BULLETIN OF THE

NATIONAL NATURAL RESOURCES MANAGEMENT SYSTEM

NNRMS (B) - 38

Satellite Remote Sensing & GIS Applications in Water Resources

December 2013

NNRMS
Department of Space Antariksh Bhavan, New BEL Road Bangalore - 560 231 INDIA

Editorial Advisors
Kiran Kumar AS, Director, SAC Dadhwal VK, Director, NRSC Shivakumar SK, Director, ISAC

Editorial Board
Raghava Murthy DVA, Director, EOS Diwakar PG, Deputy Director (RSA), NRSC Saha SK, Dean (Academics), IIRS

Technical Guidance
Shantanu Bhatawdekar, Associate Director (A), EOS

Technical Support and Compilation

Paul MA, Scientist/Engineer SF, EOS Bandyopadhyay S, Scientist/ Engineer SF, EOS Sameena M, Scientist/ Engineer SE, EOS

For details and inputs, please write to

Director Earth Observations System ISRO Headquarters Antariksh Bhavan New BEL Road Bangalore 560 231 Email: raghava@isro.gov.in Fax: 91-80-2341 3806

Published by
P&PR Unit, ISRO Headquarters on behalf of National Natural Resources Management System (NNRMS) Antariksh Bhavan, New BEL Road Bangalore 560 231

Designed by
Imagic Creatives Pvt. Ltd., Bangalore 560 071

Printed at
Brilliant Printers, Bangalore

PREFACE
Water is a vital component of the planet earth and an essential resource for human life, health, economic growth, and the vitality of ecosystems. All the human civilizations are built around the major rivers. One of the major tasks in 21st century is to conserve and protect the water resources form the threat of pollution, unsustainable use and unscientic wastages towards providing water security and establishing foundation for sustainable biomass-based economy. Therefore, achieving water security is very essential for existence of future humanity on the earth surface. In India, water resources vary unevenly with space and time. India is known for varied climatic zones, covering the areas of highest to lowest rainfall and areas covered by cold desert as well as hot deserts. Further, it is coupled with rapid population growth, intensive agricultural and industrial activities, expanding urbanization - creating enormous pressure on surface and ground water resources. Managing water resources is, therefore, a major challenge to protect resources from unsustainable use in order to ensure water security. Satellite remote sensing has really become a boon for the management, monitoring and exploration of both surface water as well as ground water scientically. The wide spread use of Information Technology (IT) tools and Geographic Information System (GIS) also helps to create an up-to-date information on water resources to manage, monitor and to conserve the nature to improve the quality of life. Extraction of hydrological parameters from multi-spectral Earth Observation data and their interpretations in terms of soil and water conservation, watershed development and ground water prospecting have made considerable impacts in ensuring water security especially in the dryland areas of the country. In these circumstances, this issue of NNRMS Bulletin on water resources attracts great signicance. This issue comprises articles related to various aspects of water resources covering, status of water resources in India, water geography, water resources management, water resources development and modeling, ood monitoring and early warning system, drought assessment and creation of water resources information system. I take this opportunity to thank all those who have contributed articles, and others who have helped in bringing out this issue of Bulletin on Water resources. The efforts made by Dr. J. R. Sharma, Chief General Manger, RCs & OSD, New Delhi and Dr. P. G. Diwakar, Deputy Director (RSA), NRSC for their support in bringing out this issue is also greatly acknowledged. D. V. A. Raghava Murthy Director Earth Observations System

CONTENTS
1 2 3 4 5

Page No. Water Resources of India - Critical Issues and Satellite Technology Options
Venkateshwar Rao V, Sharma JR and Dadhwal VK

1-9

Rainfall Estimation using Satellite Data Pal PK, Atul Kumar Varma and Gairola RM Inventory, Mapping and Monitoring of Surface Water Bodies Suresh Babu AV, Shanker M and Venkateshwar Rao V Geospatial Technology for Inventory and Monitoring of Glacial Lakes and Water Bodies in the Himalayan Region Abdul Hakeem K and Siva Sankar E Monitoring of Irrigation Projects using High Resolution Cartosat Satellite Data
Shanker M, Suresh Babu AV, Simhadri Rao B and Venkateshwar Rao V

11-22

23-33

34-42

43-50

Remote Sensing and GIS for River Morphology Studies

Manjusree P, Satyanarayana P, Bhatt CM, Sharma SVSP and Srinivasa Rao G

51-56

7 8 9

Water Resources and Hydrology of the Western Ghats: Their Role and Signicance in South India
Mysooru R. Yadupathi Putty and Madhusoodhanan CG

57-66

Use of Earth Observation Data to Unearth Sub-Surface Drainages: Potential Groundwater Source in Arid Region of North West India
Bhadra BK and Sharma JR

67-74

Modeling Hydrological Water Balance in the Forested Watershed for Water Management
Gupta PK, Singh RP, Panigrahy S, Chauhan JS, Sonakia A and Parihar JS

75-80

10

Flood Monitoring and Management using Remote Sensing

Srinivasa Rao G, Bhatt CM, Manjusree P, Sharma SVSP and Asiya Begum

81-88

 11

Page No. Geospatial and Hydro-Met Approach for Flood Management in Assam
Diganta Barman, Kundu SS, Jonali Goswami, Ranjit Das, Singh NGR, Arup Borgohain, Rekha Bharali Gogoi, Victor Saikhom, Suranjana B Bora and Sudhakar S

89-92

12

Agricultural Drought: Assessment & Monitoring

Sesha Sai MVR, Murthy CS, Chandrasekar K, Mohammed Ahamed J and Prabir Kumar Das

93-105

13 14

Irrigation Command Area Management using Remote Sensing

Raju PV, Abdul Hakeem K and Venkateswar Rao V

107-116

Remote Sensing Inputs for Feasibility Assessment Studies of Proposed Water Resources Projects
Simhadri Rao B, Suresh Babu AV, Shanker M and Venkateswar Rao V

117-123

15 16

Modeling the Impact of Land Use/Cover Change on the Runoff Water Availability: Case Study for the Narmada River Basin
Gupta PK, Punalekar S, Singh RP, Panigrahy S and Parihar JS

124-129

Hydrological Modeling Approach for Annual Water Resources Assessment- A Pilot Study in the Godavari and Brahmani-Baitarani Basins, India
Durga Rao KHV, Raju PV, Simhadri Rao B, Venkateshwar Rao V and Sharma JR

130-140

17 18 19 20

Snow Melt Runoff Modeling in Himalayan River Basins

Siva Sankar E, Abdul Hakeem K and Simhadri Rao B

141-151

Remote Sensing in Groundwater Modeling

Sudhir Kumar and Sanjay Kumar Jain

152-162

Role of Earth Observation for Grass Root Level Water Resources Planning- Technology Demonstration for a Cluster of Villages in Semi-Arid Region of Rajasthan
Rama Subramoniam S, Manoj Joseph, Bera AK and Sharma JR

163-169

India-WRIS Webgis Design and Development of web Enabled Water Resources Information System of India
Sharma JR and Project Team

170-179

Introduction
Water is the most important renewable and nite natural resources. It regenerates and redistributes through evaporation and rainfall. In India, there is a mismatch between the endowment of natural resources and the population to be supported (Table 1). Based on per capita renewable water availability, India the second most populous country in the world has water enough to meet its peoples needs. The per capita availability of fresh water in the country, which was a healthy 5,177 cubic meters in 1951, has dropped to 1,869 cubic meters in 2001. It is estimated to further decline to 1,341 by 2025 and 1,140 by 2050. The demand by 2050 AD is likely to reach the level of the full utilisable quantum. With the ever-increasing population, there is staggering increase in water requirements for agriculture, domestic and industrial sectors. Table 1: Water Availability (In Billion Cubic Metre) Total Precipitation Total Water Availability Total Utilisable Water Indias Population Indias Water Resources Indias Land Resources 4000 1869 1122 (690 Surface water + 432 Ground Water) 16% of World Population 4% of Global Water Resources 2.5% of Global Land Resources

Source: Annual report 2004-05 of Ministry of Water Resources, Govt. of India.

The spectacular growth in food production in India (from merely 50.8 million tonnes in 1950-51 to an estimated production of more than 235 million tonnes in 2010- 11) is mainly attributed to three major factors; namely, development of high yielding varieties, large expansion in irrigated area and fertilizer use. Among these, irrigation may be considered as a key driver of modern intensive agriculture, since it not only ensures physiological water needs of crop plants for desired growth of photo synthetically efficient new genotypes but also supports efficient utilization of plant nutrients. Realizing the fact that spatial and temporal distribution of monsoon rains is highly uneven and that development of agriculture is very difcult without irrigation, the Government laid major emphasis on creation of irrigation potential during different plan periods.

NNRMS

BULLETIN

Venkateshwar Rao V, Sharma JR and Dadhwal VK National Remote Sensing Centre ISRO, Department of Space, Hyderabad - 500 037, India Email:vvrao@nrsc.gov.in

DECEMBER

WATER RESOURCES OF INDIA CRITICAL ISSUES AND SATELLITE TECHNOLOGY OPTIONS

2013

Agriculture sector is the largest consumer of water (80%). With growing population, urbanization and industrialization in the country the requirements of water from competing sectors like domestic and industrial needs, are increasing.

Precipitation Variability
The long-term average annual rainfall for the country is 1160 mm, which is the highest anywhere in the world for a country of comparable size. The annual rainfall in India however uctuates widely. The highest rainfall in India of about 11,690 mm is recorded at Mousinram near Cherrapunji in Meghalaya in the northeast. In this region rainfall as much as 1040 mm is recorded in a day. At the other extreme are places like Jaisalmer, in the west, which receives barely 150 mm of rain. Though the average rainfall is adequate, nearly three-quarters of the rain pours down in less than 120 days, from June to September. As much as 21% of the area of the country receives less than 750 mm of rain annually while 15% receives rainfall in excess of 1500 mm. Precipitation generally exceeds 1000 mm in areas to the east of Longitude 780E. It reaches nearly to 2500 mm along almost the entire west coast and over most of Assam and sub-Himalayan West Bengal. Large areas of peninsular India receive rainfall less than 600 mm. Annual rainfall of less than 500 mm is experienced in western Rajasthan and adjoining parts of Gujarat, Haryana and Punjab. Rainfall is equally low in the interior of the Deccan plateau, east of the Sahyadris. A third area of low precipitation is around Leh in Kash. The rest of the country receives moderate rainfall. Snowfall is restricted to the Himalayan region (Rakesh Kumar, et al., 2005).

Surface Water Resources of India

From the point of view of surface water resources, India has been divided into 20 river basins. These comprise of 12 major basins each having catchment areas exceeding 20,000 sq km and 8 composite river basins combining suitably together all the other remaining medium and small river systems. The total water potential of these basins is estimated at 187.9 mha.m. A break up of this resource reveals that 105 mha.m. is the runoff from rainfall that ows into rivers and streams including reservoir and tanks. Additional water is received from snow melt (10 mha.m.), ow from outside India (20 mha.m.), from groundwater (37 mha.m.) and from irrigated areas (11 mha.m.) making a total of about 183 mha.m. The largest potential of water is available in Ganga/ Brahmaputra/Barak and others making a total of 117 mha.m. followed by Godavari and by west owing rivers from Tapi to Tadri each having an average annual potential of more than 10 mha.m. Due to extreme variability in precipitation, which disallows assured storage of all the water, due to non-availability of storage space in hills and plains, evaporation losses and water going to the sea and outside India, it is anticipated that utilizable surface water resources would be 69 mha.m. which will be utilized by the year 2025. It is assessed that on full development, 76 mha.m. area would be irrigated through surface water resource

Water Resources Management Issues

Planning and development of water resources related aspects need to be gover ned by National Perspectives. India is endorsed with a large network of 12 major river basins covering 256 mha, 46 medium river basins covering about 25 mha besides other water bodies like tanks and ponds covering 7 mha with the ultimate irrigation potential of 140 mha. Signicantly large gaps exist in the ultimate irrigation potential, creation and the utilization and the water-use efciency in the country is reported to be only 25-30 per cent, which has adequate scope for improvement by employing various modern techniques. Floods and drought affect vast areas of the country, transcending state boundaries. A third of countrys geographical area is drought-prone. Floods affect an average area of around 9 mha. The drinking water needs of people and livestock have also to be met. Domestic and industrial water needs have largely been concentrated in or near the principal cities, but the demand from rural society is expected to increase

for water for Hydro and Thermal power generation and for other industrial uses is also likely to increase substantially. Another important aspect is water quality. It is required to eliminate the pollution of surface and ground water sources, to improve water quality and to step up the recycling and re-use of water. Projections show, in the coming decade there may be acute water crisis in terms of availability and supply. The problems associated with the water resources development are varied and diversied. They are; spatial and temporal variations in water availability of the country; Falling per capita water availability of the country; Rising multi-sectoral water demand; Complex nature of command area problems; Signicantly large gaps exist in the ultimate irrigation potential, creation and the utilization; Rapid rate of reservoir sedimentation; Frequent oods severely affecting the development process in ood prone areas; Recurring drought in certain regions; Over-exploitation and depletion of the ground water resources; Deteriorating water quality and environment To address the above issues, it calls for use of available water resources through surface water capture and storage, long distance conveyance and inter-basin transfer, ground water exploitation, watershed management, conjunctive use of surface and ground water and de-salinisation.

Assessment of Water Resources

At present, the natural water resource of India is estimated at around 1869 km3, and out of this 690 km3 is considered useable from surface waters and another 433 km3 is considered useable from the ground water. The main deciency of the current assessment is that it ignores the principle of the unied nature of the water resources. The important interactions between the surface waters and the ground waters, through articial recharge from irrigated areas and possible recharge from river beds on one hand, and the contribution of the groundwater to the surface source through base ows on the other hand are somewhat ignored. A re-assessment, based on the complete modeling of the hydrologic cycle in the land phase is very important on its own and has to be done by setting up a hydrologic model for the whole basin. The model either has to be distributed one, or is to be made as a distributed model through an assembly of a large number of lumped models. The NRSC and CWC have come together to develop a tailor made model for the Indian water assessments using a simplied Thornthwaite and Mather model with satellite derived spatial inputs.

Agricultural Water Management Irrigation Status in India

Out of the 141.84 mha of net sown area in the country, 63.19 mha (44.5%) is under irrigation in 2009-10. Out of this 35% is under canal irrigation and more than 55% is through tube wells and other wells. Tank irrigation is only less than 10% and conned mainly to the peninsular portion of the country. Amongst the different states, the net irrigated area varies from 12.3% in Mizoram to as high as 98% in Punjab. In Assam, Himachal Pradesh, Odisha and other 3

NNRMS

BULLETIN

DECEMBER

2013

sharply as the development programmes improve economic conditions in the rural areas. The demand

North Eastern States, it is predominantly canal system while in Gujarat, Punjab, Rajasthan and Uttar Pradesh the ground water is the major source. In Tamil Nadu, tank irrigation is an important irrigation source with almost a quarter of the irrigated area under this system. In the other southern states as well as Odisha, tank system is also in vogue to a considerable extent. As regards groundwater, though only 58% ground water development of the utilizable resources is used at present, water tables are progressively declining at an alarming rate of 1 m per year due to excessive extraction. Presently about 435 blocks/ Mandals/ watersheds are grouped into critical/ dark category and 422 blocks/ mandal/ watersheds semi critical/ grey category. Construction of new wells in 435 dark areas is not allowed while a cautious slower development is required in 422 Grey areas. Over different Five Year Plans (FYPs), Net Sown Area (NSA) in India has increased from 125.95 mha in rst FYP to 139.66 mha in Fourth FYP and remained almost constant thereafter at 141.84 mha . However, the Gross Sown Area (GSA) in India has increased consistently from about 140 mha in rst FYP to 189.84 mha in tenth FYP with the growth of 0.58 % per year. Consistent increase in GSA was due to improvement in the cropping intensity over the planning periods from 111.17% in rst FYP to 136.05% in tenth FYP. One of the reasons for improvement in cropping intensity might be assured irrigation supply through various major, medium and minor irrigation projects. This is reected through signicant increase in the net as well as gross irrigated area of the country with the growth of 2.08 and 2.52% per annum, respectively, during 1950-2007. Comparatively higher growth in Net Irrigated Area (NIA) and Gross Irrigated Area (GIA) led to their increasing share in NSA and GSA, respectively over different FYPs. At the end of tenth FYP, about 42% of the GSA as well as NSA was irrigated. Irrigation intensity also witnessed increasing trend during the period under consideration, Cropping intensity, irrigation intensity and share of NIA and GIA in NSA and GSA, respectively witnessed increasing trend in all the regions of the country over different FYPs reecting overall improvement in irrigation status and consequently, agriculture.

Strategies to Enhance Water Productivity in Irrigated Agriculture

Most of the irrigation projects are operating at dismally low efciency of 35%, wasting enormous amounts of water and increasing waterlogged and saline areas and depriving the tail end farmers of their legitimate right. With increasing pressure on land and water, immediate action is needed to improve and increase irrigation efciency to 60% for surface water and 75-80% for groundwater. The data presented above shows that, with almost no increase in NSA likely in near and distant future there will be greater pressure to cultivate the available land intensively by increasing the cropping intensity from 132% to 145% by 2025 to step up Gross Cultivated Area (GCA) from 194 mha to 210 mha. In fact in the last 25 years the increase in GCA has been for non-food grain crops and area under grain hovering between 123 mha and 128 mha. Likewise, not much change in switch over from rainfed to irrigated conditions could be expected and around 80 mha areas would continue to be rainfed.

Groundwater Resources Management

In India more than 85% of Indias rural domestic water requirements, 50% of its urban water requirements and more than 50% of its irrigation requirements are being met from ground water resources. The increasing dependence on ground water as a reliable source of water has resulted in its large-scale and often indiscriminate development in various parts of the country, without due regard to the recharging capacities of aquifers and other environmental factors. Ground water extraction for various uses and evapotranspiration from shallow water table areas constitute the major components of ground water draft. In general, the irrigation sector remains the main consumer of ground water.

Water Board, 2006), about 92 % of which is utilized for irrigation and the remaining 8% for domestic and industrial uses. Hence, the stage of ground water development, computed as the ratio of ground water draft to total replenishable resource, works out at about 58% for the country as a whole. However, the development of ground water in the country is highly uneven and shows considerable variations from place to place. Though the overall stage of ground water development is about 58%, the average stage of ground water development in North Western Plain States is much higher (98%) when compared to the Eastern Plain States (43%) and Central Plain States (42%). As a part of the resource estimation following the Groundwater Estimation Committee (GEC) norms, the assessment units have been categorized based on the stage of ground water development and long term declining trend of ground water levels. As per the assessment, out of the total of 5723 assessment units in the country, ground water development was found to exceed more than 100 % of the natural replenishment in 839 units ( 14.7%) which have been categorized as Over-exploited. Ground water development was found to be to the extent of 90 to 100 percent of the utilizable resources in 226 assessment units ( 3.9 %), which have been categorized as Critical. 550 assessment units with stage of ground water development in the range of 70 to 100 % and long-term decline of water levels either during pre- or post-monsoon period have been categorized as Semi-Critical and 4078 assessment units with stage of ground water development below 70% have been categorized as Safe. 30 assessment units have been excluded from the assessment due to the salinity of ground water in the aquifers in the replenishable zone. In addition to the resources available in the zone of water level uctuation, extensive ground water resources have been proven to occur in the deeper conned aquifers in the country, a major chunk of which is in the Ganga-Brahmaputra alluvial plains (Romani, 2006). Such resources are also available in the deltaic and coastal aquifers to a lesser extent. These aquifers have their recharge zones in the upper reaches of the basins. The resources in these deep-seated aquifers are termed Instorage ground water resources. The quantum of these resources has been tentatively estimated as ~10,800 bcm. Though the ground water resources in these aquifers are being exploited to a limited extent in parts of Punjab, Haryana and western Uttar Pradesh, detailed studies are to be taken up to fully understand the yield potentials and characteristics of these aquifers (Jha, 2009).

Adoption of Space Technology in Water Resources Management

Space borne multispectral measurements at regular intervals have helped evaluating the performance in many irrigation projects across the country. The anticipated increase in irrigated area, equitable distribution, crop productivity, extent and severity of water logging and salinity/alkalinity have been studied in major irrigation command projects in India (NRSA, 1998). High resolution satellite data (LISS IV) is used in evaluating the impact of National Project for Repair, Restoration and Renovation (NPRRR) program aimed at restoration of lost irrigation potential under minor irrigation schemes. Use of high resolution remote sensing data for monitoring the irrigation infrastructure is a major recent initiative. Countrys program of Accelerated Irrigation Benet Program ( AIBP) of large scale creation of irrigation facilities towards achieving food security is being monitored through high resolution satellite data from Cartosat-1 & Cartosat-2 for its effective implementation (NRSC, 2010). 5

NNRMS

BULLETIN

DECEMBER

2013

The ground water draft for the country as a whole has been estimated as 231 bcm (Central Ground

Realistic appraisal of reservoir capacity is essential for appropriate utilisation plans. Multi temporal satellite data have been used as an aid for capacity surveys of many reservoirs in a cost and time effective manner in India.

A National action plan of sedimentation survey of 124 reservoirs using remote sensing technology is being implemented. Airborne Laser based Terrain Mapping (ALTM), in conjunction with detailed mapping through digital camera provide valuable information on terrain characteristics in terms of topography, association, detailed land use/ cover, geological features, etc., which were effectively used in water resources infrastructure projects viz. Interlinking of Rivers Project (ILR) . Space technology derived terrain (DEM) and land use-land cover information along with hydro-met data are being used in hydrological modeling for a fresh assessment of national water resources in the country (NRSC, 2012). Space inputs are operationally used in hydrologic modeling for ood forecasting and ood inundation simulation purposes. River Basin-specic ood forecasting models have been developed and are used in real-time by the concerned departments. Snow and glacier investigations and snow melt runoff forecasting are yet other areas where satellite remote sensing imagery is providing information on retreating glaciers as well as possible potential snow melt run-off. Seasonal and short term (weekly) forecasts of snowmelt runoff are being provided for Sutlej and Beas and Parabati basins in Western Himalayas by the NRSC since1980s. Development of snow melt runoff models for ve major Himalayan river basins is under development for providing both short term (fortnightly) and seasonal snow melt runoff forecasts. Inventory and monitoring of glacial lakes and water bodies in entire Himalayas covering Indian river basins is underway. Satellite Remote Sensing in conjunction with collateral eld data has made possible to provide drinking water sources to the rural India using groundwater prospects maps under the umbrella of Rajiv Gandhi Drinking Water Mission (RGDWM). So far ~ 3,00,000 bore wells and 12,000 recharge structures have been implemented by the respective State Governments (NRSC, 2012). Multi-year satellite data is also used to monitor the impact of the implementation of watershed management programmes. Implementation of appropriate rain water harvesting structures in selected watersheds under Integrated Mission for Sustainable Development (IMSD) programme has demonstrated the signicant benets by way of increased ground water recharge and agricultural development of once barren areas.

