. NDHD is defined as
⎡ ρ − ρd ⎤ NDHD = ⎢ h ⎥ ⎣ ρh + ρd ⎦
(25)
where ρh and ρd are the hot-spot and dark-spot reflectance values, respectively. The relationship between NDHD and clumping Index is inverted using POLDER data for regional and global clumping index mapping (Chen et al., 2005; Leblanc et al., 2005). For this purpose, POLDER reflectance data for a particular location observed at different angles (up to 14 in one single overpass) were fitted with a simple kernel based model or simple exponential function to find the most reliable hotspot and dark spot values for a given set of observations. Using these methodologies, for the first time, Chen et al. (2005) produced a global clumping index map at 7 km resolution (Fig. 10g). From multiple anglular data taken
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
269
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
globally, it is apparent that forests are most clumped (clumping index much smaller than unity) and grassland is least clumped (clumping index close to unity), and also within the same forest type, there are large variations in the index. The multiple angle view can also tell apart the shrub-land and grassland, which are generally inseparable in a single view image. With the availability of MISR data, further studies are plausible in this direction. In mountainous areas, the main challenge is to separate the topographical effect from the angular effects on the remotely sensed spectral signatures.
Fire Scar Wildfires (especially in boreal forests) and human-induced biomass burning (in tropical forests and shifting cultivation areas) are some of the major causes of forest disturbance affecting stand dynamics, renewals, biomass and soil C pools, which in turn are important in forest C cycling ( Kurz and Apps, 1999). Emission of large amounts of C to the atmosphere from the vegetation can occur during forest fires (Amiro et al., 2002; Soja et al., 2004; Amiro et al., 2004) or biomass burning (Fernandez et al., 1999; Tanaka et al., 2001) . Forest fires and biomass burning also affect the nutrient status of the soil which could have positive effects on succeeding vegetation (Prietofernandez et al., 1993; Deluca and Sala, 2006). Many countries have maintained databases of forest fire occurrences (http://www.fire.uni-freiburg.de/inventory/database/statistic.html). For example, Canada has the Large Fire Polygon Database, which was created by the Canadian Forest Service based on provincial reports (Amiro et al., 2002), this database covers the periods from 1959 to 1995 and misses a considerable northern fire scars, in remote areas. Remote sensing data were used to fill in the gaps both in space and time in a study by Amiro and Chen, (2003), who assumed that after a forest is burned, leaf area is greatly damaged and the standing liquid water is drastically reduced, causing the reflectance in near-infrared (NIR) to decrease and the reflectance in mid-infrared (MIR) to increase. The ratio of MIR to NIR reflectance therefore increases dramatically immediately after fire. When images acquired in multiple years were used, the accuracy in fire scar mapping and dating was further improved. Fire occurrence and time is an extremely important dataset for regional scale modeling of historical C dynamics using models such as InTEC (described later). These scars represent ‘disturbance’ events that release significant amounts of terrestrial C to the atmosphere quite rapidly. Disturbance type and time result in variations in the current stand age which has a great role in determining the productivity of a forest stand. A similar ‘fire scar’ and ‘year of occurrence’ map is very useful for tropical forest ecosystems that suffer from human-induced burning as a result of shifting cultivation. If any one venture to do long term C modeling in these ecosystems, this information is very important. Autotrophic Biomass Both above and below ground biomass (roots) are living C. Hitherto, large scale biomass products have not been made available to the scientific community although global scale GPP and NPP products exist. Unlike GPP and NPP, biomass data are an accumulated product over the recent history and are needed in some C cycle models, as an initialized product in order to estimate the plant maintenance respiration. It is also required to calculate various soil C pools that are in turn required as substrates to simulate the heterotrophic respiration. Synthetic aperture radar (SAR) and LiDAR data can be used to retrieve biomass information using back scatter strength and coherence information. However, regional mapping of biomass using
270
Ajit Govind and Jing Ming Chen
SAR or LiDAR may incur additional complexity due to variations in topography, surface dielectric constant, vegetation structure, saturation of signals at high biomass values, etc., in addition to the high costs involved, and therefore, has not yet been produced for large areas.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Wetland Wetland has unique biogeochemical cycles involving both aerobic and anaerobic processes under fluctuating soil moisture conditions (water-table). Globally there are three types of wetlands: temperate wetlands or peatlands, tropical wetlands and agricultural wetlands (paddy fields). C cycling in wetlands is highly controlled by the position of water table. C release can be in both forms of CO2 and methane. Moreover, lateral transport in the form of dissolved organic C is also possible. Due to the unique nature of wetlands, special strategies are required to model wetland C dynamics because of its unique biogeochemical processes. It is possible to map wetland areas using remote sensing so that this mask can be used within models that handle wetland C cycling processes. This can be mapped either by optical RS or through radar. Although SAR is capable of regional mapping of wetlands, it suffers from variable size and shape of wetland areas as well as seasonal variation in water table affecting SAR signals. It is an area where remote sensing techniques are underutilized. Leaf Pigments Leaf pigments, such as chlorophyll, carotene and xanthophylls, absorb visible radiation and therefore affect optical remote sensing signals. Among these pigments, chlorophyll is of particular interest to C cycle modeling because its concentration per unit leaf surface area determines the photosynthesis rate under given radiation conditions (Bondada and Syvertsen, 2003). For pristine ecosystems, leaf chlorophyll content is closely related to the leaf nitrogen levels, an important parameter in C cycle models. In order to retrieve leaf pigments, leaf-level radiative transfer models has been applied to closed canopies, for leaf chlorophyll retrieval (Zarco-Tejada et al., 2001). Combining a leaf-level model with a canopy-level model, in conjunction with hyperspectral remote sensing images, Jacquemoud et al. (1995) successfully retrieved leaf chlorophyll content. However, reliable techniques are yet to be developed and tested for deriving leaf-level chlorophyll information for open canopies using hyperspectral remote sensing data. Using MERIS sensor data on board ENVISAT (270 m resolution, 15 bands), efforts were made to map leaf chlorophyll globally (Baret et al., 2005), and it is likely that this parameter is available for regional and global C cycle modeling in the near future. These are some of the commonly used spatially explicit datasets specifically used for C modeling in terrestrial ecosystems, derived using remote sensing, GIS and modeling approaches. However, there are even more datasets that can augment the present database, with increased attention to low latitude ecosystems where anthropogenic influences are dominant. They include datasets pertaining to DEM, agro ecosystems such as land degradation, deforestation, land degradation, shifting cultivation, biomass burning, irrigated area mapping, etc., to name a few, all of which facilitate the prediction of the global C cycle.
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
271
Table 1. Examples of some spatially explicit datasets useful for carbon cycle modeling Parameters
Type of satellite data Multi-spectral (R, NIR, SWIR) Multi-spectral (R, NIR, SWIR) Multi-angular (NIR) Multi-spectral (NIR, SWIR)
Major usages in carbon cycle modeling Differentiate between plant functional types Photosynthesis and respiration Radiation distribution in canopies Disturbance emission and forest age and regrowth
Biomass
SAR, LiDAR
DEM
Radar, LiDAR
Vegetation carbon stock, maintenance respiration Hydrological modelling Near Surface radiation
Problem soil
Multi-spectral (R, NIR, SWIR, TIR) Multi-spectral (R, NIR, SWIR), Photogrammetry Multi-spectral (R, NIR, SWIR), Photogrammetry Multi-spectral (R, NIR, SWIR) Multi-spectral (R, NIR, SWIR) Multi-spectral (R, NIR, SWIR)
Salt affected soils, sodic soils. Additional info to model plant growth
Clear cut, deforestation, partial cut, forest fire, insect attack, diseases. Carbon modelling Of agricultural lands Carbon modelling of agricultural lands Nutrient controls on carbon cycle
No
Multi-spectral (R, NIR, SWIR)
Hydrology Soil respiration Carbon modelling Unique biogeochemistry, soil carbon cycling Historical carbon reconstruction
http://www.globalsoilmap.net/
Land cover LAI Clumping index Fire scar
Forest age
Disturbance type
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Irrigated area Cropping intensity Global nutrient deposition map Soil texture
Wetland
SAR, Optical
Disturbance history
Multi-spectral (R, NIR, SWIR), Photogrammetry Hyperspectral
Availability at regional global scale (source) Yes
and
Yes Yes Yes http://www.na.unep.net/globalfire/ indofire/firepaper.html No Yes, Shuttle radar topography mission http://srtm.usgs.gov/ http://www.fao.org/AG/AGL/agll/ dsmw.stm Globally rare, regionally available eg. Canada.
Leaf Photosynthesis, nitrogen pigments (Partially adopted from Chen, 2005)
leaf
http://www.iwmigiam.org/info/mai n/index.asp Not yet Not yet
http://www.isric.org/ Not yet Not yet
Yes, but preliminary
272
Ajit Govind and Jing Ming Chen
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Combined use of RS and Process Models towards reconstructing the Historical Carbon Dynamics On a long term basis, the magnitudes of current C stocks (both in the biomass and soil) in an ecosystem are affected by: (i) ecosystem disturbance history (deforestation, shifting cultivation, logging, fire, effects of pest and diseases, land use change, forest management etc.); (ii) Dynamics of non-disturbance factors (climate change, CO2 fertilization and atmospheric deposition of nitrogen and nitrogen limitation), and (iii) Age-related effects of the vegetation which determine the current levels of photosynthesis, respiration and C allocation patterns. It is critical to reconstruct the long-term dynamics of C balance of an ecosystem as a function of the above mentioned factors. Firstly, in order to accurately initialize currently (e.g. 2004) existing quantity and quality (CN ratios) of various ecosystem C pools that act as substrates for the current rates and patterns of biogeochemical processes such as autotrophic and heterotrophic respiration. These reconstructed C pools are invaluable products for driving intensive biogeochemical models in a spatially distributed manner (Govind et al., 2007d). Accurate simulation of the current rates of biogeochemical processes is critical in determining the magnitudes of the spatial distribution of terrestrial C in terms of NEP or NBP rather than NPP. Secondly, the historical trend of C balance as affected by disturbance and non-disturbance factors at a regional or national scale at fine spatial resolutions is useful for geo-political aspects that tailors towards development of policy for mitigating the effects of climate change in accordance with Kyoto commitments. This is otherwise not pragmatic using forest inventory-based approaches alone. Detailed information on the sizes and nature of C pools cannot be directly obtained only based on remote sensing approach, especially the soil C pools that play a critical role in the current biogeochemical processes. However, it is possible to reconstruct the historical variations of the C pools using the current estimates of NPP and disturbance history along with the information on the historical change in climate, atmospheric CO2, nitrogen deposition, using an equilibrium model. Current estimates of NPP and disturbance history can be realistically estimated using remote sensing techniques. The Integrated Terrestrial Ecosystem Carbon Model InTEC (Chen et al., 2000) is an example for long-term C balance modeling (Chen et al., 2000; 2000a; 2000b; Ju and Chen, 2005; Ju et al., 2006b). InTEC iteratively reduces the uncertainties in C balance based on the assumption that heterotrophic respiration and NPP0, the NPP at pre-industrial period are in equilibrium. NPP0 is calculated by employing a retrospective-prospective modeling strategy based on the current estimates of NPP derived using remote sensing and other spatial data (Chen et al., 2003). Based on the inter-annual variability in the disturbance and non disturbance factors, historical C budget of an ecosystem can be reconstructed for each pixel. Processes considered within InTEC include nitrogen deposition, fixation and deposition, C release due to disturbances, forest re-growth, change in the length of the growing season, CO2 fertilization, heterotrophic and autotrophic respiration, and changes in CN ratios of different components in biomass and soil C pools. The main idea behind InTEC is the mechanistic spatio-temporal integration of the modified version of the two widely used biogeochemical models, Century (Parton et al., 1993) which simulates C and nitrogen cycles in soils, and the biochemical model of leaf photosynthesis (Farquhar et al., 1980). Since this model runs at an annual time step, a temporal integration of leaf-level instantaneous model implemented through elegant spatial and temporal upscaling algorithms using statistics of variance and
Mapping of Terrestrial Carbon Sources and Sinks Through Remote Sensing…
273
covariance of various meteorological and biological factors involved in photosynthesis. Figure 8 shows the schematic representation of the InTEC modeling framework. Disturbances events such as fire, clear cut, partial cut, insect defoliation, etc., are considered explicitly in this model as processes that directly release C and nitrogen into the atmosphere, and modify the total NPP by changing forest age class. Forest age is an important factor that affects the productivity of a forest stand. Young trees grow vigorously and sequester C, whereas in older stands, the productivity is much lower due to large amount of respiratory losses of C.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 8. Schematic representation of the InTEC modeling framework.
Remote sensing-derived LAI, disturbance type and time (implied in a current age map) are used to run InTEC. Additionally, it requires NPP, ET estimates in the current year based on process modeling. Historical trends in disturbance and non disturbance factors in the form of ‘NPP response functions’, which are used to tune the annual NPP modeled using InTEC. Figure 9a and Figure 9b show how InTEC reconstructs the C balance on two boreal ecosystems that differ in their current ages. The historical C balance is demonstrated in the form of NBP. The use of remote sensing for regional scale C modeling of this sort not only greatly improves the spatial resolution of C balance (unlike forest inventory-based estimates), but also allows estimation of changes in forest growth conditions when simulations are performed using historical climate and current vegetation data are used in process-based modeling that uses a equilibrium assumption. One unique aspect of this modeling approach based on remote sensing is that not only non-disturbance factors (nitrogen and CO2 fertilization effects, climate variables) are considered, as many process models do, but also disturbance factors (fire, insect, harvest) are explicitly considered. Table 2 demonstrates the error analysis
274
Ajit Govind and Jing Ming Chen
performed by Chen et al. (2003) to identify the influence of various spatially explicit datasets on the accuracy of the reconstructed historical C balance in terms of NBP and NEP.
Figure 9. Historical variation of NEP simulated for an undisturbed conifer stand in a boreal ecosystem in Saskatchewan, central Canada, using the InTEC model. Note the decline in productivity of this mature stand when the system reaches carbon neutral. The dark solid circles indicate the eddy covariance measured annual NEP values (-NEE). Panel b shows the historical carbon balance of a younger stand (disturbed in 1921 and re-grown). Note the positive NEP indicating its sink nature.
Table 2. Spatial datasets used to run the InTEC model to reconstruct the historical carbon balance of pristine ecosystems and vulnerability of error induction in the final NBP and NEP (Chen et al., 2003)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Dataset LAI Landcover Ref. year NPP Ref. year ET Disturbance Type Disturbance Event Nitrogen depositon Biomass (inventory) Forest Age
Error Estimate 25 20 25 25 0. Give a fuzzy c-partition U(0). 2. Compute αj(t), wj(t), with U(t-1) using equation (6). Calculate the membership matrix U= [μi,j] with αj(t), wj(t) using equation (7). 3. Compute Δ=max(|U(t-1) - U(t)|). If Δ > ε, then go to step 2; otherwise go to step 4. 4. Find the results for the final class centroids.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In the last step, a defuzzification process should be applied to the fuzzy partition data to obtain the final segmentation. A pixel is assigned to a cluster when its membership grade in that cluster is the highest.
Case Study Fuzzy clustering algorithms explained in the previous section are applied to identify Paddy, Semi-dry and Sugracane crops using IRS LISS I (Linear Imaging Self Scanner) data in the Bhadra command area for Rabi season of 1993. Bhadra dam is located in Chickmagulur District of Karnataka state. The dam is situated 50 km upstream of the point where Bhadra river joins Tunga, another tributary of Krishna river, and intercepts a catchment of almost 2000 sq km. Bhadra reservoir system consists of a storage reservoir with a capacity of 2025 M m3, a left bank canal and a right bank canal with irrigable areas of 7,031 ha and 92,360 ha respectively. Figure 2 shows location map of Bhadra command area. Major crops cultivated
Fuzzy Clustering Algorithms for Irrigated Area Classification…
311
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
in the command area are Paddy, Semi-dry and Sugarcane. Paddy transplantation is staggered over a period of more than a month and semi-dry crops are sown considerably earlier to Paddy. The command area is divided into three administrative divisions, viz., Bhadravati, Malebennur and Davangere. Satellite imageries used for the study are acquired from IRS LISS I (with spatial resolutions of 72.5 m) on dates 20th February, 14th March and 16th April in the years 1993. Figure 3 shows the standard FCC (False colour composite) of Bhadra commad area on 16th April 1993. For the study area, ground truth was collected for various crops by scientists from National Remote Sensing Agency, Hyderabad, during Rabi 1993, by visiting the field.
Figure 2. Location map of Bhandra command area
Results and Discussion Use of penalized fuzzy c-means algorithm requires selection of values for the number of clusters c, weighting exponent m, and constant v. The algorithm is implemented with c= 20,
312
D. Nagesh Kumar
15, 9, 6 and 5 clusters, the value of m between 1.4 and 1.6, and the value of v between 1.0 and 1.5. The algorithm gave good result with c = 6, 9; m =1.5 and v =1.0. Since paddy transplantation is staggered across the command area, satellite data of any one date does not represent the same growth stage at all locations. In view of this heterogeneity in crop calendar, in order to obtain complete estimate of area under any crop as well as to ensure better discriminability, satellite data of three dates as mentioned in the previous section are used to reflect the following features.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. When only semi-dry crops exist (image of 20th February, 1993) 2. When paddy is being transplanted with semi-dry crops already sufficiently grown (image of 14th March, 1993) 3. At the time of maximum ground cover and canopy growth of Paddy (image of 16th April, 1993 shown in Fig. 3).
Figure 3. False color composite of study area
Classification of Semi-Dry Crops with Single Date Imagery Penalized Fuzzy C-Means algorithm was implemented on the 20th February and 14th March multi-spectral images separately to identify semi-dry crops, with c=5 and 15 and the results are presented in Table 1. As can be noticed c=5 is performing better. Areas occupied by semi-dry crops obtained using c=5 are given in Table 2.
Fuzzy Clustering Algorithms for Irrigated Area Classification…
313
Table 1. Semi-dry crop classified using single date imagery with c = 5 and 15 Date
c
20th February, 1993 14th March, 1993
5 15 5 15
Available ground truth 86 86 86 86
Correctly Classified
Misclassified 84 63 82 70
2 23 4 16
Accuracy (%) 98 73 95 81
Table 2. Area occupied by Semi-dry crop with c = 5 Date
Number of pixels
Resolution (m)
Cropped Area (ha)
th
67,304
72.5
35,376.67
th
75,250
72.5
39,553.28
20 Feb, 1993 14 Mar, 1993
Using 14th March data, with c=5, majority of the Paddy locations were classified into water cluster and some locations to Semi-dry crop because at that time paddy is just transplanted in most of the areas and therefore; water is dominating compared to the crop seedlings. With 16th April data and c=5; 42 Paddy locations were correctly classified out of 53 available ground truth locations. Detailed results are given in Laxmi Raju (2003).
Classification of Paddy and Semi-Dry Crops Using Multi-Date Imagery To discriminate Paddy and Semi-dry crops from other classes, NIR (near infrared) band data of the three dates was used with c=5. Only 5 locations were classified into other clusters out of available 53 ground truth locations with 91% accuracy. Similarly for semi-dry crops 95% accuracy is obtained. Classified image is shown in Figure 4. Area of Paddy and Semidry crops obtained from the NIR band data multi-date imageries are shown in Table 3. Table 3. Area of Paddy and Semi-dry crop Crop
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Paddy Semi-dry
Number of pixels
Resolution (m)
Cropped Area (ha)
1,35,405
72.5
71,172
55,065
72.5
28,943
Classification of Sugarcane Crop By increasing the number of clusters (c=15), it was possible to classify sugarcane crop. However, better results were obtained by using Normalised Difference Vegetation Index (NDVI) images of the three dates, instead of raw data. NDVI is (NIR - R)/(NIR + R) or (B4B3)/(B4+B3) for IRS LISS I. Paddy, Sugarcane and semi-dry crops could be identified from this analysis and the misclassification results (confusion matrix) are presented in the Table 4. This table shows the number of pixels from the ground truth, classified into appropriate classes.
314
D. Nagesh Kumar
Figure 4. Classified multi-date imagery
Table 4. Misclassification table
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Paddy Paddy Sugarcane Semidry crop
Sugarcane 46 3 0
Semidry crop 3 14 12
Other 3 7 73
1 0 1
CONCLUSION An unsupervised fuzzy classification technique viz., Penalized Fuzzy c-means algorithm (PFCM), is evaluated to classify paddy, semi-dry and Sugarcane crops in the Bhadra command area; Karnataka, using multi temporal remote sensing imagery. By varying the number of clusters, an accuracy of more than 90% is achieved. By using multi date NDVI imageries Sugarcane crop could be distinguished from other crops. Fuzzy clustering algorithms are found to be superior to conventional classification techniques as there will number of mixed pixels due to small farm holdings in most of the command areas in India.
Fuzzy Clustering Algorithms for Irrigated Area Classification…
315
ACKNOWLEDGMENTS Author wishes to thank his former postgraduate student, Mr. K. Laxmi Raju, who has implemented PFCM algorithm and performed the analysis of Bhadra imageries as part of his M.Tech. project. Author profoundly thanks his former colleagues in the water resources group, National Remote Sensing Agency, Hyderabad viz., Mr PV Raju, Mr.S Jonna, Mr.CS Murthy and Dr. ST Chari for providing the data and ground truth.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REFERENCES Bastiaanssen, W.G.M. (1998). Remote sensing in water resources management: The state of the art, International Water Management Institute, Colombo, Sri Lanka, pp. 118. Bezdek, J.C. (1974). Fuzzy Mathematics in Pattern Classification, Ph.D. dissertation, Applied Mathematics, Cornell University, Ithaca, New York. Campbell, J. B. (1987). Introduction to Remote Sensing, Guilford, New York. Dunn, J.C. (1973). A fuzzy relative of the ISODATA process and its use in detecting compact well separated clusters. Journal of Cybernautics 3: 13-20. Laxmi Raju, K. (2003). Crop Classification using Multitemporal Imagery with Penalized Fuzzy C-Means Algorithm, M.Tech Dissertation report, Indian Institute of Science, Bangalore. Nagesh Kumar, D. (2002). Remote Sensing Applications to Water Resources. In: Research Perspectives in Hydraulics and Water Resources Engineering (Eds Rama Prasad and S. Vedula). World Scientific Publications, Singapore, pp.287-316. Space Application Centre. (1990). Crop Acreage and Production Estimation, Status Report, ISRO, Dept. of Space, Govt. of India. Strahler, A.H. (1980). The use of prior probabilities in maximum likelihood classification of remotely sensed data. Remote Sensing of Environment 10:135-163. Wang, F. (1990). Fuzzy supervised classification of remote sensing images. IEEE Transactions on Geoscience and Remote Sensing 28:194-201. Yang, M.S. (1993). On a Class of Fuzzy Classification Maximum Likelihood Procedures. Fuzzy Sets and Systems 57(3): 365-375. Yang, M.S., and Su, C.F. (1994). On Parameter Estimation for Normal Mixtures based on Fuzzy Clustering Algorithms. Fuzzy Sets Systems 68:13-28. Zadeh, L. A. (1965). Fuzzy Sets. Information and Control 8;338-353.
QUESTION BANK True or False 1. 2. 3. 4.
Irrigation is the largest consumer of fresh water. Spectral reflectance of healthy vegetation in green band is very low. Spectral reflectance of healthy vegetation in red band is very low. Spectral reflectance of healthy vegetation in near infrared band is very high.
316
D. Nagesh Kumar 5. Image classification based on maximum likelihood approach is a fuzzy clustering method. 6. Penalized Fuzzy c-means algorithm (PFCM) is a supervised fuzzy classification technique. 7. To estimate area under any crop as well as to ensure better discriminability, satellite data of multiple dates within the crop season is essential. 8. Normalized Difference Vegetation Index (NDVI) is the ratio of (NIR - R) and (NIR + R).
Answer the Following
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. What are the advantages of fuzzy clustering over hard clustering methods? 2. What are mixed pixels? Explain the challenges in image classification because of mixed pixels. 3. What are the differences between FCM and PFCM algorithms? 4. Explain misclassification table and its utility. 5. Discuss the advantages of fuzzy classification of multi-date satellites imageries for irrigated area identification over traditional methods.
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 14
REMOTE SENSING AND GIS AIDED LAND DEGRADATION ASSESSMENT IN THE GREATER MEKONG SUB-REGION Rajendra P. Shrestha*,1 and Kingshuk Roy2 1
Asian Institute of Technology School of Environment, Resources and Development, Thailand 2 Nihon University, College of Bioresource Sciences, Japan
ABSTRACT
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Land degradation has been an important concern while talking about food security and environmental conservation. Assessment of land degradation is important to prioritize the land conservation efforts as a measure for increasing food production and ecosystem services. Although there have been attempts to produce baseline desertification and land degradation information at the global and regional level, there is a need for more precise large-scale data as the earlier attempts were based on expert judgments and of coarse-scale in nature. This regional level study employs remote sensing and GIS techniques to assess land degradation in the Greater Mekong Subregion. The indicators used to asses composite land degradation were vegetation cover, runoff, water use efficiency and soil loss.
Key words: Land degradation, indicators, GIS, remote sensing.
INTRODUCTION Land degradation is believed to be a composite term describing how the land resources have altered overtime with reducing resource potential. Land degradation can be defined as any form of deterioration of the natural potential of land that affects ecosystem integrity either in terms of reducing its sustainable ecological productivity or in terms of its native biological *
Email: [email protected]
318
Rajendra P Shrestha and Kingshuk Roy
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
richness and maintenance of resilience (GEF, 1999) arising from the causes, like deforestation, inappropriate agricultural practices, overgrazing. The term ‘desertification’ is found widely used to indicate land degradation in arid, semi-arid and dry sub-humid areas resulting from various factors, including climatic variations and human activities. The issue of desertification and land degradation is of global importance as we increasingly face the challenge of food insecurity and declining ecosystem services, particularly in South and East Asia, where more than 80 percent of the increase in production will have to come from yield increases in the scarcity of arable land to meet the growing food demand in the coming decades. Land degradation can further aggravate the situation (FAO, 2002) since the region as a whole may already have passed the safe limits for agricultural expansion (Eswaran et al., 2001). Continuation of land degradation is contributing to climate change and thus can have serious impact on livelihood of the people by threatening food security and can be a hindrance to poverty alleviation. Despite the global importance of land degradation, the available data on the extent of land degradation in drylands are limited (FAO, 2005) and this stands true with areas other than drylands. To date, the best known land degradation data are of the Global Assessment of Human Induced Soil Degradation (GLASOD) study at the global scale (1:10 million scale) and Soil Degradation in South and Southeast Asia (ASSOD) for the regional scale (1:5 million scale). These data are compiled from variety of available data of various types and scales, and information gathered through expert opinion. ASSOD data classifies various land degradation types according to five impact categories. Besides being coarse resolution data, both GLASOD and ASSOD intend to reflect the actual situation in the field by expressing soil degradation in qualitative terms as impact on productivity, the spatial extent classified under each impact category vary tremendously between GALSOD and ASSOD (van Lynden and Oldeman, 2007). This suggests that relatively precise data on the spatial extent of degradation is still lacking in the region. Since the earlier studies completed long ago are often based on the expert judgments, there is the need of updating land degradation database using current information. Such information can serve as quantified baseline information on land degradation for the regional level planning and strategies formulation for natural and land resources conservation. In this study, we use geospatial data available from secondary sources to asses the land degradation based on four major indicators in the Greater Mekong Subregion (GMS) in order to produce quantified information on land degradation.
Study Area The study area, the Greater Mekong Subregion, comprises six countries, namely Myanmar, Thailand, Cambodia, Lao PDR, Vietnam, and Yunnan province of People’s Republic of China covering about 2.34 million sq km (Fig. 1a). The area is largely under severe pressure of land degradation as indicated by ASSOD assessment in Figure 1b. The Mekong basin, irrigated by Mekong River, is one of the world’s largest river basins shared by these six countries. Although the initial concept of GMS was to enhance economic cooperation among these countries, the environmental concern has always been emphasized among the GMS countries.
Remote Sensing and GIS Aided Land Degradation …
319
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 1. Location of the study area (a), degradation types and severity (b).
The climate is governed by monsoon-steady winds that blow alternately from the northeast and the southwest, each for about half of the year. The southwest monsoon begins in May and continues until late September. The northeast monsoon is from November to March. April and October are transitional periods with unstable wind speed and direction. The regional rainfall varies significantly from the driest region in the basin (Northeast Thailand) where annual rainfall is mostly between 1,000-1,600 mm, to the wettest region (Northern and Eastern Highlands) with 2,000-3,000 mm annually. Cyclonic disturbances may cause widespread rainfall of long duration during July to September, which can cause serious flooding. Maximum temperature range from 30°C up to 38°C while minimum temperature range from 15°C in lower plain to subzero in the higher altitude during winter. More than 200 million people live in the GMS. The great majority of these people live in rural areas where they lead subsistence or semi-subsistence agricultural lifestyles. Rice along with many other agricultural crops is grown in different parts of the subregion. Inappropriate agricultural practices are highly prevalent in the study area thus contributing to several types of land degradation. The Subregion is among those basins in the world which has suffered the highest amount of forest loss (WRI, 2002). The forest cover has declined from over 70% to below 30% due to excessive commercial logging, shifting cultivation, encroachment of reserves for settlements, farming and infrastructure development during the last fifty years. Heavy fuel wood use and unclear land ownership have also contributed to this development (UNEP, 1990).
320
Rajendra P Shrestha and Kingshuk Roy
Materials and Methods Base Data Data were collected from variety of sources in digital and analog formats as well as tabular data. The basic data used were: (1)
(2)
(3) (4)
(5)
(6)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
(7)
Moderate Resolution Imaging Spectroradiometer (MODIS) composite Normalized Difference Vegetation Index (NDVI) data, available as 15-day composite, were downloaded from the University of Tokyo website (http://webmodis.iis.utokyo.ac.jp/Asia/) for the year 2002. The data has 15 arc-second resolution. One scene of National Oceanic Atmospheric Administration-Advanced Very High Resolution Radiometer (NOAA-AVHRR) data covering study area was obtained from Asian Center for Research on Remote Sensing (ACRoRS), Bangkok. This 10day composite image of 21-30 September 2000 has 1.1 km resolution. The image was digitally classified to prepare land cover map of the area. FAO-UNESCO soil map (1:5M) was downloaded from the Food and Agriculture Organization (FAO) website (//www.fao.org/ag/agl/agll/wrb/soilres.stm#down). Rainfall data were downloaded from the website of the University of Tokyo (http://hydro.iis.u-tokyo.ac.jp/GAME-T/GAIN-T/updates.html). Downloaded data were monthly rainfall of 266 weather stations in the subregion for the period of 1999-2002, from which average annual rainfall was computed and interpolated to prepare a GIS layer. Shuttle Radar Topography Mission (SRTM) GTOPO30 elevation data (1-km resolution) was downloaded from the Global Land Cover Facility website (http://glcf.umiacs.umd.edu/data/srtm/) of the University of Maryland. The data was used to prepare the slope map. Mean monthly evapo-transpiration (ET) data was downloaded from http://climate.geog.udel.edu/~climate/html_pages/README.wb_ts2.html, produced and documented by Willmott and Matsuura (2001). The data gives mm of evapotranspiration at 0.5 degree spatial resolution. ASSOD data was also obtained from the website of International Soil Research Information Center. It was used to gain an initial understanding of the degradation types and their impacts but the data was not used for the analysis in this study.
The importance of indicators in monitoring desertification and degradation has been recognized for more than two decades. Up until now, several researchers have proposed such indicators of various category of both resource- and process-based in nature. Rubio and Bochet (1998) have compiled a list of indicators proposed by several researchers. We used four major indicators of land degradation, namely vegetation cover, runoff, water use efficiency and soil loss, for assessing the land degradation status in the GMS. The simplified research procedure is presented in Figure 2. ERDAS IMAGINE Version 3.4TM and ENVI 4.0 software were used for the image processing of remotely sensed data and all overlay analyses ware carried out in Geographic Information System (GIS) environment using ArcMapTM 9.0 and ArcviewTM 3.3 software.
Remote Sensing and GIS Aided Land Degradation …
MODIS NDVI data
Vegetation cover
Evapotranspiration map Rainfall point data
NPP
321
WUE Status of land degradation
Rainfall map Runoff
Soil Map NOAA-AVHRR
Land use map
SRTM data
Slope map
Soil loss
Figure 2. Research procedure
Estimation of Vegetation Cover Vegetation cover is an important indicator of land degradation (Rubio and Bochet, 1998) because of their important role in protecting the land surface from direct action of casual factors of land degradation. Besides protecting the surface, vegetation is also important for maintaining the physical condition of soils. This indicator is found expressed in various forms, one being percentage cover of vegetation. Most often, it is found that remote sensing derived NDVI data are used to compute vegetation cover. In this study, we computed annual integrated NDVI from MODIS NDVI fortnightly composite data and subsequently the vegetation cover (Vc) using the equation suggested by Zhang (1999).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Vc = 1.333 + 131.877 × NDVI
(1)
Estimation of Water Use Efficiency Water use efficiency (WUE) and rain use efficiency (RUE) are important indicators of land degradation (Le Houerou, 1984, 1989). RUE, which quantifies NPP (in units of biomass per unit time per unit area) normalized to the rainfall for that time period (Prince et al., 1998), is used to assess land condition or productivity. RUE is believed to be not suffered from short-term fluctuations (Symeonakis and Drake, 2004) compared to vegetation cover, hence considered better indicator over vegetation cover alone. WUE has similar concept as of RUE. WUE can be applied at the leaf level, whole-plant level, or ecosystem level. At the ecosystem level, it is referred to as the grams of dry weight gained by plants during the growing season per unit land area with respect to water lost through plant and soil. WUE can be computed as the ratio of NPP and evapotranspiration (ET). NPP can be used to quantify the net carbon absorption rate by living plants (NRC, 2006) and thus serves as a measure of dryland condition (WRI, 2002). Terrestrial NPP is one of the most-modeled ecological processes with models that differ markedly in approach and complexity, often yielding comparable global estimates (Schimel et al., 1994). Since direct observations of NPP are not available in many instances, it can be estimated with other
322
Rajendra P Shrestha and Kingshuk Roy
available techniques and models. NPP can be calculated using NDVI also. We used MODIS NDVI to calculate NPP using the following regression model suggested and calibrated by Prince et al. (1998).
NPP (Mg ha-1a-1) = 3.139 ∑ NDVI-3.852
(2)
Given that the majority of the study area is dominated by agricultural ecosystem, we assume WUE to be a better indicator over RUE as the total water input includes the water sources other than rainfall. WUE is based on ET, which greatly vary according to crop types, and thus giving better picture of water use potential.
Estimation Of Runoff Runoff can be defined as the movement of water, usually from precipitation, across the earth's surface towards stream channels, lakes, oceans, or depressions or low points in the earth's surface. Runoff is an important indicator of land desertification (Sharma, 1998) and the notion of increased water run-off with land degradation is well established. The USDA Soil Conservation Service runoff curve number (CN) procedure is the best-known and widely used model of this type to compute runoff (SCS, 1972). Although the method is designed for a single storm event, it can be scaled to find average annual runoff values. The following CN equation was used to compute runoff. 2 (P-0.2S) Q= P+0.8S
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
where, Q = runoff (mm); P = rainfall (mm); S = amount of rainfall (mm) which can soak into the soil during the storm, given by S = (25400/CN) – 254 (when water depths expressed in mm); CN = dimensionless runoff index (25-100). CN can be extracted from published tables and is dependent on land use, crop, management and hydrological soil group. The CN value was estimated based on the NOAAAVHRR derived land use map and FAO-UNESCO soil map and rainfall map discussed earlier.
Estimation of soil loss Soil erosion, one of the major land degradation problems, is a highly complex problem and serious environmental threat to the human society because of its significant effects on soil productivity on-site as well as downstream effects. Most often, it is argued that it results from incorrect land use practices. Hence, soil loss is important indicator of desertification identified as ‘Rate’ indicator in Pressure-State-Response framework by FAO (1995). Since erosion due to water is the most prominent land degradation problem in the South and Southeast Asia, including the GMS, soil loss over the study area using Universal Soil Loss Equation (USLE) given by Wischmeier and Smith (1978) was estimated. The equation can be expressed as:
E = R×K×L×S×C×P
(4)
Remote Sensing and GIS Aided Land Degradation …
323
where, E is the mean annual soil loss (ton/ha/yr); R is the rainfall erosivity factor; K is the soil erodibility factor; L is the slope length factor; S is the slope steepness factor; C is the crop management factor; and P is the erosion control practice factor. USLE is most widely used elsewhere for the assessment and prediction of soil erosion due to water. Satisfactory results from USLE are reported from various researchers in the watershed scale too, such as Mellerowicz et al. (1994). However, it is desirable to modify the factors of USLE as necessary while using at different localities. In our case, derivation of individual USLE factor was carefully done drawing upon the findings of previous studies (Shrestha et al., 1996). Rainfall erosivity (R-factor) was encoded in rainfall layer; K in FAO-UNESCO soil map; S in slope map derived GTOPO-30 DEM data; L estimated from land cover map; and CP in land cover map of 2000 interpreted from NOAA-AVHRR data. Finally, all the GIS layers containing USLE factors were overlaid to compute average annual soil loss.
Assessment of land degradation Each individual indicator described above are indeed indicators of degradation on their own capacity even including some other variables, e.g. NDVI, evapotranspiration, are used as the indicators in several various studies. To assess the status of land degradation, above four indicators, viz. vegetation cover, runoff, WUE, soil loss, were combined to come up with a composite land degradation status. First of all, each indicator, based on their computed values, were grouped in to four categories from best to worse based on the range of the values for each indicator. Each class were given the score from 1 to 5 depending upon their nature of relationship in describing land degradation (Table 1). Finally, all four indicators were combined to compute the overall land degradation status. Land degradation severity was based on the computed scores – higher the computed score lesser the severity and vice versa. All the analyses were done using GIS.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Results and Discussion Distribution of vegetation cover The distribution of vegetation cover varies in different areas in the GMS with observed NDVI values ranging from -0.11 to 0.64 with the mean value of 0.57. As described in methodology section, it is to note here that these NDVI values were the average values aggregated for the whole year obtained using MODIS 15-day composite data. The computed maximum and minimum vegetation cover was 87 percent and 14 percent, respectively. In general, the integrated annual vegetation cover fairly showed high percentage of vegetation cover in relatively substantial area in the region implying the fact that higher the vegetation cover lesser the risk of degradation as the vegetation keeps the soil undamaged. The proportion area under each class of individual indicator is presented in Table 1. The majority areas were found to have more than 60 percent vegetation cover and similar finding has been reported by Ogawa et al. (2005) in their study conducted at the Mekong basin using MODIS data that the vegetation density was higher than 60% in the areas where at least two crops are taken a year although there is fallow period in between the crop seasons. However, it should be noted that the NDVI distribution is based on the annual aggregation
324
Rajendra P Shrestha and Kingshuk Roy
and monthly vegetation cover distribution might be much less in the areas of rainfed agriculture. Table 1. Area distribution under different degradation indicators.
% 80 Runoff % 80
Vegetation Cover Score % area 0.50 1 0.07 2 0.44 3 70.36 4 28.62 5 Score 5 4 3 2 1
% area 83.04 10.12 6.77 0.06 0.01
(Mg/ha/yr/mm)
30 scenes). The starting point of the procedure is a set of differential interferograms that use the same master image. The DEM used for differential interferogram generation can either be obtained from processing a close time interval suitable InSAR data pair or a-priory DEM with accuracy ≥ 20m. The temporal analysis of the differential phases is restricted to point scatterers with a high SNR and a long-time stable backscattering behaviour. Thresholding of correlation coefficient values of the pixels in coherence images associated with the interferograms could be a simple approach for identification of the stable targets or permanent scatterers. If a target pixel exhibits coherence always greater than a suitable value, that would be considered as PS candidate. However, due to large spatial and temporal baselines in many of the InSAR pairs, several coherence images appear useless for the said purpose. Alternatively, identification of the stable targets or permanent scatterers can be performed by analyzing the time series of the amplitude values of each pixel in the coregistered images. Pixels exhibiting a very “stable” sequence of amplitude values in spite of high geometrical and temporal baselines may be considered as PS candidates. The latter method always gives better feasibility for selection of the PS candidate. It also provides better results in terms of resolution. It is desirable to find as many PSs as possible because a subsidence pattern and the atmosphere need to be sampled as spatially dense as possible. On the other hand, it is necessary to avoid unreliable points causing incorrect estimation. Due to the limitations of the methods for PS candidate selection, many PSs could be neglected. Therefore, PS identification can be carried out by a time series analysis of the phase values. In this procedure, the atmospheric phase screen (APS) is estimated for all the PS candidates and
Land Subsidence Phenomena: Recent Developments…
395
resampled on the uniform image grid. Subsequently, the unwanted phase contribution due to atmospheric effect is removed from the data. Finally, DEM error and target velocity due to land deformation are computed on a pixel-by-pixel basis. Pixels with appreciable residual phase values potentially represent additional PSs. Now, we have the phase information of the PSs in stead of all the pixels in the image. It reduces the size of the data drastically from 10s of GB to only a few hundred MB. Determination of the differential phase history of the PSs gives the deformation pattern of the area. The measured differential phase of each PS has contributions from different sources as follows:
φ D − InSAR = φ herr + φ displ + φ atm + φorbit + φ noise where,
φ herr , φ displ , φ atm, φorbit and φ noise
represent
phase
contributions
from
uncompensated topography, displacement, atmosphere, orbit and noise. The equation can be solved by utilizing the contributions of spatial baseline, temporal baseline, range and azimuth location of the scatterers. A relative estimation between points scatterers located closely to each other reduces the influence of the atmosphere and orbital errors. Utilizing a periodogram, a constant relative subsidence rate can be estimated. These relative estimates are transformed into global subsidence map by a 2-D integration procedure based on a least square adjustment.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Case Studies Case Study 1: Land Subsidence in Paris City, France (Chatterjee, 2001) In Paris city, ground deformation is characteristically a slow-rate deformation and of small spatial extent. Heterogeneous atmospheric effects over the area are a major problem to differentiate ground deformation from atmospheric artifacts. Geologically, the area is composed of chalk, limestone, marl, gypsite, clay and loam, scree stone, loose sand, old and recent alluvium. Except limestone layers, all other rock formations are loose and soft, and easily compressible. Geomorphologically, Paris city and its surroundings show 4-tier planation surfaces described by four resistant limestone layers in the vertical profile. In the history of development and expansion of Paris city since its inception, the excavation of quarries for extraction of building materials, viz., limestone (as building stone), gypsite (as plastering material), shale and clay (for making bricks and tiles), and loose sand and alluvial stones (as mortar) is closely associated. In the first reported document of 1292, there were 18 quarries inside Paris city. The number and size of the quarries were increased till 1810 and stopped afterwards. Many of the quarries are now abandoned and subjected to intense mechanical fracturing in addition to regional tectonic fracture system (Moiriat et al., 2000). The quarries are now at various stages of damage having high potential for causing land subsidence. For extension of underground transport network namely East Ouest Liason Express (EOLE) or RER line-E, pumping of ground water was performed continually along this segment during the 1990s (Fig. 1a). A sudden drop in the piezometric level was observed
R.S. Chatterjee
396
during 1994-’95. Subsequently, the piezometric level was raised, close to the previous level by artificial recharge of the aquifer when installation of the metro line and underground construction was over (Fig. 1b). Abrupt piezometric fluctuation due to dewatering of the aquifer and artificial recharge leads to subsidence and absidence of the land surface. In this study, it is attempted to decipher the land deformation scenario of Paris city during the 1990s by D-InSAR technique.
N
(b) Piezometric level at point ‘9’ from November, 1978 to November, 1999.
0
3 km Scale
Trace of EOLE Point ‘9’
Point ‘62’
(a) Part of Paris city showing the trace of EOLE and locations for piezometric evolution study
(b) Piezometric level at point ‘62’ from November, 1978 to November, 1999.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 1. Piezometric fluctuation associated with Metro (underground railway) installation in Paris City during the 1990s.
In this study, twenty three InSAR data pairs, acquired during 1993-1999, were processed. However, some of the interferograms are not sufficiently coherent due to geometric and temporal decorrelation of the data pairs. The rate of ground deformation is substantially low and atmospheric heterogeneity is often very strong over the area, which pose great difficulty to obtain clear deformation fringes over noise in long temporal baseline interferogram and to separate deformation fringes from atmospheric artifacts. In this work, the procedure namely, summing of temporally overlapping independent interferograms in complex domain was employed to separate deformation fringes from atmospheric artifacts. The phases of ground deformation have been studied by judicious consideration of the interferograms according to their acquisition time. Three sets of interferograms were categorized to describe subsidence and absidence phases, and net deformation during the entire period of observation: (i) subsidence phase during 1993-1995, (ii) absidence phase during 1996-1998, and (iii) net deformation during 1993-1999. It is observed that ground deformation in this area during the observation period has occurred principally due to ground water pumping and recharge related to EOLE (transport
Land Subsidence Phenomena: Recent Developments…
397
network) project execution. Other causes like fracturing and damages associated with age-old subsurface quarries appear insignificant. During the first phase i.e., during 1993-1995, ground deformation was mainly subsidence in nature. The maximum cumulative subsidence during this phase was estimated to be 25mm in patches along the EOLE trace (Fig. 2a). During the next phase, i.e., during 1996-1998, absidence or uplift of the already subsided zones along the EOLE trace was the major deformation. The maximum cumulative absidence was found to be 20mm in patches (Fig. 2b). On the other hand, the interferograms spanning over the entire observation period (1993-1999) show residual deformation fringes with maximum subsidence of 7.5mm (Fig. 2c). The residual deformation basically represents permanent ground deformation which resulted from the compaction of the aquitards during the period of huge ground water extraction for execution of the EOLE project.
Sum of the interferograms dated 28.07.93 - 22.07.95 & 03.02.93 -
Interferogram dated: 13.01.96 -
Sum of theinterferogram: 10.03.93 - 13.03.99 & 19.05.93 -
30 50
Subsidence 10
Absidence
(a) Subsidence phase during 1993 – 1995.
Absidence
(b) Absidence phase during 1996 – 1998.
Subsidence
(c) Net deformation during 1993 – 1999.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 2. Evolution of land subsidence phenomenon in Paris City during the 1990s as obtained from satellite-based D-InSAR technique.
Case Study 2: Land Subsidence in Kolkata City, India (Chatterjee et al., 2006) The phenomenon of land subsidence resulting from heavy withdrawal of ground water for municipal supply, industrial or irrigational use has been observed in various parts of the world. In Kolkata City, India, land subsidence is believed to have been occurring due to overdrafting of ground water to meet the water-demand of a rapidly growing population and other ancillary activities (Biswas and Saha, 1985; Sikdar et al., 1996). In a confined aquifer condition, as in the case of Kolkata City, the over-extraction of ground water causes lowering of piezometric pressure. Ground water contributes as support to the overlying confining layer material. The resulting reduction in artesian pressure causes tensional forces to develop in overlying confining layer material. As a result, compaction of overlying confining layer material occurs which leads to land subsidence. The area around Kolkata City represents a typical deltaic flat country. Geologically, the area forms a part of Bengal Basin and is underlain by Quaternary sediments of fluvio-deltaic origin consisting of a succession of clay, silt and sand. The area is characterized by the occurrence of a thick surface clay layer with an average thickness of 40m. Below the clay layer, coarse clastic sediments consisting of sand and gravel interspersed with clay and silt occur down to a depth of ~300m from the surface. The principal water bearing horizons occur
398
R.S. Chatterjee
at depths between 60m and 180m in medium to coarse sand and pebble layers (Biswas and Saha, 1985). The overlying clay layers with wood stumps and peat beds are highly compressible which can facilitate land subsidence due to over extraction of ground water. Two types of deformation can occur due to piezometric fall. The aquifer sand layer undergoes elastic compression as long as piezometric fall continues during pre-monsoon time which however reverts back when the piezometric head rises during post-monsoon time. On the other hand, due to rapid piezometric fall as a result of heavy withdrawal of ground water, a hydraulic gradient is created between aquifer and overlying clay layer which leads to the leakage of pore water from the confining clay layer. The draining out of pore water from the clay layer reduces pore pressure and therefore inelastic compression of confining layer takes place.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Interferogram ‘a’ = ERS1-7335 (10.12.92) & ERS2-17064 (26.07.98); (B⊥ = 14m; BÔ = 11m; Δt= 2044days)
Interferogram 'c' = ERS1-10842 (12.08.93) & ERS2-6543 (21.07.96); (B⊥ = -35m; BÔ = -36m; Δt= 1074days)
Interferogram 'b' = ERS1-10842 (12.08.93) & ERS1-24713 (06.04.96) (B⊥ = -131m; BÔ = -79m; Δt= 968 days)
Interferogram 'd' = ERS1-11343 (16.09.93) & ERS1-25214 (11.05.96); B⊥ = -25m; BÔ = 29m; Δt= 968 days
Figure 3. Filtered interferograms ‘a’, 'b', 'c' and 'd' displayed in rainbow colour scheme are showing several fringes (deformation fringes and atmospheric artifacts).
Land Subsidence Phenomena: Recent Developments…
399
0.0 2.5 5.0
L1
L3
L2
Average subsidence rate = ~6.5mm/year (Max.)
1.0 mm
L1 : Machhua Bazar L2 : Calcutta University L3 : Rajabazar Science College
Subsidence contour with figures in mm/year
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 4. IHS colour composition of the sum-interferograms (H) average backscattered amplitude image of one of the interferograms of the sum interferogram (I) and coherence image of one of the interferograms of the sum interferogram (S) to highlight the areas underwent subsidence during the 1990s.
In this work, a space-based D-InSAR technique has been attempted to study the land subsidence scenario of Kolkata City during the 1990s. In this study, 20 InSAR data pairs spanning over 1992 – 1998 were processed. The interferograms were categorized into three groups: (i) post-monsoon (t1) – pre-monsoon (t2), (ii) pre-monsoon (tt) – post-monsoon (t2), and (iii) pre-monsoon (t1) – pre-monsoon (t2) and post-monsoon (t1) – post-monsoon (t2). The data pairs acquired during post-monsoon (t1) – pre-monsoon (t2) periods, when piezometric lowering was the maximum, would potentially represent maximum possible land subsidence as it includes both elastic and inelastic deformations resulting from the compression of the aquifer and overlying confining layer respectively. We have highlighted our study using mainly the post-monsoon (t1) – pre-monsoon (t2) interferograms. The interferograms possess low to moderate coherence and give rise to poor to fairly defined subsidence fringes superposed with atmospheric artifacts. For slowly subsiding areas, as in the present case, temporal decorrelation and atmospheric artifacts appear major difficulties in the analysis of the differential interferograms. We have attempted adaptive filtering of the noisy interferograms (Goldstein and Warner, 1998) to highlight the fringes from temporal
400
R.S. Chatterjee
decorrelation noise in the interferograms. To identify the deformation fringes from atmospheric artifacts, we have followed the approach of enhancing the deformation fringes and at the same time diluting the atmospheric fringes by summing independent interferograms in complex domain (Fig. 3). The study shows that the area in and around Machhua Bazar, Calcutta Universtiy and Rajabazar Science College had been undergoing land subsidence during the 1990s (Fig. 4). The fluctuation of piezometric level during the observation period has been found to have a positive relation during the observation period. However, for understanding the nature of land subsidence during the study period, it is required to procure and process more number of InSAR data pairs with reasonable coherence and at closer temporal baselines. For modelling of land subsidence phenomenon due to over-drafting of ground water, it appears to be essential to separate the components of land subsidence resulting from elastic compression of the aquifer and inelastic compression of the confining layer.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Case Study 3: Land Subsidence in Jharia Coalfield, India (Chatterjee et al., 2007) Jharia Coalfield, Jharkhand, India, is the exclusive store house of prime coking coal in the country with a mining history dating back to 1890 (Mukherjee et al., 1991). In this coalfield, underground mining and subsurface coal fire cause land subsidence in different parts of the coalfield leading to frequent roof collapse. Roof collapse often takes the lives of people and cause loss of property. The most recent incidence of roof collapse was occurred in Kusunda Basti on March 15, 2007 (Fig. 5) which took the lives of 10 people. Depillaring and caving methods of underground mining of existing galleries are mainly responsible for highintensity land subsidence. The nature of land subsidence due to underground mining would be in general non-linear in nature. Caving is mostly carried out in virgin areas which neither causes any casualty nor any loss of structure. Depillaring is generally accompanied by proper sand stowing particularly in the settlement areas which therefore does not cause any appreciable land subsidence or roof collapse unless otherwise deliberately opted for in case of virgin areas. However, the existence of many unknown and unmapped galleries in the coalfield, as a result of private mining prior to nationalization of the coalfield, and water logging of the empty galleries lead to major land subsidence and roof collapse in different parts of the coalfield in an unpredictable fashion.
Figure 5. Recent incidence of roof collapse in Kusunda Basti on March 15, 2007 due to underground mining and subsurface coal fire.
Land Subsidence Phenomena: Recent Developments…
401
Besides, the Jharia Coalfield is known for hosting the maximum number of known coal fires among all the coalfields in India. In this coalfield, majority of the subsurface coal fires occur at shallow depth ( n, (a) solution may not exist (b) a unique solution exists. The solution of x is obtained by minimizing the function ||AX – Y||. Various solutions can be obtained when ||AX – Y|| is minimized with different constraints on x.
Unconstrained Case x1+x2+…+xn ≠ 1, where x is any arbitrary real number. Now our aim is to minimize the difference between AX and Y so that the error term is zero. Since A, X and Y are vectors, so the difference (length) or norm of these vectors can be given by
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
||AX – Y||2 = (AX – Y)T (AX – Y) = ((AX)T – YT) (AX – Y) = ATXTAX – XTATY – YTAX + YTY Considering the 3rd term YTAX, let AX=Z, YTAX= YTZ=ZTY. Therefore, YTAX = (AX) TY= XTATY. Now, equation 29 can be written as = ATXTAX – XTATY – XTATY + YTY = ATXTAX + YTY – 2XTATY
(Equation 29)
T. V. Ramachandra and Uttam Kumar
446
Assume J= ATXTAX + YTY – 2XTATY Differentiating w.r.t X
∂J = 2 AT AX − 2 AT Y = 0 ∂X ATAX = ATY
X unconstrained = ( AT A) −1 AT Y
(Equation 30)
which is termed as the Unconstrained Least Squares (ULS) estimate of the abundance. The negative abundance values can be considered to be zero and the values exceeding one can be considered to be 1.
Constrained Case Imposing the sum-to-one constraint on the abundance values i.e. x1+x2+…+xn = 1 or n
∑x i =1
i
= 1.
The solution to the above problem can be solved by minimizing the function J(X, λ) = AX – Y + λ (1TX – 1) where λ is the Lanrangian multiplier. The purpose of introducing the Lanrangian multiplier is to convert the constrained case to the unconstrained case for which we have the above solution.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Proceeding as earlier, ||AX – Y||2 + λ (1TX – 1) = (AX – Y)T (AX – Y) + λ (1TX – 1) = ((AX)T – YT) (AX – Y) + λ (1TX – 1) = ATXTAX – XTATY – YTAX + YTY + λ (1TX – 1) = ATXTAX + YTY – 2XTATY + λ (1TX – 1) Differentiating J(X, λ) w.r.t. X
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
∂J = 2 AT AX − 2 AT Y − λ1 = 0 ∂X
447
(Equation 31)
Differentiating J(X, λ) w.r.t. λ
∂J = X T 1 − 1 = 0 or, X T 1 = 1 ∂λ Now, from equation 9, 2 A AX = 2 A Y − λ1 T
T
λ
or, A AX = A Y − ( )1 T
T
2
−1
λ
or, X = ( A A) ( A Y − ( )1) T
T
2
or,
X constrained = ( AT A) −1 AT Y −
λ 2
( AT A) −11
(Equation 32)
Multiplying both sides by 1T, −1
or, 1 X = 1 ( A A) A Y − T
therefore,
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
so,
T
λ=−
λ=−
T
T
λ 2
1T ( AT A) −11
2(1T X − 1T ( AT A) −1 AT Y 1T ( AT A) −11
2(1 − 1T ( AT A) −1 AT Y (since 1TX=1) T T −1 1 ( A A) 1
Equation 33 can also be written as
2(1 − 1T X unconstrained ) λ=− 1T ( AT A) −11
Substituting the value of λ from equation (33) in equation (32), we get,
X constrained = ( AT A) −1 AT Y +
2(1 − 1T (( AT A) −1 AT Y )) T −1 ( A A) 1 2(1T ( AT A) −1 )
(Equation 33)
T. V. Ramachandra and Uttam Kumar
448
X constrained
(1 − 1T (( AT A) −1 AT Y )) T −1 = ( A A) A Y + ( A A) 1 1T ( AT A) −1 T
−1
T
(Equation 34)
Equation 34 gives the Constrained Least Squares (CLS) estimate of the abundance.
Catchment Yield and Land Use Flows at the catchment were indirectly estimated by the land use area, runoff coefficient and precipitation. Land use is the use of land by humans, usually with emphasis on the functional role of land such as land under buildings, plantation, pastures, etc. Land use pattern in the catchment has direct implications on hydrological yield. The yield of a catchment area is the net quantity of water available for storage, after all losses, for the purpose of water resource utilization and planning (Ramachandra et al. 1999). Runoff is the balance of rainwater, which flows or runs over the natural ground surface after losses by evaporation, interception and infiltration. The runoff from rainfall was estimated by rational method that is used to obtain the yield of a catchment area by assuming a suitable runoff coefficient. Yield = C×A×P
(35)
where, C: runoff coefficient, A: catchment area and P: rainfall. The value of C varies depending on the soil type, vegetation, geology, etc. from 0.1 to 0.2 (heavy forest), 0.2 to 0.3 (sandy soil), 0.3 to 0.4 (cultivated absorbent soil), 0.4 to 0.6 (cultivated or covered with vegetation), 0.6 to 0.8 (slightly permeable, bare) to 0.8 to 1.0 (rocky and impermeable).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
RESULTS AND DISCUSSION Shannon’s entropy computed for Bangalore city, Peri-urban and outskirts (hence n = 3), together comprising Greater Bangalore for 1973, 1992, 2000 and 2006 are listed in Table 3. The entropy values obtained for 2000 and 2006, (1.0325 and 1.0782) are closer to the upper limit of log n, i.e. 1.0986, showing the higher degree of dispersion of built-up in the city. The urbanisation process increased in 2000 and 2006, indicating higher entropy value as the distribution of built-up during 2006 was more dispersed than in 1973 or 1992. Table 3. Shannon’s entropy for Greater Bangalore YearÆ
1973
1992
2000
2006
Entropy
0.9007
0.9023
1.0325
1.0782
ln(n)
1.0986
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
449
Supervised classification was performed using Bayesian classifier and was verified with field knowledge, visual interpretation and Google Earth image. The supervised classified images of 1973, 1992, 1999, 2000, 2002, 2006 and 2007 with an overall accuracy of 72%, 75%, 71%, 77%, 60%, 73% and 55% were obtained by using the open source programs (i.gensig, i.class and i.maxlik) of Geographic Resources Analysis Support System (http://wgbis.ces.iisc.ernet.in/ grass) as displayed in Figure 4. The class statistics is given in table 4. The implementation of the classifier on Landsat, IRS and MODIS image helped in the digital data exploratory analysis as were also verified from field visits in July, 2007 and Google Earth image. Table 4. Greater Bangalore land cover statistics Class Æ Year È 1973 1992 1999 2000 2002 2006
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
2007
Ha % Ha % Ha % Ha % Ha % Ha % Ha %
Built up
Vegetation
Water Bodies
Others
5448 7.97 18650 27.30 23532 34.44 24163 35.37 26992 39.51 29535 43.23 30876 45.19
46639 68.27 31579 46.22 31421 45.99 31272 45.77 28959 42.39 19696 28.83 17298 25.32
2324 3.40 1790 2.60 1574 2.30 1542 2.26 1218 1.80 1073 1.57 1005 1.47
13903 20.35 16303 23.86 11794 17.26 11346 16.61 11153 16.32 18017 26.37 19143 28.01
Figure 4. Temporal land use changes in Greater Bangalore
450
T. V. Ramachandra and Uttam Kumar
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
From the classified raster data, urban class was extracted and converted to vector representation for computation of precise area in hectares. There has been a 466% increase in built up area from 1973 to 2007 as evident from temporal analysis leading to a sharp decline of 61% area in water bodies in Greater Bangalore mostly attributing to intense urbanisation process. Figure 5 shows Greater Bangalore with 265 water bodies (in 1972). The rapid development of urban sprawl has many potentially detrimental effects including the loss of valuable agricultural and eco-sensitive (e.g. wetlands, forests) lands, enhanced energy consumption and greenhouse gas emissions from increasing private vehicle use. Vegetation has decreased by 32% from 1973 to 1992, by 38% from 1992 to 2002 and by 63% from 2002 to 2007. Disappearance of water bodies or sharp decline in the number of waterbodies in Bangalore is mainly due to intense urbanisation and urban sprawl. Many lakes were unauthorisedly encroached for illegal buildings (54%). Field survey (during July-August 2007) shows that nearly 66% of lakes are sewage fed, 14% surrounded by slums and 72% showed loss of catchment area. Also, lake catchments were used as dumping yards for either municipal solid waste or building debris. The surrounding of these lakes have illegal constructions of buildings and most of the times, slum dwellers occupy the adjoining areas. At many sites, water is used for washing and household activities and even fishing was observed at one of these sites. Multi-storied buildings have come up on some lake beds that have totally intervene the natural catchment flow leading to sharp decline and deteriorating quality of waterbodies.
Figure 5. Greater Bangalore with 265 water bodies.
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
451
Urbanisation has telling influences on the natural resources evident from the sharp decline in number of water bodies and also from depleing groundwater table. Temporal analyses indicate the decline of 34.48% during 1973 to 1992, 56.90% during 1973-2002 and 70.69% of waterbodies during 1973-2007 in the erstwhile Bangalore city limits. Similar analyses done for Greater Bangalore (i.e Bangalore city with surrounding 8 municipalities) indicate the decline of 32.47% during 1973 to 1992, 53.76% during 1973-2002 and 60.83% during 1973-2007 (Table 5). This is correlated with the increase in built up area from the concentrated growth model focusing on Bangalore, adopted by the state machinery, affecting severely open spaces and in particular waterbodies. Some of the lakes have been restored by the city corporation and the concerned authorities in recent times. Figure 1, given earlier shows the rate of increase in built up from 1973 to 2007 and its implication on the decline of vegetation and water bodies. Table 5. Status of water bodies in Bangalore city limits and Greater Bangalore
SOI 1973 1992 2002 2007
Bangalore City Number of Area (in ha) Water bodies 58 406 51 321 38 207 25 135 17 87
Greater Bangalore Number of Area (in ha) Water bodies 207 2342 159 2003 147 1582 107 1083 93 918
The land cover features that have been classified using the ML estimates and Bayesian theory shows that builtup and vegetation are negatively correlated given by: y = -8E-06x2 – 0.7362x + 50472, R2 = 0.9286
(36)
where, y is the independent variable (builtup) and x is the dependent variable (vegetation). Builtup and water bodies are also negatively correlated as given by:
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
y= - 8E-07x2 – 0.0148x + 2111.9, R2 = 0.9953
(37)
where, y is builtup (independent) and x is water bodies (dependent). Table 5 gives the details of the number of wetlands from 1973 to 2007. Pattern classifiers were used to map waterbodies automatically from NIR bands of MODIS and Landsat remote sensing data. MODIS provided data of 2002 to 2007, while for 1973 and 1992, IR Bands of Landsat (79m and 30m spatial resolution) data were used. Principal Components of IR bands of MODIS (250 m) were fused with IRS LISS-3 NIR (23.5 m). To extract waterbodies, statistical unsupervised learning of IR bands for the respective temporal data was performed using Bayesian approach based on prior probability, mean and covariance. Spatial distribution of waterbodies on temporal scale is given in Figure 6. Temporal analysis of waterbodies indicate sharp decline of 58% in Greater Bangalore attributing to intense urbanisation process, evident from 466% increase in builtup area from 1973 to 2007.
452
T. V. Ramachandra and Uttam Kumar
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 6. Spatio-temporal analysis of wetlands of Greater Bangalore. Water bodies are represented in blue and the vector layer of water bodies generated from SOI Toposheet is overlaid in red. The inner boundary (in black) is the Bangalore city limits and the outer boundary represents the spatial extent of Greater Bangalore.
Urbanisation and the consequent loss of lakes has led to decrease in catchment yield, water storage capacity, wetland area, number of migratory birds, flora and fauna diversity and ground water table. Studies in selected lake catchments in Bangalore reveal the decrease in depth of the ground water table from 10-12 m to 100-200 m in 20 years due to the disappearance of wetlands. Reclamation of lakes for various developmental activities has resulted in the loss of interconnectivity in Bangalore district leading to higher instances of floods even during the normal rainfall. Analyses of Bellandur and Ulsoor drainage network (Fig. 7) showed that the network is lost due to conversion of Chelgatta tank into a golf course. Similarly the drainage network between Madivala and Bellandur revealed of encroachment and conversion that has resulted in the loss of connectivity between Yelchenhalli kere and Madivala (Fig. 8).
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
453
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 7. Ulsoor–Bellandur–Varthur (a) drainage network (b) lakes overlaid on 10 m DEM showing their missing interconnectivity
Figure 8. Madivala–Bellandur–Varthur (a) drainage network (b) lakes overlaid on 10 m DEM showing their missing interconnectivity
Increased peak discharge and higher frequency of floods are the consequences of urbanisation. As land is converted from fields to built up, it loses its ability to absorb rainfall. Urbanisation has increased runoff 2 to 6 times over what would occur on natural terrain in some pockets of Bangalore. During periods of urban flooding, streets become swift moving rivers, while low lying residential areas and basements become death traps as they fill with water. Conversion of water bodies to residential layouts has further exaggerated the problem. Flooding in urban areas causes large damage at buildings and other public and private infrastructure (evident during 1997, 2002 and 2007). Besides, street flooding can limit or completely hinder the functioning of traffic systems and has indirect consequences such as
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
454
T. V. Ramachandra and Uttam Kumar
loss of business and opportunity. The expected total damage; direct and indirect monetary damage costs as well as possible social consequences is related to the physical properties of the flood, i.e. the water level above ground level, the extent of flooding in terms of water volume escaping from or not being entering the drainage system, and the duration of flooding. Ulsoor–Belandur catchment: This catchment has 6 lakes – Sankey, Ulsoor, Chalghata, Chinnagara and Varthur and was classified into three major land use types – built up, vegetation and others (comprising open land, waste land etc). The total rainfall yield in this catchment is 240 Mm3, percolated water is 90 Mm3 and water overflow is 150 Mm3. The SRTM DEM data were resampled to 10 m resolution and the volume of each lake was computed assuming the depth to be 1 m and the mean annual rainfall to be 850 mm. The total volume of all the 6 lakes in this catchment is 73 Mm3. Hence there is surplus overland flow of 77 Mm3, which cannot flow to downstream due to disruption of natural drainage (removal of lakes and blockage of storm water drains) resulting in flooding (even during normal rainfall). Madivala–Varthur catchment: Similar analysis was done for Madivala catchment which has 14 lakes – Venkatapura, Yellakunte, Bandepalya, Begur Doddakere, Madivala, Hulimavu, Marenahalli, Govindanaikana kere, Tank north of Doresanipalya, Gittigere and Vaddarpalya. The total rainfall yield is 247 Mm3, percolated water is 97 Mm3 and the remaining 150 Mm3 water flows as overland flow and storage in lakes. The total volumes of all the lakes considering 1 m depth is 110 Mm3 resulting in the excess of 40 Mm3 from the catchment leading to artificial floods. In addition to rainfall, Belandur-Varthur watershed receives untreated municipal sewage to the order of 500MLD. Landsat ETM+ data of 15 m spatial resolution (on fusing with Landsat ETM+ PAN) have been used to estimate impermeable areas that does not distinguish between types of urban land use (industrial, commercial and residential) but consist of a sample from a mixture of residential and commercial areas. Figure 9 shows the relation found between impermeable area to urban population density. Consequence of increase in built up pixels (evident from the increase in paved surface or impermeable area) is the increase of population density in a region. Also, due to increased paved surface and concentrated human activities the magnitude of the difference in observed ambient air temperature between urban pockets (artificial land surface) and the regions covered with vegetation (natural area), which is ascribed as urban heat island effect. The urban heat-island effect results in increased local atmospheric and surface temperatures in urban pockets compared to the surrounding open spaces, etc. Specifically, surface and atmospheric temperatures are increased by anthropogenic heat discharge due to energy consumption, increased land surface coverage by artificial materials having high heat capacities and conductivities, increased vehicular and industrial emissions and the associated decreases in vegetation and water pervious surfaces, which reduce surface temperature through evapotranspiration. An attempt is made here to understand the implications of land cover changes on local climate.
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
455
100
Impermeable Area
90 80 70 60 50 40 30 20 0
200
400
600
800
1000
1200
1400
Urban Density in hab/ha
Figure 9. Impermeable area and urban density for Bangalore city
Computation of LSTfrom Landsat TM/ETM+ and MODIS Thermal Bands
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
LST were computed from Landsat TM and ETM thermal bands. The minimum and maximum temperatures from Landsat TM data of 1992 were 12 and 21 with a mean of 16.5±2.5 while for ETM+ data were 13.49 and 26.32 with a mean of 21.75±2.3. MODIS Land Surface Temperature/Emissivity (LST/E) data with 1 km spatial resolution with a data type of 16-bit unsigned integer were multiplied by a scale factor of 0.02 (http://lpdaac.usgs.gov/modis/dataproducts.asp#mod11). The corresponding temperatures for all data were converted to degree Celsius. Figure 10 shows the LST map and NDVI of Greater Bangalore in 1992, 2000 and 2007. The minimum and maximum temperatures were computed as 20.23, 28.29 and 23.79, 34.29 with a mean of 23.71±1.26, 28.86±1.60 for 2000 and 2007 respectively. Data were calibrated with in-situ measurements.
Computation of NDVI from Landsat TM/ETM+ and MODIS Data NDVI was computed from visible Red (0.0.63 – 0.69 μm) and NIR (0.76 – 0.90 μm) bands of Landsat TM (1992)/ETM+ (2000) and MODIS data (620 – 670 μm (Red) and 841 – 876 μm (NIR)) of 2007, so that the relationship between LST and NDVI can be studied (Fig. 10). NDVI for Landsat TM was 0.04±0.4543, for ETM+ was 0.0252±0.5369 and for MODIS was -0.0917±0.5131. The correlation between NDVI and temperature of 1992 TM data was 0.88, 0.72 for MODIS 2000 and 0.65 for MODIS 2007 data respectively, suggesting that the extent of land
456
T. V. Ramachandra and Uttam Kumar
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
cover with vegetation plays a significant role in the regional LST. Respective NDVI and LST for different land uses is given in table 6 and further analysis was carried out to understand the role of respective land uses in the regional LST’s. Temporal analysis showed a linear growth of 466% in number of urban pixels from 1973 to 2007 and a decline of 61% in the number of water bodies with a 63% decrease in vegetation cover. Similarly during 1992 to 2007, the increase in built pixels was 63% while vegetation and water bodies declined by 45 and 43.8% respectively. The increase in LST during 2002 and 2007 compared to 1992 could be mainly attributed to the increase in built up area and decline of area under vegetation and water bodies in the region
Figure 10. LST and NDVI from Landsat TM (1992), MODIS (2002 and 2007) (Note: pixelisation of MODIS 2002 and 2007 is mainly due to coarse spatial resolution ~ 1 Km)
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
457
Table 6. LST (°C) and NDVI for various land uses Land use
Built up Vegetation Water bodies Open ground
1992 (TM) LST NDVI Mean±SD Mean±SD 19.03± -0.162± 0.096 1.47 15.51± 0.467± 1.05 0.201 12.82± -0.954± 0.62 0.055 17.66± -0.106± 2.46 0.281
2000 (MODIS) LST NDVI Mean±SD Mean±SD 26.57± -0.614± 1.25 0.359 22.21± 0.626± 1.49 0.27 21.27± -0.881± 1.03 0.045 24.73± -0.016± 1.56 0.283
2007 (MODIS) LST NDVI Mean±SD Mean±SD 31.24± -0.607± 2.21 0.261 25.79± 0.348± 0.44 0.42 24.20± -0. 81± 0.27 0.27 28.85± -0.097± 1.54 0.18
It is clear that urban areas that include commercial, industrial and residential land exhibited the highest temperature followed by open ground. Lowest temperatures were observed in water bodies across all years and vegetation. Spatial variation of NDVI is not only subject to the influence of vegetation amount, but also to topography, slope, solar radiation availability, and other factors (Walsh et al., 1997). A closer look at the values of NDVI by land use category (table 6) indicates that the relationship between LST and NDVI may not be linear. Clearly, it is necessary to further examine the existing LST and vegetation abundance relationship using fraction as an indicator, which is discussed next.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Generation of Abundance Maps Through Linear Spectral Unmixng The Landsat ETM+ images (band 1, 2, 3, 4, 5 and 7) were unmixed through Linear unmixing to get the abundance maps of 5 classes (1) dense urban (commercial/industrial/residential), (2) mixed urban (that has some amount of vegetation present in between), (3) vegetation, (4) open ground and (5) water bodies. We considered only dense urban, mixed urban and vegetation abundance for further analysis as shown in Figure 11. Minimum and maximum temperature from ETM+ data were 13.49 and 26.32 with a mean of 21.75±2.3. These abundance images were further analysed to see their contribution to the UHI by separating the pixels that contains 0-20%, 20-40%, 40-60%, 60-80% and 80-100% of urban pixels. Table 7 gives the average LST for various land cover (LC) abundance classes. Table 7. Mean LST for various Land use classes with varying abundances Class Æ Abundance È 0-20%
Mean Temperature ± SD of dense urban 21.99±2.37
Mean Temperature ± SD of mixed urban 21.57±2.36
Mean Temperature ± SD of vegetation 17.91±2.19
20-40%
22.06±2.15
21.58±2.36
17.39±1.37
40-60% 60-80% 80-100%
22.27±2.00 22.33 ±2.22 22.47±1.96
21.67±2.41 22.28±2.02 22.37±2.17
17.22±0.89 17.13±0.85 17.12±0.91
458
T. V. Ramachandra and Uttam Kumar
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 11. Abundance maps and LST obtained from Landsat ETM+ data.
The relationship between LST and NDVI was investigated for each LC type through the Pearson’s correlation coefficient at a pixel level and are listed in table 8. The significance of each correlation coefficient was determined using a one-tail Student’s t-test. It is apparent that values tend to negatively correlate with NDVI for all LC types. NDVI values for built up ranges from -0.05 to -0.6. Temporal increase in temperature with the increase in the number of urban pixels during 1992 to 2007 (63%) is confirmed with the increase in “r” values for the respective years. The NDVI for vegetation ranges from 0.15 to 0.6. Temporal analyses of the vegetation show a decline of 45%, which is reflected in the respective temperature increase and the respective ‘r’ values confirm the trend.
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
459
Table 8. Correlation coefficients between LST and NDVI by Land use type (p=0.05) Land use Built up
1992 -0.7188
2000 -0.7745
2007 -0.7900
Vegetation Open ground Water bodies
-0.8720 -0.6817 -0.4152
-0.6211 -0.5837 -0.4182
-0.6071 -0.6004 -0.4999
Relationship of Population Density with LST
Temperature (in degree celcius)
The population data used in the analysis were from Census, 2000 coinciding with the acquisition of ETM+ data of 2000 from which LST were derived. Figure 12 provides the temperature profile for 100 wards with different population densities.
25 24 23 22 21 20 19 0
200
400
600
800
1000
1200
1400
Population density (per ha)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 12. Ward wise temperature and population density
Increased urbanization has resulted in higher population densities in certain wards, which incidentally have higher LST due to higher level of anthropogenic activities, which corroborate with the reports for Fukuoka City, Japan (Tanaka et al., 2005).
Effect of Different Land use Classes Derived from Combined Outputs of Unmixed Image and Bayesian Classification on Lst The abundance output of the Landsat ETM+ images were combined with the signatures of the five classes that were collected from the ground and the Baye's classifier was used to obtain new improved land use map. The output of the unmixing algorithm controls the output of the Baye's classifier by providing the prior probability of each class for every pixel apart
T. V. Ramachandra and Uttam Kumar
460
from the training pixels. Given a set of multispectral images the computation is done as follows: •
• •
The abundances of all categories in the data are computed using an unmixing algorithm discussed earlier. For each pixel, abundance of each category is used as a prior probability of the class. Let x denote the multispectral observation vector and k any class. In the Baye's classifier for the multispectral data, the posterior probability of the class given the observation is computed by multiplying the prior probability of the class (obtained from the unmixing algorithm) with the conditional probability P(x|k). The class label assigned to the pixel is
l = arg min k
P (k | i, j ) P ( x | k ) P( x)
(36)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In the conventional approach, the prior probability of all classes is assumed to be same for all pixels through out the image. This unmixing decision based approach, systematically exploits the information from both the sources for achieving more reliable classification. The output of this approach is shown in Figure 13. Table 9 shows the LST, NDVI and correlation coefficients land use wise.
Figure 13. Classified image obtained from combining unmixed images and training data as input to Baye’s classifier with 6 multispectral bands of Landsat ETM+
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
461
Table 9. LST, NDVI and correlation coefficient for different land use classes Land use
LST (Mean±SD)
NDVI (Mean±SD)
Dense built up Mixed built up Vegetation Open Ground Water Bodies
23.09±1.16 22.14±1.06 19.27±1.59 22.40±1.97 19.57±1.72
-0.2904±0.395 -0.138±0.539 0.3969±0.404 -0.0193±0.164 -0.301±0.47
Correlation coefficient between LST and NDVI -0.7771 -0.6834 -0.8500 -0.6319 0.2319
8 transacts were laid across the city in eight directions (north [N], north-east [NE], east [E], south-east [SE], south [S], south-west [SW], west [W] and north-west [NW]) as shown in Figure 14 and LST were analysed to understand the temperature dynamics in different directions. Table 10 gives the LST Mean and SD in various directions.
Figure 14. Transect lines superimposed on Greater Bangalore boundary
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Table 10. LST across eight directions Direction N NE E SE S SW W NW
Mean LST±SD 21.30±2.39 22.15±2.22 21.01±2.47 21.34±2.30 21.71±2.07 22.19±1.92 22.97±1.72 22.07±2.25
462
T. V. Ramachandra and Uttam Kumar
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Temperature profile was further analysed by overlaying the LST map on the Baye’s classified map to visualise the effect of different land uses. The temperature profile given in Figure 15, show that for vegetation patch or water body temperature fell below the mean on the transact (marked with circle) beginning from the center of the city and moving outwards along the transact. Regions in SW, W, N and NE coincides with urban growth poles (due to intense urbanisations consequent to setting up of IT corridors and industrial plots in the region; likely increase in urban pixels would be 48% in N, 51% in NE, 41% S and 38% in SE directions as listed in table 11).
Figure 15. Temperature profile across N, NE, E, SE, S, SW, W and NW (X axis – Movement along the transact from the city centre (see Fig. 14), Y axis - Temperature (°C))
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
Built up Pixels (in 10000)
6 5 4 3 2 1 0 1973
1992
1999
2000
2002
2006
2007
Year N
E
S
NW
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 16. Temporal growth of built up pixels in various directions.
Figure 17. Temporal changes in built up area in Greater Bangalore (eight directions)
463
T. V. Ramachandra and Uttam Kumar
464
The dynamics of built up pixels from 1973 to 2007 in Greater Bangalore is presented in Figure 16. The growth poles are evident in Figure 17, which are false colour composites for 1973, 1992, 1999, 2000, 2002 and 2007 towards N, NE, S and SE indicating the intense urbanisation process due to growth agents like setting up of IT corridors, Peenya industrial units, etc. The growth in northern direction can be attributed to the new International Airport, encouraging other commercial and residential hubs. The southern part of the city is experiencing new residential and commercial layouts and the north-western part of the city outgrowth corresponds to the Peenya industrial belt along the Bangalore-Pune National Highway 4. The forecast of the growth of built up pixels in various directions for 2010, 2015 and 2020 are listed in table 11. Projected increase in urban pixels during the next 15 years would be 48% in N, 51% in NE, 41% S and 38% in SE directions. Table 11. Projections of growth (built up pixels) 2010, 2015 and 2020 Direction N NE E SE S SW W NW
Equation y = 2E-64e0.0786x y = 8E-69e0.0833x y = 7E-47e0.0585x y = 1E-52e0.0648x y = 4E-57e0.0701x y = 8E-22e0.0294x y = 1E-21e0.0294x y = 8E-37e0.0468x
R2 0.98 0.97 0.99 0.99 0.97 0.99 0.97 0.95
SE (ha) 505 383 373 293 244 58.3 196 411
Increase in Builtup Area (ha) 2010 2015 2020 7374 10924 16184 3737 5668 8597 7343 9838 13180 3313 4581 6334 5608 7963 11305 3323 3849 4459 4154 4812 5573 5135 6489 8200
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
CONCLUSION Field verification showed that newly built up areas obtained from digital classification of remotely sensed data, had the maximum number of small-scale industries, IT companies, multistoried building and private houses that came up in the last one decade. However, to account for the micro-level study of the land use with a detailed growth model, the classification process would require further refinement with other techniques like bootstrap, a computer implementation of the nonparametric or parametric ML estimation that provide direct computational way of assessing uncertainty and estimates of standard errors (Hastie et al., 2001) along with high spatial resolution data like IRS-1C PAN (with 5.8 m) or Cartosat1/2 (with 2 m/1 m spatial resolution). Supervised learning seeks to extract information from labeled samples. If the underlying distribution comes from a mixture of component densities described by a set of unknown parameters Ө, then Ө can be estimated by ML methods. Pattern classifiers along with the advances in geo-informatics coupled with the availability of higher spatial, spectral and temporal resolution data help in extracting spatial features of interest like land cover classes such as built up. In this context, an important application of pattern classifiers would be to estimate accurate temporal land cover statistics that are useful in monitoring the status and extent of these features. The analysis showed a rapid growth of urban pockets and consequent decline of water bodies and vegetation in Greater Bangalore.
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
465
Shannon’s entropy computed for Bangalore city for 2000 and 2006, (1.0325 and 1.0782) are closer to the upper limit of log n, i.e. 1.0986, showing the higher degree of dispersion of built-up in the city. Urbanisation and the consequent loss of lakes has led to decrease in catchment yield, water storage capacity, wetland area, number of migratory birds, flora and fauna diversity and ground water table. Temporal analyses of waterbodies in Greater Bangalore indicate the decline of 32.47% during 1973 to 1992, 53.76% during 1973-2002 and 60.83% during 19732007 consequent to a linear growth of 466% of built up/urban area. As land is converted, it loses its ability to absorb rainfall. Urbanisation has increased runoff 2 to 6 times over what would occur on natural terrain in some pockets of Bangalore. The relationship between LST and NDVI investigated through the Pearson’s correlation coefficient at a pixel level and the significance tested through one-tail Student’s t-test, confirms the relationship for all LC types. Also, increased urbanisation has resulted in higher population densities in certain wards, which incidentally have higher LST due to higher level of anthropogenic activities. The growth poles are towards N, NE, S and SE of the city indicating the intense urbanisation process due to growth agents like setting up of IT corridors, Peenya industrial units, etc. The growth in northern direction can be attributed to the new International Airport, encouraging other commercial and residential hubs. The southern part of the city is experiencing new residential and commercial layouts and the north-western part of the city outgrowth corresponds to the Peenya industrial belt along the Bangalore-Pune National Highway 4. The forecast of the growth of built up pixels in various directions during the next 15 years would be 48% in N, 51% in NE, 41% S and 38% in SE directions.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
SUMMARY Urbanisation is the increase in the population of cities in proportion to the region's rural population. Urbanisation in India is very rapid with urban population growing at around 2.3 percent per annum. Urban sprawl refers to the dispersed development along highways or surrounding the city and in rural countryside with implications such as loss of agricultural land, open space and ecologically sensitive habitats. Sprawl is thus a pattern and pace of land use in which the rate of land consumed for urban purposes exceeds the rate of population growth resulting in an inefficient and consumptive use of land and its associated resources. This unprecedented urbanisation trend due to burgeoning population has posed serious challenges to the decision makers in the city planning and management process involving plethora of issues like infrastructure development, traffic congestion, and basic amenities (electricity, water, and sanitation), etc. In this context, to aid the decision makers in following the holistic approaches in the city and urban planning, the pattern, analysis, visualization of urban growth and its impact on natural resources has gained importance. This communication, analyses the urbanisation pattern and trends using temporal remote sensing data based on supervised learning using maximum likelihood estimation of multivariate normal density parameters and Bayesian classification approach. The technique is implemented for Greater Bangalore – one of the fastest growing city in the World, with Landsat data of 1973, 1992 and 2000, IRS LISS-3 data of 1999, 2006 and MODIS data of
466
T. V. Ramachandra and Uttam Kumar
2002 and 2007. The study shows that there has been a growth of 466% in urban areas of Greater Bangalore across 35 years (1973 to 2007). The study unravels the pattern of growth in Greater Bangalore and its implication on local climate and also on the natural resources, necessitating appropriate strategies for the sustainable management.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REFERENCES Artis, D. A., and Carnahan, W. H. (1982). Survey of emissivity variability in thermography of urban areas. Remote Sensing of Environment 12: 13-329. Arumugam, M., Emerson, C. W., Siu-Ngan Lam, N., and Quattrochi, D. A. (2003). Classifying urban land covers using local indices of spatial complexity. Proceedings of the ASPRS 2003 Annual Conference, Anchorage, AK, May 2003, CD-ROM. Balling, R. C., and Brazell, S. W. (1988). High resolution surface temperature patterns in a complex urban terrain. Photogrammetric Engineering and Remote Sensing 54: 12891293. Barnsley, M. J., and. Barr, S. L. (1997). Distinguishing urban land-use categories in fine spatial resolution land-cover data using a graph-based, structural pattern recognition system. Computers Environment and Urban Systems 21(3/4): 209-225. Barr, S., and Barnsley, M. (2000). Reducing structural clutter in land cover classifications of high spatial resolution remotely-sensed images for urban land use mapping. Computers & Geosciences 26: 433-439. Becker, F., and Li, Z. -L. (1990). Temperature-independent spectral indices in TIR bands. Remote Sensing of Environment 32: 17–33. Berry, B. J. L. (1990). Urbanisation. In The Earth as Transformed by Human Action (B. L. Turner II, W. C. Clark, R. W. Kates, J. F. Richards, J. T. Mathews, and W. B. Meyer, eds.), Cambridge University Press, Cambridge, U. K. pp. 103-119. Betts, A., Ball, J., Beljaars, A., Miller, M., and Viterbo, P. (1996). The land surface– atmosphere interaction: A review based on observational and global modeling perspectives. Journal of Geophysical Research 101: 7209-7225. Boegh, E., Soegaard, H., Hanan, N., Kabat, P., and Lesch, L. (1998). A remote sensing study of the NDVI–Ts relationship and the transpiration from sparse vegetation in the Sahel based on high resolution satellite data. Remote Sensing of Environment 69: 224-240. Campana, N. A. and Tucci, C. E. M. (2001). Predicting floods from urban development scenarios: case study of the Diluvio Basin, Porto Alegre, Brazil. Urban Water 3: 113124. Carnahan, W. H., and Larson, R. C. (1990). An analysis of an urban heat sink. Remote Sensing of Environment 33: 65-71. Carper, W.J., Lilesand, T.W., and Kieffer, R.W. (1990). The use of intensity–hue–saturation transformation for merging SPOT panchromatic and multispectral image data. Photogrammetric Engineering and Remote Sensing 56(4): 459–467. Carson, T. N., Gillies, R. R., and Perry, E. M. (1994). A method to make use of thermal infrared temperature and NDVI measurements to infer surface soil water content and fractional vegetation cover. Remote Sensing Reviews 9: 161-173.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
467
Compana, N. A., and Tucci, C. E. M. (1994). Estimativa de area Impermeavel de macro bacias urbanas. RBE, Caderno de Recursos Hidricos 12(2): 79-94. Dash, P., Gottsche, F. -M., Olesen, F. -S., and Fischer, H. (2002). Land surface temperature and emissivity estimation from passive sensor data: Theory and practice-current trends. International Journal of Remote Sensing 23(13): 2563-2594. Diaz, O., and Tucci, C. E. M. (1987). Regionalizacao de Hidrogramas Unitarios de Bacias Urbans. Caderno de Recursos Hidricos Revista Brasileira de Engenharia 7(2): 19-30. Douglas I. (1994). Human Settlements. In: William B. Meyer and B.L. Turner II (Editors), Changes in Land Use and Land Cover: A Global Perspective. Cambridge University Press, Cambridge, U.K. Duda, R. O., Hart, P. E., and Stork, D, G. (2000). Pattern classification, Wiley-Interscience Publication, New York (ISBN 9814-12-602-0). Elena G. I., and Nancy E. B. (2004). Land use externalities, open space preservation, and urban sprawl. Regional Science and Urban Economics 34: 705-725. Franca, G. B., and Cracknell, A. P. (1994). Retrieval of land and sea surface temperature using NOAA-11 AVHRR data in north-eastern Brazil. International Journal of Remote Sensing 15: 1695-1712. Friedl, M. A. (2002). Forward and inverse modeling of land surface energy balance using surface temperature measurements. Remote Sensing of Environment 79: 344-354. Friedl, M. A., and Davis, F. W. (1994). Sources of variation in radiometric surface temperature over a tallgrass prairie. Remote Sensing of Environment 48: 1-17. Fukushima, K. (1988). A Neural Network for Visual Pattern Recognition, IEEE Computer, 65-74. Gallo, K. P., and Owen, T. W. (1998) Assessment of urban heat island: A multi-sensor perspective for the Dallas-Ft. Worth, USA region. Geocarto International 13: 35-41. Gallo, K. P., McNab, A. L., Karl, T. R., Brown, J. F., Hood, J. J., and Tarpley, J. D. (1993) The use of NOAA AVHRR data for assessment of the urban heat island effect. Journal of Applied Meteorology 32: 899-908. Galster, G, Hanson, R, Ratcliffe, M R, Wolman, H, Coleman, S and Freihage, J. (2001). Wrestling sprawl to the ground: defining and measuring an elusive concept. Housing Policy Debate 12(4). Gillespie, A. R. (1985). Lithologic mapping of silicate rocks using TIMS. Proceedings TIMS Data User’s Workshop. JPL Publ., 86–38: 29-44, Pasadena, CA: Jet Propulsion Laboratory. Gillespie, A. R., Rokugawa, S., Matsunaga, T., Cothern, J. S., Hook, S. J., and Kahle, A. B. (1998). A temperature and emissivity separation algorithm for advanced spaceborne thermal emission and reflection radiometer (ASTER) images. IEEE Transactions on Geoscience and Remote Sensing 36: 1113-1126. Gillies, R. R., and Carlson, T. N. (1995) .Thermal remote sensing of surface soil water content with partial vegetation cover for incorporation into climate models. Journal of Applied Meteorology 34: 745-756. Gillies, R. R., Carlson, T. N., Cui, J., Kustas, W. P., and Humes, K. S. (1997). A verification of the ‘triangle’ method for obtaining surface soil water content and energy fluxes from remote measurements of the Normalized Difference Vegetation index (NDVI) and surface radiant temperature. International Journal of Remote Sensing 18: 3145-3166.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
468
T. V. Ramachandra and Uttam Kumar
Gluch, R. (2002). Urban growth detection using texture analysis on merged Landsat TM and SPOT-P Data. Photogrammetric Engineering and Remote Sensing 68: 1283-1288. Goetz, S. J. (1997). Multisensor analysis of NDVI, surface temperature and biophysical variables at a mixed grassland site. International Journal of Remote Sensing 18: 71-94. Gong, P., and Howarth, P. J. (1990). The use of structural information for improving landcover classification accuracies at the rural–urban fringe. Photogrammetric Engineering and Remote Sensing 56: 67-73. Gori, M., and Scarselli, F. (1998). Are Multilayer Perceptrons Adequate for Pattern Recognition and Verification?, IEEE Transactions on Pattern Analysis and Machine Intelligence 20(11): 1121-1132. Goward, S. N., Xue, Y., and Czajkowski, K. P. (2002). Evaluating land surface moisture conditions from the remotely sensed temperature/vegetation index measurements: An exploration with the simplified simple biosphere model. Remote Sensing of Environment 79: 225-242. Guindon, B., Zhang, Y., and Dillabaugh, C. (2004). Landsat urban mapping based on a combined spectral–spatial methodology. Remote Sensing of Environment 92: 218-232. Hastie, T., Tinshirani, R., and Friedman, J. (2001). The Elements of Statistical Learning – Data Mining, Inference, and Prediction, Springer, Canada. Homer, C., Huang, C., Yang, L., and Wylie, B. (2002). Development of a Circa 2000 land cover database for the United States. Proceedings of the 2002 ASPRS Annual Convention, April 2002, Washington, DC, CD-ROM. Kato, S., and Yamaguchi, Y. (2005). Analysis of urban heat-island effect using ASTER and ETM+ Data: Separation of anthropogenic heat discharge and natural heat radiation from sensible heat flux. Remote Sensing of Environment 99: 44-54. Kealy, P. S., and Gabell, A. R. (1990). Estimation of emissivity and temperature using alpha coefficients. Proceedings of Second TIMS Workshop. JPL Publ., 90– 95: 11-15. Pasadena, CA: Jet Propulsion Laboratory. Kidder, S. Q., and Wu, H. T. (1987). A multispectral study of the St. Louis area under snowcovered conditions using NOAA-7 AVHRR data. Remote Sensing of Environment 22: 159-172. Konig, A., Sægrov, S., and Schilling, W. (2002). Damage assessment for urban flooding, Ninth International Conference on Urban Drainage, Portland, Oregon, USA. Kulkarni, V. and Ramachandra T.V. (2006). Environmental Management, Commonwealth Of Learning, Canada and Indian Institute of Science, Bangalore. Kulkarni, V. and Ramachandra T.V. (2006). Environmental Management, Commonwealth Of Learning, Canada and Indian Institute of Science, Bangalore, Printed by Capital Publishing Company, New Delhi. Kwan, H. K., and Cai, Y. (1994). A Fuzzy Neural Network and its Application to Pattern Recognition. IEEE Transactions on Fuzzy Systems 2(3): 185-193. Lambin, E. F., and Ehrlich, D. (1996). The surface temperature– vegetation index space for land cover and land-cover change analysis. International Journal of Remote Sensing 17: 463-487. Landsat Project Science Office. (2002). Landsat 7 science data user’s handbook. Goddard Space Flight Center, Available from: www.address: http://ltwww.gsfc.nasa.gov/IAS/ handbook/ handbook_toc.html.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
469
Lee, C. H. N., Liu, A., and Chen, W. S. (2006). Pattern Discovery of Fuzzy Time Series for Financial Prediction. IEEE Transactions on Knowledge and Data Engineering 18(5): 613-625. Li, F., Jackson, T. J., Kustas, W, Schmugge, T., J., French, A. N., Cosh, M. L., and Bindlish, R. (2004). Deriving land surface temperature from Landsat 5 and 7 during SMEX02/SMACEX. Remote Sensing of Environment 92: 521-534. Lillesand, T.M. and Kiefer, R. W. (2002). Remote Sensing and Image Interpretation, Fourth Edition, John Wiley and Sons, New York. (ISBN 9971-51-427-3) Masek, J. G., Lindsay, F. E., and Goward, S. N. (2000). Dynamics of urban growth in the Washington DC Metropolitan Area, 1973–1996, from Landsat observations. International Journal of Remote Sensing 21: 3473-3486. Masser, I., Cheng, J. (2003). Urban growth pattern modeling: a case study of Wuhan city, PR China. Landscape and urban planning 62: 199-217. Mesev, V. (2005). Identification and characterisation of urban building patterns using IKONOS imagery and point-based postal data. Computers Environment and Urban system 29: 541-557. Moller-Jensen, K. (1990). Knowledge-based classification of an urban area using texture and context information in Landsat-TM imagery. Photogrammetric Engineering and Remote Sensing 56: 899-904. Nancy, T., Chad, H., and Russell, G. C. (2003). A Comparison of Urban Mapping Methods Using High-Resolution Digital Imagery. Photogrammetric Engineering & Remote Sensing 69(9): 963-972. Nichol, J. E. (1994). A GIS-based approach to microclimate monitoring in Singapore’s highrise housing estates. Photogrammetric Engineering and Remote Sensing 60: 1225-1232. Nikolakopoulos, K. G, Vaiopoulos, D. A, Skianis, G. A. (2003). Use of multitemporal remote sensing thermal data for the creation of temperature profile of Alfios river basin. Geoscience and Remote Sensing Symposium, 21-25 July 2003, IGARSS '03. Proceedings,IEEE International 4: 2389-2391. Oke, T. R. (1982). The energetic basis of the urban heat island. Quarterly Journal of the Royal Meteorological Society 108: 1-24. Owen, T. W., Carlson, T. N., and Gillies, R. R. (1998). An assessment of satellite remotelysensed land cover parameters in quantitatively describing the climatic effect of urbanization. International Journal of Remote Sensing 19: 1663-1681. Pajares, G., and Cruz, J.M.D.L. (2004). A wavelet –based image fusion tutorial. The Journal of the Pattern Recognition Society 37: 1855-1872. Peiser, R. (2001). Decomposing urban sprawl. Town Planning Review 72(3): 275-298. Phinn, S., Stanford, M., Scarth, P., Murray, A. T., and Shyy, P. T. (2002). Monitoring the composition of urban environments based on the Vegetation–Impervious surface–Soil (VIS) model by Subpixel Analysis Techniques. International Journal of Remote Sensing 23: 4131-4153. Porto, R.L., Zahel, F. K., Tucci,C. E. M., and Bidone, F. (1993). Drenagem Urbana. In C. E. M. Tucci (org), Hydrologia: Ciencia e Aplicacao (pp. 805-847). Editora da UFRGS, EDUSP, ABRH. Prata, A. J., Caselles, V., Coll, C., Sobrino, J. A., and Ottle, C. (1995). Thermal remote sensing of land surface temperature from satellites: Current status and future prospects. Remote Sensing Reviews 12: 175-224.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
470
T. V. Ramachandra and Uttam Kumar
Quattrochi, D. A., and Goel, N. S. (1995). Spatial and temporal scaling of thermal remote sensing data. Remote Sensing Reviews 12: 255-286. Ramachandra, T. V., Jagadish, K. S., Sudhira, H. S., Raj, K. S., and Jha, S. K., Urban Sprawl Pattern Analysis Using GIS, CES Technical Report No. 99, Centre for Ecological sciences, Indian Institute of Science, Bangalore. http://wgbis.ces.iisc.ernet.in/energy/urban/ Ramachandra, T. V., Subramanian, D. K. and Joshi, N. V. (1999). Hydro electric resource assessment in Uttara Kannada district, Karnataka state, India. Journal of Cleaner Production 7(3):195-211. Rashed, T., Weeks, J. R., Gadalla, M. S., and Hill, A. G. (2001). Revealing the anatomy of cities through spectral mixture analysis of multispectral satellite imagery: A case study of the Greater Cairo Region, Egypt. Geocarto International 16: 5-15. Rawashdeh, S. A., and Saleh, B. (2006). Satellite Monitoring of Urban Spatial growth in the Amman Area, Jordan, Journal of urban planning and development Pp. 211-216. Ridd, M. K. (1995). Exploring a V–I–S (Vegetation-Impervious surface– Soil) model for urban ecosystem analysis through remote sensing: Comparitive anatomy of cities. International Journal of Remote Sensing 16: 2165-2185. Roth, M., Oke, T. R., and Emery, W. J. (1989). Satellite derived urban heat islands from three coastal cities and the utilisation of such data in urban climatology. International Journal of Remote Sensing 10: 1699-1720. Sandholt, I., Rasmussen, K., and Andersen, J. (2002). A simple interpretation of the surface temperature/vegetation index space for assessment of surface moisture status. Remote Sensing of Environment 79: 213-224. Schmitt, T.G., Schilling,W., Sægrov, S. and. Nieschulz, K.-P. (2002). Flood risk management for urban drainage systems by simulation and optimisation, Ninth International Conference on Urban Drainage, Portland, Oregon, USA. Schmugge, T., Hook, S. J., and Coll, C. (1998). Recovering surface temperature and emissivity from thermal infrared multispectral data. Remote Sensing of Environment 65: 121-131. Shi, P-J., Yuan, Y., Zheng, J., Wang, J-A., Ge, Yi., and Qiu, G-Y. (2007). The effect of land use/cover change on surface runoff in Shenzhen region, China. Catena 69: 31-35. Singh, S., M. (1998). Brightness Temperatures Algorithms of Landsat Thematic Mapper Data. Remote Sensing of Environment 24: 509-512. Snyder, W. C., Wan, Z., Zhang, Y., and Feng, Y. -Z. (1998). Classification based emissivity for land surface temperature measurement from space. International Journal of Remote Sensing 19: 2753-2774. Sobrino, J. A., and Raissouni, N. (2000). Toward remote sensing methods for land cover dynamic monitoring: Application to Morocco. International Journal of Remote Sensing 21: 353-366. Stathopoplou, M., Cartalis, C. and Petrakis, M. (2006). Integrating CORINE land cover data and landsat TM for surface emissivity definitions: an application for the urban area of Athens, Greece, International Journal of Remote Sensing 28(15): 3291-3304. Stathopoulou, M., and Cartalis, C. (2007). Daytime urban heat island from Landsat ETM+ and Corine land cover data: An application to major cities in Greece. Solar Energy 81: 358-368.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
471
Streutker, D. R. (2002). A remote sensing study of the urban heat island of Houston, Texas. International Journal of Remote Sensing 23: 2595-2608. Sudhira H.S., Ramachandra T.V., Bala Subramanya M.H. (2007). City Profile: Bangalore., Cities l24(4): 379-390. Sudhira, H.S., Ramachandra, T.V., and Jagadish, K. S. (2003). Urban sprawl: metrics, dynamics and modelling using GIS, International Journal of Applied Earth Observation and Geoinformation 5(2004): 29-39. Tanaka, Y.; Shibata, S.; Gotoh, K. (2005). Appearance characteristic analysis of Heat Island phenomenon by using satellite remote sensing and GIS. Geoscience and Remote Sensing Symposium, 25-29 July 2005, IGARSS '05, Proceedings, IEEE International, 3: 18551858. Tian, G., Liu, J., Xie, Y., Yang, Z., Zhuang, D., and Niu, Z. (2005). Analysis of spatio temporal dynamic pattern and driving forces of urbanland in China in 1990s using TM images and GIS. Cities 22(6): 400-410. Valor, E., and Caselles, V. (1996). Mapping land surface emissivity from NDVI: Application to European, African, and South American areas. Remote Sensing of Environment 57: 167–184. Walsh, S. J., Moody, A., Allen, T. R., and Brown, D. G. (1997). Scale dependence of NDVI and its relationship to mountainous terrain. In D. A. Quattrochi, and M. F. Goodchild (Eds.), Scale in Remote Sensing and GIS, pp. 27-55, Boca Raton, FL: Lewis Publishers. Wang, J. (1993). LINDA—a system for automated linear feature detection and analysis. Canadian Journal of Remote Sensing 19: 9 – 21. Watson, K. (1992). Spectral ratio method for measuring emissivity. Remote Sensing of Environment 42: 113-116. Weng, Q. (2001). A remote sensing-GIS evaluation of urban expansion and its impact on surface temperature in the Zhujiang Delta, China. International Journal of Remote Sensing 22: 1999-2014. Weng, Q. (2003). Fractal analysis of satellite-detected urban heat island effect. Photogrammetric Engineering and Remote Sensing 69: 555-566. Weng, Q., Lu, D., and Schubring, J. (2004). Estimation of land surface temperature – vegetation abundances relationship for urban heat island studies. Remote Sensing of Environment 89: 467-483. World Urbanization Prospects. (2005). Revision, Population Division, Department of Economic and Social Affaris, UN. Yang, L., Huang, C., Homer, C. G., Wylie, B. K., and Coan, M. J. (2003). An approach for mapping large-area impervious surfaces: Synergistic use of Landsat-7 ETM+ and high spatial resolution imagery. Canadian Journal of Remote Sensing 29: 230– 240. Yeh and Li, 2001. A.G.O. Yeh and X. Li. (2001). Measurement and monitoring of urban sprawl in a rapidly growing region using entropy. Photogrammetry. Engineering and Remote Sensing 67(1): 83. Yu, X. J., and Ng, C. N. (2007). Landscape and urban planning. Spatio and temporal dynamics of urban sprawl along two urban-rural transects: A case study of Guamgzhou, China 79: 96-109. Zha, Y., Gao, J., and Ni, S. (2003). Use of normalized difference built-up index in automatically mapping urban areas from TM imagery. International Journal of Remote Sensing 24: 583-594.
T. V. Ramachandra and Uttam Kumar
472
Zhang, Q., Wang, J., Peng, X., Gong, P., and Shi, P. (2003). Urban built-up land change detection with road density and spectral information from multi-temporal Landsat TM data. International Journal of Remote Sensing 23: 3057-3078.
Question Bank Fill in the Blanks
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
(1) Urban sprawl is ___________________________________________________ . (2) Urbanisation is _____________________________________________________ . (3) Urban population is increasing at ______ % per annum and ___ % population resides in urban India. (4) Remote sensing data is useful in ________ _________ management. (5) Consequences of urbansiation are ________, _______, ________, etc. (6) Spatial resolution of Landsat MSS data (1973) is ______ and MSS data (1999) is ___________. (7) Accuracy assessment of a classified image is done using __________/__________ matrix and generally __________, __________, __________ accuracy are used to define the accuracy of each class. (8) Reflectance for vegetation is maximum in _________ band. (9) __________, __________, __________ are the three different types of resampling techniques. (10) __________ classification technique works on neurons whereas __________ classification technique is based on the principle of genetics. (11) __________ is the procedure that is used to assign coordinate system to each pixel in the image. (12) Mixture of red and green produces __________, green and blue produces __________, blue and red produces __________and mixture of red, green and blue produces __________ colour. (13) __________ is often used to remove data redundancy and also used in image fusion. (14) __________ is a hyperspectral sensor. (15) UTM stands for _____ _____ _____. It is divided into _____ zones. India lies in UTM zone _____.
State True or False (1) The most important constraints of pixel-based image classification are that it results in spectral classes and each pixel is assigned to one class only. (2) In the classification process a spectral class may be represented by several training classes. (3) Thermal remote sensing is not based on the measuring of electromagnetic radiation in the infrared region of the spectrum. (4) Principal Components represents maximum variability of the original datasets.
Geoinformatics for Urbanisation and Urban Sprawl Pattern Analysis
473
(5) (6) (7) (8)
In RGB-HIS algorithm, I is replaced by low resolution panchromatic data. Landsat TM and ETM+ have the same spatial resolution for thermal bands. Digital elevation model gives the elevation and presents a 3-D view of the terrain. Tone, shape, size, pattern, texture, site and association are the keys used in visual interpretation of remote sensing data. (9) After resampling the shape of the image changes. (10) Two different images can be fused without georeferencing.
Short Answer Questions
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
(1) (2) (3) (4)
What are the different types of urban sprawl? How the analysis of urban sprawl pattern help regional planning? How geoinformatics helps in optimal management of natural resources? List out and discuss briefly the impacts of urbanisation in a developing country like India? (5) How does urban study help in minimising the environmental impacts of urbanisation? (6) Do you think urbanisation is important? If answer is yes, then how one can achieve sustainable development with urbanisation. (7) What is the difference between land use and land cover? (8) What is false colour composite (FCC) and true colour composite? How does FCC help in image interpretation? You are shown a picture in which grass looks green and houses are red- what is your conclusion? Now, you are shown a picture in which grass shows purple and houses are black – what is your conclusion now? (9) What information is contained in a histogram of image data? (10) How many possibilities are there to visualize a 4 band image using a computer monitor? (11) What is Principal component analysis? How is it useful? Can it be applied in any braches of science? Explain with an example? (12) What is data fusion? List any 3 techniques of data fusion. (13) Explain the difference between supervised and unsupervised classification? (14) Define endmember. List any 3 techniques of endmember extraction and explain them briefly. (15) State Baye’s Theorem. What is Bayesian classifier? How is it dependent on conditional probability?
Long Answer Questions (1) Differentiate between urbanisation and urban sprawl? (2) What is the principal behind NDVI? How does NDVI help in discriminating vegetation, built up, water bodies? Explain. (3) What is geometric correction and resampling? Why the images should be geometrically corrected. Explain in detail?
474
T. V. Ramachandra and Uttam Kumar
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
(4) What is digital image classification? Explain one classification technique (algorithm) in detail? (5) What is Linear unmixing? How is this different from hard classification technique such as Baye’s classifier?
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 20
POTENTIAL OF THE INFORMATION TECHNOLGOY FOR THE PUBLIC PARTICIPATION IN THE URBAN PLANNING Malgorazata Hanzl* Institute of Architecture and Urban Planning Technical University of Lodz, Lodz, Poland
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ABSTRACT The contemporary planning theory evolves towards planning through communication and debate. The image of the city in the citizens’ minds depends, largely, on real city appearance and on information absorbed from different sources. It is influenced by city images and visions of designed transformation published in different media. There are many kinds of city representation created for different purposes: maps, plans and city models of various kinds and applications. The basis for physical planning is data base concerning the city. 3D models, which are easy to read, enable nonprofessional addressees to understand complex planning issues. There are various methods of creating them, as well as numerous applications. The paper includes the collection of sample city models which represent the most important kinds. Current technology also facilitates public participation in the city data base completion. To sum up, the city models play a vital part in the contemporary planning process. New information technology has many areas of application in urban planning: data management and collaboration in project groups are most important. Net participation, that is, taking part in the process via the Internet, grows on popularity. The communication may be one-directional – then we may talk of informing – or twodirectional – then it becomes real. It is also possible to communicate in working groups over a specific subject, for example using an Internet forum or a chat room. The usual rules of delivering information should be applied if the process is to be successful. There are two main streams of presentation of urban models at WWW sites: 3D models presentation and Participatory Planning GIS. Different methods, joining both of the above, are also present. A non-professional addressee should become attracted, if he/she *
Email: [email protected]
476
Malgorzata Hanzl is supposed to visit a participatory planning site. Achieving this requires both the clarity of presentation forms used and the possibility of interaction. Game technology may be used as a tool in urban planning communication; it may serve as a communication channel and as a method of visual simulation. The paper contains also a collection of examples of using new information technology in urban planning with citizens’ participation. Apart from the Internet, there are also other new computer technologies which may be used in direct debates over planning issues. Great potential lies in the Augmented Reality technology, which is currently being tested.
Keywords: GIS, Information technology, public participation, urban planning,
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
INTRODUCTION Contemporary planning theory evolves towards planning through communication and debate (Healey, 1997). The contemporary urban planning practice aims to solving planning issues with the participation of civil society and a consensus between citizens, planning professionals, officials and all the parts engaged in the planning process. The philosophical basis for the communicative planning theory derives from the philosophy of the communicative action of Jürgen Habermas and Karl-Otto Apel. According to this theory: ‘The validity of a norm is justified only intersubjectively in processes of argumentation between individuals; in dialectic. The validity of a claim to normative rightness depends upon the mutual understanding achieved by individuals in argument’ (http://en.wikipedia.org/wiki/Discourse_ethics). The current philosophy trends are related to the increasing activity of people. The development of the civil society, which is one of the core principles of European Union politics, is another aspect of the rising civic consciousness. The Spanish philosopher José Ortega y Gasset in his famous book The Revolt of the Masses (1982) predicts the rise to power and action of a mass-man in society. He defines ‘the mass-man as an ordinary man, “a certain type of European, mainly by analysing his behaviour as regards the very civilization into which he was born’ (http://en.wikipedia.org/wiki/The_Revolt_of_the_Masses). One of the basic fields of human activity is shaping the environment of one’s life, which remains the essence of participatory design, defined by Sanoff (2007) as ‘an attitude about a force for change in the creation and management of environments for people’. Sanoff (2007) indicates at ‘(…) strength (of the participatory approach – author’s add. - which) lies in being a movement that cuts across traditional professional boundaries and cultures. Its roots lie in the ideals of a participatory democracy, where collective decision-making is highly decentralised throughout all sectors of society, so that all individuals learn participatory skills and can effectively participate in various ways in the making of all decisions that affect them’. Fischer et al. (2005) suggests another factor as being favourable for participatory design approach, which is described as collective intelligence. ‘Collective intelligence (CI) (is – author’s add.) shared insight that comes about through the process of group interaction, particularly where the outcome is more insightful and powerful than the sum of individual perspectives. When people align their individual intelligences in shared undertakings, instead of using their intelligence to undermine each other in pursuit of individual status, they are much more able to generate collective intelligence’ (Atlee, 2003).
Potential of the Information Technology for the Public Participation …
477
The rapid growth of Internet based services and the emergence of Web 2.0 which bases on users’ activity may be perceived as another aspect of the phenomena described above. The Web 2.0 had been first defined by O’Reilly (2005) as a phenomenon which may be described by ‘a set of principles’ with ‘the Web as platform’ principle at the first place. The definition by Kaye (2006) after Stern H. describesWeb 2.0 as ‘the writable web’ contrary to the Web 1.0, which is described as “the read-only web”, which underlines the users’ activity as the core of this phenomena. Now the technology experts speak about the third generation of Web, including such elements like: the Web databases, non-browser applications, AI technologies, the semantics of the Web, the Geospatial Web and the 3D web as well (http://en.wikipedia.org/wiki/Web_3.0). Box 1 The principles of Web 2.0 Use of the Web as platform, which offers users a lot of possibilities to fulfill theirs own task online. The long tail – a lot of small sites adressing the sophisticated needs of many users. Use of the network effect and user contribution - Collective Intelligence. Data as a driving force for the Web application - data is the next intel inside. Development of applications with the users’ contribution - users add value, Web applications designed for collective adoption should use licenses with as few restrictions as possible - some rights reserved. Cooperation replaces control. End of the software release cycle, perpetual beta. Applications become easy to distribute and reuse, thanks to lightweight programming model. Software above the level of a single device – This means mobile phones, TVs and other devices used in computer-like manner. Source: (O’Reilly 2005)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
PROBLEM IDENTIFICATION/CONCEPTUAL BACKGROUND The changes in real urban environment are preceeded by creation of the coherent vision of what is planned. Therefore the city representation in the citizens’ minds plays an essential role in reshaping real space. The image depends, largely, on real city appearance and on information absorbed from different sources. It is also influenced by visions of designed transformation published in media. As Lynch says (1960, p.120): ‘In the development of an image, education will be quite as important as the reshaping of what is seen. Indeed, they together form a circular, or hopefully a spiral, process: visual education impelling the citizen to act upon his visual world, and this action causing him to see even more acutely. A highly developed art of urban design is linked to the creation of a critical and attentive audience. If art and audience grow together, then our cities will be a source of daily enjoyment to millions of their inhabitants.’ The perception process is shown in Figure 1.
478
Malgorzata Hanzl Real world 3D model Multimedia Media
Perception Inner image of the world in one’s mind
Knowledge about environment Decisions concerning changes Consensus
Discussion tools, Opinions gathering, Different techniques of discussion and informing
Vision
Model of vision
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 1. The role of vision and city models in perception process.
The aim to build city models is to understand and to represent the processes which take place in the city and to support discussion (Guhathakurta, 2002). Models may show current events, the ones which took place in the past or will take place in the future. Models which address the future include elements of visions. The development of e-society as an effect of new technologies development is connected with accessibility of data concerning planning issues. The text sites, with interaction ensured via references, are unsufficient in planning situations where most of data use maps and plans. 2D plan remains legally recognized way of presentation of planning regulation. Information addressed to laypersons should however be understandable for people without professional formation, so some additional forms of presentation are indispensable. The modern information technology offers indispenable tools for the creation of the city vision. The basis for physical planning is data base concerning the city. 3D models, which are easy to read, assist non-professional addressees understanding of complex planning issues. There are various methods of creating them, as well as numerous applications. The Internet is used as a communication tool in a group decision making process. The methods were developed in local networks and successfully transferred into global network. The bases for project are real world models and GIS data bases (Jankowski and Nyerges, 2001) Geographic Information Systems – georelational data bases – is the tabular data set related to geometric objects representing real world objects. Systems are used to gather, store, analyze and represent data. The GIS systems require high level of proficiency of users and thus they are not the most suitable form for planning with public participation. The real participation may take place in more interactive systems, where there is not predefined hierarchy of users, such as: chat rooms and internet forums. The other forms of communication are virtual worlds. The avatars of users play their roles in virtual scenes. They may communicate, walk and visit virtual world (Evans and Smith, 2001). There are also trials to enable avatars to move items in
Potential of the Information Technology for the Public Participation …
479
virtual worlds, for example to construct common spaces. Examples of projects of that kind are rare and function more as experiments, not as real practice.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REVIEW OF LITERATURE The PPGIS and the use of IT for the participatory planning is the area of research which remains at the edge of three disciplines: IT, sociology and urban planning. The index of literature contains chosen entries from these three fields. The comprehensive literature study would exist a separate review, thus the author restricts herself to description of the most important works and databases concerning the subject and taken into account when writing the present article. Among the authors who deal with the issues of city representation and city spaces the best known are: Edward Hall, Kevin Lynch, Françoise Choay, Claude Levi-Strauss and Umberto Eco. The new directions in the planning theory are the subject of the ouvre of (among many others): Patsy Healey, Stephen Graham, Manuel Castells. There are also the works of Susan Fainstein, Susan Innes and Subhrajit Guhathakurta, which remain important for this article. The community participation methods are the theme of many hanbooks and theoretical works, the significant in author’s opinion are: Sanoff (2000), Forester (1999), Gindroz (2003) and Wrona (1981). The important research in the area of PPGIS and planning models is managed by the Centre of Advanced Spatial Analysis at the University College London. CASA presents the results of the research in the form of working papers, accessible via the Web at http://www.casa.ucl.ac.uk/publications/workingpapers.asp . Some of the CASA works are also presented in (Brail and Klosterman, 2001), which is an important handbook concerning PPGIS. Development of the computer tools for public participation in urban planning is a subject of interest of a few research centers, PICT - Planning Inclusion of Clients through eTraining – an international research realized within Leonardo Da Vinci EC education programme is an example (Bourdakis, 2004; Stellingwerff and Kuhk, 2004). The annual conferences of: eCAADe (Education and Research in Computer Aided Architectural Design in Europe), CAAD Futures Foundation, ACAADIA (Association for Computer Aided Design in Architecture – North America), CAADRIA (Association for Computer-Aided Architectural Design in Asia) and SIGRADI (La Sociedad Iberoamericana de Gráfica Digital which is held in Latin America) gather the IT expert from the field of architecture and urban planning, the papers of which are compiled and accessible via the bibliographic index CUMINCAD at http://cumincad.scix.net . The precious sources of information are also the Annals of Geomatics published by Polish Association for Spatial Information (www.ptip.org.pl).
METHODOLOGY Different kinds of online participation may be ordered in analogy to the ladder of citizens’ participation by Arnstein (1969). The lowest step of the ladder describes an utterly
Malgorzata Hanzl
480
passive behavior and concerns public right to know, while the full interactivity occurs at the top of the ladder as the participation in making decisions.
Online services Online discussion forums Communication barrier Online surveys Basic WWW site
Level of communication
Online comments on chosen solutions
2 directions
PPGIS
1 direction
Increasing participation
Making decisions online
Figure 2. e-participation Ladder (Kingston, 2002)
In the schema in Figure 2, devised by the scientists from Leeds, the lowest step is passive supporting of information and the highest one are systems supporting decisions working via the Internet (Jankowski and Nyegerges, 2001; Kingston, 2002).
Virtual project studio Community project system Systems supporting decisions online Online surveys Online discussions
2 directions
Communication barrier Online services
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Increasing communication
Increasing participation
Virtual worlds
1 direction
Figure 3. Broadened ladder of e-participation (Hudson-Smith et al., 2002 after Kingston, 2002)
The schema devised by the scientists from CASA, London (Figure 3) includes also the design by local society and virtual worlds. Such actions may engage more participants – actors of virtual scenes – than in systems supporting decisions (Hudson-Smith et al., 2002).
Potential of the Information Technology for the Public Participation …
481
Providers / senders
Form of communication
Level of communication
Addressees
IT experts
Collaborative software
Cooperation
IT experts
Professionals authors
Games
Discussion
Professionals authors
Professionals milieu
3D VRML/ Animation
Voting
Professionals milieu
Officials/ Politicians
Forum/ chat/ communicators
Opinions
Officials/ Politicians
Local organizations
PPGIS
Transactions/ services
Local organizations
Stakeholders
Formularies/ email
Education
Stakeholders
Citizens/ Groups of citizens
Text/illustrations / downloadable data
Information
Citizens/ Groups of citizens
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Fig. 4: Net participation - classification. Schema author after (Hudson-Smith et al., 2002)
A new net participation typology has appeared in the last few years (Hudson Smith et al., 2002). Different kinds of participation may use different communication channels in the network. There are seven types of receivers and supporters of information (Figure 4). They may exchange their roles. The list contains professionals of planning, professionals of computer science and lay persons. There are: IT experts – system designers; urban professionals – people responsible for communication of information in planning; group of professionals – people who are interested in planning and influence urban professionals; administrators, councilors, politicians – groups interested in planning and disposing some knowledge on the subject; politicians – individuals with rights to make decisions; groups of citizens – people sharing common engagement, able to organize and act together and stakeholders – the most often those who are directly touched by the issue. Looking from the point of view of the theory of information a slightly different approach may be assumed with the division for data providers and data users. In a really interactive system users become data providers. In a well functioning system without manipulation the technical functions like preliminary processing of data and data providing should be completed in the background. In the realized systems these functions remain the most important ones because of the experimental character and status of the prototype of systems.