Institutional arrangements for use of Remote Sensing for Water Resources Management
The space technology application activities in the country are co-coordinated by the National Natural Resources Management System (NNRMS) established under the Planning Commission, for which Department of Space (DOS) is the nodal agency. NNRMS has been facilitating inventory of the countrys natural resources for its optimal utilization through application of Remote Sensing technology in conjunction with the conventional methods. The Planning Committee of NNRMS (PC-NNRMS) is the apex of NNRMS, which is chaired by Member (Science) Planning Commission with Secretaries of Ministries/Departments as members. Ten thematic Standing Committees support the PC-NNRMS, of which Standing Committee on Water Resources (SC-W) is one. Standing Committee on Water Resources (SC-W) since its formation in the year 1984 has been providing the essential thrust and guidelines for many important projects in water resources sector and operationalisation of remote sensing using IRS satellite data. Ministry is periodically implementing the plan scheme on remote sensing applications in water resources development and management, through which several important remote sensing projects are taken up

creation of national digital water data bases, development of national water resources information system, snow & glacier studies. Use of high resolution remote sensing data for monitoring the irrigation infrastructure and feasibility studies for interlinking of river basins are major recent initiatives. Development of a national Water Resources Information system (India WRIS) is underway jointly by the Department of Space and Ministry of Water Resources. There is a need to institutionalise the space technology applications in Water Resources by integrating and strengthening the existing Central and State level agencies. This initiative would help to adopt improved methodological framework for measuring/monitoring the progress of the countrys water sector towards sustainable development and management

Strategies for Improved Water Management

Structural and non-structural measures are required for mitigating the floods and droughts. Mathematical models are needed for forecasting the monsoon rainfall accurately, which may be utilized by the decision makers and farmers for adopting appropriate strategies for management of droughts and oods. There is a need for increasing the availability of water by better management of existing storages and creation of additional storages by constructing small, medium and large sized dams, considering the economical, environmental and social aspects. The availability of water resources may be further enhanced by rejuvenation of dying lakes, ponds and tanks and increasing the articial means of groundwater recharge. In addition to these measures, Inter-basin transfer of water provides one of the options for mitigating the problems of the surplus and decit basins. However, for inter-basin transfer of water, scientic studies need to be carried out for establishing its technical and economic feasibility considering the environmental, social and ecohydrological aspects. Integrated and coordinated development of surface water and groundwater resources and their conjunctive use should be envisaged right from the project planning stage and should form an integral part of the project implementation. Mathematical modeling of hydrological processes would provide an opportunity to both the research hydrologists and the water resources engineers involved in developing the integrated approaches for planning, development and management of water resources projects for sustainable development as well for preserving the ecosystems. Concerted and holistic efforts are required in increasing the overall water use efciency at system level which would be achieved through various measures like timely execution of projects, AIBP monitoring, minimizing the losses, better operational efciency through stake holders participation, implementation of on farm water management technologies, conjunctive use of water and changes in irrigation policy. Regular water quality monitoring programme has to be launched for identifying the areas likely to be affected the water quality problems. For maintaining the quality of freshwater, water quality

NNRMS

BULLETIN

DECEMBER

2013

in the elds of reservoir sedimentation, irrigation command area monitoring, river basin development,

management strategies are required to be evolved and implemented. Minimum ow must be maintained in the rivers for meeting the criteria of Environmental Flow Requirements (EFR). As the country has embarked upon creating national digital water data bases and with availability of high resolution satellite data, the stage is now set for developing locale-specic strategies for beneting at the grass-root level. Such systems provide an integrated approach for water resources management considering the various water-related disciplines together with socio-economic aspects. Recent studies have brought out the adverse impact of climate change on water resources in the country. Realistic estimation of water resources in spatial and temporal domain is required. Hence, in the coming years efforts would be made to address the above issues with advanced technologies. Periodic Basin-wise water resources assessment, ood forecasting and ood inundation modeling of all ood prone areas, periodic monitoring of Irrigation projects for monitoring Irrigation infrastructure and irrigated area assessment, development of Decision Support System (DSS) for real-time irrigation water management, micro level ground water aquifer mapping and modeling for sustainable management of groundwater resources, snow cover dynamics monitoring and snow melt runoff forecasting, glacial inventory & monitoring and climate change impact are some of the key areas requiring critical inputs from satellite remote sensing in the near future

Conclusion
Space borne multi spectral measurements have in some cases replaced ground based observations and in others complemented at varying levels.The various issues related to topographical surveys, irrigation infrastructure, water resource assessment in both rainfed and snowfed baisns, information on command area expansion, planning of new storage reservoirs, stabilizing existing enroute command areas, reservoir sedimentation, glaciers, river conguration, ood forecasting and ood inundation studies, etc. have been suitably addressed through satellite data during the last two and half decades. The water and water related data bases generated using satellite data over the last two decades under various programmes under NNRMS umbrella were used together with large hydromet and environmental data from CWC, for developing a standardized national information system viz.India-WRIS jointly by ISRO and CWC. With the improved eld observational networks and Remote sensing and GIS capabilities including India-WRIS web portal, the focus would be on scientic assessment, development and monitoring the available water resources.

Acknowledgements
Authors place on record various ministries viz. Ministry of Water Resources, Ministry of Environment and Forest, Ministry of Drinking water Sanitation and Ministry of Rural Development for the valuable support in executing several projects mentioned in the manuscript. Authors also acknowledge the support and help provided by CWC, NIH and other organizations in preparing the manuscript. Authors sincerely place on record the contributions of all the concerned scientists across various ISRO centers who have executed several projects which have been mentioned in the manuscript and their contribution is acknowledged.

References
CGWB. (2006). Dynamic groundwater Resources of India, Central Groundwater Board, New Delhi. Jha, B.M., (2009). Management of Groundwater resources for ensuring food security in India, National Groundwater Congress, New Delhi. NRSA, (1998). Project report : Performance Evaluation of 13 irrigation command areas in 5 States, Study commissioned by, Ministry of Water Resources, GOI.

infrastructure using high resolution satellite data in 53 AIBP Projects. NRSC, (2011). Assessment of Water Resources at Basin Scale using Space Inputs (December, 2011): NRSC / RS&GISAA / WRG / WRD / NRSC-CWC Pilot Study / R1 /Dec 2011/ TR 369. NRSC, (2012). Project Report on Groundwater Prospects mapping Under Rajiv Gandhi Drinking Water Mission. Study commissioned by Ministry of Rural Development, GOI. Rakesh Kumar, Singh R.D. and Sharma K.D., (2005). Water Resources of India, Current science,vol.89, No.5

Romani, (2006). Groundwater management- Emerging challenges, Groundwater Governanceownership of Groundwater and its pricing. Proceedings of the 12th National Symposium on Hydrology, November 14-15, New Delhi.

NNRMS

BULLETIN

10 September 2005.

DECEMBER

2013

NRSC, (2010). Project Report on Assessment of irrigation potential through mapping of irrigation

Pal PK, Atul Kumar Varma and Gairola RM Space Applications Centre ISRO, Department of Space, Ahmedabad - 380015, India Email: varma.atul@gmail.com

Introduction
Rainfall is one of the most important atmospheric parameters, which affects the lives and economies of a majority of the earths population. Its amount and distribution during summer and winter monsoon seasons are crucial to sustain the rural economy of the country. While the deciency of rain causes drought and failure of the crop, excessive rainfall is often a curse that causes ood, property and crop damages, loss of human lives and other livestock. Rainfall is an indispensable component of global hydrological cycle and the energy exchange that maintains the planet liveable. Conventionally rainfall over the ground is measured using rain gauges and radars. Rain gauges with inadequate ground network, which provide point measurements at discrete random locations and are ineffective to capture the variability of the precipitation and vary from few meters to several kilometers on spatial scale and few seconds to several days on temporal scale (Piyush et al., 2012). The sparsely available distribution of the rain gauges makes it difcult to measure areal averaged rainfall, which is necessary to study various rain-induced events/processes, like ash ood, dam failure, river catchments, etc. On the other hand, radars are better representative of the areal rain, but their coverage is limited due to their sophistication and high cost. The radar measurements often suffer due to poor calibration of radar reectivity and also of relationship between effective radar reectivity (Ze) and rain rate (Ze-R relationship). The usability of radars is also marred by ground clutter, anomalous propagation, uncertainty in the drop size distribution, melting precipitation, attenuation, etc. Both gauge and radar observations mostly absent over the vast oceans and often fail during severe weather conditions. The most convenient means to measure the precipitation over large area is by using the satellite-based methods that offer frequent uniform coverage over large area. However, the satellite measurements suffer from large errors. The satellite based rainfall estimation is carried out using observations in Visible/IR and microwave spectral bands. While Visible/IR based methods suffer from their inability to sense hydrometeors directly (Barrett and Martin, 1981, Bhandari and Varma, 1995), microwave measurements suffer due to so called beam lling problem, and uncertainty in the rain type, drop-size distribution, drop temperature, fall velocity and shape and orientation of the drops, etc. (Varma et al., 2003, 2004 and Varma and Liu, 2006, 2010, Mishra et al., 2010, Varma and Pal, 2012). The clouds are essentially opaque

NNRMS

BULLETIN

DECEMBER

RAINFALL ESTIMATION USING SATELLITE DATA

2013

at IR frequencies, and the IR measurement of brightness temperature from satellite refers only to Cloud Top Brightness Temperature (CTT). Thus their use for the measurement of precipitation suffers from lack of a direct physical relationship between CTT and rain falling below the cloud base. Inspite of these limitations, it is still possible to develop a statistical relationship between spatially and temporally averaged CTT and precipitation. On the other hand, microwaves directly interact with the hydrometeors and thus offer more direct measurement of precipitation, but they have large footprint (~ 25-50 km at 19 GHz) from low earth orbiting satellites that also restrict their frequency of observation over any given area to 1-2 observations a day. In this paper, an attempt has been made to provide an account of all the rainfall estimation techniques from both IR and microwave observations developed and employed at Space Applications Centre.

IR based rain estimation techniques

In the IR based methods, the CTT is the key parameter to identify and estimate the rainfall. The basic assumption is that a cloud with low CTT is a deep convective cloud and hence it gives higher rain rate. Unfortunately neither all cold clouds rain nor the rain always comes from cold clouds (Barrett and Martin, 1981). In this section an account of the all the major rain measurement techniques employed for rain estimation using IR observations is provided.

Quantitative Precipitation Estimation (QPE) with Arkins Method

This is a very simple technique, yet gives satisfactory results for rainfall measurement over larger spatial and temporal domains. This technique is not meant for ner scale rain estimation. It was initially proposed by Arkin (1979) and Richard and Arkin (1981) when they established a linear relationship between fractional rain-cover over a large area and rain rate. They used data from GARP (Global Atmospheric Research Program) Atlantic Tropical Experiment (GATE) to show that for large spatial scales (e.g., 1.5o X 1.5o or 2.5o X 2.5o) the fraction of an area covered by cold clouds with CTT < 235oK in hourly imagery is highly correlated (>0.7) with area averaged hourly rainfall accumulations in the same area. The temporal averaging for 24 hr. or more improved the correlations to as high as 0.9. According to Arkin and Meisner (1987), the probability of rainfall is a step function of equivalent black body brightness temperature (EBBT) for each pixel and the rain rate is a constant. The probability is 0 for EBBT > 235oK and 1 for lower temperatures. Using 3 hourly images, they found that maximum rain rate for a day is 71.2 mm/day. Arkin et al., (1989) applied their scheme to INSAT-2B observations over Indian region and found that correlation between weekly meteorological sub-divisional rainfall from Arkins method and ground observations varies from 0.33 to 0.91 for different subdivisions. The quantitative precipitation estimation of rainfall using Arkins method is employed for operational rain estimation. Figure 1 shows the rainfall estimation on 7 June 2007 using IR observations from Kalpana-1 VHRR. Mishra et al., (2010) have compared daily averaged rain from Meteosat using Global Precipitation Index (GPI) with surface observations in 0.25oX0.25o over Indian region and reported correlation coefficient of 0.59 and 0.47 and rms difference of 28.67 and 16.53 mm/day for S-W monsoon and N-E monsoon seasons, respectively. A more advanced variant of
Fig. 1: QPE for 7 June 2007 using Arkins method

Quantitative Precipitation Estimation (QPE)

model is also attempted and found to provide improved results.

Hydro-estimator
While QPE technique, holds good for large scale rain measurement for climatic studies, for precipitation measurements at ner scale the indirect relationship between CTT and rain rate is always an inhibiting factor. Also the QPE measured with Arkins method still has the known problem with warm rain and orographic rain. The QPE method described above is also not able to provide corrections due to evaporation of rain drops below the cloud base which is important when air below cloud base is dryer. ancillary knowledge/observations of the cloud environment and provided a scheme for instantaneous rainfall estimation and outlooks for next 30 minutes called IFFA (Interactive Flash Flood Analyzer). The scheme is build upon a hypothetical model of precipitation utilizing skill of a trained meteorologist along with complementary information about environment from Numerical Weather Prediction (NWP) model elds. This scheme was a good success but required experience and skill of a meteorologist who is able to calculate the precipitation estimates. In order to eliminate human intervention in IFFA, Vicente et al. (1998) proposed a new scheme called Auto-Estimator (A-E). Though it simplied IFFA, it often produced false rain from cirrus clouds. It also required radar observations in near real time, which are not readily available in many parts of the world. This has restricted the use of A-E outside North America, and also over the vast oceans. In order to further improve the A-E method, Scoeld et al., (2005) proposed further modications to A-E called Hydro-Estimator (H-E), which is completely automated and eliminates the limitations associated with A-E. In H-E method the cirrus clouds are avoided by considering the pixel under consideration with respect to its neighboring pixels. Rain from convective and non-convective cores is identied and different rain rates (R) versus brightness temperature (Tb) relationships are suggested for them. This allows higher precipitation rates for the convective cores. For convective core, the following relationship between brightness temperature (Tb) in 0K and convective rain (Rc) (mm h-1) is given: In order to overcome these problems, Scoeld (1987) suggested blending of IR measurements with

The coefcients a and b are dynamically calculated for each pixel for given value of Precipitable Water (PW) from National Centre for Environmental Prediction (NCEP/NOAA) model. Also, in H-E method, the maximum possible rain value at any pixel is limited depending upon availability of PW. Thus the maximum precipitation becomes a function of available moisture in the environment. For a non-convective core, the relationship between brightness temperature (Tb) and stratiform rain (Rs) is given as: and In H-E method, the precipitation at a pixel is allowed to be combination of both convective and non-convective core. This is worked out by considering an array of 101 X 101 pixels surrounding the pixel under consideration, and the mean () and standard deviation () of Tb in this box is determined. The and are used to determine standard score (also called Z score) for each pixel, where, 13

NNRMS

BULLETIN

DECEMBER

2013

with inclusion of precipitable water and relative humidity from National Weather Prediction (NWP)

The maximum value of Z score is restricted to 1.5. convective, Otherwise,

If Z < 0; R = 0, i.e., pixel either cirrus or inactive

If Z=1.5, the pixel rain rate R reduces to convective type only. On the other hand, if Z=0, the pixel rain rate R is determined by purely non-convective rain. The above relation provides the rst guess of precipitation amounts, which is further modied for the wetness/dryness of the atmosphere and for warm rain situations. The following corrections are applied in H-E estimations: Correction for Wet/ dry Environment Correction for Equilibrium Level / Warm Top Orography correction:
Simplied Block Diagram of Hydro-Estimator
NWP derived T and RH prole Generating Dew Point Temp Proles NWP derived 850 mb wind Elevation Model NWP derived TPW NWP derived 1000-500 RH

A block diagram of H-E method is shown in Figure 2. Figure 3 shows hourly rain measurement using H-E by SAC and NOAA/ NESDIS. The H-E at NOAA is run on MaCIDAS (Man Computer Interactive Data Assessment System), whereas it has been coded in FORTRAN on Linux / windows. In SAC, Kalpana thermal IR observations are used, whereas NOAA uses Meteosat observations

Atmospheric Thermodynamic Model Equilibrium Level Correction (EL) extrapolated at each Kalpana Pixel Orographic Correction (ORO) interploated at each Kalpana Pixel TPW Correction (TPW) interpolated at each Kalpana Pixel RH Correction (RH) interpolated at each Kalpana Pixel

at TIR1 channel (10.7 m band). Also SAC uses the 6 hourly forecast elds from NOAA Global Forecasting System (GFS), whereas the NOAA uses the derived elds from high resolution North America Model (NAM). Figure 3 shows a good quantitative and qualitative comparison in the rain measured at SAC and NOAA using H-E method on
Next Pixel

Tb corrections for EL, ORO, RH and TPW (Teff)

Kalpana IR Tb No Rain

If Teff<235 K Yes
Tmin, , mean surrounding 101x101 pixels Apply EL correction procedure to Tb Apply ORO correction procedure to Tb Apply TPW correction procedure to Tb

No

13th August 2008. The minor differences in two are possibly due to difference in the input parameters from satellite and NWP model elds as discussed above. Figure 4 shows a more recent rain measurement from H-E on 21st August 2012 at 2200 Z during Jaipur intense rain event. The gure shows a very high rain rate of >100 mm/h at 2200 Z on 21st August 2012, which according to meteorological reports continued till 22nd August and caused ash oods, and casualties of human and other lives.

Get corrected Tadj

Get core rain (Rc) by function t with Rn=12.7mm/h at 240 k and Rc=f(TPW) at 210k Get non-core rain (Rn); Rn = f(TPW, Tadj) and Rn < Rc Rn < 12 mm/h Zi=(mean-Tadj) , for101x101 and 30x30 windows Apply Rh correction procedure Rain

Zi=o
No

Yes

No Rain

Rain

Done

Fig. 2: A schematic diagram of H-E method

INSAT Multispectral Rainfall Algorithm

Apart from H-E, the other IR based algorithm that is presently operational at NOAA/ NESDIS is referred as GOES Multispectral Rainfall Algorithm (GMSRA) which makes use of all the 5 channels of GOES for rain

channels are used for removing the thin cirrus clouds, identifying overshooting tops from anvil cirrus, determining cloud particle size, and cloud growth rate, etc. The GMSRA offers different relationships for determining the rain for different regions. Inspired by GMSRA, Gairola et al., (2010) deloped a scheme called INSAT Multispectral Rainfall Algorithm (IMSRA). Technically the IMSRA is not inclusive of all the features of GMSRA. In IMSRA, a regression based non-linear relationship
Fig. 3: Comparison of H-E hourly rain using Kalpana data and NOAA Meteosat data

between the brightness temperature and the rain rate is established using near concurrent observations from Kalpana and TRMM Precipitation Radar (PR). Prior to rain measurement, a cloud classification scheme is applied which essentially screens a pixel for probable rain using brightness temperature measured at water vapour absorption band (6.7 m) and thermal infrared band (10.5-12.5 m). The IMSRA is used for daily/ 3-hourly averaged rain measurement in 0.25 o x 0.25 o grid. Unlike H-E method, the IMSRA method lacks in its ability to identify orographic/warm rain, for example, those observed in the N-E region and Western Ghats during monsoon season. Figure 5 shows rain measurement on 21st and 22nd August 2012 using IMSRA method. The GPI and IMSRA are operationally implemented at SAC and products are being regularly distributed through MOSDAC (Meteorology and Oceanography Satellite Data Archival Centre) website (www.mosdac. gov.in). The H-E products will also be very soon available through this site. It is also proposed to provide through website the high rain areas with rain rate, affected districts and met-subdivisions, etc. As an example, a typical H-E based such information is provided in

Fig. 4: H-E hourly Rain on 21st August 2012 at 2200 Z during intense Jaipur intense rain event

Fig. 5: Daily rain measurement using IMSRA method on (a) 21 st and (b) 22nd August 2012

Figure 6 for rain occurred during last one

15

NNRMS

BULLETIN

DECEMBER

2013

estimation (Ba and Gruber, 2001). These

hour on 3rd September 2012 at 1400 Z. This information may be useful for assessing of ashood situations for better coordination of relief work. Further developments are underway to provide 3 hour outlook of intense rain using half-hourly H-E rain estimates.

Passive Microwave based Rain Estimation Techniques

In the microwave regime, at low frequencies (e.g., 10 and 19 GHz) the emission from the hydrometeors dominates over the scattering, whereas at higher frequencies (e.g., 85 GHz) the scattering is dominating. Thus at low frequencies the brightness temperature usually increases with rain rate, whereas at higher frequencies it decreases. Both scattering and emission based algorithms are used for rain measurement from passive microwave observations. An account of the rain estimation techniques employed by SAC using measurements from microwave instruments onboard Indian satellite missions is provided herein.

Fig. 6: The location of high rain events during an hour ending at 1500 Z on 3rd Sep 2012, along with tabular report on area of each rain event of > 10 mm/h with highest rain and its geolocation (Table 1), and name of the district along with corresponding metrological subdivision experienced high rain (> 5 mm/h) and the actual rain amount.

Oceansat-1 MSMR Rain Algorithm

The Oceansat-1 launched in May 1999 carried onboard a passive microwave radiometer called Multichannel Scanning Microwave Radiometer (MSMR) and an

Ocean Colour Monitor (OCM). The MSMR provided measurements of brightness temperatures at 6.6, 10, 18 and 21 GHz frequencies in both horizontal and vertical polarisations. The operational geophysical parameters available from MSMR measurements are wind speed, cloud liquid water, water vapour and surface temperatures over the global oceans (Gohil et al., 2000). After the satellite was launched, a detailed campaign for validation of MSMR derived geophysical parameters was carried out using both in-situ and other satellite data (Varma et al., 1999, 2002a). Varma et al., (2000, 2002b, and 2003), developed a non-linear empirical algorithm for rain measurement by regression using concurrent TMI rain rate with MSMR brightness temperatures measured at 10 and 18 GHz V and H polarized channels. Their algorithm explained a multi-correlation of 0.82 and error of estimation of 1.61 mm/h. Figure 7 shows the monthly rain map from MSMR of global oceans. The MSMR could not be used for rain estimation over land as it did not carry high frequency channels for rain estimation over land.

MT-MADRAS Rain Algorithm

A more recent launch of Indo-French Megha-Tropiques (MT) satellite on 12th October, 2011 provided a strong impetus for precipitation measurement over tropical regions. The MT satellite carries a microwave radiometer called MADRAS (Microwave Analysis and Detection of Rain and Atmospheric Systems) that is specically designed to measure

other geophysical parameters over tropical oceans (Varma et al., 2011). In addition to MADRAS, MT mission also carries a number of other instruments referred as radiation budget instrument (SCAnner for RAdiative Budget: SCRAB), an atmospheric sounding instrument Sounder for Probing Vertical Proles of Humidity (SAPHIR), and a GPS receiver for occultation based prole measurements. The MT MADRAS measures the Earths radiation at 18.7, 23.8, 36.5, 89 and 157 GHz V and H polarized frequencies (except 23.8 GHz which is received with V polarization only) over the entire global
Fig. 7: Monthly Rain from (a) MSMR and (b) SSM/I for the month of August 2000

tropics (within ~ 28o latitudes). Gohil et al., (2013) described the

operational rain algorithm for MSMR which is based on Mishra et al. (2009). Following Grody (1991) and Ferraro and Mark (1995), Mishra et al., (2009) used scattering index for rain measurement. The Scattering Index (SI) is dened as depression in brightness temperature at 85 GHz SSM/I channel due to rain. The SI is calculated by taking difference of the observed brightness temperature from the expected brightness temperature during rain free conditions. The expected brightness temperature at 85 GHz during rain free conditions is obtained by regression involving non-scattering (low-frequency) channels (19 and 21 GHz). Mishra et al., (2009) proposed new set of coefcients for the Indian region for calculating expected 85 GHz brightness temperature for SI. They further correlated SSM/I brightness temperatures with surface rain observations from TRMM PR, and established relationship between SI and rain rate for Indian region. A radiative transfer approach is adopted by Varma et al., (2011), in which an attempt was made to radiometrically calibrate MT-MADRAS channels with corresponding SSM/I channels using radiative transfer model simulations. After calibrating MT-MADRAS to SSM/I channels, they used the rain retrieval scheme of Ferraro et al., (1996). The steps involved in their scheme are as follows; rst establishment of radiative transfer (RT) simulation based relationship between SI for SSM/I 85 GHz channel (SI-SSMI) with SI
Fig. 8: A schematic diagram of rain estimation from MT-MADRAS as proposed by Varma et al. (2011)

for MADRAS 89 GHz channel (SI-MADRAS). A very high correlation is found between SI-SSMI and SI-MADRAS. In the second

step, SI-SSMI and SI proposed by Grody (1991) are calculated using actual SSM/I observations and a relationship is established between them which accounted for any difference in the development

17

NNRMS

BULLETIN

DECEMBER

2013

rainfall over the land and oceans, along with

procedure of the two SI models (e.g., difference in the RT model, etc.). Further procedure for rain estimation as developed by Ferraro and Mark (1995) are applied. Over the land area the identication of deserts and arid land as proposed by Ferraro and Mark (1995), required calibration of MT-MADRAS 18.7 V, 18.7 H and 89V channels to corresponding SSM/I channels, which is carried out using RT simulated measurements for SSM/I and MT channels. Over the oceans, the Cloud Liquid Water (CLW) along with SI is used by Ferraro and Mark (1995) for identifying the rain. CLW from operational MT products (Varma et al., 2011) was considered. The scheme developed by Varma et al. (2011) for rainfall measurement is shown in Figure 8.

Active Microwave based Rain Estimation Techniques

The Precipitation Radar onboard TRMM is known to provide most accurate measurement of the precipitation. The PR rain is available as a nished product in TRMM operational products, e.g., 2A23 and 2A25. PR observations for development as well as validation of various rain algorithms have been extensively used. An account of the rain algorithms that have been developed in the past using active microwave instrument, the scatterometer and the altimeter are provided herein.

Rain from Oceansat-2 Scatterometer

Scatterometer is essentially a Radar that measures the radar cross-section at larger incidence angles. The dominant mechanism for radar return in scatterometer is Bragg scattering from the wind generated capillary waves on the ocean surface. In September 2009, India launched a Ku-band pencil beam
Fig. 9: 3-day averaged (10-12 Dec. 2012) rainfall from (a) scatterometer (b) AMSRE

radar scatterometer onboard Oceansat-II satellite for global wind vector measurements.

The Oceansat-II scatterometer has an antenna with dual feed assembly generating two rotating beams at two different incidence angles and polarizations. The Oceansat-II scatterometer is operating at 13.515 GHz frequency in VV and HH polarizations at incidence angles of 48.9o and 57.6o while sweeping a swath of 1400 km and 1840 km by its inner and outer beams, respectively. The scatterometer instrument also provides the apparent brightness temperature measurements through measurement of instrument noise using an appropriate model. Ghosh et al. (2012) developed a technique to detect and estimate precipitation over the global oceans using the radar back scattering coefcient and brightness temperature measurements from Oceansat-II scatterometer along with rain sensitive parameters from NWP model through a neural network (NN) based setup. Rain / No-Rain label was produced by using concurrent Tropical Rainfall Measuring Mission (TRMM) and Advanced Microwave Scanning Radiometer for Earth Observation Satellite (AMSR-E) rain measurements. NN was applied in two stages: (1) rain identication and (2) rain measurement with training samples from ve different geographical regions over 5 different latitudinal bands. Rain identication accuracy of about 93%, 87.2%, 90.1%, 78.9% and 85.5% and no-rain detection accuracy of about 96.9%, 87.4%, 87.5%, 84.1% and 85.5% for these 5 regions was reported. It was found that the missing rain cases are few compared to the size of no-rain samples and are largely from the low rain regime. RMS error of rain estimation for regions I to V (rain rates varying from > 0 to approximately 45, 25, 25, 45, and 20 mm/h) to 1.86, 0.69, 0.47, 0.56, 0.46 mm/h, respectively was reported. The qualitative comparisons of rain rates from scatterometer and AMSR-E demonstrate a good agreement between them (Figure 9).