IT TOOLS FOR PUBLIC PARTICIPATION IN URABN PLANNING 3D models presentation via the network 3D models are presented as static pictures, animations and as Virtual Reality models. The methods of project presentation may be ranked according to growing user interaction. 2D elevations and 3D presentations – photos of a project – allow presentation from assumed
482
Malgorzata Hanzl
points of view. Photos of a model used as WWW illustration are the most conventional form of presentation. Animation allows better perception of the model. It is performed according to scenario and allows perception of chosen scenes. The most effective form of presentation of planning solutions is animated 3D graphics showing building development from a passer-by’s point of view and from a bird’s point of view. User chooses what to see and from which direction – thus such presentation remains interactive. The digital model may be discussed using space and time, project variants and LOD (Level of Details) or LOOP (Level of Object Presentation) which depends on the proximity observer/object (Voigt et al., 2003). This type is represented by VRML presentation or presentations using commercial software. XML language (GML - Graphic Modelling Language) offers possibilities to record 2D or 3D graphics.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
VRML VR models enable unconstrained moving within virtual city and through acquaintance of proposed development. There are a few standards of record of this kind of presentation. The most popular is VRML (Virtual Reality Modelling Language) which is being developed for some time. Files may be viewed using simple software accessible as freeware or shareware, or using internet browsers’ plug-in (for instance: freeware browser by Viewpoint, http://www.viewpoint.com/). VRML similarly to HTML sends only scene description as language commands. It fits the requirements of dial up connection with small data transfer. Visualization takes place on the client’s computer according to commands in the transferred file. VRML files may refer to sound data, to external sites, they enable interaction. Popular modeling software allows exporting files to this format. Some of the examples of some of the most popular and oldest modes are (i) Glasgow model (Abacus, Department of Architecture, The University of Strathclyde, Glasgow; http://iris.abacus.strath.ac.uk/glasgow); (ii) Los Angeles model (Urban Simulation Team, Department of architecture and Urban Design, University of California, Los Angeles; www.ust.ucla.edu ); (iii) New York model (Environmental Simulation Center, Ltd., New York; www.simcenter.org/); and (iv) London model (Centre for Advanced Spatial Analysiss , University Central London; http://www.casa.ucl.ac.uk/3dcities/london3d.htm). XML Another method is the use of XML (eXtensible Markup Language). XML is a language for data exchange whose standards are specified by Word Wide Web Consortium (W3C). Data codification in XML consists of two parts: schema describing rules and data. XML allows groups of people for creation of their own tools for data exchange according to users’ needs. Environments like XSLT allow for easier transposing of data between different codifications. JSP technology allows for creation of dynamic or changeable content (Szalapaj 2003). In the last few years object oriented technology like JAVA and XML influence the WWW contents and exchange. Rapid evolution of technology is connected with their applicability within WWW sites. In literature the role of XML in architectural and urban data transfer is emphasized. Preparing of presentation with the use of XML syntax is possible from modeling software, for example 3Dstudio MAX. Interactivity and selection of elements for automated generation of simulation may be ensured with the use of JAVA programming language.
Potential of the Information Technology for the Public Participation …
483
The geographical information is recorded with the use of GML (Geography Markup Language) which is an XML application (Michalak 2003). The bases for predefined elements of language are the definitions originated from OGC Abstract Specification. The actual 3.1.1 version of GML was published 18th July 2006 at the OpenGIS Consortium site. GML responds to most ISO standards. The rules for coding the geographical information are specified in ISO 19118 – Geographic Information – Encoding standard worked out by Technical Committee ISO/TC 211 Geographic Information / Geometrics.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Multimedia Interactive presentation of proposed development is also possible using commercial software. Shockwave of Macromedia Director allows recording 3D files with the possibility of moving within the virtual reality. Qualities: accessibility – plug-in enabling reading Shockwave is joined to Windows XP and preinstalled along with the operating system (after Macromedia Director 8.5 Shockwave Studio Feature Tour, available at www.macromedia.com), texture serving, built in reduction of number of polygons in geometrical models in relation to zoom, collision detection, import of models prepared in a few most popular modeling modules. User interaction is ensured by Lingo programming language, enabling shaping of virtual elements from user’s level like in computer games. It is possible to change future development position or alterations of location of building facades. The data in data base may be read by clicking on an object. Information may be different: text, videos, details of chosen object. Other software which offers similar functionality was Atmosphere by Adobe, which allowed for moving around with the use of virtual personages. It was used for example in the project of regeneration of Park Noerrebro in Copenhagen (Holmgren et al., 2004). Another application is Anark (www.anark.com ) used since 1994 for commercial presentation of products. Like other systems its use needs plug-in installation. The presentation and panoramas may use Viewpoint software as well (http://www.viewpoint.com/). For example in Woodberry Down project by CASA University College London: http://www.casa.ucl.ac.uk/woodberry, a website providing 3D presentation of different variants of development was delivered (Hudson-Smith et al., 2002). Other methods of vision presentation All applications are restricted by data transfer conditions, which remain limited for most users. Thus in practice the most popular are WWW sites relying on html or php technology with static illustrations showing 2D plans and 3D models of development. Another possibility which does not need fast transfer is animation recorded as gif file, which allows for fluent transition between views. Interaction potential along with fast data transfer is offered by technology Flash by Macromedia. Swift 3D plug-in allows for graphic simulation – in techniques visually similar to comics – of 3D objects (after www.macromedia.com). One of the ways of presentation of static model is fitting it into a photograph of a site where it is to be. An alternative for construction of 3D models of urban areas is the use of film as a background for project (Chen et al., 2003). Software which offers such possibilities is for example Adobe AfterEffects. The technical difficulty which has to be overcome is precise reproduction of camera location when recording for further synchronization with location of camera to register movement around planned objects.
Malgorzata Hanzl
484
GEOGRAPHICAL INFORMATION SYSTEMS
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
GIS in the network (PPGIS) Participatory Planning Geographic Information Systems serve data with spatial reference to wide audience via the Internet. Common characteristic feature of all PPGIS projects is the use of GIS technology. Data is presented in network with the use of Internet GIS systems: ArcIMS; Autocad MapGuide, PostGIS which enable showing spatial data without a need to install software at the users’ station or with installation of simple plug-in. An example of this kind of sites is www.mapa.lodz.pl. The use of system needs qualifications both to operate an application and to manage the data contents. The data presented in this way include: property data, demographic data, investment areas location, master plans, and information on cultural and natural heritage. The system may be an interface to access multimedia: VRML panoramas, photographs or video films. The common fault of most PPGIS is an interface which often needs proficiency in planning issues to get real access to data. Allowing for data manipulation is a method to get user’s engagement (Han and Peng, 2003; Geertman, 2001). User looks through data and makes decisions about the way of their presentation using tools built in his/her interface. An example is http://www.casa.ucl.ac.uk/newtowns/. The site allows user to show chosen layers of data representing different parameters of sustainable development, and then to measure and combine them to get general indicator, which then may be used to range different centers according to level of their sustainability (Hudson-Smith et al., 2002). Sites providing data without treatment or partly treated such like neighbourhoods’ statistics allow user to download these data, showing them and treatment in software like Office (www.neighbourhood.statistics.gov.uk/home.asp). An example is a site presenting actual projects for London suburb Wandsworth. Visitor may verify last decisions and display data on a map (http://www.wandsworth.gov.uk/gis/map/mapstart.aspx). There are also sites accessible for registered users providing data in GIS/CAD format (for example: http://census.ac.uk/casweb). Data may be used by specialists with appropriate software and proficiency. The example of sites which enable users’ comments on different kinds of development for city’s areas is Virtual Slaithewaite Participatory Planning System, Slaithewaite, West Yorkshire, Great Britain (Kingston, 2002), http://www.ccg.leeds.ac.uk/slaithwaite/, worked out by Centre for Computational Geography, Leeds. All former comments may be displayed to facilitate opinion making. There are also examples of mixed use of different systems like SUCoD, http://sucod.shef.ac.uk, which offers presentation of urban forms of different historical periods (Han, Peng, 2003). 3D visualization is accessible after choosing plan elements and period of time. System enables user to pose questions and to acquire data, which may be used for further analysis. System makes use of: 3D visualization and GIS. The presentation is prepared with the use of VRML and Java. Cortona VRML Client is required to gain access to model. Examples of PPGIS are also GoogleMap and GoogleEarth applications, which allow for creation and publication of users’ maps, and for completion of maps with the users’ data in the form of text, graphics, photographs and 3D objects modelled with Google Sketchup.
Potential of the Information Technology for the Public Participation …
485
Planning Support Systems as a tool for public participation The Planning Support Systems is general notion describing software which supports urban planning. The software enables displaying data in forms which are easy to understand by a layperson. The examples are: The Community Viz and What if (Brail and Klosterman, 2001). Both applications allow for simulation of future state of a site after introducing parameters describing current state and planning conditions. The basis for analysis is the model of current land use. The parameters required to show future state are: intensity of development, accepted height of buildings, buildings’ placement on the plot and other conditions that the buildings must fulfill. The software allows envisioning land use alternatives and understanding their potential environmental, economic, and social impacts. The example is Paint the Town system by Northeastern Illinois Planning Commission (Dieber, 2003). Until July 2003 project was applied in 77 communities and 271 suburbs of Chicago. During meetings in city hall officials and citizens worked over propositions of use of land and the character of development. A tool allowed for prognosis presentation, creation of scenarios and gathering information on local expectations. Paint the Town is a GIS tool used to support discussion. After introducing desired development the system generates feedback information concerning characteristics of future population, habitats and social structure. Box 2 PPGIS
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Participatory Planning Geographic Information Systems (PPGIS) enable presentation of spatial data concerning planning issues via Internet. The systems enable also data analyzes and users’ interaction. Games as a tool in physical planning Games are a tool with great educational potential. They offer great possibilities to generate 3D graphics in real time using hardware support. They enable object manipulation within a scene. The games are the most developed branch of computer graphics: their interactivity decides about educational possibilities. A user is rewarded for practical use of skills like movement coordination, logics, memorizing, imagination, problem solving (Hanzl and Wrona, 2004). Games are often perceived as related with violence and developing indifference to human death. Such a negative connotation is, in the author’s opinion, unfounded. There are logic games, strategic games and many other kinds. An example of a typical educational game is Immersive Education by Media Stage, where users play historical persons www.mediastage.net. Another branch is restoring cultural heritage: www.virtualheritage.net (O’Coill and Doughty, 2004). Two games: SimCity and Civilization, whose idea is town planning, gained great popularity. SimCity, invented by William Wright, is used in Massachusetts Institute of Technology as one of elements in urban planning education (http://www.geocities.com/edit6100/Task_4/SimCity.html). Similar possibilities are offered by simulations which are constructed for needs of a specific project and basing on modules identical with these used in network games. The games use avatars to communicate with other users regarding simulation at the same time. Furthermore they allow introducing changes into virtual environment.
486
Malgorzata Hanzl
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Game technology as a communication medium An example of cyberspace basing on VRML is Active Worlds (http://www.activeworlds.com). Active Worlds browser enables access to 3D VRML scenes. Every user is represented by a 3D personage – an avatar. Appearance of a character depends on user’s choice. Active Worlds are a combination of an interactive game with a kind of Internet forum with rich interaction possibilities. The biggest virtual world is AlphaWorld – ‘a town-similar world, where every piece of terrain belongs to individually registered user – virtual world inhabitant.’ Moving within AlphaWorld is realized on foot – for short distances and by teleportation – for long ones. To move to another place one must write down coordinates and the required location is gained. Center of virtual world is in (0,0) point of the co-ordinate system – this is a place where user gets every time after logging into system. ‘Buildings’ development takes place around this point. An example of an experiment with similar functionality is ‘Wired Whitehall’ (Batty et al., 1998). The site allows for ‘visiting’ of the centre of London by choosing points on a plan, which refer to photorealistic panoramas of characteristic places in the town. The site uses mechanisms similar to ‘Active Worlds’. Logged users may speak to each other or use simple gestures of their avatars to communicate. There is a possibility of interaction with objects in a scene – in this respect it differs from AlphaWorld, where construction was possible only within own plot. The trials to move virtual objects independently on property are managed – the idea of experiments is to enable common decisions making in matters of real world development. In case of virtual public space the best place for discussion would be a vision of future development. An example is Electronic Neighborhood – virtual agora prepared with Adobe Atmosphere software allowing for communication and site visiting – used in project of regeneration of surroundings of a Noerrebro Park in Copenhagen by a team of School of Architecture at the Royal Danish Academy of Fine Arts (Holmgren et al., 2004). The project was connected with public discussion which took place at public meeting and at internet forum. A photosafari for pupils of local schools was also organized to show environment quality, www.e-kvarter.dk. Simulation games ‘(…) simulation games, using game model, imitate real decision process in conflict situation. The games represent, in less or more formal way, situations accompanying project creation, base on people participation as on an important operational factor, thus allowing for taking into consideration behaviors of design participants in conflict situation.’ (Wrona, 1981) The earliest games simulated conflicts taking place in a town and applied to allocation of resources. In 1969 Richard Duke, considered one of simulation games inventors, designed a game Metropolis for Lansing City Council, Michigan. Computer was used to find out effects taken by decision makers. In the mid-1970 in the later version of the game - Metro-Apex computer simulation took central place. A list of over 200 simulation games concerning architecture and urban planning along with comments can be found in (Wrona, 1981). Changes of planning thought of the beginnings of the 1980’s, going away from scientific planning towards developing of communication processes had great impact on the crisis of phenomenon modeling in town scale. Human participation got on importance. Role Playing Games (RPG) became standard technique for conflict solving.
Potential of the Information Technology for the Public Participation …
487
Development of computers, software and data bases in the next two decades changed both scientific and practical possibilities of computer simulations. At the end of the 1980’s a renaissance of great scale models in spatial planning was observed, despite the fact that in practical situations predictive possibilities of models are questioned. Early simulations and games required access to servers and skills which had been rare. Improved graphics caused simulations and graphics to be not only accessible and absorbing but also intuitive and consequently easy to play. Thus they may become a form of entertainment.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Role Playing Games RPGs are used as sociological technique supporting mediations. The aim is to aid formulate citizens’ opinions during debate. The RPG is a useful tool in consensus building programs for decision makers and planning professionals. Computer may be used both as a tool for simulation and communication. In order to arrange the process in a reliable way there should be a meeting of all participants at the beginning. Further actions may take place with the use of digital media. The process should be accompanied by periodical meetings and possibility of real contacts to avoid anonymity of participants. The idea of the game is, on the basis of an example of RPG, used in planning for Landscape Park in Senegal, that people participating exchange their roles. After a period of being responsible for other people’s tasks, consensus was easier to reach. A GIS system was used to present game course (D'Aquino et al., 2003). Collaborative software New paradigm of social participation in planning assumes collaboration of all interested parties (Innes and Booher, 2000; Sanoff, 2000). Both citizens and planners become providers and recipients of information. Such collaboration takes place in design groups and in internet systems where users are actively engaged in design process. A rapid development of new ways of use of network has been observed in last few years. A term groupware – software for group work had been introduced for ‘computer-based systems that support groups of people engaged in a common task (or goal) and that provide an interface to a shared environment’ (Ellis et al., 1991). It is also called collaborative software, which means software for computer supported collaboration. This group includes tools enabling mutual contacts (for example email or internet communicators: AOL or MSN), process management applications - calendaring, and software which enables creation and managing of information on a website by several users at the same time, used to group creating of knowledge bases. There are wiki (WikiWikiWeb, Wikipedia and Everything2) and Slashdot in this group. Great popularity is also gained by interactive forms of group communication: chat rooms and http://www.wie.org/collective/resources.asp; http://www.collectivewisdominitiative.org/;http://www.co-intelligence.org/; http://www.the dialoguegrouponline.com/;http://www.theworldcafe.com/; http://www.solonline.org/; http:// www.openspaceworld.com/ etc. Deliberation online describes interdisciplinary discussion via the Internet. It includes: consultations, voting, debate, and online facilitation – software which enables creation of online communities, dialogue on citizens’ issues via internet forums and chat rooms and also group decision making with the use of software enabling collaboration within working groups. All above-mentioned forms use electronic media in a way which improves mutual understanding of users. According to research on collective intelligence – which means
488
Malgorzata Hanzl
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
intelligence of a group or of a community - group work is not a sum of effects of work of single participants but provides new values, which appears as an effect of collective work. As it was said before communicating of people is accompanied by creation of new values (Kamiński, 2002). Some of the WWW sites concerning group intelligence and the use of groupware:http://www.wie.org/collective/resources.asp; http://www.collectivewisdom initiative.org/;http://www.co-intelligence.org/; http://www.thedialoguegrouponline.com/; http://www.theworldcafe.com/; http://www.solonline.org/; http://www.openspaceworld.com/ etc. A condition of efficiency of these forms of communication is continuous activity of responders and thus reliability of presented information. According to he Metcalfe’s law, which states that the more people who use something the more valuable it becomes, the potential of WWW site is related to its popularity (http://en.wikipedia.org/wiki/ Robert_Metcalfe). Techniques for public participation in planning: charette, synectic session, Brainstorming and Buzz Session, take-part workshop, scenario method, simulation games or RPG (Wrona, 1981) may be successfully used within net participation, with the use of groupware. Augmented Reality and advanced Virtual Reality systems Augmented Reality (AR) – reality combined with some virtual elements - used for group working, public participation in planning, project consultations. Lonsing (2004) proposes the definition: ‘(…) an augmented reality system generates a composite view in real time. The composition is a combination of a real scene viewed by a user and a virtual scene generated by a computer, where the real scene is submerged with additional information in order to enhance the perception of the user’. In an aim to get such ‘composition’ a position of an eye must be followed and virtual elements – 2D or 3D - generated in a way as if they were seen from a fixed point. An example of Augmented Reality system is ARTHUR - an Augmented Reality (AR) enhanced round table to support complex design and planning decisions for architects (Fatah et al., 2004). The system used wireless devices for vision creation (CV – computer vision). Input data are taken from glasses shaped cameras worn by users during the experiment, and from static camera to catch hands movements. Virtual objects manipulated by users are displayed as stereoscopic visualization. Use of CAD system (Microstation) allows shaping objects according to needs. Apart from Augmented Reality systems in literature there is also a conception of tangible devices – which function on a basis of touch. An example of such system was constructed in Hong Kong (Seichter, Kvan, 2004). Moving virtual objects is connected with moving real objects – moving boxes on a table causes moving objects in virtual space. Another example of tools of this kind is Illuminating Clay – user changes a model made of clay, which influences changes of virtual model (Ishii et al., 2004). MouseHaus Table is an interface for pedestrian movement simulation in a city (Huang et al., 2003). Mobile devices include a video camera, a coloured sheet of paper, scissors and a table with projection. Introducing buildings is achieved by use of models cut out of coloured paper. The advanced tool to display Virtual Reality is CAVE. It is a rectangular room, which walls, floor and ceilings are screens to display video projection (http://www.evl. uic.edu/paper/CAVE). Sensors verify observer’s position and match display to it. Interaction is possible via joysticks, touch panels or specially designed devices (Wœssner et al., 2004). There may be more than one person inside CAVE at the same time, which allows for
Potential of the Information Technology for the Public Participation …
489
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
discussion and decision making (Voigt et al., 2004). Simulated environment may be displayed at any scale. CAVE enables presentation of variants of planned development. Researches preceded in laboratory of Faculty of Architecture of Swiss Federal Institute of Technology in Zurich (Lang and Hovestadt, 2004) concerned experimental methods of interaction with model using 3D video record of people taking part in meetings. The holograms of participating persons are displayed in virtual space. User moves which in turn make his hologram move in virtual reality. 3D video technology enables communication between users in virtual space. User may influence virtual environment by pushing appropriate place on a wall which causes moving of an object in 3D space. There are trials to overlap design simulation with real surroundings (Kieferle and Wœssner, 2003). It is done using special markers, which allows for precise location of points in real and virtual space and then matching the two images. User provided with special glasses sees projected development looking at the real place. Superposition of real and virtual world may be seen also at the screen. Switching between variants is enabled via keyboard. Another way is moving sensors on real, reduced scale models of buildings (Kieferle and Wœssner, 2003). Manipulating of models causes changes in virtual world – they may be observed also in CAVE in 1:1 scale. Such device makes working in group much easier. Tabak and de Vries (2003) describe an intuitive device used for shaping of buildings in urban scale which acts in a similar way. Buildings are represented by rectangular boxes. User manipulates objects and system introduces information about proposed position of buildings in interactive way in relation to each other and to surroundings. A project built in this way is displayed simultaneously on screens at three sides of the device. Beside visual simulation of the project including elements which characterize perception: lines of sight of observer, positions of shadows, information on accessibility, silhouette of buildings, the system provides also numeric values: dimensions, area of objects, cubature, estimated costs, objects’ functions and also statistics of above mentioned parameters. Box 3 AR Augmented Reality (AR) combines reality with virtual elements. The real scene is completed with additional, virtual elements generated by computer, which are perceived together. The AR systems are used for group working, public participation in planning, project consultations. The interaction within the Virtual Reality environments is achieved with the use of the tangible devices. The CAVE system, which is a set of video projection screens in the form of a room, allows for the perception of Virtual Reality in the real scale (or in the desired dimensions). 3D video technology enables communication between users in virtual space.
CONCLUSION New information technology offers citizens new possibilities of participation in the planning process. Essential goals and tasks to achieve with the use of new media: 1. provide communication platform suppressing a barrier of non professionalism;
Malgorzata Hanzl
490
2. allow for distant contacts; and 3. manage a participatory planning process. A detailed list of techniques and computer tools available to use in physical planning is contained in the Table 1. Table 1. Use of computer communication in local authorities’ activities in physical planning: kinds, forms and valorization of communication. Communication types viz., (i) one direction, broadcasted by authorities and addressed to citizens; (ii) one direction, senders - citizens, addressee - authorities; (iii) two directions – both citizens and authorities become senders or addressees of the communication Kind of communication
Informing
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Education
Citizens opinions
Referendum Transactions
Forms of communication
Text Text and graphic attachments – bitmaps or pdf files Text and graphic attachments – use of PPGIS Documents with commentaries Drawing and plan records presented with the use of PPGIS – a form of abstract 3D simulation of planning document Static images representing vision from bird’s or from passing-by person’s point of view. Animations representing vision from bird’s or from passing-by person’s point of view. 3D interactive simulation of proposed development. Education games – interactive presentation of planning documents. The simplest form of interaction is ability to displace within virtual world. Interactive www sites with 2D graphics. Interactive www sites with 3D graphics. Questionnaires: close – options choosing, or open – answers for questions Survey – a form of voting Opinions, observations or demands sent by email Forms allowing for opinions gathering Competitions Observation and recording of actions and phenomena Voting Mechanism of decision-making voting. It requires authentication of persons taking part in voting to get reliability. Mechanisms enabling arranging issues concerning the participation in planning via the net
Type of communication 1 2 3 × × × × ×
-
-
×
-
-
×
-
-
×
-
-
×
×
-
×
×
-
× ×
× ×
-
-
×
-
-
× × × × × ×
× × ×
-
×
×
×
×
×
Potential of the Information Technology for the Public Participation … Kind of communication
Discussion
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Cooperation with use of groupware
Forms of communication
Chat room – virtual platform of discussion with the use of text messages in Real Time. Synchronic form of communication. Message boards, discussion boards - virtual platform of discussion with the use of text messages. Asynchronic form of communication Expositions connected with discussion Panel discussion – synchronic or asynchronic Virtual public space, where users are represented by avatars who may discuss with each other, with professionals and with officials. Common use of application Data conferencing – users may modify common “board". Voice conferencing Video conferencing Electronic meeting system (EMS) – conference system built into a special purpose room (screen, projector, a few computers). Collaborative management tools: Electronic calendar (time management software) – automatically starts events, informs and reminds group members about it, Project management system – starts, track and monitor project realization , Knowledge management system — collects, organizes, manages and allows for sharing different forms of information, Social software – organizes social life of a group Collaborative software basing on WWW (UseModWiki, Scoop), or independent software (CVS - Concurrent Versions System, RCS - Revision Control System). This kind of software enables creation of final version of project by a few users simultaneously.
491
Type of communication 1 2 3 ×
×
×
×
×
×
× ×
× ×
× ×
×
×
×
× × × ×
× × × ×
× × × ×
×
×
×
×
×
×
×
×
×
Most of the examples of IT use described in literature are still experimental. The projects verify available technical possibilities and do not match real actions connected with social participation in planning. Constraints are an effect of: lack of funds, politics of authorities and technical factors: data transfer restrictions or lack of network. In most cases the main function of system is the informative one. The restriction of use of 3D graphics is also an effect of costs and time consuming of 3D modeling, and thus difficulties of current updates introduction. Most of examples show how computer tools may be used for visualizing the new development and not for constructive process of continuous public participation. Great potential lies in the use of collaborative software and groupware. The applications of this kind may be used in the citizens’ activity independently from the official authorities’ actions. The potential of experimental IT tools like Augmented Reality systems or telepresence opens utterly new ways of non-professional participation in urban planning.
492
Malgorzata Hanzl Box 4
New possibilities for participation in the planning process provided with IT: communication platform suppressing a barrier of non professionalism, distant, synchronous or asynchronous contacts, participatory planning process management.
SUMMARY
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The information technology offers new potentials of citizens’ participation in the urban planning. The essential tasks to achieve with the use of new media are: providing communication platform which suppresses a barrier of non professionalism, allowing for distant contacts and enabling participatory process management. The chapter contains a review of experiments and prototypes of different IT applications: Participatory Planning GIS, 3D models, communication platforms and computer games. Technology facilitates also collaborative distant work and citizens’ participation in the city data base completion. The most cited examples remain experimental. Great potential lies in the Augmented Reality technology, which is currently being tested. Close future of use of IT in participatory planning belongs to Web 2.0 technology. Planning, which is closely related with taking democratic decisions may become a field of use of Web 2.0 tools like: collaborative software, A glimpse into the future allows to foretell the use of automatically created virtuality showing real world along with the design conception, discussed both in real and virtual space by teams of people present in the site of discussion both: locally and virtually. The difference becomes a question of individual choice. Mitchell (2000) describes this effect as a new economy of presence: ‘In conducting our daily transactions, we will find ourselves constantly considering the benefits of the different grades of presence that are now available to us, and weighting these against the costs.’ and ‘(…) digital telecommunications infrastructure and smart spaces are now completing the system, and as a result they are introducing new possibilities and radically restructuring comparative benefits and costs.’
REFERENCES Arstein, S. (1969). The ladder of citizen participation. Journal of Institute of American Planners 35(4): 216-240. Atlee, T (2003) The Tao of democracy The Writers Collective, Cranston, RI. Batty, M. (2007). Model Cities Working Paper Series Paper 113, UCL Centre for Advanced Spatial Analysis London. http://www.casa.ucl.ac.uk/working_papers.htm accessed on February 2007. Batty, M. (2007). Planning Support Systems: Progress, Predictions, and Speculations on the Shape of Things to Come. In : Seminar on Planning Support Systems for Urban and Regional Analysis, Lincoln Institute of Land Policy, Cambridge MA, 27-28 September
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Potential of the Information Technology for the Public Participation …
493
2007, Working Paper Series Paper 122, UCL Centre for Advanced Spatial Analysis London. http://www.casa.ucl.ac.uk/working_papers.htm accessed on September 2007. Batty, M. Steadman, P., and Xie, Y. (2004). Visualization in Spatial Modeling Working Paper Series Paper 79, UCL Centre for Advanced Spatial Analysis London. http://www.casa.ucl.ac.uk/working_papers.htm accessed on March 2004. Batty, M., Dodge, M., Doyle, S., and Smith, A. (1998). Modeling Virtual Urban Environments. Working Paper Series Paper 1, UCL Centre for Advanced Spatial Analysis London. http://www.casa.ucl.ac.uk/working_papers.htm accessed on January 1998. Bourdakis, V. (2004). Developing VR tools for an urban planning public participation ICT Curriculum: The PICT Approach. In: Architecture in the Network Society (Eds. Rüdiger, B., Tournay, B., and Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe. The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen, pp 601-607. Brail, R., and Klosterman, R. (eds.) (2001). Planning Support Systems: Integrating Geographic Information Systems, Models and Visualization Tools, ESRI Press, Redlands California. Chen, G.Y.M., Shih, C.H., and Liu, Y.T. (2003). Realizing various urban spaces in Year 2050 Taiwan: Camera-matching in helicopter movement animations. In: Digital Design - 21th eCAADe Conference Proceedings Graz, Austria, 17-20 September 2003, pp 311-314 D'Aquino, P., Le Page, Ch., Bousquet, F., and Bah, A. (2003). Using self-designed roleplaying games and a multi-agent system to empower a local decision-making process for land use management: The self cormas experiment in senegal. Journal of Artificial Societies and Social Simulation 6(3) http://jasss.soc.surrey.ac.uk/6/3/5.html. Dieber, M. (2003). Paint the Town. In: Proceedings of The 2nd Annual Public Participation GIS (PPGIS) Conference, July 20-22nd 2003, Portland, Oregon, USA, www.urisa.org/PPGIS/2003/papers/Dieber.pdf Ellis, C. A., Gibbs, S. J., and Rein, G. (1991). Groupware: some issues and experiences. Commun. ACM 34(1): 39-58 Evans, S., and Hudson-Smith, A. (2001). Information Rich 3D Computer Modeling of Urban Environments. Working Paper Series Paper 35, UCL Centre for Advanced Spatial Analysis London. http://www.casa.ucl.ac.uk/working_papers.htm accessed on August 2001. Fatah, G., Schieck, A., Penn, A., Mottram, Ch., Strothmann, A., Ohlenburg, J., Broll, W., and Aish, F. (2004). Interactive Space Generation through Play, Exploring Form Creation and the Role of Simulation on the Design Table. In: Architecture in the Network Society (Eds. Rüdiger, B., Tournay, B., Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen, pp 148-158. Fischer, G, Giaccardi, E, Eden, H, Sugimoto, M and Ye, Y. 2005. Beyond binary choices: integrating individual and social creativity. Human-Computer Studies 63: 482-512 Forester, J. (1999). The Deliberative Practitioner, Encouraging Participatory Planning Processes. The MIT Press, Cambridge, Massachusetts, London England. Geertman, S. (2001). Participatory Planning and GIS: a PSS to bridge the gap. Environment and Planning B: Planning and Design 29(1): 21-35
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
494
Malgorzata Hanzl
Gindroz, R. (2003). The Urban Design Handbook, Techniques and Working Methods. Urban Design Associates, W.W. Norton & Company, New York, London. Guhathakurta, S. (2002). Urban modeling as storytelling: using simulation models as a narrative. Environment and Planning B: Planning and Design 29: 895-911 Han, S., and Peng, Z. (2003). Public Participation GIS (PPGIS) for town council management in Singapore. Environment and Planning B: Planning and Design 30: 89-111 Hanzl, M., and Wrona, S. (2004). Visual Simulation as a Tool for Planning Education, Computer Aided Participation Support. In: Architecture in the Network Society (Eds. Rüdiger, B., Tournay, B., and Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe, The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen, pp 500-507. Healey, P. (1997). Collaborative Planning: Shaping Places in Fragmented Societies University of British Columbia.. Holmgren, S., Rüdiger, B.,,Storgaard, K., and Tournay, B. (2004). Virtual Environment and Participatory Design The Electronic Neighborhood - A New Urban Space. In: Architecture in the Network Society Society (Eds. Rüdiger, B., Tournay, B., and Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen, pp 24-34. http://en.wikipedia.org/wiki/Discourse_ethics accessed on 10.09.2007 http://en.wikipedia.org/wiki/Robert_Metcalfe, accessed on 18.08.2006 http://en.wikipedia.org/wiki/The_Revolt_of_the_Masses accessed on 24.10.2007 http://www.co-intelligence.org/ accessed on 20.08.2006. http://www.collectivewisdominitiative.org/ accessed on 20.08.2006. http://www.openspaceworld.com/ accessed on 20.08.2006. http://www.ptip.org.pl accessed on 20.08.2006. http://www.solonline.org/ accessed on 20.08.2006. http://www.thedialoguegrouponline.com/ accessed on 20.08.2006. http://www.theworldcafe.com/ accessed on 20.08.2006. http://www.wie.org/collective/resources.asp accessed on 20.08.2006. Huang, Ch., Yi-Luen, Do E., and Gross, M. (2003). Mouse-Haus Table: a Physical Interface for Urban Design, University of Washington, USA. Hudson-Smith, A. (2007). Digital Urban - The Visual City, Working Paper Series, Paper 124, UCL Centre for Advanced Spatial Analysis; London. http://www.casa.ucl.ac. uk/working_papers.htm accessed on September 2007. Hudson-Smith, A., Evans, S., Batty, M., and Batty, S. (2002). Online Participation: The Woodberry Down Experiment. Working Paper Series, Paper 60, UCL Centre for Advanced Spatial Analysis; London. http://www.casa.ucl.ac.uk/working_papers.htm accessed on December 2002. Hudson-Smith, A., Milton, R., Batty, M., Gibin, M., Longley, P., and Singleton, A. (2007). Public Domain GIS, Mapping & Imaging Using Web-based Services, Working Paper Series, Paper 120, UCL Centre for Advanced Spatial Analysis; London. http://www.casa.ucl.ac.uk/working_papers.htm accessed on September 2007. Innes, J., and Booher, D. (2000). Public Participation in Planning: New Strategies for the 21st Century, University of California at Berkeley, Institute of Urban and Regional
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Potential of the Information Technology for the Public Participation …
495
Development. Working Paper 2000-07; http://www-iurd.ced.berkeley.edu/pub/WP-200007.pdf Ishii, H., Ratti, C., Piper, B., Wang, Y., Biderman, A., and Ben-Joseph, E. (2004). Bringing Clay and sand into design – continuous tangible user interfaces. BT Technology Journal 22(4): 287-299. Jankowski, P., and Nyerges, T. (2001). GIS for Group Decision Making , Taylor & Francis, New York. Kamiński, Z. (2002). Pojęcie konfliktu w planowaniu przestrzennym Architektura z.40; Politechnika Śląska, Zeszyty Naukowe Nr 1553; Wydawnictwo Politechniki Śląskiej, Gliwice. Kaye, R. (2006). ETech Day 2: What is Web 2.0? O'Reilly Emerging Technology Conference March 6-9, 2006 San Diego, California, U.S. http://www.oreillynet. com/digitalmedia/blog/2006/03/etech_day_2_what_is_web_20.html accessed on October 2007. Kieferle, J. and Wœssner, U. (2003). Combining Realities, Designing with Augmented and Virtual Reality in Digital Design - 21th eCAADe Conference Proceedings, Graz Austria, 17-20 September, pp 29-32 Kingston, R. (2002). The role of e-government and public participation in the planning process; Proceedings of XVI AESOP Congress, Volos, Greece, July 10th –14th 2002, http://www.geog.leeds.ac.uk/papers/ Lang, S. and Hovestadt, L. (2004). Interaction in Architectural Immersive Applications Using 3D Video. In: Architecture in the Network Society (Eds. Rüdiger, B., Tournay, B., and Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen pp 74-81. Lonsing, W. (2004). Augmented Reality, Augmented Reality as Tool in Architecture. In: Architecture in the Network Society (Eds. Rüdiger, B., Tournay, B., and Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen, pp 495-499 Lynch K (1960) The Image of the City MIT Press, Cambridge Massechussets Martin, A., Dean, R., and Ingle, J. (2002). A Picture Tells a Thousand Words, Community Empowerment for Regeneration, Virtual Reality and Computer Animation, HACAS Chapman Hendy Report. www.hch.eu.com accessed on February 2002. Michalak, J. (2003). Podstawy metodyczne i technologiczne infrastruktur geoinformacyjnych; Roczniki Geomatyki 2003, Tom I, Zeszyt 2; Polskie Towarzystwo Informacji Przestrzennej, Warsaw. Mitchell, W. J. (2000). e-topia Massachusetts Institute of Technology, Cambridge O’Coill, C., and Doughty, M. (2004). Computer Game Technology as a Tool for Participatory Design. In: Architecture in the Network Society (Eds. Rüdiger, B., Tournay, B., and Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe. The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen, pp 12-23. O'Reilly, T. (2005). What Is Web 2.0 Design Patterns and Business Models for the Next Generation of Software, http://www.oreillynet.com/lpt/a/6228. Ortega, Y., and Gasset, J. (1982). Bunt mas i inne pisma socjologiczne PWN Warsaw Poland.
496
Malgorzata Hanzl
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Sanoff, H. (2000). Community Participation Methods in Design and Planning John Willey and Sons, New York USA. Seichter, H., and Kvan, T. (2004). Tangible Interfaces in Design Computing. In: Architecture in the Network Society (Eds. Rüdiger, B., Tournay, B., and Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen, pp 159-166. Stellingwerff, M., and Kuhk, A. (2004). 3D/4D Communication Tools for Facilitators in Public Participation. In: Architecture in the Network Society (Eds. Rüdiger, B., Tournay, B., and Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen, pp 593-600. Szalapaj, P.J. (2003). Architectural GIS: Interoperable and Integrated Information Environments. In: Digital Design - 21th eCAADe Conference Proceedings, Graz Austria, 17-20 September, pp 319-325 Tabak, V., and deVries, B. (2003). Interactive Urban Design using integrated planning requirements control. In: Digital Design - 21th eCAADe Conference Proceedings, Graz Austria, 17-20 September, pp 295-300 Voigt, A., Achleitner, E., Linzer, H., Schmidinger, E., and Walchhofer, H. (2003). Multidimensional Digital City Models. In: Digital Design - 21th eCAADe Conference Proceedings, Graz Austria, 17-20 September, pp 253-257. Voigt, A., Martens, B., and Linzer, H. (2004). City Simulator: A Multi-dimensional VRSimulation Environment. In: Architecture in the Network Society (Eds. Rüdiger, B., Tournay, B., and Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe. The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen, pp 586-592 Wœssner, U., Kieferle, J., and Drosdol, J. (2004). Interaction Methods for Architecture in Virtual Environments. In: Architecture in the Network Society (Eds. Rüdiger, B., Tournay, B., and Ørbæk, H.). Proceedings of the 22th Conference on Education and Research in Computer Aided Architectural Design in Europe. The Royal Danish Academy of Fine Arts School of Architecture, Copenhagen, pp 66-73. Wrona, S. (1981). Participation in Architectural Design and Urban Planning Wydawnictwo Politechniki Warszawskiej, Warsaw, Poland.
QUESTION BANK Short Answer Questions 1. Explain the term e-participation. 2. Explain VRML and XML. What is their importance? 3. How information communication technology can contribute to better planning and implementation? 4. What are augmented reality and virtual reality systems? 5. What do you understand by the term collaborative software?
Potential of the Information Technology for the Public Participation …
497
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
6. What is importance of public participation in geoinformatics? How it an help in planning and management of natural resources? 7. What is net participation? What are its different components? 8. What is PPGIS? How it strengthens the geospatial modeling and planning? 9. What is visualization? What are different techniques of enriching visualization? 10. What is WebGIS? What is its role in natural resource management?
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 21
EXPLORING ALTERNATE FUTURES WITH GIS: A PLANNING SUPPORT SYSTEM Richard E. Klosterman* What if?, Inc., Hudson, Ohio 44236, USA
ABSTRACT The powerful spatial analysis and display capabilities of today’s geographic information systems (GIS) have proven to be extremely useful for urban and regional planners around the world. However, while useful for considering current and past conditions, the challenge with today’s GIS lies in determining the effect of current development trends and alternative public policies on the future conditions. This chapter describes a GIS-based planning support system (PSS) called What if?TM that uses database on land use, land cover, projected population and trends to consider alternative scenarios for the suitability of land and alternative public policies to explore alternative land use, population, and employment trends. What if?TM is currently being applied in dozens of locations around the world. The role that What if?TM can play in public decision making is illustrated with two applications in the United States.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Keywords: GIS, land use, land cover, predictive modeling, What if?TM
INTRODUCTION The planning profession’s methodological foundations are in question. The combined forces of academic theorizing and political expedience have made public participation “the cry of the age” (Campbell and Marshall, 2000), as private citizens, elected officials, and planners are increasingly abandoning the ideal of neutral professional expertise for a new *
Email: [email protected] Originally published as: “Deliberating About the Future," by Richard E. Klosterman, pp. 199-219. In: Engaging the Future: Forecasts, Scenarios, Plans, and Projects, edited by Lewis Hopkins and Marisa Zapata, published by the Lincoln Institute of Land Policy, Cambridge, Massachusetts
500
Richard E. Klosterman
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
emphasis on public involvement. On the other hand, technically oriented practitioners and academics continue to rely on the methods and models that are based solidly on the technocratic ideal of value-neutral science. The mismatch between planning education and practice is revealed clearly in many studies, such as by Kaufman and Simons (1995), which found that the “supply” of methods offered by 43 methods courses matched the “demand” for methods expressed by 106 planning practitioners for only one-quarter of the 53 methods considered. Methods for which supply exceeded demand, i.e., methods that were more heavily taught than practitioners felt was necessary, included the projection techniques and statistical methods that underlie the applied science model. Methods for which demand exceeded supply, i.e., the methods not taught as often as practitioners believed they should be, included forecasting, scenario construction, and impact analysis. In academia, the inherently political nature of planning, the importance of informal communication, and the need for combining technical competence with political sophistication have occupied the core of planning theory courses since the 1970s (Klosterman, 1981; 1992). However, these lessons have been largely neglected in planning curricula, which all too often treat analytical methods as value-neutral objective techniques of scientific analysis. This image of value-neutral technical competence is reflected in professional practice by the continued reliance on sophisticated analytical techniques and increasingly complex computer-based models that portray planners’ forecasts and analysis in dry images of technocratic expertise. The sophistication of planners’ analysis and forecasts is, however, often more apparent than real. The core assumptions that underlie a forecast (e.g., whether past growth rates will continue, increase, or decrease) play a much larger role in determining the forecast outcomes than the sophistication of the tools used to prepare them. Other equally important choices must be made in the selection of data, the application of computational procedures, and the analysis, presentation, and distribution of results. These choices are inherently political because they determine not only the analysis results but also the perception of problems and the identification of solutions, helping determine who gets what, when, and how (Klosterman, 1987; Wachs, 1982). Recognizing that forecasts are both politically influential and difficult, if not impossible to evaluate; planners and their clients can all too easily claim to be carrying out a value-neutral process of forecasting, while adjusting the underlying assumptions to produce their preferred outcomes (Wachs, 2001).
PRINCIPLES GOR DELIBERATIVE FORECASTING A realistic assessment of forecasting must begin by recognizing that planners’ information and knowledge are limited, their data and models often contain large amounts of error, and their forecasts are almost always wrong. Many of the points in the remainder of this section were first outlined by Klosterman (1987). More importantly, it must acknowledge that the future is impossible to predict, particularly for small areas and long time periods. As a result, planners should abandon the unrealistic goal of preparing exact predictions of an unknowable future. Instead they should prepare a range of forecast scenarios describing a range of possible futures. Recognizing that there is no way to evaluate a forecast until the forecast date arrives, planners must recognize that accuracy cannot be the proper criterion for
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Exploring Alternate Futures with GIS: A Planning Support System
501
evaluating forecasts. Instead, planning forecasts should be evaluated with respect to their ability to inform the policy-making process, facilitate community understanding, and prepare the public to deal with an uncertain future. Judged on these grounds, good forecasts will incorporate as many different kinds of information from as many perspectives as possible and help reduce the influence of expedient or self-serving viewpoints and overly optimistic or pessimistic thinking (Skaburskis, 1995). Planners must also acknowledge that their forecasts are ultimately dependent on their underlying assumptions, and consciously adopt the “what if” metaphor popularized by electronic spreadsheets. That is, they must explicitly acknowledge that the results of their analysis only indicate what would happen if the underlying assumptions were correct. This suggests that planning models should explicitly state their underlying assumptions concerning future trends and alternative policy choices. These assumptions should be easily modified and the effects of alternative assumptions and policy choices should be easily identified In addition, planners must recognize that they have no special knowledge about the future and that sophisticated projection models and methods will probably be no more accurate than simpler ones (Skaburskis, 1995; Smith, 1997; Wachs, 1989). They must also acknowledge that their models will only be useful in a policy context if policy makers and the public understand and trust them. As a result, planners should attempt to develop models that are as simple rather than as complex as possible. While the model’s detailed computational procedures will generally be too involved for non-experts to understand, their underlying structure, assumptions, and limitations should be as explicit and clear as possible. Finally, planners’ analysis, methods, and models must be documented fully so that policy makers, the public, and other experts can understand them. Planning forecasts and the models that are their bases all too often fail to describe the data, computational procedures, and assumptions on which they are based. As a result, elected officials and the public are forced to rely on the undocumented professional expertise of the often unknown individuals who prepared them. Only by making these foundations explicit and one can properly design and document models and methods that allow policy makers, public, and other experts to evaluate adequately the policy analysis and recommendations they receive. On this model, public policy–making would not be based on the assumed expertise of professionals or the presumably objective analysis of technical experts. Instead, it would be based on explicitly political processes of deliberative modeling in which community members use mutually agreed-upon models, techniques, and data to examine policy questions from their own, perhaps fundamentally different, perspectives (Klosterman, 1987; Wachs, 2001). By helping reveal the possible outcomes, different assumptions, and actions, a new generation of computer tools may thus help provide the technical foundations for community-based processes of collective design, collaborative planning, and consensus building that attempt to achieve collective goals and deal with common concerns (Klosterman, 1997).
IMPLEMENTING THE FORECASTING PRINCIPLES This chapter describes an attempt to incorporate these ideals into an operational planning support system (PSS) named “What if?TM”. For a more complete description of the model see (Klosterman 1999; 2001) and the What if? website (www.What-if-PSS.com). As its name
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
502
Richard E. Klosterman
suggests, What if? does not attempt to predict a future as if we could get it right. Instead, it is an explicitly policy-oriented planning tool that can be used to determine what would happen if clearly defined policy choices are made and assumptions concerning the future prove to be correct. Policy choices that can be considered in the model include the staged expansion of public infrastructure, the implementation of alternative land use plans or zoning ordinances, and the establishment of farmland or open space–protection programs. Assumptions about the future that can be considered in the model include future population and employment trends, household characteristics, and development densities. What if? is a relatively simple, rule-based model that does not attempt to duplicate the complex spatial interaction and market clearing processes that shape the urban fabric (Klosterman and Pettit, 2005). Instead, it incorporates a set of explicit decision rules for determining the relative suitability of different locations, projecting future land use demands, and allocating the projected demands to suitable sites. The model is a stand-alone product that adapts to the GIS data for any area and can be used to project the population, housing, and employment for census enumeration areas, political jurisdictions, school districts, traffic analysis zones, and other user-defined areas. It has been applied in the United States at the township, county, and regional levels (Klosterman et al., 2002; 2006) and internationally (Kweon and Kim, 2002; Pettit, 2005). What if? includes four major components – (i) current, (ii) suitability, (iii) demand, and (iv) allocation. The current component allows the user to view maps showing the GIS layers that are used in the analysis and reports listing current land use, population, housing, and employment information for the study area and its subareas. The suitability component considers the supply of land by allowing the user to specify: (i) the importance of different factors in determining the relative suitability of different locations for accommodating future land use demands; and (ii) public policies that limit the amount of developable land. The demand component considers the demand for land by allowing the user to prepare scenarios projecting the amount of land that will be required to accommodate future population and employment growth. Finally, the allocation component, jointly considers supply and demand by allowing the user to create allocation scenarios that project future land use, population, and employment patterns by allocating the projected land use demands (as determined by a demand scenario) to the most suitable locations (as determined a suitability scenario). The allocation scenarios can also incorporate public policies such as the implementation of a land use or open space preservation plan or the staged expansion of public infrastructure. The role that What if? suitability, demand, and allocation scenarios can play in deliberative planning practice will be illustrated with two examples: Waupaca County, Wisconsin, and Dublin, Ohio.
SUITABILITY SCENARIOS In the spring of 2003, Waupaca County, Wisconsin, received a state planning grant to support the preparation of a countywide comprehensive plan and 33 local plans. The planpreparation process included representatives from each local community, the county’s economic development director, planning specialists from the University of Wisconsin Stevens Point Center for Land Use Education, the University of Wisconsin Extension
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Exploring Alternate Futures with GIS: A Planning Support System
503
Community Resource Development Educator, and a consulting firm. Working closely with the county’s professional staff, the group used What if? and GIS data from the county land information officer to create over 250 suitability scenarios for the county and 33 municipalities. The scenarios were created at fifteen public meetings, conducted over a sixmonth period, involving hundreds of participants. The process of creating suitability scenarios engaged the local community-planning commissions more concretely in planning for their communities than at any previous point in the process. Questions such as what does it mean for development if we protect farmland? had previously been abstract because they lacked a solid connection to the landscape, and there was little information on the implications of alternative choices. The process of creating the What if? suitability maps encouraged people to think realistically about their goals and how to achieve them. By preserving some areas and opening other areas to development, the planning commission members could readily see the benefits and the costs of preserving farmland and the implications these choices would have on future development patterns. The suitability analysis process was based on maps that had been reviewed and corrected by the local planning commissioners. The process of reviewing and correcting the maps helped the commissioners become more familiar with their communities and created a sense of ownership in the data on which the suitability decisions were based. The geography expressed in the maps provided a shared foundation for understanding the local landscape and the implications of implementing alternative actions. Perhaps most importantly, the suitability maps were developed through a consistent and open process of decision making in public meetings involving the local planning commissions. This meant that the resulting suitability maps and the decisions on which they were based were defensible and transparent. For example, if asked why particular areas were or were not available for development, the commissioners could point to the What if? suitability maps and say, “We allowed development to occur here and not over there because we prefer to protect prime farmlands and expressed this in a policy to do so, and we believe that developers are more likely to develop here than there because of steep topography and access to public services.” The What if? suitability scenarios can thus consider the assumed behavior of developers, community preferences about the future, and policies expressing these community preferences. In deliberative practice, it is important to be clear about which of these three factors is incorporated in a particular suitability scenario. For example, Figure 1a shows the suitability map for low-density residential development in the town of Union, Wisconsin, under a scenario that assumes no public policies are implemented to limit future development. Towns in Wisconsin are similar to townships elsewhere in being a complete regionalization of the state, and thus include rural as well as urban areas. The areas shown in green are assumed to be suitable for residential development from the perspective of developers; areas in dark green are assumed to be more suitable than areas shown in light green. Areas in grey are currently developed, and areas shown in white are water bodies or have high slopes, which prohibit residential development. This scenario simulates the behavior of developers considering only the ability of alternative sites to accommodate future growth, unconstrained by community preferences or public policies for directing that growth. Under this scenario, over 20,000 acres of land are available to accommodate future residential growth.
504
Richard E. Klosterman
(a)
(b)
(c)
(d)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Not Not Developable Convertible
Not Suitable
Low
Moderately Low
Moderate Moderately High
High
Figure 1. Residential Suitability for the Town of Union in different scenarios (a) No Controls Scenario (b) Farmland-Protection Scenario; (c) Environmental Protection Scenario; (d) Agricultural and Environmental Protection
Three alternative scenarios consider both the assumed behavior of developers and the effect of alternative public policies for guiding development. Figure 1b shows the low-density residential suitability map for a scenario that assumes the town implements a farmlandpreservation policy that prohibits development in areas with prime agricultural soils or within a quarter mile of a dairy farm. Under this scenario, only 12,300 acres are available for residential development, and large portions of the town will be protected from development, as the map indicates. Figure 1c shows the low-density residential suitability map for a scenario that assumed that the town implements an environmental protection policy that prohibits development in wetlands or within 100 feet of rivers and streams, wetlands, and managed forestry areas. The total quantity of land that is available for residential development
Exploring Alternate Futures with GIS: A Planning Support System
505
is nearly identical to the farmland-preservation scenario, but the location of developable land is substantially different. Figure 1d shows the low-density residential suitability map for a scenario that assumed that the town implements both the agricultural preservation and environmental protection policies. Under this scenario, only 5,400 acres are available for development, severely limiting the town’s ability to accommodate future residential development. Together, the four scenarios reflect the assumed behavior of land developers and demonstrate clearly the tradeoffs between the community’s desires to accommodate future growth, preserve prime agricultural land, and protect the environment.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
DEMAND SCENARIOS The role What if? can play in considering the future demand for land can be considered by examining Dublin, Ohio, a rapidly growing community northwest of Columbus. The information provided in the next two sections is illustrative and only approximates the situation in Dublin. The city’s population was less than 4,000 in 1980, quadrupled by 1990, and nearly doubled again to reach 31,400 by 2000. As a result, the community’s residents are worried that this growth will continue and, if it does, the community will have difficulties dealing with the implications of this growth. A conservative projection of past growth trends suggests that the city’s population and employment will grow by only 14 percent in the next 25 years. Assuming that the community’s housing mix and the residential and employment-related land use densities do not change, this will create a demand for roughly 2,100 more acres of residential land and 300 more acres of land for employment-related land uses. A moderate growth projection assumes that the city’s population will grow by 52 percent and its employment will grow by 29 percent, requiring 4,680 acres of additional residential land and 630 acres of additional employment-related land. A high projection of past growth trends suggests that the community’s population will nearly double over the next 25 years and its employment will increase by 63 percent. Again, assuming that the community’s housing mix and land use densities do not change, 7,700 additional acres of residential land and 1,400 acres of employment-related land would be required to accommodate the anticipated growth. To put this in perspective, if the high growth trends are observed, the quantity of land devoted to residential uses will more than double and the amount of land devoted to employment-related uses will increase by roughly 50 percent in 25 years. What if? recognizes that future population growth is not the only factor that must be considered in creating demand scenarios for future land use demands. The demand for residential land is also dependent on the average household size and the average housing density. That is, if the average household size (persons per household) continues to decline (as it has over the last few decades), the number of housing units will increase more rapidly than the population, increasing the demand for land. Similarly, if the average household densities (housing units per acre) decrease, as it has in most suburban areas, the demand for land will increase more rapidly than the population. The amount of land required to accommodate future employment growth is likewise dependent not only on projected employment growth but also on future employment densities
506
Richard E. Klosterman
(employees per acre). The demand for land is also dependent on pubic policies concerning the quantity of land to be devoted to recreational and other public uses and set aside for open space preservation or agricultural protection uses. These factors can easily be considered in What if? by modifying future development densities and by specifying the amount of land to be protected from development or reserved for parks and recreation and other local land uses. The demand scenarios incorporate assumptions about both the behavior of individuals, such as changes in the average household size, and public policies, such as the permitted development densities, that convert the projected population and employment trends into the equivalent land use demands. As was true for the suitability scenarios, it is important to be clear about the behavioral assumptions and public policies that are expressed in a particular demand scenario.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ALLOCATION SCENARIOS What if? projects future land use, population, and employment patterns for up to five projection years (for example, 5, 10, 15, 20, and 25 years) by allocating the projected land use demands—as specified by a demand scenario—to different locations on the basis of their relative suitability, as defined by a suitability scenario. An allocation scenario adds two kinds of rules to the suitability and demand scenarios on which it is based. It sets the order in which land will be allocated to each use at each projection year, such as assigning commercial demand before residential demand. It also allows the availability of developable land to be modified in successive projection years to account for the staged expansion of infrastructure, such as extending sewer and water service or building new major roads and freeway interchanges at particular times. An allocation scenario can thus incorporate the public policies expressed in a suitability scenario and other policies for implementing land use plans or zoning restrictions and for extending infrastructure. It can also include the assumptions concerning the behavior of developers expressed in a suitability scenario (concerning, for example, the desirability of developing sites with different natural features). It can also incorporate assumptions about the behavior of developers in the form of assumed growth patterns, such as a GIS layer that numbers buffers around currently developed areas in increasing order by their distance from urban concentrations. The model can then use the growth pattern values in conjunction with the suitability of different locations to specify the order in which different areas will be developed. For example, a user could assume that developers are more concerned with sites’ natural features than with their accessibility and order the development of land parcels first by their suitability scores and then by their growth pattern scores. Conversely, a user may assume that developers value nearby sites more than they value the sites’ natural features and specify that the allocation is guided first by the growth pattern scores and then by the suitability scores. The second option may also express a desire for—or a public policy of—encouraging compact development. Here again it is important to clearly identify the behavioral, community preference, and policy assumptions that generate the projected development patterns for a particular allocation scenario.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Exploring Alternate Futures with GIS: A Planning Support System
507
Consider, for example, an allocation scenario that assumes (i) the suitabilities defined by a preservation suitability scenario; (ii) the projected demands for the high-growth demand scenario; and (iii) public policies requiring new industrial development to be located in areas that are zoned industrial and have sewer service in a given year as specified by a 2030 Plan allocation scenario. To allocate the projected industrial demand in the first projection year, the model (i) selects all of the land use polygons that are vacant or can be converted to industrial uses as defined by the preservation suitability scenario; (ii) selects all of the polygons in this set that are zoned industrial and have sewer service in the first projection year as specified by the 2030 Plan allocation scenario; (iii) rank orders these polygons in decreasing order of their suitability scores for industrial use as computed for the preservation suitability scenario; (iv) determines the projected demand for industrial land in the first projection year as computed by the high-growth demand scenario; (v) converts the land use for the polygon with the highest industrial suitability score to industrial use and deducts the polygon’s area from the projected industrial demand for industrial land; and (vi) converts the use for the developable polygons with progressively lower industrial suitability scores to industrial use and deducts the areas for these polygons from the projected industrial demand. The process continues until the demands for all land uses in the first projection year have been satisfied. The model then repeats this process to allocate the projected demand for all land uses in the remaining projection years. The process stops when all of the demand has been allocated for all of the projection years or the model runs out of land, meaning that there is not enough suitable land to satisfy the projected demand in a given year. If this happens, the model issues a warning statement, and the user must modify the model assumptions to increase the supply of suitable land or reduce the demand for land. For example, Figure 1(a) shows the current land uses for Dublin. The western half of the city is still vacant or devoted to agricultural uses (shown in green). The city’s commercial and industrial areas (shown in purple and red) are concentrated near the circumferential road around Columbus that cuts through the lower right-hand corner of the city and along the interstate highway that leads to the northwest. The residential areas (shown in yellow) are located in the center of the city and inside the circumferential road. Figure 1(b) shows the projected land uses in 2030 under a scenario that assumes the high population and employment projections and no effort to protect agricultural land or environmentally sensitive land. As the map shows, under these assumptions the vacant land in the western and southern parts of the city has been converted largely to residential uses, and industrial and commercial uses are scattered throughout the southern part of the city. This scenario suggests that the city’s rural character (shown in Fig. 1(a)) will be lost if the community continues to grow rapidly and nothing is done to protect its vacant and agricultural lands. Figure 1(c) shows the projected land uses in 2030 under a scenario that assumes the medium population and employment projections and no efforts to protect agricultural land or environmentally sensitive land. As the map shows, there is much less residential development under these assumptions than there is under the high-growth scenario, but large portions of the vacant land in the western part of the city have still been converted to residential and employment-related uses, severely reducing the area’s rural character. Figure 1(d) shows the projected land uses in 2030 under a scenario that assumes the medium population and employment projections and the implementation of an open space preservation plan that does not allow development in the western quarter of the study area. As the map shows, under these assumptions the vacant land in the western part of the city is
Richard E. Klosterman
508
preserved; residential development is concentrated in the central parts of the city; and new employment-related uses are located near the expressway. It is important to note that the projected land use demands for this scenario are identical to those for the no-growth scenario shown in Figure 3. The open space preservation plan accommodates the same amount of growth; it just concentrates in the central part of the city, preserving the area’s agricultural character. Unfortunately, the preservation plan does not provide enough land to accommodate the high-growth demand scenario; this growth can only be accommodated by sacrificing the city’s vacant and agricultural land or by substantially increasing its residential densities.
a)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
(c)
(b)
(d)
Figure 2. (a) Current Land Uses, Dublin Ohio; (b) 2030 Dublin Land Uses: High-Growth with No Controls Scenario; (c) 2030 Dublin Land Uses: Medium Growth with No Controls Scenario; (d) 2030 Dublin Land Uses: Medium-Growth and Open Space–Preservation Plan Scenario
Exploring Alternate Futures with GIS: A Planning Support System
509
Figures 3(a); 3(b) and 3(c) show the changes over time in the land uses, population, and employment for the moderate-growth with no growth controls scenario shown in Figure 3 for the city’s eleven traffic analysis zones (TAZs). Figure 3(b) shows that the rapid residential growth in the southwestern part of the study area continues throughout the 25-year projection period. In contrast, the residential growth in the northwestern part of the city occurs largely after 2025. Figures 3(b) and 3(c) show that the city’s population and employment growth over the 25-year projection period occurs almost exclusively in the western part of the city and that the residential population growth is much more dramatic than the employment growth.
(b)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
(a)
(c) Figure 3.(a) Low-Density Residential Land for Dublin Traffic Analysis Zones, 2005–2030: Medium Growth with No Growth Controls; (b) Total Population for Dublin Traffic Analysis Zones, 2005–2030: Medium Growth with No Growth Controls; (c) Total Employment for Dublin Traffic Analysis Zones, 2005–2030: Medium Growth with No Growth Controls
510
Richard E. Klosterman
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
CONCLUSIONS Unlike more theoretically sophisticated and complex models, What if? does not include measures of spatial interaction, represent the interlinked markets for land, labor, and infrastructure, or explicitly model the behavior of households, businesses, and developers. The model does not pursue theoretical sophistication for its own sake or attempt to find one correct projection of an unknowable future. Instead, it has been designed to provide an understandable, transparent, adaptable, and fully operational model, which helps a community understand its present, consider its future, and evaluate alternative policies for achieving its collective goals. It also provides a useful framework for framing and determining the implications of assumptions about the behavior of developers, community preferences about a desired future, and public policies for achieving those preferences that underlie different conceptions of an area’s future. In this way, it attempts to implement the forecasting goals outlined by Moore (2007) and extend the deliberative forecasting described by Isserman (2007) to the sub-county level. The model’s procedures for balancing the supply of and demand for land by determining the relative suitability of different locations, projecting the demand for land, and allocating the projected demand to the most suitable sites subject to public policies for directing growth can be readily understood by planners, elected officials, and the public. The model’s simplicity is reflected in its flexible and rather modest data requirements: GIS layer describing an area’s current land uses, natural features, administrative boundaries, and growth-management policies and information on its current and projected population and employment trends. This allows the model to be used to project up to 50 different land uses and an area’s future residential population, housing units, and households. It can also be used with commercially available data for the United States to project the employment by place of work for two-digit NAICS (North American Industrial Classification System) sectors. These projections can be prepared for subcounty areas such as TAZs, political jurisdictions, and taxation districts, providing the information needed for transportation planning and fiscalimpact studies. However, it can also be used in areas with limited spatially related information. Like the LEAM model described by Deal and Pallathucheril (2007), What if? is an example of a new generation of computer tools that attempts to use the dramatic improvements in computing power and the availability of spatially related data to support community-based processes of deliberative decision making, collaborative planning, and consensus building. However, tools like this are not enough. Planning practitioners and academics must be willing to apply them in practice, providing the practical experience and financial support needed to refine and improve them. Until that happens, computer technology’s potential for supporting community-based deliberative forecasting may never be realized.
Exploring Alternate Futures with GIS: A Planning Support System
511
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REFERENCES Aziza, Parveen. (2006). What if? Evaluation of growth management strategies for a declining region. International Journal of Environmental Technology and Management 6(1/2): 79– 95. Campbell, Heather J., and Marshall, R. (2000). Public involvement and planning: Looking beyond the one to the many. International Planning Studies 5: 321–343. Deal, Brian and Varkki George Pallathucheril. (2007). Developing and Using Scenarios. In: Engaging the Future: Forecasts, Scenarios, Plans, and Projects (Eds. Lewis D. Hopkins and Marisa A. Zapata), Cambridge, Massachusetts: Lincoln Institute of Land Policy, pp. 221-242. Isserman, Andrew M. (2007). Forecasting to Learn How the World Works. In: Engaging the Future: Forecasts, Scenarios, Plans, and Projects (Eds. Lewis D. Hopkins and Marisa A. Zapata), Cambridge, Massachusetts: Lincoln Institute of Land Policy, pp. 175-198. Kaufman, Sanda, and Simons, R. (1995). Quantitative and research methods in planning: Are schools teaching what practitioners practice? Journal of Planning Education and Research 15: 17–34. Klosterman, Richard E. (1981). Contemporary planning theory education: Results of a course survey. Journal of Planning Education and Research 1: 1–11. Klosterman, Richard E. (1987). Politics of computer-aided planning. Town Planning Review 58(4): 441–452. Klosterman, Richard E. (1992). Planning theory education in the 1980s: Results of a course survey. Journal of Planning Education and Research 11: 130–140. Klosterman, Richard E. (1997). Planning support systems: A new perspective on computeraided planning. Journal of Planning Education and Research 17(1): 45–54. Klosterman, Richard E. (1999). The what if? Collaborative planning support system. Environment and Planning B: Planning and Design 26: 393–408. Klosterman, Richard E. (2001). The what if? Planning support system. In Planning support systems: Integrating geographic information systems, models, and visualization tools. Richard K. Brail and Richard E. Klosterman, eds., 263–284. Redlands, CA: ESRI Press. Klosterman, Richard E. and Pettit, C.J. (2005). Guest editorial: An update on planning support systems. Environment and Planning B: Planning and Design 32(4): 477–484. Klosterman, Richard E., Loren Siebert, Mohamned Ahmadul Hoque, Jung-Wook Kim, and Aziza Parveen. (2002). Using an operational planning support system to evaluate farmland preservation policies. In Planning support systems in practice. Stan Geertman and John Stillwell, eds. Heidelberg: Springer. Kweon, Ihl, and Jung-Wook Kim. (2002). Urban land use planning with a PSS-based land use change model. Journal of the Geographic Information System Association of Korea 10(4): 512–532. Moore, Terry. (2007). The Use of Forecasts in Creating and Adopting Visions for Urban Growth. In: Engaging the Future: Forecasts, Scenarios, Plans, and Projects (Eds. Lewis D. Hopkins and Marisa A. Zapata), Cambridge, Massachusetts: Lincoln Institute of Land Policy, pp. 19-38.
Richard E. Klosterman
512
Pettit, Christopher J. (2005). Use of a collaborative GIS-based planning support system to assist in formulating a sustainable-development scenario for Hervey Bay, Australia. Environment and Planning B: Planning and Design 32(4): 523–546. Skaburskis, Andrejs. (1995). Resisting the allure of large projection models. Journal of Planning Education and Research 14: 191–202. Smith, Stanley K. (1997). Further thoughts on simplicity and complexity in population projection models. International Journal of Forecasting 13: 557–565. Wachs, Martin. (1982). Ethical dilemmas in forecasting for public policy. Public Administration Review 42: 562–567. Wachs, Martin. (1989). When planners lie with numbers. Journal of the American Planning Association 55: 476–480. Wachs, Martin. (2001). Forecasting versus envisioning: A new window on the future. Journal of the American Planning Association 67(4): 367–372.
QUESTION BANK Define Following Terms 1. 2. 3. 4. 5.
TAZ PSS Land Use Deliberative modeling. Rule-based model.
Short Answer Questions
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. What the challenges of Land use land cover change to planners? 2. What is planning support system? 3. What is land use policy? What are different spatio-temporal data requirements for this? 4. Write a brief note on “What if?TM”. What are its different components? 5. What is predictive modeling? Why is this important?
Long Answer Questions 1. Why are planners’ analyses and forecasts more political than they appear? 2. What principles should guide forecasting? 3. What advantages did the process of creating Suitability scenarios have for the planning process in Waupaca County? 4. What assumptions and policy choices can be considered in the “What if?TM”? 5. How does the “What if?TM” project future land use patterns?
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 22
USING SDI AND SERVICE BASED SYSTEMS TO FACILITATE NATURAL RESOURCE MANAGEMENT Ali Mansourian*, Mohammad Taleai† and Narges Babazadeh‡ Faculty of Geodesy & Geomatics Engineering K.N.Toosi University of Technology, Iran
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ABSTRACT It has been accepted as a fact that the quantity and quality of natural resources and environment affect the overall quality of life, economic growth, and, in some instances, the mankind very existence. Meanwhile, many developing countries have rapidly lost or degraded their valuable nature resources base as a price for rapid economic growth. As a result, today there are several environmental problems and natural resource management related issues that represent immediate concerns for sustainable development of the countries. Due to diversity of natural resources, broad spectrums of governmental and private organizations from different political/administrative levels (local, state/provincial and national levels) are involved for decision-making and planning process of managing natural resources. Therefore, the delivery of a sustainable natural resource management relies on collaborative effort and coordination of the involved organizations in such a management. Spatial information and related technologies have proven crucial for natural resource management. More specific, sharing spatial data among different organizations involved in natural resource management is essential for coordinated planning and decision-making, based on common datasets. However, current studies show there are currently substantial problems with availability, access, sharing and usage of reliable, upto-date and accurate data for natural resource management. Therefore, it is necessary to utilize appropriate frameworks and technologies to resolve current spatial data problems for managing natural resources. This paper aims to address the role of Spatial Data Infrastructure (SDI) as a framework for facilitating natural resource management by resolving current problems with spatial data. It is argued that the design and implementation of an SDI model and consideration of SDI development factors and *
[email protected] [email protected] ‡ [email protected] †
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
514
issues, together with development of a web-based system, can assist natural resource management agencies to improve the quality of their decision-making and increase efficiency and effectiveness in natural resource management activities. This is based on an ongoing research project on the development of an SDI conceptual model and a prototype web-based system which can facilitate sharing, access and usage of spatial data in natural resource management.
Key Words: Geospatial tools, NRM, SDI, service based system, WebGIS
INTRODUCTION
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
A natural resource may be defined as a supply of raw materials that are furnished by nature and bring a country wealth. Natural resources are frequently classified as renewable and non-renewable (Osundwa, 2001; Lojala, 2003, EEA, 2005). Renewable resources are defined as resources that are regenerated on a human time scale. These include substance such as water, fisheries, forests etc. These types of resources are often connected in ecological systems; for example, water is necessary for forest growth and fisheries. By contrast, nonrenewable resources are matters which have evolved over geological periods of time and therefore effectively cannot be replaced. These types of resources are less likely to participate in the circular flows of the ecosystem, and exploitation of one resource typically does not affect the availability of the other resources (as long as the extraction does not destroy the other resource). These include substances such as metal ores, coal, oil, and natural gas. It is notable that renewable resources are only self-generating if they are allowed to be so. Overexploitation effectively converts some renewable resources into non-renewable (Osundwa, 2001). It has been accepted as a fact that the quantity and quality of natural resources and environment affect the overall quality of life, economic growth, and in some instances, the very existence of mankind. However, with such an importance of natural resources, many countries have rapidly lost or degraded their valuable nature resources base as a price for rapid economic growth (Schmidt-Vogt and Shrestha, 2006). In brief, three persistent concerns are consistently raised regarding natural resources, particularly in developing countries (World Bank, 2000; EEA, 2005): • • •
Renewable resources are utilized beyond their regenerative capacity; Non-renewable resources are depleted with insufficient savings in man-made, human, or social capital, and with minor benefits directed; and The ‘sink’ capacity of the environment is overburdened by pollution, which in turn damages human health and ecosystem functions.
The last thirty years have witnessed a growing understanding that the earth cannot sustain current levels of pollution and utilization of natural resources and hence a continuous effort to survey and monitor the quantitative and qualitative characteristics of the environmental and natural resources conditions is essential (Osundwa, 2001). As a result, Natural Resource Management (NRM) has been brought into front as part of sustainable development activities of the countries.
Using SDI and Service Based Systems to Facilitate Natural Resource…
515
NRM studies the management of natural resources with a particular focus on how management affects the quality of life for both present and future generations. It is about making and implementing decisions to develop, maintain or protect ecosystems and natural resources to meet human needs and values. One of the priority concerns in NRM is gaining a better understanding of the ways that our environments are changing and how our activities interact and influence such processes. This can be achieved by systematically conducting the following NRM-related operations (Schmidt-Vogt and Shrestha, 2006): • •
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
•
Assessment of the availability and condition of natural resources (mainly through inventory and monitoring techniques); Analysis of problems, potentials, and perspectives (through a wide variety of analytical methods from natural, social, economic, and political sciences; e.g. policy analysis, institutional analysis, livelihood analysis, analysis of ecosystem processes, impact analysis, cost and benefit analysis); and Development and application of solutions and alternatives (through planning and management strategies and processes).
Because of diversity of natural resources, different governmental and private organizations from different political/administrative levels are involved for decision-making and planning process of managing natural resources. Therefore, proper conduction of the above-mentioned operations highly relies on collaborative efforts of NRM organizations as well as coordination of the activities of these organizations. Spatial information and related technologies have been proven crucial for NRM. Having reliable and up-to-date spatial data about natural resources, current plans and activities, and the effect of activities on other natural resources can considerably facilitate conduction of the above-mentioned operations. Such spatial data can also significantly facilitate coordination of the NRM organizations. However, there are currently substantial problems with collection, access, dissemination and usage of required spatial data for NRM. This is a fact which has been highlighted by NLWRA (2002), Hartanta (2006) and Taylor et al. (2006) as well. It is suggested that Spatial Data Infrastructure (SDI) as an initiative in spatial data management, along with appropriate service-based systems, can be used as an integrated framework for resolving current problems with spatial data. This paper aims to describe the development of an SDI conceptual model and a prototype service-based system that facilitates spatial data collection, access, dissemination, management and usage for effective and efficient NRM. This is based on an ongoing research and case study in Iran which investigates the role of SDI and service-based systems in NRM.
A General Model for Spatial Data Management There are different organizations that are conducting their daily businesses, based on their defined missions and goals. If some parts of these activities relates to or affect natural resources (directly or indirectly), they would be a member of NRM organizations and should obey NRM policies and laws. Meanwhile, due to diversity of natural resources and diversity of activities that may affect natural resources, broad spectrums of organizations are involved
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
516
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
in NRM. In a country, these organizations are from local, provincial/state, and national levels, each of which are in charge of decision-making, planning, and managing specific/relevant natural resources at the relevant administrative/political level. Generally, operational activities are conducted at the lower levels and organizations at the higher level are in charge of policy making. Each country has it own mechanisms for coordination, supervision, and controlling natural resource related activities of NRM organizations. As mentioned earlier, NRM organizations and coordinating body need reliable and up-todate spatial data for proper decision-making, planning, and managing. Location, quantity, and quality of available natural resources, side effect of different activities on natural resources, current exploitation status of the available resources, plans for exploration and exploitation of natural resources, and forbidden zones for exploiting natural resources are some examples of required datasets for NRM. These datasets need to be regularly collected and updated to be used for decision-making and management process. NRM organizations must use these data in order to prepare and execute relevant plans and programs, by considering their effects (and side effects) on environment and whole natural resources. So, each NRM organization should have access to the data of other NRM organizations in addition to own natural resource datasets. Coordination body must also have access to all NRM datasets and use them for decision-making, coordination, supervision, monitoring and controlling purposes. However, because of the variety of required datasets for NRM, no individual NRM organization can collect and keep up-to-date all of NRM-related datasets by itself. Also, just one organization (being assigned as responsible for data collection) cannot collect and update all of the required datasets for all NRM organizations. Therefore, collecting and updating datasets for NRM should be done jointly, through a collaborative effort and partnership of organizations in spatial data collection/production and sharing. In this context, NRM organizations must be the main stakeholders for producing, updating and maintaining required spatial datasets. If each of the involved organizations collects some part of the required spatial datasets for NRM (relevant to its tasks) during everyday business, required spatial datasets can always be available to the organization. If this data is shared and exchanged, then datasets are accessible to the other NRM organizations. This collaborative environment is based on the concept of partnership in spatial data production and sharing. Although a partnership model for spatial data collection and sharing can resolve the problem with collection, access and dissemination of required spatial data for NRM, relevant research into collaborative efforts in spatial data production, sharing, and exchange shows that there are different technical (such as standards and interoperability models) and non-technical (such as social, cultural, and institutional) issues that create barriers for such participation (Nedovic-Budic and Pinto, 1999; McDougall et al., 2002; Rajabifard and Williamson, 2003; Mansourian et al., 2006). Therefore, by creating an environment in which such issues are taken into consideration and resolved, the concept of partnership in data production, sharing and exchange can become a reality. In this way, Spatial Data Infrastructure (SDI, as an initiative in spatial data management) with related concepts and models together with service based systems can be used as integrated frameworks for creating such an environment and consequently, facilitating NRM.
Using SDI and Service Based Systems to Facilitate Natural Resource…
517
Framework and Tools for Implementing the General Model This section describes the concept of SDI, web services technologies, geospatial web services and native XML databases as the promising frameworks and tools that can resolve spatial data problems for NRM. At the next section, in the context of a case study, utilization of the mentioned frameworks and tools to facilitate NRM is depicted.
Role of SDI for NRM The growing need to organize data across different disciplines and organizations and also the need to create multi-participant, decision-supported environments has resulted in the concept of Spatial Data Infrastructure (SDI). SDI is an initiative intended to create an environment that will enable a wide variety of users to access, retrieve and disseminate spatial data in an easy and secure way. In principle, SDI allow the sharing of data, which is extremely useful, as it enables users to save resources, time and effort when trying to acquire new datasets by avoiding duplication of expenses associated with generation and maintenance of data and their integration with other datasets. As illustrated in Figure 1, an SDI encompasses the policies, access networks and data handling facilities (based on the available technologies), standards, and human resources necessary for the effective collection, management, access, delivery and utilization of spatial data for a specific jurisdiction or community. Dynamic
Access Network
PEOPLE
Policy
Data
Standard
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 1. SDI Core Components (Rajabifard et al., 2002)
Viewing the core components of SDI, different categories of components can be formed based on the different nature of their interactions within the SDI framework (Rajabifard et al., 2002). Considering the important and fundamental role between people and data as one category, a second category can be considered to consist of the main technological components: the access networks, policy and standards. The nature of these two categories is very dynamic due to the changes occurring in communities (people) and their needs, as well as their ongoing requirement for different sets of data. Additionally, with the rapid development of technology, the need for the mediation of rights, restrictions and responsibilities between people and data are also constantly subject to change. According to this view, anyone (data users through to producers) wishing to access datasets must utilize the technological components. SDI is also an integrated, multi-leveled hierarchy of interconnected SDIs based on collaboration and partnerships among different stakeholders, which is known as SDI hierarchy (Fig. 2).
518
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
Less Detailed Data Global Planning
Global SDI
Regional Planning
Regional SDI
National Planning
National SDI
State Planning
State SDI
Local Planning
Local SDI
Organizational Planning
Organizational SDI More Detailed Data
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 2. SDI Hierarchy (Rajabifard et al., 2003)
There are two views on nature of SDIs, which helps in better describing and understanding of SDI hierarchy (Rajabifard et al., 2003). The first view is an Umbrella View, which describes the SDI at a higher level, say the national level, encompasses all the components of SDIs at levels below. This suggests that ideally at a national level, the necessary institutional framework, technical standards, access network and people are in place to support sharing of fundamental spatial datasets at lower levels, such as state/provincial level. The second view is the Building Block View. According to this view, any level of SDI, say state/provincial level, serves as the building blocks supporting the provision of spatial data needed by SDIs at higher level in the hierarchy such as national and regional levels. This describes the aggregation property of SDI hierarchy. Based on these views, the SDI hierarchy creates an environment in which decision-makers working at any level can draw on data from other levels, depending on the themes, scales, currency and coverage of the data needed. With this in mind, due to the need for collaboration in spatial data production and sharing for NRM, an SDI can be used as an appropriate framework to facilitate NRM. This is due to the fact that SDI can define the relationship between people and data (Fig. 1) and the relation between organizations at different political/administrative levels and hence can create an environment in which people can access, retrieve, and disseminate data, based on SDI’s core component and relevant models. By clarifying each of the core components of SDI, an SDI conceptual model can be developed which can contribute to facilitating the availability, access and usage of spatial data for NRM and hence facilitation of NRM.
Role of Web Services Technologies for NRM The Web as it exists today is intended for human consumption. Consequently, data is presented in a form that is human-readable, but this form of representation is error prone and difficult for applications to examine, extract and use both automatically and programmatically. So there is a need for application to application communication and this is the idea of application centric web rather than human centric web (the web as it works today).
Using SDI and Service Based Systems to Facilitate Natural Resource…
519
Automatic and direct communications among functional units which are running on heterogeneous platforms are the unique characteristics of the application centric web (as the next generation of web). The promising technologies for this kind of communications are web services technologies (Newcomer, 2002). In fact, web services technologies are implementation of a conceptual architecture, which is called Service Oriented Architecture (SOA). Some IT experts state that SOA is the next generation of software development model which is emerged after distributed technologies paradigm (Marks and Werrell, 2003). SOA is a conceptual architecture for implementing services with characteristics (i) loosely coupled; (ii) self-describing; (iii) standard-based and (iv) dynamic discoverable (Newcomer and Lomow, 2005). In SOA, the central elements are services. A service is a location on the network that has a machine-readable description of the messages it can receive and reply. As illustrated in Figure 3, SOA consists of three primary roles and three primary tasks (i) service provider; (ii) service requester and (iii) service broker. These are distributed computational nodes in the network environment. Service provider publishes its own service with service broker. Service requester uses the service broker to find desirable services and then binds to a service provider to invoke the service.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 3. Major components of Service Oriented Architecture (SOA)
The actual implementation of SOA using open, standard and widely used protocols and technologies is called web services (Marks and Werrell, 2003). In more technical point of view, web services are about delivering distributed applications via programmable URL (Uniform Resource Locators) (Amirian and Mansourian, 2006). In this case, using an XML document in the form of a message, an application sends a request to a web service across the network, and receives a reply also in the form of an XML document. Web services are based on open standards, so they can provide interoperability in decentralized and distributed environments like web. These new technologies can be developed by using any software platform, operating system, programming language and object model. More precisely, web services are loosely coupled, self-describing services that are accessed programmatically across a distributed network, and exchange data using vendor, platform, and language neutral protocols (Marks and Werrell, 2003). These are implemented by using a collection of standards and technologies. These standards and technologies when
520
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
considered together, establish what is widely referred to as the web services protocol stack. Figure 4 illustrates the eight distinct layers of the Web services protocol stack (Gailey, 2004).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 4. Web services protocol stack
These eight layers are grouped into three distinct levels (Fig. 3); each level indicates a level of maturity for the layers it contains. The enabling standards level contains two layers: the network transport protocols and meta-language. The layers within the enabling standards level contain well-defined and accepted standards and protocols that are widely used in internet and web such as HTTP and XML. The evolving standards level contains layers for SOAP (Simple Object Access Protocol) and WSDL (Web Service Description Language) and UDDI (Universal Description, Discovery and Integration). Collectively, these layers form the core technologies for implementing web services. SOAP is a lightweight, XML based protocol for exchanging information in decentralized, distributed environments. It is used for messaging among various SOA’s components in a web services platform. It is also platform independent and can be used virtually with any network transport protocols such as FTP, HTTP, HTTP-S and HTTP-R. WSDL is XML-based specification for describing the capabilities of a service in a standard and extensible manner. Technically WSDL defines the software interface of web service in platform independent approach. UDDI is a set of specifications and APIs (Application Programming Interfaces) for registering, finding and discovering services. These three layers establish an explicit mapping between elements of SOA as a conceptual and technology independent architecture and web services as specific collection of standards, protocols and technologies. In this context, web service provider publish its own service description using WSDL, then web service requester take advantage of search API’s (Application Programming Interface) of UDDI to find appropriate web services and finally, Web service requester bind to the service provider using SOAP. The last level of standards or emerging standards level represents proposed standards that are promoted by individual vendors (such as Microsoft, IBM and Sun Microsystems). This level consists of specifications which have not yet gained broader endorsement or acceptance in the wider web services community, and have not been adopted as open standards for development by key standards bodies such as the W3C (World Wide Web Consortium.
Using SDI and Service Based Systems to Facilitate Natural Resource…
521
http://www.w3c.org) and OASIS (Organization for the Advancement of Structured Information Standards. http://www.oasis-open.org). System heterogeneity would be one of the main problems of NRM organizations for data sharing. From an IT perspective, web services technologies can be an appropriate solution for this problem. Because, systems that are developed using web services technologies can easily communicate with each other. Therefore, utilization of these technologies can also provide access interoperability among NRM organizations for data and decision sharing.
Role of Geospatial Web Services for NRM Spatial data has their own characteristics and specifications from format, data structure, etc. perspectives. As a result data interoperability would be another problem of NRM organizations for sharing spatial data. Nowadays, geospatial web services have been considered as the promising technology to overcome the non-interoperability problem associated with current spatial processing systems. They are particular kind of online services which deal with spatial data. Geospatial web services can do the following (Lake et al., 2004):
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
• • •
Provide access to spatial information stored in database; Perform spatial analysis; and Return messages that contain spatial information, which can be delivered as image, text, numeric data or geographic features.
In this context, OGC has defined a comprehensive framework of geospatial web services which is known as OGC web service framework. OGC web service framework consists of interface implementation specification and encodings which are openly available to be implemented by developers. The interface implementation specifications are software technology neutral details about various operations of each geospatial web service. The encodings provide the standard glue among different parts of geospatial web services. Each service of this framework can be implemented using various software technologies and systems. The most fundamental services and encodings of the OGC web service framework are Web Map Service (WMS), Web Feature Service (WFS), Geography Markup Language (GML), Common Query Language (CQL) etc. In short, WMS intends to share and request vector and raster geospatial data in plain image format. WFS is the main geospatial web service for publishing and requesting vector geospatial data in GML format. WMS and WFS and other geospatial web services supply standard access to geospatial data thus provide access interoperability in spatial information community. According to GML specification (http://portal.opengeospatial.org/ files/?artifact_id=4700), GML is an XML grammar written in XML Schema for modeling, transporting, and storing spatial data. GML provides data interoperability among heterogeneous spatial processing systems. CQL is XML-based format for defining queries on spatial data. This language provides standard grammar for making query statements, independent of any software system. Geospatial web services can be considered as technology independent services. A common misunderstanding about the geospatial web services is that they are exactly the same as web services. However, geospatial web service differs from the web service. The most
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
522
important distinction between these two kinds of services is the fact that web services are particular set of technologies and protocols but geospatial web services are composed of defined set of interface implementation specifications which can be implemented with diverse technologies. With respect to the above descriptions, developing geospatial web services using web services technologies can provide data and access interoperability among NRM organizations.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Role of XML Database Systems for NRM XML, as a rich set of technologies, is playing an important and increasing role in share and exchange of data over the web. The more XML has been used in share and exchange of data, the more XML data management issues have to be considered. So, database researchers have actively participated in developing technologies centered on XML data management, in particular data models and query languages for XML. As a result of these researches, many XML data management systems have been implemented. In general, XML data management systems can be categorized as XML enabled databases and native XML databases (Chaudhri et al., 2003). An XML enabled database stores XML data in its internal data models while allows retrieval of stored data as XML documents or fragments through some proprietary methods. Typically, an XML enabled database is a relational database which provides storage of hierarchical XML documents in relational model. In addition, it provides proprietary methods for relational to hierarchical data mapping (or conversion) for retrieving stored data as XML. The mentioned proprietary methods vary from extension to standard SQL language to implementation of a full featured XML query language. On the other hand, a native XML database has an XML document as its fundamental unit of logical storage, just as a relational database has a row in a relation as its fundamental unit of logical storage. A native XML database defines a logical model for its fundamental unit of storage and stores and retrieves XML documents according to that model (eg. the XPath data model, the XML Infoset, and the models implied by the Document Object Model (DOM)) (XML db, http://xmldb-org.sourceforge.net). Chaudhri et al. (2003) conducted a comparative performance and scalability analysis between the two technologies for XML data management. This test included: • • • • •
evaluating insert, update and delete operations performances; measuring the running time in searching for record by using index key; measuring the running time for bulk load, bulk mass delete and bulk mass update; evaluating the efficiency in executing a complicated query using regular expression and a complicated query using join (one-to-many relationship); and investigate the performance of more complicated queries that combine selection and sorting
After analyzing the results of the above tests, Chaudhri et al. (2003) concluded that a native XML database has better performance and scalability than a XML enabled database for handling XML documents with larger data sizes. In other words, a XML-enabled database has better performance for small document sizes (number of records less than 1,000) but it cannot handle large-sized documents as efficiently due to conversion overhead. The native
Using SDI and Service Based Systems to Facilitate Natural Resource…
523
XML database engine directly accesses XML data without conversion (Chaudhri et al., 2003). As mentioned earlier, GML is an XML based technology for modeling, transporting, and storing spatial data. Since GML is based on XML, the same technology for managing XML data can be used to manage spatial data stored in GML. As a result, using native XML databases for storing spatial data (as GML), provides an efficient solution for storing and accessing high volume spatial data in multi-user enterprise environments. In addition, as more and more data is stored and exchanged using GML format, by using XML technologies which are easy to integrate with native XML databases more spatial capabilities can be added to native XML databases. Adding spatial capabilities to native XML databases allow them to manipulate spatial data thus make them spatial databases. Considering the above descriptions, coupling native XML database systems, as efficient means for storing and managing spatial data (in GML format), with geospatial Web services that are developed using Web services technologies, can provide an appropriate open and interoperable spatial data sharing mechanism for NRM in an SDI environment.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Case Study - Implementation As part of the research, a case study has been designed and conducted in Iran, to investigate the role of SDI and service-based systems for NRM. Iran extends over a total area of 1,648,000 sq km, in middle-east region. The country lies between 44.5° and 63° in longitude and between 24.5°and 40° in latitude. Preliminary results of 2007 census estimates its total population as 71 millions. Iran is a country which has different natural resources including oil and natural gas, various mines and ores, forests, agriculture, different types of plants, animal types, water (rivers, underground waters, lakes, sea and golf), fisheries, deserts (i.e. sources of wind and solar energies) etc. Ministry of Agricultural Jihad and its branches at the provincial and local levels are the organizations that take more concern about natural resources in Iran, with emphasis on forests, watersheds, pasture, agriculture, and deserts. Ministry of Energy (with emphasis on water resources) and Iranian Department of Environment are also other organizations that have interest towards NRM in their activities. Although considerable effort has been undertaken for NRM in Iran, many activities still remain to be done particularly with respect to (i) assessing available natural resources; (ii) preparing an strategic plan for integrated management of natural resources; (iii) investigating and monitoring side effects of different activities (e.g. factory construction, sewage repelling, etc.) on natural resources; (iv) executing comprehensive supervision and controlling programs; and (v) effective and efficient use of new technologies, tools, innovations, and concepts to improve NRM" in Iran. Spatial data can considerably improve NRM; however initial studies show that they are not yet effectively and efficiently used. Such situation causes decision-making, planning, and coordination of activities to be poor and improper without paying attention to all factors that must be considered for a sustainable NRM. In this respect and based on what has been descried earlier, current research is being conducted to investigate the role of SDI and service based systems in NRM. Two important outputs through this research are:
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
524 •
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
•
an SDI conceptual model as a framework to facilitate the development of an infrastructure for NRM; and a geospatial Web Service developed by Web services technologies as a tool for data sharing and data exchange using an SDI conceptual model.
The first output is to create an environment in which spatial data can be more available and accessible and used more appropriately to facilitate decision-making, planning, and coordination. The second output is to create relevant software and tools for accessing and retrieving data. As described before, developing SDI for NRM requires different technical and non-technical factors to be met, which creates a multidisciplinary environment for research. In this regard, a seven step case study approach has been taken for this research. At the first step a literature review was conducted on theories relevant to the research with emphasis on NRM, SDI, spatial information sciences, Information Communication Technology (ICT), and organizational behavior. With this background, at the second step, the NRM of Iran and relevant organizations were assessed with respect to spatial data from an SDI perspective. In this assessment, based on current SDI models, factors that influence spatial data sharing, and factors that influence participation of organizations in SDI development (Kevaney, 1995; McDougall et al., 2002; Rajabifard and Williamson, 2003; Mansourian et al., 2006), the NRM communities of Iran were evaluated. As a result of the assessment, an initial conceptual model for SDI was developed at the third step. The conceptual model included establishing a data framework; developing standards for data collection, storing, and sharing; developing access network specifications; policy considerations; and identifying responsible organizations for data collection and updating for NRM. According to the results of the assessment and initial SDI conceptual model, at the fourth step, a prototype geospatial web service was developed using web services technologies and native XML databases, as a tool for data storage, access, and sharing and consequently facilitation of the decision-making, based on the SDI concept. The developed service-based system was installed on different computers, each of which presenting a data custodian geospatial web service. At the fifth step, a small prototype GIS based applications for NRM were also developed. This system could integrate spatial data retrieved from multiple services to make comprehensive, integrated, and coordinated decisions and plans. To test and evaluate the developed SDI conceptual model and the prototype service-based system, they were demonstrated to some of the NRM experts (the sixth step). Prior to the demonstration, required spatial datasets (sample data) were collected from different organizations. They were then prepared based on developed standards and stored in different native XML databases corresponding to spatial data custodians services. During the demonstration, it was supposed that different custodians produce, update, and store their own spatial datasets and share them with the NRM organizations. Therefore, within an SDI environment and using service-based systems, each organization had access to its required spatial datasets to integrate and analyze those, using GIS functionalities to support its own decisions and planning. With this in mind, it was found that such a service-based system that utilizes SDI can facilitate and improve NRM, by having all information available and accessible for planners and decision-makers. Based on the results of the demonstration and the initial SDI conceptual model, the participants could gain better understanding on the concept of SDI and service-based systems.
Using SDI and Service Based Systems to Facilitate Natural Resource…
525
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Therefore, they could share better information about the status of NRM from technical and non-technical perspectives (that affect participation of organizations in SDI initiative) with the authors. As a result of that, the initial SDI conceptual model together with the prototype system were updated (the seventh step) and was found to be an appropriate framework for NRM. The organizational assessment, SDI conceptual model and prototype service-based systems are described below.
Organizational Assessment Considering governmental system of Iran, most of the NRM activities are conducted at the provincial level. However, existing mechanisms for planning, executing and coordinating NRM related activities are traditional and very poor. Additionally, as highlighted earlier, spatial information and related technologies are not effectively used in NRM processes. In this paper, the NRM organizations and relevant committees are named as NRM community. In order to assess NRM in Iran, three provinces were selected as the case study and assessed with respect to spatial data from a SDI perspective. The basic organizational behavior model (Robbins et al., 1994) was recognized as an appropriate framework and was utilized for organizational assessment. According to basic organizational behavior model, the NRM community consists of three tiers including (i) individual (people); (ii) group (involved NRM organizations); and (iii) organizational system (NRM community). Assessment was conducted at all three levels with respect to technical and non-technical variables that affect an organization at each level. The results of organizational assessment show that the NRM communities do not have a clear regime for partnerships in data production and sharing. In this respect, development of SDI for NRM is a matter of social, technical and technological, political, institutional and economical challenges. There are various social challenges that were identified as aspects that need to be emphasized in order to facilitate the SDI initiative. The first is the need to increase peoples awareness on the value of utilization of spatial data and related technologies (particularly GIS) for a NRM community at policy, management and operational level (based on organizational pyramid model: Petch and Reeve, 1999). Other challenges include skill formation (working with and interpreting GIS data) and cultural issues in the use and sharing of spatial data. Increasing the awareness is a key factor in developing SDI for NRM as it affects other factors in any three levels of NRM community (people as individual level, involved organizations as group level, and NRM community as organizational level). Having a clear strategic plan with respect to spatial data for NRM is essential but currently neither exists at the group level nor at the organizational level of the NRM communities. Also, clear regulations and policies are required at both group and organizational level to clarify the custodianship of datasets, access policies, and dissemination policies in the context of copyright and privacy rules as well as security considerations. In addition, policies are required with respect to inclusion of private and academic sector and appropriate use of their capabilities in terms of spatial data. In terms of institutional challenges, four main areas were identified as needing emphasis: •
establishing a formal structure for spatial data affairs at group and organizational level of NRM communities;
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
526 •
• •
establishment of good relationships between NRM organizations for data sharing by appropriate agreements that respond to needs of organizations and increase their willingness to participate; establishment of appropriate relationships between governmental, private and academic sectors to better utilize their capabilities; and including those organizations and groups (e.g. national mapping agencies) that can provide required data for NRM.
With respect to technical and technological challenges, several areas were identified as needing emphasis including: •
• • •
Producing and collecting reliable and up-to-date spatial data required for comprehensive NRM is essential, since there are no appropriate framework data for NRM particularly with respect to availability and status of natural resources as well as side effects of different activities on natural resources; development of appropriate standards that support applicability of organizations’ datasets in GIS environment with respect to NRM requirements, providing the interoperability of systems of NRM organizations and integratability of their datasets with each other; and increasing the technological capabilities of NRM organizations particularly with respect to space technologies (GIS, GPS, remote sensing, thermal imagery, etc.), telemetry solutions (for real-time monitoring of some phenomena), networking, and communication facilities.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
With respect to political challenges many factors affect participation of organizations in SDI initiative such as the current political environment of the country, and NRM organizations; interest of national mapping agencies and the historical relationship between NRM organizations. Finally, financial resources are required as important support for an SDI initiative in the context of economic factors.
SDI - Conceptual Model As was described earlier the five core components of SDI establish the relation between people and data through technological components (standards, policy and access network) (Fig. 1). In light of standards and policy components, producers can produce data free of duplication of efforts and share them to be accessible and applicable for users (including value-adders and end-users). Value-adders can access and enrich data for end-users, other value-adders, and their own use and end-users can easily access and use data during their business. This is done through the access network component, which provides a physical environment for dissemination of data and access for use. The five core components of SDI and their relationships (Fig. 1) can also be regarded as a conceptual model as it describes a system in generic terms without reference to particular implementations (Davies, 2003; Mansourian, 2006). With this in mind, the SDI conceptual model for NRM was developed by expanding and clarifying each of the core components of SDI with respect to the results of the organizational
Using SDI and Service Based Systems to Facilitate Natural Resource…
527
assessment (Fig. 5). This SDI conceptual model is a framework that defines a clear regime for partnership of organizations in spatial data production and sharing. STANDARDS Semantic Interoperability
Syntactic Schematic
Mapping Agencies Data Provider
SDI Model
NRM Org.s Private Sector
Value-Adders
Private Sector
Policy for Access
NRM Org.s Coordinating body
End-User
DATA
POLICIES
PEOPLE
NRM Org.s
Product-Based Copyright Privacy Pricing Security
Standard Development
Policy for Standards
Mandate for Usage Facilitating the Usage
Forcing Mandates Data Custodianship Private Sector Role Academic Sector Role Financial Flow Type of Partnership
Coordinating Body
Secretary Institutional Arrangement Working Groups
SDI Organization
Other decision-makers Environment Preparation SDI Coordinating Committee
Cultural Political
Technological Economical Partnership Human Factors
Capacity Building
Coordination with NSDI and PSDI
Scale & Resolution Metadata Content Content
Capture Access and Analysis Tools Databases Management
Multi-Scale Spatial Res. Spectral Res. Radiometric Res.
Fundamental Data NRM Data GPS R.S Photogrammetry Telemetry Service-based Systems GIS Distributed DBs
ACESSING NETWORK Communicati on System Network Mechanism
Fiber Optics ADSL LAN WAN Internet Intranet
Figure 5. Schematic Presentation of SDI Conceptual Model for NRM
As Figure 5 shows, with respect to people, four categories were identified including data providers, value-adders, end-users, and SDI coordinating committee and for each category involved organizations and their responsibilities were clarified. In brief:
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
•
• •
•
data providers are NRM organizations as well as mapping agencies (national, provincial, and local). NRM organizations should produce, update and store spatial data describing current NRM situation, during their daily activities, and mapping agencies provide base datasets for NRM; value-adders are those organizations (generally from private sector) that get data from data providers and integrate them to prepare an value-added datasets; end-users are those organizations (generally NRM organizations and coordinating body) that use spatial datasets for decision-making, planning, and coordination of activities; and SDI coordinating committee consists of representatives form data providers, valueadders, and end-users. The committee is in charge of policy-making and coordinating SDI activities for NRM.
With respect to data, scale and resolution, content, capture (tools and mechanisms), access and analysis tools, database management, and metadata were identified as important
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
528
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
factors and each one has been clarified (Fig. 5). Multi-scale datasets are required for NRM due to the need for different details of information for different kinds of decisions. Satellite and aerial imageries with different spatial, spectral (e.g. Visible, thermal and infrared) and radiometric resolutions are required for collecting different kinds of data. Identifying required data for an integrated and comprehensive NRM is a very complicated task, which should be conducted regarding the content factor. With respect to database, designing and implementation of distributed database management systems have been recommended. Implementing service-based systems to facilitate access to available datasets (which are available in distributed databases) and practical utilization of GIS in decision-making, planning and coordinating activities is an essence and highly recommended. Production of metadata for datasets is essential in order to facilitate searching data and ensuring its quality and applicability for users. With respect to standard components, interoperability was identified as the important requirement. Interoperability is an important subject that needs to be emphasized in the context of standard component. Due to the collaboration of different organizations in data production, produced data should be easily applicable and integratable in value-adders' and end-users' systems. No additional work should be spent for data preparation (such as transferring format, structuring, and referencing data) to be applicable in their systems. Therefore, there should be no heterogeneity between data providers, value-adders and endusers systems. In this respect, there are three sources of heterogeneity that should be brought into consideration during standardization including semantic, syntactic and schematic heterogeneities. Semantic heterogeneity is relevant to differences in definition, primitives, structure, quality and coordinate system of data layers. Syntactic heterogeneity relates to difference in software, hardware, DBMS, and data format which is used by data provider and analyzer. Schematic heterogeneity relates to differences in data model, data coding, and topology. Utilization of standardization outcomes of international initiatives especially ISO, OGC, and W3C is recommended for achieving interoperability among NRM organizations systems. Meanwhile, using web services technologies and geospatial web services is recommended for resolving some parts of synthetic heterogeneities. Developing proper standards and specifications based on NRM requirements as well as using ontology concepts is also essential to achieve semantic and schematic interoperability. More specific, it is essential to develop standards and specifications for metadata storage and management as well as quality control and quality assurance of data and relevant metadata. Policy is a very important component as it can facilitate participation of organizations in SDI initiative. These are appropriate SDI policies that can create incentive and willingness in organizations and encourage them to keep their long-term partnerships. SDI development model, forcing mandates and approvals, institutional arrangements, policy for standards, policy for access, environment preparation, capacity building, SDI organization, and coordination with national/provincial SDI initiatives were identified as nine important requirements in the context of policies. As Figure 5 shows, regarding the SDI development model, a product based approach (Rajabifard et al., 2002) was selected for SDI development because of current technological capacity of NRM community and current status of required spatial data for NRM with respect to availability. In order for organizations to participate in the SDI initiative, a formal and forcing mandate is essential. However, besides formal mandate, policies are required to facilitate participation of organizations in data production and sharing by removing barriers and also increasing the incentive and willingness of
Using SDI and Service Based Systems to Facilitate Natural Resource…
529
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
organizations to form partnerships. In this respect policy for access, environment preparation, capacity building and institutional arrangements are four categories of required policies that can facilitate participation of organizations in SDI initiative. Clarifying and approving copyright and privacy laws and related concerns for spatial data are important tasks with respect to policy for access that should be considered. Clarifying data pricing and security concerns for data access are other factors in this category need to be emphasized. Capacity building should be conducted at all three levels of the NRM communities. In this respect, economical and technological capacity building at the group and organizational levels and skill formation and increasing awareness at the individual level are important subjects that should be emphasized. In addition there is a need to increase the capacity of partnerships between private, academic and governmental organizations. Cultural aspects of data sharing and providing an appropriate political environment are important tasks that should be noted in the context of environmental preparation. Institutional arrangements with respect to data custodianship, financial flow, type of partnership, and role of private and academic sectors need to be considered to facilitate partnerships by removing institutional barriers. Having policies with respect to the practical utilization of standards by data providers and value-adders in order to meet interoperability of systems and integration of datasets with accepted quality is another important item that should be noted. As mentioned earlier an appropriate organization for SDI is required to coordinate and follow SDI activities. A coordinating body, secretary and working groups are three tires which are recommended for this organization. The SDI coordinating body should have a peak position to be able to mandate relevant SDI approvals to other participants. In addition, policies are required that make organizations execute the mandates. The NRM SDI coordinating body should keep itself fully coordinated with provincial SDI and National SDI initiatives. Sharing data relies on the physical relocation of data from providers and value-adders to users. Two major factors that need to be considered with respect to the accessing network are communication system and network mechanism.
Prototype System, Based on SDI for NRM Figure 6 illustrates general structure of the developed prototype system. As Figure 6 shows each of data custodians, stores its relevant data in an XML database with GML format. Each organization has also WFS, connected to a XML database that provides the ability of disseminating spatial data at the feature level. There is also a catalog service which includes brief information about available datasets in XML databases (metadata of custodians). Users can search their required datasets through catalog service. After finding the spatial data, they can connect to the relevant WFSs (directly or via. catalog service) to retrieve their required datasets. With such a system, NRM organizations can easily retrieve all spatial data required for decision-making and planning activities. NRM coordinating body can also retrieve spatial data for supervision, coordination, and decision-making.
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
530
XML DB1 (GML
Web Feature Service (WFS)1
XML DB2 (GML
Web Feature Service (WFS)2
XML DBn (GML
Web Feature Service (WFS)n
Catalog Service
Client (User)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 6. General structure of the developed prototype system
Implementing WFS involves querying spatial data using CQL, delivering spatial data as GML and transporting spatial data over the Web. According to WFS Implementation Specification (OGC 2005, Open GIS Web Feature Service implementation specification 1.1.0. available at: https://portal.opengeospatial.org/files/?artifact_id=8339), WFS allows a client to retrieve spatial data encoded in GML from multiple Web Feature Services. There are three operations for basic WFS: (i) GetCapabilities - The purpose of this operation is to obtain service metadata, which is a machine readable (also human-readable) description of the required technical information for consuming WFS. The most important part of the service metadata (or capabilities document) indicates which feature types the WFS can provide and what operations are supported on each feature type; (ii) DescribeFeatureType WFS describes the structure or schema of any feature type it can provide using DescribeFeatureType operation. This structure will be used for retrieving spatial data; (iii) GetFeature - This operation is used for retrieving spatial data. Making use of CQL, spatial and non-spatial criteria can be introduced to retrieve the necessary GML data from WFS. These operations provide the software interface of the WFS system. In other words, internal details of the functional units and software components as well as communications are transparent to consumer applications; the consumer application just can communicate with the WFS system through the operations and defined set of parameters which are specified in WFS implementation specification. Software components, communication among them and physical location of each component are specified in logical and physical architecture of the WFS system. Physical architecture is quite different from a logical architecture. The physical architecture is about the number of machines or network hops involved in running the application. Rather, a logical architecture is all about separating different types of functionality in software components (Lhotka, 2006). Traditionally logical architecture of software applications consists of three tiers; (i) presentation and user interface tier; (ii) business logic tier; and (iii) data management tier (Barnaby, 2002). With the advent of new technologies and software design patterns the traditional logical architecture is rarely
Using SDI and Service Based Systems to Facilitate Natural Resource…
531
efficient for the modern software applications. Today, the business logic tier is often physically splits among a client, Web server and application server. In addition, with new software design patterns (such as façade, flyweight, adapter and composite) the business logic breaks up into multiple parts and components. In this research the WFS designed in four logical tiers (Fig. 7).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 7. Four tier logical architecture of the WFS
As the name implies, the presentation and user interface tier (Fig. 7) provides the end user an appropriate and friendly user interface which hides details of local and remote computational tasks from user. This tier is responsible for gathering the user inputs, validating the user inputs, composing CQL statements based on the user inputs to make requests, validating requests against proper schemas, sending validated requests to WFS server and displaying the returned geospatial data. The business logic tier includes all business rules for the WFS system. For the implemented WFS theses rules consist of translate requests to DBMS specific query language statements and dispatch them to the next tier. Data access tier (Fig. 7) interacts with the data management tier to retrieve, update, and remove information. It doesn't actually manage or store the data; it merely provides an interface between the business logic and the database. Logically, defining data access as a separate tier enforces a separation between the business logic and how application interacts with a database (or any other data source). This separation provides the flexibility to choose later whether to run the data access code on the same machine as the business logic, or on a separate machine. It also makes it much easier to change data sources or data access technologies without affecting the application. This is important because there may be a need to switch from one database vendor to another at some point. The forth tier (data management tier in Fig. 7) handles the physical retrieval, update, and deletion of data. This is different from the data access tier,
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
532
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
which requests the retrieval, update, and deletion of data. Often these operations are implemented within the context of a full fledged database management system. The mentioned four tier logical architecture have been developed using Microsoft .NET framework and SQL Server 2005 DBMS. .NET windows forms were used to implement the client side application (user interface and presentation tier). Windows forms provide much more flexibility and capability to use the client machine resources when compared with browser based applications. Web services infrastructure was utilized in all interactions between client side application and WFS server. In other words, WSDL was used to create proxy and skeleton in client side application and business objects of WFS server respectively and SOAP was used to transport every interaction (request and response) between the proxy and skeleton. In client side application, responses to these operations (client's requests) specify which feature types and attributes can be requested. Then CQL statements can be created by client using various logical and comparison operators which were provided as a part of user interface. The created CQL then is sent to the WFS server and the requested geospatial data is sent back to client side application. In order to allow user to visualize the GML data, a basic GML viewer was developed as part of Client side application using GDI+ (Graphics Device Interface+ is a part of .NET framework for 2D drawings). The GML viewer just represents the retrieved GML data and provides basic functionalities such as Zoom, Pan and Identify. In addition to developed client side application, any Web browser or application can consume the implemented WFS using WFS specifications. In order utilize retrieved datasets, users may use GIS software, which can process GML data directly, or convert GML data to the data format, which their GIS software accepts. Two famous software patterns were used for implementing business logic and data access components; façade and flyweight (Some authors refer to this pattern as Resource Pool pattern (Nock, 2003) or Object Pool pattern (Shalloway and Trott, 2004)) patterns respectively. Façade pattern provides a unified interface to a set of interfaces in a subsystem. Façade defines a higher-level interface that makes the subsystem easier to use (Gamma, 1995). A Façade pattern encapsulates a design feature where there is a simple interface that acts as a central point of reference for many interfaces (Horner, 2006). In general, façade pattern forces interfaces to communicate with use of chunk of data as method parameters. This way of communication ensures the minimal network roundtrips and thus increases the performance of the application drastically. Flyweight pattern is a pattern that allows client programmers to think that they are using a factory method to create their own object, when ‘their’ object is actually being shared by multiple clients. Normally, this is done to save memory and improve performance, by avoiding the creation of many equivalent objects (O’Docherty, 2005). In other words, the overhead required to continuously release and create any server-side resource that is frequently used and expensive to create for each client, limits the performance and reduces the maximum number of clients that can be served simultaneously; Flyweight pattern Manage the reuse of objects when it is either expensive to create an object or there is a limit on the number of objects of a particular type that can be created and thus resolves the mentioned problem (Shalloway and Trott, 2004). In the developed system, objects and components in business logic and data access tiers work together to prepare an appropriate response message. More accurately, user supplied parameters are parsed by business objects to determine which methods have to be executed. In the case of GetFeature operation, user
Using SDI and Service Based Systems to Facilitate Natural Resource…
533
supplied CQL statements are translated to appropriate XQuery statements. Then XQuery statements are delivered to objects and components in data access tier to be sent to DBMS. In the last tier of the architecture, geospatial data were stored as GML 3.1 in the back end native XML database. For retrieving geospatial data, XQuery statements which were sent by data access components are executed and result are sent back to the data access component. Data access components dispatch retrieved geospatial data to business logic components. Afterwards geospatial data are prepared to be valid against WFS specifications. Finally, prepared GML data are sent back to the client using Web services infrastructure (using SOAP). For implementation of the data management tier Microsoft SQL Server 2005 DBMS were used. Microsoft SQL Server provides the best performance and compatibility when .NET framework is used to develop the data driven applications. This DBMS is by far considered as one of the most powerful commercial relational databases. In contrast with earlier versions (such as SQL Server 2000), SQL Server 2005 defines a model based on SQL/XML (standard extensions to the SQL language specification) for an XML document and stores and retrieves documents according to that model (Amirian, 2006). As a result, SQL Server 2005 can be considered as a native XML database. The catalog service was also developed using Microsoft .Net tools.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
CONCLUSION This paper described a prototype service-based system as a tool and an SDI framework to facilitate NRM by providing a better way of spatial data collection, sharing, access, usage and management. The results of the case study showed how useful such a service-based system that works based on SDI can be for effective and efficient NRM. Having all information about the current situation of natural resources and different activities that affect them, how easily NRM coordinating body can plan for natural resources and control the situation. In addition each of the NRM organizations could improve their decisions and plans, having all of the information describing the current situation and limitations and also by modeling the future (using GIS). The main areas that this research can assist are relevant to availability, accessibility and interoperability of spatial datasets for NRM. Using an SDI conceptual model (which has been developed based on different technical and non-technical characteristics of NRM community), data custodians produce and store their relevant spatial data and share them to the wider NRM community. Data production and storage are based on SDI standards, so available data are easily applicable and integratable in users' systems. SDI policies, removes barriers and facilitates partnership of NRM organizations in NRM SDI initiative. So the spatial data problems can be better resolved. Using the developed service-based system, NRM organizations can access to required spatial data and use them in their planning and decisionmaking. On account of using Web services technologies, interoperability among heterogeneous platforms is achieved. Since Web services are the foundation of new type of application-to-application communication, they provide an unprecedented opportunity to connect heterogeneous platforms and applications. With the help of Web services technologies, it is an easy task for a developer to utilize WFS functionality into almost any
534
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
type of spatial processing system. Besides, spatial data and access interoperability is achieved through the use of standard interfaces and data format. Since native XML databases outperform other types of databases for storing and retrieving XML data, storing spatial data as GML in native XML database, retrieving spatial data from WFS no longer needs time and resource consuming process of data conversion.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REFERENCES Amirian, P. (2006). Design and development of a distributed geospatial web services using XML and .NET technologies. M.Sc. Thesis, K.N. Toosi University of technology, Tehran, Iran. Amirian, P. and Mansourian, A. (2006). Integration of GML and web services technologies for implementing distributed GIServices. In: Proceedings of GeoWeb 2006, British Columbia, Canada. Barnaby, T. (2002). Distributed .NET programming in visual basic .NET. Apress Publishing, California, USA. Chaudhri, A., Rashid, A., and Ziecari, R. (2003). XML data management: native XML and XML enabled database systems. Addison Wesley, Inc. Wokingham, England. Davies, J. (2003). Expanding spatial data infrastructure model to support spatial wireless applications. Ph.D. Dissertation, Department of Geomatics, The University of Melbourne, Melbourne, Australia, 210pp. EEA (2005). Sustainable use and management of natural resources. European Environment Agency. http://reports.eea.europa.eu/eea_report_2005_0/en/eea_report_9_2005.pdf accessed on Sep 2007. Gailey, J, H. (2004). Understanding web services specifications and the WSE. Washington, USA, Microsoft Press. Gamma, E., Helm, R., Johnson, R. and Vlissides, J. (1995). Design patterns: elements of reusable object-oriented software. Addison-Wesley, Boston. Hartanta Y. (2006). Strategy of Spatial Data Infrastructure (SDI) development in Indonesia as an anticipation action for future global change. M.Sc. Thesis, Pertanian Bogor. Horner, M. (2006). Pro .NET 2.0 Code and Design Standards in C#. Apress Publishing, California, USA. Kevany, M. (1995). A proposed structure for observing data sharing. In: Sharing Geographic Information (Eds. Onsrud, H.J. and Rushton, G.). Rutgers, New Brunswick, NJ 510pp. Lake, R., Burggraf, D., Trinic, M. and Rae, L. (2004). Geography Markup Language. John Wiley and Sons, Chichester, England. Lhotka, R. (2006). Expert C# 2005 Business Objects. second edition. APress Publishing, California, USA. Lojala, P. (2003). Classification of natural resources 2003 ECPR Joint Session of Workshops, Edinburgh, UK 28.3 – 2.4. www.essex.ac.uk/ECPR/events/jointsessions/paperarchive/ edinburgh/ws9/Lujala.pdf accessed on Sep. 2007. Mansourian, A., Rajabifard, A., Valadan Zoej, M. J. and Williamson, I. P. (2006). Using SDI and web-based systems to facilitate disaster management. Journal of Computers and GeoSciences 32(6): 303-315.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Using SDI and Service Based Systems to Facilitate Natural Resource…
535
Marks, E. and Werrell, M. (2003). Executive's guide to web services. John Wiley & Sons, Inc., New Jersey, USA, McDougall, K., Rajabifard, A. and Williamson, I. P. (2002). From little things big things grow: building the SDI from local government up. In: Proceedings Joint AURISA and Institution of Surveyors Conference, Adelaide, South Australia, 9pp. Nedovic-Budic, Z. and Pinto, J. K. (1999). Understanding inter-organizational GIS activities: a conceptual framework. Journal of Urban and Regional Information Systems Association 11(1): 53–64. Newcomer, E. and Lomow, G. (2005). Understanding SOA with web services. Addison Wesley, Inc., Maryland, USA, NLWRA (2002). Australia natural resource information. National Land and Water Resources Audit, http://audit.ea.gov.au/ANRA/data/docs/national/Data_Contents.html. Osundwa, J. (2001). The role of spatial information in natural resource management. In: International conference on spatial information for sustainable development, Nairobi, Kenya, 2–5 October 2001. O’Docherty, M. (2005). Object-oriented analysis and design: understanding system development with UML 2.0. John Wiley & Sons, Inc., New Jersey, USA. Petch, J. and Reeve, D. (1999). GIS Organizations and People: A Socio-Technical Approach. Taylor & Francis, London, UK, 214pp. Rajabifard, A., Feeney, M.E.F. and Williamson, I.P. (2002). Future directions for SDI development. International Journal of Applied Earth Observation Geoinformation 4 (1): 11–22. Rajabifard, A., Feeney M. E. F. and Williamson I. P. (2003). Spatial Data Infrastructure: Concept, nature and SDI hierarchy, in developing spatial data infrastructure: from concept to reality. (Eds Williamson, I. P., Rajabifard, A. and Feeney, M. E. F.), Taylor and Francis, London & New York. Rajabifard, A. and Williamson, I. P. (2003). Anticipating the cultural aspects of sharing for SDI development. In: Proceedings Spatial Science 2003 Conference, Canberra, Australia, 9pp. Robbins, S.P., Watters-Marsh, T., Cacioppe R. and Millett, B. (1994). Organizational behavior: concept, controversies and applications. Prentice Hall, New York 753pp. Schmidt-Vogt, D. and Shrestha R. P. (2006). Role of geospatial technologies in training for sustainable natural resources management in Asia. Geospatial Resource Portal, GIS Development, http://www.gisdevelopment.net/application/nrm/overview/over001_1.htm. Shalloway, A. and Trott, J. (2004). Design patterns explained a new perspective on objectoriented design. Second Edition. Addison Wesley, Inc., Maryland, USA, Taylor, B., McDonald, G., Heyenga, S., Hoverman, S., Smith, T., and Robinson, C. (2006). Evaluation of regional planning arrangements for natural resource management 2005– 06: Benchmark Report II, Healthy Savanna Planning Systems Project, Tropical Savannas Management CRC, Australia. World Bank (2000). Natural resources management. http://info.worldbank.org/etools/docs/ library/110135/nrm.
536
Ali Mansourian, Mohammad Taleai and Narges Babzadeh
QUESTION BANK Fill in the Blanks 1. Natural resources are frequently classified as ……….... and ……….... . 2. Based on ……..…and …….…. views, the SDI hierarchy creates an environment in which decision-makers working at any level can draw on data from other levels. 3. In Service Oriented Architecture (SOA), Service ………… uses the service ……… to find desirable services and then binds to a service ……… to invoke the service. 4. GML is an ……… grammar written in XML Schema for modeling, transporting, and storing spatial data. 5. According the proposed SDI Conceptual Model for NRM, with respect to standards component ………… was identified as the important requirement.
True and False 1. Just one organization should be responsible for collecting and updating needed data for NRM. 2. Natural resource management is a multi-disciplinary and multi-stakeholder activity. 3. Non-technical issues do not create barriers for participating in spatial data production and sharing. 4. Web services technologies are considered as a solution for system heterogeneity problems in data sharing activities. 5. According the proposed SDI Conceptual Model for NRM, NRM organizations and mapping agencies (national, provincial, and local) data providers or value adders.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Short Answer Questions 1. What is natural resource management? 2. What are three persistent concerns regarding natural resources in developing countries? 3. What is SDI definition? 4. What are SDI core components? 5. What is GML? 6. What are eight distinct levels of Web services protocol stack? Just mention the names. 7. Why XML databases are suitable for managing GML data? 8. Describe the role of SDI in NRM. 9. What are major operations of basic WFS? 10. What is difference between Web services technologies and Geospatial Web services technologies?
Using SDI and Service Based Systems to Facilitate Natural Resource…
Long Answer Questions
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. 2. 3. 4. 5.
Describe the role of spatial data in NRM. What are web services technologies? How they can facilitate NRM? What are geospatial web services? How they can facilitate NRM? Compare native XML databases and XML-enabled databases. Describe briefly, how SDI and service-based systems can facilitate NRM.
537
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 23
FUNDAMENTALS FOR USING GEOGRAPHIC INFORMATION SCIENCE TO MEASURE THE EFFECTIVENESS OF LAND CONSERVATION PROJECTS Robert Gilmore Pontius Jr1,*, Shaily Menon2, Joseph Duncan1 and Shalini Gupta1 1
School of Geography, Department of International Development, Clark University, Worcester, Massachussetts, USA 2 Department of Biology,Grand Valley State University, Allendale, Michigan, USA
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ABSTRACT This chapter describes a general approach to use Geographic Information Science to assess the effectiveness of conservation projects that are designed to prevent anthropogenic land development from threatening ecosystem services. We illustrate the approach with an application to measure the effect of land protection on the preservation of biodiversity in part of the Indo-Malayan realm. The approach requires maps that show: initial land cover, independent variables associated with the drivers of anthropogenic land cover change, protected areas, and suitability for conservation. The land change model Geomod produces a map of suitability for development, which is then used to produce maps of extrapolated land development under three scenarios: Baseline, Prevention, and Leakage. Maps of these three scenarios are combined with the map of suitability for conservation to measure effectiveness of protection. The approach examines the consequence of leakage, in which conservation at protected locations has the effect of shifting anthropogenic land development from protected locations to unprotected locations. If the shift is from places of higher suitability for conservation to places of lower suitability for conservation, then the protection has an overall positive net effect at preserving ecosystem services relative to the baseline. However, the results for this chapter’s application indicate that the effect of the protected areas is to shift development from places of lower biodiversity to places of higher biodiversity, because there are high biodiversity locations that have high suitability for development and are not protected. These results illustrate a situation where a conservation strategy can backfire when it *
Email: [email protected]
540
Robert Gilmore Pontius Jr.1, Shaily Menon, Joseph Duncan et al. aims to protect only the locations that are under threat and not the locations that maintain the most important ecosystem services. The methodology has been designed for use in a variety of contexts, specifically for policy applications that award credits for offset projects, for example carbon offset projects as called for by the Bali Roadmap for climate change.
Key Words: Conservation, GIS, land change, modeling, protected area
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
INTRODUCTION Some humans spend a tremendous amount of effort to change landscapes from a “natural” state to a “developed” state for a variety of desirable economic uses, such as urban, agriculture, transportation, and mining. Others spend a tremendous amount of effort to prevent such development in order to conserve the landscapes for a variety of important environmental uses, such as biodiversity maintenance, carbon storage, water filtration, and landslide prevention. It would be efficient in theory if a society were to focus its development efforts at locations that give the largest economic utility per area developed, and to focus its conservation efforts at locations that give the largest environmental utility per area conserved. However this is not necessarily the strategy of some important conservation policies. Some policy approaches, such as those proposed by the Clean Development Mechanism of the Kyoto Protocol on climate change and the subsequent Bali Roadmap, call for conservation on land that is under imminent threat of new development, not necessarily on land that gives the largest environmental utility (Sedjo et al., 1998; Clémençon, 2008). The apparent motivation to focus policy strategies on land under immediate threat is to prevent development before it exerts its environmental impact. This strategy is nearly a perfect equation for escalation of conflict, because it motivates conservationists to prevent the actions that are highest priority for developers. If conservation is effective in preventing development, then conservationists win and developers lose. If conservation is not effective in preventing development, then developers win and conservationists lose. A third plausible outcome of this policy strategy is that a conservation project might inspire developers to shift their future development from their first priority locations to their second priority locations. The process whereby conservation at one location causes development to shift from that location to another location is known as leakage. Leakage can undermine the overall effectiveness of a conservation project in terms of total environmental utility (Schwarze et al., 2002). This chapter presents a general conceptual framework to assess the effectiveness of land conservation projects by using Geographic Information Science (GIS) and land change modeling to analyze development and conservation in the presence of leakage.