Rain from Radar Altimeter

TOPEX an ocean topography mission launched in 1992 carried onboard a dual frequency Radar Altimeter (RA) working in C and Ku band. The primary objective of the radar altimeter was to provide ocean surface topography.

for estimating Total Electron Content (TEC) and thereby estimating the ionospheric delay in the altimetric measurements for better accuracy of the sea surface height. The TOPEX satellite also carried a 3 frequency (18, 21 and 37 GHz) nadir looking TOPEX Microwave Radiometer (TMR) for wet
Fig. 10: Rain for the month of July 2000 from TOPEX -TMR and RA

tropospheric correction to sea surface height measurements. Using differential attenuation

BULLETIN

at C and Ku band, Bhandari and Varma (1996) demonstrated the use of altimeter for rain measurement. The study was further extended by combining both TMR and RA for rain measurement (Gairola et al., 2005). Figure 10 shows a monthly rain map over ocean for July 2000 using TMR and RA combined algorithm. The study is useful in view of forthcoming dual frequency AltiKa mission.

Conclusion
A number of rain estimation schemes developed at SAC/ISRO using thermal IR, passive and active microwave observations are presented in this paper. These schemes are presently employed with IR observations from Kalpana satellite, but will also be employed with future INSAT-3D mission. All schemes have their own advantages and disadvantages. For example, while IR based QPE technique is simple to implement, which, provides good estimation at climatic scales. It is not suitable for providing high rain rates that occur over small area for short time. On the other hand, the H-E is too complex, but is worth pursuing as that can provide measurement of even very intense rain at pixel-scale and at the frequency of satellite-image-acquisition. The passive microwave observations are more direct in nature but suffer due to their coarser time sampling and spatial scale. Both IR and microwave have poorer estimation over land where measurements are most desirable for applications like crop monitoring, ash ood, surface hydrology, etc. The most challenging task over land is to account for warm rain and the rain falling over hilly terrain. Perhaps, the H-E method is the only satellite based rain measurement scheme which is designed to take care of warm rain and rain over topography by blending the NWP model elds and earth elevation model. Keeping in view the advantages and disadvantages of IR and passive microwave measurements, it is thus desirable to develop a comprehensive rain algorithm that takes into account advantages of IR (e.g., its spatial resolution and high frequency of observations), passive microwave (e.g., its direct interaction with hydrometeors) and the ancillary datasets from other sources (e.g., NWP elds and earth elevation model) to optimally generate the estimation. A number of algorithms that blend IR and microwave observations are already developed and made operational by many other agencies (e.g., TRMM 2B42 by NASA; GsMAP by JAXA). It is found that the existing IR-Microwave blended algorithms do not consider any of the NWP model elds into the algorithm. With the experience gained towards blending IR measurements with such observations in H-E method, an algorithm that combines IR, Microwave and model elds for the best rain estimation under an approved NASA-GPM (Global Precipitation Mission) project is being pursued at present.. The rain algorithms presented herein for scatterometer can also be combined with other measurements in the comprehensive rain algorithm as discussed above. Like passive microwave radiometers (e.g., SSM/I, MADRAS, etc.), the wind scatterometer has a large swath that provides a very good spatial

19

NNRMS

DECEMBER

2013

The two frequencies were used to account

coverage. However, the operational continuity of the scatterometer similar to one onboard Oceansat-2 has to be ensured for making it meaningful for blending it in any comprehensive rain algorithm. The presently available scatterometer based rain is useful for providing better sampling of rain for climatic studies and for agging the scatterometer wind measurements. Unlike scatterometer with very large swath, the Radar Altimeter provides rain only at the nadir which may be of limited use. However, the altimeter rain is essential for possible correction and/or agging off the rain affected sea surface height measurements. Recently, with the launch of Indo-French AltiKa mission with a dual frequency altimeter operating at Ku and Ka band rain measurements with better expected accuracy is being attempted. The most advanced sensors for rain measurement till date is ying on TRMM satellite since 1997. The TRMM carries onboard a passive microwave radiometer (TMI), a precipitation radar (PR), and other instruments. The TRMM is still providing useful observations to global scientic community. The continuity of TRMM has been ensured through more advanced Global Precipitation Mission (GPM) with proposed launch in 2013. The GPM is a constellation with a core and several candidate satellites with each of them to carry a passive microwave radiometer to ensure passive microwave observation of most parts of the globe in < 3 hour. The core satellite will be carrying a very advanced radiometer called GPM Microwave Imager (GMI) with frequencies ranging from 10 to 157 GHz in both V and H polarizations and a twofrequency (Ku and Ka band) precipitation radar. The dual frequency radar will help retrieving the vertical rain prole with differential attenuation. The availability of GPM and the INSAT-3D simultaneously will open up possibility of providing more accurate rain estimations.

References
Arkin, P.A., (1979). The relationship between fractional coverage of high cloud and rainfall accumulation during GATE over the B- scale array, Monthly Weather Review, 107, 1382 1387 Arkin, P. A., and Meisner, B.N., (1987). The relationship between large-scale convective rainfall and cold cloud over the western hemisphere during 1982-84, Monthly Weather Review, 115, 51-74. Arkin, P.A., Krishna Rao, A. V. R., and Kelker, R. R., (1989). Large Scale Precipitation and Outgoing Longwave Radiation from INSAT-1B during the 1986 Southwest Monsoon Season, Journal of Climate, 2, 619-618. Ba, M. B., and Gruber, A., (2001). GOES multispectral rainfall algorithm (GMSRA). J. Appl. Meteor., 40, 15001514. Barrett, E. C. and Martin, D. W., (1981). The use of satellite data in rainfall monitoring, Academic Press, London. Bhandari, S.M and Varma, A.K. , (1995). On Estimation of Large Scale Monthly Rainfall Estimation Over the Indian Region Using Minimal INSAT-VHRR Data, International Journal of Remote Sensing, 16, 2023-2030. Bhandari, S.M and Varma, A.K., (1996). Potential of Simultaneous Dual-Frequency Radar Altimeter Measurements from TOPEX/Poseidon for Rainfall Estimation Over Ocean, Remote Sensing of Environment, 58 (1), 13-20. Ferraro R. R, Weng F., Grody, N.C., and Basist, A., (1996). An eight-year (1987-1994) time series of rainfall, clouds, water vapour, snow cover and sea ice derived from SSM/I measurements, Bulletin of American Meteorological Society, 77 (5), 891-905. Ferraro R .R, and Mark, G. F., (1995). The development of SSM/I rain rate algorithms using ground-based radar measurements, Journal of Atmospheric and Oceanic Technology, 12, 755-770. Gairola, R.M., Pokhrel, S., Varma, A.K., and Agarwal, V.K., (2005). A combine passive-active microwave retrieval of quantitative rainfall from TOPEX/Poseidon altimeter and TMR, International Journal of Remote Sensing, 26 (8), 1729-1753. Gairola R.M, Mishra A., Prakash S., and Mahesh C., (2010). Development of INSAT multi-spectral rainfall algorithm (SRA) for monitoring rainfall events over India using alpana-IR and TRMM-precipitation radar observations, SAC/ISRO Report, SAC/EPSA/AOSG/INSAT/SR-39/2010.

measurement using Oceansat-II Scatterometer observations, Remote Sensing of Environment, under revision. Gohil, B S, Gairola, R.M., Mathur, A.K., Varma, A.K., Mahesh, C., Gangwar R.K., and Pal, P.K., (2013). Algorithms for retrieval of Geophysical Parameters from MADRAS and SAPHIR sensors of Megha-Tropiques Satellite Indian Scenario, Quarterly Journal of Royal Meteorological Society, 139: 954963. Gohil, B.S., Mathur A.K., and Varma, A.K., (2000). Geophysical Parameters Retrieval over Oceans from NIO, Goa, 207-211. Grody, N.C., (1991). Classication of snow cover and precipitation using the special sensor microwave imager (SSM/I), J. Geophys. Res., 96, 7423-7435. Mishra, A, Gairola, R.M., Varma, A K., Sarkar, A., and Agarwal, V.K., (2009). Rainfall retrieval over Indian land and oceanic regions from SSM/I microwave Data, Advances in Space Research, 44, 815823, doi: 10.1016/j.asr.2009.05.010.01. Mishra, A., Gairola, R.M., Varma, A.K., and Agarwal, V.K., (2010). Remote Sensing of precipitation over Indian Land and Oceanic regions by synergistic use of multi satellite sensors, Journal of Geophysical Research, 115, D06106, doi:10.1029/2009JD012157. Piyush, D N, Varma, A.K., Pal, P.K., and Liu, G., (2012). An Analysis of Rainfall Measurements over Different Spatio-Temporal Scales and Potential Implications for Uncertainty in Satellite Data Validation, Journal of Meteorological Society of Japan, 90 (4), 439 - 448, DOI: 10.2151/JMSJ.2012-408. Richard, F. and Arkin, P.A., (1981). On the relationship between satellite observed cloud cover and precipitation, Monthly Weather Review, 109, 1081 1093. Scoeld, R. A., (1987). The NESDIS Operational Convective Precipitation Estimation Technique, Mon. Wea Rev., 115, 1773-1792. Scoeld, R.A., Kuligowskiand R.J., Davenport, J.C., (2005). The satellite-derived Hydro-Estimator and Hydro-Nowcaster for Mesoscale Convective Systems and Landfalling Tropical Systems, in Applications with Weather Satellites II, edited by W. Paul Menzel, Toshiki Iwasaki, Proceedings of SPIE Vol. 5658 (SPIE, Bellingham, WA, 2005) 0277-786X/05/$15 doi: 10.1117/12.577850 Varma, A.K., and Pal, P.K., (2012). Use of TRMM Precipitation Radar to address the problem of rain detection in Passive Microwave Measurements, Indian Journal of Radio and Space Physics, 41 411-420. Varma, A K, Gohil, B.S., Prashant Kumar and Pal, P.K., (2011). Retrieval of Ocean Surface Wind Speed, Total Precipitable Water and Cloud Liquid Water from Megha-Tropiques MADRAS: Initial Brightness Temperature Observations and Retrievals, internal report no.: SAC/EPSA/AOSG/MT/SR/75/2011, p 25. Varma, A.K, Gohil, B.S., Gairola R.M., and Pal, P.K., (2011). A new radiative transfer based rain retrieval algorithm for rain identication and measurement from MT-MADRAS: Initial Results, internal report no.: SAC/EPSA/AOSG/MT/SR/76/2011, p 26. Varma, A.K., and Liu, G., (2010). On Classifying Rain Types Using Satellite Microwave Observations, Journal of Geophysical Research, 115, D07204, doi:10.1029/2009JD012058.

21

NNRMS

BULLETIN

IRS-P4 MSMR, Proc. of Pacic Ocean Remote Sensing Conference (PORSEC-2000), 5-8 Dec, 2000 at

DECEMBER

2013

Ghosh, A., Varma, A.K., Shivani Shah, Gohil, B.S., and Pal, P.K., (2012). Rain identication and

Varma, A.K., and Liu, G., (2006). Small-Scale Horizontal Rainrate Variability by Satellite, Monthly Weather Review, 134 (10), 2722-2733. Varma, A.K., Liu, G., and Noh, Y. J., (2004). Sub-Pixel Scale Variability of Rainfall and Its Application to Mitigate the Beam-Filling Problem, Journal of Geophysical Research, 109, D18210, doi:10.1029/2004JD004968,. Varma, A.K., Gairola, R.M., Kishtawal, C.M., Pandey, P.C., and Singh, K.P., (1999). Rain Rate Estimation from Nadir Looking Microwave Radiometer (TMR) for Correction of Radar Altimetric Measurements, IEEE Transactions on Geosciences and Remote Sensing, 35 (5), 2556-2568. Varma, A.K., Gairola, R.M., Mathur, A.K., Gohil B.S., and Agarwal, V.K., (2002a). Intercomparison of IRS-P4-MSMR derived geophysical products with DMSP-SSM/I and TRMM-TMI nished products, Proceedings of Indian Academy of Sciences Earth and Planetary Sciences, 111 (3), 247-25,. Varma, A.K., Gairola, R.M., Samir Pokhrel, Mathur, A.K., Gohil B.S., and Agarwal, V.K., (2002b). Rain Rate Measurements over global oceans from IRS-P4 MSMR, Proceedings of Indian Academy of Sciences Earth and Planetary Sciences, 111 (3), 257-266. Varma, A.K., Samir Pokhrel, Rakesh Mohan Gairola, and Vijay K. Agarwal, (2003). An Empirical Algorithm for Cloud Liquid Water From MSMR and Its Utilization in Rain Identication, IEEE Transactions on Geosciences and Remote Sensing, 41 (8), 1853-1858. Varma, A.K., Pokhrel, S., Gairola, R.M., and Agarwal, V.K., (2000). Rain Rate Measurements from IRS-P4 MSMR, Proc. of 5th Pacic Ocean Remote Sensing Conference (PORSEC-2000), 5-8 Dec, 2000 at NIO, Goa, India, 240-243. Vicente, G. A., Scoeld, R. A., and Menzel W. P., (1998). The operational GOES infrared rainfall estimation technique. Bulletin of the American Meteorological Society, 79, 18831898.

Introduction and Objective

Surface water bodies are essential water storage units and play an important role in efcient tapping of huge water quantities from rainfall, runoff events. For irrigated agriculture in India, water is drawn from surface water bodies such as major/medium reservoirs, all irrigation tanks, river reaches where lifting of water is possible, etc. Huge infrastructure / establishments were created in India for creation of irrigation facilities. Irrigation potential was created about ~102.77 Mha out of ultimate irrigation potential of ~140 Mha [Annual Report MoWR, 2011]. There is continuous development and increase in number of large dams from 3600 in 1999 to 4710 in 2009; 382 large dams are also under construction [Dam Registry, CWC, MoWR]. In addition, there are huge number of smaller irrigation tanks / water bodies, and also many are under construction apart from the above number of major reservoirs in the country. Mapping and monitoring of surface water bodies is necessary as they are dynamic in nature in terms of water spread area and volume of water. The dynamic nature results into inter/intra annual/ seasonal/ monthly variations in surface water spread and there by the availability of water for agricultural use varies (Roberts et al., 1993, Voeroesmarty et al., 1997, Bastiaanssen et al., 2000 and Eric et al., 2003). The utilisation of satellite images for the inventory, mapping and monitoring of surface water bodies would lead to the assessment of inter and intra seasonal as well as annual surface water spread dynamics vis--vis progress of cropping area and irrigated agriculture and water utilisation patterns. This article describes the need for spatio-temporal information on surface water bodies, brief overview of the existing scenario of popular satellite data interpretation / image processing / classication / methods for extraction of surface water bodies, need for development of automated image processing methods for extraction of surface water bodies from satellite images etc. Possibilities for the use of automated methods for quick processing of satellite images and, further opportunities for surface water area dissemination in near real time through web services is discussed along with demonstrated results. Details of automated algorithm for extraction of water bodies from optical and microwave data are explained and demonstrated, combined use of these datasets in a crop year to capture the dynamic range of surface water spreads. Results from the demonstrative studies are presented to represent the national scenario of surface water spread dynamics. The water body information generated from these efforts would support the studies on spatio-temporal surface water spread dynamics, water resources planning and management, etc.

NNRMS

BULLETIN

Suresh Babu AV, Shanker M and Venkateshwar Rao V Water Resource Group, National Remote Sensing Centre ISRO, Department of Space, Hyderabad - 500 037, India Email: sureshbabu_av@nrsc.gov.in

DECEMBER

INVENTORY, MAPPING AND MONITORING OF SURFACE WATERBODIES

2013

Inventory, Mapping and Monitoring of Surface Water Bodies from Satellite Images Present Scenario
The availability of reliable and higher frequency of earth observation capabilities from various satellite missions has increased the scope for regular database generation and analysis on surface water area dynamics. There are several techniques used for delineation of surface water pixels / surface water area from satellite images using both optical and microwave sensors. This paper describes the overview of current scenario and typical case studies. Further, summary on efforts made by NRSC / ISRO on development and utilization of quick image processing techniques and algorithms for delineation of surface water bodies is also brought out.

Optical Satellite Images

There are traditional image classication algorithms for detection of water pixels, namely supervised classication, unsupervised classication, applying band value thresholds, generation of popular spectral indices like Normalized Difference Water Index (NDWI = Green-NIR / Green + NIR), Modied Normalized Difference Water Index (MNDWI = Green-SWIR /Green + SWIR) etc. Several case studies are mentioned here. ISRO has mapped wet lands at national level in India and atlas was released under Natural Resources Census (NRC) programme. Lenher, et .al, 2004 have created National Global lakes and wetlands database at levels viz. large lakes, reservoirs, smaller water bodies, and wetlands. Li et al., 2005 have published the maps of Canadas wetlands using optical, radar and DEM data. Ma et al., 2011 published a dataset of China lakes, which were constructed using 11004 satellite images from CBRES CCD camera data and Landsat TM/ETM sensors. There are several case studies in literature who have used multi temporal satellite images for delineating the water spread area of several reservoirs at high frequency, and estimated the live storage capacity useful for reservoir planning, operations, and sedimentation assessment (Frazer et al., 2000, Suresh Babu et.al., 2003, Manavalan et.al., 1993, Hui, et al., 2008). Normalized Difference Water Index (NDWI) and Modied Normalized Difference Water Index (MNDWI) were used by McFeeters et al., 1996 and Hanqiu, et al., 2006. Subramaniam et al., 2011 have developed automated extraction algorithm for the delineation of surface water bodies using IRS Resourcesat AWiFS, LISS III and implemented on national datasets.

Microwave Satellite Images

There were several case studies on use of microwave remote sensing datasets for delineation of surface water bodies. The application potentials of microwave datasets for several applications including hydrological perspective was described by Van der Sanden 2004, Sokol et al., 2004 and Brisco et al., 2008. There are several mapping techniques being used by researchers considering various parameter viz. polarization ratios (Brisco et al., 2011, Schroeder et al., 2010), concurrent use of optical and radar images (Laura Brown et al., 2006), ISODATA algorithm (Maria et al., 2002), adaptive thresholding (Liu et al., 2004) and active contour model (Horritt et al., 2002, Thomas et al., 2010) for delineation of surface water bodies from the microwave images. Rajiv Kumar Nath et al., 2010 have also described several techniques for delineation of surface water from satellite data. There are several applications where in multi-polarization radar images were utilized for ood inundation mapping and monitoring (Van der Sanden 2004, Hu et al., 2007 and Panchagnula et al., 2012). Several wetland mapping applications were also conducted using optical and radar remotely sensed data (Junhua et al., 2005 and Jean-Robert et al., 2010). The inuence of incidence angle on backscatter varies according to ground layer characteristics, including surface roughness, presence/absence of standing water, soil moisture, and the forest structure (Megan et al., 2008). The sigma naught being chiey a function of incidence angle and varying by about 5 dB for the smooth targets such as bare soil, surface water (Ross et al., 1998) and 10dB for other landuse features as the incidence angle varies between 200 and 500 (Horritt et al., 2002). Separability between different classes was performed by the visual investigation of, incident angle based HH and HV scatter plots (Brisco et al., 2011) to nalize the thresholds.

hydrological applications. There is a need for automation of such data processing techniques. NRSC has made initiation in this direction which is briefed in this paper.

Information Requirements vis--vis Image Processing for Delineation of Surface Water Bodies
Satellite data processing for extraction of surface water bodies is required for assessment of water resources at regional level in different seasons. India is a large country and hence, huge number of satellite datasets are to be processed. Development of techniques for quick processing of these datasets and in specic automated algorithms will help in generation of near real time information and would enable web based information service. Most of the contemporary image processing softwares have modules for image processing. Techniques involved in these softwares are extraction of inherent spectral pattern in satellite data and categorization of pixels into various classes. Interpretation and selection of these classes representing water surface are manually carried and grouped to derive water spread. For example, in supervised classication, spectral signature sets are generated from water bodies and is used as basis for the classication which may not accommodate all types of water bodies that might exist in other images across the region. Similarly, in unsupervised classication, number of classes are dened and classied. The corresponding classes pertaining to water pixels have to be determined by visual interpretation techniques by the analyst to arrive at water layer. In band threshold method, digital number (DN) ranges for a water pixel in different bands are dened and are used for extraction of water bodies. These methods are however, scene specic, and may not be applicable to other areas. Assigning of these parameters or threshold limits are highly subjective in nature and vary from image to image and across the seasons / years and also in different geographical settings of India. This has lead to requirement of manual intervention, processing time and commitment of errors in surface water area delineation. Popular indices such as NDWI (Normalized Difference Water Index), MNDWI (Modied Normalized Difference Water Index) are being used in general which uses spectral information available in any of two spectral bands such as Green, SWIR or Green, NIR and range for water pixel varies from one water body to other and also from one image to another image. In this method, spectral information available in other bands of a sensor are not fully utilized. Concept of automatic extraction of water features from Resourcesat-1 AWiFS, LISS was developed using spectrally knowledge based and further implementation on series of seasonal national databases. The emphasis was made here for explaining the concept of automated delineation of surface water bodies using an hierarchical multi-logic execution using all four spectral bands available in AWiFS / LISS III.

Automated Techniques for Delineation of Surface Water Bodies

The increasing number of Satellites / Sensors available at present, provides data at higher spatio-temporal resolutions and better repetivity. Improved computational infrastructure for handling huge satellite data sets is enabling the development of quick image processing techniques facilitating analysis of multi-date satellite data in near real time to derive dynamic thematic information. Surface water spread is highly dynamic in nature and varies with rainfall pattern, storage capacity in reservoirs, tanks, etc. of a region and water utilisation patterns. Hence, inventory, mapping and monitoring 25

NNRMS

BULLETIN

DECEMBER

2013

RISAT-1 launched on April 26, 2012 is providing microwave datasets on regular basis for operational

is necessary for the assessment of water resources planning and management with respect to supply-demand. The availability of medium spatial resolution data from Advanced Wide Field Sensor (AWiFS) onboard Resourcesat-1/ Resourcesat-2 (56m spatial resolution, 5 days repetivity,740km swath) enabled the prospects of near-real time monitoring of water bodies. In addition, all weather capable RISAT-1 MRS/CRS data is also helping in capturing of water bodies information in cloud condition and has also increased the frequency of observation. Logical steps followed for the development of automatic algorithm for quick processing of satellite images for extraction of water surface are briey described in this section. Various reservoirs, tanks are depicted through satellite images (Figure.1a, Figure.1b)
Fig. 1a: Satellite image showing water bodies (Resourcesat AWiFS -Optical data)

from Resourcesat MRS data .

AWiFS and RISAT-1

Automated Extraction of Water Bodies from Optical Satellite Images

AWiFS/LISS III sensor onboard Resourcesat missions provide data in four spectral bands Green (0.52-0.59 m), Red (0.62-0.68 m), NIR (0.77-0.86 m), SWIR (1.5-1.70 m). Identification of distinct spectral reectance characteristics of water surface in these four bands has led to development of the automatic extraction algorithm for extraction of water bodies. A new knowledge based algorithm was developed (Subramaniam, et al., 2011) using multi-temporal spectral information available in four bands of AWiFS / LISS III. Knowledge base was created from collection and analysis of visually known water pixels (extracted from various water bodies) from multiple satellite data sets of different time periods and regions. Spectral
Fig. 1b: Satellite image showing water bodies (Risat-1 MRS-Microwave data)

characteristics of water in the visible-SWIR

Visible-SWIR spectral region compared to other surface cover features due to the absorption of NIR and SWIR radiation (ii) high spectral contrast between green and NIR /SWIR. Distinct spectral behaviour of water pixels are identied and a knowledge based hierarchical algorithm was developed indices for quick processing of satellite data and extraction of water bodies. Hierarchical, logical steps were developed for building algorithm and utilise all the information available in all four bands of the sensor. Quantitative evaluation of water spread obtained using the algorithm and by visual analysis was carried out with respect
Fig. 2: Water spread derived from Resourcesat AWiFS Images

to size and shape by comparing the results.

Visual interpretation was chosen as it is well established technique to capture the landwater boundary, though one or two pixels along the periphery could be ambiguous. The accuracy of delineation of water bodies using automated algorithm vis--vis visual interpretation has been found to be in the range of 92 - 97% for large water bodies and varied for the other water bodies depending on the size, area, area /perimeter. Figure.2 shows satellite image and corresponding water body information derived from automatic extraction algorithm.

Automated Extraction of Water Bodies from Microwave Satellite Images

Microwave satellite images are useful to capture the information during cloud cover. An experimental study was conducted using RISAT -1 MRS data (dual polarization-HH, HV, 18 m spatial resolution) and developed an automated algorithm for classication of water bodies with the development of knowledge base on sigma naught of HH, HV values. The methodology consists of characterization of microwave response to various types of water features, observations on behaviour of back scatter coefcient (o) from these features, generation of spectral plots, identication of sigma naught thresholds, identifying the back scatter coefcient thresholds which are distinct to water surface. The accuracy was improved by referring to legacy water bodies mask. Initial
Fig. 3: Water spread derived from RISAT-1 MRS Images

results are shown in Figure 3.

27

NNRMS

BULLETIN

with band thresholds, band ratios, spectral

DECEMBER

2013

spectral bands are (i) lower reectance in

Monthly Monitoring of Surface Water Bodies

Water spread is dynamic in nature and requires continuous monitoring, but there are limitations due to cloud cover. Experimental studies indicated that the cloud free data is available during October month only, where as, capturing the same in Kharif season (Jun, Jul, Aug, Sep) months is difcult with optical data and hence there is a need to make use microwave data to ll the gap. In view of this, the automated approach presented in above sections would provide scope for capturing the dynamics of water spread in India at monthly interval.

Operational Utilization of Automated Extraction Algorithm for Surface Water Bodies

Results from operational Implementation of automated extraction algorithm revealed that the quick processing of optical satellite data is feasible. Algorithm was implemented on multiple datasets in different time scales. Resourcesat -2 AWiFS derived Water Body Area (WBA) in different months (Sep, 2011-May 2012) is shown in Figure 4.

Utilization of Satellite Derived Surface Water Bodies Information

Huge number of satellite data sets were used for extraction of dynamic water body area information in addition to inventory of surface water bodies. The focus is being made for specic utilization of such
Fig. 4: Monthly surface water bodies area using Resourcesat-2 AWiFS Sensor data

dynamic surface water bodies. The immediate applications are : Monitoring of surface water

bodies at regional level at dened interval (fortnight/month) and analysis of the same variations and development of hydrological drought indices, input to hydrological modeling, climate models as surface water area is one of critical parameter in land surface process modeling, representation of grid wise surface water area fraction, capacity estimation of major & medium reservoirs, etc.

Utilisation at National Scale

A scenario of water body area dynamics is described here along with the satellite derived information. Resourcesat AWiFS data sets were used for the implementation of automated water spread extraction algorithm for the months of Oct, Feb, May of 2004-2011 for the demonstration of the concept and its operational utilization. Analysis was carried out in two categories namely, Water Bodies Area (WBA) and WBA-Res (only reservoirs, tanks, lakes, ponds excluding rivers/streams). The dynamic behaviour of WBA was estimated at national scale and statistics revealed that WBA was ranging from 2.386 M.ha (May) to 5.107 M.ha (October) during 2004-2011. The dynamic behaviour of WBA was quantied

was ranging from 0.72%-1.55% of total geographical area during 2004-2011 in India. WBA was estimated maximum in October, reduced by 4-17% in February, 30-46% in May analyzed over 7 years period. Overall scenario indicates that WBA
Fig. 5: Variations in surface water spread during 2004-2011 over India

is around 35%, 15%, 50% from larger water bodies (>500 ha), small water bodies, rivers / streams respectively. The rainfall was higher during Jun-Sep in all the years which is 70-87% of annual RF (890 mm-1212 mm) and the rest in Oct-Dec, and rainfall was negligible during Jan-May months and WBA variations were found to be in correspondence with rainfall variations. The results shown in Figure 5 indicate that, the automated methods for consistent information generation on surface water bodies can be operationally utilized. Though, the study was done for three months representing 3 seasons namely Kharif, Rabi and summer, the same can be extended at higher frequency with the combination of microwave and optical datasets.
BULLETIN

140000
126336 115971 112779 105272

WSA: Total Waterbodies WSA: Excluding new Waterbodies after 2006 WSA: New Waterbodies after 2006
109302 105829 121801

120000

Development of Water Spread Indicators Scale

Hydrological drought is expressed in terms of availability of water - change / deviations in discharges which are ultimately
Oct 2006
Reference year

100000 Water spread area (H a)

84341 80232

Representing

80000
66749 61433

Hydrological Drought at Regional

60000

40000
15557 10699 5316 (28) (21) 4109 (15) 3473 (10)

20000

(37)

0 Oct 2011 Oct 2010 Oct 2009 Oct 2008 Oct 2007

linked to rainfall intensity and amounts. However, the impact can be observed from surface water spread in various water bodies. The concept demonstrated here is the use of satellite derived surface water spread information to understand the deviations

Note : Number of new water bodies formed after 2006 are Indicated within brackets

Fig. 6: Spatial variations in satellite derived surface water bodies current & historic data

from normal water spread area of water bodies and also the visibility of new water bodies in the region. An example shown in Figure 6 indicates the regional analysis of surface water spread over Vidarbha region of Maharashtra State. 29

NNRMS

DECEMBER

2013

at national level and revealed that WBA

Inputs for Land Surface Process Modeling

Water Body Fraction (WBF) was estimated as fraction of Water Body Area (WBA) and Grid Area (3 x 3 grid) over India. An example is shown in Figure 7. WBF derived over India during 2004-2011 indicated that majority of grids are having <5% of WBF and signicant number grids are having WBF in the range of 5-30%. Grid (3x3) wise surface water area fraction was generated, which is useful as an input to climate and weather studies.

Capacity Estimation of Major Reservoirs from Satellite Data

Monitoring the changes in capacities of several reservoirs in cropping seasons is essential in India as we have huge number of reservoirs and irrigated agriculture. This will help us to estimate reservoir capacities from time to time for better water resources planning and management. Utilisation of satellite data for extraction of water spread area of reservoir(s) at frequent intervals and
Fig. 7: Example showing the satellite derived Water Body Fraction (WBF) at 3 x 3 grid interval

linking the water spread area to the AreaCapacity curves of the concerned reservoir for

capacity estimation in near real time through automation would provide the scope for estimation of reservoir capacities as soon as satellite data is acquired. In this direction, an integrated set of software tools were developed for extraction of water layer from satellite data, tagging the identication to cluster of reservoirs available in satellite image, linking lookup table for capacity estimation (pre dened area capacity curve from the eld data). The development of these tools enabled the use of Resourcesat AWiFS / LISS III effectively . Operational utilisation of such tools is in progress. The concept will also be useful for effective utilization of RISAT-1 CRS/MRS dual polarization data as we can overcome cloud cover problem in monsoon season when there are more dynamic changes in water spread due to rainfall and subsequent water withdrawals. Figure 8, depicts the demonstration of the concept. Blue line indicates the capacity of reservoir based eld based observations. The magenta dots on the blue curve indicates the satellite derived water spread based capacity
Fig. 8: Water Spread Area, Capacity of a reservoir derived from GIS Data model

of reservoir.

The scope for use of optical and microwave data for the purpose of inventory, mapping and monitoring of surface water bodies has been improved with the frequency of data available with the combination of optical and microwave satellite data sets. The updated scenario on the techniques being used and the demonstration on the utilization and implementation of quick data processing methodologies, derived results and examples on the utilization of datasets are explained. It can be concluded that, the surface water bodies can be delineated from satellite data at fortnight interval with the use Resourcesat-2 AWIFS data and RISAT-1 MRS/ CRS datasets. The datasets can be served through web services like ISRO Bhuvan and India WRIS.

References
Annual Report, Ministry of Water Resources, India. 2011. Bastiaanssen, W.G.M., Molden, D.J., and Makin, I.W., (2000). Remote sensing for irrigated agriculture: examples from research of possible applications. Agricultural Water Management, vol.46, pp. 137155. Brisco, B., Touzi, R., van der Sanden, J.J., Charbonneau, F., Pultz, T.J., DIorio, M. (2008). Water resource applications with RADARSAT-2 a preview, Int. J. of Digital Earth. 1(1):130-147 Brisco, B., Kapfer, M., Hirose, T., Tedford, B., and Liu, J. (2011). Evaluation of C-band polarization diversity and polarimetry for wetland mapping, Can. J. Remote Sensing. 37(1):82-92 Dam Registry, Central Water Commission, Ministry of Water Resources, Govt. of India. Eric, S. Kasischke, Kevin B. Smith, Laura L. Bourgeau-Chavez, Edwin A. Romanowicz, Suzy Brunzell, Curtis J. Richardson (2003); Effects of seasonal hydrologic patterns in south Florida wetlands on radar backscatter measured from ERS-2 SAR imagery, Remote Sensing of Environment 88, 423441. Frazier, P.S. and Page K.J., (2000). Water Body Detection and Delineation with Landsat TM Data, Photogrammetric Engineering & Remote Sensing, vol. 66, pp.1461-1468. Hanqiu and Xu., (2006). Modication of normalize difference water index (NDWI) to enhance open water features in remotely sensed imagery. Int. J. of Remote Sensing, vol.27, pp.3025-3033. Horritt D Mason., and Luckman, A J., (2002). Flood boundary delineation from Synthetic Aperture Radar imagery using a statistical active contour model, Int. J. of Remote Sensing. 22(13):2489-2507 Hu Zhuowei., Gong Huili., Zhu Liying. (2007). Fast Flooding Information Extraction in Emergency response of ood Disaster, ISPRS Workshop on Updating Geo-spatial Databases with Imagery & The 5th ISPRS Workshop on DMGISs Hui,F., Xu, B., Huang, H., Yu, Q,. and Gong, P., (2008). Modeling spatial-temporal change of Poyang Lake using multitemporal Landsat imagery. Int. J. of Remote Sensing, vol.29, pp. 57675784. Jean-Robert B Bwangoy., Matthew C Hansen., David P Roy., Gianfranco De Grandi., Christopher O Justice. (2010). Wetland mapping in the Congo Basin using optical and radar remotely sensed data and derived topographical indices, Remote Sensing of Environment. 114:7386 Junhua, Li., and Wenjun, Chen. (2005). A rule-based method for mapping Canadas wetlands using optical., radar and DEM data, Int. J. of Remote Sensing. 26(22):5051-5069.

31

NNRMS

BULLETIN

DECEMBER

2013

Conclusion

Laura Brown., and Kathyl L Young. (2006). Assessment of Three Mapping Techniques to Delineate Lakes and Ponds in a Canadian High Arctic Wetland Complex,ARCTIC. 59(3):283293. Lehner,B., and Doell,P. (2004) Development and validation of a global database of lakes, reservoirs and wetlands J. of Hydrology, vol. 296, pp. 122. Liu, H., Jezek K C. (2004). Automated extraction of coastline from satellite imagery by integrating Canny edge detection and e thresholding methods, Int. J. of Remote Sensing. 25(5):937-958 Li , J. and Chen, W. (2005).A rule-based method for mapping Canadas wetland using optical, radar and DEM data. Int. J. of remote sensing, vol.26, no.22, pp. 5051-5069. McFeeters, S.K.(1996).The use of normalized difference water index (NDWI) in the delineation of open water features. Int. J. of Remote Sensing, vol.17, pp. 1425-1432. Ma, D.J. , Yang, G. , Duan, H. , Jiang, J. , Wang, S., Feng, X., Li, A., Kong, F., Xue, B., Wu, J. and Li, S. (2011). Chinas lakes at present: number, area and spatial distribution. Science China Earth Sciences, vol. 54, pp. 283289. Maria Gabriela Parmuchi., Haydee Karszenbaum., and Patricia Kandus. (2002). Mapping wetlands using multi-temporal RADARSAT-1 data and a decision-based Classier, Can. J. Remote Sensing. 28(2):175186 Manavalan, P., Sathyanath, P., and Rajegowda, G.L. (1993). Digital Image Analysis Techniques to Estimate Water spread for Capacity Evaluations of Reservoirs. Photogrammetric Engineering & Remote Sensing, vol. 59, pp.1389-1395. Megan W. Lang, Philip A. Townsend, Eric S. Kasischk (2008). Inuence of incidence angle on detecting ooded forests using C-HH synthetic aperture radar data, Remote Sensing of Environment 112 pp. 38983907 Panchagnula Manjusree, Prasanna Kumar, L., Chandra Mohan Bhatt, Goru Srinivasa Rao, and Veerubhotla Bhanumurthy. (2012). Optimization of Threshold Ranges for Rapid Flood Inundation Mapping by Evaluating Backscatter Proles of High Incidence Angle SAR Images . Int. J. Disaster Risk Sci. 2012, 3 (2): 113122. Rajiv Kumar Nath., Deb S.K. (2010). Water-Body Area Extraction from High Resolution Satellite Images-An Introduction, Review, and Comparison, 2010. Int. J. of Image Processing. 3(6):353-372. Roberts, N., Taieb, M., Barker, P., Damnati, B., Icole, M., and Williamson, D. (1993). Timing of the Younger Dryas event in East Africa from lake-level changes. Nature, vol.366, pp. 146148. Ross, S., Brisco, B., Brown, R.J., Yun, S., and Staples, G. (1998). Temporal Signature Analysis of Rice Paddies using RADARSAT-1, Preliminary Results. 20th Canadian Symposium on Remote Sensing, Calgary, May 11-14. Schroeder, R., Rawlins, M A., McDonald, K C., Podest, E., Zimmermann, R., and Kueppers, M. (2010). Satellite microwave remote sensing of North Eurasian inundation dynamics, development of coarse-resolution products and comparison with high-resolution synthetic aperture radar data, Environ. Res. Lett. 5 015003. Sokol, J., NcNairn, H., and Pultz, T J. (2004). Case studies demonstrating the hydrological applications of C-band multi polarized and polarimetric SAR, Can. J. Remote Sensing. 30(3):470483. Subramaniam, S., Suresh Babu, A.V. and Roy, P.S.(2011). Automated Water Spread Mapping Using ResourceSat-1 AWiFS Data for Water Bodies Information System. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol.4, pp. 205 215.

of water Spread contours for Tungabhadra reservoir during low water levels of the year 2002 using satellite remote sensing technique, proc. GIS India, Jaipur, India. Thomas Hahmann, Birgit Wessel (2010); Surface Water Body Detection in High-Resolution TerraSAR-X Data using Active Contour Models, EUSAR 2010 Van der Sanden J J. (2004). Anticipated applications potential of RADARSAT-2 data, Technical Note, Can. J. Remote Sensing. 30(3):369379. Voeroesmarty, C J., Sharma, K P., Fekete, B M., Copeland, A H., Holden, J.,. Marble, J., and Lough, J A. (1997). The storage and aging of continental runoff in large reservoir systems of the world, Ambio.26:210219.

33

NNRMS

BULLETIN

DECEMBER

2013

Suresh Babu, A.V., Shanker, M., Venkateshwar Rao, V., and Bhanumurthy, V. (2003). Generation

GEOSPATIAL TECHNOLOGY FOR INVENTORY AND MONITORING OF GLACIAL LAKES AND WATER BODIES IN THE HIMALAYAN REGION
Abdul Hakeem K and Siva Sankar E Water Resource Group, National Remote Sensing Centre ISRO, Department of Space, Hyderabad - 500 037, India Email: abdulhakeem_k@nrsc.gov.in and sivasankar_e@nrsc.gov.in

Introduction
The Hindu-Kush Himalaya (HKH) region extends about 3500 km from Afghanistan in the west to Myanmar and China in the east, and runs through Pakistan, Nepal, India, Bangladesh and Bhutan (Nina Behrman, 2007). The HKH region is one of the most geologically active zones on Earth. Such is the size, elevation and climate of the HKH region that it hosts the largest areas of glaciers, snow and permafrost beyond high latitudes. These huge reservoirs of frozen fresh water represent the sources of a number of the worlds greatest rivers, and is often described as the water tower of Asia. The glaciers of the Hindu KushHimalayan (HKH) region are one of natures greatest renewable storehouses of fresh water; and they benet hundreds of millions of people downstream. The common natural hazards in the region are earthquake, landslide (due to seismicity), landslide induced ood, cloudburst, ice/snow avalanches, etc.

Glacial Lakes and Water Bodies in Himalayas

In general, the area higher than 4,000 m in elevation is mostly covered by snow and ice throughout the year. The glaciers, some of which consist of a huge amount of perpetual snow and ice, are found to create many glacial lakes. These glaciers as well as glacial lakes are the sources of the headwaters of many great rivers in the region. Most of these lakes are located in the down valleys close to the glaciers. They are formed by the accumulation of vast amounts of water from the melting of snow and ice cover and by blockage of end moraines. A glacial lake is a lake with origins in a melted glacier. Towards the end of the last glacial period, roughly 10,000 years ago, glaciers began to retreat. A retreating glacier often left behind large deposits of ice in hollows between drumlins or hills. As the ice age ended, these melted to create lakes. These lakes are often surrounded by drumlins, along with other evidence of the glacier such as moraines, eskers and erosional features such as striations and chatter marks. A water body is any signicant accumulation of water, usually covering the Earth. It most often refers to large accumulations of water, such as oceans, seas, and lakes, but it includes smaller pools of water such as ponds, puddles or wetlands. Some bodies of water are man-made (articial), such as reservoirs or harbors, but most are naturally occurring geographical features. In addition to several glacial lakes, Himalayan region also hosts many water bodies of varying size.

Glacial Lake Outburst Floods

A Glacial Lake Outburst Flood (GLOF) is a type of outburst ood that occurs when the dam containing a glacial lake fails. The dam can consist of glacier ice or a terminal moraine. Failure

earthquake or cryoseism, volcanic eruptions under the ice, or if a large enough portion of a glacier breaks off and massively displaces the waters in a glacial lake at its base. Catastrophic failure of the containing ice or glacial sediment can release this water over periods of minutes to days. Peak ows as high as 15,000 m3/sec have been recorded in such events, suggesting that the v-shaped canyon of a normally small mountain stream could suddenly develop an extremely turbulent and fast-moving torrent about 50m deep. On a downstream oodplain, it suggests a somewhat slower inundation spreading as much as 10 km wide. Both scenarios are signicant threats to life, property and infrastructure. Apart from landslides and river erosion, the mountainous Himalayan region is also quite susceptible to disastrous hazards due to GLOF (Pradeep K. Mool et al., 2001). There have been several occurrences of GLOF events in different parts of the Hindu Kush-Himalayan region. Downstream impacts of these GLOFs are reported to be highly destructive in nature and to lead to long-term secondary environmental degradation in the valleys, both physically and socio-economically. A report by International Centre for Integrated Mountain Development (ICIMOD) (Ives, J.D. et al., 2010) lists 34 GLOF events that have occurred in Nepal, TAR/ China, and Bhutan. The United Nations has adopted a series of monitoring efforts to help prevent death and destruction in regions that are likely to experience GLOFs. The importance of this situation has magnied over the past century due to increased population, and the increasing number of glacial lakes that have developed due to glacier retreat. While all countries with glaciers are susceptible to this problem, central Asia, the Andes regions of South America and those countries in Europe that have glaciers in the Alps, have been identied as the regions at greatest risk (http://en.wikipedia.org, 2013). The study of glacial lakes is also very important for the planning and implementation of any water resources development projects.

Geospatial Technology for Inventory of Glacial Lakes/ Water Bodies

Glaciers and glacial lakes are generally located in remote areas, where access is through tough and difcult terrain. Creating inventories and monitoring of the glacial lakes can be done quickly and correctly using satellite images and aerial photographs. Visual and digital image analysis techniques integrated with Geographic Information Systems (GIS) are very useful for the study of glacier, glacial lakes. Satellite remote sensing offers several unique advantages quick data collection, reliability, more accurate, repetitive collection, geometric integrity and digital storage, which makes it an ideal tool for mapping, inventorying and monitoring the natural resources.

Earlier Studies on Glacial Lakes Inventory in Himalayan Region

Many studies have been carried out for inventorying glacial lakes in different parts of the Himalayan region. International Centre for Integrated Mountain Development (ICIMOD) has carried out inventory of glaciers and glacial lakes in the Himalayan regions in cooperation with other organisations (Ives, J.D. et al., 2010). ICIMOD has created a comprehensive inventory and GIS database of glaciers and glacial lakes in Nepal and Bhutan using available maps, satellite images, aerial photographs, reports, and eld data on different scales. Along with national partner institutions, ICIMOD has also carried out similar studies in parts of India (Himachal Pradesh, Sikkim and Uttarakhand), and a few basins in China. (http://www.icimod.org). Space Applications Centre (SAC), ISRO has mapped glacial lakes in Himalayan region (SAC, 2011). National Remote Sensing Centre (NRSC), ISRO has also carried out inventory of glacial lakes 35

NNRMS

BULLETIN

DECEMBER

2013

can happen due to erosion, a build-up of water pressure, an avalanche of rock or heavy snow, an

and water bodies in Sutlej basin. Mangde Chu River Basin and Tawang river basin using medium resolution satellite data (NRSA, 2007a, NRSA, 2007b, & NRSC, 2009). GLOF study was carried out in Alaknanda valley through inventorying of glacial lakes using satellite images (Sanjay K. Jain et al., 2012). The efforts put by different organisations for inventory of glacial lakes in the Himalayan region were not covering entire Himalayas. In addition to this, Landslide Lake Outburst Floods (LLOF) occurred during 2000 and 2005 caused by the breaching of landslide lakes created in the Trans-Himalayan region along the Satluj River and Paree Chu (stream), respectively, both in the Tibetan region of China (Vikram Gupta and M. P. Sah, 2008). This LLOF had severe impact on the channel and infrastructure in the Kinnaur district of Himachal Pradesh. The lake formed due to the landslide across the Pareechu river in 2004 and the subsequent breach in 2005 was continuously monitored by NRSC using satellite data of different resolutions (NRSA, 2005). The occurrence of LLOFs and threat of GLOFs lead to the need for a comprehensive inventory and monitoring of glacial lakes and water bodies in the Himalayan region of Indian river basins. Based on the request from Central Water Commission under Ministry of Water Resources, NRSC carried out a study (NRSC, 2011) during 2010-11 for inventorying of glacial lakes/ water bodies in the Himalayan region of Indian River basins using satellite data of the recent year (2009). Glacial lakes with spatial extent greater than 50 ha were considered and inventoried. In addition to this, NRSC also monitored (NRSC, 2012) the spatial extent of the glacial lakes/ water bodies (identied/inventoried earlier) on monthly basis from June to October during 2011 and 2012.

Inventory and Monitoring of Glacial Lakes and Water Bodies

The study was carried out for the area covering Himalayas under the major river basins of Indus, Ganga and Brahmaputra. The study area extends across different countries namely India, Nepal, Bhutan and China. The index map showing study area is given in Figure.1.

Data Used for the Inventory

Fig. 1: Map showing the extent of Himalayan region of Indian River basins

The basic data required for the inventory of glacial lakes and water bodies are large-scale topographic maps, aerial

photographs, and multi-date satellite images. Large-scale topographic maps and aerial photographs are generally not available for this area and also the study area extends across international boundaries. The study mainly used the satellite images of the Advanced Wide Field Sensor (AWiFS) Resourcesat-1. Reports of earlier studies carried out by different organizations like International Centre for Integrated Mountain Development (ICIMOD), Geological Survey of India (GSI), on glaciers and glacial lakes inventory in the Himalayan region were also referred. For glacial lake identication from satellite images, it is preferable to have images with least snow cover and cloud cover. Generally the least snow cover occurs in the period between May and September in the Himalayas. If snow precipitation is late in the year, winter images are also suitable except for the problem of long relief shadows in the high mountain regions. The cloud free satellite data for the period of May-Dec, 2009 were procured and orthorectied with the reference image (Figure 2). Apart from the above satellite data, historic AWiFS data were also used.

Thematic Mapper (ETM) on-board Landsat-7 satellite available in the Global Land Cover Facility website (http://www.landcover.org) were also used as reference image as well as in supporting the identication of glacial lakes & water bodies.

Preparation of Input Data

Fig. 2: Mosaic of satellite images covering the study area

Mission (SRTM) based digital elevation model was used to delineate basin and

sub-basin boundaries with the help of AcrHydro tool. The SRTM Digital Terrain Elevation Data (DTED) one degree tiles covering the study area were mosaiced and lled for sinks. Then ow direction and ow accumulation images were generated. The default threshold value was used in delineating catchment boundaries using the ow accumulation grid image. Drainage lines were subsequently derived automatically. The southern boundary of the study area was restricted up to the Himalayan foot hills & administrative/political boundary, whereas northern boundary follows the natural basin boundaries of Indus, Ganga and Brahmaputra rivers as shown in Figure 1. The total geographical area of the study area is 98.6 mha. Since, the study area is highly undulating mountainous terrain, orthorectication of satellite data is essential for accurate water spread area estimation as well as for better location accuracy. In this study, orthorectication of AWiFS data was carried out using Projective Transform model available in ERDAS Imagine software. The orthorectied Landsat ETM images were used as reference image for collections of GCPs and the elevation values for GCPs were collected from SRTM DEM.