PROBLEM IDENTIFICATION / CONCEPTUAL BACKGROUND There are two important challenges in the problem of measuring the effectiveness of a conservation project. The first is philosophical, the second is technological. The philosophical challenge is to design a procedure by which one can measure objectively the effectiveness of a land conservation project. Let us assume the land conservation project’s mechanism is to
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Fundamentals for using Geographic Information Science …
541
place a legal restriction on a patch of undeveloped land in order to prevent the threat posed by future development. How can we measure whether the conservation project is effective in its overall goal to maintain the ecosystem services of undeveloped land? We need to consider two components. First, we need to assess the threat by development to that particular patch of undeveloped land. This can be done by envisioning a baseline scenario which portrays the development that would occur if the conservation project did not exist. Second, we must envision the influence of the conservation project on the actions of developers. If the conservation project prevents developers from their intended development on that patch, then the conservation project is successful in its goal to maintain ecosystem services on the protected patch. However, the conservation project might cause the developers to develop other parcels of land beyond the protected patch, through a process of leakage. We must consider this leakage when assessing the overall effectiveness of the conservation project. The project’s overall effectiveness is the difference between the baseline scenario that lacks a conservation project and a scenario that has both the conservation project and leakage. This puts us in a challenging situation whereby we must design and assess a counterfactual scenario. The baseline scenario without the conservation project is counterfactual in the sense that it portrays how the developers would behave if there were no conservation project, while in fact there is a conservation project. Therefore, it is difficult to devise an objective method to assess the accuracy of the baseline scenario (Oreskes et al., 1994). Furthermore, we must also assess how the conservation project influences the developers’ differential impacts beyond the area of the conservation project. This is something that would probably be difficult for the developers themselves to quantify, due to the philosophical problem that it depends in part on a counterfactual baseline scenario. Some have proposed to address this problem by using GIS and land change modeling to generate the scenarios (Kerr et al., 2003). Even if the philosophical challenges can be overcome, there are technological challenges to using GIS and land change modeling to measure the effectiveness of conservation projects. GIS-based modeling techniques can be used to produce a map of the counterfactual baseline scenario to portray how the landscape would appear if there were no conservation project. GIS can be used also to generate a map to portray the landscape under the assumption that the conservation project completely prevents the hypothetical baseline development. Additional land change modeling could generate a map that portrays the landscape under a leakage scenario that assumes the conservation project causes developers to shift their development beyond the conservation project area. If scientists could produce all of these maps accurately, then they could measure the effectiveness of the conservation project by considering the differences among the maps. This would require a clear definition of development, a comprehensive understanding of the behavior of developers, a quantifiable definition of ecosystem services, a proper understanding of how development influences the environment, accurate data for all of the preceding, and a land change model that accurately predicts the behavior of developers under counterfactual scenarios. There exist many GIS-based models that are designed to generate maps of such land change scenarios. Whether there is any objectively measured justification to trust the usefulness of large complex models is another story (Lee, 1973). Herein lie many technological challenges. These philosophical and technological challenges are enormous. If scientists fail to address these challenges promptly, then important environmental initiatives will fail because implementation of major policy proposals requires these types of measures of effectiveness. For example, policies that encourage offset credit trading require a project to quantify its
542
Robert Gilmore Pontius Jr.1, Shaily Menon, Joseph Duncan et al.
effectiveness. If we wait for all the philosophical and technological challenges to be resolved perfectly, then it will be too late, because ecosystems around the world are already being degraded rapidly. Therefore, this chapter proposes a path forward to use the burgeoning fields of GIS and land change modeling to address the philosophical and technological challenges of measuring the effectiveness of efforts to protect the world’s natural resources. This paper illustrates the concepts with an application to biodiversity conservation in the Indo-Malayan region.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REVIEW OF LITERATURE For many years, scientists have been developing methods for land change modeling (Veldkamp and Lambin, 2001). There are a variety of approaches, each with its own data requirements. Such models typically predict a quantity of future anthropogenic development and then predict the spatial allocation of that development. It is possible to separate conceptually and mathematically the quantity of the area of future development from the spatial allocation of that development. For example, Pontius and Batchu (2003) and Pontius and Pacheco (2004) give methods to analyze the spatial allocation of development distinctly from the quantity of development for applications in the Western Ghats of India. Agrawal et al. (2002) review many different land change models and expose a large range of techniques. Pontius et al. (2008) performed a validation exercise that compared the accuracy of the output maps for 13 different modeling applications from 9 different models and found that 12 of the 13 cases contained more error than correctly predicted change at the fine resolution of the raw data. They also found that the accuracy of the output maps seemed to be influenced more by the selection of the study site and the format of the data, than by the modeling technique, so there is no guarantee that more complex models will necessarily have higher levels of accuracy. In fact, Pontius et al. (2007) found that a very simple model that allocated deforestation adjacent to the Transamazon Highway was more accurate on a pixelby-pixel basis than a more complex model that attempted to simulate the behavior of farmers in the Amazon. GIS and land change modeling have been applied to generate baselines for land use and carbon. Applications include the prediction of sequestration for carbon-offset markets (Pfaff et al. 2000) and the estimation of the potential for carbon mitigation by agriculture (Smith et al. 2000). Sathaye and Andrasko (2007) give several case studies of land change modeling applied to measuring the effectiveness of conservation projects and the related leakage in the context of carbon offset projects. These methods build on the framework of Aukland et al. (2003). McDonald et al. (2007) examine historical maps to reveal how the protection of land in the past has influenced subsequent protection and development of neighboring lands at three sites in the United States. Menon et al. (2001) show how land change modeling can be used to target locations for conservation with an application in northern India. A common complication is that there are usually many goals for any single conservation project. Stier and Seibert (2002) examine how efforts to manage carbon can be related to protection of biodiversity. Pearce and Perrings (1995) make explicit the link between development and biodiversity and call for a shift of focus from biodiversity as an asset
Fundamentals for using Geographic Information Science …
543
unrelated to development to one that values biodiversity as an integral part of the development process.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Study Area The study site for this chapter’s example is the part of the Indian subcontinent for which we have data. It is helpful to use available empirical data to illustrate the overall approach because presentation of the data can illustrate clearly both the strengths and weaknesses of the analysis. Regardless of the quality of the data, the purpose of this chapter is to illustrate the concepts of conservation, development, and leakage in a general framework. The reader is encouraged to consider how these concepts play out in the context of biodiversity in the IndoMalayan region and whether the available data are up to the task. For leakage analysis, selection of the extent of the study area is extremely important. The study area should be the union of the land that is originally targeted for development, the land that is intended to be conserved, and the land where leakage could possibly occur. Hence, the selection of the extent can be a challenge since it can be difficult to know where the leakage might occur, especially before one performs the analysis. The selection of study extent is easiest when: (i) there is exactly one specific threat of development; (ii) there is exactly one conservation project; and (iii) the possibility of leakage is confined to be close to both the development threat and the conservation project. These three conditions typically occur in the case where slash & burn farming by small scale farmers is the threat, and there is a single proposed conservation project designed to conserve the threatened forests. There is a distinct likelihood that the farmers will simply walk outside the conservation area to begin slash & burn farming in the forests that are adjacent to the conservation area. In this case, the areas of development, conservation, and leakage are contained within a small contiguous extent. This design is common for some types of projects implemented by non-governmental organizations such as Conservation International. However, if the process of development and conservation is much broader, then it can be more difficult to determine the appropriate study extent. For example, if the planned development is by multi-national corporations who plan to convert large segments of the Amazon forest to soy bean cultivation, then the conservation plan must be correspondingly larger, and the leakage could occur internationally. Conservation efforts in the Amazon could inspire agribusinesses to shift agricultural production from Brazil to Indonesia, in which case Indonesia would need to be included in the study area. One should select a study site that captures the entire phenomenon including leakage. Therefore, the selection of this chapter’s study extent is appropriate for the type of development that can shift within the Indian subcontinent.
Tools / Materials A spatial dataset for this analysis was compiled from maps of vegetation, protected areas, species richness, and elevation obtained from the World Wildlife Fund (WWF-US), Environmental System Resources Institute, World Conservation and Monitoring Centre, and the United States Geological Survey (online). The study area is constrained to be those parts
544
Robert Gilmore Pontius Jr.1, Shaily Menon, Joseph Duncan et al.
of the Indo-Malayan realm shown in the maps from MacKinnon (1996). All data layers are georegistered in GIS as raster images with a resolution of one square kilometer per pixel. The values in the elevation map are binned in intervals of 100 meters. The map of vegetation was reclassified to produce a map of cleared and non-cleared areas. Various vegetation categories, such as sub-tropical dry evergreen and evergreen forest, were reclassified as non-cleared areas and the remaining categories (cleared, barren and cultivated) were reclassified as cleared areas. The resulting map serves as the initial time land cover map in which the non-cleared pixels indicate locations that have the potential for future anthropogenic development. Figure 1 superimposes the map of protected status on the map of initial cleared versus non-cleared.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 1. Protected areas overlaid on the initial time land cover
We also produced a map of suitability for conservation by aggregating three species layers (plant, mammal, and bird richness) and by comparing the result with a map of current status of biodiversity in the Indian subcontinent. The various biodiversity layers, with values that show species richness and endemism, were created by WWF-US during their ecoregions project (Olson and Dinerstein 1998; 2001). The values for bird, mammal, and plant richness show the number of species found in the various regions. Z-scores were computed for each of the input maps and then weights of 0.2, 0.2 and 0.6 were assigned to bird richness, mammal richness and plant richness respectively to create a weighted average of z-scores. We then rescaled the resulting map such that each pixel expresses an index for biodiversity between 0 and 100. Figure 2 shows the resulting product that serves as the suitability for conservation map, which gives high indices at locations of relatively high biodiversity concentrations and lower indices at locations of lower biodiversity concentrations.
Fundamentals for using Geographic Information Science …
545
Figure 2. Suitability for conservation.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Methodology Figure 3 indicates the flow of information in the overall analysis. The land-use change model Geomod reads the initial time land cover map and factor maps, which are vegetation type and elevation for this case study. Geomod uses empirical analysis of these maps to generate a map of suitability for development, which shows the order of the priority of the locations for the simulation of additional future clearing. Figure 4 shows this map where the value of each pixel is an index ranging from 0 to 100, where the relatively higher values indicate a combination of vegetation and elevation that makes those pixels relatively more attractive for development, according to Geomod’s empirical analysis (Pontius et al., 2001). Geomod then use the suitability for development map to generate maps of future cleared status by searching among the non-cleared pixels for the largest suitability values and converting them from non-cleared to clear. All pixels that are cleared in the initial map remain cleared in the future, so Geomod simulates a one-way gain of cleared pixels as time progresses. Geomod’s first run produces a map in which half of the numbers of pixels that are non-cleared at the initial time become cleared in the future, where the selection of the location of the pixels is based exclusively on the suitability for development map. We select the quantity of future clearing as equal to half of the initial time non-cleared pixels to illustrate the procedure. The time at which half the existing land cover would be cleared depends on the assumed rate of future development, and we do not specify the exact time in the future that is portrayed by the map of the scenario. Figure 5 gives the map of the future cleared status, which is called the baseline scenario.
546
Robert Gilmore Pontius Jr.1, Shaily Menon, Joseph Duncan et al.
Factor 1 Factor 2
Land Use Change Model
Baseline Scenario Conservation Protection
Prevention Scenario
Initial Time Land Cover Suitability for Development
Leakage Scenario
Utility Assessment Biodiversity 1
Suitability for Conservation
Biodiversity 2 Biodiversity 3
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 3. Flow of information in the analysis. Solid rectangles represent maps and bold italics specify input maps. Dashed rectangles indicate analytical procedures that are not maps.
This baseline scenario map is then overlaid with the map of protected areas to produce a map called the prevention scenario. The prevention scenario portrays the case where the protected status is perfectly effective at preventing the new clearing in the baseline scenario at those locations that are protected. This prevention scenario is shown in a map where each pixel is assigned a category of cleared or non-cleared, but it has fewer newly cleared pixels than the baseline scenario, because newly cleared pixels in the baseline scenario that are also protected are reclassified as non-cleared in the prevention scenario. Alas, we suspect that the protection status would not simply eliminate future development, since the effect of the protection is likely to displace future development from protected locations to unprotected locations, through a process of leakage. Therefore, we design a leakage scenario by first counting the difference in the number of newly cleared pixels between the baseline scenario and the prevention scenario. Then we modify the map of the prevention scenario to allocate this number of pixels of new clearing at unprotected locations that have the largest available suitability for development values. Hence the baseline scenario and the leakage scenario have the same quantity of newly cleared pixels, but some of the pixels are in different locations due to leakage. Figure 6 shows these shifts in spatial allocation that are associated with leakage.
Fundamentals for using Geographic Information Science …
547
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 4. Suitability for development.
Figure 5. Baseline scenario where future clearing occurs at locations of high suitability for development regardless of protected status.
548
Robert Gilmore Pontius Jr.1, Shaily Menon, Joseph Duncan et al.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 6. Effects of conservation and leakage where baseline clearing inside protected areas (gray) is displaced to locations outside protected areas (black)
If these shifts in spatial allocation cause the newly cleared pixels to move from a place of relatively high biodiversity to a place of lower biodiversity, then the consequences for conservation are positive relative to the baseline, in spite of the fact that the protection did not reduce the total quantity of area of new clearing. However, if this shift in location causes the newly cleared pixels to move from locations of lower biodiversity to locations of higher biodiversity, then the consequences for conservation are negative. In order to measure this effect, we overlay a map of suitability for conservation on each of the four land cover maps, i.e. the initial time and the three future scenarios. Figure 7 gives the theoretical framework to quantify the information in the four land cover maps. The vertical axis shows the utility for conservation, which is conceptualized as the value of the ecosystem services that derive from the non-cleared areas; it is computed as the sum of the suitability for conservation values in the pixels that are non-cleared in a map of land cover. The horizontal axis shows the utility for development, which is conceptualized as the value of the economic services that derive from the cleared areas; it is computed as the sum of the suitability for development values in the pixels that are cleared in a map of land cover. There are four land cover maps and each map corresponds to a single point in the space, with the map of the initial time residing at the dot in the upper left of Figure 7 and the maps of the three future scenarios at the arrow heads below and to the right of the initial dot. These points are important in their position relative to each other, and not necessarily in terms of particular numbers along the axes for this particular application.
Fundamentals for using Geographic Information Science …
549
initial time
baseline scenario
Dodge
Leakage
Intended Benefit
leakage scenario
Gain
Utility for Conservation
prevention scenario
Loss
Posed Threat
Utility for Development
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure7. Theoretical intended arrangement of points
It is helpful to describe each of the points in sequence. The point of the initial time shows the overall utility of the initial landscape in terms of both conservation and development. All three future scenarios are to the right and below the initial time point because the scenarios portray landscapes that demonstrate an increase in future cleared land due to human development, so they necessarily have an increase in total utility for development and a decrease in total utility for conservation. The baseline scenario is farthest to the right because it places the new clearing at locations of highest suitability for development. Figure 7 shows the baseline scenario as lowest on the utility for conservation axis because it envisions that the baseline clearing occurs on land parcels that have relatively high values of utility for conservation. Figure 7 portrays a situation where some of the land cleared in the baseline scenario occurs on protected land. This portion of the baseline clearing is eliminated in the prevention scenario, which is the next scenario in the sequence. Consequently, the point for the prevention scenario is to the left and above the point for the baseline scenario, since the prevention scenario has less newly cleared land than the baseline scenario. Lastly we examine the position of the point for the leakage scenario relative to the other points. The quantity of newly cleared pixels in the leakage scenario is identical to the quantity in the baseline scenario but the spatial allocation of those cleared pixels is different, therefore the positions of the points in Figure 7 are different. Figure 7 portrays a situation where the overall effect of the protected network is to displace future clearing from locations of relatively high suitability for conservation to locations of lower suitability for conservation. The leakage scenario point is to the left of the baseline point because the displacement necessarily causes new clearing to shift from locations of high suitability for development to locations of lower suitability for development. The leakage point is above the baseline point when the displacement causes new clearing to occur on land that has lower suitability for conservation. The leakage scenario has more newly cleared land than the prevention scenario; consequently the leakage point is below and to the right of the point for the prevention scenario. It is helpful to assign names to the horizontal and vertical differences among the points. The vertical distance between the baseline point and the prevention point is the “intended benefit”. This is the amount of decrease in utility for conservation that the protection would
550
Robert Gilmore Pontius Jr.1, Shaily Menon, Joseph Duncan et al.
prevent, if its effect were to eliminate new clearing in protected areas. The vertical distance between the points of the prevention scenario and the leakage scenario is a measure of the decrease in total utility for conservation due to leakage from the prevention scenario. The vertical distance between the leakage point and the baseline point is the resulting overall combined effects of protection and leakage on utility for conservation from the baseline. If the leakage scenario point is above the baseline scenario, then the overall effect is positive, as portrayed as “gain” in Figure 7. Figure 7 expresses the gain in utility for conservation from the baseline as the intended benefit minus leakage. If the pixels of highest suitability for conservation are not protected, then it is possible in practice to have an unintended consequence that leakage causes clearing to shift from locations of lower suitability for conservation to locations of higher suitability for conservation, in which case the leakage point would be below the baseline point. Similar analysis of the differences among the points can be made in terms of the utility for development. The horizontal difference between the baseline and prevention points is the initially posed threat to development, since it represents a reduction in the projected future baseline of growth in development. Developers can attempt to avoid this threat by shifting future clearing to unprotected locations, in which case they would not necessarily suffer the entirety of the initially projected threat to utility for development. This shifting activity allows them to “dodge” the initial threat posed by the conservation project. To generate the leakage scenario, they shift their new clearing to their next best alternatives, so they are likely to recoup only partially the lost utility for development. Consequently, the horizontal difference between the leakage scenario and the baseline scenario is the effect of the protection in terms of loss in utility for development from the baseline. Figure 7 expresses the loss in utility for development from the baseline as the posed threat minus the dodge.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
RESULTS The results of our modeling application to biodiversity illustrate a situation in which the leakage causes a displacement of future clearing from the baseline locations to other locations of relatively higher suitability for conservation. Figure 6 shows how the leakage causes clearing to move from locations scattered about the subcontinent to locations clustered in three regions that Figure 2 shows as having very high suitability for conservation: Western Ghats, Western Himalaya, and Northeast India. Consequently, the resulting arrangement of points in Figure 8 is not consistent with the theoretical intended arrangement of points shown in Figure 7. Figure 8 shows that the resulting leakage point is below the baseline point. This portrays a situation where the efforts at protection have backfired, meaning that the combined effects of protection and leakage result in a landscape that has less utility for conservation than shown in the baseline scenario, since the leakage is larger than the intended benefit. At the same time, the leakage point is to the left of the baseline point, indicating that the combined effect of protection and leakage causes a loss in utility for development, since the protection prevents development from occurring at locations that have the highest suitability for development. Hence, the results portray a lose-lose situation, where the leakage point shows a lower utility for both conservation and for development relative to the baseline point.
Fundamentals for using Geographic Information Science …
551
initial time
baseline scenario leakage scenario
Dodge
Loss Intended Benefit Leakage
Utility for Conservation
prevention scenario
Loss
Posed Threat
Utility for Development
Figure8. Observed arrangement of points for the biodiversity case study where the only point that is different from Figure 7 is the leakage scenario point.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
DISCUSSION In our case study, we measured utility for conservation and utility for development as the sum of indices in pixels. This approach is somewhat simplistic, since it equates indices with utilities and assumes that the utility of the entire landscape is the sum of the indices within individual pixels. Future research should further develop these concepts in the context of specific applications. For example, if the concepts of this chapter are to be used to compute credits for offset trading, then the utility for conservation must be converted to monetary units. This is necessary for carbon dioxide offset projects that are designed to Reduce Emissions due to Deforestation and forest Degradation (REDD). For REDD projects, the suitability for conservation map would be a map of potential carbon emissions, where the utility for conservation in each pixel would be the mass of carbon dioxide emission equivalent that would be emitted from that pixel if it were to be cleared. For this case, the total amount of carbon emissions in the study extent is the sum of the carbon emissions from each pixel in the study extent. Pricing mechanisms have been designed to translate units of carbon into monetary units, so that offset credits could be awarded and traded. If both axes of Figure 7 have the same units, it may be reasonable to compare the cost and benefit tradeoffs between utility for conservation and utility for development. It can be more challenging to specify the units for conservation when the application is biodiversity. One reason is that the overall threat to biodiversity is not necessarily the sum of the threat in each pixel. If two pixels contain very high levels of biodiversity of the same species, then the elimination of one of those pixels does not necessarily cause the elimination of the entire species, if the other pixel is protected effectively and the species uses a spatial range that is smaller than the size of an individual pixel. However, if the species demands a range that is larger than the size of an individual pixel, then the development of one pixel can cause species loss in neighboring pixels, even when neighboring pixels are protected. In this manner, biodiversity can be influenced by the spatial pattern of development, for example by
552
Robert Gilmore Pontius Jr.1, Shaily Menon, Joseph Duncan et al.
whether large patches become fragmented, and not only by whether a particular location is developed. In many situations, there are various simultaneous goals for conservation, such as both carbon maintenance and biodiversity protection (Hardner et al., 2000; Hecht and Orlando, 1998). In these cases, the situation is even more complex.
CONCLUSIONS This paper offers some bricks in the conceptual foundation for using GIS and land change modeling to assess effectiveness of conservation. Our example of biodiversity protection illustrates how a conservation strategy to prevent developers from pursuing their first priorities can back fire and cause a lose-lose situation for both developers and conservationists when one considers the possible effect of leakage. If the land that has the largest value for conservation is protected regardless of the immediacy of the threat from development, then this lose-lose situation can be avoided. However, a policy of protection at only the locations with large conservation values may leave unprotected the locations where there is the largest immediate threat posed by development. It is quite challenging to quantify the effectiveness of conservation projects because there are enormous philosophical and technological challenges to produce accurate baseline scenarios and to track leakage. Scientists should dedicate energy to address these challenges, because scientists are being asked increasingly to use GIS and land change modeling to advise land policy in order to address some of the planet’s most pressing economic and environmental crises. It is necessary to measure the effectiveness of conservation in the manner outlined in this paper in order to implement international agreements for offset credit trading of ecosystem services. The earth’s natural ecosystem services are already disappearing rapidly, so we urgently need to make progress on these issues in a process of learning by doing, because if we delay conservation until all these challenges are completely resolved, it will be too late.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REFERENCES Agrawal, C., Green, G.M., Grove, J.M., Evans, T.P., and Schweik, C.M. (2002). A review and assessment of land-use change models: dynamics of space, time, and human choice. General Technical Report NE-297. Newton Square, PA: United States Department of Agriculture. Forest Service, Northeastern Research Station. 61 p. Aukland, L., Costa, P.M., and Brown, S. (2003). A conceptual framework and its application for addressing leakage: the case of avoided deforestation. Climate Policy 3(2): 123-136. Clémençon, R. (2008). The Bali Road Map: A first step on the difficult journey to a postKyoto protocol agreement. The Journal of Environment Development 17(1): 70-94. Hardner, J. J., Frumhoff, P. C., and Goetze, D.C. (2000). Prospects for mitigating carbon, conserving biodiversity, and promoting socioeconomic development objectives through the clean development mechanism. Mitigation and Adaptation Strategies for Global Change 5(1): 61-80.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Fundamentals for using Geographic Information Science …
553
Hecht, J.E. and Orlando, B. 1998. ELR DIALOGUES: Can the Kyoto Protocol Support Biodiversity Conservation? Legal and Financial Challenges. ELR: NEWS AND ANALYSIS.Vol. XXVIII, No. 9. Environmental Law Institute, Washington, DC. Kerr, S., Liu, S., Pfaff, A.S.P., and Hughes, R.F. (2003). Carbon dynamics and land-use choices: building a regional-scale multidisciplinary model. Journal of Environmental Management 69(1): 25-37. Lee Jr, D.B. (1973). Requiem for large-scale models. Journal of the American Institute of Planners 39(3):163-178. MacKinnon, J. (1996). Review of the Protected Areas System of the Indomalayan Realm: Remaining Habitat Map, based on 1980-1990's AVHRR and LANDSAT images. Scale 1:1,000,000. Retrieved from http://www.unep-wcmc.org/. McDonald, R. I., Yuan-Farrell, C., Fievet, C., Moeller, M., Kareiva, P, Foster, D., Gragson, T., Kinzig, A., Kuby, L., and Redman, C. (2007). Estimating the protected land on the development and conservation of their surroundings. Conservation Biology 21(6): 15261536. Menon, S., Pontius Jr, R.G., Rose, J., Khan, M.L., and Bawa, K.S. 2001. Identifying conservation priority areas in the tropics: a land-use change modeling approach. Conservation Biology 15(2): 501-512. Olson, D.M. and Dinerstein, E.T. 1998. The Global 200: A representation approach to conserving the earth’s most biologically valuable ecoregions. Conservation Biology 12(3): 502-515. Olson, D.M., Dinerstein, E., Wikramanayake, E.D., Burgess, N.D., Powell, G.V.N., Underwood, E.C., Amico, J.A., Itoua, I., Strand, H.E., Morrison, J.C., Loucks, C.J., Allnutt, T.F., Ricketts, T.H., Kura, Y., Lamoreux, J.F., Wettengel, W.W., Hedao, P., and Kassem, K.R. (2001). Terrestrial Ecoregions of the World: A New Map of Life on Earth. BioScience 51(11): 933-938. Oreskes, N., Shrader-Frechette, K., and Belitz, K. (1994). Verification, Validation, and Confirmation of Numerical Models of Earth Sciences. Science 263(5147): 641-646. Pearce, D.W. and Perrings, C.A. (1995). Biodiversity conservation and economic development: local and global dimensions. In: Biodiversity Conservation - Problems and Policies. Ed. Perrings, C. A., Maler, K. G., Folke, C., Holling, C. S., and Jansson, B. O. Kluwer Academic Publishers. pp. 23-44. Pfaff, A.S.P., Kerr, S., Hughes, R.F., Shuguang L., Sanchez-Azofeifa, G.A., Schimer, D., Tosi, D. and Watson, V. (2000). The Kyoto protocol and payments for tropical forest: An interdisciplinary method for estimating carbon-offset supply and increasing the feasibility of a carbon market under the CDM. Ecological Economics 35(2): 203-221. Pontius Jr, R.G. Boersma, W., Castella, J.-C., Clarke, K., de Nijs, T., Dietzel, C., Duan, Z., Fotsing, E., Goldstein, N., Kok, K., Koomen, E., Lippitt, C.D., McConnell, W., Mohd Sood, A., Pijanowski, B., Pithadia, S., Sweeney, S., Trung, T.N., Veldkamp, A.T., and Verburg, P.H. (2008). Comparing the input, output, and validation maps for several models of land change. Annals of Regional Science 42(1): 11-47. Pontius Jr, R. G., Walker, R., Yao-Kumah, R., Arima, E., Aldrich, S., Caldas, M., and Vergara, D. (2007). Accuracy assessment for a simulation model of Amazonian deforestation. Annals of the Association of American Geographers 97(4): 677-695. Pontius Jr, R.G., and Pacheco, P. (2004). Calibration and validation of a model of forest disturbance in the Western Ghats, India 1920 - 1990. GeoJournal 61(4): 325-334.
554
Robert Gilmore Pontius Jr.1, Shaily Menon, Joseph Duncan et al.
Pontius Jr, R.G., and Batchu, K. (2003). Using the relative operating characteristic to quantify certainty in prediction of location of land cover change in India. Transactions in GIS 7(4): 467-484. Pontius Jr, R.G., Cornell, J. and Hall, C. (2001). Modeling the spatial pattern of land-use change with GEOMOD2: application and validation for Costa Rica. Agriculture, Ecosystems & Environment 85(1-3): 191-203. Sathaye, J.A. and Andrasko, K. (2007). Special issue on estimation of baselines and leakage in carbon mitigation forestry projects. Mitigation and Adaptation Strategies for Global Change 12(6): 963-970. Sedjo, R.A., Sohngen, B., and Jagger, P. (1998). Carbon sinks in a post-Kyoto world. Internet edition. Resources for the Future Climate Issue Brief #12. Washington, DC. http://www.rff.org/rff/Documents/RFF-CCIB-12.pdf. Schwarze, R., Niles, J.O. and Olander, J. (2002). Understanding and managing leakage in the forest-based greenhouse-gas-mitigation project. Philosophical Transactions of the Royal Society London A 360(1797): 1685-1703. Smith, P., Powlson, D.S., Smith, J.U., Falloon, P. and Coleman, K. (2000). Meeting Europe’s climate change commitments: quantitative estimates of the potential for carbon mitigation by agriculture. Global Change Biology 6(5): 525-539. Stier, S.C. and Siebert, S.F. (2002). The Kyoto Protocol: an Opportunity for Biodiversity Restoration Forestry, Conservation Biology, 16(3): 575-576. United States Geological Survey. GTOPO30 DEM Digital Elevation Model. Retrieved from http://edc.usgs.gov/products/elevation/gtopo30/gtopo30.html Veldkamp, T.A. and Lambin, E. (2001). Predicting land-use change. Agriculture, Ecosystems and Environment 85(1-3): 1-6.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ACKNOWLEDGMENTS The theses of Clark University students Amy Nelson, Nick Pieri, and Patrick Morris helped to establish the intellectual foundation of this paper. We acknowledge the support of Professor Kamal Bawa, University of Massachusetts, Boston, especially in acquiring data. The Michigan Space Grant Consortium and Grand Valley State University’s Research and Development program contributed seed grants. The United States’ National Science Foundation supported this work through the Center for Integrated Study of the Human Dimensions of Global Change at Carnegie Mellon University. Clark Labs has made the building blocks of this analysis available in the GIS software Idrisi®.
QUESTION BANK True Or False 1. The baseline scenario is designed to portray a future in the absence of a conservation project.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Fundamentals for using Geographic Information Science …
555
2. The prevention scenario is designed to portray a future where a conservation project is successful at completely preventing the additional disturbance that the baseline scenario portrays. 3. The leakage scenario is designed to portray a future in which there are no conservation projects. 4. Leakage is the process whereby conservation at one location causes development to shift from that location to another location. 5. When there is zero leakage, then conservation efforts that were intended to prevent development will backfire, resulting in less utility of conservation than in the baseline scenario. 6. The suitability for development values of the pixels always indicate the monetary value that the developers could obtain if the locations at those pixels were to be developed. 7. The utility of development for any scenario’s map is computed as the sum of suitability for development values of the pixels that are cleared in that scenario’s map. 8. The utility of conservation for any scenario’s map is computed as the sum of suitability for conservation values of the pixels that are not cleared in that scenario’s map. 9. Kyoto Protocol on climate change and the subsequent Bali Roadmap call for conservation of land that has the largest density of carbon storage. 10. Land change models that simulate future anthropogenic disturbance predict a quantity of future development and then allocate the future development randomly in space. 11. There is evidence that land change models predict accurately because validation exercises show that most models produce more correctly predicted change than error at a fine spatial resolution. 12. For leakage analysis, selection of the extent of the study area has negligible influence on the results. 13. This chapter’s map of protected areas in the Indian subcontinent show that the protected areas are not on the pixels that have the largest values in this chapter’s map of suitability for conservation. 14. This chapter shows that the network of protected areas in the Indian subcontinent has been poorly designed. 15. This chapter shows how the present network of protected areas in the Indian subcontinent could backfire in terms of biodiversity protection if the network causes development to shift via a process of leakage from locations of lower biodiversity to locations of higher biodiversity. 16. This chapter tests how sensitive the results are to the accuracy of the land change model and finds that its conclusions are robust. 17. Geomod simulates both the transition from non-cleared to cleared due to development, and the reverse transition from cleared to non-cleared due to vegetation regrowth on abandoned land. 18. This chapter shows that it is possible to have a lose-lose situation where developers are denied their top priorities thus clear the land that has higher suitability for conservation.
556
Robert Gilmore Pontius Jr.1, Shaily Menon, Joseph Duncan et al. 19. This chapter describes how the quality of the input maps has been assessed and concludes that the maps are sufficiently accurate. 20. The foundations concerning the theory and practice of how to assess the effectiveness of conservation are sufficiently developed so that the only remaining challenge is to obtain high quality maps to perform the analysis.
Short Answer Questions 1. The intended benefit of a conservation project in terms of utility for conservation is the difference between which two scenarios? 2. For what types of conservation applications does it make most sense to compute the total utility of conservation as the sum of the individual pixel values of suitability for conservation? 3. For what types of applications does it make least sense to compute the total utility of conservation as the sum of the individual pixel values of suitability for conservation? 4. The concept of dodge in terms of utility of development is analogous to what concept in terms of utility of conservation? 5. How was the suitability for conservation map created in this chapter? 6. How was the suitability for development map created in this chapter? 7. What are the most important aspects of the relative positions of the points in figures 6 and 7? 8. What are the advantages and disadvantages of following a strategy to protect the locations that have the largest suitability for conservation, regardless of the threat posed by development? 9. What are the advantages and disadvantages of following a strategy to protect the locations that have the largest threat posed by development, regardless of suitability for conservation? 10. How can a conservation project that is intended to preserve ecosystems cause a loselose situation for both conservationists and developers?
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Long Answer Questions 1. Describe the range of applications in natural resources management for which the principles of this paper could be applied. 2. Describe the types of analysis that would be necessary to express both utility of development and utility of conservation in the same monetary units in order to examine the tradeoffs between development and conservation. 3. Describe how the extent of the study area can influence the results of the analysis of effectiveness. 4. Describe the likely strategies that a developer would use to circumvent a conservationists plan to prevent development.
Fundamentals for using Geographic Information Science …
557
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
5. Describe the processes by which decisions are made in practice concerning where to locate protected areas and how the methods of this chapter could influence those processes.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 24
VALIDATION OF GIS-PERFORMED ANALYSIS Montserrat Gómez Delgado* and Joaquín Bosque Sendra† Departamento de Geografía, Universidad de Alcalá. Calle Colegios, 2, 28801 Alcalá de Henares, España.
ABSTRACT
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The contribution of the combined use of the Geographical Information Systems (GIS) and spatial modelling to various land planning processes has been widely demonstrated. Nevertheless, in the execution of these models, methods and techniques, little attention has been paid to the robustness of the final results and the model itself. Traditionally evaluation has become more of a verification process (that is to say, a process which verifies the results, rather than a procedure to test its robustness), which is related exclusively to the problem of the error assessment. The aim of this methodology is to replace this process for a risk control procedure of the decision making based on an uncertainty analysis of the data and the model used. This procedure has been applied in two different types of spatial models. A normative model based on multicriteria evaluation for the disposal of a hazardous landfill and a prospective model of land use changes based on multi-agent systems.
Keywords: Agent based models, decision making, GIS, land use changes, multicriteria evaluation, uncertainty analysis.
INTRODUCTION Data used in GIS and their analysis are affected by uncertainty, which is understood to be any (known or not) error, ambiguity or variation in a database and in any analytical model (Eastman, 2001), and this lack of accuracy or reliability may invalidate the results of said * †
Email: [email protected] [email protected]
560
Montserrat Gómez Delgado and Joaquín Bosque Sendra
analysis and processes. GIS enclose many sources of uncertainty. First of all, data may present many different errors; moreover any analytical process can increase errors and uncertainty because it is not fully defined nor its error propagation range is known. Uncertainty usually generates distrust, so results of a GIS-performed analysis may seem to be inaccurate or weak. It is relevant, therefore, to have procedures available to support and strengthen statements derived from a GIS analysis. Until now, the most usual procedure was to deal with errors existing in initial data. Thus, user needs to know the type and quantity of errors in any data. In 1991, the International Cartographic Association created the Committee on Spatial Data Quality and this committee worked, among other tasks, to establish in a rigorous way the components involved in quality evaluation of geographical data (see Table 1). Table 1. Evaluation matrix for data quality Space
Time
Theme
Accuracy Completeness Logical Consistency Currency Resolution
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Source: adapted from Veregin and Hargitai, 1995.
Following this structure, the analysis of the error in a database should follow these steps viz., (i) Identification of source of error; (ii) error detection and measuring; (iii) modelling of error propagation; and (iv) proposals for error management and reduction strategies. Identifying errors in every geographical data dimension (space, theme and time) is not possible due to the different and separate sources of error: from measurement errors (due for instance, to tool malfunction) to conceptual errors (that is, errors appearing from translating onto map objects real- world entities that are almost impossible to measure), and to, among others, class generalisation errors, space generalisation, etc. Once errors are identified, their accuracy is usually measured while any other components are qualitatively assessed. The most usual measuring systems are based on the application of statistical and geo-statistical methods, stochastic simulation or retrospective validation with independent data. Most of them are based on sample taking and on root mean square error or confusion matrix setting. In relation to error propagation modelling, there are not any relevant advances. Current propagation models start on the simple idea that error has a normal distribution with no spatial variations. The most used statistical methods are the Monte Carlo analysis and, in a second place, Taylor’s Series (Heuvelink, 1998). Strategies to deal with or reduce error are not encouraging either. Among the reduction strategies, there are proposals to set a threshold beyond which assessed data are so low-quality that they can not be used, or other more general proposals such as the one suggested by P. Burrough (Burrough and McDonnell, 1998), for instance, about removing or checking data sub-groups or systemic biases. In any case at the end, it is the user who has to decide the error level he is willing to accept, always depending on the goal set to such data employment. In short, it is clear that many drawbacks will appear because not every GIS-treated data errors will be identified, measured and
Validation Of GIS-Performed Analysis
561
managed in the same way, while their removal and reduction will not be always simple nor complete.
CHECKING OR VALIDATION OF ANY GIS ANALYSIS RESULTS Until now, the usual procedure to deal with the above problems was to CHECK results, understanding it as a procedure that verifies or examine the truth of something. It is, however, a procedure difficult to implement because it is always put into practice in part (on some of the initial data). Gómez and Bosque (2004) suggest an alternate centred on the uncertainty analysis as a result validation and decision making risk control process that tries to CORROBORATES, that is, to fortify or strengthen results. The idea is to find out how GIS analysis results are affected by the different uncertainties (errors in initial data and any new errors added due to any applied analysis) involved in every GIS analysis. At the end, we obtain the analysis results as well as an assessment of said results robustness.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The Uncertainty Analysis Figure 1 presents a possible solution to this problem. The uncertainty analysis is about studying any possible errors affecting any initial data (at first not very well known) to find out how they modify the results of just applying the analytical model. Likewise, the possible variations of the analytical model and its effects on any final results ought to be taken into account, a procedure that is not usual. It is clear that the proposed procedure makes an intensive use of computers to generate the model a large number of times and to analyse whatever variations of the final results may be obtained. At the end, if variations are not too many, user may think that results are reliable and validated. On the other hand, if variations are too many, user ought to try to improve certain model-building aspects or to collect betterquality data. These procedures are even more adequate when we obtain results that can not be verified, for different reasons, with real-world facts. For instance, using multicriteria evaluation models (MCE) to determine the best location for a certain activity on a region (Gómez and Barredo, 2005). In this type of situations, it is not possible to corroborate results. On the other hand, there are many different MCE that can be applied with a large probability of getting different results depending on the method followed. Another instance is applying dynamic simulation models to generate different future scenarios related to changes in land use in general, and to town growth in concrete. The application of cellular robot-based or multiagent-based models has grown exponentially recently, but no methodologies have been developed to validate results related to a middle/long-term future and, therefore, those results can not be checked with some past or present data. In any case, these data can be used to carry out a model calibration process, to make sure that any predictions are accurate in function of model results for present time on the basis of the available data from the past.
562
Montserrat Gómez Delgado and Joaquín Bosque Sendra
UNCERTAINTY ANALYSIS
In the initial data
Introducing “noise” in the spatial
Introducing “noise” in the thematic
In the specifications of the model
Factors modelling
Analysis results
Using fuzzy methods
of
the
Analysis of the results
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 1. Uncertainty analysis process
Dealing with Errors in the Database Data can be affected by many different kinds of errors. Simplifying these errors can introduce important modifications on the spatial position of objects as well as on the attribute values of the variables measured in one point of space. The application of uncertainty analysis to a GIS data has increased over the last decade. This analysis is necessary to check the incidence of errors on both components of geographic data. How must we proceed? Since we usually do not know the accuracy of the database that we are using, one alternate can be to introduce random errors in the initial maps and ascertain its incidence on the results. To do this we have to (a) Insert random errors in a sample of the initial data. It must be done in both the spatial component and thematic component of the geographic data. Geometric correction techniques can be used to perturb the spatial component. After selecting a significant number of control points, we can perturb their coordinates X and Y. The size of the spatial error to be introduced must be determined regarding the scale of the original data. Figure 2a shows a soil map perturbed with randomly errors in the interval 0-500m and 0-1000m. Different procedures must be developed for the thematic component depending on the sort of variable represented. If it is a qualitative variable, we can randomly change the identifiers of the different categories. For quantitative variables, we have to decide the range of perturbation regarding the distribution function of the original variable. Figure 2b shows the effect of this “thematic noise” introduced in two different thematic variables. The first example shows the effect of the categories random change on a land use map. The second one shows a perturbed probability map of a land use change from arable to holm oak plantation. In this case, values of “noise” were extracted in the range ±40 % of the maximum value registered in the original probability map. Another important decision is concerning the size of the sample. We can establish different scenarios of “noise” to better analyse the effect of the uncertainties in the initial data. 4% of the total amount of cells of the study area is perturbed in the land use map in Figure 2b, while 8% have been perturbed in the probability map. (b) Following a Monte Carlo simulation, execute the model (or analysis, geoprocessing operation, etc.) a large
Validation Of GIS-Performed Analysis
563
number of times, selecting each time different error-perturbed maps. After all these multiple actions, we will have a set of new results. Assess the variation of all the new results obtained with the perturbed data with respect to the original result. All this information is to be used to adopt a final decision on the global level of variation of results and the confidence degree that it provides. (c) Similar procedures and others based on the use of fuzzy logic have been proposed by several authors, specially because of the lack of better data to verify the accuracy of the data used (Emmi and Horton, 1996; Davis and Keller, 1997; Kiiveri, 1997; Arbia and Haining, 1998; Hunter, 1999; Krivoruchko and Crawford, 2005; Burnik et al., 2007). However, in many cases the method is very complex and difficult to implement. In this case proposals must be developed to carry out it with the conventional tools of a GIS framework.
A
B
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
C
Figure 2. Introducing spatial and thematic “noise” in different variables of a land use change model (Amendoeira, Portugal): a) Farmer parcels map perturbed (500m and 1000m); b) Land use map; c) Probability change map to holm oak plantation.
The Uncertainty Analysis in the Geoprocessing Operations A second important source of error that we do not usually take into account is derived from the decisions underlying any analytical process applied to data. How could we deal with this? In this case the procedure can be the following, though it could be adapted to each geoprocessing operation. Firstly, we should modify the way to model the variables implied in the process. It could be by using a different mathematical or statistical procedure from which we can obtain a similar result. Figure 3 shows two different ways to model the accessibility to industrial areas. This factor is one of the 11 factors of a multicriteria evaluation process developed to find out the better location for a new hazardous waste disposal facility. In the
Montserrat Gómez Delgado and Joaquín Bosque Sendra
564
first simulation (Fig. 3a), only the distance to the industrial areas (the potential demand of this kind of installations) is assessed. Another way to model this factor is to calculate the travel distance through the road network, giving different weights to different sorts of roads (Fig. 3b). Figure 3c shows the differences between both suitability maps. We can observe that the spatial distribution of the suitability values (being 0 the worst and 255 the better) is not the same and that a large number of pixels registers high differences. The negative values are concentrated in the north and northwest part of the region, where the presence of industrial areas is low. Process should be repeated again a large number of times, selecting each time the variables to be re-modelled. In spite of the important differences found in the distribution of the above suitability level instance, results of the optimal final location of the disposal facility given by the multicriteria evaluation did not change substantially.It is possible to find more information about this planning problem and the validation procedure applied in Gómez and Bosque (2004).
Suitability
a)
0
0
27 Km
c)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Suitability
b)
27 Km
Suitability
0
27 Km
Figure 3. Accesibility to industrial areas: (a) suitability in terms of distance; (b) suitability in terms of travel distance; (c) differences between the two suitability maps.
Validation Of GIS-Performed Analysis
565
Once we get the set of new results derived from the multiple realizations made, the variation of the results should be analysed. Table 2 shows the differences in size and suitability level of the parcels obtained in our location problem. In this case, not only the accessibility factor was changed. Factors related to socio-economic variables were modified using information of different time periods. In addition, information about undesirable facilities (power lines, pipelines, wastewater treatment plants, etc) was introduced with different updating levels. Finally the decision making process should be based on the global variation level of the results and the confidence degree that the uncertainty analysis has provided. Table 2. Results in terms of size and average suitability of the parcels obtained in the location of a new hazardous waste disposal
Parcel (1) 1 2 3 4 5 6 7 8
Original scenario Pixels Average suitability 79 113 446 36 676 41 36 36
210.5 210.8 217.2 210.2 221.7 214.2 216.6 210.2
62 46 84 119 444
Scenario 1 Average suitability 201.0 201.1 201.4 201.7 207.8
Average suitability (2) 208.9 209.0 210.3 210.6 217.3
675 40 36 31 20
212.9 204.1 208.0 201.3 202.6
221.7 214.3 216.6 210.3 210.3
Parcel
Pixels
1 2 3 4 5 6 7 8 9 10
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
(1) The parcels that match spatially in both scenarios are located in the same row. (2) Value extracted of the original suitability map. Highlight parcels are those with the greater number of pixels selected to locate de hazardous waste disposal.
As already shown, the first type of uncertainty analysis (related to errors in the database) could be managed with standard procedures (inserting random perturbations in the spatial position and the thematic value of the geographical data), simulating what takes place in the real world. However, the uncertainty analysis related to geoprocessing operations is more subjective (based on user decision), and it is determined by the procedure, mathematical model, etc. that are employed in the study. Therefore, uncertainty analysis must take into account that any analysis or models developed involve a number of choices made by the user, based on his knowledge about the problem under study: variables selected and the modelling process, the statistical or mathematical procedure employed, etc. The effect of these choices in the final results should be included in the uncertainty analysis. One way to do this is by executing the analysis or model changing the mathematical procedures, the transformations of the initial variable and so on. Thus we obtain a set of results and estimate the level of variation in respect to the original analysis. It could determinate reliability of our process and the results reached with it.
Montserrat Gómez Delgado and Joaquín Bosque Sendra
566
Assessing Uncertainty Once we obtain a significant number of results with these modifications in the database or in the modelled process of variables, it is necessary to develop indicators to assess the global variation in respect to the original results. One process is based in the procedures used in the statistical treatment of the error. For instance, the root mean square error can be adapted to our analysis in the form of “Root Mean Square Uncertainty”, RMSU (Gómez Delgado, 2003).
RMSUj =
∑ (X i
ij
− X tj ) 2
(1)
n −1
where Xij is the result for j in the execution i of the model (each i executions with errors of modifications in the original data set or variables derived), Xtj is the original result of the geoprocessing operation and n the number of executions. RMSU
GUADALAJARA
0
RMSU
27 Km
< 25 25 – 50 50 - 100
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
> 100
Parcels with greater suitability values Administrative boundaries
Reference system: UTM 30
-
Figure 4. Root Mean Square Uncertainty (RMSU) of the 30 realizations respect to the results of the original model (location of hazardous waste facility).
RMSU shows the average difference of the n realizations with disturbed data or different modelled variables in respect to the results of the original model or geoprocessing operation. Small values of RMSU indicate the robustness of our results. RMSU can be assessed and mapped at any location (for example for each pixel of one raster map); thus we can assess the confidence level for each part of our study area. Figure 4 shows one example of this procedure applied to the location problem mentioned before. 30 realizations of the model were generated, each one with different perturbed-map combinations. Results show that the final parcels selected after the execution of the original model are very robust, because of the
Validation Of GIS-Performed Analysis
567
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
low IMC registered in this part of the study area. Another statistical tool can help to establish confidence on our model or geoprocessing operation. For example, the mean or population standard deviation of the set of new results can be used to determine the areas more and less affected by variations in the original values and variables. A correlation analysis between each of the results obtained and the original one can be also used to check the robustness of our model. The correlation coefficient obtained would confirm (or not) the stability of the results. Following with the location problem mentioned, Figure 5 shows the higher and the lower correlation coefficient found among the 30 executions done of the model. Depending on the model, it is possible to make a cartography that shows the stability of the results, once again one for each part of the study area. For example, if we are searching the better place to locate an equipment or activity (usually carried through multicriteria evaluation approaches), it is easy to generate a frequency map that shows the times that one pixel have been selected in the n executions of the model. It enables us to determine which locations are more times selected as a solution for the location problem, that is to say, the more robust solution. This sort of cartography is one of the techniques frequently used to explore the results of an uncertainty analysis (Lodwick, 1989; Brundsden, 1990; Lowry et al., 1995; Canters et al., 2002). In Figure 6 we find a frequency map that is part of the uncertainty analysis applied to the location problem illustrated. Up to 8 different executions of the model were obtained modifying the modelling of the different factors and using different methods of multicriteria evaluation. However, a significant part of the same parcel was systematically selected. This indicates the robustness and stability of the results, which is very convenient in a conflictive planning process like the location of a hazardous waste disposal.
Low frequency
High frequency
R² = 0.35
Low frequency
High frequency
R² = 0.93
Figure 5. Correlation coefficient (by pixel) of two realizations and the original result.
Nevertheless, it is not always possible to apply the same methodology to all of the geoprocessing operations to be carried out, or to all of the different models that can be implemented in a GIS (which have substantially grown in the last years). It is very important to take into account the nature of the results that are obtained through each operation or analysis, with the aim to adapt the uncertainty analysis procedure to its special characteristics.
568
Montserrat Gómez Delgado and Joaquín Bosque Sendra
For example, the result of a dynamic land use change simulation carried out through an agentbased model is a new land use structure for a number of years later, that is to say a qualitative result. The initial conditions and variables are also different in respect to the location problem showed before. In this case, one of the uncertainties that should be analysed is related to the influence of different spatial configurations of farmer parcels at the initial time period.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 6. Frequency maps showing the number of times that one pixel is selected trhoug multiple realizations of the model, modifying the terms of variables modeling
Figure 7 shows one example related to this. Two different initial parcels structured were introduced in the model to find out if a different spatial configuration of the farm parcels at the initial time period could generate modifications in the land use configuration at the final time period (fifty years later). Figure 7b shows a new configuration that reproduces a parcel concentration process, in such way that now the parcels of the same farmer are contiguous. Figure 7c shows a new structure of the parcels in terms of size and shape. The uncertainty analysis had to be adapted to the characteristics of this kind of models. The procedure was based in comparing the amount of land allocated to each land use with the original results (table 3). This analysis shows that several land uses suffer important modifications (montado and unmanaged lands), while others remain the same. However, the more important conclusion is that the land uses that are going to reduce substantially their presence in the study area (holms oaks and cork oaks plantations) and the land uses that are going to experiment the greater increments (montado and dense shrub) do not depend on the initial spatial configuration conditions of the farmer parcels, though they are systematically the same.
Validation Of GIS-Performed Analysis
569
A
B
C
Figure 7. A) Original ownership structure; B) Ownerhip concentration scenario; C) New farmer parcels structure. Dynamic simulation land use change model, Amendoeira, Portugal.
Table 3. Results in terms of hectares occupied for each land use in the original output for 2050 and for the model with modifications in the ownership structure (dynamic simulation land use change model, Amendoeira, Portugal)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Land Use Agricultural area Holm oak plantation Cork oak plantation Pine plantation Montado Olive groove Dense shrub Natural grassland Unmanaged lands
Original result (A) 415 13 8 338 1733 37 781 180 415
Ownership B - A New parcels C-A concentration configuration scenario (B) scenario (C) 407 -8 421 6 13 0 15 2 8 0 6 -2 149 -189 259 -79 1164 -569 1535 -198 29 -8 38 1 915 134 862 81 67 -113 92 -88 1222 807 696 481
CONCLUSIONS This paper presents a general application or methodology that brings elements to validate the results of a GIS-based informed mathematical model. It is shown that, although the initial data error is unknown, it is possible to test the reliability and validity of the solutions proposed by said model. It is then possible to ascertain how initial data errors (either in the spatial or the thematic components) can affect final results. Likewise, there are proposals to determine how certain model specifications affect final results. All these possibilities must be
570
Montserrat Gómez Delgado and Joaquín Bosque Sendra
adapted to each case, in function of the analytical model used and, specially, of the type of (qualitative or quantitative) result derived from said application. However, authors present techniques and simple indicators such as the most selected pixel cartography or the “root mean square uncertainty” to show initial data and model specifications uncertainty impact on results. These are tools easily applied to any conventional GIS, against other more sophisticated uncertainty-viewing techniques. Therefore, there is no need to develop complex tools to carry out this analysis, and it can be performed with any of the available options in a conventional GIS or with any other free-access software. There is no doubt that this availability will help to promote these analyses so little used at present. Finally, this validation procedure is specifically relevant when, for instance, we are trying to find the most suitable location for a hazardous, noxious or dangerous activity (waste disposal, polluting industries, airports, etc.) that has almost always the opposition of its negative-externalities-potentially-affected population. In these cases, validating the proposed solution showing its robustness is essential to achieve more easily an always-difficult agreement.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REFERENCES Arbia, G., and Haining, R. (1998). Error propagation modelling in raster GIS: overlay operations. International Journal of Geographical Information Science 12(2): 145-167. Brunsden, C., Carver, S., Charlton, M., and Openshaw, S. (1990). A review of methods for handling error propagation in GIS. Proceedings of the European Conference on GIS, pp. 106-116. Burnicki, A.C., Brown, D.G., and Goovaerts, P. (2007). Simulating error propagation in landcover change analysis: the implications of temporal dependence. Computers, Environment and Urban Systems 31: 282-302. Burrough, P.A., and McDonnell, R.A. (1998). Principles of Geographical Information Systems, Oxford University Press. Canters, F., Genst, W. De, and Dufourmont, H. (2002). Assessing effects of input uncertainty in structural landscape classification. International Journal of Geographical Information Science 16(2): 129-149. Davis, T.J., and Keller, C.P. (1997). Modelling uncertainty in natural resource analysis using fuzzy sets and Monte Carlo simulation: slope stability prediction. International Journal of Geographical Information Science 11(5): 409-434. Eastman, R. (2001). IDRISI32 Release 2. Guide to GIS and Image Processing. Volume 2, MA: Graduate School of Geography, Clark University, Worcester. Emmi, P.C. and Horton, C.A. (1996). Seismic risk assessment, accuracy requirements, and GIS-based sensitivity analysis. In: GIS and Environmental modeling: progress and research issues (Eds. Goodchild, M.F; Steyaert, L.T. Y Parks, B.O.). Fort Collins, CO: GIS World Books, pp. 191-195. Gómez Delgado, M. (2003). Sistemas de Información Geográfica y toma de decisiones: control del riesgo a partir de análisis de sensibilidad y análisis de incertidumbre. PhD thesis. Universidad de Alcalá.
Validation Of GIS-Performed Analysis
571
Gómez Delgado, M., and Barredo Cano, J.I. (2005). Sistemas de Información Geográfica y Evaluación Multicriterio en la ordenación del territorio. Paracuellos de Jarama, Editorial Ra-Ma. Gómez Delgado, M., and Bosque Sendra, J. (2004). Análisis de incertidumbre para la gestión del riesgo. GeoFocus (Artículos) 4:179-208. Heuvelink, G.B.M., (1998). Error propagation in environmental modelling with GIS. London, Taylor & Francis. Hunter, G.J. (1999). Managing uncertainty. In: Geographical Information Systems (Eds. Longley, P.A., Goodchild, M.F., Maguire, D.J., and Rhind, D.W.). Volume II, John Wiley & Sons, Inc., New York, pp. 633-641. Kiiveri, H.T. (1997). Assessing, representing and transmitting positional uncertainty in maps. International Journal of Geographical Information Science 11(1): 33-52. Krivoruchko, K., and Gotway Crawford, C.A. (2005). Assessing the uncertainty resulting from geoprocessing operations. In: GIS, spatial analysis and modelling (Eds. Maguire, D.J., Batty, M., and Goodchild, F.). Redlands, CA, ESRI Press. Lodwick, W. (1989). Developing confidence limits on errors of suitability analysis in GIS. In: The Accuracy of Spatial Databases (Eds. in Goodchild, M.F. and Gopal, S.). Taylor & Francis, London, pp.69-78. Lowry, J.H., Miller, H.J., and Hepner, G.F. (1995). A GIS-based sensitivity analysis of community vulnerability to hazardous contaminants on the Mexico/US border. Photogrammetric Engineering and Remote Sensing 61(11): 1347-1359. Veregin, H., and Hargitai, P. (1995). An evaluation matrix for geographical data quality. Elements of spatial data quality (Eds Guptill, S.C. and Morrison, J.L.). Oxford, International Cartographic Association, Elsevier Science, pp. 167-188.
Question Bank
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. Define the term Uncertainty. Why this is important in geospatial analysis? 2. What is accuracy assessment? What are various types of accuracies related to quality of GIS database? 3. Define the term error? What are various sources of error in GIS database? 4. What is multi-criteria evaluation? What is its importance? 5. What are various steps to handle errors in a database?
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
In: Geoinformatics for Natural Resource Management Editors: P.K. Joshi, P. Pani, S.N. Mohapatra et al.
ISBN: 978-160692-211-8 ©2009 Nova Science Publishers, Inc.
Chapter 25
DIGITAL GOVERNANCE, HOTSPOT GEOINFORMATICS, AND SENSOR NETWORKS FOR MONITORING, ETIOLOGY, EARLY WARNING, AND SUSTAINABLE MANAGEMENT Ganapati Patil*,1, Sanjay Pawde†,2, Shashi Phoha‡,3, Vijay Singhal#,4 and Raj Zambre±,5 1
Center for Statistical Ecology and Environmental Statistics, Department of Statistics, The Penn State University, University Park, Pennyslvania, USA 2 Watershed Surveillance and Research Institute, JalaSRI, M.JCollege, Jalgaon, MS, India 3 Information Science and Technology Division, ARL, The Penn State University, University Park, Pennyslvania, USA 4 District Collectorate, Jalgaon, MS, India 5 Erallo Technologies Inc., Littleton, Massachusetts, USA
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ABSTRACT A declared purpose of digital governance is to empower public with information access to enable transparency, accuracy, and efficiency for societal good at large. Spotting what is hot and prioritizing become natural undertakings as a result of the spacetime information access. Naturally, hotspot geoinformatics for natural resources monitoring, etiology, early warning, and management has become a critical need for the 21st century. The chapter introduces and exposits geoinformatics of geospatial and spatiotemporal hotspot detection and prioritization for this age of ecological, environmental, and socio-economic indicators and geospatial information technology. The discussion provides examples and illustrations of regional issues involving water resources at *
Email:[email protected] [email protected] ‡ [email protected] # [email protected] ± [email protected] †
574
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al. watershed levels, bioreserves and corridors for biodiversity and endangered species, poverty patches and their temporal trajectories, etc. The presentation concludes with an important case study of geoinformatic digital governance of innovative river connectivity in Jalagon District and the related societal impacts, inclusive of scientific issues and potential solutions.
Key words: Digital governance, geoinformatics, India, JalaSRI, sensor, watershed
INTRODUCTION This chapter is prepared in the spirit of inviting the attention of the readership to some of the initiatives of the authors that have presently culminated into a novel and innovative project for digital governance and hotspot geoinformatics. Geoinformatics of geospatial and spatio-temporal hotspot detection and prioritization is a critical need for the 21st Century. A declared need is around for statistical geoinformatics and software infrastructure development. A hotspot can mean an unusual phenomenon, anomaly, aberration, outbreak, elevated cluster, critical area etc. The declared need may be for monitoring, etiology, early warning, or sustainable management. The responsible factors may be natural, accidental or intentional. The five year National Science Fiundation Digital Government Research Program project has been instrumental to conceptualize hotspot geoinformatics partnership among several interested cross-disciplinary scientists in academia, agencies, and communities around the world. Our efforts are driven by a wide variety of case studies involving a wide variety of critical societal issues. The JalaSRI, Watershed Surveillance and Research Institute, Jalgaon, India has been instrumental to initiate several case studies at the district level. Today the group finds themselves in the knowledge society and knowledge economy. To begin with, consider the following three stimulating scenarios followed by a brief overview of the initiative in digital governance and hotspot geoinformatics.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Statistics and Significance Science strives for the discovery of significant scientific truth. It is statistics that takes care of the uncertainty of the scientific methods consisting of design, analysis, and interpretation, and even the assessment of significance. The society in which we live has chosen to fully use statistics as a decisive instrument to deal with societal crises, whether they are related to environment, education, economy, energy, engineering or excellence. While it is exciting that we are alive in the age of information, and while it is unfortunate that we find ourselves in the crisis of environment, it is only a bliss to have the opportunity to more effectively serve the cross-disciplinary cause of statistics, ecology, environment, and society in the research, training, and outreach setting.
Raster Map and Change Map What message does a remote sensing derived land cover land use map have about the large landscape it represents? At what scale and at what level of detail? Does the spatial pattern of the map reveal any societal, ecological, environmental condition of the landscape?
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
575
Can it be an indicator of change? How do we automate the assessment of the spatial structure and behavior of change to discover critical areas, hot spots, and their corridors? Is the map accurate? How accurate is it? How do you assess the accuracy of the map and of the change map over time for change detection? What are the implications of the kind and amount of change and accuracy on what matters, whether climate change, carbon emission, water resources, urban sprawl, biodiversity, indicator species, or early warning? And with what confidence, even with a single map/change-map? Research is expected to find answers to these questions and a few more that involve multi-categorical raster maps based on remote sensing and other geospatial data. It is also expected to design a prototype advanced raster map analysis system for digital governance.
Surveillance Geoinformatics and Digital Governance Geoinformatic surveillance for spatial and temporal hotspot detection and prioritization is a critical need for the 21st century digital government. A hotspot can mean an unusual phenomenon, anomaly, aberration, outbreak, elevated cluster, or critical area. The declared need may be for monitoring, etiology, management, or early warning. The responsible factors may be natural, accidental or intentional, with relevance to both infrastructure and homeland security. This involves critical societal issues, such as carbon budgets, water resources, ecosystem health, public health, drinking water distribution system, persistent poverty, environmental justice, crop pathogens, invasive species, bio-security, bio-surveillance, remote sensor networks, early warning and homeland security. The geo-surveillance provides an excellent opportunity, challenge, and vehicle for synergistic collaboration of computational, technical, and social scientists.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Brief Overview of the Initiative of Digital Governance and Hotspot Geoinformatics This initiative describes a multi-disciplinary research program based on novel methods and tools for hotspot detection and prioritization, driven by a wide variety of case studies of direct interest to several government agencies. These case studies deal with critical societal issues. Our methodology involves an innovation of the popular circle-based spatial scan statistic methodology. In particular, it employs the notion of an upper level set and is accordingly called the upper level set scan statistic, pointing to the next generation of a sophisticated analytical and computational system, effective for the detection of arbitrarily shaped hotspots along spatiotemporal dimensions. We also propose a novel prioritization scheme based on multiple indicator and stakeholder criteria without having to integrate indicators into an index, using revealing Hasse diagrams and partially ordered sets. Responding to the Government’s role and need, we propose a cross-disciplinary collaboration among federal agencies and academic researchers to design and build the prototype system for surveillance infrastructure of hotspot detection and prioritization. The methodological toolbox and the software toolkit developed will support and leverage core missions of federal agencies as well as their interactive counterparts in the society. The research advances in the allied sciences and technologies necessary to make such a system work are the thrust of this initiative. A multi-disciplinary multi-institution research team will
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
576
address the issues in an integrated manner, a crucial element of success. The team comprises several leading researchers with track records from research universities. Information technologies promise to make Government more efficient and responsive. The purpose of this initiative is to help that happen (Fig. 1).
NSF Digital Government Surveillance GeoInformatics Project, Federal Agency Partnership and National Applications for Digital Governance. Homeland Security
Disaster Management
Public Health
Ecosystem Health
Federal Agency Partnership
Other Case Studies
Surveillance Geoinformatics of Hotspot Detection, Prioritization and Early Warning
Statistical Processing: Hotspot Detection, Prioritization, etc.
NSF Digital Government Project #0307010
CDC DOD EPA NASA NIH NOAA USFS USGS
Arbitrary Data Model, Data Format, Data Access
PI: G. P. Patil
Application Specific De Facto Data/Information Standard
National and International Applications
[email protected]
Standard or De Facto Data Model, Data Format, Data Access Data Sharing, Interoperable Middleware
Linear extension decision tree Agency Databases
Thematic Databases
Poset
Other Databases
(Hasse Diagram) b
a
Cellular Surface
c e
c e
d f
b
a b b
d
a d
c
c e
d
d
a
c
c
b
e
d
d
e
f
d
e
f
e
f
d
e
f
e
f
e
f
f
f
f
e
f
f
e
f
e
f
f
e
f
e
f
e
d
e
d
c
• Biosurveillance • Carbon Management • Coastal Management • Community Infrastructure • Crop Surveillance • Disaster Management • Disease Surveillance • Ecosystem Health • Environmental Justice • Environmental Management • Environmental Policy
• Homeland Security • Invasive Species • Poverty Policy • Public Health • Public Health and Environment • Robotic Networks • Sensor Networks • Social Networks • Syndromic Surveillance • Tsunami Inundation • Urban Crime • Water Management
Geoinformatic Surveillance System
Websites: http://www.stat.psu.edu/~gpp/ http://www.stat.psu.edu/hotspots/ http://www.stat.psu.edu/%7Egpp/DGOnlineNews2006.mht
Geoinformatic spatio-temporal data from a variety of data products and data sources with agencies, academia, and industry
Spatially distributed response variables
Hotspot analysis
Prioritization
Masks, filters
Decision support systems
Masks, filters
Indicators, weights
Figure 1. Schematic diagram of digital government surveillance geoinformatics project
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Statistical geoinformatics and geospatial data mining can help with more conceptual and methodological background. The following two monographs provide insights to this:
Landscape Pattern analysis for Assessing Ecosystem Condition One of our current challenges is the preservation and remediation of ecosystem integrity. This requires monitoring and assessment over large geographic areas, repeatedly over time, and cannot be practically fulfilled by field measurements alone. Remotely sensed imagery plays a crucial role by its ability to monitor large spatially continuous areas. This technology increasingly provides extensive spatial-temporal data; however, the challenge is to extract meaningful environmental information from such extensive data. This presents a new method for assessing spatial pattern in raster land cover maps based on satellite imagery in a way that incorporates multiple pixel resolutions. This is combined with more conventional singleresolution measurements of spatial pattern and simple non-spatial land cover proportions to assess predictability of both surface water quality and ecological integrity within watersheds of the state of Pennsylvania (USA) (Johnson and Patil, 2007).
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
577
Pattern-Based Compression of Multi-Band Image Data for Landscape Analysis This describes an integrated approach to using remotely sensed data in conjunction with geographic information systems for landscape analysis. Remotely sensed data are compressed into an analytical image-map that is compatible with the most popular geographic information systems as well as freeware viewers. The approach is most effective for landscapes that exhibit a pronounced mosaic pattern of land cover. The image maps are much more compact than the original remotely sensed data, which enhances utility on the internet. As value-added products, distribution of image-maps is not affected by copyrights on original multi-band image data (Myers and Patil, 2007).
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Background We propose a multi-disciplinary research program to develop infrastructure for geoinformatic surveillance based on novel methods and tools, tightly coupled with case studies of critical importance to several government agencies. In particular, we propose to enhance and broaden the popular spatial scan statistic method, which has been widely used for medical surveillance. For example, during the summer of 200l, it was successfully used for the early detection of dead bird clusters to localize West Nile virus epicenters in New York City. Cluster findings led to preventive measures such as targeted application of mosquito larvicide (Mostashari et al., 2003). Our enhancement is called the upper level set (ULS) scan statistic (Patil, 2002; Patil et al., 2004; Myers et al., 2006; Patil et al., 2004a; Patil et al., 2004b; Patil and Taillie, 2004a). Some of its attractive features include: (i) identification of arbitrarily shaped clusters; (ii) data-adaptive zoning of candidate hotspots; (iii) applicable to data on a network; (iv) yields both a point estimate and a confidence set for the hotspot; (v) uses hotspot-membership rating to map hotspot boundary uncertainty; (vi) computationally efficient; (vii) applicable to both discrete and continuous syndromic responses; (viii) identifies arbitrarily shaped clusters in the spatial-temporal domain; and (ix) provides a typology of space-time hotspots with discriminatory surveillance potential. The ULS scan statistic ranks hotspots according to their statistical significance (likelihood values). But, other factors need to be considered in prioritizing hotspots, such as mean response, peak response, geographical extent, population size, economic value, political and social considerations, etc. We therefore envision a suite of indicator values attached to each hotspot with large indicator values signifying greater importance. Different indicators reflect different criteria and may rank the hotspots differently. Therefore, we also propose a prioritization tool based on multiple indicator and stakeholder criteria without having to subjectively integrate indicators into an index. The prioritization tool employs Hasse diagrams for visualization purposes and partially ordered set for analytical purposes (Patil and Taillie, 2004b). Our team involves researchers with a solid track record in a number of complementary areas that are at the core of this project. Our approach will develop and combine appropriate methodologies paying particular attention to the related computational aspects. We will integrate the resulting advances into a decision support system to be used on a rich set of large-scale case studies. The project goals and results will be achieved in a well-integrated disciplinary and cross-disciplinary effort coupled with matching educational abilities.
578
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
Illustrative Study Areas The proposed geosurveillance project identifies studies in health, environment, persistent poverty, environmental justice on the one hand, and in biosurveillance, crop surveillance, and security on the other. This section describes some of these illustrative applications and case studies: Network Analysis of Biological Integrity in Freshwater Streams This study will employ the network version of the ULS scan statistic to characterize biological impairment along the rivers and streams of Pennsylvania and to identify subnetworks that are badly impaired. The state Department of Environmental Protection is determining indices of biological integrity (IBI) at about 15,000 sampling locations across the Commonwealth. Impairment will be measured by a complemented form of these IEI values. We will also use remotely sensed landscape variables and physical characteristics of the streams as explanatory variables in an attempt to account for impairment hotspots. Hotspots that remains unaccounted for after this filtering exercise become candidates for more detailed modeling and site investigation.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 2. Watershed disturbance vs. vulnerability
Watershed Prioritization for Impairment and Vulnerability This study will develop a prioritization model for watersheds (12-digit HUCs) of the Mid-Atlantic Highlands. A suite of indicators will be identified to assess each watershed's susceptibility to impairment (vulnerability). A second suite of indicators will measure actual stress or disturbance for each watershed. The watersheds will then be ranked according to each of the two separate sets of indicators. The proposed prioritization methodology will be used for ranking purposes. Each watershed is thus assigned a pair of ranks indicating its vulnerability status and its disturbance status. The pairs of ranks yield a scatter plot in the disturbance × vulnerability plane. The four quadrants in this plot have distinctly different management implications, as depicted in the accompanying diagram. Disturbance will be measured by stressor variables such as: excess sediment, riparian degradation, mine drainage, excess nutrients, exotic species, agriculture (esp. on slopes), road crossings, forest fragmentation, and indices biological impairment. Vulnerability primarily reflects physical characteristics and natural features of the watershed and can be measured by: hydrogeomorphology (HGM), climate, aspect, slope, stream sinuosity, soil type, bedrock, and
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
579
water source. Products include: a procedure for classifying watersheds by their features and condition, taxonomy of MidAtlantic watersheds, and a set of monitoring and restoration options for each watershed class that can assist managers in developing total maximum daily load (TMDL) plans.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Spatial-Temporal Patterns of Poverty in US Metropolitan Areas Poverty has been a persistent problem for the US and a costly target of federal policy interventions for many decades. This study is driven by four questions concerning urban poverty: (i) What explains the persistence of poverty over time? (ii) What explains the growth of high poverty neighborhoods? (iii) What explains the geographic concentration of the poor? and (iv) How have policy interventions affected the patterns of urban poverty? We hypothesize that the explanations of urban poverty will vary, depending on the different patterns of persistence, growth and concentration, and that examination of these patterns will provide clues for improved policy interventions. A principal information source will be the 1970-2000 census tract data with boundaries rectified for temporal comparisons. Approximately 45,000 metropolitan tracts have complete poverty data for all four census years. We will employ the proposed ULS scan statistic to identify Y space-time clusters of metropolitan poverty, to track their time-slice trajectories, and to develop a spatial-temporal typology for metropolitan poverty in the US. Poverty is a household, instead of a per capita, characteristic so appropriate modifications will be made to the scan statistic methodology to account for statistical clustering and variable household sizes. Dead Bird Clustering - Early Warning System for West Nile Virus Since the 1999 West Nile (WN) virus outbreak in New York City (NYC), health officials have been searching for an inexpensive and real-time early warning system that could signal increased risk of human WN infection, and provide a basis for targeted public education and increased mosquito control. Laboratory evidence of WN virus preceded most human infections in 2000, but sample collection and laboratory testing are time-consuming and costly. We have evaluated the cylinder-based space-time scan statistic for detecting small area clustering of dead bird reports and have found it useful in providing an early warning of West Nile virus activity in NYC. All unique non-pigeon dead bird reports were geocoded, and categorized as ‘cases’ if occurring in the prior 7 days, ‘controls’ if occurring during a historic baseline, or censored. The proposed case study would revisit the analysis using the ULS space-time scan statistic. Since the latter allows for arbitrarily shaped clusters in both the spatial and temporal dimensions, there is potential for earlier detection with more accurate delineation as well as a reduced false alarm rate. Mapping Priority Hotspots of Vegetative Disturbance for Carbon Budgets Hotspot detection can complement existing approaches to remote measuring and mapping vegetation disturbance for global change research. Existing data products either strive to reduce 'false alarms' by relying on multi-year comparisons of matched 'best quality' data or restrict information to one type of disturbance (e.g., MODIS fire products). National and global carbon budgets, at time scales relevant to inversion of atmospheric transport models, require data that are both more timely and more comprehensive. Producing such data in an operational mode would be well beyond the scope of this case study. Nonetheless it is vital to
580
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
investigate approaches that could fill this critical gap. The proposed toolkit for hotspot detection and ranking shows great promise for identifying significant disturbance events and providing a 'front-end' to a collaborative system for characterizing their carbon cycle consequences. This case study will sample BOS data streams (primarily from MODIS instruments) and test proposed hotspot algorithms for their value in carbon cycle research and potential for support of carbon management decisions and technology.
Oceanic Surveillance using a Remote Mobile Sensor Network This study will validate empirical methods for dynamic feedback in sensor networks including biological, chemical and physics-based mechanisms. Our application is the mapping of oceanographic fields such as bathymetry, temperature and currents using unmanned undersea vehicles. ULS scan statistic theory will be used to guide the vehicles by estimating the location of hotspots based on the data previously taken by the surveillance network. In our case, hotspots are areas of high variation in the data fields. By detecting only the significant variations, resources are not wasted on mapping areas of little change. As mobile sensor platforms move toward estimated hotspot locations, more data will be taken and used to update the locations. The Autonomous Ocean Sampling Network Simulator will be used for high resolution, spatio-temporally coordinated surveys. Oceanographic data fields will be determined by the Harvard Ocean Prediction System.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Proposed Tools Scan Statistic Three central problems arise in geographical surveillance for a spatially distributed response variable. These are (i) identification of areas having exceptionally high (or low) response; (ii) determination of whether the elevated response can be attributed to chance variation (false alarm) or is statistically significant; and (iii) assessment of explanatory factors that may account for the elevated response. Although a wide variety of methods have been proposed for modeling and analyzing spatial data (Cressie, 1991), the spatial scan statistic (Kulldorff and Nagarwalla, 1995; Kulldorff, 1997) has quickly become a popular method for detection and evaluation of disease clusters. When applied in space-time, the scan statistic can provide early warning of disease outbreaks and can monitor the spatial spread of an outbreak. With innovative modifications, the scan statistic approach can be used for hotspot analysis in any field. We propose to develop methodology and corresponding software for applications of the scan statistic to critical areas of concern for the digital government of the 21st century. Spatial scan statistic - The spatial scan statistic deals with the following situation. A region R of Euclidian space is tessellated or subdivided into cells that will be labeled by the symbol a. Data is available in the form of a count Ya (non-negative integer) on each cell a. In addition, a ‘size’ value Aa is associated with each cell a. The cell sizes Aa are regarded as known and fixed, while the cell counts Ya are random variables. In the disease setting, the response Ya is the number of diseased individuals within the cell and the size Aa is the total number of individuals in the cell. Generally, however, the size variable is adjusted for factors such as age, gender, environmental exposures, etc., that might affect incidence of the disease. The disease rate within the cell is the ratio Ya / Aa. The spatial scan statistic seeks to identify
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
581
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
‘hotspots’ or clusters of cells that have an elevated rate compared with the rest of the region, and to evaluate the statistical significance (p-value) of each identified hotspot. These goals are accomplished by setting up a formal hypothesis-testing model for a hotspot. The null hypothesis asserts that there is no hotspot, i.e., that all cells have (statistically) the same rate. The alternative states that there is a cluster Z such that the rate for cells in Z is higher than for cells outside Z. An essential point is that the cluster Z is an unknown parameter that has to be estimated. Likelihood methods are employed for both the estimation and significance testing. Candidate clusters for Z are referred to as zones. Ideally, maximization of the likelihood should search across all possible zones, but their number is generally too large for practical implementation. Various devices (e.g., expanding circles) are employed to reduce the list of candidate zones to manageable proportions. Significance testing for the spatial scan statistic employs the likelihood ratio test; however, the standard chi-squared distribution cannot be used as reference or null distribution-in part because the zonal parameter Z is discrete. Accordingly, Monte Carlo simulation (Dwass, 1957) is used to determine the needed null distributions. Explication of a likelihood function requires a distributional model (response distribution) for the response Ya in cell a. This distribution can vary from cell to cell but in a manner that is regulated by the size variable Aa. Thus, Aa enters into the parametric structure of the response distribution. In disease surveillance, response distributions are generally taken as either binomial or Poisson, leading to comparatively simple likelihood functions. The scan statistic that we propose allows continuous response distributions and complex likelihood functions. Limitations of current scan statistic methodology - Available scan statistic software suffers from several limitations. First, circles have been used for the scanning window, resulting in low power for detection of irregularly shaped clusters (Fig. 3). Second, the response variable has been defined on the cells of a tessellated geographic region, preventing application to responses defined on a network (stream network, water distribution system, highway system, etc.). Third, reflecting the epidemiological origins of the spatial scan statistic, response distributions have been taken as discrete (specifically, binomial or Poisson). Finally, the traditional scan statistic returns only a point estimate for the hotspot but does not attempt to assess estimation uncertainty. We propose to address all these limitations.
Figure 3. Circular spatial scan statistic zonation (left) and cylindrical space-time zonation (right)
In our approach to the scan statistic, the geometric structure that carries the numerical information is an abstract graph consisting of (i) a finite collection of vertices and (ii) a finite set of edges that join certain pairs of distinct vertices. A tessellation determines such a graph: vertices are the cells of the tessellation and a pair of vertices is joined by an edge whenever
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
582
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
the corresponding cells are adjacent. A network determines such a graph directly. Each vertex in the graph carries three items of information: (i) a size variable that is treated as known and non-random; (ii) a response variable whose value is regarded as a realization of some probability distribution; and (iii) the probability distribution itself, which is called the response distribution. Parameters of the response distribution may vary from vertex to vertex, but the mean response (i.e., expected value of the response distribution) should be proportional to the value of the size variable for that vertex. The response rate is the ratio Response / Size and a hotspot is a collection of vertices for which the overall response rate is unusually large. ULS scan statistic - We will develop a new version of the spatial scan statistic designed for detection of hotspots of arbitrary shapes and for data defined either on a tessellation or a network. Our version looks for hotspots from among all connected components of upper level sets of the response rate and is therefore called the ULS scan statistic. The method is adaptive with respect to hotspot shape since candidate hotspots have their shapes determined by the data rather than by some a priori prescription like circles or ellipses. This data dependence will be taken into account in the Monte Carlo simulations used to determine null distributions for hypothesis testing. We will also compare performance of the ULS scanning tool with that of the traditional spatial scan statistic. The key element here is enumeration of a searchable list of candidate zones Z. A zone is, first of all, a collection of vertices from the abstract graph. Secondly, those vertices should be connected (Fig. 4) because a geographically scattered collection of vertices would not be a reasonable candidate for a ‘hotspot’. Even with this connectedness limitation, the number of candidate zones is too large for a maximum likelihood search in all but the smallest of graphs. We propose to reduce the list of zones to searchable size in the following way. The response rate at vertex a is Ga = Ya / Aa. These rates determine a function a → Ga defined over the vertices in the graph. This function has only finitely many values (called levels) and each level g determines an upper level set U g defined by Ug = { a : Ga ≥ g}. Upper level sets do not have to be connected but each upper level set can be decomposed into the disjoint union of connected components. The list of candidate zones Z for the ULS scan statistic consists of all connected components of all upper level sets. This list of candidate zones is denoted by ΩULS. The zones in ΩULS are certainly plausible as potential hotspots since they are portions of upper level sets. Their number is small enough for practical maximum likelihood search-in fact, the size of ΩULS does not exceed the number of vertices in the abstract graph (e.g., the number of cells in the tessellation). Finally, ΩULS becomes a tree under set inclusion, thus facilitating computer representation. This tree is called the ULS-tree (Fig. 5); its nodes are the zones Z Є ΩULS and are therefore collections of vertices from the abstract graph. Leaf nodes are (typically) singleton vertices at which the response rate is a local maximum; the root node consists of all vertices in the abstract graph. Finding the connected components for an ULS is essentially the issue of determining the transitive closure of the adjacency relation defined by the edges of the graph. Several generic algorithms are available in the computer science literature (Carmen et a1., 2001; Knuth 1973; Press et al., 1992).
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
583
Figure 4. Connectivity for tessellated regions. The collection of shaded cells on the left is connected and, therefore, constitutes a zone. The collection on the right is not connected.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 5. A confidence set of hotspots on the ULS tree. The different connected components correspond to different hotspot loci while the nodes within a connected component correspond to different delineations of that hotspot all at the appropriate confidence level.
Hotspot confidence sets - The hotspot MLE is that-an estimate. Removing some cells from the MLE and replacing them with certain other cells can generate an estimate that is almost as plausible in the likelihood sense. We will express this uncertainty in hotspot delineation by a confidence set of hotspot zones-a subset of the ULS tree (Fig. 5). We will determine the confidence set by employing the standard duality between confidence sets and hypothesis testing (Lehmann 1986) in conjunction with the likelihood ratio test. The confidence set also lets us assign a numerical hotspot-membership rating to each cell (e.g., county, zip code, census tract). The rating is the percentage of zones (in the confidence set) that include the cell under consideration (Fig. 6). A map of these ratings, with superimposed MLE, provides a visual display of uncertainty in hotspot delineation. Typology of space-time hotspots - Scan statistic methods extend readily to the detection of hotspots in space-time. The space-time version of the circle-based scan statistic employs cylindrical extensions of spatial circles and is unable to detect the temporal evolution of a hotspot. The space-time generalization of the ULS scan statistic will be able to detect arbitrarily shaped hotspots in space-time. This will allow us to classify space-time hotspots into various evolutionary types-a few of which appear on the left hand side of Figure 7. The merging hotspot is particularly interesting because, while it comprises a connected zone in space-time, several of its time slices are spatially disconnected.
584
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
Figure 6. Hotspot-membership rating. Cells in the inner envelope belong to all plausible estimates (at specified confidence level); cells in the outer envelope belong to at least one plausible estimate. The MLE is nested between the two envelopes
Figure 7. The four diagrams on the left depict different types of space-time hotspots. The spatial dimension is represented schematically on the horizontal axis while time is on the vertical axis. The diagrams on the right show the trajectory (sequence of time slices) of a merging hotspot.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Prioritization Methodology We address the question of ranking a collection of objects, such as initial hotspots, when a suite of indicator values is available for each member of the collection. The objects can be represented as a cloud of points in indicator space (Filar and Ross, 2001), but the different indicators (coordinate axes) typically convey different comparative messages and there is no unique way to rank the objects. A conventional solution is to assign a composite numerical score to each object by combining the indicator information in some fashion. Every such composite involves judgments (often arbitrary or controversial) about tradeoffs or substitutability among indicators. Rather than imposing such a composite, we take the view that the relative positions in indicator space determine only a partial ordering (Fishburn, 1985; Neggers and Kim, 1998; Trotter, 1992) and that a given pair of objects may not be inherently comparable. Working with Hasse diagrams (Neggers and Kim, 1998; Di Battista et al., 1999) of the partial order, we propose to study the collection of all rankings that are compatible with the partial order Multiple Indicators and Partially Ordered Sets (Posets). The scan statistic
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
585
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
ranks hotspots based on their statistical significance (likelihood values). But, other factors need to be considered in prioritizing hotspots, such as mean response, peak response, geographical extent, population size, economic value, etc. We therefore envision a suite of indicator values attached to each hotspot with large indicator values signifying greater hotspot importance. Different indicators reflect different criteria and may rank the hotspots differently. In mathematical terms, the suite of indicators determines a partial order on the set of hotspots. Thus, if a and bare hotspots, we say that b is inherently more important than a and we write a < b if I(a)≤ I(b) for all of the indicators I. If distinct hotspots are distinct in indicator space, the relation has the three defining properties of a partial order: (i) transitive: a band b c implies a c ; (ii) antisymmetric: a band b a implies a = b; and (iii) reflexive: a a. Certain pairs a, b of hotspots may not be comparable under this importance ordering since, for example, there may be indicators such that I1 (a) < I1(b) but I2 (a) > I2 (b). In this case, hotspot b would be located in the fourth quadrant of Figure 6. Because of these inherent incomparabilities, there are many different ways of ranking the hotspots while remaining consistent with the importance ordering. A given hotspot a can therefore be assigned different ranks depending upon who does the ranking. It turns out that these different ranks comprise an interval (of integers) called the rank interval of a. Rank intervals can be calculated directly from the partial order. First, define B(a) to be the number of hotspots b for which a b, i.e., the count of the first quadrant in Figure 6. Next, define W(a) as B(a) plus the number of hotspots that are not comparable with a; this is the total count for quadrants I, 2, and 4 in Figure 8. The rank interval of a then consists of all integers r such that B(a)≤ r ≤ W(a). The length, W(a) - B(a), of this interval is called the rankambiguity of hotspot a.
Figure 8. Regions of comparability and incomparability for the inherent importance ordering of hotspots. Hotspots form a scatterplot in indicator space and each hotspot partitions indicator space into four quadrants
Hasse Diagrams and Linear Extensions - Posets can be displayed as Hasse diagrams (Fig. 9). A Hasse diagram is a graph whose vertices are the hotspots and whose edges join vertices that cover one another in the partial order. Hotspot b is said to cover a in the partial order if three things happen: (i) a b; (ii) a ≠ b ; and (iii) if a x b then either x = a or x = b. In words, b is strictly above a and no hotspots are strictly between a and b. Each of the many possible ways of ranking the elements of a poset is referred to as a linear extension. The Hasse diagram of each linear extension appears as a vertical graph (Fig. 9). Enumeration of all possible linear extensions can be accomplished algorithmically as follows. The top
586
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
element of a linear extension can be anyone of the maximal elements of the Hasse diagram. Select anyone of these maximal elements and remove it from the Hasse diagram. The second ranked element in the linear extension can be any maximal element from the reduced Hasse diagram. Select any of these and proceed iteratively. The procedure can be arranged as a decision tree (Fig. 9) and each path through the tree from root node to leaf node determines one linear extension.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Figure 9. Hasse diagram of a hypothetical poset (left), some linear extensions of that poset (middle), and a decision tree enumerating all 16 possible linear extensions (right). Links shown in dashed/red (called jumps) are not implied by the partial order. The six members of the poset can be arranged in 6!=720 different ways, but only 16 of these orderings are valid linear extensions
Figure 10. (Left) Rank-frequency table for the poset of Figure 9. Each row gives the number of linear extensions that assign a given rank r to the corresponding member of the poset. Each row is referred to as a rank-frequency distribution. (Right) Cumulative rank-frequency distributions for the poset of Figure 9. The curves are stacked one above the other giving a linear ordering ofthe elements: a> b > c >d>e>f
Linearizing a Poset - The suite of indicators determines only a partial order on the hotspots, but it is human nature to ask for a linear ordering of those hotspots. We ask the question: Is there some objective way of smoothing the partial order into a linear one? Our proposed solution treats each linear extension in Figure 9 as a voter and we apply the principle of majority rule. Focus attention on some member of the poset, say element a, and ask how many of the voters give a rank of l? Rank of 2? Rank of 3? etc. The results are
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
587
displayed in Figure 8, where each row of the table is called a rank-frequency distribution. The cumulative forms of these rank-frequency distributions form a new poset with stochastic ordering of distributions as the order relation. For this example, the new poset is already a linear ordering (see Fig. 10).
Figure 11. (Left) Two iterations of the CRF operator are required to transform this partial order into a linear order. (Right) A poset for which the CRF operator produces ties
We refer to the above procedure as the cumulative rank-frequency (CRF) operator. In general, it does not transform a partial order into a linear order in a single step; instead, multiple iterations may be required (Fig. 11). The CRF operator can also produce ties in the final linear ordering.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Sensors Envisaged for Water Conservation The availability of a variety of inexpensive micro-sensors with embedded wireless communications has enabled real-time monitoring of natural phenomena that span temporal and spatial scales. This enables in-situ information fusion for comprehension and scientific prediction of spatial-temporal events, which in turn supports scientific decision models that adapt to predicted events. For example, autonomous networks of unmanned undersea vehicles with embedded sensor systems have been designed to formulate high fidelity newcasts and forecasts of the ocean through time-space coordinated sampling to support collaborative undersea mine-hunting missions (Phoha et. al., 2006; Phoha et al., 1999). The National Ecological Observatory Network (NEON) is another effort of the US National Science Foundation to create a national observing system for ecological measurements and monitoring to support research (Schimel, 2007). In this section we present recent research on sensor networking architectures that enable in-situ scientific decision making with the goal of exploring possible value added enhancements to current plans of water linking by the JalaSRI project in Jalgaon, India. This research will enable the project to establish the appropriate
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
588
regional infrastructure for utilizing the transformational power of information to support situation aware adaptive control of natural resources, such as optimal water conservation. Other possible uses of such a network are delineated by the NEON project in areas of land use and agriculture, spatial patterns of climate-change that affect eco-hydrology and bio-geochemistry, and bio-diversity (Schimel, 2007). The important characteristics of the decisionsupport sensor network architecture are its quality of fusion support, low total cost of ownership, scalability, portability of nodes, and system dependability. The architectural design of the infrastructure for an adaptive sensor network has generated a lot of research interest and experimentation. The following subsections discuss some of the design issues for a cost effective, flexible and reconfigurable sensor network. The major new research addressed here is the fusion driven dynamic adaptation of the decision support network. The chapter presents innovative analytical models to support regional decision-making. The methodology is extendible and has the potential of influencing the design of a national scale environment monitoring network such as the INDOFLUX (Srinivasan et al., 2007).
Cyberinfrastructure Architecture A sensor network operates on an infrastructure of sensing, computation, and communication, through which it perceives the evolution of physical dynamic processes in its environment. Sensors require physical interaction with the sensed phenomena and are subject to a number of noise factors. Sensor data is therefore highly correlated in the vicinity of a stimulus. To get reliable performance from individually less reliable sensors, time-critical collaborative inference in the vicinity of a stimulus is necessary to circumvent limitations of sensing, communications, power, and equipment faults. We call this dynamic clustering. Characterizing dynamic events in multiple spatial-temporal scales under operational constraints requires the tactful capture of coarse and fine grained system dynamics. A large resource constrained sensor network must, therefore, dynamically switch from coarse to fine grained topologies to support progressively segmented analyses to localize emerging hotspots, correct spatial-temporal misalignments through statistical analyses, and discover distributed higher level associations of emerging patterns in diverse multisource asynchronous data sources. A typical sensor network should consist of the following components:
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
•
• • • • •
A variety of sensors types including: acoustic doppler, oxygen optode, ecolab analyzer (NO3, Si, PO4, NH4) , turbidity sensor, temperature sensor, IR gas analyzer, pH probes, RH probes, quantum sensors, fluorometer, barometer, wind sensor, rain gauge, prynometer, soil temperature, soil moisture, sonic anamometer, gas analyzer, etc. Sensing actuators - Interact with the environment to gather data. Local nodes - Log sensor data do some signal processing. Store data temporarily. Network - Transfer data from field site to data portal. Data portal - Store data permanently. Manage and manipulate data. User interface - Manipulate data on portal, download for local processing, or use custom tools.
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
589
Spatial and temporal sensor sampling rates will vary greatly. Real-time data interactions are necessary. Some local nodes are remote data logging devices that store information for later retrieval. The network physical layer may use long-range 802.11 and/or cell phone connections. The data portal provides a grid computing environment. Data signatures certify the sensor hardware that produced the original data and provide assurance that the data is not tampered with. The exact processing history of all derived data can be verified using cryptographic primitives. Sensors interact with their environment and degrade over time, leading to loss of precision and/or accuracy. With minimal knowledge of degradation modes, it is possible to detect and compensate for calibration problems. Distributed calibration considering a variety of noise models is described in (Brooks 1996). Several design issues for such dynamic sensor networks have been addressed in (Phoha et al., 2006). These include: • • • • •
Sensor deployment, self-organization and localization Purposeful mobility and scalability Network routing and protocol design Power and resource management, and Network security.
Fusion Driven Design Concept The dynamic adaptation of the sensor network is necessary to support regional decisionmaking and action-oriented control. The goal is to formulate analytical models by using the non-stationary statistics of the information dynamics of the sensor data to drive in-situ changes in the network design space as depicted in (Fig. 12). This figure illustrates the concept of closed-loop network control that manipulates the network topology ( ), based on
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
feedback information of evolving statistical patterns ( ) derived from sensor data sequences (yk).
Figure 12. Solution Concept
590
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
To enhance quality of data fusion and resilience, a distributed sensor network needs to be adaptively reconfigured, where the network topology is updated in real time based on the spatial -temporal information derived from the ensemble of sensor data. We proceed to present methods for in-situ construction of statistical models of sensor information and fusion processes in the next two sections.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Multilevel Fusion Model The Information space of the dynamical system is represented by spatial-temporal statistics of the ensemble of sensor data. In this context, sensor data fusion is posed as a multi-time-scale problem under the following assumptions: (i) quasi-stationarity over the fasttime scale (i.e., stationary over a sufficiently long duration) and (ii) possible non-stationarity caused by small parametric or non-parametric deviations in the system behavior due to accumulating changes in the slow -time scale. We characterize multi-level fusion processes as hierarchical dynamic processes that detect signal patterns in multiple, diverse, and spatially dispersed sensor data streams at four hierarchical levels of abstraction. Symbolization is the first atomic level of fusion akin to feature extraction. It captures causal information, communications and computational patterns embedded in the underlying physics. Higher levels of abstraction represent higher levels of fusion. Models of Information Dynamics The discrete event dynamics of sensor data is modeled as hybrid multilayered interacting probabilistic automata (Phoha et al., 1999; Phoha et al., 2002). Continuously varying dynamics capture the physical processes at the lowest level of abstraction while discrete event models integrate sensing, computation and communication events in a formal language representation (Phoha et al., 2002). A formal language measure has been developed for measuring operational deviations from specified behavioral representations (Ray et al., 2005). This analytically captures the structural dynamics of the information space at various levels of abstraction and develops a measure for its deviations from normalcy. In its simplest form, the information space is modeled as cellular automata with its cells representing sensors that are on or off and interacting with neighboring cells organized as a lattice (Brooks et al., 2002). More complex high fidelity models capture the nonlinear interactive and multi scale dynamics of the sensor network at multiple layers of abstraction (Biswas et al., 2006; Phoha et al., 2006) to assess coverage, connectivity and coordination. As discussed below, these models enable the analytical formulation and empirical evaluation of a networked decision support system. However, to rigorously address the inverse problem, we incorporate algorithms for in-situ derivation of statistical characteristics of the information space. In-Situ Model Construction and Approximation of Information Statistics We first formulate mathematical techniques for local processing at a sensor node into semantic information with flexible resolution. We represent nodes by traveling wave packets of sensor energy, enabling a semantic interpretation (Friedlander et al., 2002):
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
591
(1)
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
where V is the velocity of the wave packet, x is its position at time t, S , is its scale and 11 is its offset. This representation treats space and time in a symmetric manner, preserving translational invariance for forecasting. Because the wavelet transform divides data into different frequency components and analyzes each component with a resolution matched to its scale, we can directly model the dependent properties of sensor data. Hence, we represent the information originating at a single sensor node as a finite set of wavelet coefficients that change due to interactions with the environment or other nodes.
Figure 13. Information, modeling and network control hierarchies
We discretize this dynamical system, both spatially and temporally, through novel symbolization and nonlinear filtering techniques that preserve the statistical characteristics of the sensor data yielding a reduced order representation of the information space (Ray, 2004). Multivariate sensor outputs are converted to univariate symbol sequences by partitioning a compact region in the wavelet coefficient space into finitely many discrete blocks. Each block in the partition is labeled as a symbol of a finite alphabet. As the dynamical system trajectory evolves in time, it travels through various blocks generating a symbol sequence. A hidden Markov model is constructed from the symbol sequence as a finite-state automaton (FSA;
592
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Fig. 123). The FSA is constructed based on the principle of sliding block codes (Lind et al., 1995) and the machine states are defined corresponding to the alphabet of symbols. The proposed Symbolic Dynamic Filtering (SDF) (Ray, 2004) technique follows the recursion and input/output structure of Rao-Blackwellised Particle Filtering (RBPF) that is a sequential Monte Carlo Markov chain method (Doucet et al., 2001) However, unlike RBPF that is constructed as a Markov process on a finite-dimensional state space, SDF is constructed on a finite-state automaton with finite memory. While the state variables in RBPF are real-valued Markov processes, the automaton states of SDF are analogous to discrete energy states in Quantum Statistical Mechanics (Pathria, 1996).
Figure 14. SDF-based level I fusion
The next step is order reduction. For each sensor cluster, a local automaton is derived as a shift space of finite type (i.e., having finitely many forbidden blocks). The resulting PerronFrobenius operator (i.e., the state transition matrix of the FSA) has an invariant algebraic structure with time dependent parameters. This algebraic structure allows order reduction of the automaton by state merging, conceptually analogous to information marginalization in RBPF (Doucet et al., 2001). The state probability distribution, represented as histograms in Figure 14 is recursively computed as an approximation of the natural invariant density, which is a fixed point of the local Perron-Frobenius operator. For in-situ data fusion and information
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
593
compression, this model of the information space has the major advantage of providing a succinct statistical characterization of the sensor data with the following benefits: •
•
• •
•
Robustness to noise - the wavelet coefficients not only help represents patterns at different scales, but also severely mitigate the effects of measurement noise and spurious signals. Thus, probability of occurrence of false symbols, which either degrade accuracy of pattern identification or produce false alarms, is significantly reduced ID Adaptive resolution - partitioning based on maximization of the Shannon entropy (Rajagopalan and Ray, 2006) makes regions with more information segmented finer. This resolution is associated to sensing density Capability for early detection of emergent behaviors with decreased probability of false alarms due to sensitivity to changes in the underlying dynamical system Compression of multi-sensor information into a code book of short packets of statistical pattern yields high throughput, low latency and error-corrected transmission over a wireless communication network Real-time execution on COTS platforms
In contrast to RBPF and other state estimation methods, our technique does not require an explicit model of the system as it extracts the intrinsic in-situ information directly from sensor data. Freedom to choose an arbitrary starting point makes it robust for statistical pattern discovery. This formalism provides the basic analytical framework for capturing effects of change in the network design space on system evolution.
Network Reconfiguration The network design space is reconfigured to adapt to the information space in a manner that preserves the statistical characteristics (predictability) of the ensemble of original sensor data at each level of fusion. In the following steps, we present how we use these concepts to build a probabilistic theory for fusion based decision support by designing flexible sensor networks that capture change in operational environments:
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
•
•
•
Network-centric sensor information is organized as a discrete-event dynamic system of interacting probabilistic automata, where sensor nodes may change their internal states through interactions with other nodes or the environment. Sensor nodes generate multivariate asynchronous data streams that interact over the network. Based on these interactions, some sensors may form collaborative clusters. The symbolization and filtering processes (fusion levels 0 and 1) for a multivariate stream of asynchronous sensor data are said to be effective to the extent that they preserve the statistics of the original data. The goal here is to design flexible network topologies for sufficiently fine -grained adaptive sensing that can detect changes in the statistics of the information space in emerging hotspots. Statistical invariance, simultaneously in space and time, is used to reduce the order of the nonlinear dynamic systems and its computational complexity, without loss of predictability.
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
594 •
We have defined a formal quantitative language measure (Ray et al., 2005), which is used to quantify statistical changes in the information space as we vary the operational setting of the network design space. We thus formulate theoretical foundations for solving the forward and backward problems of network adaptation by analytically associating a measure of the effect of changes in the network's topological structure to forecasts of system evolution.
The actuation of network reconfiguration for large sensor networks is achieved through adaptive sampling at individual sensors, sensor mobility, turning existing sensors on or off, bandwidth reallocation, protocol modification, or through redeployment of resources. Urban topologies may further constrain such actuation, resulting in approximate solutions. This research presents issues in the design of a distributed in-situ decision support system at the regional level that is capable of multi-level environmental monitoring and resource management using sensor networks. Diversity of sensor modalities is recommended for effective sensing, identification, and cross-cluster association of complex scenarios and environmental conditions. Analytical fusion models automate change detection and prediction as human oversight for low level sensing is not feasible in this amorphous networking environment.
Sensor Network to Determine Drinking Water Quality and Security Finding patterns in large, real, spatial-temporal data continues to be of great interest. Ailamaki et al. (2003) describe a cross-disciplinary research effort to couple knowledge discovery in large environmental databases with biological and chemical sensor networks. They describe a distribution and operation protocol for the placement and utilization of in situ environment sensors by combining new algorithms for spatial-temporal data mining, new methods to model water quality and security dynamics, and a sophisticated analysis framework.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Experimental exercise in Jalgaon The objective of the program proposed in Jalgaon is to deploy a network of low cost, smart sensors to reduce or eliminate paper work, save field staff time and duplication efforts, improve the operational efficiency and accuracy of data, and provide timely data and reports to decision makers and to administrators, researchers, farmers, and the public. Jalgaon district is active in establishing digital governance process (http://www.jalgaon.nic.in). Just 20 years earlier, Jalgaon had a rich and healthy forest and ample natural resources. Back then the water table was just 80 feet deep – compared to the water table depth of 200 feet that is reality today. Over the last 20 years, the natural resources have been used in random and unmanaged ways, resulting in the current condition of resource scarcity, especially water. The ‘alarms have gone off’ in Jalgaon, and this progressive district in the state of Maharashtra is eager to adopt modern sensor technology to better manage its natural resources. The government has initiated an innovative district level river linking program (http://www.jalgaon.nic.in) to address water shortage problem.
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
595
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
The center of administration in Jalgaon is the District Collector’s office. This office was ISO 9000 certified in 2000, which demonstrates that the district is progressive and already has a well-defined process. There is a strong desire and eagerness; at all levels of the administration and with farmers, researchers, and the community, to aggressively improve the processes and implement more effective services for the people. Jalgaon has already adopted and implemented various modern tools and techniques to improve the workings of the administration. For example, it has implemented e-District, which is a district level website (http://www.jalgaon.nic.in) where reports and data are made available to the public. Jalgaon now has the desire to use smart sensors to collect and digitize data for critical tasks, like infrastructure monitoring, healthcare development, disaster management, development projects, improved agricultural productivity, and water quality monitoring. The district level water resource management requires a multidisciplinary approach. Jalgaon district administration is working closely with JalaSRI, a research institute contributing significantly to the Surveillance and Research activities for Natural Resources Monitoring and Management for the Jalgaon District (http://www.jalasri.kces.in). JalaSRI’s focus is on geoinformatics and hot spot detection. The needs and requirements of developing countries are very different than in developed countries. The availability of communication infrastructure and cost factor is unique to each country and to each district within the countries. The wireless sensor network and system design expertise for the Sensor Network Program is being provided by Erallo Technologies, Inc. USA (http://www.erallo.com); which is active in research and development of wireless sensor networks and ad hoc mesh networks.
Water Resource Management In the district of Jalgaon, water management is critical for drought conditions as well as for flood situations. 60% of the land in Jalgaon is classified as Drought Prone Areas (DPA). The remaining 40% has been classified as Assured Rain Fall Areas (ARFA); however, this classification was done many years ago and due to multiple factors, including global warming, they’ve seen an increase in the Drought Prone Areas (information gathered through personal interaction). This makes it even more critical for the government to provide effective water services and an efficient water management infrastructure to villages, farmers, and industry. The district of Jalgaon is a highly productive agricultural area; thus, the economy and politics of the area are a function of water availability. In this area, water is managed using a series of interconnected dams of 3 major dams (the Girna, Hatnur and Waghur), 10 medium-sized dams, and 70 minor dams; along with 178 inspection-classified wells. Monitoring water availability and levels is a critical component for water management and irrigation projects. In the rainy season, measurements are taken every two to four hours (water depth, overflow, and amount of water). Based on these measurements, estimations are calculated for water over-flow predictions and the potential for down-stream flooding. In the dry season, the water depth is taken once a day. The measurements are used for irrigation and pumping rates and for drinking water availability. For example, when sufficient water is available, approximately 120 liters per person per day is made available; however, in drought conditions only 20 liters per person/day is provided. Thus, the accurate reading and reporting of the water depth is critical and fundamental for forecasting and decision making. With data from the sensor networks - complete, concise, and real-time district level data on water entering the system (lakes, rivers, dam reservoir, and
596
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
aquifers) will be used in a model to predict the availability of water within the district. This will be key information for the decision makers to formulate accurate water usage policies.
A Solution for Water Resource Management
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Sensor devices could easily eliminate the limitations associated with conventional style of data gathering, recording, reporting, database generation, decision making etc. Most smart sensors (whether digital or analog) maintain an accurate clock; thus, the time-of-day that the data is collected is reported along with the measurement. Additionally, since sensors can be programmed to record measurements on a schedule or in response to a series of events, the data is always collected when required. Using cell phone technology, the sensor data can also be processed and automatically transmitted to a central office for further correlation and calculations in near real-time. Thus, the data at the district level as well as at the state level could be accurately and promptly processed and reported. The district of Jalgaon is primarily an agro-economy – the entire agricultural planning for the district, including water shed management, management of drought prone areas, and resource scarcity management – are all heavily dependent on this critical set of data. A map of the Jalgaon district and its major and minor water harvesting projects are shown in Figure 15.
Figure 15. Map of the district of Jalgaon in the state of Maharashtra, India
Sensors for Integrated Water Resource Management An integrated network of various types of water sensors could provide a very comprehensive system for managing all the aspects of water resources: drinking and farming
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
597
use; rain gauge measurements, dam depth and inspection well monitoring; water quality; planning for water harvesting, irrigation and pumping; and emergency responses for flood conditions. All of these aspects are interrelated – so then too should be the system to monitor them. Today, many of these tasks are not even routinely performed, further complicating the management of this growingly scarce resource. Measurements that are used come with very tedious manual and error-prone processes of recording. Additionally, the time lag between data collection and reporting to authorities is too late to enable pro-active responses. Automated collection and reporting via sensors would improve the processes to provide this critical resource information on time and when needed.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Rain Fall Monitoring Data from a sensor system to monitor rain fall would help officials anticipate the rain fall’s effect on the water table, water table recharge, and dams and reservoirs where water collects by providing the exact amount of water recharge due to rain. Sensors for monitoring dam reservoir depths and wells could be used to predict, manage, and prevent water overflow and flood potentials. More than just preventing floods, an integrated sensor network could also prevent the loss of a precious resource: rain water. Currently, there are only 48 manuallyread rain gauges spread throughout the district of Jalgaon. The measurements are taken at the village level and reported by phone to the tashil level (county level) at various times. Then, over 48 hours pass by before a district report is compiled and distributed to administrators, radio stations, and other authorities. With the introduction of smart, low cost, wireless rain gauge sensors, hundreds of sensors could be deployed to provide the desired granularity at a micro level. The sensors would be programmed to accurately read the rainfall measurements, on a pre-defined schedule or in response to a pre-programmed series of events. At each sampling event, the rain gauge sensor would record the rain fall and time stamp of the reading. The data would also be automatically transmitted (using cell phone technology) to the district offices and/or state level offices – in real-time or near real-time - on a regular schedule or in response to predefined events. A timely report could then be generated by software and stored in an open standards-based SQL database. By making the database structure public, any interested party could access the data within the Decision Support System (DSS). Rainfall Sensors A smart, electronic, digital rain sensor is designed to take accurate rainfall measurements from 0 to 300 mm per hour. The principle it uses is simple: water is collected by a funnel which is then routed to a measuring chamber. An electronic circuit is connected to the solidstate level sensor to output the measurement value into a programmable parameter value. When the capacity of the measuring chamber is met, the water is siphoned out (in about 1 second) onto the ground; and the process then repeats indefinitely. The electronics of the sensors are also programmed to filter out false readings, making the gauges very precise. The digital logger on the sensor can store the data for transmission. Many devices also include the time of day and outside temperature at the time of reading. Rain monitoring sensor stations will be distributed throughout the district. The sensors will provide accurate hourly rain fall in millimeters, thus providing time series rain fall data and the rate of water storage recharge. With the time series measurement, the total amount of
598
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
additional water in a given area can be accurately determined. The addition of water and rate data can also provide critical information on the rain’s effect on the aquifer as well as potential soil erosion damage. The sensors will be programmed to alter the sampling parameters once they detect rain and all sensor data will be sent to a central site; thus providing real-time data on rain and its distribution. The increase in aquifer levels could then be estimated and verified by sensors in the test wells. Time series data from each sensor would include the sensor ID, time stamp, and the amount of rain fall in millimeters/hour. Thus the data base would be populated with spatially distributed rainfall data. Temporal and spatial rainfall patterns could then be analyzed to describe the distribution of rail fall across the district.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Well Monitoring The 178 inspection-class wells located throughout the Jalgaon district are used to monitor ground water levels. The depth of the water in the wells is manually measured once a month by the Groundwater Survey Development Agency (GSDA). The monthly data is compiled and a report is provided by the GSDA once a year. Thus, this critical data is not available to authorities on a timely basis. Further, the manual methods of measurement and recording the data are prone to error, skewing the report and any applications or calculations made from the data. The limited number and locations of the wells monitored also adds to the difficulty of providing a clear picture of the district. Additionally, in this agro-industrial area, the water table is greatly affected by excessive pumping, which in turn impacts farmers’ yields as well as precious drinking water. Improved monitoring of ground water discharge and recharge is critical. An economical solution to inspection well monitoring is the use of low cost, wireless, smart sensors and software to automatically measure water depths, transmit the data, and generate frequent reports. The use of sensors could also increase the number of inspectionclass wells; which today is restricted by the number of technicians available to take and report the manual measurements. The well monitoring sensor system will include district wide, spatially distributed, time series aquifer depth data -- enabling a district wide aquifer map. The long term analysis of the aquifer data will enable estimations and predictions on water availability in the future. The model could also be refined by using the farmer electricity usage as an indicator of aquifer discharge (by irrigation pumping). Sensors for Well Water Depth There are two main technologies used to measure water depth in wells and dams. The first type of sensor emits ultra sonic waves and then measures the time required for the waves to bounce back; similar to the sonar technique used by ships to measure the ocean depth. The sensors float on the surface of the water and send a ping at specified intervals. The time required for the wave to travel down and then bounce back is recorded and the water depth is accurately calculated. As these sensors float on the surface of the water, they are easily accessible for routine maintenance. The second technology uses piezo electric pressure sensors. These sensors are submersed to the bottom of the water body. From there they measure the water pressure on the sensor. The depth of the water is then extrapolated from the pressure exerted by the water. The sensor uses an analog to digital converter to translate the analog signal into a digital format. The digital data is then processed and the pressure is converted to engineering units. Independent of the sensing technology used, all sensors
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
599
provide spatial time series data on water levels in wells. It the well is equipped with an electric pump, the electricity usage of that pump will provide the well discharge rate. The amount of time required to recharge the well is an indicator of water availability in the aquifer.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Reservoir and Dam Monitoring Monitoring is a very important task in appropriate dam management due to the economic, social and environmental significance of structures. It is fundamental in order to guarantee not only the safety of the structure and its users, but also to optimize the exploitation and maintenance of the dam. Monitoring the water level in dams is critical for many water uses; including agricultural, community, and drinking water for both humans and livestock. Dam monitoring is also used to manage the water between dams to maintain their capacity and intactness. If up-to-date reports are made available on a timely basis, they could be used to predict water levels, better manage water availability, improve warning systems, and take corrective measures when necessary. Accuracy of the data is critical; any false data or false alarms causes excessive burdens on the first responders and wastes precious water. Smart digital sensors are ideal for dam and reservoir monitoring. The automated sensors can sample the water level accurately and frequently so that up-to-date data is always available. The sensors can be configured to report on a regular schedule or report only special events, like the rate of change of in water levels. Dam monitoring will be functionally categorized into two groups: water reservoir monitoring and dam safety monitoring. These two groups are interrelated and the database for both will be common. The sensor data will be mined in central data center. The hydrological model will extract data from the central database and generate customized reports for the dam operators and policy makers. The Jalgaon district has complete cellular wireless coverage, thus, the backend connectivity can be economically achieved by using the existing cellular networks. A hardwired phone line could be used as a backup system for data transmission. To eliminate false alarms, backup sensors will be deployed to continuously compare data with each other and generate an alarm only if all applicable sensors in the network agree with the event. A combination of fuzzy logic and voting algorithms will be implemented to reduce false alarm and identify/isolate any faulty sensors. Sensors for Dam Monitoring Commercial grade, water ultrasonic depth sensors (described above) are also ideal for dam and reservoir monitoring. Alternatively, the submersible piezo electric pressure sensors could also be used dam monitoring. These sensors measure the pressure of the surrounding water while subtracting the atmospheric pressure to determine an exact pressure exerted by a column of water. The depth of the water is then extrapolated from the pressure measurement. These sensors also have a dynamic temperature compensation system, enabling high accuracy measurements over a wide temperature range. The submersible pressure sensors have a solid state transducer which is encapsulated in stainless steel housing and fully encapsulated with marine-grade epoxy to prevent moisture from leaking in. One commercially available submersible water level sensor uses a unique, highly flexible silicon diaphragm to interface between the water and the sensing element. The silicon diaphragm protects the sensor's electronics from moisture, providing each sensor with reliable linearity and eliminating issues associated with metal foil diaphragms (which tend to crinkle
600
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
and stretch out over time causing drift and linearity problems). This sensor also has automatic barometric compensation by using an attached vent cable which is protected by a stainless steel micro-screen cap to prevent fouling with silt, mud, or sludge. Both ultrasonic and submersible sensors can be easily integrated with embedded processors, data loggers, and telemetry equipment. The processor would control the operations of the sensor as well as backend communication with the database. For applications in Jalgaon, the ideal backend connectivity may be GSM digital cellular.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Girna Dam Hydrologic Monitoring and Observing System (GDHMOS) The primary objective of the Girna river sensor monitoring system is to design and implement a closed system to monitor and process water availability and usage of water in the Girna Dam water system. The term ‘closed system’ refers to the scope of the system. The scope of the monitoring process will start with the rain fall into the river, its flow into the dam, and its passage all the way through the exit channels of the dam. The scope could also be expanded to take corrective actions based on sensor and user inputs. Although the heart of the system will be the Girna Dam and Girna Reservoir, sensing and monitoring will include recharging from the Girna River and remote monitoring of the rain fall in the recharging area, as well as, water discharge from the dam gates and monitoring of the rivers down stream. The sensors will monitor the water volume (depth) of the reservoir, enable the calculation of hydraulic conductivity for dam safety, and keep track of the structural integrity of the dam. The sensors in the Girna River (the river feeding the reservoir) will include multi-layered, side-looking, Doppler current meters with variable range. These sensors can be easily installed on the bank of the river to measure the velocity of the water in horizontal layers so that accurate water flow can be measured. They can also be used to determine the volume of water entering the reservoir. Rain sensors deployed within a 5000 sq km recharge zone area (or catch basin) will monitor the amount of rain fall and provide a measurement of “anticipated additional water” expected to arrive at the reservoir and the anticipated time when the water will arrive to the reservoir. Sensors at the dam water discharge end (tow end) will measure the amount of water leaving the Girna Dam system to the rivers/channels down stream and will enable an estimate of equilibrium or shift in equilibrium in the water level. Dam Safety Monitoring The second objective of GDHMOS is to monitor the safety of the dam. The sensor monitoring system will include safety sensors to monitor the structural integrity of the dam, as well as the hydraulic conductivity of the foundation and surrounding walls, to ensure dam safety. A wireless network of ultrasonic and piezo electric depth sensors installed at the foundation of the dam and in the inspection wells will continuously monitor the ground water flow under the dam. The dam design parameters and the constant monitoring of hydraulic conductivity will be used to determine the safety of the dam and avoid piping condition. The structural safety of the dam will be monitored by a network of crack measuring sensors, strain gauges, and fiber optic foundation deformation sensors. Any minute changes in the dam structure could be quickly identified and reported. The sensor system will help ensure that the dam is operating under safe conditions and provide information to anticipate any adverse events within given conditions; like weather, water volume, discharge rate, cracks in the dam, and seepage variation.
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
601
With the set of data discussed above, engineers and policy makers could better assess, estimate, and plan water flow and usage – and the effects it may have on people down stream from the dam. For example, early warnings could be issued when the dam gates must be opened to eliminate excessive water. People living in a ‘danger zone’ could be put on alert, and with pre-planning, the water could be released gradually. Thus, any disruption downstream could be effectively eliminated or managed.
Intelligent Sensors and Software The data from the sensors will be sampled at rates specified by the user or sampled on events-based conditions. In an event-driven situation, the sensors would detect an abnormal situation and then automatically alter the sampling rate, thereby providing data a higher resolution. The heart of the monitoring system will be specialized intelligent sensors that can react to the environmental conditions and central data mining software capable of performing analysis and compiling reports. The report distribution facility will include computerized custom reports.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Agricultural, Forest and Weather Monitoring The economy of the Jalgaon district is based on agriculture and the agro industry in the region – which makes the accurate monitoring of rain fall and water levels for agricultural irrigation and food processing even more critical. Monitoring needs to be on both a micro and macro level and is the foundation for related applications, like ground moisture, disease outbreaks or forecasts, and pest control. For instance, cropping patterns are based on the prevailing rainfall as measured through the current rain gauge methods and the GSDA inspection wells (which are often inaccurate and out-of-date). The forest department in this region also depends on this critical (yet often flawed today) data to manage its resources. Sensors for Agro/Forest Monitoring Moisture sensors can play a vital role in helping the farmer determine planting patterns for gram and rabbi crops. In addition to aiding the farmers on the most advantageous sowing times, other benefits include improved planning for seed availability and fertilizers at appropriate times and places, depending on the moisture condition at specific locations. In addition to the moisture sensors for the soil, improved methods for monitoring rainfall, dams, wells, and irrigation reservoirs with sensor networks would be a great service to the farming community and economy of the Jalgaon district. This information could easily be distributed to farmers and forest managers. Sensors for Micro-Weather Station and Early Warning System A typical weather station includes anemometer sensors to measure wind speed and wind direction; electronic rain gauges; sensors for barometric pressure changes; and humidity sensors for relative humidity, heat index, and dew point. Micro-weather station sensors can provide timely and critical weather measurement data. For instance, a plunging barometer suggests an incoming storm and an anemometer sensor can capture threateningly-high wind gusts or dangerously low wind chill conditions. A chain of micro weather stations could be established in the district and integrated with the water management network. This would allow the establishment of a robust early warning system. These sensors can be integrated
602
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
with an embedded controller to locally process the sensor data and issue warnings/alarms if the data presented by the sensor exceeds normal threshold. The data collection and reporting scheme could also be remotely monitored and changed in real-time. The location of each sensor node would be identified by its sensor address. As this would be a static network, the location of each sensor would be pre-determined, enabling DSS to conduct analyses based on the geographic location of each sensor. Since each sensor in the sensor network would have a unique address, each sensor could be individually addressed and configured. This configuration could be done locally or remotely from a central unit. The remote connectivity and the local microprocessor based controller would provide two way command and control capabilities to each sensor within a sensor node and within the sensor network. A sensor package could be customized based on the regional needs. Each weather sensor node would contain one or more analog or digital sensors. Each individual sensor would be controlled by a microprocessor responsible for recording and transmitting the data. The software in the controller could also be configured to apply intelligent knowledge and generate critical alarms if the weather sensor readings deviate from a normal, pre-defined range. The efficiency of a digital governance system for policy decisions and availability of data to the public could be significantly improved with the use of smart wireless sensors nodes and sensor networks. In a dynamic environment, data becomes ineffective if it is not processed as soon as it is received. Sensors and sensor networks will provide timely data and make the decision support systems used by policy makers more robust; include emergency and disaster responses, forecasts and planning, and resource management of water, irrigation, and forests. The services of the administration could be better focused and tuned towards the needs of the community and people and the administration could more quickly respond to existing and emerging hot-spots because of the availability of fast and accurate data. Thus, the use of sensors and sensor networks in government operations would all greatly benefit the regional public, agricultural industry, and commerce.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Futuristic Vision Geoinformatic Surveillance Decision Support System Computational structure, system integration and database management - This component of the project focuses on the development of efficient data structures and algorithms coupled with effective visualization techniques for hotspot detection and prioritization using statistical methodologies developed in the project. In fact, we have recently addressed the problem of quickly identifying regions for large scale multivariate maps for which a number of geospatial parameters satisfy certain conditions (See JaJa and Shi (2001)). We will extend these techniques in a number of directions suggested by the proposed scanning techniques and prioritization tools. Information visualization, user interface design, and GIS linkage - A major goal is to develop a visualization interface integrated with the statistical software tools developed in this project. Information visualization and interface design are critical for effective use of these tools. A phased implementation will allow us to implement simple algorithms at first and then embed more sophisticated algorithms. As our implementations mature, we will conduct
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
603
usability tests in coordination with the specialists to refine the interfaces and demonstrate efficacy.