Identification & Mapping of Glacial Lakes/Water Bodies

The glacial lakes & water bodies were delineated based on the visual interpretation of satellite images of Resourcesat AWiFS sensor. Identication of features was done through panchromatic mode and/or different colour combinations of the multi-spectral bands namely green, red, near infrared and shortwave infrared. To identify the glacial lakes & water bodies, different image enhancement techniques are used to improve the visual interpretation. This method is complimented with the knowledge and experience of the Himalayan terrain conditions for inventorying glacial lakes and water bodies. With different spectral band combinations in False Colour Composite (FCC) and in individual spectral bands, glacial lakes and water bodies can be identied. Figure 3 and Figure 4 shows how typical water bodies and glacial lakes respectively are seen in satellite image. The water spread area of the lakes in images ranges in appearance from light blue to blue to black. The frozen lakes appear white in colour. They are generally associated with glaciers in the case of high lying areas, or rivers in the case of low lying areas. The boundary of glacial lakes and water bodies are digitized using on-screen digitisation techniques as polygon feature. The polygons are geoprocessed and the water spread area of glacial lakes/water bodies were computed digitally. The lakes are identied and digitized as on the date of satellite data. There is a possibility that some lakes that are frozen or overlaid with snow might have been omitted in this inventory. However, during monitoring phase of this study, the inventory was 37

NNRMS

BULLETIN

The Shuttle Radar Topographic

DECEMBER

2013

The ortho-rectied images from Enhanced

updated. Although it was planned to map only glacial lakes & water bodies that are larger than 50 ha in area, glacial lakes and water bodies having area more than 10 ha were also digitised.

Organisation of Digital Database

A digital database is necessary for the monitoring of these lakes & water bodies and to identify the potentially dangerous lakes. GIS is the most appropriate tool for spatial
Fig. 3: Water bodies as seen in satellite image

data input and attributes data handling. The digital database has the following attributes (Table 1). Glacial Lake Number: Each glacial lake & water bodies will have unique number in the digital database. The numbering is done sequentially within each 1:250,000 reference grid. The rst two digits indicate the basin number (01 - Indus, 02 - Ganga and 03 - Brahmaputra). The next three characters depict the reference number of the 1:250,000 SOI toposheet. The last three digit

Fig. 4: Glacial lakes as seen in satellite image

number indicates lake number within a grid of 1:250,000 SOI toposheet. The attributes

of the databases are Latitude and Longitude, Area, Length, Width and Altitude In addition to above, the attributes on basin, sub-basin name, country toposheet No. (1:250,000 scale) and name of the glacial lake/water body (where ever available) were also populated.

Salient features of inventory of glacial lakes/water bodies

Table 1: Sample record showing the GIS database attributes Basin_Name Lake ID River Country Toposheet No Latitude Longitude SRTM Elevation (m) Area (ha) Length Width GL/WB Lake Name Indus 01_520_001 Shyok China 52 O 330 4500 7901423.93 5064 65825.15 147347.41 29207.60 WB Pangong Tso The inventory of glacial lakes and water bodies in the Himalayan region of Indian river basins carried out using satellite images shows presence of 2028 glacial lakes and water bodies within the study area (Table 2). Out of these, 503 are glacial lakes and 1525 are water bodies. Brahmaputra basin part of the Himalayan region contains 294 glacial lakes and 1099 water bodies whereas Indus basin has 31 glacial lakes and 321 water bodies. There are 178 glacial lakes and 105 water bodies under Ganga basin (Figure 5).

Basin Name Brahmaputra Ganga Indus Total

Glacial Lakes Count Area (ha) 294 178 31 503 11,371 8,476 771 20,619

Water Bodies Count Area (ha) 1099 105 321 1525 1,94,562 54,247 2,94,791 5,43,602

Total Count Area (ha) 1393 283 352 2028 2,05,935 62,724 2,95,562 5,64,221

The distribution of glacial lakes and water bodies based on its water spread area is given in Table 3. It is observed that around 1600 glacial lakes/water bodies are having water spread area between 10 and 50 ha and around 200 water bodies have water spread area between 50 and 100 ha. There are 14 water bodies with water spread area more than 10,000 ha. All the glacial lakes are having water spread area less than 600 ha and 80% of them have less than 50 ha. The largest water body (Pangong Tso) with water spread area of 65,825 ha is located in Shyok sub basin of Indus basin. The largest glacial lake with water spread area of 542 ha is located in Brahmaputra basin. It is also observed from the Table 4 that more than 50% of glacial lakes/ water bodies (1169 nos.) are located within the elevation range of 4,000 to 5,000 m.
Numbers

The water body at the highest elevation of 5,810 m is located in the Karnal sub-basin of Ganga basin. The glacial lake at the highest elevation of 5,743 m is located in

Fig. 5: Distribution of glacial lakes/water bodies

Brahmaputra basin. A glacial lake at the lowest elevation of 2,744 m is located in the Gilgit sub-basin of Indus basin. The analysis of water spread area of glacial lakes/water bodies distributed within different elevation zones (Table 5) shows the presence of more number of smaller water bodies at higher elevation. It is also observed that 75% of the larger water bodies (water spread area > 1,000 ha) are located in the higher altitude.

Monitoring of Glacial Lakes & Water Bodies

The monitoring of glacial lakes and water bodies in the Himalayan region of Indian river basins was carried out through visual interpretation of satellite images from AWiFS sensor for the months of June to October during 2011 and 2012. All the water bodies that are larger than 50 ha in size only were considered (433 nos). Due to persistent cloud cover during the monsoon period, all the water bodies could not be monitored. 39

NNRMS

BULLETIN

DECEMBER

2013

Table 2: Basin wise details of glacial lakes & water bodies

Table 3: Distribution of glacial lakes & water bodies based on area Water Spread Area (ha) 10-50 50-100 100-1000 1000-10000 > 10000 Total 503 20,934 Glacial Lakes Count 404 62 37 Area (ha) 8,999 4,370 7,565 Water Bodies Count 1191 139 146 35 14 1525 Area (ha) 25,279 9,891 44,165 91,632 3,72,635 5,43,602 Total Count 1595 201 183 35 14 2028 Area (ha) 34,278 14,261 51,730 91,632 3,72,635 5,64,536

The maximum water spread area for each water body among the different dates of satellite for the month was considered for the nal analysis of the change in water spread. The following criteria were followed while monitoring the water bodies.

Table 4: Elevation zone wise distribution of glacial lakes & water bodies Elevation Zone (m) < 1000 1000-2000 2000-3000 3000-4000 4000-5000 > 5000 Total 1 11 212 279 503 12.85 1,099 7,578 12,243 20,934 Glacial Lakes Count Area (ha) Water Bodies Count 20 13 14 236 957 285 1525 Area (ha) 54,244 2,717 1,626 14,393 3,36,452 1,34,168 5,43,602 Total Count 20 13 15 247 1169 564 2028 Area (ha) 54,244 2,717 1,639 15,493 3,44,031 1,46,412 5,64,536

Table 5: Relation between elevation and water spread area Water Spread Area (ha) < 1000 1000-2000 2000-3000 3000-4000 4000-5000 > 5000 Total < 50 50-100 100-1000 100010000 5 1 1 2 24 2 35 8 3 14 > 10000 Total

Elevation Zone (m) 5 6 11 205 929 439 1595 3 3 2 21 104 68 201 4 3 1 19 104 52 183 3 20 13 15 247 1169 564 2028

A change in water spread area within +/- 5% is considered to be normal (insignificant) in remote sensing derived inventory studies.

Partly or fully cloud covered or frozen water bodies have not been considered in monitoring Only the maximum spatial extent of water spread area during each month has been mapped and compared with the spatial extent of water spread area mapped for the inventory year (2009). The status of monitoring carried out during the years 2011 and 2012 is given in Table 6.

Month

Increased 49 36 73 93 114 40 48 16 5 15

Decreased 20 17 23 56 97 126 73 128 200 228

No Change 109 72 57 94 149 101 96 96 100 123

June 2011 July 2011 August 2011 September 2011 October 2011 June 2012 July 2012 August 2012 September 2012 October 2012

178 125 153 243 360 267 217 240 305 370

Conclusion
The inventory and monitoring of glacial lakes and water bodies in the Himalayan region was carried out using satellite images from AWiFS sensor. The study reveals that the water spread area is uctuating over time due to either snow melt or rainfall. Regular monitoring the change in water spread area of the glacial lakes will be useful in identifying the potential dangerous lakes prone for GLOF. The 56 m resolution data of AWiFS is found to be sufcient for monitoring water bodies of size more than 50 ha. However, for monitoring other water bodies of size less than 50 ha, medium resolution images with better repetivity will be useful.

References
Geological Survey of India, (1999). Inventory of the Himalayan glaciers, a contribution to the international Hydrological Programme, Special publication no 34, GSI, India. http:// http://en.wikipedia.org/wiki/Glacial_lake_outburst_ood, Last accessed: 08-Aug-2013 Ives, JD., Shrestha, RB., Mool, P.K. (2010). Formation of glacial lakes in the Hindu Kush-Himalayas and GLOF risk assessment. Kathmandu: ICIMOD - ISBN978 92 9115 137 0 (printed) 978 92 9115 138 7 (electronic) Nina Behrman (Edited by), (2007). The Waters of the Third Pole: Sources of Threat, Sources of Survival:, Published by Aon Beneld UCL Hazard Research Centre, University College London. NRSA, (2005). Study of Pareechu Lake in Spiti basin using Satellite Remote Sensing Data, Technical Report, Published by National Remote Sensing Agency, Hyderabad. NRSA, (2007a). Inventory of Glacial Lakes and Water Bodies in Sutlej Basin, Technical Report, Published by National Remote Sensing Agency, Hyderabad. NRSA, (2007b). Inventory of Glaciers and Glacial Lakes in Mangde Chu River Basin, Technical Report, Published by National Remote Sensing Agency, Hyderabad. NRSC, (2009). Inventory of glacial lakes and water bodies in Tawang river basin, Technical Report, Published by National Remote Sensing Centre, Hyderabad. NRSC, (2011). Final Report of Inventory and Monitoring of Glacial Lakes / Water Bodies in the Himalayan Region of Indian River Basins, Technical Report Published by National Remote Sensing Centre, Hyderabad. 41

NNRMS

BULLETIN

DECEMBER

No. of glacial lakes/water bodies monitored

Water Spread Area

2013

Table 6: Status of glacial lakes/water bodies monitored

NRSC, (2012). Report on Monitoring of Glacial Lakes/Water Bodies in the Himalayan Region of Indian River Basins during 2011, Technical Report Published by National Remote Sensing Centre, Hyderabad. Pradeep K. Mool., Samjwal R., Bajracharya and Sharad P. Joshi, (2001). Inventory of Glaciers, Glacial Lakes and Glacial Lake Outburst Floods - Monitoring and Early Warning Systems in the Hindu Kush-Himalayan Region Nepal, ISBN 92 9115 331 1 Published by ICIMOD, Kathmandu, Nepal. Sanjay K. Jain., Anil K. Lohani., Singh R.D., Anju Chaudhary and Thakural, L.N. (2012). Glacial lakes and glacial lake outburst ood in a Himalayan basin using remote sensing and GIS, Natural Hazards, Volume 62, Issue 3, pp 887-899 SAC, (2011). Snow and Glaciers of the Himalayas, Technical Report, ISBN 978-81-909978-7-4, Published by Space Applications Centre, Ahmedabad. Vikram Gupta and Sah, M.P. (2008). Impact of the Trans-Himalayan Landslide Lake Outburst Flood (LLOF) in the Satluj catchment, Himachal Pradesh, India, Natural Hazards, Volume 35, Issue 3, pp 379-390

Introduction
Irrigation development is the key for ensuring water and food security of the country. Irrigation has been the major factor in increasing agricultural production in India. The massive development of a vast irrigation network in India has been recognized as a landmark in the history of agriculture. The development of water resources for irrigated agriculture received high priority in the different Plan periods. Expansion of irrigation facilities, along with consolidation of the existing systems, has been the main strategy for increasing production of food grains. The irrigation projects are classied into three categories viz major, medium and minor. Projects which have a Cultivable Command Area (CCA) of more than 10,000 ha are termed as major projects, those which have a CCA of less than 10,000 ha but more than 2,000 ha are termed as medium projects and those which have a CCA of 2,000 ha or less are known as minor projects. Irrigation support is provided through major, medium and minor irrigation projects and command area development in India. Total irrigation potential of the country is 139.90 mha. (Figure 1). Of this, the irrigation potential from Major and Medium (M&M) irrigation projects is 58.47 mha and 81.43 mha from Minor irrigation (64.05 mha from groundwater and 17.38 mha from surface water). The inter-basin transfer of water from surplus to decit basins envisaged to bring additional 35 mha under irrigation.

Irrigation Infrastructure in India

Minor irrigation projects have both surface and ground water as their source, while major and medium projects mostly exploit surface water resources. The plan-wise proliferation of schemes in major and medium sector is provided in Figures 2a & 2b. There are about 442 Major projects, 1230 Medium projects and another 215 Extension, Renovation and Modernization (ERM) projects taken up, out of which 276, 1008 and 126 projects have been completed respectively till the end of Xth Plan (INCID Publication, 2009).

Irrigation Potential Development in India

During the pre-plan period prior to1951, irrigation potential created through Major and Medium (M&M) sectors was 9.70 mha. In the 1st FYP (1951-56) period, the country launched a major irrigation programme and a number of multipurpose and major projects were taken up. During II and III Plan periods, several new projects were started. During the

NNRMS

BULLETIN

Shanker M, Suresh Babu AV, Simhadri Rao B and Venkateshwar Rao V Water Resource Group, National Remote Sensing Centre ISRO, Department of Space, Hyderabad - 500 037, India Email: shanker_m@nrsc.gov.in or muthyamshanker@gmail.com

DECEMBER

MONITORING of IRRIGATION PROJECTS USING HIGH RESOLUTION cartosat SATELLITE DATA

2013

IV Plan (1969-74), emphasis was shifted to the completion of ongoing projects, integrated use of surface and ground water, adoption of efcient management techniques and modernization of existing schemes. In the V Plan, Command Area Development Program (CADP) was launched with the objective of reducing the lag between potential created and optimum utilization of available land and water. During the VIII Plan, new projects were restricted considerably and greater emphasis
Fig. 1: Total irrigation potential in India

was laid on completion of projects. By the end of the X Plan, the projects completed, along with minor irrigation and ground water development, have created an estimated potential of about 102.77 mha including 42.35 mha. under Major & Medium projects (Figure 3). However, the target of 58.47 mha is expected to be achieved by the end of the XII Plan under Major & Medium projects. Irrigation potential created in the country from Major & Medium and Minor irrigation projects, which stood at 22.60 mha in 1951, has risen to 102.77 mha till the

Fig. 2a: Status of completion of irrigation projects taken up in India (Major)

end of X Plan period (2002-07). An irrigation potential of 7.3 mha has been created by the end of March, 2009 during XI plan. Thus, the total cumulative potential created in the country reached to 110.07 mha by the end of March 2009. The temporal irrigation potential creation and its utilization among Major & Medium irrigation projects in India is shown in Figure 4.

Accelerated Irrigation Benefit Program (AIBP)

A large number of river valley projects, both multipurpose and irrigation
Fig. 2b: Status of completion of irrigation projects taken up in India (Medium)

have spilled over from plan to plan mainly because of nancial constraints being faced

by the State Governments. As a result of this, despite huge investment having already been made on these projects, the country was not able to derive the desired benets. There were 171 major, 259 Medium and 72 ERM on-going Irrigation projects in the country at various stages of construction with spillover cost of Rs. 75,690 crore, at the end of VIII plan (i.e end of March 1997). This was a matter of great concern for the Union Government and remedial measures for expeditious completion of some of the projects which were in advanced stage of completion became necessary

State Governments was identied as the single most important factor for the inordinate delay in irrigation projects execution. To o v e r c o m e t h e f i n a n c i a l resources constraints, a special scheme named Accelerated Irrigation Benefits Programme (AIBP) was launched in Central Loan Assistance (CLA) to the States
Fig. 3: Status of IP creation since 1951 to 2007 (Area in mha)

for accelerating the implementation of large irrigation and multipurpose projects. AIBP was aimed at twin benets of accelerating the ongoing irrigation / multipurpose water resources projects and realization of bulk benets from completed irrigation projects quickly. The loan assistance under this scheme was extended to selected irrigation projects in the country, with the objective to accelerate the implementation of those projects, which were beyond resource capability of the States or were in advanced

Fig. 4: Irrgation potential creation and utilisation among Major & Medium (mha)

stage of construction and could yield irrigation benets in the next few agricultural seasons (A R MoWR 2007-08). Since its formulation, the terms of the program have been widened and liberalized over the time. As per present pattern of assistance under the AIBP, the Central Government provide grant to the irrigation projects as an incentive to the States for creating irrigation infrastructure in the country. As on date, Major, Medium and ERM projects are eligible for central assistance under AIBP. The surface water minor irrigation schemes of special category States as well as schemes in drought prone and tribal areas in non-special category. States are also eligible for central assistance under AIBP. So far, a total of 293 Major and Medium irrigation projects (with an I.P of 13.79077 mha) have been included under AIBP, out of which, 140 projects have been completed, another 148 projects (with an I.P of 10.382989 M.ha) are ongoing as on March 2012 and remaining 5 projects were deferred.

Conventional Monitoring Mechanism and Need for Spatial Information

A comprehensive physical and nancial periodical monitoring of the Major and Medium projects is being carried out by the Central Water Commission / Ministry of Water Resources with the emphasis on quality control through CWC regional ofces situated all over the country. 45

NNRMS

BULLETIN

1996-97 by Government of India, for providing

DECEMBER

2013

(www.wrmin.nic.in). The lack of funds with

Conventional monitoring reports are prepared based on inputs provided by eld authorities without the spatial content and random eld checks necessitated by time constraint. Thus, the project monitoring is limited in nature, lacking the synoptic view of the critical gaps and the quantitative progress in actual irrigation infrastructure creation. However, the complete spatial information about the project status is essential for objective monitoring, identication and prioritisation of critical gaps for efcient utilisation of resources.

Role of Satellite Remote Sensing data in Monitoring of Irrigation Projects

Satellite remote sensing is an ideal tool for mapping, inventorying and monitoring purposes. High resolution satellite data like Cartosat-1 provides excellent opportunities to capture the existing irrigation infrastructure and for monitoring the project implementation progress. High resolution satellite data acquisition can be planned to match the eld reporting for effective monitoring of the project. Monitoring visit can be effectively planned and carried out with the help of critical gap areas identied in irrigation infrastructure creation using satellite data and thus minimizing the time required for each visit. Thus, satellite based monitoring addresses both inadequacy in number of monitoring visits and total area monitoring rather than random checks carried out presently.

Conceptualisation and Methodology Development

At the instance of Planning Commission, NRSC, ISRO proposed the concept on Satellite Technology Applications in Irrigation Infrastructure Mapping including the scope of monitoring the progress made and potential created through AIBP. Subsequently, NRSC, ISRO has developed the methodology for assessment of irrigation potential created through inventory and mapping of irrigation infrastructure using high resolution satellite data through a pilot study carried out in two selected irrigation projects (Upper Krishna Project in Karnataka and Teesta Project in West Bengal during 2004-05. Basic approach (Figure 5) consists of inventory and mapping of existing irrigation infrastructure (such as canal network, irrigation and other related structures) from high resolution satellite data in an on-going irrigation project and comparing the physical progress & status to the design with proposed irrigation infrastructure. Based on the completion status of irrigation infrastructure derived from the satellite data and considering the hydraulic connectivity from source to the outlet, the irrigation potential created in the project is assessed. The pilot study had captured the
Fig. 5: Methodology for monitoring satellite based irrigation projects

ground reality of the irrigation infrastructure

8). Based on the satellite derived information, percentage progress of AIBP works along with critical gap areas were identied and an assessment of irrigation potential created in the project was carried out. Thus, the satellite data can capture the existing irrigation infrastructure and completion status. It also provides possibility to monitor the entire irrigation project yet with
Fig. 6: Canal (lined) as observed in Cartosat data

limited eld checks overcoming the limitation of physical visits to entire project area. The encouraging and satisfactory results of the pilot study (in terms of progress, status of infrastructure and associated irrigation potential created) were objective in nature and compared well with ground realities which were veried and reported by CWC eld ofces and State Government departments. Thus, the pilot study during 2004-2005 demonstrated the utility of high resolution satellite data for monitoring the progress of ongoing irrigation projects and its application potential for assessment of Irrigation Potential (I.P.) created.

Operationalisation
The availability of Cartosat-1 (2.5 m resolution) data from 2005-06 provided a cost effective solution for upscaling the study. In view of the importance and utility of results arising out of satellite data based pilot study, Planning Commission in consultation with NRSC and MoWR decided to upscale the study to national scale covering all AIBP Projects with an estimated irrigation potential of 10 mha spread across different
Fig. 7: Gaps / Pending work in Canal network as observed in Cartosat data

States in India in a phased manner. A c c o rd i n g l y, u n d e r p h a s e - I ,

Assessment of Irrigation Potential Created in 53 AIBP funded Irrigation Projects in India using Cartosat-1 Satellite data with an I.P. target area of 5.45 mha spread across 18 States in India during 2007-09 (Figure 9) was carried out. 47

NNRMS

BULLETIN

its physical status for time stamping its

DECEMBER

2013

and its status in spatial domain (Figures 6,7 &

Cartosat data derived spatial irrigation canal network and structures along with assessed Irrigation Potential created for Upper Wardha (Maharashtra) is shown in Figure 10. Satellite data based I.P. created was compared with eld based report for 50 projects (Summary Report, February 2010). As per the eld report, the I.P. created is about 25% more than the satellite data based study. Assessment also indicated large deviation (>25%) in eld reporting in 15 projects out of 50 projects studied (Figure 11).
Fig. 8: Pending cross Drainage structure as observed in Cartosat data

CWC and MoWR utilized the study result for reconciliation of figures on I.P creation through verications and clarications from respective State departments. The spatial irrigation infrastructure information generated in the study is utilized for further monitoring by CWC. Thus, the technology has been well received by the Central Water Commission (CWC) and MoWR and is being used as a tool for effective AIBP programme implementation. Further, the technology has been recognized by the Planning Commission and suggested for monitoring of all projects funded under Accelerated Irrigation Benet Porgramme (AIBP) using high resolution Cartosat satellite data.

Capacity Building
To create awareness among planners, managers, engineers and other stake holders,
Fig. 9: Distribution 53 projects in AIBP Phase-I

a two day workshop was conducted during May 2011 on the use of high resolution Cartosat data for monitoring irrigation

infrastructure and potential creation. The workshop recommended for adoption of technology among Central and State Govt. departments after detailed deliberation on conventional monitoring mechanism and results of the operational phase study by NRSC. To further the capacity building process, NRSC has executed additional 50 irrigation projects (Figure 12) essentially to transfer technology in the domain of satellite based monitoring of AIBP funded irrigation projects during 2011-12.

Institutions consisting of State Remote Sensing Centres and academic institutions located across different states. During this phase, training and guidance were provided to all teams from Partner Institutes and the technology was transferred to CWC by providing intensive two projects. The study was completed in December 2012.
Fig. 10: Cartosat derived irrigation infrastructure and potential created in Upper Wardha project

In continuation of demonstration and operational use of application technology, capacity building / technology transfer to CWC and other institutions on the satellite based monitoring by NRSC, adoption of technology by line departments is next step
Fig. 11: Satellite derived study result in Phase-I projects

for institutionalization. There are 148 irrigation projects currently ongoing under AIBP in India. CWC does monitoring of these projects twice a year and needs two time period Cartosat data (pre and post monsoon) in near real time for use by their monitoring ofces located across the country. In this regard, ISRO-BHUVAN platform meant for Earth Observation visualisation provided an excellent opportunity for online monitoring of irrigation projects. It facilitates hosting of satellite data with user access control and provides multiple access facility from various locations. NRSC demonstrated the potential of online monitoring through ISRO-BHUVAN platform using near real time Cartosat data during 8th NNRMS SC-W (National Natural Resources Monitoring System Standing Committee on Water Resources) meeting held on 18th June 2012. SC-W recommended for institutionalization 49

Fig. 12: Distribution of 50 projects in AIBP Phase-II

NNRMS

Online Monitoring through ISRO-BHUVAN and Institutionalization

BULLETIN

on the job training during execution of

DECEMBER

2013

This study was carried out with 14 Partner

of the technology among the Central and State Govt. line departments for online monitoring using Cartosat data through ISRO-BHUVAN web services platform. A team headed by Chief Engineer, Project Monitoring Organisation, CWC had thoroughly evaluated the ISROBHUVAN services for online monitoring of AIBP projects and recommended for its implementation. Accordingly, CWC (MoWR) had prepared a road map for implementation of satellite data based online monitoring of AIBP projects through ISRO BHUVAN platform. NRSC has provided on the job training to 30 CWC ofcers through two training programs during December 2012 and February 2013 for implementation of online monitoring of AIBP projects.

Conclusion
Satellite based monitoring provides an excellent opportunity to monitor the entire irrigation project yet with limited eld checks overcoming the limitation of physical visits to entire project area. ISRO-BHUVAN platform is an excellent facility for online monitoring of AIBP implementation by CWC. The adoption of this new application and using ISRO-BHUVAN platform for online monitoring by CWC and other line departments would go a long way in institutionalization of technology. This particular application will be one of the largest user driven utilization of high resolution Cartosat satellite data for their operational program implementation.