Integration of Research, Education and Dissemination An essential part of this project is to introduce methods and tools at the core of the upper level scan statistic system to hotspot analysis researchers in various agencies. Constant interactions among the participating researchers and partners will ensure the development of techniques and tools tailored to address the needs of the involved federal agencies and other partners. In graduate education, we will integrate the techniques and methods into the wide range of related graduate courses offered. Graduate students will test and validate various tools as they become available through the project. Also, the graduate students will contribute to the tutorials offered during each summer workshop, in addition to presenting their research progress. Every effort will be made to iteratively accomplish the upward spiral of horizontal and vertical research and training integration. For effective technology transfer, we plan: monographs, case books, thematic journal issues, research workshops and tutorials, and distributed information management.
ACKNOWLEDGMENTS This material is based upon work partially supported by (i) The National Science Foundation under Grant No.0307010 (ii) The US Environmental Protection Agency under Grant No. CR -83059301 and No. R-828684-01; and (iii) The US Army Research Office Award W911NF-07-0376. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the sponsoring agencies.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
REFERENCES Ailamaki, A., Faloutsos, C., Fischbeck, P. S., Small, M. J. and VanBriesen, J. (2003). An environmental sensor nework to determine drinking water quality and security. SIGMOD Record 32(4): Aldous, D. (1987). On the Markov chain simulation method for uniform combinatorial distributions and simulated annealing. Probability in the Engineering and Informational Sciences 1:33-46. Biswas, P.K. and Phoha, S. (2006). Self-organizing sensor networks for integrated target surveillance. IEEE Transactions on Computers 55(8):1033-·1047. Bithell, I. F., Dutton, S. I., Neary, N. M., and Vincent, T. I. (1995). Controlling for socioeconomic confounding using regression methods. Community Health 49:S15-Sl9. Brightwell, G. and Winkler, P. (1991). Counting linear extensions. Order 8: 225-242. Brooks, R. and Griffin, C. (2002). Traffic model evaluation ol ad hoc target tracking algorithms. Internet. J. High Performance Computer Applications (Special issue on Sensor Networks) Hi(3): 221-234.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
604
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
Brooks, R. (1996). Reliable sensor fusion algorithms: Calibration and cost minimization, Ph.D. Dissertation, LSD. Charnes, A., Cooper, A. Y., Lewin, A. Y., and Seiford, L. M. (1994). Data Envelopment Analysis: Theory, Methodology, and Applications. Kluwer, Boston. Cormen, T. H., Leierson, C. E., Rivest, R. L. and Stein, C. (2001). Introduction to Algorithms, Second Edition. MIT Press, Cambridge, Massachusetts. Cressie, N. (1991). Statistics for Spatial Data. Wiley, New York. Di Battista, G., Eades, P., Tamassia, R., and Tollis, I. G. (1999). Graph Drawing: Algorithmsfor the Visualization of Graphs. Prentice Hall, Upper Saddle River, New Jersey. Doucet, A., de Freitas, N. and Gordon, N. (2001). Sequential monte carlo methods in practice. Springer, New York. Dwass, M. (1957). Modified randomization tests for nonparametric hypotheses. Annals of Mathematical Statistics 28: 181-187. Filar, I. A., and Ross, N. P. (2001). Generalized data envelopment analysis, and environmental indicators. Invited Paper: Plenary Forum on Environmental Indicators and their Integration for Quality of Life. Index 2001 Congress. Rome, Italy. Fishburn, P. C. (1985). Interval Orders and Interval Graphs: A Study of Partially Ordered Sets. Wiley, New York. Friedlander D. and Phoha, S. (2002). Semantic information fusion for coordinated signal processing in mobile sensor networks. Int. Journal of High Performance Computing Applications (Special Issue on Sensor Networks) Hi (3): 235-241. Haggstrom, O. (2002). Finite Markov Chains and Algorithmic Applications. Cambridge University Press, Cambridge. JaJa, J. and Shi, Q. (2001). Efficient Techniques for Exploring Geospatial Data. Johnson, G. D. and Patil, G. P. (2007). Landscape Pattern Analysis for Assessing Ecosystem Condition. Environmental and Ecological Statistics Series 1:99-127. Knuth, D. E. (1973). The Art of Computer Programming. Volume 1, Fundamental Algorithms, Second Edition. Addison-Wesley, Reading, Massachusetts. Kulldorff, M. (1997). A spatial scan statistic. Communications in Statistics: Theory and Methods 26:1481-1496. Kulldorff, M. and Nagarwalla N. (1995). Spatial disease clusters: Detection and inference. Statistics in Medicine 14: 799-810. Kulldorff, M., Feuer, E. J., Miller, B. A., and Freedman, L. S. (1997). Breast cancer clusters in Northeast United States: A geographic analysis. American Journal of Epidemiology146:161-170. Kumbhakar, S.C. and Knox Lovell, C. A. (2000). Stochastic Frontier Analysis. Cambridge University Press, Cambridge. Lehmann, E. L. (1986). Testing Statistical Hypotheses, Second Edition. Wiley, New York. Lind, D. and Marcus, M. (1995). An introduction to symbolic dynamics and coding. Cambridge University Press, Cambridge. Mostashari, F., Kulldorff, M., Hartman J.J., Miller, J.R., Kulasekera V. (2003). Dead bird clusters as an early warning system for West Nile virus activity. Emerg. Infect. Dis. 9: 641-646.
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Digital Governance, Hotspot Geoinformatics, And Sensor Networks …
605
Myers W.L. and Patil G.P. (2007). Pattern-Based Compression of Multi-Band Image Data for Landscape Analysis. Environmental and Ecological Statistics Series: Springer, Boston Volume 2, 170p. Myers, W. L., Kurihara, K., Patil, G. P., and Vraney, R. (2006). Finding upper level sets in cellular surface data using echelons and SaTScan. Environmental and Ecological Statistics 13(4): 379-390. National Academy of Sciences (2002). Predicting invasions of non-indigenous plants and plant pests. National Research Council, 198 p. Neggers, J. and Kim, H. S. (1998). Basic Posets. World Scientific, Singapore. Pathria, R. K. (1996). Statistical mechanics. Butterworth Heinmann, Oxford. Patil, G. P. (2002). Next Generation of Potential Outbreak Detection and Prioritization System. Invited comment and discussion. National Syndromic Surveillance Conference, September 2002, New York City. Patil, G. P., and Taillie, C. (2004b). Multiple indicators, partially ordered sets, and linear extensions: multi-criterion ranking methods. Environmental and Ecological Statistics 11(2): 199-228. Patil, G. P., and Taillie, C. (2004a). Upper Level set scan statistic for detecting arbitrarily shaped hotspots. Environmental and Ecological Statistics 11(2): 183-198. Patil, G. P., Balbus, J., Biging, G., JaJa, J., Myers, W. L., and Taillie, C. (2004a). Multiscale Advanced Raster Map Analysis System: Definition, Design and Development. Environmental and Ecological Statistics 11(2): 113-138. Patil, G. P., Bishop, J., Myers, W. L., Taillie, C., Vraney, R., and Wardrop, D. H. (2004b). Detection and delineation of critical areas using echelons and spatial scan statistics with synoptic cellular data. Environmental and Ecological Statistics 11(2): 139-164. Patil, G. P., Myers, W. L., Taillie, C., and Wardrop, D. (2002) Hotspot Detection and Early Warning for Synoptic and Network-Based Syndromic Surveillance. Invited Poster Presentation, National Syndromic Surveillance Conference, September 2002, New York City. Phoha, S., Koch, J., Grele, E., Griffin, C. and Madan, B. (2005). Space time coordinated distributed sensing algorithms/f) I' resource efficient narrowband target localization and tracking. Int. J. of Distributed Sensor Networks 1(1): 81-99. Phoha, S., LaPorta, T. and Griffin, C. (eds.). Sensor network operations. Ch. 14, IEEE/Wiley, March 2006, ISBN 0-471-71976-5, EAN 9780471719762, 783 pages. Phoha, S., Peluso, E. and Brooks, R.R. (1999). A constructivist theory of distributed intelligent control of complex dynamic systems. JFACC Symposium, Nov. 15-16 1999. San Diego, CA. Phoha, S., Peluso, E. and Culver, R.L. (2002). A high fidelity ocean sampling mobile network (samon) simulator. IEEE Journal of Oceanic Engineering (Special Issue on Autonomous Ocean Sampling Networks) 26(4): 646--653. Press, W. H., Teukolsky, S. A., Vetterling, W. T., and Flannery, B. P. (1992). Numerical Recipes in C, Second Edition. Cambridge University Press, Cambridge. Rajagopalan, V. and Ray, A. (2006). Wavelet-based space partitioning for symbolic time series analysis. Signal Processing 86(11): 3309-3320. Ray, A. (2004). Symbolic dynamic analysis of complex systems for anomaly detection. Signal Processing 84(7):1115-1130.
606
Ganapati Patil, Sanjay Pawde, Shashi Phoha et al.
Ray, A., Phoha, V. and Phoha, S. (2005). Quantitative measure for discrete event supervisory control. Springer, New York. Rogerson, P. A. (2001). Monitoring point patterns for the development of space-time clusters. Journal of the Royal Statistical Society, Series A 164: 87-96. Schimel, D. (2007). National ecological observatory network (neon): Research design basis for the neon relocatable systems. NSF Document Number: NEONDESIGN-0001vA. Srinivasan, G., Sundareshwar, P.V., Murtugudde, R., et al., Environmental monitoring network for India. AAAS Science, Policy Forum, www.sciencemag.org 316 (13 April 2007). Trotter, W. T. (1992). Combinatorics and Partially Ordered Sets. Johns Hopkins University Press, Baltimore. Waller, L. (2002). Methods for detecting disease clustering in time or space. In: Monitoring the Health of Population: Statistical Methods and Principles in Public Health Surveillance (Eds. Brookmeyer and Stroup), Oxford University. Walsh, S. J., and DeChello, L. M. (2001). Geographical variation in mortality from systemic lupus erythematosus in the United States. Lupus 10: 637-646. Walsh, S.I. and Fenster, I.R. (1997). Geographical clustering of mortality from systemic sclerosis in the Southeastern United States, 1981-90. The Journal of Rheumatalogy 24(12): 2348-2352.
QUESTION BANK Short Answer Questions
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
1. Explain the term watershed prioritization. Why is this important? 2. What are the linkages between disturbance and vulnerability for developing conservation strategies? 3. What is importance of pattern based compression of multi-band image data? 4. What is scan statistics? What are its limitations? 5. Write a short note on landscape pattern analysis for assessing ecosystem condition.
Long Answer Questions 1. What is sensor network? What are the building blocks of sensor network? How could it be used to water resource management? 2. What is importance of multi-disciplinary research using geospatial tools? Explain the importance of this in watershed management. 3. What is geoinformatic surveillance decision support system? What are its challenges and applications?
INDEX # 1G, 119
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
A Aβ, 307 AAAS, 606 AAS, 329 abiotic, 245 absorption, 33, 37, 38, 62, 254, 262, 306, 321, 338, 339, 432, 442 Abundance, 457, 458 academic, 77, 499, 525, 526, 529, 575 academics, 500, 510 accessibility, 82, 159, 168, 415, 419, 478, 483, 489, 506, 533, 563, 565 accidental, 574, 575 accounting, 2, 148, 210, 259, 261 accuracy, 2, 66, 74, 75, 101, 115, 134, 146, 147, 159, 171, 189, 191, 200, 203, 206, 209, 220, 221, 222, 225, 239, 269, 274, 294, 295, 305, 306, 313, 314, 328, 360, 377, 384, 386, 387, 388, 389, 390, 393, 394, 415, 418, 449, 472, 500, 541, 542, 555, 559, 560, 562, 570, 571, 573, 575, 589, 593, 594, 599 acid, 38 ACM, 493 acoustic, 99, 103, 588 acquisitions, 342, 393, 402 activation, 53, 362 activation energy, 362 actuation, 594 actuators, 588 acute, 306, 412 Adams, 46, 61, 63, 217, 243 adaptation, 23, 588, 589, 594 adaptive control, 352, 588
adequate housing, 428 adjustment, 302, 344, 386, 395, 406 administration, 426, 595, 602 administrative, 6, 170, 311, 417, 430, 433, 510, 513, 515, 516, 518 administrators, 481, 594, 597 adult, 87 adverse event, 600 aerobic, 270 aerosol, 266 aerosols, 75 aesthetics, 421 Africa, 10, 164, 330, 413 age, 242, 249, 255, 269, 271, 273, 275, 397, 499, 573, 574, 580 agent, 493, 559, 568 agents, 52, 339, 464, 465 aggregates, 344 aggregation, 77, 83, 87, 103, 323, 518 aging, 274 agrarian, 131 agricultural, 1, 22, 34, 61, 71, 128, 131, 133, 137, 156, 169, 187, 191, 209, 245, 270, 271, 277, 281, 291, 292, 293, 295, 299, 305, 306, 318, 319, 322, 324, 426, 430, 450, 465, 504, 506, 507, 543, 595, 596, 599, 601, 602 agricultural crop, 61, 319 agriculture, ix, 4, 5, 60, 111, 127, 139, 146, 170, 219, 226, 227, 232, 233, 237, 291, 292, 295, 296, 299, 303, 306, 324, 329, 330, 429, 523, 540, 542, 554, 578, 588, 601 agroforestry, 246 aid, 78, 129, 147, 249, 281, 421, 425, 428, 432, 465, 487 aiding, 601 air, 22, 75, 113, 114, 123, 124, 250, 252, 390, 429, 432, 454 Air Force, 381 Aircraft, 278, 287
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
608
Index
airports, 570 Alberta, 342 algae, 24, 25, 63 Algal, v, 4, 9, 10, 27, 30 algorithm, 14, 53, 54, 104, 122, 128, 209, 210, 212, 257, 265, 266, 267, 275, 287, 305, 306, 309, 310, 311, 312, 314, 315, 316, 354, 356, 365, 374, 377, 438, 439, 459, 460, 467, 473, 474 alluvial, 300, 302, 395 alpha, 468 Alps, 346 alternative, 1, 6, 35, 60, 76, 92, 177, 215, 257, 261, 420, 483, 499, 501, 502, 503, 504, 510, 581 alternatives, 2, 485, 515, 550 alters, 430 Amazon, 38, 212, 542, 543 Amazonian, 553 ambient air, 114, 454 ambient air temperature, 114, 454 ambiguity, 136, 559 ammonium, 338 amorphous, 338, 594 amplitude, 81, 392, 394, 399, 404 Amsterdam, 27, 30, 104, 105, 106, 346 anaerobic, 270 analog, 320, 596, 598, 602 analysis of variance, 61 analytical framework, 593 analytical models, 588, 589 analytical techniques, 500 anatomy, 470 Andes, 241 animations, 481, 493 anisotropy, 96, 254 ANN, 52, 53, 54, 55, 56, 58, 349, 377, 382 annealing, 603 annual rate, 212, 245 anomalous, 24 ANOVA, 355 antenna, 387, 388, 391, 392 anthropogenic, 71, 186, 193, 216, 250, 270, 281, 385, 391, 407, 427, 454, 459, 465, 468, 539, 542, 544, 555 API, 520 Appalachian Mountains, 246 application, xi, 1, 2, 4, 5, 6, 35, 41, 42, 67, 68, 72, 74, 80, 83, 88, 89, 90, 96, 97, 100, 102, 103, 104, 114, 161, 181, 185, 249, 291, 292, 349, 354, 374, 381, 382, 404, 406, 408, 464, 470, 475, 477, 483, 484, 491, 500, 515, 518, 519, 530, 531, 532, 533, 535, 539, 542, 548, 550, 551, 552, 554, 560, 561, 562, 569, 577, 580, 581 appropriate technology, 291
aquaculture, 9, 10, 22, 24, 27, 29, 72, 105 Aquifer, v, 151, 154, 160 aquifers, 152, 160, 407, 596 Arabia, 405 Argentina, 4, 33, 34, 35, 46, 62, 70, 345, 415, 422 argument, 476 arid, 245, 299, 318, 330, 333, 334, 344 Arizona, 242 Army, 603 arrest, 1 artificial intelligence, 52 ash, 219 ASI, 30 Asia, 10, 27, 29, 133, 186, 209, 291, 318, 320, 322, 329, 330, 413, 479, 535 Asian, 5, 27, 29, 182, 284, 302, 320, 328, 329, 413 Asian countries, 328, 413 assessment, ix, 2, 3, 4, 5, 62, 66, 68, 104, 129, 132, 149, 163, 166, 171, 185, 186, 191, 193, 208, 210, 212, 213, 222, 225, 233, 285, 292, 294, 297, 299, 302, 318, 323, 326, 329, 330, 342, 350, 442, 467, 468, 469, 470, 472, 500, 524, 525, 527, 552, 553, 559, 561, 571, 574, 575, 576, 580 assessment models, 350 assignment, 91 assimilation, 100, 152, 256, 259 assumptions, 59, 114, 328, 377, 429, 500, 501, 502, 506, 507, 510, 512, 590 ASTM, 346 asymptotic, 52, 122 asymptotically, 57 asynchronous, 492, 588, 593 Athens, 470 Atlantic, 10, 24, 27, 34, 85, 86, 98, 104, 286, 578 Atlas, 329 atmosphere, 3, 39, 74, 75, 112, 122, 186, 211, 249, 250, 252, 259, 263, 264, 269, 273, 277, 278, 280, 281, 387, 392, 393, 394, 395, 432, 433, 466 atmospheric deposition, 272 atmospheric pressure, 386, 599 attribution, 74 Australia, 30, 242, 245, 276, 382, 429, 512, 534, 535 Austria, 346, 493, 495, 496 authentication, 490 authenticity, 149 autocorrelation, 126, 127 automata, 3, 590, 593 automation, 33 autotrophic, 251, 272 availability, ix, 6, 34, 59, 79, 115, 131, 133, 146, 181, 268, 269, 278, 279, 293, 301, 328, 335, 341, 344, 384, 393, 408, 457, 464, 506, 510, 513, 514,
Index 515, 518, 526, 528, 533, 570, 587, 595, 598, 599, 600, 601, 602 averaging, 40, 76, 125 awareness, 180, 525, 529
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
B backfire, 539, 555 background information, 27 backscattered, 13, 392, 399 backscattering, 394, 408 bacterial, 38 Bali, 182, 540, 552, 555 bandwidth, 594 Bangladesh, 240 banks, 137, 209 barrier, 489, 492 barriers, 160, 169, 516, 528, 529, 533, 536 base year, 191 bathymetric, 95 Bayesian, 159, 374, 381, 425, 429, 439, 449, 451, 459, 465, 473 Bayesian methods, 159 Bayesian theory, 451 behavior, 5, 53, 59, 125, 159, 355, 361, 362, 480, 503, 504, 506, 510, 525, 535, 541, 542, 575, 590 Beijing, 27, 28, 147, 181, 430 bell, 363 benchmark, 386 benefits, 5, 71, 73, 275, 292, 295, 492, 503, 514, 593, 601 Best Practice, 182 bias, 209, 210 biochemistry, 255 biodiversity, xi, 4, 5, 68, 101, 186, 189, 200, 206, 210, 215, 217, 218, 234, 239, 240, 241, 242, 244, 539, 540, 542, 543, 544, 548, 550, 551, 552, 555, 574, 575 biological interactions, 107 bioluminescence, 27 biomass, 5, 10, 29, 33, 34, 68, 92, 93, 94, 95, 100, 107, 122, 186, 217, 218, 224, 239, 245, 249, 250, 251, 257, 264, 269, 270, 272, 281, 301, 321, 442 biophysics, 259, 264 biosphere, 211, 249, 250, 251, 252, 264, 276, 277, 278, 280, 281, 442, 468 biotic, 34 birds, 219, 242, 246, 452, 465 birth, 420 black body, 443 Black Sea, 218 blocks, 167, 591, 592 blog, 495
609
body temperature, 462 boils, 254 Bolivia, 61, 244 bootstrap, 62, 464 boreal forest, 265, 266, 269 boreholes, 161 Boston, 534, 554, 604, 605 boundary conditions, 259, 288 brain, 52 Brazil, 415, 422, 431, 466, 467, 543 Brazilian, 212 breakdown, 228, 239 breeding, 242 Britain, 105 British Columbia, 10, 245, 494, 534 broad spectrum, 6, 513, 515 browser, 477, 482, 486, 532 Brussels, 423 Buenos Aires, 4, 33, 34, 35, 41, 61, 62, 63 buffer, 178 building blocks, 518, 554, 606 buildings, 139, 168, 173, 178, 413, 415, 417, 418, 419, 430, 448, 450, 453, 485, 488, 489 burn, 543 burning, 250, 269, 270, 281
C Ca2+, 35 CAD, 484, 488 calibration, 70, 74, 76, 159, 259, 444, 561, 589 Cambodia, 318, 324, 326, 327, 328 Cambodian, 326 Canada, 27, 68, 104, 263, 266, 268, 269, 271, 274, 275, 276, 277, 278, 279, 280, 282, 283, 285, 286, 330, 333, 335, 339, 340, 341, 342, 343, 345, 346, 468, 534 canals, 169 cancer, 421, 604 candidates, 394, 578 Canonical Correlation Analysis, 51 capacity building, ix, 528, 529 carbon, vi, 5, 79, 186, 212, 213, 215, 216, 217, 218, 221, 224, 225, 234, 237, 238, 239, 240, 242, 244, 245, 246, 247, 249, 250, 251, 255, 263, 271, 272, 274, 275, 276, 277, 280, 281, 282, 283, 284, 285, 286, 287, 289, 321, 540, 542, 551, 552, 553, 554, 555, 575, 579 carbon cycling, 271 carbon dioxide, 250, 280, 551 carbon emissions, 217, 551 carbonates, 338 Caribbean, 413
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
610
Index
carotene, 270 carrier, 383, 388, 389, 407, 408 CAS, 26 case study, 73, 133, 167, 182, 212, 215, 225, 241, 243, 246, 297, 302, 330, 335, 346, 355, 358, 422, 466, 469, 470, 471, 515, 517, 523, 524, 525, 533, 545, 551, 574, 579 cast, 417 catastrophes, 169 catchments, 131, 133, 144, 245, 430, 450, 452 categorization, 33, 41, 428, 439 cattle, 139, 212 cell, 50, 75, 87, 143, 160, 161, 334, 393, 394, 580, 581, 583, 589, 596, 597 cement, 174 Census, 187, 211, 414, 459 Central America, 285 Central Asia, 413 central planning, 178 CERES, 251 certificate, xi CES, 470 Chad, 469 channels, 35, 322, 356, 366, 431, 432, 481, 600 children, 167 Chile, 338 China, v, 4, 9, 10, 11, 12, 13, 24, 25, 27, 28, 29, 30, 31, 169, 241, 245, 246, 268, 274, 275, 276, 279, 284, 287, 318, 324, 327, 328, 431, 469, 470, 471 Chl, 12, 14, 20, 21, 24, 77, 99 chloride, 156, 160 chlorophyll, 28, 29, 34, 45, 51, 52, 56, 58, 59, 61, 71, 74, 75, 79, 80, 87, 98, 100, 107, 109, 110, 122, 264, 270, 339, 442 Chlorophyll-a, 35, 57, 77 circular flow, 514 circulation, 11, 12, 20, 22, 28, 29, 70, 71, 79, 80, 81, 85, 103, 105 citizens, 4, 6, 475, 476, 477, 479, 481, 485, 487, 489, 490, 491, 492 civil engineering, 350 civil society, 476 classes, 54, 136, 141, 147, 185, 186, 191, 206, 207, 209, 222, 223, 225, 232, 233, 237, 293, 294, 295, 298, 305, 306, 307, 309, 313, 341, 352, 382, 429, 430, 433, 440, 442, 444, 457, 459, 460, 461, 464, 472 classical, 85, 92, 159, 164, 307, 354, 388, 407 classification, 4, 5, 33, 35, 41, 45, 48, 49, 53, 54, 55, 60, 62, 77, 90, 91, 102, 136, 138, 191, 206, 209, 210, 212, 213, 220, 222, 225, 226, 234, 237, 263, 267, 293, 294, 295, 296, 302, 305, 306, 307, 309, 314, 315, 316, 339, 340, 341, 344, 381, 415, 425,
428, 429, 430, 432, 436, 437, 439, 440, 449, 460, 464, 465, 467, 468, 469, 472, 473, 474, 481, 570, 595 clay, 38, 137, 139, 219, 338, 341, 385, 395, 397, 398, 488 Clean Development Mechanism, 186, 540 clients, 500, 532 climate change, 3, 5, 24, 27, 69, 105, 166, 169, 186, 217, 249, 250, 272, 280, 281, 285, 318, 425, 540, 554, 555, 575 Climate Change Science Program (CCSP), 246 climatic factors, 1 climatology, 83, 470 closed-loop, 589 closure, 94, 114, 125, 225, 228, 229, 238, 239, 386, 582 clouds, 70, 75, 84, 109 cluster analysis, 429 clustering, 293, 309, 310, 314, 316, 427, 579, 588, 606 clusters, 306, 309, 311, 313, 314, 315, 431, 577, 579, 580, 581, 593, 604, 606 Co, 103, 107, 129, 284, 341, 364, 365, 377, 382, 470, 477, 491 CO2, 249, 250, 259, 264, 270, 272, 273, 274, 276, 277 coal, xi, 139, 381, 385, 400, 401, 406, 407, 514 coal mine, 381, 407 coalfields, 401 coastal areas, 385 coastal zone, 81 codes, 159, 388, 592 coding, 77, 483, 528, 604 coefficient of variation, 146, 224 coffee, 246 coherence, 269, 342, 344, 394, 399, 404, 408 collaboration, 475, 487, 517, 518, 528, 575 Colombia, 241 colors, 38 combined effect, 550 commerce, 4, 602 communication, 2, 4, 177, 209, 292, 425, 465, 475, 476, 478, 481, 486, 487, 488, 489, 490, 491, 492, 496, 500, 518, 526, 529, 530, 532, 533, 588, 590, 593, 595, 600 communication processes, 486 communication systems, 209 communication technologies, 292 communities, 23, 30, 34, 166, 168, 169, 177, 180, 181, 183, 246, 485, 487, 503, 517, 524, 525, 529, 574 community, ix, 4, 10, 69, 70, 89, 108, 139, 154, 165, 166, 167, 168, 171, 177, 180, 182, 183, 184, 219,
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Index 259, 269, 479, 488, 501, 502, 503, 505, 506, 507, 510, 517, 520, 521, 525, 528, 533, 571, 595, 599, 601, 602 compaction, 385, 397, 407 compatibility, 6, 27, 533 compensation, 104, 401, 402, 403, 599, 600 competence, 500 competition, 1, 306 complement, 333, 579 complex systems, 52, 605 complexity, 2, 69, 112, 132, 223, 249, 256, 270, 321, 326, 338, 356, 366, 377, 433, 466, 512, 593 components, 2, 39, 46, 99, 112, 132, 217, 239, 272, 295, 339, 340, 384, 390, 400, 402, 403, 405, 429, 430, 431, 437, 440, 444, 497, 502, 512, 517, 518, 519, 520, 526, 528, 530, 532, 533, 536, 541, 560, 562, 569, 582, 583, 588, 591 composites, 464 composition, 1, 27, 30, 74, 132, 216, 223, 239, 241, 242, 335, 341, 399, 469, 488 compounds, 38 comprehension, 71, 587 compressive strength, 355, 356, 357 computation, 53, 54, 86, 140, 150, 159, 161, 385, 390, 432, 450, 460, 588, 590 computer graphics, 485 computer science, 180, 481, 582 computer simulations, 487 computer software, 151 computer systems, 152 computer technology, 510 computing, 5, 132, 133, 139, 147, 148, 149, 331, 349, 350, 351, 355, 356, 380, 381, 382, 442, 510 concentrates, 508 concentration, 10, 12, 14, 34, 35, 45, 56, 58, 59, 61, 71, 74, 75, 77, 78, 79, 94, 98, 99, 109, 110, 249, 250, 259, 270, 276, 336, 355, 413, 426, 427, 431, 568, 569, 579 conception, 488, 492 conceptual model, 133, 134, 244, 514, 515, 518, 524, 526, 533 conceptualization, 260, 261 concordance, 49, 51 concrete, 174, 561 conductance, 129, 259, 260, 261, 264, 279, 288 conduction, 515 conductivity, 35, 391, 408, 600 confidence, 59, 60, 85, 179, 249, 380, 563, 565, 566, 571, 575, 577, 583, 584 confidentiality, 92 configuration, 56, 75, 217, 223, 233, 239, 364, 392, 568, 569, 602 conflict, 1, 216, 486, 540
611
conformity, 294 confusion, 40, 222, 313, 355, 430, 560 Congress, iv, 244, 346, 495, 604 conifer, 263, 266, 268, 274 coniferous, 278 connectivity, 103, 210, 223, 452, 574, 590, 599, 600, 602 consciousness, 476 consensus, 476, 487, 501, 510 conservation, 1, 5, 6, 69, 101, 132, 134, 138, 148, 187, 206, 215, 234, 239, 240, 242, 243, 244, 296, 297, 302, 317, 318, 324, 326, 328, 329, 539, 540, 541, 542, 543, 544, 545, 548, 549, 550, 551, 552, 553, 554, 555, 556, 588, 606 consolidation, 385 constraints, 151, 153, 154, 168, 261, 295, 299, 445, 472, 588 construction, 33, 77, 100, 131, 138, 169, 174, 178, 239, 306, 339, 344, 396, 416, 420, 421, 483, 486, 500, 523, 590 construction materials, 169, 178 constructivist, 605 consulting, 335, 503 consumption, 10, 427, 450, 454, 518 contaminant, 161 contaminants, 44, 571 contamination, 74, 100, 114, 420 continental shelf, 78, 82, 85, 99 contingency, 49, 50 continuity, 70, 93, 95, 96, 427 control, 6, 35, 74, 87, 88, 151, 152, 154, 155, 222, 223, 241, 250, 260, 278, 298, 323, 351, 352, 356, 364, 365, 382, 405, 437, 477, 496, 533, 559, 561, 562, 570, 579, 588, 589, 591, 600, 601, 602, 605, 606 convergence, 53, 79, 80, 354, 364, 418 conversion, 39, 77, 100, 169, 224, 243, 250, 452, 522, 534 conviction, 166 Copenhagen, 483, 486, 493, 494, 495, 496 Coping, 181 copper, 335, 338 copyrights, 577 Coriolis effect, 81 corporations, 543 correlation, 86, 93, 125, 126, 127, 128, 132, 166, 240, 264, 356, 365, 374, 376, 377, 378, 393, 394, 405, 438, 455, 458, 460, 461, 465, 567, 596 correlation analysis, 86, 128, 567 correlation coefficient, 356, 365, 374, 394, 458, 460, 461, 465, 567 correlation function, 93, 125, 126 correlations, 51
Index
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
612
corridors, 427, 462, 464, 465, 574, 575 cosine, 257, 258, 393 cost minimization, 604 Costa Rica, 212, 554 cost-effective, 100, 101, 392, 402 costs, 101, 177, 270, 292, 429, 430, 454, 489, 491, 492, 503 cotton, 302 couples, 278, 280 coupling, 133, 249, 279, 289, 523 covering, 45, 71, 86, 94, 188, 200, 306, 318, 320, 394 cows, 139 crab, 108 crack, 600 CRC, 103, 104, 105, 106, 535 creativity, 493 credit, 541, 552 creep, 360, 361, 362, 364, 365, 366, 367, 368, 369, 370, 371, 374, 375, 376, 381, 402 creep tests, 360 crime, 413 criminals, 413 critical value, 394 crop production, 295, 301 crops, 4, 5, 34, 40, 60, 61, 111, 114, 124, 125, 137, 170, 262, 268, 292, 293, 295, 301, 305, 306, 310, 311, 312, 313, 314, 319, 323, 601 cross-validation, 94 crust, 5, 384, 390 crustaceans, 94 cryptographic, 589 cultivation, 10, 117, 137, 146, 269, 270, 272, 281, 319, 324, 543 cultural factors, 217 cultural heritage, 485 cultural values, 169 culture, 10, 167 cumin, 479 cumulative distribution function, 93 currency, 518 customers, 77 cyberspace, 486 cycles, 5, 35, 186, 270, 272, 274, 279, 289, 293, 342, 392, 426 cycling, 260, 263, 264, 269, 270, 271, 278, 279, 281 cyst, 27, 30 cysts, 23, 26
D dairy, 504 Dallas, 467
danger, 100, 156, 171, 239, 374, 601 Darjeeling, 205 data analysis, 53, 75, 89, 167, 209 data collection, 167, 168, 180, 515, 516, 524, 533, 597, 602 data gathering, 99, 239, 596 data mining, 91, 576, 594, 601 data processing, 70, 74, 91, 189, 303, 392 data set, 53, 70, 73, 83, 93, 101, 113, 119, 121, 122, 187, 209, 213, 263, 364, 374, 377, 378, 389, 393, 394, 414, 432, 437, 440, 444, 478, 566 data structure, 73, 77, 110, 521, 602 data transfer, 482, 483, 491 database, 2, 24, 73, 88, 89, 90, 97, 104, 109, 161, 168, 180, 203, 215, 221, 225, 269, 270, 292, 318, 328, 330, 404, 407, 412, 418, 468, 499, 521, 522, 523, 527, 529, 531, 533, 534, 559, 560, 562, 565, 566, 571, 596, 597, 599, 600, 602 database management, 73, 292, 527, 532, 602 dating, 269, 400 death, 453, 485 deaths, 171 decay, 81, 82, 126 deciduous, 187, 193, 263 decision makers, 152, 165, 425, 427, 430, 465, 486, 487, 594, 596 decision making, 1, 2, 5, 6, 125, 128, 163, 350, 364, 381, 411, 478, 487, 489, 499, 503, 510, 559, 561, 565, 587, 595, 596 Decision Support Systems, v, 151 decision-making process, 177, 181, 183, 493 decisions, 100, 101, 152, 217, 350, 355, 364, 428, 429, 476, 480, 481, 484, 486, 492, 503, 515, 524, 528, 533, 557, 563, 580, 602 decomposition, 252, 280 deep-sea, 10 defects, 362 deficit, 113, 124, 383, 390, 391 deficits, 35, 278 definition, 72, 76, 82, 94, 99, 151, 156, 264, 265, 267, 352, 377, 384, 411, 412, 429, 477, 488, 528, 536, 541 deflation, 34, 402 deforestation, 169, 186, 202, 203, 206, 209, 210, 211, 212, 213, 216, 224, 227, 237, 240, 242, 244, 245, 246, 252, 270, 271, 272, 281, 318, 542, 552, 553 deformation, 342, 344, 360, 381, 384, 387, 389, 390, 392, 393, 394, 395, 396, 398, 400, 402, 403, 404, 405, 406, 408, 600 degradation, 1, 5, 157, 169, 198, 202, 240, 246, 270, 291, 292, 317, 318, 319, 320, 321, 322, 323, 324, 326, 327, 328, 329, 330, 331, 412, 578, 589
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Index delinquency, 413 delivery, 6, 513, 517 democracy, 492 demographic data, 484 demography, 168 denitrification, 279 density, 22, 23, 27, 36, 68, 70, 104, 112, 113, 114, 159, 193, 200, 224, 234, 239, 252, 254, 259, 292, 306, 323, 356, 359, 378, 393, 414, 425, 427, 430, 431, 433, 454, 455, 465, 472, 503, 504, 505, 555, 592, 593 Department of Agriculture, 140, 148, 552 Department of Defense, 387 dependent variable, 377, 378, 451 deposition, 249, 271, 272, 342 deposits, 35, 137, 335, 345 depression, 87 derivatives, 339, 390, 440 desert, 187, 385 designers, 481 destruction, 175 detection, 3, 30, 68, 74, 82, 85, 87, 88, 89, 91, 99, 187, 191, 209, 210, 213, 242, 391, 392, 430, 468, 471, 472, 483, 560, 573, 574, 575, 577, 579, 580, 581, 582, 583, 593, 594, 595, 602, 605 detention, 325 detritus, 82 developed countries, 292, 595 developed nations, 413 developing countries, 5, 132, 133, 134, 149, 183, 184, 291, 292, 295, 326, 411, 412, 513, 514, 536, 595 developmental process, 412 deviation, 81 dew, 601 diaphragm, 599 diatoms, 24, 28 dielectric constant, 270 differentiation, 38, 422 diffusion, 362 digitization, 418 dimensionality, 73, 437 dinoflagellates, 16, 23, 27, 28, 30 direct action, 321 direct observation, 68, 321 directionality, 254 disaster, 4, 165, 166, 167, 168, 169, 170, 171, 175, 177, 179, 180, 181, 182, 183, 184, 534, 595, 602 discharges, 9, 21, 24, 79 discourse, 307 Discovery, 469, 520 discriminant analysis, 441 discrimination, 3, 292, 293, 294, 295, 302, 303
613
discriminatory, 577 disease rate, 580 diseases, 10 dislocation, 361, 362 dislocations, 361, 362 disorder, 413 dispersion, 33, 38, 59, 78, 82, 268, 448, 465 displacement, 123, 342, 392, 393, 395, 402, 408, 549, 550 disposition, 338 disseminate, 100, 292, 517, 518 dissolved oxygen, 10, 98 distributed applications, 519 distribution function, 93, 433, 562 diversity, 6, 11, 35, 186, 187, 243, 246, 295, 338, 452, 465, 513, 515, 588 division, 37, 147, 299, 481 dominance, 34 doors, 174 DOP, 389 doppler, 74, 99, 588, 600 download, 122, 329, 484, 588 draft, 172 drainage, 34, 116, 132, 140, 149, 169, 325, 339, 344, 415, 416, 418, 419, 420, 421, 430, 431, 435, 437, 452, 453, 454, 470, 578 drinking, 139, 427, 434, 575, 595, 596, 598, 599, 603 drinking water, 139, 427, 575, 595, 598, 599, 603 drought, 3, 128, 171, 595, 596 droughts, 34, 35 duality, 583 dumping, 450 duplication, 517, 526, 594 duration, 134, 148, 171, 175, 319, 389, 431, 454, 590 dust, 75 dykes, 335 dynamic environment, 602 dynamic systems, 605 dynamical system, 381, 590, 591, 593 dynamical systems, 381
E early warning, 573, 574, 575, 579, 580, 601, 604 earth, 5, 69, 73, 75, 76, 88, 99, 113, 216, 251, 263, 302, 322, 345, 417, 436, 442, 514, 552, 553 Earth Science, xi, 139, 345, 346, 349, 553 earthquake, 5 East Asia, 318 ecological, 1, 2, 4, 5, 28, 82, 152, 210, 215, 216, 217, 244, 249, 251, 276, 280, 289, 296, 299, 300, 301, 302, 317, 321, 339, 411, 420, 421, 514, 573, 574, 576, 587, 606
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
614
Index
Ecological Economics, 553 ecological systems, 5, 215, 514 ecology, xi, 30, 60, 70, 240, 242, 244, 247, 330, 420, 574 economic activity, 170, 174 economic cooperation, 318 economic development, 1, 182, 425, 502, 553 economic growth, 10, 513, 514 economic indicator, 415, 573 economic losses, 9, 10, 430 economic performance, 292 economic status, 419 economics, 101 ecosystem, 5, 10, 11, 31, 59, 68, 81, 83, 101, 106, 152, 215, 216, 234, 239, 240, 249, 250, 251, 252, 255, 258, 260, 263, 264, 266, 272, 274, 276, 279, 280, 288, 289, 292, 299, 317, 321, 322, 324, 427, 429, 470, 514, 515, 539, 541, 548, 552, 575, 576, 606 ecosystems, ix, 4, 10, 31, 60, 62, 68, 81, 109, 152, 185, 186, 210, 216, 217, 219, 228, 234, 239, 240, 241, 242, 250, 260, 263, 265, 266, 268, 269, 270, 273, 274, 276, 278, 281, 284, 296, 299, 328, 419, 426, 427, 515, 542, 556 Ecuador, 27 eddies, 76, 78, 80, 82, 83, 85, 86, 100 Eden, 493 Education, 479, 485, 490, 493, 494, 495, 496, 502, 511, 512, 603 EEA, 514, 534 effluents, 156, 420 Egypt, 429, 470 eigenvector, 438 elders, 167 election, 46 electric conductivity, 35 electrical conductivity, 391, 408 electricity, 425, 427, 428, 465, 598, 599 electromagnetic, 39, 74, 75, 113, 255, 306, 389, 391, 392, 393, 431, 472 electromagnetic wave, 75, 392 electromagnetic waves, 392 electron, 259 email, 487, 490 emergency response, 597 emission, 75, 113, 271, 343, 432, 467, 551, 575 empirical methods, 259, 580 employees, 506 employment, 10, 420, 499, 502, 505, 506, 507, 509, 510, 560 employment growth, 502, 505, 509 encapsulated, 599 Encoding, 355, 483
encouragement, 210 end-users, 526, 527, 528 energy, 1, 10, 68, 75, 112, 113, 123, 129, 251, 252, 263, 264, 275, 276, 277, 281, 291, 362, 364, 365, 366, 367, 368, 369, 372, 374, 377, 391, 416, 425, 427, 433, 450, 454, 467, 470, 552, 574, 590, 592 energy consumption, 427, 450, 454 engagement, 481, 484 England, 493, 534 enterprise, 523 entertainment, 487 entropy, 437, 448, 465, 471, 593 environment, ix, 4, 5, 29, 60, 68, 73, 80, 100, 131, 133, 153, 154, 160, 165, 166, 169, 170, 177, 182, 184, 234, 281, 291, 296, 299, 320, 411, 418, 419, 421, 422, 423, 429, 433, 476, 477, 485, 486, 487, 489, 505, 513, 514, 516, 517, 518, 519, 523, 524, 526, 528, 529, 536, 541, 574, 578, 588, 589, 591, 593, 594, 602 environmental change, 100, 169 environmental conditions, 24, 26, 34, 216, 389, 594, 601 environmental control, 259, 278 environmental degradation, 291 environmental effects, 174 environmental factors, 11, 24, 26, 128, 249, 251, 253, 256 environmental impact, 10, 27, 71, 246, 426, 473, 540 Environmental Impact Assessment, 3 environmental influences, 102 environmental issues, 243 environmental protection, ix, 504 Environmental Protection Agency, 603 environmental resources, 419 epoxy, 599 equating, 259 equilibrium, 85, 272, 273, 600 ER, 36, 139, 189, 191, 222, 320, 417 erosion, 3, 136, 168, 178, 245, 296, 298, 299, 302, 322, 323, 326, 330, 335 error detection, 560 error management, 560 estates, 431, 469 estimating, 4, 111, 112, 114, 140, 142, 147, 275, 553, 580 ethics, 476, 494 etiology, 573, 574, 575 euphotic zone, 87 Europe, 242, 276, 285, 330, 479, 493, 494, 495, 496, 554 European Commission, 181 European Environment Agency, 534 European Space Agency, 263
Index European Union (EU), 89, 388, 476 eutrophication, 10, 21, 22, 24 evacuation, 168, 171, 175, 178 evaporation, 112, 132, 264, 448 evapotranspiration, 111, 112, 113, 114, 122, 123, 124, 127, 264, 275, 320, 321, 323, 324, 427, 454 evolution, 1, 41, 42, 44, 70, 80, 82, 88, 106, 108, 232, 244, 299, 335, 354, 404, 405, 482, 583, 588, 593, 594 execution, 397, 559, 566, 593 exercise, 209, 542, 578, 594 expert systems, 52, 345 expertise, ix, 165, 499, 500, 501, 595 exploitation, 1, 88, 99, 154, 159, 160, 514, 516, 599 exporter, 34 exposure, 175, 179, 184 externalities, 467, 570 extraction, 10, 77, 147, 191, 347, 385, 391, 395, 397, 398, 405, 423, 429, 438, 473, 514, 590
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
F fabric, 502 factorial, 61 failure, 299, 342, 344, 355, 374, 383, 384, 385, 391, 407 false alarms, 579, 593, 599 family, 89, 93, 166, 175, 183, 259, 413 FAO, 10, 27, 105, 299, 302, 318, 320, 322, 323, 328, 329 farmers, 116, 139, 177, 542, 543, 594, 595, 598, 601 farming, 10, 72, 295, 299, 319, 543, 596, 601 farmland, 153, 502, 503, 504, 511 farmlands, 326, 503 farms, 10 fatalities, 342 fatigue, 360 fault tolerance, 377 faults, 385, 402, 588 fauna, 452, 465 FCC, 311, 473 February, 15, 79, 80, 293, 311, 312, 313, 492, 495 feces, 24 feedback, 2, 10, 255, 257, 278, 279, 485, 580, 589 feeding, 24, 83, 84, 103, 600 feet, 504, 594 fertilization, 272, 273, 274 fertilizer, 306 fertilizers, 22, 601 fiber, 50, 600 fidelity, 587, 590, 605 filament, 79, 82, 107 film, 75, 483
615
films, 75, 88, 484 filters, 97, 405 filtration, 540 financial resources, 166, 183, 526 financial support, 329, 510 fire, 211, 250, 269, 272, 273, 276, 278, 400, 401, 406, 552, 579 fires, 207, 212, 250, 269, 401, 406 firms, 335 first responders, 599 fish, 10, 24, 27, 30, 34, 44, 68, 72, 73, 81, 83, 87, 88, 89, 90, 92, 93, 102, 107 fisheries, 4, 10, 28, 67, 68, 71, 72, 73, 83, 88, 89, 96, 97, 98, 100, 101, 102, 105, 107, 109, 514, 523 fishing, 10, 35, 72, 76, 83, 84, 88, 89, 90, 91, 92, 94, 100, 101, 102, 434, 450 fitness, 366 fixation, 256, 272 flexibility, 91, 531, 532 float, 598 flood, 4, 41, 147, 165, 166, 167, 169, 171, 172, 173, 174, 175, 177, 178, 179, 182, 430, 454, 595, 597 flood hazards, 168, 175 flooding, 41, 167, 169, 171, 172, 178, 179, 319, 385, 406, 430, 453, 454, 468, 595 flora, 210, 244, 452, 465 flora and fauna, 452, 465 flow, 4, 31, 35, 69, 71, 82, 117, 131, 132, 140, 141, 146, 147, 148, 154, 159, 160, 161, 164, 167, 303, 326, 385, 431, 450, 454, 529, 545, 595, 600, 601 flow rate, 385 fluctuations, 321, 390 fluid, 87, 385, 391 focus group, 172, 173 focusing, 46, 218, 451 food, 5, 10, 29, 81, 291, 306, 317, 318, 328, 601 food production, 291, 317 Ford, 345 forecasting, 3, 71, 350, 355, 377, 500, 510, 512, 591, 595 forest ecosystem, 185, 210, 216, 217, 219, 228, 234, 238, 239, 240, 243, 246, 258, 266, 269 forest fire, 209, 212, 250, 252, 269, 271 forest fires, 212, 250, 269 forest fragments, 223, 234, 242 forest management, 71, 218, 223, 224, 226, 230, 232, 233, 239, 240, 241, 245, 272 forest resources, 218, 223, 239, 240, 243 Forest Service, 269, 552 forestry, 210, 211, 215, 241, 242, 243, 246, 281, 282, 504, 554
Index
616
forests, 4, 60, 186, 187, 193, 210, 217, 218, 219, 226, 228, 233, 237, 241, 243, 244, 245, 262, 264, 265, 266, 269, 274, 287, 450, 514, 523, 543, 602 fouling, 600 fracture, 361, 362, 382, 395 fractures, 335, 342 fragmentation, 186, 187, 191, 203, 207, 208, 209, 210, 212, 213, 215, 217, 218, 221, 223, 233, 234, 239, 240, 241, 242, 243, 244, 245, 578 fragmented landscapes, 187 framing, 510 France, xi, 181, 395, 406 frequency distribution, 586, 587 fresh water, 79, 156, 306, 315 freshwater, 22 friction, 123 FSA, 591, 592 fuel, 139, 319, 426, 427 fugitive, 413 funds, 491 fusion, 263, 333, 335, 336, 337, 347, 415, 416, 430, 438, 469, 472, 473, 587, 588, 590, 592, 593, 594, 604 futures, 500 fuzzy logic, 52, 352, 364, 381, 563, 599 fuzzy sets, 307, 352, 381, 570
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
G Gadus morhua, 68 games, 483, 485, 486, 487, 488, 490, 492, 493 Gamma, 532, 534 gas, 112, 277, 350, 385, 405, 514, 523, 554, 588 gases, 280, 342 gauge, 384, 588, 597, 601 Gaussian, 95, 363 gender, 580 generalization, 52, 77, 263, 358, 377, 433, 583 generation, 3, 41, 76, 77, 83, 141, 161, 366, 394, 415, 477, 482, 501, 510, 517, 519, 596 genetic algorithms, 52, 354 genetics, 472 geochemical, 151, 156 geochemistry, 338 geography, 71, 170, 426, 503 geology, 1, 4, 11, 99, 333, 339, 344, 345, 448 geophysical, 5, 335, 336, 344, 384, 391, 402, 406, 408 Georgia, 30 geothermal, 391, 405 geothermal field, 405 Germany, 185, 243, 390, 406, 407 germination, 23
gestures, 486 Gibbs, 493 glass, 50 glasses, 488, 489 global climate change, 186, 250 global warming, 250, 280, 425, 595 globalization, 169 goals, 180, 217, 292, 489, 501, 503, 510, 515, 542, 552, 577, 581 google, 436, 442 governance, 1, 6, 103, 180, 181, 426, 573, 574, 575, 594, 602 government, iv, 13, 134, 166, 177, 180, 183, 421, 426, 428, 495, 535, 575, 576, 577, 580, 594, 595, 602 GPP, 251, 255, 256, 257, 269 Global Positioning System (GPS), ix, 2, 7, 74, 101, 168, 171, 173, 180, 182, 183, 291, 344, 383, 385, 387, 388, 389, 390, 391, 394, 401, 403, 407, 408, 435, 526 grades, 307, 308, 309, 492 graduate education, 603 graduate students, 603 grain, 362 grain boundaries, 362 grains, 362 granites, 136 grants, 26, 554 graph, 38, 44, 59, 89, 90, 127, 356, 359, 360, 369, 371, 375, 376, 466, 581, 582, 585 grass, 268, 449, 473 grasses, 262 grassland, 206, 263, 269, 277, 468, 569 grasslands, 34, 219 gravitational field, 390 gravitational force, 384 gravity, 339, 389, 390, 391 grazing, 218, 244, 246 Great Britain, 484 Greece, 432, 470, 495 Green Revolution, 291 greenhouse, 250, 284, 450, 554 greenhouse gas, 250, 450 grid computing, 589 grid resolution, 113 grids, 220, 239 ground water, 1, 35, 71, 118, 132, 152, 153, 154, 156, 157, 159, 160, 161, 163, 164, 385, 391, 395, 396, 397, 398, 400, 405, 407, 408, 420, 427, 451, 452, 465, 598, 600 ground-based, 266, 389, 401, 403 group work, 487, 488, 489 grouping, 45, 141, 293
Index groups, 48, 51, 52, 62, 100, 136, 167, 217, 232, 257, 262, 263, 267, 268, 307, 399, 413, 475, 481, 482, 487, 526, 529, 560, 599 growth, 6, 10, 22, 23, 24, 29, 72, 82, 131, 152, 169, 170, 249, 251, 256, 271, 272, 273, 274, 278, 292, 293, 301, 312, 328, 412, 425, 426, 428, 429, 430, 434, 451, 456, 462, 463, 464, 465, 468, 469, 470, 477, 500, 502, 503, 505, 506, 507, 509, 510, 511, 513, 514, 550, 561, 579 growth dynamics, 426 growth rate, 131, 292, 500 GSA, 346 GSM, 600 Guangdong, 10, 24, 28, 29 Guangzhou, 431 guidance, 81, 133 guidelines, 153, 163, 335, 346 Gujarat, 195
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
H habitat, 71, 73, 87, 217, 242, 243, 354, 413 handling, 33, 71, 73, 517, 522, 570 hands, 488 harbour, 90 hardness, 36 hardwood forest, 222, 225, 226, 228, 232, 233 hardwoods, 224 harm, 175 Harvard, 580 harvest, 273 harvesting, 35, 131, 147, 193, 228, 238, 239, 596, 597 hazards, 167, 168, 169, 171, 174, 175, 183, 351, 360, 391, 405 haze, 75 health, 4, 10, 111, 114, 153, 154, 175, 411, 428, 514, 575, 578, 579 healthcare, 595 heart, 600, 601 heat, 75, 76, 86, 112, 113, 114, 122, 123, 128, 129, 275, 278, 279, 280, 301, 391, 427, 429, 431, 432, 433, 454, 466, 467, 468, 469, 470, 471, 601 heat transfer, 114, 123, 280 height, 76, 100, 113, 114, 123, 140, 179, 276, 277, 342, 386, 389, 392, 402, 403, 417, 433, 485 hemisphere, 63, 254 herbicides, 156 heterogeneity, 186, 223, 312, 390, 396, 521, 528, 536 heterogeneous, 46, 250, 296, 306, 392, 519, 521, 533 heterotrophic, 252, 269, 272, 277, 289
617
high resolution, 37, 68, 76, 87, 104, 211, 250, 293, 329, 334, 344, 389, 390, 393, 412, 414, 415, 422, 431, 466, 580 high risk, 173, 333, 334, 345 high temperature, 22, 23, 362 highlands, 298 high-tech, 292 highways, 426, 465 HIS, 438, 439, 473 histogram, 266, 473, 592 holistic, 1, 2, 425, 465 holistic approach, 425, 465 Holland, 330 hologram, 489 holograms, 489 homeland security, 575 homogeneity, 59 Honduras, 243, 245 Hong Kong, 10, 13, 27, 488 horizon, 292, 416, 417 host, 174 hot spots, 575 House, 173, 174 household, 166, 171, 173, 174, 175, 179, 416, 417, 450, 502, 505, 506, 579 households, 115, 170, 172, 175, 178, 179, 510 housing, 168, 169, 171, 172, 178, 412, 415, 416, 421, 428, 431, 469, 502, 505, 510, 599 HRV, 422 hub, 416 hue, 336, 439, 466 human, ix, 1, 2, 4, 5, 10, 11, 31, 52, 151, 153, 164, 165, 166, 169, 170, 175, 180, 181, 183, 186, 187, 216, 217, 237, 240, 241, 250, 269, 276, 318, 322, 328, 350, 352, 354, 385, 426, 438, 454, 476, 485, 514, 515, 517, 518, 530, 549, 552, 579, 586, 594 human activity, 476 human brain, 52, 354 human development, 549 human motivation, 164 human nature, 586 human resources, 180, 181, 517 human welfare, 1 humanity, 428 humans, 10, 216, 426, 448, 540, 599 humidity, 1, 123, 135, 386, 601 Hungarian, 351 Hungary, 245 hunting, 35, 587 hybrid, 279, 364, 377, 380, 590 hydro, 133, 152, 160, 280, 320, 330, 391 hydrocarbon, 335, 337 hydrocarbons, 391
Index
618
hydrogen, 387 hydrologic, 112, 132, 133, 135, 139, 140, 141, 143, 147, 148, 159, 246, 261, 325, 431 hydrological, 4, 20, 33, 35, 132, 133, 148, 151, 152, 156, 157, 158, 159, 160, 161, 164, 171, 187, 249, 263, 278, 279, 280, 306, 322, 330, 339, 448, 599 hydrological cycle, 132 hydrology, 34, 147, 154, 244, 245, 264, 280, 588 hydrosphere, 68 hydrothermal, 335, 338, 345, 385 hydrothermal system, 335, 338 hydroxyl, 338 hypothesis, 51, 52, 84, 85, 93, 125, 444, 581, 582, 583 hypothesis test, 582, 583
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
I Iberian Peninsula, 79, 85, 86, 87, 104, 107, 108 IBM, 520 ice, 76, 339, 386, 406, 433 ice pack, 76 ice packs, 76 ICT, 493, 524 id, 82, 83, 86, 329, 418, 521, 530 identification, 1, 33, 38, 42, 72, 74, 91, 94, 160, 167, 168, 182, 292, 316, 335, 382, 391, 394, 412, 415, 416, 422, 430, 500, 577, 580, 593, 594 identity, 191, 438 IHS, 335, 399, 439 Illinois, 485 illumination, 121, 254, 255, 268, 339, 442 image analysis, 245 imagery, 5, 60, 62, 68, 74, 78, 80, 82, 85, 88, 100, 103, 109, 212, 217, 225, 245, 253, 254, 264, 305, 313, 314, 336, 339, 344, 412, 414, 415, 416, 417, 418, 419, 420, 421, 422, 423, 432, 469, 470, 471, 526, 576 imagination, 485 imaging, 37, 102, 266, 335, 339, 383 imbalances, 250 immigrants, 412 impact analysis, 500, 515 implementation, 2, 3, 4, 6, 7, 101, 151, 154, 156, 157, 246, 281, 381, 449, 464, 496, 502, 507, 513, 519, 521, 522, 526, 528, 530, 533, 541, 581, 602 in situ, 68, 75, 80, 81, 82, 85, 88, 98, 100, 101, 103, 106, 107, 187, 355, 429, 433, 594 incentive, 528 incidence, 39, 335, 346, 393, 400, 562, 580 inclusion, 261, 294, 415, 525, 582 income, 178, 413, 415, 420 incomes, 169
independence, 47, 48, 59, 88 independent variable, 377, 451, 539 India, v, vi, xi, 4, 5, 6, 107, 131, 133, 134, 139, 147, 185, 186, 187, 189, 193, 208, 210, 211, 212, 244, 245, 246, 282, 283, 286, 287, 291, 292, 295, 296, 299, 302, 305, 306, 314, 315, 329, 381, 383, 397, 400, 401, 405, 406, 411, 412, 414, 417, 419, 422, 423, 426, 433, 435, 465, 470, 472, 473, 542, 550, 553, 554, 573, 574, 587, 596, 606 Indian, vi, xi, 5, 27, 28, 70, 131, 133, 140, 149, 150, 185, 186, 188, 189, 205, 210, 211, 212, 243, 263, 276, 281, 284, 291, 302, 305, 315, 349, 381, 383, 401, 405, 407, 411, 425, 436, 468, 470, 543, 544, 555 Indian Ocean, 243 Indiana, 62 indication, 88, 93, 179, 228, 239, 328 indicators, 35, 217, 238, 239, 244, 249, 317, 318, 320, 321, 323, 324, 325, 326, 328, 330, 415, 426, 442, 566, 570, 575, 577, 578, 584, 586, 604, 605 indices, 40, 122, 128, 187, 206, 224, 233, 254, 256, 257, 265, 266, 429, 432, 466, 544, 551, 578 indigenous, 605 individual character, 427 individual characteristics, 427 Indochina, 29 Indonesia, 11, 182, 534, 543 Indo-Pacific, 28 induction, 274 industrial, 152, 156, 250, 272, 382, 397, 431, 433, 454, 457, 462, 464, 465, 507, 563, 564, 598 industrial emissions, 454 industrial revolution, 250 industrialization, 10, 169, 212, 228, 239, 425 industry, 10, 153, 170, 186, 416, 420, 595, 601, 602 inelastic, 398, 399 infection, 579 infections, 579 inferences, 39, 52 inflation, 405 informal sector, 416 information and communication technologies, 292 information communication technology, 496 information sharing, 165, 166 Information System, ix, 2, 68, 103, 105, 108, 131, 139, 240, 484, 535, 559, 570, 571 information systems, 6, 105, 147, 499, 511, 577 information technology, 2, 6, 73, 475, 478, 489, 492, 573 infrared, 37, 38, 46, 67, 70, 74, 75, 80, 81, 82, 83, 85, 100, 109, 121, 129, 220, 239, 266, 267, 269, 302, 313, 315, 335, 339, 431, 466, 470, 472, 528 infrared spectroscopy, 335
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Index infrastructure, 6, 170, 171, 178, 209, 319, 421, 425, 426, 427, 428, 430, 453, 465, 492, 502, 506, 510, 524, 532, 533, 534, 535, 574, 575, 577, 588, 595 inheritance, 354 inherited, 2, 389 inhomogeneities, 390, 394 inhomogeneity, 86 injury, iv, 174 innovation, 575 inorganic, 21, 22, 34, 38, 306 insecurity, 170, 318, 328 insight, 73, 181, 476 inspection, 89, 595, 597, 598, 600, 601 institutions, ix, 72, 73, 96, 166, 241, 421 instruments, 37, 68, 69, 70, 71, 81, 166, 265, 266, 277, 289, 342, 385, 421, 431, 580 integration, ix, 1, 2, 7, 153, 167, 168, 180, 184, 272, 297, 301, 335, 337, 395, 415, 422, 428, 517, 529, 602, 603 integrity, 244, 292, 317, 576, 578, 600 intelligence, 52, 381, 476, 487, 494 interaction, 3, 39, 85, 169, 466, 476, 478, 481, 482, 483, 485, 486, 489, 490, 502, 510, 532, 595 interaction process, 85 interactions, 4, 6, 29, 106, 107, 112, 152, 174, 295, 517, 532, 589, 591, 593, 603 interactivity, 480, 485 interdisciplinary, 77, 487, 553 interface, 73, 87, 157, 158, 484, 487, 488, 520, 521, 522, 530, 531, 532, 588, 599, 602 interference, 389 international law, 99 Internet, 97, 99, 180, 475, 477, 478, 480, 482, 484, 485, 486, 487, 520, 554, 577, 603 interoperability, 516, 519, 521, 522, 526, 528, 529, 533 interrelationships, 339, 344 interstate, 507 interval, 53, 91, 126, 132, 133, 388, 391, 392, 393, 394, 402, 404, 562, 585 intervention, 89 interview, 172 interviews, 88 intrinsic, 85, 93, 254, 426, 593 invasive, 575 invasive species, 575 inventories, 244, 251, 328 inventors, 486 inversion, 250, 275, 287, 404, 579 Investigations, 381 investment, 99, 178, 484 IPCC, 250, 285 Iran, 382, 513, 515, 523, 524, 525, 534
619
iris, 482 iron, 338 irrigation, 4, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 125, 127, 128, 129, 131, 132, 153, 156, 160, 306, 434, 595, 597, 598, 601, 602 IRS, vi, 70, 186, 187, 189, 193, 209, 212, 213, 293, 294, 296, 302, 305, 310, 311, 313, 333, 339, 344, 425, 428, 436, 437, 449, 451, 464, 465 ISCO, 147 Islam, 62 island, 3, 344, 427, 429, 431, 433, 454, 467, 469, 470, 471 ISO, 99, 483, 528, 595 isolation, 217, 413 Israel, v, 151, 152, 154, 155, 156, 158, 160, 163, 422 Italy, 105, 211, 302, 604 ITC, 211 iterative solution, 159
J Jamaica, 242 Japan, 9, 71, 102, 165, 243, 244, 317, 329, 342, 343, 351, 387, 406, 459 Japanese, 70, 387 JAVA, 482, 484 Jerusalem, 151 Jet Propulsion Laboratory, 14, 467, 468 jobs, 416 Jordan, 216, 223, 241, 245, 470 judgment, 2, 152 Jung, 511 jurisdiction, 433, 517 jurisdictions, 502, 510 justice, 575, 578 justification, 541
K K+, 35 kappa, 206, 294 kappa coefficient, 294 Kashmir, 193, 195 Kenya, 535 kernel, 268, 433 killing, 10, 24, 30 kinetics, 259 knots, 90 knowledge capital, 433 knowledge economy, 574 Korea, 276, 511 Kyoto protocol, 212, 250, 540, 552, 553, 554, 555
Index
620
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
L L1, 388 L2, 388 labor, 510 labour, 384, 385, 430, 510 Lafayette, 62 lagoon, 27, 28, 170, 171 lakes, xi, 4, 33, 34, 35, 36, 38, 39, 42, 43, 45, 46, 47, 48, 49, 52, 54, 55, 56, 57, 58, 60, 63, 157, 219, 322, 434, 450, 451, 452, 453, 454, 465, 523, 595 Land Use Policy, 240, 242, 243 landfill, 559 Landsat 7, 56, 64, 65, 119, 222, 254, 293, 294, 295, 333, 339, 344, 468 landscapes, 135, 187, 241, 242, 243, 429, 433, 540, 549, 577 language, 89, 91, 167, 482, 483, 519, 520, 521, 522, 531, 533, 590, 594 Laos, 169, 324, 326, 327, 328 large-scale, 5, 83, 107, 129, 209, 217, 219, 292, 317, 336, 390, 553, 577 larvae, 79, 87, 98, 106, 108 larval, 80, 82 laser, 386 latency, 593 Latin America, 413, 479 lattice, 590 law, 75, 87, 99, 132, 488 laws, 515, 529 leaching, 279 leakage, 6, 161, 398, 539, 540, 541, 542, 543, 546, 548, 549, 550, 551, 552, 554, 555 learning, 52, 53, 54, 55, 56, 57, 279, 280, 354, 355, 364, 365, 381, 425, 429, 451, 464, 465, 552 learning process, 52 lens, 79, 266 licenses, 477 life cycle, 28, 87 life forms, 4 life sciences, 60 lifestyles, 319 light conditions, 87 light scattering, 442 likelihood, 191, 213, 222, 306, 307, 309, 315, 316, 425, 429, 440, 441, 465, 543, 577, 581, 582, 583, 585 limitation, 143, 160, 272, 360, 408, 582 limitations, ix, 2, 7, 59, 66, 109, 133, 150, 184, 209, 261, 393, 394, 408, 409, 501, 533, 581, 588, 596, 606 Lincoln, 6, 492, 499, 511
linear, 40, 46, 47, 48, 59, 87, 93, 104, 122, 132, 148, 175, 185, 187, 189, 190, 201, 252, 256, 352, 358, 364, 384, 400, 402, 403, 404, 405, 406, 408, 409, 419, 429, 432, 444, 445, 456, 457, 465, 471, 585, 586, 587, 603, 605 linear regression, 59, 185, 187, 189, 190, 201 linear systems, 352 linguistic, 364 linkage, 426, 602 links, 42, 52, 53 liquid water, 269 literacy, 139 lithologic, 334 livestock, 244, 599 living conditions, 169 local authorities, 165, 172, 177, 178, 181, 490 local community, 177, 502, 503 local government, 177, 535 localization, 589, 605 location, 2, 74, 76, 77, 78, 80, 82, 87, 94, 99, 103, 109, 112, 113, 115, 132, 134, 140, 172, 220, 221, 222, 265, 268, 276, 297, 310, 344, 386, 387, 388, 391, 395, 413, 414, 418, 430, 483, 484, 486, 489, 505, 519, 530, 540, 545, 548, 552, 554, 555, 561, 563, 565, 566, 567, 568, 570, 580, 602 LOD, 482 logging, 186, 272, 306, 319, 400, 486, 589 London, 62, 103, 108, 183, 285, 286, 422, 479, 480, 482, 483, 484, 486, 492, 493, 494, 535, 554, 571 long distance, 132, 174 Los Angeles, 482 losses, 4, 10, 100, 112, 118, 134, 154, 170, 172, 175, 212, 232, 252, 273, 430, 448 low power, 581 low risk, 173 low temperatures, 24 low-density, 427, 503, 504 LSD, 604 lupus erythematosus, 606 lying, 170, 172, 261, 453
M M and A, 28 M.O., 63, 286 machine learning, 355 machine-readable, 519 machinery, 451 machines, 350, 530 magnetic, iv, 121, 335, 336, 337, 390, 416 maintenance, 154, 251, 269, 271, 318, 517, 540, 552, 598, 599 maize, 296
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Index major cities, 432, 470 Malaysia, 9, 11, 12, 13, 28, 240 mammal, 35, 544 management practices, 1, 295 mandates, 528, 529 manipulation, 76, 77, 89, 97, 98, 481, 484, 485 man-made, 514 manufacturing, 350 map unit, 224 mapping, ix, 2, 3, 5, 73, 89, 94, 163, 164, 165, 166, 167, 168, 171, 172, 173, 174, 175, 177, 179, 180, 181, 183, 184, 186, 191, 210, 212, 213, 244, 249, 251, 263, 267, 268, 269, 270, 296, 297, 299, 303, 306, 333, 334, 335, 336, 337, 338, 339, 344, 345, 346, 347, 350, 391, 406, 414, 415, 416, 423, 428, 429, 431, 432, 437, 438, 466, 467, 468, 471, 520, 522, 526, 527, 536, 579, 580 marginal distribution, 49 marginalization, 592 marine environment, 10, 71 marine mammals, 73, 81 maritime, 274 market, 34, 90, 169, 178, 502, 553 market economy, 169, 178 market prices, 34 markets, 510, 542 Markov, 3, 591, 603, 604 Markov chain, 592, 603 Markov model, 591 Markov process, 592 Mars, 346 Maryland, 219, 320, 535 mask, 40, 45, 188, 194, 198, 270, 339 masking, 39, 40, 66, 437 mass transfer, 390 Massachusetts, 485, 493, 495, 499, 511, 554, 604 Massachusetts Institute of Technology, 485, 495 mathematics, 431 matrix, 46, 77, 191, 208, 209, 222, 223, 225, 226, 230, 231, 247, 294, 310, 313, 361, 438, 440, 441, 445, 472, 560, 571, 592 Maximum Likelihood, 293, 315 measurement, 62, 74, 75, 103, 112, 114, 115, 123, 124, 223, 251, 267, 268, 277, 278, 281, 307, 383, 385, 386, 387, 388, 389, 390, 393, 402, 408, 409, 433, 442, 470, 560, 593, 596, 597, 598, 599, 600, 601 measures, 3, 10, 35, 40, 123, 124, 156, 163, 169, 178, 179, 222, 276, 277, 294, 296, 329, 356, 366, 374, 385, 510, 541, 577, 598, 599 mechanical properties, 349, 381 media, 152, 154, 165, 475, 477, 487, 489, 492 mediation, 517
621
Mediterranean, 30, 85, 86, 107, 156, 218, 245 melt, 342 melting, 406 membership, 305, 307, 308, 309, 310, 352, 356, 358, 360, 363, 364, 365, 577, 583, 584 memorizing, 485 memory, 532, 592 mesh networks, 595 messages, 491, 519, 521, 584 meta-analysis, 241 metaphor, 501 meteorological, 69, 86, 133, 170, 171, 250, 259, 260, 273, 277 methane, 250, 270 metropolitan area, 427 Mexico, 212, 241, 381, 405, 407, 571 Mg2+, 35 microbial, 252 microclimate, 112, 128, 431, 433, 469 Microsoft, 520, 532, 533, 534 microwave, 2, 67, 70, 75, 76, 82, 85, 86, 100, 106, 107, 109, 347, 391, 392, 393 microwaves, 67 middle-aged, 353 migrant, 242 migration, 108, 411, 416, 426 migratory birds, 452, 465 mineral resources, 5 mineralization, 279, 280, 333, 344 mineralized, 335, 338 minerals, 336, 338, 385, 391 mines, 335, 336, 523 mining, xi, 35, 91, 186, 335, 346, 381, 385, 391, 400, 401, 402, 405, 406, 540, 576, 594, 601 Ministry of Environment, 211 Minnesota, 111 misleading, 60 missions, 69, 70, 71, 72, 101, 390, 515, 575, 587 misunderstanding, 2, 521 MIT, 62, 493, 495, 604 mixing, 86, 430, 432 mixture analysis, 46, 470 MLC, 293, 294 MLE, 583, 584 mobile phone, 477 mobility, 168, 171, 178, 342, 344, 345, 589, 594 modalities, 594 model specification, 569 modeling, ix, 2, 3, 4, 6, 52, 53, 61, 151, 152, 157, 158, 164, 249, 250, 251, 258, 260, 263, 264, 266, 269, 270, 271, 272, 273, 275, 278, 279, 281, 289, 296, 350, 351, 352, 355, 381, 382, 384, 466, 467, 469, 482, 483, 486, 491, 494, 497, 499, 501, 512,
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
622
Index
521, 523, 533, 536, 540, 541, 542, 550, 552, 553, 568, 570, 578, 580, 591 models, ix, 6, 46, 52, 53, 60, 61, 73, 100, 112, 132, 133, 147, 152, 156, 161, 211, 250, 254, 255, 256, 258, 259, 260, 261, 263, 264, 268, 269, 270, 272, 273, 276, 278, 297, 321, 339, 350, 355, 360, 376, 377, 388, 389, 390, 433, 467, 475, 478, 479, 481, 482, 483, 487, 488, 489, 492, 494, 500, 501, 510, 511, 512, 516, 518, 522, 524, 541, 542, 552, 553, 555, 559, 560, 561, 565, 567, 568, 579, 587, 588, 589, 590, 594 modules, 89, 158, 483, 485 modulus, 362, 366, 371, 382 modus operandi, 91 moisture, 40, 75, 112, 123, 133, 141, 148, 149, 257, 270, 279, 299, 301, 306, 340, 342, 432, 433, 468, 470, 588, 599, 601 moisture content, 432 momentum, 75, 123 money, 178, 180, 183 monsoon, 9, 11, 12, 20, 21, 22, 25, 29, 117, 118, 137, 139, 145, 170, 265, 319, 398, 399 Monte Carlo, 62, 560, 562, 570, 581, 582, 592 Monte Carlo method, 62 Moon, 346, 417 Morocco, 470 morphology, 5, 27, 78, 171, 334, 374, 411, 419, 421, 433 mortality, 10, 606 MOS, 70 mosaic, 299, 577 Moscow, 381 motion, 87, 161, 342, 343, 385, 386, 387, 407 motivation, 51, 72, 164, 167, 540 mountains, 170, 198, 245, 299 mouth, 10, 20, 171 movement, 81, 90, 216, 279, 322, 342, 362, 383, 384, 387, 402, 407, 476, 483, 485, 488, 493 Mozambique, 243 MPS, 223, 224, 234, 235, 236 MSS, 189, 191, 202, 206, 209, 243, 302, 435, 436, 437, 438, 439, 472 multidisciplinary, 524, 553, 595 multimedia, 484 multiple factors, 595 multiplication, 146, 147 multiplier, 446 multivariate, 60, 93, 378, 425, 441, 465, 593, 602 municipal area, 417 municipal sewage, 454 municipal solid waste, 450 mutation, 354 Myanmar, 5, 318, 324, 326, 327, 328
N Na+, 35 NASA, 14, 30, 39, 62, 69, 70, 102, 104, 213, 218, 247 nation, 34 National Academy of Sciences, 605 National Research Council (NRC), 68, 106, 321, 330, 605 National Science Foundation, 554, 587, 603 National Strategy, 181 nationalization, 400 NATO, 30 natural disasters, 165, 166, 168, 169, 170, 182, 183 natural environment, 169, 170 natural gas, 385, 514, 523 natural hazards, 169, 174, 360 natural resource management, ix, xi, 1, 2, 3, 4, 5, 6, 7, 186, 187, 428, 497, 513, 535, 536 natural resources, ix, 1, 2, 3, 4, 5, 6, 7, 153, 203, 215, 216, 217, 223, 281, 291, 292, 299, 356, 425, 427, 428, 451, 465, 473, 497, 513, 514, 515, 516, 523, 526, 533, 534, 535, 536, 542, 556, 573, 588, 594 neglect, 413 neighbourhoods, 179, 484 neon, 606 Nepal, 133, 182, 243, 246 Netherlands, 104, 105, 106, 139, 242, 406 network, 33, 52, 53, 54, 55, 58, 60, 66, 116, 131, 164, 169, 173, 178, 180, 188, 276, 281, 336, 354, 356, 362, 363, 364, 365, 366, 374, 375, 376, 386, 394, 395, 397, 408, 418, 420, 429, 431, 437, 452, 453, 477, 478, 481, 484, 485, 487, 491, 518, 519, 520, 524, 526, 529, 530, 532, 549, 555, 564, 577, 578, 580, 581, 582, 588, 589, 590, 591, 593, 594, 595, 596, 597, 599, 600, 601, 605, 606 networking, 526, 587, 594 neural network, 33, 41, 52, 53, 55, 60, 66, 354, 364, 374, 375, 376, 429 neural networks, 41, 52, 53, 364 neurons, 52, 53, 55, 56, 374, 472 New Jersey, 61, 103, 244, 535, 604 new media, 489, 492 New Mexico, 381 New South Wales, 30 New York, iii, iv, 61, 62, 63, 103, 106, 108, 183, 244, 245, 246, 315, 346, 381, 422, 467, 469, 482, 494, 495, 496, 535, 571, 577, 579, 604, 605, 606 newspapers, 13 Newton, 242, 552 next generation, 519, 575 NGOs, 166 Ni, 29, 471
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Index nickel, 335 Nigeria, 243 Nile, 577, 579, 604 NIR, 37, 38, 66, 121, 122, 266, 269, 271, 313, 316, 416, 417, 442, 451, 455 nitrogen, 261, 270, 271, 272, 273, 275, 278, 279, 280, 288 NOAA, 31, 69, 72, 75, 82, 84, 104, 186, 320, 322, 323, 331, 431, 467, 468 nodes, 42, 53, 519, 582, 583, 588, 589, 590, 591, 593, 602 noise, 75, 392, 394, 395, 396, 400, 402, 403, 562, 563, 588, 589, 593 nonlinear dynamic systems, 593 nonlinearities, 260, 279 non-linearity, 122, 249, 256 nonparametric, 464, 604 non-random, 582 non-renewable, 514 non-renewable resources, 514 non-uniform, 443 normal, 21, 169, 307, 377, 385, 392, 403, 407, 425, 452, 454, 465, 560, 602 normal distribution, 377, 560 normalization, 47, 253, 267 North America, 263, 286, 351, 479, 510 North Atlantic, 83, 85, 86, 105, 107, 108 Northeast, 12, 20, 26, 193, 198, 201, 319, 550, 604 Northern Ireland, 244 Norway, 88, 94 novelty, 152 NPP, 251, 252, 255, 256, 257, 269, 272, 273, 274, 275, 278, 321, 322, 331 NRM, 514, 515, 516, 517, 518, 521, 522, 523, 524, 525, 526, 527, 528, 529, 533, 536, 537 nuclear, 374 null hypothesis, 581 nutrient, 23, 24, 34, 78, 81, 82, 87, 212, 269, 271, 274, 278, 279, 289 nutrients, 21, 22, 27, 34, 79, 156, 291, 578 nutrition, 21, 278
623
offshore, 22, 28, 78, 79, 81, 82, 83, 85, 86 Ohio, 502, 505, 508 oil, 68, 71, 73, 75, 76, 77, 88, 106, 132, 133, 139, 296, 306, 322, 350, 385, 405, 514, 523 oil spill, 71, 73, 88, 106 omission, 209, 222 online, 331, 477, 479, 487, 521, 543 open space, 426, 451, 454, 465, 467, 502, 506, 507 open spaces, 451, 454 openness, 264, 278 operating system, 483, 519 operator, 440, 587, 592 opposition, 570 optical, 2, 4, 33, 62, 74, 75, 109, 255, 265, 266, 267, 268, 270, 333, 335, 336, 344, 347 optical parameters, 75, 109 optical properties, 265, 266 optics, 70 optimization, 81, 152, 354, 364 oral, 167 oral tradition, 167 orbit, 37, 72, 74, 76, 388, 390, 395 Oregon, 242, 243, 284, 468, 470, 493 ores, 335, 514, 523 organ, 491 organic, 35, 38, 62, 250, 252, 270, 280, 339, 426 organic C, 270 organic compounds, 39 organic matter, 62, 280 organizational behavior, 524, 525 orientation, 387, 422, 430 oscillations, 53 overgrazing, 318 oversight, 594 overtime, 217, 317 overtraining, 53 ownership, 177, 242, 319, 419, 503, 569, 588 ownership structure, 569 oxides, 338 oxygen, 10, 34, 98, 588 oysters, 10, 24 ozone, 266
O P object recognition, 438 object-oriented design, 535 observations, 28, 51, 68, 69, 70, 74, 75, 76, 81, 82, 88, 94, 100, 103, 106, 107, 187, 211, 228, 239, 268, 321, 401, 403, 469, 490 obsolete, 168 obstruction, 34 ocean waves, 76 oceans, ix, 1, 4, 67, 322, 387, 390
Pacific, 11, 27, 28, 29, 102, 104, 108, 274 packets, 87, 593 Pakistan, 133 PAN, 416, 417, 418, 436, 454, 464 parameter, 53, 74, 76, 83, 112, 129, 228, 256, 257, 262, 263, 264, 268, 270, 288, 303, 310, 356, 363, 412, 431, 440, 581, 597 parameter vectors, 440
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
624 Paris, xi, 151, 329, 395, 396, 397, 406, 422 Parkinson, 70, 106 partial differential equations, 132 participatory democracy, 476 particles, 38, 75, 87 particulate matter, 38, 75 partition, 307, 308, 310, 352, 591 partnership, 166, 180, 350, 516, 527, 529, 533, 574 partnerships, 517, 525, 528, 529 passive, 37, 74, 75, 106, 432, 467, 480 pasture, 523 pastures, 187, 212, 448 pathogens, 575 pathways, 85 pattern recognition, 429, 466 PCA, 51, 66, 222, 437 peat, 279, 340, 398 pedestrian, 488 peer, xi penalty, 309, 310 Penn State University, 573 Pennsylvania, 381, 576, 578 per capita, 579 perception, 82, 438, 477, 478, 482, 488, 489, 500 perceptions, 2 percolation, 118 periodic, 74, 223 permit, 6, 35, 51, 83, 134, 159 perturbation, 562 perturbations, 128, 405, 565 Peru, 243 pest control, 601 pests, 605 Petroleum, 401 pH, 35, 588 Philippines, 9, 10, 11, 12, 13, 21, 22, 27, 28, 30 philosophical, 476, 540, 541, 552 philosophy, 476 phone, 589, 596, 597, 599 phonological, 256 phosphorus, 35 photographs, 2, 69, 102, 109, 217, 339, 403, 484 photon, 259 photons, 252 photosynthesis, 87, 250, 251, 258, 259, 260, 261, 262, 263, 264, 270, 272, 278, 280, 288, 289 photosynthetic, 256, 259, 261, 264 phyllosilicates, 338 physical environment, 167, 180, 183, 216, 423, 526 physical factors, 173, 326 physical interaction, 588 physical properties, 333, 339, 344, 430, 454 physics, 132, 254, 580, 590
Index physiological, 84, 153, 256, 257, 258, 260, 261, 288 physiology, 249, 256, 259, 278 phytoplankton, 11, 12, 21, 22, 23, 24, 27, 28, 29, 30, 35, 38, 62, 63, 75, 78, 79, 81, 82, 87, 100, 104, 106, 108 phytoplanktonic, 34 pigments, 62, 78, 270, 271 pigs, 139 pipelines, 565 planetary, 345 plankton, 87 planning decisions, 488 plants, 112, 114, 123, 219, 259, 306, 321, 339, 523, 565, 605 plastic, 414, 418 platforms, 73, 167, 179, 250, 257, 492, 519, 533, 580, 593 play, 11, 69, 85, 161, 217, 250, 272, 279, 416, 475, 478, 485, 487, 499, 500, 502, 505, 543, 601 Pleistocene, 154 plug-in, 482, 483, 484 PLUS, 94, 95 poisoning, 10 Poisson, 581 Poland, 406, 495, 496 polarization, 341, 344, 345, 406 policy choice, 501, 502, 512 policy makers, 4, 501, 599, 601, 602 policy making, 516 political aspects, 250, 272 politicians, 481 politics, 476, 491, 595 pollutants, 152, 156, 161 pollution, 10, 24, 27, 29, 31, 33, 71, 161, 421, 514 polygons, 76, 91, 483, 507 polynomial, 363 polynomials, 267 pools, 212, 249, 269, 272, 289 poor, 38, 70, 78, 82, 138, 167, 172, 178, 179, 180, 256, 324, 325, 344, 399, 413, 415, 421, 426, 523, 525, 579 population, 1, 10, 11, 34, 102, 139, 151, 152, 154, 156, 159, 163, 169, 170, 177, 183, 187, 209, 217, 237, 241, 291, 328, 355, 397, 411, 412, 413, 414, 420, 423, 425, 426, 428, 431, 433, 434, 454, 459, 465, 472, 485, 499, 502, 505, 506, 507, 509, 510, 512, 523, 567, 570, 577, 585 population density, 209, 413, 434, 454, 459 population growth, 10, 169, 292, 328, 412, 426, 465, 505, 509 population size, 577, 585 pore, 398 porosity, 356, 359, 378
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Index porous, 138, 154 porous media, 154 portability, 588 Portugal, 67, 84, 88, 94, 95, 99, 102, 107, 563, 569 positive feedback, 10 positive relation, 400 poverty, 5, 170, 171, 178, 179, 292, 318, 413, 414, 415, 420, 574, 575, 578, 579 poverty alleviation, 5, 292, 318 poverty line, 414, 420 power, 13, 34, 75, 149, 389, 427, 476, 510, 565, 581, 588 power lines, 389, 565 pragmatic, 268, 272 precipitation, 24, 112, 118, 140, 187, 188, 219, 322, 326, 431, 448 predators, 34 predictability, 111, 115, 127, 128, 129, 576, 593 prediction, 52, 55, 56, 58, 59, 60, 103, 152, 270, 323, 349, 358, 374, 377, 378, 382, 542, 554, 570, 587, 594 predictive model, 499, 512 preference, 506 preparedness, 167, 168, 182 pre-planning, 601 preprocessing, 63, 253, 437 present value, 94 pressure, 22, 113, 114, 124, 169, 186, 193, 233, 237, 246, 318, 328, 386, 397, 398, 425, 598, 599, 601 prevention, 166, 179, 181, 183, 349, 540, 546, 549, 550, 555 preventive, 178, 351, 577 prices, 34, 416, 419 primary data, 173, 174, 417 primitives, 528, 589 Principal Components Analysis, 51, 66 pristine, 250, 256, 265, 270, 274 privacy, 525, 529 private, 6, 178, 339, 400, 430, 450, 453, 464, 499, 513, 515, 525, 526, 527, 529 private citizens, 499 private sector, 527 probability, 33, 41, 45, 48, 50, 92, 174, 217, 222, 247, 309, 355, 377, 429, 440, 442, 451, 459, 460, 473, 561, 562, 582, 592, 593 probability distribution, 33, 45, 48, 377, 582, 592 problem solving, 52, 485 producers, 81, 206, 251, 264, 517, 526 production, 1, 6, 10, 22, 24, 34, 76, 78, 79, 87, 104, 109, 147, 239, 280, 306, 318, 326, 335, 344, 430, 516, 518, 525, 527, 528, 533, 536, 543
625
productivity, 1, 5, 79, 84, 107, 246, 251, 252, 269, 273, 274, 278, 284, 292, 296, 305, 317, 318, 321, 322, 324, 326, 595 professionalism, 489, 492 professions, 167 profit, 292 profitability, 72 prognosis, 485 program, 159, 180, 200, 244, 288, 355, 554, 575, 577, 594 programming, 89, 91, 182, 354, 477, 482, 483, 519, 534 projector, 491 propagation, 85, 87, 401, 560, 570, 571 property, iv, 93, 170, 174, 253, 333, 335, 336, 337, 344, 382, 393, 400, 430, 442, 484, 486, 518 prosperity, 153 protected area, 73, 76, 106, 185, 188, 203, 210, 539, 543, 546, 548, 550, 555, 557 protected areas, 73, 76, 106, 203, 210, 539, 543, 546, 548, 550, 555, 557 protection, ix, 186, 233, 502, 504, 506, 539, 542, 546, 548, 549, 550, 552, 555 protein, 10 protocol, 212, 250, 254, 264, 265, 266, 520, 536, 552, 553, 589, 594 protocols, 267, 519, 520, 522 prototype, 481, 514, 515, 524, 525, 529, 530, 533, 575 proxy, 175, 223, 532 pseudo, 46, 388 PSS, 6, 7, 174, 175, 493, 499, 501, 511, 512 psychologist, 151 public, 1, 6, 96, 153, 171, 172, 173, 178, 339, 412, 413, 430, 453, 475, 476, 478, 479, 480, 485, 486, 488, 489, 491, 493, 495, 497, 499, 501, 502, 503, 504, 506, 507, 510, 512, 573, 575, 579, 594, 595, 597, 602 public education, 579 public health, 575 public policy, 501, 506, 512 public service, 413, 503 public support, 153 pulse, 386, 391, 393 pulses, 76, 384, 386, 391, 408 pumping, 152, 154, 156, 159, 395, 396, 595, 597, 598 pupils, 486 pure water, 66 pyramidal, 151, 153, 156, 164
Index
626
Q qualifications, 484 quality assurance, 528 quality control, 528 quality of life, 411, 421, 513, 514, 515 quanta, 256 quantum, 588 Quebec, 280 Quercus, 219 query, 97, 161, 168, 521, 522, 531
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
R R&D, 99 radar, 74, 76, 85, 88, 102, 269, 270, 271, 333, 334, 345, 384, 391, 392, 393, 406, 407, 408 radial distance, 384 radiation, 37, 38, 39, 74, 75, 112, 113, 114, 122, 127, 252, 254, 256, 257, 260, 261, 262, 264, 265, 268, 270, 271, 285, 442, 457, 468, 472 radio, 74, 387, 597 radio station, 597 radius, 276, 388 rail, 598 rain, 70, 109, 148, 159, 160, 187, 276, 321, 330, 588, 597, 600, 601 rain forest, 187 rainfall, 1, 22, 117, 118, 132, 133, 134, 135, 139, 140, 141, 148, 149, 150, 170, 171, 266, 297, 319, 320, 321, 322, 323, 342, 427, 434, 448, 452, 453, 454, 465, 597, 598, 601 rainwater, 448 random, 92, 93, 140, 222, 257, 265, 268, 288, 294, 326, 377, 392, 393, 440, 562, 565, 580, 594 random assignment, 223, 294 random errors, 562 range, ix, 2, 4, 5, 10, 23, 34, 37, 46, 53, 70, 84, 94, 101, 113, 121, 126, 135, 167, 168, 170, 175, 186, 187, 192, 200, 228, 239, 249, 255, 275, 299, 319, 323, 338, 339, 363, 383, 386, 387, 388, 392, 395, 402, 403, 413, 417, 443, 484, 500, 542, 551, 556, 560, 562, 589, 599, 600, 602, 603 ratings, 583 raw material, 514 raw materials, 514 Rayleigh, 39 REA, 201, 210 reaction rate, 360 readership, 574 reading, 159, 386, 483, 595, 597
real time, 2, 4, 6, 33, 35, 52, 60, 81, 82, 88, 485, 488, 590 reality, 134, 256, 260, 415, 483, 488, 489, 496, 516, 535, 594 reasoning, ix, 52, 350, 363 recession, 405 reclamation, 1 Reclamation, 452 recognition, 33, 69, 80, 96, 101, 257, 350, 422, 429, 438, 466 recombination, 354 reconstruction, 51, 271 recovery, 245, 361 recreation, 35, 506 recreational, 35, 506 rectification, 220, 221 recurrence, 41 recursion, 592 redundancy, 52, 389, 404, 472 reference frame, 386 reflectance spectra, 255, 338 reflection, 38, 76, 220, 467 refuge, 178 regeneration, 483, 486 regenerative capacity, 514 regional, ix, 1, 4, 5, 6, 9, 22, 24, 25, 27, 29, 30, 34, 40, 63, 81, 83, 107, 152, 156, 171, 187, 217, 241, 242, 243, 246, 250, 251, 256, 259, 260, 263, 264, 266, 268, 269, 270, 271, 272, 273, 275, 276, 278, 281, 292, 295, 305, 317, 318, 319, 326, 328, 329, 333, 335, 337, 338, 339, 344, 383, 390, 391, 395, 428, 430, 456, 473, 499, 502, 518, 535, 553, 573, 588, 589, 594, 602 regional problem, 29 regression, 59, 93, 179, 185, 187, 189, 190, 201, 322, 377, 378, 603 regression analysis, 59, 179, 377, 378 regression equation, 377 regression line, 59 regression method, 603 regrowth, 271, 555 regular, 13, 69, 73, 78, 82, 91, 95, 100, 102, 228, 239, 268, 391, 522, 597, 599 regulation, 374, 478 regulations, 34, 99, 525 regulators, 35 rehabilitate, 328 rehabilitation, 384, 421 rehabilitation program, 384, 421 relational database, 88, 109, 522, 533 relationship, 84, 86, 103, 112, 113, 114, 122, 124, 132, 134, 140, 175, 177, 179, 185, 200, 216, 217, 256, 268, 279, 295, 323, 333, 334, 344, 352, 356,
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Index 377, 385, 406, 431, 432, 455, 457, 458, 465, 466, 471, 518, 522, 526 relationships, 52, 53, 68, 72, 132, 149, 216, 255, 256, 257, 267, 279, 339, 526 relaxation, 83 relevance, ix, 71, 281, 380, 575 reliability, 352, 380, 488, 490, 559, 565, 569 remediation, 576 renewable resource, 514 rent, 420 replication, 181 representativeness, 241 reproduction, 23, 483 research and development, 554, 595 reservation, 511 reserves, 319 reservoir, 115, 117, 134, 164, 186, 250, 306, 310, 346, 595, 597, 599, 600 reservoirs, 152, 597, 601 residential, 169, 172, 173, 413, 416, 427, 431, 453, 454, 457, 464, 465, 503, 504, 505, 506, 507, 509, 510 residuals, 59 residues, 404, 405 resilience, 165, 166, 183, 184, 239, 318, 590 resistance, 113, 122, 362, 420 resource allocation, 167 resource management, ix, 1, 2, 4, 6, 7, 151, 153, 186, 187, 292, 295, 299, 513, 536, 589, 594, 595, 602, 606 resources, ix, xi, 1, 2, 3, 4, 5, 6, 7, 10, 11, 35, 71, 72, 94, 99, 100, 101, 110, 147, 151, 152, 153, 154, 156, 157, 163, 164, 165, 166, 167, 170, 171, 180, 181, 183, 203, 215, 216, 217, 218, 223, 239, 240, 243, 281, 291, 292, 296, 299, 315, 317, 318, 330, 356, 419, 425, 426, 427, 428, 451, 465, 466, 473, 486, 487, 488, 494, 497, 513, 514, 515, 516, 517, 523, 526, 532, 533, 534, 535, 536, 542, 556, 573, 575, 580, 588, 594, 596, 601 respiration, 250, 251, 252, 257, 258, 259, 269, 271, 272, 277, 289 respiratory, 273 responsibilities, 2, 517, 527 restructuring, 492 retention, 79, 80, 82, 108, 134, 140, 141, 145, 149, 150 returns, 581 rice, 306 Río de la Plata, 63 Rio Earth Summit, 1, 7 risk, 3, 4, 5, 6, 11, 41, 132, 165, 166, 167, 168, 171, 172, 173, 174, 175, 177, 178, 179, 180, 181, 183,
627
184, 297, 298, 299, 302, 303, 323, 330, 333, 334, 345, 350, 470, 559, 561, 570, 579 risk assessment, 41, 297, 298, 299, 302, 303, 330, 350, 570 risk factors, 11 risk management, 4, 165, 166, 167, 168, 177, 180, 181, 183, 184, 470 risks, 94, 165, 166, 167, 169, 172, 174, 177, 183, 184 river basins, 170, 318, 326 rivers, 1, 4, 12, 21, 22, 38, 63, 71, 79, 157, 169, 170, 172, 174, 178, 419, 420, 453, 504, 523, 578, 595, 600 robustness, 559, 561, 566, 567, 570 rocky, 137, 245, 448 role-playing, 493 rolling, 135, 298 Rome, 27, 105, 302, 329, 604 roughness, 68, 76, 86, 113, 123, 339, 432 routines, 89, 90, 152 routing, 589 rovers, 389 Royal Society, 285, 554 RPGs, 487 runoff, 5, 22, 78, 132, 133, 134, 139, 140, 141, 142, 145, 147, 148, 149, 150, 245, 317, 320, 322, 323, 324, 325, 326, 328, 331, 420, 430, 431, 448, 453, 465, 470 rural, 168, 169, 171, 178, 181, 183, 319, 411, 416, 420, 425, 426, 428, 431, 433, 465, 468, 471, 503, 507 rural areas, 169, 178, 319, 411, 416 rural people, 181, 183 rural population, 420, 425, 426, 465 Russia, 276, 388 Russian, 219
S SAC, 406 safety, 169, 179, 360, 381, 391, 599, 600 sales, 90 saline, 170, 171 salinity, 24, 34, 70, 73, 76, 79, 98, 104, 105, 107, 306 salinization, 306 salt, 86, 380, 381, 385 salts, 156 sample, 45, 50, 59, 75, 86, 93, 140, 175, 222, 239, 276, 309, 356, 377, 414, 437, 438, 441, 454, 475, 524, 560, 562, 579, 580, 599 sample mean, 437, 441
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
628
Index
sampling, 11, 13, 36, 45, 46, 63, 64, 68, 73, 74, 76, 82, 85, 94, 102, 104, 109, 122, 221, 578, 587, 589, 594, 597, 598, 601, 605 sand, 94, 170, 219, 297, 340, 341, 395, 397, 398, 400, 495 sanitation, 412, 413, 425, 427, 428, 465 SAR, 76, 86, 87, 88, 103, 105, 109, 269, 270, 271, 333, 334, 336, 337, 339, 341, 342, 344, 345, 346, 385, 391, 392, 393, 394, 402, 405, 406, 407, 408 satellite imagery, 5, 62, 74, 80, 85, 103, 109, 212, 217, 225, 245, 305, 412, 414, 416, 417, 419, 421, 423, 470, 576 satellite-borne, 100 saturation, 113, 127, 270, 336, 439, 466 Saudi Arabia, 405 savannah, 265 savings, 178, 514 scaffold, 276 scalability, 522, 588, 589 scalar, 47, 48, 257 scaling, 120, 169, 221, 289, 331, 470 Scandinavia, 276 scarcity, 70, 306, 318, 594, 596 scatter, 256, 269, 306, 341, 433, 578 scatter plot, 433, 578 scattering, 39, 75, 76, 254, 255, 264, 334, 394, 442 scheduling, 129 schema, 480, 482, 530 schemas, 531 Schmid, 282 scholarship, 101, 181 school, 486, 502, 511 scientific community, 4, 69, 259, 269 scientific method, 574 scores, 323, 506, 507, 544 SCS, 9, 10, 11, 12, 13, 14, 15, 16, 17, 20, 21, 22, 23, 24, 25, 26, 131, 132, 133, 140, 141, 147, 148, 149, 150, 322, 330, 431 sea ice, 71, 106 sea level, 71, 82, 85, 219, 384, 406, 434 seabirds, 81 search, 2, 338, 420, 520, 529, 581, 582 searching, 335, 522, 528, 545, 567, 579 seasonal factors, 433 seawater, 152, 155, 156 second generation, 69 secondary data, 180 security, 5, 175, 306, 317, 318, 525, 529, 575, 578, 589, 594, 603 sediment, 23, 26, 30, 35, 38, 63, 134, 148, 242, 339, 341, 407, 578 sediments, 34, 38, 71, 100, 339, 341, 344, 385, 397 seed, 554, 601
seeding, 306 seedlings, 313 segmentation, 310, 438 seismic, 402, 405 selecting, 158, 160, 389, 562, 564 Self, 310, 436, 603 self-organization, 589 semantic, 528, 590 semantic information, 590 semantics, 477 semiarid, 132, 147, 245, 300, 301, 318 Senegal, 487 sensitivity, 120, 121, 122, 276, 344, 366, 375, 570, 571, 593 sensor nodes, 593 sensor technology, 594 sensors, 5, 33, 35, 37, 39, 46, 52, 54, 60, 68, 69, 70, 74, 75, 76, 80, 83, 100, 101, 106, 109, 213, 250, 252, 255, 263, 267, 281, 288, 294, 295, 335, 344, 390, 414, 416, 431, 432, 489, 587, 588, 590, 593, 594, 595, 596, 597, 598, 599, 600, 601, 602 separation, 392, 394, 433, 467, 531 series, 27, 30, 37, 59, 69, 70, 71, 83, 99, 100, 114, 119, 121, 125, 126, 127, 152, 181, 187, 189, 192, 193, 201, 209, 240, 250, 306, 386, 391, 392, 394, 404, 595, 596, 597, 598, 599, 605 service provider, 519, 520 services, iv, 71, 99, 102, 212, 317, 318, 412, 413, 415, 427, 477, 503, 517, 519, 520, 521, 522, 523, 524, 528, 532, 533, 534, 535, 536, 537, 539, 541, 548, 552, 595, 602 settlements, 5, 219, 222, 232, 233, 319, 411, 412, 413, 414, 415, 416, 418, 419, 420, 421, 422, 423, 426 severity, 169, 183, 319, 323, 326, 327, 328 sewage, 22, 450, 454, 523 Shanghai, 102 shape, 77, 135, 216, 223, 224, 234, 270, 306, 325, 334, 335, 414, 415, 418, 433, 473, 502, 568, 582 shaping, 168, 476, 483, 488, 489 sharing, 165, 166, 481, 491, 513, 516, 517, 518, 521, 523, 524, 525, 526, 527, 528, 529, 533, 534, 535, 536 shear, 355 shear strength, 355 sheep, 139 shellfish, 10 shelter, 173, 174, 178, 180 shocks, 402 shoot, 224, 264 short period, 81 shortage, 117, 594 short-term, 153, 178, 276, 321
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Index shoulder, 339 shrimp, 10, 88, 94, 95, 240 shrubs, 193, 207, 268 Siberia, 287 side effects, 516, 523, 526 sigmoid, 53 sign, 179 signals, 74, 76, 85, 239, 270, 334, 387, 389, 391, 392, 593 signal-to-noise ratio, 394 silicate, 467 silicates, 338 silicon, 599 similarity, 341 simple linear regression, 59 simulation, 95, 132, 133, 148, 152, 161, 250, 255, 263, 272, 278, 281, 470, 476, 482, 483, 485, 486, 487, 488, 489, 490, 494, 545, 553, 560, 561, 562, 564, 568, 569, 570, 581, 603 simulations, 81, 100, 159, 267, 268, 273, 276, 485, 487, 582 Singapore, 315, 431, 469, 494, 605 sites, 4, 30, 94, 129, 191, 200, 222, 264, 267, 278, 334, 338, 344, 345, 386, 391, 401, 420, 450, 475, 477, 478, 482, 483, 484, 488, 490, 502, 503, 506, 510, 542 skeleton, 532 skills, 177, 180, 181, 476, 485, 487 skin, 68, 75 Slovakia, 242 sludge, 600 slums, 412, 413, 414, 415, 419, 450 Smithsonian, 242 Smithsonian Institution, 242 smoothing, 430, 586 SNR, 394, 408 SO2, 343 social activities, 420 social capital, 514 social conflicts, 233, 240 social factors, 174 social impacts, 485 social life, 491 social organization, 413 social participation, 487, 491 social structure, 485 socioeconomic, 552, 603 sociological, 487 sociology, 479 software, 36, 76, 88, 89, 90, 91, 95, 98, 102, 151, 152, 154, 157, 160, 161, 163, 164, 180, 181, 183, 189, 209, 210, 222, 320, 341, 477, 482, 483, 484, 485, 486, 487, 491, 492, 496, 519, 520, 521, 524,
629
528, 530, 532, 534, 554, 570, 574, 575, 580, 581, 597, 598, 601, 602 softwoods, 224 SOI, 134, 435, 437, 451, 452 soil, 1, 5, 34, 35, 38, 40, 61, 112, 133, 134, 136, 137, 139, 140, 141, 143, 148, 170, 187, 215, 217, 218, 219, 240, 242, 245, 246, 249, 261, 269, 270, 271, 272, 277, 278, 279, 281, 289, 296, 297, 299, 300, 302, 303, 306, 317, 318, 320, 321, 322, 323, 324, 325, 326, 328, 329, 331, 338, 415, 429, 431, 432, 433, 442, 448, 466, 467, 562, 578, 588, 598, 601 soil erosion, 218, 297, 298, 303, 323, 326, 598 soils, 136, 137, 141, 148, 149, 242, 252, 271, 272, 295, 297, 300, 301, 321, 326, 385, 407, 504 solar, 39, 74, 75, 113, 253, 254, 256, 257, 258, 260, 264, 267, 390, 457, 523 solar energy, 264 solid state, 597, 599 solid waste, 430, 450 sorting, 522 South Africa, 10, 28 South America, 471 Southeast Asia, 27, 186, 318, 322, 329, 330 soy, 543 soy bean, 543 Soyuz, 86, 102 space-time, 573, 577, 579, 580, 581, 583, 584, 606 spatial analysis, 2, 171, 173, 222, 233, 240, 499, 521, 571 spatial frequency, 405 spatial heterogeneity, 223 spatial information, 218, 292, 295, 299, 521, 524, 525, 535 spatial location, 76 spatial processing, 521, 534 spatiotemporal, 575 specialization, xi species, 3, 4, 9, 10, 11, 13, 16, 17, 19, 20, 22, 23, 24, 26, 27, 29, 30, 35, 44, 68, 81, 87, 88, 90, 94, 97, 98, 102, 108, 109, 138, 186, 187, 193, 217, 219, 224, 233, 239, 243, 245, 256, 338, 354, 543, 544, 551, 574, 575, 578 species richness, 187, 217, 543, 544 specific heat, 113 specificity, 259 spectral signatures, 38, 45, 47, 48, 66, 265, 269 spectroscopy, 345 spectrum, 6, 37, 46, 74, 75, 113, 121, 256, 306, 392, 416, 431, 444, 472 speed, 1, 13, 53, 75, 76, 88, 89, 90, 113, 147, 175, 319, 386, 387, 388, 601 speed of light, 386, 388 spills, 71
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
630
Index
sports, 35 spreadsheets, 501 SQL, 91, 522, 532, 533, 597 squatter, 412, 423 Sri Lanka, 111, 115, 116, 117, 128, 315 St. Louis, 468 stability, 123, 133, 342, 350, 355, 356, 374, 380, 382, 404, 567, 570 stabilization, 170 stages, 34, 55, 70, 82, 87, 151, 153, 154, 156, 217, 225, 228, 238, 239, 256, 293, 360, 395 stainless steel, 599, 600 stakeholder, 536, 575, 577 stakeholders, 2, 4, 6, 165, 168, 177, 180, 481, 516, 517 standard deviation, 366, 404, 567 standard error, 464 standard of living, 4, 152 standardization, 267, 528 standards, 99, 346, 350, 413, 482, 483, 516, 517, 518, 519, 520, 521, 524, 526, 528, 529, 533, 534, 536, 597 state planning, 502 statistical analysis, 77 statistics, 13, 95, 128, 189, 193, 225, 272, 293, 294, 296, 328, 449, 464, 484, 489, 574, 589, 590, 593, 605, 606 steady state, 159, 360, 361, 390 steel, 599, 600 stigma, 412, 421 stimulus, 588 stochastic, 163, 354, 360, 560, 587 Stochastic, 604 stochastic model, 360 stochastic ordering, 587 stock, 3, 102, 104, 178, 218, 224, 228, 239, 271 storage, 2, 5, 76, 131, 140, 148, 149, 152, 168, 178, 216, 217, 218, 221, 225, 237, 238, 239, 240, 245, 247, 310, 430, 448, 452, 454, 465, 522, 524, 528, 533, 540, 555, 597 storms, 170, 178 strain, 360, 361, 362, 364, 365, 366, 367, 368, 369, 370, 371, 374, 375, 376, 377, 600 strains, 361 Strait of Gibraltar, 85 strategies, 2, 5, 7, 35, 74, 179, 180, 215, 249, 259, 260, 261, 270, 289, 318, 333, 344, 425, 440, 466, 511, 515, 540, 556, 560, 606 strategy use, 259 stratification, 34, 79 stratified sampling, 93 streams, 35, 135, 170, 243, 426, 475, 504, 578, 580, 590, 593
strength, 75, 121, 127, 166, 254, 269, 292, 307, 355, 356, 357, 358, 360, 378, 380, 381, 382, 390, 391, 476 stress, 3, 4, 10, 34, 86, 111, 114, 124, 125, 128, 129, 261, 262, 306, 355, 360, 361, 362, 366, 375, 578 stress factors, 261 stretching, 78, 79 structural characteristics, 216 structuring, 158, 528 students, ix, 97, 180, 554, 603 subjective, 112, 565 subjectivity, 377 sub-Saharan Africa, 330, 413 subsistence, 319 substances, 38, 514 substrates, 269, 272 subtraction, 147 suburban, 505 suburbs, 485 sugar, 305 sugarcane, 293, 295, 296, 302, 305, 313 Sumatra, 246 summer, 10, 12, 29, 78, 79, 87, 94, 258, 434, 577, 603 Sun, 38, 39, 253, 284, 287, 288, 416, 417, 520 sunlight, 75 supervision, 516, 523, 529 supply, 4, 87, 116, 118, 119, 120, 125, 127, 152, 154, 306, 397, 427, 500, 502, 507, 510, 514, 521, 553 surface area, 252, 264, 270, 431 surface energy, 111, 112, 113, 114, 128, 264, 467 surface layer, 87, 109, 276 surface ocean, 85, 100 surface roughness, 76, 86, 113 surface structure, 254 surface water, 81, 153, 157, 266, 297, 420, 576 surface wave, 75 surplus, 454 surprise, 134 surveillance, 575, 576, 577, 578, 580, 581, 603, 606 survival, 80, 108, 153, 299 susceptibility, 216, 234, 578 sustainability, 34, 153, 156, 177, 239, 241, 291, 292, 484 sustainable development, ix, 5, 6, 7, 10, 131, 291, 302, 473, 484, 513, 514, 535 swelling, 374 symbolic, 604, 605 Symbolic Dynamic Filtering (SDF), 592 symbols, 592, 593 synchronization, 483 synchronous, 492
Index synergistic, 575 syntactic, 528 syntax, 482 system analysis, 295 systemic lupus erythematosus, 606 systemic sclerosis, 606 systems, 2, 3, 4, 5, 6, 36, 52, 71, 72, 77, 81, 82, 86, 88, 99, 102, 103, 105, 132, 147, 152, 154, 155, 156, 159, 164, 175, 209, 215, 259, 278, 291, 292, 296, 297, 302, 306, 350, 351, 352, 354, 363, 364, 381, 382, 422, 430, 433, 453, 470, 478, 480, 481, 483, 484, 485, 487, 488, 489, 491, 496, 499, 511, 514, 515, 516, 521, 522, 523, 524, 526, 528, 529, 533, 534, 537, 559, 560, 577, 587, 599, 602, 605, 606
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
T T and C, 29 Taiwan, 12, 21, 29, 244, 493 Taiwan Strait, 21, 29 tangible, 488, 489, 495 tanks, 115, 434 targets, 74, 338, 344, 394 taxation, 510 taxonomy, 4, 579 teachers, 97 teaching, 511 technicians, 598 technocratic, 500 technology, ix, xi, 2, 7, 71, 97, 99, 100, 101, 105, 132, 165, 168, 179, 181, 212, 218, 291, 292, 350, 411, 414, 426, 429, 475, 476, 477, 482, 483, 484, 486, 489, 492, 517, 520, 521, 523, 534, 576, 580, 594, 596, 597, 598, 603 technology transfer, 603 Tehran, 534 Tel Aviv, 422 telecommunications, 281, 492 temperature, 1, 3, 20, 21, 22, 23, 24, 25, 35, 68, 69, 74, 75, 77, 80, 82, 83, 84, 85, 86, 98, 100, 104, 105, 107, 109, 113, 114, 117, 122, 123, 124, 125, 127, 129, 135, 187, 188, 219, 250, 257, 279, 319, 360, 362, 382, 386, 427, 431, 432, 433, 434, 442, 443, 444, 454, 455, 457, 458, 459, 461, 462, 466, 467, 468, 469, 470, 471, 580, 588, 597, 599 temperature gradient, 83, 113 temporal distribution, 193, 194, 198, 278, 404 tensile, 355, 356, 357, 360, 378, 380 tensile strength, 355, 356, 357, 360, 378, 380 tenure, 412 terraces, 298 territory, 34, 35
631
Texas, 407, 471 Thailand, 11, 22, 27, 246, 287, 317, 318, 319, 324, 326, 327, 328, 329, 330 thawing, 342 thermal properties, 113, 432 thermodynamic, 360 theta, 362 thinking, 501 third order, 386 threat, 10, 217, 291, 297, 322, 328, 540, 541, 543, 550, 551, 552, 556 threatened, 4, 169, 543 threatening, 217, 318, 539 threats, 428 three-dimensional, 73, 159 threshold, 40, 44, 53, 74, 93, 94, 185, 187, 189, 201, 209, 404, 560, 602 thresholds, 93 tides, 70, 87 timber, 178, 224, 237, 239, 245 time consuming, 147, 167, 356, 362, 380, 491 time frame, 128, 171 time lags, 90 time periods, 132, 149, 421, 500, 565 time series, 27, 69, 83, 100, 114, 121, 125, 126, 127, 152, 187, 209, 240, 306, 394, 404, 597, 598, 599, 605 timing, 156 tin, 149 TIR, 271, 431, 432, 436, 442, 443, 466 Tokyo, 9, 320, 329, 330, 342 tolerance, 350, 377 top-down, 177 topographic, 78, 82, 121, 140, 159, 160, 170, 221, 279, 335, 336, 344, 392, 402, 403, 404, 408, 442 topological, 594 topology, 77, 528, 589, 590 total product, 154 total utility, 549, 550, 556 toughness, 382 tourism, 35 toxic, 10, 26, 27, 28, 30 toxic effect, 10 toxin, 27 tracking, 73, 85, 88, 389, 390, 394, 603, 605 trade, 1, 2, 281 trade-off, 1, 2 trading, 541, 551, 552 traffic, 412, 425, 427, 430, 453, 465, 502, 509 training, 53, 54, 55, 56, 58, 180, 191, 222, 294, 306, 307, 308, 309, 356, 363, 364, 365, 366, 374, 376, 377, 429, 440, 460, 472, 535, 574, 603 trajectory, 584, 591
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
632
Index
trans, 160 transactions, 492 transducer, 599 transfer, 61, 99, 113, 114, 123, 133, 183, 258, 260, 264, 268, 270, 280, 289, 342, 387, 432, 433, 482, 483, 491, 603 transference, 53 transformation, 1, 40, 61, 66, 171, 221, 439, 466, 475, 477 transformations, 213, 565 transition, 1, 7, 12, 85, 215, 221, 223, 230, 231, 233, 238, 243, 247, 292, 352, 355, 483, 555, 592 transition period, 12 transitions, 78, 217, 223, 232, 241 translation, 93, 221 translation transformation, 221 translational, 591 transmission, 38, 257, 265, 268, 593, 597, 599 transmits, 88, 389 transparency, 2, 573 transparent, 34, 503, 510, 530 transpiration, 112, 320, 433, 466 transplantation, 311, 312 transport, 22, 68, 79, 81, 82, 85, 86, 87, 104, 106, 107, 113, 250, 259, 270, 385, 395, 396, 426, 520, 532, 579 transportation, 1, 131, 173, 239, 427, 510, 540 transpose, 438 traps, 453 travel, 37, 87, 386, 426, 564, 598 travel time, 386, 426 trawlers, 89 trawling, 96 treaties, 186 tree cover, 5, 185, 186, 187, 188, 189, 190, 193, 194, 198, 200, 201, 212 trees, 138, 220, 221, 238, 273 triangulation, 388 Triassic, 242 tribal, 139 Trichodesmium, 19, 23 triggers, 342 trophic chains, 33 tropical forest, 186, 211, 217, 242, 269, 442, 553 tropical forests, 186, 211, 217, 269 tropical rain forests, 193 troposphere, 386 trust, ix, 177, 501, 541 Tsunami, 350 tundra, 263 tunneling, 385 turbulence, 75 turbulent, 123, 276
Turkey, vi, 5, 128, 215, 218, 219, 220, 237, 241, 242, 243, 244, 245 turnover, 34 two-dimensional, 393 two-way, 386 typhoon, 167, 171, 175, 178 typology, 481, 577, 579
U U.S. Department of Agriculture (USDA), 148, 322, 330, 331 U.S. Geological Survey, 164 ultra-fine, 334, 344 UML, 535 uncertainty, 2, 6, 93, 94, 95, 96, 309, 350, 352, 377, 464, 559, 561, 562, 565, 567, 568, 570, 571, 574, 577, 581, 583 UNDP, 182 unemployment, 420 UNEP, 185, 188, 203, 213, 319, 327, 329, 330 UNESCO, 27, 77, 108, 320, 322, 323 UN-Habitat, 412, 423 unification, 154 uniform, 254, 276, 299, 385, 395, 443, 603 United Kingdom (UK), 62, 95, 182, 242, 244, 534, 535 United Nations (UN), 99, 182, 329, 412, 413, 423, 471 United States, 26, 27, 106, 132, 133, 139, 140, 276, 346, 429, 468, 499, 502, 510, 542, 543, 552, 554, 604, 606 univariate, 60, 440, 591 universe, 307 universities, 180, 576 updating, 53, 318, 516, 524, 536, 565 urban areas, 4, 171, 423, 425, 426, 428, 430, 432, 453, 457, 466, 471, 483, 503 urban centers, 411, 420, 421 urban population, 413, 426, 454, 465 urban settlement, 233 urbanisation, 425, 426, 427, 428, 431, 448, 450, 451, 453, 464, 465, 473 urbanization, 6, 10, 168, 169, 186, 218, 219, 227, 237, 242, 246, 412, 415, 459, 469 urbanized, 275 user-defined, 354, 404, 502
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
Index
633
V
W
validation, 6, 60, 91, 189, 209, 210, 249, 266, 275, 281, 364, 384, 436, 542, 553, 554, 555, 560, 561, 564, 570 validity, 380, 476, 569 vapor, 112, 113, 114, 123, 124, 266, 276 variability, 4, 27, 34, 71, 74, 78, 80, 82, 83, 94, 95, 96, 103, 104, 106, 111, 112, 113, 125, 127, 129, 133, 134, 146, 186, 233, 253, 260, 261, 264, 266, 272, 276, 295, 299, 301, 326, 378, 430, 437, 438, 466, 472 variables, 31, 33, 51, 52, 53, 59, 66, 75, 82, 93, 100, 140, 171, 222, 227, 228, 260, 273, 323, 329, 362, 363, 374, 439, 440, 468, 525, 562, 563, 565, 566, 568, 578, 580, 592 variance, 59, 61, 93, 179, 185, 189, 201, 272, 377, 378 variation, 30, 44, 51, 52, 59, 104, 122, 129, 146, 179, 257, 258, 259, 268, 270, 274, 306, 336, 366, 385, 390, 391, 392, 393, 402, 415, 457, 467, 559, 563, 565, 566, 580, 600, 606 vector, 14, 46, 48, 54, 76, 78, 93, 109, 110, 143, 307, 309, 389, 437, 438, 440, 441, 445, 450, 452, 460, 521 vegetables, 138 vehicles, 73, 281, 580, 587 velocity, 22, 114, 123, 276, 349, 382, 387, 395, 402, 431, 443, 591, 600 versatility, 250 vessels, 76, 81, 88, 89, 90, 91, 92, 97 Vietnam, 10, 11, 12, 13, 14, 21, 22, 25, 28, 29, 30, 169, 181, 318, 324, 326, 327, 328 Vietnamese, 22, 166, 180, 182 village, 167, 168, 172, 173, 177, 437, 597 violence, 485 virtual reality, 483, 489, 496 virtual world, 478, 480, 486, 489, 490 virus, 579 visible, 37, 38, 46, 62, 67, 70, 74, 75, 80, 82, 85, 109, 121, 167, 178, 183, 221, 233, 256, 270, 306, 335, 389, 416, 422, 442, 455 vision, ix, 477, 478, 483, 486, 488, 490 visualization, 36, 42, 44, 52, 60, 71, 73, 91, 97, 105, 152, 160, 337, 425, 465, 484, 488, 497, 511, 577, 602 voids, 159, 384, 385, 391, 401 voters, 586 voting, 487, 490, 599 vulnerability, 166, 167, 168, 169, 170, 171, 175, 178, 179, 181, 182, 183, 184, 274, 351, 571, 578, 606
W3C, 482, 520, 528 Wales, 30 war, 417, 459, 465 warning systems, 599 Warsaw, 495, 496 waste disposal, 152, 563, 565, 567, 570 wastes, 169, 599 wastewater, 565 wastewater treatment, 565 water absorption, 37 water quality, 5, 10, 62, 63, 152, 154, 156, 215, 217, 218, 426, 576, 594, 595, 597, 603 water resources, 4, 94, 147, 151, 152, 154, 156, 164, 296, 315, 523, 573, 575, 596 water table, 159, 261, 270, 452, 594, 597, 598 water vapor, 123, 266, 276, 277 watershed, xi, 131, 132, 133, 134, 135, 136, 137, 139, 143, 147, 150, 168, 243, 244, 245, 246, 266, 268, 280, 296, 302, 323, 454, 574, 578, 606 watersheds, 133, 134, 140, 157, 302, 434, 523, 576, 578 wave packet, 87, 590, 591 wave packets, 87, 590 wave propagation, 87 wavelengths, 37, 38, 74, 75, 335, 388, 443 wavelet, 469, 591, 593 wealth, 169, 514 web, 29, 70, 81, 477, 495, 514, 517, 518, 519, 520, 521, 522, 524, 528, 534, 535, 537 Web 2.0, 477, 492, 495 web service, 517, 519, 520, 521, 522, 524, 528, 534, 535, 537 web-based, 514, 534 websites, 13, 329 weight gain, 321 wells, 115, 139, 153, 157, 159, 161, 595, 597, 598, 600, 601 West Nile virus, 577, 579, 604 wetlands, 34, 270, 430, 450, 451, 452, 504 wheat, 293, 295, 296, 302, 306 wildlife, 71, 186, 239, 242, 339 wind, 1, 9, 14, 20, 21, 22, 23, 24, 25, 26, 34, 35, 75, 76, 79, 81, 82, 86, 87, 88, 107, 113, 114, 123, 171, 175, 178, 319, 339, 523, 588, 601 wind gusts, 601 windows, 113, 174, 532 winter, 22, 24, 29, 78, 79, 106, 107, 160, 319, 433, 434 wintertime, 79 wireless, 488, 534, 587, 593, 595, 597, 598, 599, 600, 602
634
Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.
wireless devices, 488 wireless sensor networks, 595 Wisconsin, 277, 284, 502, 503 wisdom, 165, 166 withdrawal, 385, 397, 398 women, 167 wood, 161, 228, 239, 319, 398, 426 wood products, 228, 239 woodland, 242, 426 workers, 2 workflow, 91 working groups, 475, 487, 529 workload, 180, 183 World Bank, 181, 182, 412, 514, 535 World Resources Institute, 331 World Wide Web, 520 WRI, 319, 321, 331 writing, 159, 479 WWW, 475, 482, 483, 488, 491
Index
X X-axis, 20, 55, 56, 58, 126 XML, 482, 483, 496, 517, 519, 520, 521, 522, 523, 524, 529, 533, 534, 536, 537
Y Y-axis, 55, 56, 58 yield, 1, 33, 34, 38, 53, 55, 94, 133, 134, 146, 148, 152, 186, 224, 259, 306, 318, 448, 452, 454, 465, 578
Z Zone 3, 219 zoning, 68, 299, 301, 302, 502, 506, 577 zooplankton, 36, 87