References
Annual Report (2007-08), Central Water Commission, MoWR, Govt. of India (http://cwc.nic.in/main/downloads/AR_07-08.pdf). Annual Report (2007-08), Ministry of Water Resources (MoWR), Govt. of India. Summary Report on Assessment of Irrigation Potential Created in AIBP funded irrigation projects in India using Cartosat satellite data (53 projects) (NRSCRS&GIS AA-WR&OG-WRD-February 2010-TR-153). Website of Ministry of Water Resources (MoWR), Govt. of India (http://wrmin.nic.in). Water Resources Development in India. INCID Publication, 2009.

Introduction
A river is the general term for a channel and develops various landforms through channel processes. The main channel/ uvial processes are erosion, transportation and sedimentation. Erosion predominates in the upper reach area of a drainage basin and valleys composed of channels and slopes are formed. The materials brought to the lower reaches in a channel are sediment load. Weathering of rocks composing slopes is the main cause of production of sediment load and it is deposited in the form of alluvial plains. River bank erosion leads to sequential changes in the position of banklines as well as various changes within the channel. Three basic channel patterns are detected in alluvial plains namely braided, meandering and straight. River morphology is explained in channel patterns and channel forms and is decided by factors which are inter-related to each other such as discharge, water surface slope, water velocity, depth and width of channel and river bed materials etc. For a scientic and rational approach to different river problems and proper planning and design of water resources projects, an understanding of the morphology and behaviour of the river is a pre-requisite. Morphology of river is a eld of science which deals with the change of river plan form and cross sections due to sedimentation and erosion. In this eld, dynamics of ow and sediment transport are the principal elements. The Morphological studies, therefore, play an important role in planning, designing and maintaining river engineering structures. In order to assist the engineers of the concerned departments and other agencies, morphological study reports with broad guidelines are prepared. Satellite Remote Sensing is being successfully used for various river morphological/ engineering studies. The present paper discusses about the role of remote space technology in mapping and monitoring the shifts in river bank line and erosion/deposition with the help of multi-sensor, multi-spectral, multi-date satellite images. The changes have been illustrated by comparing pre and post event satellite images for better understanding.

Remote Sensing and GIS for River Morphological Studies

There are direct and indirect methods for monitoring the river bank erosion. The direct method is taking measurements from the eld in terms of linear rates of erosion, volumes of erosion and channel cross section. The indirect method is by analyzing the archival sources that exist at various timescales with the sediment records. The archive sources can be conventional survey maps, aerial photos or satellite images. Using multi-temporal

NNRMS

BULLETIN

Manjusree P, Satyanarayana P, Bhatt CM, Sharma SVSP, Srinivasa Rao G Remote Sensing Applications Area, National Remote Sensing Centre ISRO, Department of Space, Hyderabad - 500 037, India Email: manjusree-p@nrsc.gov.in

DECEMBER

REMOTE SENSING AND GIS FOR RIVER MORPHOLOGY STUDIES

2013

high-resolution satellite data, the latest river conguration, shift in the river courses, formation of new channels/oxbow lakes, bank erosion/deposition, drainage-congested areas, etc. can be mapped at different scales. Since accurate river conguration is obtained, it can be used for laying models for conducting river behaviour studies. Information derived from remote sensing can be used for other river morphological application studies like monitoring the existing ood control works and identication of vulnerable reaches, planning bank protection works, drainage improvement works etc. Works have been carried out worldwide using remote sensing and GIS for river morphological studies. Sarma and Basumallick (1980) studied the bank line migration of the Burhi Dihing River using topographic maps and eld survey. Bardhan (1993) studied the channel behavior of the Barak River using satellite imagery. Naik et al. (1999) studied the erosion at Kaziranga National Park using remote sensing data. Goswami et al. (1999) carried out a study on river channel changes of the Subansiri (northern tributary of Brahmaputra River) in Assam, India using information of topographic sheet and satellite data.

Methodology
In the present study application of remote sensing and GIS for the identication of the changes in the river bankline, erosion and deposition and in quantication of the changes occurred has been demonstrated taking stretches along the Brahmaputra River, Ganga River, Kosi River and Gandak River. Figure 1 shows the methodology.

Brahmaputra River, Assam

The Brahmaputra River is one of the largest alluvial rivers in the world characterized by frequent bank erosion, exceedingly large ow, enormous volume of sediment load, continuous changes in channel morphology, rapid bed aggradations and bank line recession and erosion. According to Assams Water Resources Department, Assam valley portion
Fig. 1: Methodology ow chart

of the Brahmaputra has lost approximately 7.4 per cent of its land area due to river bank erosion and channel migration.

Satellite images for the year 2002 and 2010 have been used for mapping of the shift in the river bank lines along the Brahmaputra River in Assam. It is observed that the Brahmaputra River has shifted its bank line drastically, causing severe damage to agriculture as well as habitat areas on its both sides. The total land area lost due to erosion has been estimated at 27,098 ha (15248 ha. on the north bank and 11850 ha. on the south bank). On the south bank, the maximum impact of erosion was noticed in the Golaghat district (3221 ha. approx), followed by Marigaon (2815 ha. approx) and Dibrugarh (1442 ha. approx) during the period. On the north bank, Dhubri (3030 ha. approx) accounted for maximum area affected by erosion, followed by Sonitpur (2823 ha. approx) and Dhemaji (2671 ha. approx). Deposition of silt has been very low, compared to the rate of erosion. On the south bank, deposition has been 18% of the total land eroded, while on the north bank it is 24%. Maximum deposition has been observed in Kamrup district (537 ha.) on the south bank and Tinsukia (1396 ha) on the north bank.

deposition during 2002 and 2010 in parts of Upper, Middle and Lower Assam. Majuli Island, Assam : Majuli, the largest inhabited river island bounded by the river Subansiri to the north and Brahmaputra River to the south, is one of the subdivisions of the Jorhat district, Assam. and posses a signicant concern. The extreme braided nature of the Brahmaputra coupled with silt and sand strata of the banks is the main cause of erosion. It was observed in 2010, the Majuli Island reduced to 448.23 km2. Erosion in the island was observed due to the shift in Brahmaputra river at Goalgaon and Haldibari. Deposition was observed at Mayadobi mainly due to shift in the river Subansiri besides erosion due to Brahmaputra. Figure 5 shows the erosion and deposition during 1996, 2002 and 2010 in Majuli Island. Erosion of the island is a continuous processes
Active river channel during 2010 and Water bodies River Bank Erosion Other Major Roads River Bank Deposition River bankine of 2002 State boundary District boundary Railway National Highway River bank Settlements

Fig. 2: Map showing the erosion and deposition during 2002 and 2010 in parts of Upper Assam

Active river channel during 2010 and Water bodies River Bank Erosion Other Major Roads River Bank Deposition River bankine of 2002 State boundary District boundary Railway National Highway River bank Settlements

Ganga River, Bihar and West Bengal

Ganges is one of the largest perennial river systems amongst the 14 major rivers in India having its source in the Gangotri glacier in the Himalayas at an elevation of about 4200 m above Mean Sea Level (MSL). River dynamics is one of the major problems in rivers draining the Ganga plains. Near Ballia and Chappra, Bihar: The present study was conned from West of Ballia to the East of Chappra and the entire region actually falls as part of U.P.- Bihar border. The Ganga River in this region exhibits a great deal of dynamicity thereby showing a continuous change in the river morphology

Fig. 3: Map showing the erosion and deposition during 2002 and 2010 in parts of Middle Assam

Active river channel during 2010 and Water bodies River Bank Erosion Other Major Roads River Bank Deposition River bankine of 2002 State boundary District boundary Railway National Highway River bank Settlements

as well as bank-line stability. Near Khawashpur in Ballia, the shift is mainly in the North-West direction by around 850 m. South West of Chappra near 53

Fig. 4: Map showing the erosion and deposition during 2002 and 2010 in parts of Lower Assam

NNRMS

BULLETIN

DECEMBER

2013

Figures 2, 3 & 4 show the erosion and

Ghazipur,the river shows shifting towards the Northern direction by 1.34 km. Near Bakhorapur, the river shifted by 1.1 km towards Southwest. A detailed mapping regarding the area under aggradation and degradation showed that both the bank-lines in this region have undergone aggradation as well as degradation with a varying amount. From Figure 6, it can be seen that the Northern bank (left bank) has undergone more aggradation than degradation which indicates the shrinking of the river bed whereas conversely the Southern bank shows much more degradation then aggradation
Fig. 5: Maps showing the erosion, deposition and the changes in the river banklines observed during 1996, 2002 and 2010 in Majuli island

which indicates the widening of the river bed. The river has undergone deposition and the shifting of the river channel has occurred mainly in the north direction.

The present study showed prominent changes in river bank-line as well as various uvial features. It showed great deal of variation in direction, dimension and magnitude. The shifting of the river channel in this region is mainly due to the erosional and depositional processes of the river. The overall direction of shifting is in the South and South-East direction. Amount of deposition near Kushwarapur in Ballia and Ghazipur in 2004 was estimated to be 491and 283 ha. respectively. Near Jalangi in West Bengal: Due to the shift in the river course of Ganga River near Jalangi, substantial erosion is observed during the last one decade 1987-1997. The analysis of IRS images of 1987-1992-1997, it is observed that the river has shown shift towards south at two locations (encircled with yellow colour). Figure 7 shows the river bankline shift in Ganga River near Jalangi in West Bengal. Upstream of Jalangi about 722 ha. of erosion was observed and near Jalangi about 5218 ha. of erosion was observed to have taken
Fig. 6: River bankline shift in Ganga River near Chappra in Bhojpur district, Bihar

place during last one decade. The Kosi River has been responsible

for some of the most devastating oods in the North Bihar. It is also known as the Sorrow of Bihar due to the frequent channel migration and the extensive ood damage it causes in the region. During the last two centuries the Kosi river has shifted its course by about 150 km (Gole and Chitale, 1996; Wells and Dorr, 1987). The presence of a number of palaeo channels all along the surface of Kosi alluvial fan clearly visible on the satellite images, bear testimony to the dynamic nature of the Kosi river.

Bihar: The change in the course of the Kosi River led to a breach in the eastern embankment causing severe oods in parts of North Bihar. Figure 8 shows the changes in the river course of Kosi River and subsequent ooding in parts of North Bihar. The breach in the eastern embankment of the Kosi river took place about 12 km upstream of Birpur barrage, near Kusaha in Nepal on August 18, 2008. The analysis of the satellite datasets
Fig. 7: River bankline shift in Ganga River near Jalangi in West Bengal

upstream of the Kosi barrage has been very dynamic and frequently changing its course as observed from the river course delineated for IRS LISS III images of 1997, 2006 and 2011 (Figure 9). At the breach location the river ow was westwards (Figure 9a, location1 ) whereas upstream the river was owing eastwards (Figure 9a, location A) during 1998. However due to sedimentation and formation of large sand bar along the main channel, the river ow shifted towards east (Figure 9b, location 2) at the breach location whereas upstream of it shifted towards west (Figure 9b, location B) as visible from the images of 2001. Although the ow had shifted eastwards, the major portion of the discharge was still passing down the west side (Figure 9c, location 1) of the bar, until 2007. However, further sediment deposition along the west side of the channel blocked the free ow of water and the river has almost abandoned the western channel, with the entire discharge flowing down the east side of the sand bar (Figure 9d, l o c a t i o n 2 ) duri ng 2008. The fl ow direction upstream of the breach location had also gradually changed prior to the breach, providing a more erosive angle of attack against the eastern embankment (Figure 9b-f, dotted lines). The river ow had become almost perpendicular in the months

Fig. 8: Flooding caused due to change in the river course of Kosi River in Bihar

Fig. 9: Change in the river course of Kosi River near breach location during 1997-2011

55

NNRMS

reveals that the Kosi river course 30 km

BULLETIN

DECEMBER

2013

Kosi River Embankment Breach,

prior to the breach (Figure 9e, dotted lines), thrusting pressure directly on the eastern embankment. From the satellite images it can be seen that the river was owing very close to the eastern embankment during 2007 (Figure 9c) and during, 2008 the river can be seen almost abutting the eastern embankment (Figure 9d & e). Measurements made along the river in this stretch show that the river was owing at a distance of less than 300 m from the eastern embankment. From the satellite based study it is observed that the reduction in the cross-sectional area of the river and channel carrying capacity due to the sedimentation together with the change in the angle of attack of Kosi river, upstream of the breach location prior to the breach, may be one of the causes that led to Kosi river avulsion. The change in the angle of attack upstream of breach location may be a surface manifestation of the sub-surface structural movements as the higher Himalayas and the foothill zone through which the Kosi river emerges into the plains near Chatra constitutes a tectonically active zone (Wells and Dorr, 1987, Agarwal and Bhoj, 1992, Sinha et al., 2005). Therefore, the ood management of the Kosi river basin requires an integrated approach, which not only considers the hydrological factor, but also the geological and geomorphological aspects of the river basin.

Conclusion
Remote sensing technology due to its various advantages could be used by the planners for taking measures required for channel stabilization and strengthening of embankments. It is essential to monitor the vulnerability of the ood control structures, identify the changes in the river course, formation of new oxbow lakes and to understand the behaviour of the river to lay physical models. Climate change would seriously affect the quality and quantity of water in the rivers particularly due to extreme events like oods and droughts, seawater intrusion and anthropogenic contamination. In this regard Remote Sensing and GIS technology can help in undertaking mitigation measures.

References
Agarwal, R P, and Bhoj, R. (1992). Evolution of Kosi River fan, India: structural implications and geomorphic signicance. International Journal of Remote Sensing, 13 (10): 18911901. Bardhan, M. (1993). Channel stability of Barak river and its tributaries between Manipur-Assam and Assam- Bangladesh borders as seen from satellite imagery, Proc. Nat. Syrup. on Remote Sensing Applications for resource Management with special emphasis on N.E. region, held in Guwahati, Nov. 25-27, 481-485. Gole CV, and Chitale SV (1996) Inland delta building activity of Kosi River. Journal of the Hydraulics Division. American Society of Civil Engineers 92: 111-126. Gole, C.V. and Chitale, S.V. (1996). Inland delta building activity of Kosi river. Journal of the Hydraulics Division, American Society of Civil Engineers 92: 111-126 Goswami, U., Sarma, J.N. and Patgiri, A.D. (1999). River channel changes of Subansiri in Assam. India. Geomorphology, 30: 227-244. Naik, S.D., Chakravorty, S.K., Bora, T. and Hussain. (1999). Erosion at Kaziranga National Park, Assam, a study based on multitemporal satellite data. Project Report. Space Application Centre (ISRO) Ahmedabad and Brahmaputra Board, Guwahati. Sarma, J.N. and Basumallick, S. (1980). Bankline Migration of Burhi Dihing river, Assam. IND. J. Ear. SCI., 11 (3&4): 199-206. Sinha R, Jain V, Prasad B G, and Ghosh, S. (2005). Geomorphic characterisation and diversity of the uvial systems of the Gangetic plains. Geomorphology, 70/3-4: 207225. Wells N.A, and Dorr, J.A. (1987). Shifting of the Kosi river, northern India. Geology, 15: 204-207.

WATER RESOURCES AND HYDROLOGY OF THE WESTERN GHATS: THEIR ROLE AND SIGNIFICANCE IN SOUTH INDIA
Mysooru R. Yadupathi Putty1 and Madhusoodhanan CG2 1 Dept. of Civil Engineering, The National Institute of Engineering, Mysuru - 570008, India 2 Dept. of Civil Engineering, Indian Institute of Technology, Bombay - 400076, India Email: puttyyadupathi@gmail.com

Introduction
The Western Ghats, locally called the Sahyadri ranges, are a series of hill ranges bordering the west coast of India, almost throughout its length. The region of Western Ghats is one of the eight hottest biodiversity hotspots in the world and has also been recently acknowledged as a World Heritage Site (Myers et. al., 2000; UNESCO, 2012). The Western Ghat mountains extend 80 198 - 210 16 24 N and 720 56 24 - 780 19 40 E and encompass an area of approximately 1,29,000 km2, in the states of Kerala, Tamil Nadu, Karnataka, Goa, Maharashtra and Gujarat (WGEEP, 2011). They run almost parallel to the coast and have a 1600 km long north south orientation, with an extremely narrow east - west prole (Figure 1). The Western Ghats are also one of the worlds most heavily populated biodiversity hotspots (Cincotta et al., 2000), providing for and supporting 245 million people with water for drinking, transport, irrigation and hydroelectric power, together with food and resources to sustain livelihoods (WGEEP, 2011). As the repository of an exceptional assemblage of life forms and human cultural diversity, the Western Ghats have a global pre-eminence. The remnant natural ecosystems of the Western Ghats are currently subject to a plethora of threats that make the region a highly sensitive zone.Though there are numerous studies related to the biodiversity and its conservation in this biologically rich hill tract, research into the hydro-climatological aspects of the Western Ghats, which ultimately sustain these unique ecosystems and the people in the whole of south India, has been negligible. This paper presents a brief account on the available knowledge of the hydrological aspects of the Sahyadris, highlighting the uniqueness of the region and bringing forth the probable elds open for research. Physical Settings: Physiography, Geology and Soils The physiographic map of the region, derived from 1 km resolution SRTM DEM data, is shown in Figure 2 (Jarvis et. al., 2008). The Sahyadris consist of mountain ranges with an average elevation of 1000 m, displaying an extremely steep western face and gently descending eastern slopes. The eastern slopes merge with Mysore plateau in the south and Deccan plateau in the north, which are contiguous and located north of Palakkad gap (between 10030-110 N). The Palakkad gap is a major discontinuity in the Sahyadri ranges, extending about 30 km with an average elevation of 200 m. South of Palakkad gap, the eastern and western slopes are equally steep with drastically different ecological conditions

NNRMS

BULLETIN

DECEMBER

2013

(Nair, 1991). The hills do not rise much beyond 1500 m in the northern tract, while in the south they tend to be rounded and rise even beyond 2000 m. The Anaimudi peak in Kerala is the highest peak (2695 m) south of the Himalayas. Geologically, the Sahyadris are a stable mark of Archaean and Pre-Cambrian formations, where the mountain building has ceased in the Pre-Cambrian times (Daniels, 2001). Most of the exposed Gneisses of the Sahyadris are 2,500 million years old. The non-metamorphic sedimentary formations are very rare and found only along the coastal belt (WGEEP, 2011) in north. Lithologically, Sahyadris vary from basaltic Deccan Trap, with relatively fragile rocks, North of 160 N, to the Precambrian Archaean crystalline hard rocks with at hill tops in the south. The continental drainage divide coincides with the crest almost throughout the length of the Western Ghats. However, in many places, the western faulting has led to river capture and diversion of the easterly drainage to the west (Radhakrishna, 1991) is noticed. The soils of Sahyadris consist mainly
Fig. 1: Location and the extent of the Western Ghats

of the derivatives of the basaltic lava and ancient metamorphic rocks, rich in iron and manganese (Pascal, 1988). There are seven

main soil groups found in the region, viz. laterites (high and low), red loam, medium black soils, hill soils, red gravelly soils and alluvial soils including coastal alluvium, mixed red and black soils. Soils vary from humus rich peat in the montane areas to laterite in the lower elevations on the western foothill belts. Soils are generally acidic. Isolated small expanses of exposed lateritic rocks, that are mostly unt for plant growth, characterise low lands along the coastal hills. The depth of soils on the western slopes of south and central zones belong to the very deep and deep categories, while in the north, they belong to shallow and very shallow categories. All over the ghats, soils are well drained with very high inltration rates (NBSS&LUP, 1996 a & b; Sivaprasad et al.,1998; Challa et al.,1999; Harindranath et al., 1999).

Land use and Land cover

With their high rainfall regime, the western slopes of the Ghats have a natural cover of evergreen forests, which changes to moist deciduous types as one comes to the eastern slopes. The vegetation reaches its highest development towards the southern tip in Kerala with rich tropical rain forests. Together, the forests cover approximately 20 percent of the total area of the Sahyadris. The majority of the area under moist forest types falls within the southern states of Kerala and Karnataka (IIRS, 2002). The entire Western Ghats have now been declared an Ecologically Sensitive Area. It is estimated that not more than about 7% of the area of the Western Ghats is presently under primary vegetation cover, though a much larger area is under

(WGEEP, 2011). The traditional land use in the non-forested tracts of the Ghats has been Paddy cultivation in the valleys, supplemented by cultivation of millets and legumes on the hill slopes. A number of horticultural and tuber crops such as tea, coffee, rubber, cashew, tapioca and potato were introduced to this region through European inuence. Pepper and cardamom, which are native to the evergreen forests of the Sahyadris were also taken up as plantation crops on a more extensive scale in the modern times. These changes have led to rampant conversion and high fragmentation of the Western Ghats forest landscape. At the same time, the deep river valleys spread across the Ghats were thrown open by various hydroelectric power projects causing signicant submersion of virgin forest tracts. Developments in remote
Fig. 2: Physiographic features of South India underlining dominance of the Western Ghats

sensing technology have revolutionised the understanding of these changes in the region. But, the implication of these large scale

landscape changes on the water resources of the region is yet to be studied.

Rainfall and Climate

The heterogeneity in the physical features of the Sahyadris has a profound inuence on the climate of the region (Nair, 1991). The region lies in the tropical South Asian monsoon tract characterised by wet summers and dry winters. The climate in general is hot and humid tropical, with mean temperatures between 20 and 240 C. However, it frequently shoots beyond 300 C during April May and sometimes falls to 00 C during winter in the higher hills. The Western Ghats present a combination of hydrological features which can be considered very rare (or even unique) in the tropics. Characteristics of rainfall the distribution over space and time and the intensity pattern, are among the primary reasons for this uniqueness. The rainfall in the region, despite being located in the tropics, is rather due to orographic lifting of South West Monsoon winds, than due to convective activities. Since the Sahyadri ranges are close to the coast and run nearly normal to the direction of the motion of moisture laden air from the vast Indian Ocean, rainfall in the region is very heavy, and a major portion of the Ghats falls under the wet category of climate classication (Strahler and Strahler, 1992). According to WMO (1983), the complete region comes under the meteorological division humid tropics, with no cyclonic activity. The distribution of Normal Annual Rainfall (NARF) over the Western Ghats is presented in Figure 3. It may be noted that there exists a systematic variation in rainfall across the region with a gradual increase in magnitude towards the crests and a more gradual decrease towards the plains in the east. The intra-annual distribution of rainfall varies over the region, depending on the cause 59

NNRMS

BULLETIN

DECEMBER

2013

secondary forest or some form of tree cover

of precipitation. The area below 150 N is inuenced by both South West and North East monsoons, North East monsoon being more effective deeper south, where tropical cyclones originating in the Bay of Bengal bring substantial rains. The inuence of South West Monsoon increases gradually towards north, parts of Kerala and the rest of the Ghats getting more than 90% of their rains during this season. The pre-monsoon rainfall that occurs in the months of April and May due to local convection is never widespread. A very important and interesting feature of rainfall in the Sahyadris is its intensity and duration. Throughout the length of the Ghats and the coast, daily intensities are very high. In areas with NARF above 5000 mm, daily rainfall exceeding 250 mm is quite a common feature and even where NARF is less than 2000 mm, falls exceeding 100 mm contribute about 25% of the annual rainfall. From a hydrological perspective, a more important feature of rainfall in the Western Ghats is its short duration intensity. On very heavy rainy days (>150 mm) in the region, it rains during 20-23 hours and the contribution of low intensity rainfall is always much greater (Tipperudrappa, 2009). Even in a very heavy rainfall area (NARF > 5000 mm) falls less than 6 mm/15 minutes (24 mm/h) last 95 % of the time and contribute 75% of the total rain. This is a feature of only the extra-tropical areas, but is found to characterise this tropical region also. Hence, the common belief that heavier rainfalls mean higher intensities does not hold water in this region.

S t re a m f l o w a n d Wa t e r Resources
The importance of the Sahyadris with regard to the water resources of South
Fig. 3: Distribution of normal annual rainfall (in mm) over the Western Ghats

India is evident from the fact that more than 60% of the surface waters in Karnataka

(Malhotra and Prasad, 1984), 70% in Kerala, 60% in Tamil Nadu, 50% in Maharashtra, 25% in Goa and 10% in Gujarat are derived from the Western Ghats. While the western slopes of the Ghats, below Tapi, drain into 10 Medium basins (> 200 km2, Rao, 1970) and 18 minor basins, the eastern slopes contribute water to the major basins of Godavari, Krishna and Kaveri and to a couple of minor basins. Figure 4 shows all the basins draining to the west and parts of the

fed by the Western Ghats. Table 1 shows the approximate quantities of yield contributed by the Western Ghats to the east owing rivers, illustrating the importance of the Sahyadris.
200N

Some of the important features of streamow in the Sahyadris are presented duration curves for daily ow within the
180N

monsoon months (June-Sept.), the variation of runoff coefcient for 10-daily ows in the season and typical hydrographs of ow from a very small catchment. These gures illustrate the very special characteristic of

160N

streamow in the region - the predominance of slow ow (commonly called baseow). This is in contradiction to the common belief that surface runoff dominates streamow
Legend

140N

Achancoil Ambika Bedthi Bharathapuzha Chalakudy Chaliyar Chandragiri Daman Ganga Kadalundi

in heavy rainfall areas. But, the fact is that it is the subsurface runoff which shapes the hydrographs in the region (Putty and Prasad, 2000a). The experience is that the term runoff needs to be dened differently in these areas if it is to be near reality, runoff should be dened to encompass all forms of ow in to the stream. However, this feature is not an exceptional characteristic of the Western Ghats, but has been found to be the case in many mountainous and forested catchments over the world (Ward and Robinson, 2011). In such areas, the runoff
Kilometers

120N

Kali Kaliada Karuvannur Mandovi Muvattupuzha Netravathi Pamba Par Periyar

100N

Purna Savitri Sharavathi Shastri Tadri Ulhas Vaitarna Valapattanam Vashishti

processes are explained by the Variable Source Areas theory (Hewlette and Hibbert, 1967) and the Jones Extended Variable Sources

80N 740E 760E 780E

Area theory (Jones, 1979). In the heavy rainfall areas of the Sahyadris, ow in natural pipes formed within the soil mantle forms

Fig. 4: River basins into which the Western Ghats drain

an important process contributing ow to the stream (Putty and Prasad, 2000b). Surface runoff due to rainfall intensities being greater than inltration rates is found only in limited areas with exposed rocks, in high lands and those with non-weathered laterites in the foot of the hills towards the coast and in areas where the human interference is too much. 61

NNRMS

BULLETIN

in Figure 5. These gures show the ow

DECEMBER

2013

major basins draining to the east and getting

Table 1: Contribution of the Western Ghat region to ow in the east owing rivers Basin area considered River Total yield (Mm )
3

Up to

Fraction of the total basin area (%) 15% 70% 40% 45% 55%

Total yield (Mm3)

Yield from WG region (Mm3) 5,000 28,000 5,000 6,500 8,000

Godavari Krishna (excluding TB) Tungabhadra Kaveri Kaveri

1,05,000 50,000 16,000 20,000

Nanded Raichur TB Dam Karnataka -TN Border Bhavani

11,000 37,000 12,000 15,000 18,000

Note: All gures in the table are indicative only

Remote sensing data is also being used in assessing runoff yield and quantities of soil loss from basins, using rainfall data and adopting watershed modeling techniques. Most of the models used for these purposes consider landuse and land-cover as a factor inuencing runoff. Further, assessment of catchment wetness conditions, ground water availability and river channel expansion are all parameters of hydrological importance, being accomplished with the help RS data. However, the application of the commonly used methods for such hydrological studies in the Western Ghat areas remains a challenging task because of two reasons (i) runoff processes in the region are different from those for which the commonly used models, like the Curve Number method, have been developed and (ii) non-availability of cloud free data during the rainy season in sufcient quantities. The fact that not much has yet been understood concerning the hydrology of the region provides ample scope for research in the areas of Western Ghats. Availability of microwave remote sensing and imageries of ner spatial and temporal resolutions may be considered to have opened up new vistas for hydrological research in the area.

Water Resources Development

The Western Ghats are obviously rich in water resources. The west owing rivers are not harnessed to the extent the east owing rivers are. However, the hydroelectric potential of the west owing rivers has been tapped to the extent possible, further development having been impeded by reasons of ecological degradation of the region. Figure 6 shows the reservoirs (both irrigation and hydroelectric) located in the region of the Ghats. These projects have led to destruction of natural vegetation to an alarming extent in Kerala, while in other states, awareness in time has prevented destruction to the point of no return. Open dug wells and springs are the other important water sources being extensively used for irrigation and drinking water purposes in the Sahyadri region, even though the Western Ghats have been identied as a region of very low groundwater potential, yielding less than 1 liter per second. Bore wells have made their entry in the recent past due to intensive irrigation patterns and lowering of water tables in the plateaus and the foothill regions.

Inter Basin Water Transfers

As mentioned earlier, there are some portions of the Sahyadris, the vast eastern slopes of which ultimately drain to the west, just because of some minor geological changes like faulting. It is often felt that ow in rivers from such areas can be diverted easily to the east, by technological interventions like damming, tunneling or lifting. Even gravity diversions, by means of garland (contour) canals, have been suggested. The common mans perception is that the yield from the basins draining to the west is being wasted, since it is not harnessed completely for consumptive uses, while the east draining basins adjacent to them have severe shortages. Hence, proposals for diverting ow from west owing rivers to the east are making rounds for a couple decades now. Even the National Water Development Authority, entrusted with the task of planning and implementing the National Water Grid, has proposed three inter-basin transfer schemes in the Western Ghats and feasibility studies are underway. On the other hand, inter-basin transfers already nd a very

map of Kerala and Tamil Nadu. There are four interstate water diversions between Kerala and Tamil Nadu, of which the Mullaperiyar and the Parambikulam - Aliyar Project (PAP) are the major ones. The Mullaperiyar project, in operation since the 1890s, transfers water from the Periyar basin in Kerala to the Vaigai basin in Tamil Nadu. The project was realised due to the political manoeuvring by the British and since Travancore was a Princely State of the British and not due to any riparian rights of Tamil Nadu. PAP diverts water from the Periyar, Bharathapuzha and Chalakudy river basins, since some parts of the upper catchments of these basins fall in the state of Tamil Nadu. Ravi et al., (2004), Madhusoodhanan and Sreeja (2010) and Sreeja et al., (2012) discussed the varied impacts of these diversions, including the challenges to river basin governance in the linked basins. Their ndings show that when not planned and implemented in true spirit and with a high order of fairness and justice, such projects not only lead to severe water shortages downstream but also to unending conicts and tensions. In fact, the experience of Kerala in more than 100 years of interlinking, fraught with continuous conicts and power tussles with Tamil Nadu,
Fig. 5: Characteristics of streamow in Western Ghats (i) Runoff coefcients during the monsoon; (ii) A typical ow duration curve of monsoon ow; and (iii) The hydrograph of ow in a small stream during a typical rainy day (rainfall amounting to about 100 mm)

can be an eye opener to the planners and a great deal of lessons can be learnt from it. Inter-basin transfer projects are always planned to divert only the so called excess

waters. It is argued with great credibility that if executed with restraint they would be a boon to the water starved plains of South India. However, once implemented with great difculty and with huge investments, people and the state nd many plausible reasons as to why the facilities created should be operated even in drier months, to draw water owing waste into the sea. The downstream effects of such drastic summer diversion can spell acute water scarcities and ecological disasters in the donor basin, which are conveniently forgotten in the analysis of benets to the recipient basin. This is one reason why the projects being planned in Karnataka are facing stiff opposition from the people in the coastal belt. The possibility of small-scale diversions that do not infringe on the needs and the riparian rights of the downstream users and the ecological needs of the river, which are always overlooked, can only be realised through a basin-wise planning effort that would consider the requirements of all the sections of the stake holders including the river itself. 63

NNRMS

BULLETIN

DECEMBER

2013

prominent position in the water resources

Water Scarcities Solutions

and

The average rainfall of the region of Western Ghats, throughout their extent, exceeds 2500 mm and the amount of ground water recharge is also very high in most parts of the Ghats. Hence, possibilities of water scarcity must have been low. However, demands for domestic needs, energy, irrigation and industrial purposes have been high in the downstream areas of the Western Ghats. Most of the rivers in the Sahyadris, except in Karnataka, are already either dammed or diverted at several sites for power generation in the upper reaches and irrigation in the lower reaches, affecting the natural flow regimes and downstream regions. In addition, east owing rivers like Krishna and Kaveri barely reach the seas due to over abstraction (Molle, et. al., 2010; Venot, et. al., 2011), and demands for river diversions are mounting. Many west draining basins are also fast closing and the impacts are felt on delta shing, farming, livelihoods and ecology (WGEEP, 2011). The aquifers in the coastal belt, being lateritic, drain out fast during the post monsoon season and as a result, sea water intrusion into the fresh water aquifers is creating serious problems. Indiscreet diversions and injudicious use of land and water for agriculture have left Kerala a water starved state. The situation in Maharashtra is not much different the ever growing urban domestic and industrial demands here have been responsible for the need to resort to countless reservoirs storing the ow from the Ghats. Yet, acute scarcities of water are conned to the summer months, while shortages in immediate post monsoon season are mounting. It is commonly agreed that measures like better management of water and implementation of conservation practices in the coastal areas should help alleviate scarcities in the winter season, while more stringent measures, presently not visible as a possibility, may be necessary to wade through the
Fig. 6: Reservoirs in the Western Ghats

summer months.

irrigation is practiced, in India, include illegal diversion and violation of indicative cropping patterns. Unscientic methods of conveyance and rampant misuse of water add to these factors. It is true that developments over the world during the last two decades, in the backdrop of the threat of a global climate change that is detrimental to the very sustenance of life, have resulted in emergence of lobby groups spreading environmental awareness and promoting sustained development even in the water sector. But, not much seems to have happened in the Sahyadri region. One of the possible reasons for such a state is the dearth of proper documentation and assessment of the evil practices and the damage being caused as a result. The extent and the type of the terrain of the area seem to be a major hurdle in mobilising a movement against forces responsible for the mess. The same is the case also with the unabated expansion of tea, coffee and rubber plantations and illegal logging for timber and rewood in those parts of the Sahyadris, which still remain fairly undisturbed. These activities have often gone unnoticed because they are executed deep inside private or encroached land. It is in these sectors, which call for urgent attention, that modern technologies of geo-information systems, including remote sensing, do come in handy for the planners and the concerned groups of enlightened citizens. Remote sensing technologies furnish the tools required for acquiring real time data from vast areas, for better management of resources, but it remains a big question as to how these are made use of and how useful the knowledge gained would be in combating the ever increasing needs.

Conclusion
The remnant natural ecosystems of the Western Ghats are currently subject to a plethora of threats that vary widely from localised threats such as quarrying, livestock grazing, and forest res to landscape-level threats such as mining, hydel projects, large-scale agricultural expansion and creation of monoculture plantations. Each of these must have directly or indirectly impacted upon the availability of water resources and led to an accelerated impoverishment of this ancient landscape. Though there are numerous studies related to the biodiversity and its conservation in these hill tracts, research into the hydro-climatological aspects of the Ghats that ultimately sustain this unique ecosystem has been negligible. Further, the available knowledge of the hydrological processes that characterise this region establishes its uniqueness and underline the need for further studies. This paper calls for an urgent need to take up detailed investigations and lists the possible research areas in hydrology and water resources pertaining to the Western Ghats, where modern tools of geo-spatial studies could be adopted with advantage.

References
Challa, O., Gajbhiye, K.S. and Velayutham, M. (1999). Soil series of Maharashtra, NBSS Publication 79, National Bureau of Soil Survey and Land Use Planning, Nagpur, Maharashtra, 428 p. Cincotta, R. P., Wisnewski, J. and Engelman, R. (2000). Human population in the biodiversity hotspots, Nature, 404, 990-992. Daniels, R.J.R. (2001). National Biodiversity Strategy and Action Plan: Western Ghats Eco-region. Report submitted to Ministry of Environment and Forests, Government of India, New Delhi. Harindranath, C.S., Venugopal, K.R., Raghumohan, N.G., Sehgal, J. and Velayutham, M. (1999). Soils of Goa for optimizing land use: Executive summary, NBSS Publication 74b (Soils of India series), National Bureau of Soil Survey and Land Use Planning, Nagpur, Maharashtra, 131 p. Hewlett, J.D. and Hibbert, R.A. (1967). Factors affecting the response of small watersheds to precipitation in humid areas. In: Proc. Inter. Symp. on Forest Hydrology. Pennsylvania State Univ., 1965. Pergamom Press. N.Y., 275-289.

65

NNRMS

BULLETIN

DECEMBER

2013

It is commonly agreed that factors contributing to shortage of water in areas where large scale

IIRS. (2002). Biodiversity Characterization at Landscape Level in Western Ghats India using Satellite Remote Sensing and Geographic Information Systems, Indian Institute of Remote Sensing, Indian Institute of Remote Sensing, Dept. of Space, Govt. of India, Dehradun. Jarvis, A., Reuter, H.I., Nelson, A., Guevara., E. (2008). Hole-lled seamless SRTM data V3, International Centre for Tropical Agriculture (CIAT), Colombia, available at http://srtm.csi.cgiar.org. Jones, J.A.A. (1979). Extending the Hewlette model of stream runoff generation. Area (Inst. of British Geographers), 11(2):110-114. Madhusoodhanan, C.G. and Sreeja, K.G. (2010). The Mullaperiyar Conict. NIAS Backgrounder on Conict resolution, B4-2010, Bangalore: National Institute of Advanced Studies. Malhotra, K. and Prasad, R. (1984). Some important aspects of water resources in Karnataka.Pub.Dept. Civil.Engg.,IISc., Bengaluru, India. Molle, F., Wester, P. and Hirsch, P. (2010). River basin closure: Processes, implications and responses, Agricultural Water Management, 97: 569-577. Myers, N, Mittermeier, R.A., Mittermeier C. G., da Fonseca, G. A. B. and Kent, J. (2000). Biodiversity hotspots for conservation priorities, Nature, 403, 853-858. Nair, S.C.(1991). Southern Western Ghats. INTACH, New Delhi. NBSS&LUP. (1996) (a). Kerala Soils.National Bureau of Soil Survey and Land Use Planning, Nagpur. NBSS&LUP. (1996) (b).Tamil Nadu Soils. National Bureau of Soil Survey and Land Use Planning, Nagpur. Pascal, J.P. (1988). Wet evergreen forests of the Western Ghats of India: Ecology, structure, oristic composition and succession, French Institute of Pondicherry (FIP), Pondicherry, India. Putty, M.R.Y. and Prasad, R. (2000a). Understanding runoff processes using a watershed modela case study in the Western Ghats in South India, J. Hydrol., 228:215-227. Putty, M.R.Y. and Prasad, R. (2000b). Runoff processes in headwater catchments An experimental study in Western Ghats, South India, J. Hydrol., 235:63-71. Radhakrishna, B.P. (1991). An excursion into the past the Deccan volcanic episode, Curr. Sci., 61 (9&10): 641647. Rao, K.L. (1970). Indias Water Wealth. Orient Longman Ltd. New Delhi. Ravi, S.P., Madhusoodhanan C.G., Latha, A., Unnikrishnan, S. and AmitaBachan, K.H. (2004). Tragedy of commons: The Kerala experience in river linking. New Delhi: SANDRP and Thrissur: RRC. Sivaprasad, C.R., Reddy, R.S., Saigal, J. and Velayutham, M. (1998). Soil resources of Karnataka for Land Use Planning, Draft Final Copy, National Bureau of Soil Survey and Land Use Planning, Nagpur, Maharashtra. Sreeja, K.G., Madhusoodhanan, C.G., Shetty, P.K., and Eldho, T.I. (2012). Inclusive spaces in Integrated River Basin Management: discerning multiple boundaries of resource relations. International Journal of River Basin Management, 10(4), 351-367. Strahler, A.H. and Strahler, A.N. (1992). Modern Physical Geography.John Wiley & Sons, Inc. NY. Thipperudrapa. (2009). Some studies on the intensity pattern of rainfall in Karnataka, with particular reference to Wetern Ghats. Unpub.Ph.D thesis, The National Inst. of Engg.,Mysuru, 570008. UNESCO, (2012). Decisions adopted by the World Heritage Committee at its 36th session, Available from: http://whc. unesco.org/en/sessions/36COM/ [Accessed 16 August 2012]. Venot, J.P., Bharati, L., Giordano M. and Molle, F. (2011). Beyond water, beyond boundaries: spaces of water management in the Krishna river basin, South India. The Geographical Journal, 177 (2), 160-170. Ward, R.C. and Robinson, M. 2011. Principles of Hydrology, 4thEdition, Tata McGrawHill, New Delhi. WGEEP, (2011). Report of the Western Ghats Ecology Expert Panel, Submitted to the Ministry of Environment and Forests, Government of India, New Delhi. WMO. (1983). (World Meteorological Organization).Operational Hydrology in the Humid Tropical Regions. In: Hydrol. of Humid Tropical regions with Particular reference to the Hydrological effects of Agriculture and Forestry practices, IAHS. Publ. No: 140: 1-25.

Use of Earth Observation Data to Unearth Sub-surface Drainages: Potential Groundwater Source in Arid Region of North West India
Bhadra BK and Sharma JR RRSC-W, NRSC, ISRO, Dept. of Space, CAZRI Campus, Jodhpur 342003, India Email: bkbhadra63@gmail.com

Introduction
In the ancient literature like Rigveda, the river Saraswati is described as the Ambitame, Naditame and Devitame ie., the best of mother, the best of river and the best of goddess. The Vedic Saraswati, a mighty and holy river of Northwest India during 6000 B.C. stretched for 1600 km long and 3-12 km wide through Punjab, Haryana and Rajasthan and nally discharged into the Rann of Kutchch in Gujarat coast and disappeared around 3000 B.C. (Chauhan, 1999; Kalyanraman, 1999; Radhakrishna, 1999, Valdiya, 2002; Lal, 2009). It was a life line of the people of ancient India, mostly in the Vedic and Puranic ages. The discovery of sites of Harappan civilization on the banks of Saraswati also indicates that the river was mighty more than 8000 years ago. Two perennial rivers of the Himalaya viz. Yamuna and Satluj might be the feeder drainage of the `Lost River Saraswati in the past (Figure 1). But due to eastward shifting of Yamuna and westward deection of Sutlej, the Saraswati nally dried off. Compounding with the neo-tectonic activity and climatic changes in North West India, the river Saraswati nally got lost under the aeolian sand cover. Several remnants of this river still exist as palaeochannels in different parts of North West India. To unearth the sub-surface drainage through modern tools and its great signicance as Indian cultural heritage, study of River Saraswati becomes a challenging task among the Geologists and Archeologists.

Remote Sensing Studies by Different Agencies

Over the past 30-35 years, taking advantage of the aerial and satellite remote sensing data, palaeochannels have been systematically mapped by the agencies working in this region mainly Central Arid Zone Research Institute (CAZRI), Jodhpur, Geological Survey of India (GSI), Jaipur, Space Application Center (SAC/ISRO), Ahmedabad and Regional Remote Sensing Center (RRSC-W/ISRO), Jodhpur to discover the course of lost Saraswati river. Most of the earlier studies have used aerial photographs and satellite images of low resolution from the Landsat MSS and IRS 1A/1B satellites to map the palaeo drainage courses in Northwest India. Mostly hardcopy in Black & White and colour prints of the satellite images have been used in these studies. Probably, Ghosh et al., (1979, 1980) of CAZRI, Yash Pal et al. (1980) of SAC/ISRO and Bakliwal and Sharma (1980) of GSI were the pioneering workers in satellite base interpretation of Saraswati palaeochannels. Based on the image interpretations, varying number of courses of river Saraswati have been suggested by these workers. Due to the lack of clear image signatures in several areas, the continuity of the river courses could not be maintained. Hence, there are discrepancies in delineating the number of Saraswati

NNRMS

BULLETIN

DECEMBER

2013

courses e.g. 5 courses by Ghosh et al., (1979, 1980); 7 courses by Bakliwal and Grover (1988) and single course by Yash Pal et al., (1980). All the three organization continued their palaeo-drainage work till 1990s with the available Landsat images and aerial photographs (Sood and Sahai, 1983; Kar and Ghosh, 1984; Bakliwal and Grover, 1988). The pace of remote sensing work gained momentum during 1991-2000 with the availability of new optical sensors as well as microwave sensors. It has been observed that microwave data has advantageous over optical data in delineating buried channels below the sand cover due to having penetration capability (Ramasamy et al.,
Fig. 1: Existence of Saraswati River System(?) between Indus and Ganges River System in NW India

1991; Rajawat et al., 1999; Kar, 1999; Sharma et al., 1999; Sahai 1999). Further, it became necessary to map afresh of the palaeodrainage network of the Saraswati river basin due the advancements in sensor resolution, image processing and GIS techniques. Taking advantages of the new data and methods, re-interpretation of satellite images was carried out by RRSC-W, Jodhpur during 20002010 to arrive at the actual visible course of river Saraswati and its tributaries (Gupta et al., 2004; Bhadra et al., 2005; Bhadra et al., 2009). Image interpretation was done bit by bit to trace the entire palaeochannels of River Saraswati, covering Punjab, Haryana, Rajasthan and Gujarat States. Finally, the entire palaeochannel network in Northwest India were reconstructed by integrating ground data such as archaeological sites, historical maps, hydrogeology, sediment character of aquifer formation through drilling data, geochronology of trapped water and sand samples (Bhadra and Sharma, 2011). Delineation of Sutlej palaeochannel and the old course of River Drishadvati helped in linking the present day `Lost River Saraswati with the Himalayan perennial sources (Sharma and Bhadra, 2012).

Fig. 2: SRTM derived DEM (90m) showing gradual change in slope along the major drainage systems of Northwest India

Physiography and drainage pattern of North West India is depicted in Figure 2 through Digital Elevation Model (DEM) from the SRTM data (90m resolution) for the year 2000. The images indicate that these channels served as passage for huge volumes of water that rushed down from the Himalayas, made their way into the ocean. The ood plains of the Saraswati River slope towards South West direction from Siwalik foothills (elevation > 270m) to the low lying marshy stretch of the Great Rann of Kutchch. Beyond the sandy expanse of the Thar Desert in Cholistan (Bahawalpur) in Pakistan, there exists the ood plain (500km long and 16-24km wide) of the dry channel of Hakra River. The ood plain was built on the sunken part of the Indian crust known as the Himalayan foredeep. Huge piles of detritus were deposited by the rivers originating from the Himalayas and some part was also contributed by the Aravalli hills.

The present study has been carried out to map the palaeochannels in the Northwestern region and establish the course of Vedic river Saraswati. Due to large study area, the delineation work of the palaeochannels is done in two phases
Fig. 3: IRS-1A LISS-I image of Aug. 1990 showing the signatures of palaeochannels (dark tone) in Thar Desert

viz. Phase-I study in western Rajasthan and Northern Gujarat during 1995-2000 and Phase-II study in Haryana, Punjab and parts of Himachal Pradesh and Uttarakhand during 2005-10. Image processing techniques such as Piece-wise Histogram Stretching and contrast enhancement have been applied in both the phases. Further, SRTM DEM is used to show the topography and present day drainage system in North-west India. (a) Phase-I Study in Rajasthan and Gujarat Initially, digital mosaic image of IRS 1A LISS-I data of August, 1990 is generated for detailed study of palaeochannels in the Northwestern region (Figure 3).The linear stretching is performed on the mosaiced LISS-I FCC image. Palaeochannels are basically the old course of river channels which appears on the satellite image as serpentine drainage course with high moisture content (dark tone). The composited image showed the course of palaeochannels (dark signature) very clearly with naked eye. Further, satellite

Fig. 4: (IRS P3 WiFS) of Dec. 1999 showing the palaeochannel network of Saraswati River in parts of Rajasthan and Gujarat

69

NNRMS

Methodology in Delineation of Palaeochannel

BULLETIN

DECEMBER

2013

Dem Study for Major Drainages in North West India

data from IRS P3 WiFS sensor of December 12, 1999 has been used for mapping of the prominent palaeochannels (Figure 4), as it may be easy to locate broad channels (minimum ~200m wide) on the ground for their exploration. The large swath of WiFS data of December, 1999 has made the mapping exercise more reliable and easier as the palaeochannels are mapped from the data with same radiometric and geometric conditions over the entire study area. This has facilitated avoiding the problems of radiometric and geometric corrections of large number of scenes. Piece-wise Histogram Stretching technique has been used to enhance the palaeochannel signatures on the image. This technique has been found unique in enhancing digital images to discriminate the moist sand (dark tone) in the sandy alluvium tract. The feature enhancement is carried out by way of loading sub-scenes on computer terminal screen in full resolution and improving the feature contrast by histogram stretching interactively. Palaeochannels have been mapped online from the enhanced digital data using the vector module of EASI/PACE image processing software. At the same time, IRS 1C LISS-III, data of November 1, 1996 and PAN scene of same date of pass have been used for selective areas in parts of Jaisalmer district. For this purpose, digital image processing techniques like hybrid product generation (merged product of LISS-III and Pan) have been used to enhance the micro-geomorphological features indicating palaeochannels. Thus, the entire course of palaeochannels along Indo-Pakistan border of Rajasthan State has been delineated. These palaeochannels have been designated as parts of `Lost Vedic Saraswati and linked to present day Ghaggar River in RajasthanHaryana Border States. (b) Phase-II Study in Punjab and In Phase-II study, the work is
Fig. 5: Satellite image showing the delineated palaeochannels using IRS P6 AWiFS and LISS-III images (Feb. 2004) in northern districts of Haryana. Inset shows highly moist zones with dark tone in Hisar and Sirsa districts

Haryana extended in upstream direction by further delineation of palaeochannels of Ghaggar, Saraswati, Drishadvati and Sutlej Rivers in Haryana and Punjab. For this purpose, multiresolution data from IRS P6 AWiFS, LISS-III, LISS-IV of (February, 2004), Landsat ETM data and Radarsat-1 SAR data (100m resolution) has been used in the study. For delineation of palaeochannels, digital image processing techniques like histogram equalization, linear stretching, contrast and brightness enhancement etc. have been applied on a small area of the satellite images (IRS P6 LISS-III). Drainage features are highlighted on applying local stretching on ~10X10 km2 area out of the full LISS-III scene (141X141 km2). In this process, palaeochannels are delineated with proper care by avoiding the canals, existing ephemeral drainages and water logged areas. The delineated course of palaeochannels pass through Kurukshetra, Kaithal, Jind, Fatehbad, Hisar and Sirsa districts of northern Haryana from east to west (Figure 5). A large number of discontinuous palaeochannels are found to lie in the above districts. In the northeast of Kurukshetra, these palaeochannels could possibly have a link with the existing / abandoned channels which originate from the Siwalik Hills. In the central Haryana, a sub-parallel drainage to Saraswati is also marked on the satellite image that passes through Karnal, Jind, and Hisar districts and is known as Drishadvati River. Presently, Western Yamuna Canal (WYC) and Hansi Branch Canal is constructed all along the natural depression of Drishadvati River. In the west of Hisar, Draishadvati River is passing through Anupgarh and joins with Ghaggar River near Kalibangan in Hanumangarh district of Rajasthan.

Punjab. For this purpose, satellite images of IRS P6 LISS-III data of February, 2004 and Radarsat SAR data (50m resolution) of December, 2002 have been used to delineate palaeochannels (Figure 6). In this area, Radarsat SAR image was used to map moist zone (torquous channel with dark tone), possibly indicating a subsurface palaeochannel. The delineated palaeochannel between Ropar and Patiala is named as Sutlej palaeochannel which is a North-South trending tortuous palaeochannel extending for a length of about 75 km with having width between ~1 to 6 km. The Sutlej palaeochannel is connecting the present day Sutlej River and the Ghaggar River in the south of Patiala. Thus, the Sutlej, Saraswati and Drishadvati palaeochannels possibly contributed to the main Vedic Saraswati to form a mighty owing River in the past.

Ground Validation as Supporting Evidence

Mapped courses of Saraswati River within Indian Territory have been validated through data from a variety of ground investigations viz. historic maps, geomorphic anomaly, archeological sites, core drilling / litholog data, ground water quality-yield-depth, age of ground water etc. Study of a number of historical maps of Indo-Pakistan region of Mughal period, prepared by the Italians, Dutch and British authors during 14th 17th Century A.D., show the presence of the dry channels of Saraswati. The merged satellite image (LISS-III+Pan) are used to identify corn-cob structures which are formed due to deposition of sand over owing as well as stagnant water bodies. They indicate the presence of uvial activity in this region in the past. Compilation of all the available archaeological sites of Early Harappan, Mature Harappan and Late Harappan sites in northern states of India indicates 436 sites in Haryana, 157 sites in Punjab, 100 sites in Gujarat and only 36 sites in Rajasthan. The association of Harappan sites and the delineated palaeochannels points towards a close link with the Vedic Saraswati civilization. Similarly, analysis of drilling data, groundwater quality and its isotopic ages support the existence of palaeochannels in the entire region of North-west India. Source of Groundwater along the Palaeochannels Using the delineated palaeochannels maps as a guide, 14 tube wells have been drilled by Ground Water Department, Rajasthan jointly with the Central Ground Water Board (CGWB) along TanotKishangarhLongewala-Ghotaru section in Jaisalmer district of Rajasthan (Figure 6). Observation of core drilled samples (Table-1) suggests that the palaeochannels consist of alternating layers of fine to medium and coarse grained sand and sometimes have gravel columns, indicating presence of uvial regime. The column thickness of palaeochannels in this area ranges from 35 to 80 m. Coarser sediments are noticed at a depth ranging from 40-125 m and have been encountered in 9 out of the 14 tube wells. Water levels in most of the bore wells drilled ranged in between 35-60 m. Similarly, litholog data analysis of 304 drilling
Fig. 6: Optical (IRS P6 LISS-III) and SAR (Radarsat-1) images showing the delineated Sutlej palaeochannels between Ropar and Shatrana in Punjab. A large number of archaeological sites can be noted lying along the palaeochannels.

sites (Source: Ground Water Cell, Govt. of Haryana, 2005-06) in 9 districts of northern Haryana indicates that most of the lithologs, 71

NNRMS

BULLETIN

DECEMBER

2013

An attempt has been made to identify the continuity of Saraswati palaeochannels in eastern

lie near the palaeochannels, have medium to coarse grained sands and associated gravel and pebbles at a depth between 10 to 100 m. Analysis of ground water samples indicates that the water quality is quite good (potable water) for most of the drilled tube wells, as compared to tube wells or dug wells away from the channel. For example, the well drilled at Tanot shows Total Disseled Solids (TDS) in the order of 2650 ppm in comparison to the existing well at Tanot (TDS~9000 ppm), which is about 1.5 km away from the channel. The isotopic age analysis (H3, O18 and C14) by BARC, Mumbai of 17 ground water samples from the existing wells along the palaeochannels in Jaisalmer district of Rajasthan shows ground water ages from 1340 to 18880 Before Present (BP) at different localities from NE to SW (Rao and Kulkarni, 1997; Nair et al., 1999).Ground Water Department, Jodhpur has identied three major aquifer zones along palaeochannels of Saraswati River. Static groundwater reserve in these aquifers is estimated as 590 Mm3 (A1 zone), 92 Mm3 (A2 zone) and 81 Mm3 (A3 zone) respectively. Table 1: Data of the tube wells drilled along the palaeochannels in the Jaisalmer district, Rajasthan (Source: Ground Water Dept, Govt. of Rajasthan:Adopted from Gupta et al, 2004) Sl. No 1. 2. 3. 4. 5. 6. 7. Drilled well name / location Tanot (3.5 km from Ghantiyali to Tanot) Ghotaru-I (12.5 km Ghotaru to Longewala) Ghotaru-II (14.5 km Ghotaru to Longewala) Ghotaru-III (10 km from Ghotaru to Longewala) Ghotaru-IV (3 km from Ghotaru to Longewala) Ghotaru-V 150 m NE of Fort Ghotaru-VI (1.5 km from Ghotaru to Asutar) Dharmi Khu (3 km from Kishengarh to Dharmi Khu) Ranau-I* (RanauTanot Road) Ranau-II* (Close to Ranau Village) Karthai* (9.5 km from Ranau on Tanot Road) Nathura Kua* (4.5 km from Tanot) Kuria Beri* Ghantiyali I* (500 m from Ghantiyali Mandir to Tanot) Yield (lph) 11250** 13500** Not Developed 2250** 32400# 33750# 22500# Quality (TDS) 2650 6506 151 4337 3554 1536 934 151 151 148 125 48 45 33 46 Depth Drilled (m) 125 151 Static Water Aquifer material as observed Level (m) in lithologs 33 43 Mainly ne grained sand, medium grained at certain levels Medium to coarse sand and gravel (Out of main channel) Fine Grained sand / S.St kankar (Out of main channel) Fine Grained sand and very coarse grained gravelly sand Medium to ne and coarse grained sands Coarse gravelly and ne to medium grained sands, occasional clayey Dominantly medium to coarse sands, ne grained and clayey sands Fine and medium grained sands Fine grained sand and silt with kankar, ne to medium sand N.A. Mostly Fine sand medium at certain levels Fine grained sand and silt with kankar Mostly ne Sand Fine grained sand

8. 9. 10. 11. 12. 13. 14.

35100# 9120** 18240** 12312* 12768** 12768** 11400**

1024 1010 1000 1800 2656 1295 2200

153 102 120 125 120 131 130

40 42 58 42 36 32 62

*Wells are drilled by the CGWB, **Compressor yield # Pump test yield

Linkage of Vedic Saraswati with Himalayan Sources

The then mighty river, the Saraswati in Vedic Period might have been contributed by any major river system of the Himalaya. Presently, Sutlej and Yamuna are the two perennial rivers which are likely to be the feeder channels of Vedic Saraswati River in the past (Figure 7). Beyond the range of Siwalik and lesser Himalaya, these two rivers are fed with the permanent glaciers in the higher Himalaya. It has been observed from the satellite images that the size of the glacier of Sutlej River is much larger than the size of the Yamunotri and Bandarpunch glaciers. But, due to tectonic

shift their courses viz. Sutlej to the west to join river Indus and Yamuna to east to join River Ganges near Allahabad. Based on the analysis of several satellite images the entire course of Saraswati palaeochannels has been delineated from the Himalayan foothills to the Rann of Kutchch, West India. Review of different literatures, archaeological findings and synthesized scientic evidences, two possible connectivity
Fig. 7: Mosaiced Landsat ETM+ image of 1999-2000 shows the Saraswati palaeochannel network and their link with the perennial sources of Sutlej and Yamuna Rivers

River sources have been emerged viz. (a) Connectivity of Vedic Saraswati with Sutlej

River (b) Connectivity of Vedic Saraswati with Yamuna/Tons Rivers.

Acknowledgements
The authors are extremely grateful to Dr. V. K. Dadhwal, Director, NRSC, Hyderabad for his valuable guidance to carry out this research work. The authors are equally grateful to Dr. A. K. Gupta, Ex. ISRO Scientist, Jodhpur, Dr. S. Kalyanraman of Saraswati Nadi Sodh Prakalp, Chennai; Shri Darshan Lal Jain of Saraswati Nadi Sodh Sansthan, Yamunanagar; Late Dr. S P Gupta, Indian Archaeological Society, New Delhi; Dr. Amol Kar, CAZRI, Jodhpur and Shri Rajesh Purohit, Archaeologist, Director, Allahabad Museum and others for their active association, encouragement and fruitful discussion at various stages of the project work.

References
Bakliwal, P.C. and Grover, A.K. (1988). Signature and migration of Saraswati River in Thar Desert, Western India, Records Geol. Surv.India, V.116, Pt.3-8, pp.77-86. Bakliwal, P.C. and Sharma, S.B. (1980). On the migration of river Yamuna.J. Geol. Soc. India, V.21, pp.461-463. Bhadra, B.K., Sharma, J.R. and Bharadwaj, Trilok (2005). Delineation of Sutlej Palaeochannel using IRS LISS-III and Radarsat SAR data in parts of Punjab: Its possible linkage with Lost River Saraswati in Northwest India. Published Abstract in Indian Geological Congress (IGC), National Conference at Delhi University during 2-4 December, 2005, pp.8-9. Bhadra, B.K., Gupta, A.K. and Sharma, J.R. (2009). Saraswati Nadi in Haryana and its linkage with the Vedic Saraswati River Integrated study based on satellite images and ground based information. J. Geol. Soc. India (Springer Co-Publ.), V.73, pp.273-288. Bhadra, B.K. and Sharma, J.R. (2011). Satellite Images as Scientic Tool for Saraswati Palaeochannel and its Archaeological Afnity in NW India. In: Proceedings of Int. Seminar `How Deep are the Roots of Indian Civilization by Draupadi Trust & Ministry of Culture, New Delhi, pp.30-49. Chauhan, D.S. (1999). Mythological observations and scientic evaluation of the Lost Saraswati River. In: Vedic Saraswati (Eds. B.P. Radhakrishna and S.S.Merh). MemoirGeol. Soc. India, V.42, pp.35-45. 73

NNRMS

of the Vedic Saraswati with the Himalayan

BULLETIN

passing through the Thar Desert in North

DECEMBER

2013

changes in the past, these two perennial rivers

Ghosh, B., Kar, A. and Hussain, Z. (1979). The lost courses of the Saraswati River in the Great Indian Desert: new evidence from LANDSAT imagery. Geograph. Jour., London, V.145(3); pp.446-451. Ghosh B, Kar A and Hussain, Z. (1980). Comparative role of Aravalli and Himalayan river systems in the uvial sedimentation of Rajasthan desert. Man and Environment, V. 4, pp.8-12. Gupta. A.K., Sharma, J.R. Sreenivasan, G. and Srivastava, K.S. (2004). New ndings on the course of River Saraswati. Jour. Ind. Soc. Remote Sensing, V.32 (1), pp.1-24. Kalyanraman, S. (1999). Saraswati River, Goddess and Civilization. In: Vedic Saraswati (Eds. B.P. Radhakrishna and S.S.Merh). MemoirGeol. Soc. India, V.42, pp.25-34. Kar, A. (1999). A hitherto unknown palaeodrainage system from the Radar Imagery of southeastern Thar Desert and its signicance.Memoir Geol. Soc. India, No.42, pp.229-235. Kar, A. and Ghose, B. (1984). The Drishadvati River System of India: An assessment and new Findings. The Geographical Journal, V.150 No.2, pp. 221-229. Lal, B. B. (2009). How deep are the Roots of Indian Civilisation? Archaeology Answers. Aryan Books International, New Delhi. Nair, A.S., Navada, S.V. and Rao, S.M. (1999). Isotope study to investigate the origin and the age of ground water along the palaeochannels in Jaisalmer and Ganganagar districts of Rajasthan. In: Vedic Saraswati (Eds.: B.P. Radhakrishna and S.S. Merh). Memoir Geol. Soc. India, No.42, pp. 315-321. Radhakrishna. (1999). Vedic Saraswati and the dawn of Indian Civilization. In: Vedic Saraswati (Eds. B.P. Radhakrishna and S.S.Merh). Memoir.Geol. Soc. India, V.42, pp.5-13. Rajawat A.S., Narain A., Navalgund, R.R., Pathak S., Sharma, J.R., Soni V., Babel, M.K., Srivastava, K.S. and Sharma, D.C. (1999). Potentials of Radar (RES-1/2 SAR) and high Resolution IRS 1-C Data in reconstructing Palaeodrainage Network of western Rajasthan. Momoir Geo. Soc. India, No.42, pp.245-258. Ramasamy, S.M., Bakliwal, P.C. and Verma, R.P. (1991). Remote Sensing and river migration in western India.Int. JRemote Sensing, V.12, pp.2597-2609. Rao, S.M. and Kulkarni, K.M. (1997). Isotope hydrology studies on water resources in western Rajasthan. Curr. Sci., V.72 (1), pp. 55-61. Sahai Baldev. (1999). Unraveling the Lost Vedic Saraswati. In: Vedic Saraswati (Eds.: B.P. Radhakrishna and S.S. Merh). Memoir Geol. Soc. of India, No.42, pp.121-142. Sharma, J.R. and Bhadra, B.K. (2012). Signatures of Palaeo Rivers Network in Northwest India Using Satellite Remote Sensing. In: Historicity of Vedic and Ramayan Eras: Scientic Evidence from the Depths of Oceans to the Heights of Skies, Book Eds. - SarojBala and Kulbhusan Mishra, Publ. Institute of Scientic Research on Vedas (I-SERVE), New Delhi, pp.171-192. Sharma J.R, Gupta A.K., Pathak S., Rajawat, A.S., Sharma D.C., Srivastava, K.S. and SoniVimal. (1999). Results of Remote Sensing based studies on Reconstruction of the course of River Saraswati in Western India. In: Proceeding of ISRS Symposium, Bangalore, pp.378-384. Sood, R. K. and Sahai, B. (1983). Hydrographic changes in northwestern India. Man and Environment, V.7, pp.166-169. Valdiya, K.S. (2002). Saraswati The River That Disappeared. University Press (India), Hyderabad, 116p. Yash Pal, Sahai, B., Sood, R.K. and Agarwal, D.P. (1980). Remote Sensing of the `Lost Saraswati River. Proc. Ind. Acad. Science (Earth and Planetary Sciences), V.89, pp.317-331.

Introduction
Severe water shortages have already occurred in several parts of the world and this situation is worsening each year. Freshwater withdrawals increase with increasing population and increasing per capita needs and both ground water aquifers and low-season stream ows are experiencing depletion. This water crisis will adversely affect water supplies for irrigated agriculture, drinking, industries and other domestic uses. Forested catchments supply a high proportion of the water for various purposes in both upstream and downstream areas. There is an urgent need to understand the interactions between forests and water. Of late studies are carried out on large-scale watershed hydrology, climate change impacts, and application of hydrological models. But, the forest hydrology in India is still in an initial stage. The information to dene the entire hydrological system and water budget of a particular forest system is not yet available. Forests management practices have the potential to alter the quantity, quality and timing of water moving through aquifers by altering the interception, evapotranspiration, soil inltration, base ow, runoff etc. (Anderson et al., 1976; Ice and Stednick, 2004). Forest hydrology deals with the role of forest over precipitation and water yield production potential of the forest. Therefore, forest inuences on various hydrological parameters viz. rainfall, interception, inltration, soil moisture, evapotranspiration, groundwater, water yield, soil loss and oods etc., forms an important area of hydrological studies. Some of the earlier studies (Mehar Homji 1986; Dutt and Manikiam, 1987; Gupta et al., 2005) indicated that forests and rainfall relationship are not linear on a regional scale. Interception studies carried out in India (Dabral and Subbarao, 1969; Mathur, 1975) indicate that the canopy interception varies from 15% to 35% of rainfall in the forest regions. There is evidence that interception varies not only with canopy density but also with intensity of rainfall. The analysis of inltration data from small forests and agriculture watershed in Doon valley indicated that the rate of inltration was twice in forest watershed (Shorea Robusta) as compared to agriculture watershed (Dhruvanayayana and Shastri 1983). It can be depicted that the inltration rates are relatively more in forested soils as compared to agricultural areas & grasslands. Much effort has not been made to quantify soil moisture storages under forests. However, forested soils have a better soil moisture retention capacity due to improved soil structure because of more humus and organic matter content. In general, forests have high evapotranspiration requirement as compared to other land uses. Groundwater relationship with forest is yet to be examined scientically on large scales. Studies conducted in India and abroad

NNRMS

BULLETIN

Gupta PK1, Singh RP1, Panigrahy S2, Chauhan JS3, Sonakia A3 and Parihar JS1 1 Environment and Hydrology Division, Space Applications Centre, Ahmedabad-380015 2 Former Group Director, Agriculture, Terrestrial Biosphere and Hydrology Group, Space Applications Centre, Ahmedabad-380015 3 Kanha National park, MP Forest Department, Mandla, MP India Email: pkgupta@sac.isro.gov.in

DECEMBER

MODELING HYDROLOGICAL WATER BALANCE IN THE FORESTED WATERSHED FOR WATER MANAGEMENT

2013

(Hibbert, 1965; Lal and Subba Rao, 1981) have shown increase in stream ow due to cutting and reduction of density of forests. Empirical modeling approach to quantify the accumulated hydrologic effects of watershed management is limited due to its complex nature of soil and water conservation practices. This paper describes a study under MIKE SHE/ MIKE 11 coupled model. The study has been done to estimate the water balance components in the forest and neighbouring regions for the conservation and management of water.

Data used
Three types of data used were such as remote sensing, in-situ measurement and soil physical and channel characteristics for the hydrological and hydraulic modeling. Brief summary of data used in modeling are as follows;

Remote Sensing data: The remote sensing derived parameters that have been used for model setup, delineation
of various thematic layers and model testing are (i) Digital Elevation Model (ASTER), (ii) Rainfall (NOAA Climatic Prediction Centre), (iii) LULC (LISS III), (iv) LAI (LISS III and MODIS) and (v) River network (LISS III, Google earth)

In-Situ measurements: The measurements that were used for model initialization, calibration and validation are
(i) River water level (2 locations), (ii) Velocity, (iii) Cross sections, (iv) Groundwater (pre and post monsoon), (v) Resistivity survey (groundwater, hydraulic conductivities, soil formations etc.), (vi) Rainfall, (vii) Temperature, (viii) Soil moisture and (ix) Leaf Area Index

Other data: The other data that were used for setting up and as calibration parameters for the coupled hydrological
and hydraulic model are (i) Soil physical properties and (ii) Channel characteristics

Study area
The watershed is located within and the buffer regions of the Kanha National Park along the south-east boundary of Madhya Pradesh State (Figure 1). Geographic extent is 80 0 42E to 81 0 01E and 22 0 02N to 220 31N. It covers 892 km2. Majority of the area is covered by forest and agriculture. Forest is mainly deciduous type and found near the southern parts. Main forest types are Sal, Sal mixed with bamboo and miscellaneous etc. Agricultural area extends towards middle and northern watershed area. Leaf area index varies from 1.5 to 6 in the study area. Elevation ranges from 523 to 900 m with
Fig. 1: Location map of the study area

an average of 689 m and the terrain slopes toward the northern side. The soil is mainly loamy. In the study area during the summer,

the temperature ranges from 11C to 43C, whereas in winter it ranges from 2C to 29C. The southwest monsoon prevails during late June to September. The maximum rainfall occurs in July and August. The average annual rainfall is 1,225 mm. Groundwater is shallow during monsoon season (surface to 20 feet) whereas pre-monsoon period groundwater table variation ranges from 4 to 49 feet. Area mainly consist of basaltic (porphyritic and non-porphyritic), quartz (Mica and k-feldspathic) and granite rock types. Forest area is dominated by the several perennial and semi perennial springs.

during the monsoon season and release it to the nearby rivers/channels as the delayed interow.

Methodology MIKE SHE/ MIKE 11 Coupled Model

MIKE SHE is a GIS-based distributed model. It is a spatially and temporally explicit, modularized modeling system. This model simulates the complete terrestrial water cycle, including saturated water movement in soils, 2-D water movement of overland ow, 1-D water movement in rivers / streams, unsaturated water movement and evapotranspiration (ET). Saturated water movement in soils is modeled using 3-D Finite Difference or Linear Reservoir. The 2-D water movement of overland ow is simulated using Finite Difference or Subcatchment-based method. The diffusive wave version of Saint Venant equations is used to simulate 1-D water movement in rivers / streams. The unsaturated water movement is simulated using either Richardss equation or Gravity Flow or Two-Layer water balance method. Detailed descriptions of the model and algorithms can be found in many publications (Abbott et al., 1986a, b; DHI, 2005).

Model Setup Parameterization

and

In this study, MIKE SHE was coupled with the river flow routing model MIKE 11 (DHI, 2005; Sahoo et al., 2006), a onedimensional river/ channel water movement model, to simulate the full hydrological cycle of the watershed, including evapotranspiration, inltration, unsaturated ow, saturated ow, overland ow and stream ow (Figure 2). The main inputs for the model included spatial data on topography, soils, vegetation,
Fig. 2: Processes for modeling hydrological cycle in the forest system

and drainage network; and temporal data on precipitation and Potential Evapotranspiration (PET).

Unsaturated Flow: The Two-Layer Water Balance model (Yan and Smith, 1994; DHI, 2005),
which is designed for the areas with a shallow groundwater table, was used to simulate the unsaturated ow for this study. The model divides the unsaturated zone into a root-zone where ET can occur, and a below-root-zone where ET does not occur (Yan and Smith, 1994).

Saturated Flow: The 3-D Finite Difference method (DHI, 2005) and linear reservoirs methods
were used to simulate the saturated ow and delayed interow for this study. The inputs needed to simulate saturated ow were soil hydraulic properties, including horizontal and vertical hydraulic conductivities, specic yield, and storage coefcient. Overland ow, subsurface ow (lateral ow) and ground water table level are signicantly affected by the values of vertical hydraulic conductivity.

Overland Flow and Stream Flow: Overland ow was simulated using diffusive wave
approximation. The inputs include initial water depth on the surface, surface detention storage, and Manning number. The measured surface water depth was used to initialize the water depth above 77

NNRMS

BULLETIN

DECEMBER

2013

These springs are located in the hilly regions and basically are geological lenses which store water

the ground surface for the model to run. Surface detention storage largely affects routing water toward the stream and water table dynamics. Large values of surface detention storage reduce the overland ow reaching the stream, but increase ponding water that may lead to a subsequent increase in water table level. Manning coefcients signicantly inuences routing overland ow toward the stream and stream ow toward the outlet of the stream with higher values leading to faster water movement.

Evapotranspiration (ET): In this study, daily actual evapotraspiration was estimated using the Kristensen
and Jensen model, 1975. Model is based on the effect of the LAI on transpiration and effect of soil and water on evaporation.

Calibration and Validation: Model is calibrated and validated based on the modeled and observed
measurements on river water level/discharges and groundwater measurements.

Simulation time Steps and Period: MIKE SHE has the exibility of using variable simulation time steps
for different hydrological modeling components and ow characteristics (DHI, 2005; Zhang et al., 2008). Simulation period was taken during 1 June 2010 to 15 August 2012.

Results Model Calibration and Validation

Daily Flows for the model calibration (2010) and validation (2011 and 2012) periods show that the model could capture the dominant runoff process and stream ow dynamics of the watershed (Figures 3
Fig. 3: MIKE SHE Model simulated at daily time scale for calibration (2010) and validation (2011 and 2012) at u/s gauging site (Kurkuti).

& 4). However, the model both over estimated and underestimated stream flow during the simulation period. This is because of perturbations in the discharge values which model could not handle and contributed for uncertainty in the simulations. Coefcients of determination (R2) of 0.82 and 0.78 and RMSE of 10.81 and 32.10 were obtained for the upstream (u/s) and downstream (d/s) gauging sites during model testing period, respectively. During the calibration period R2 values were of 0.78 and 0.81, whereas RMSE values were of 8.8 and 24.0 considering u/s and d/s gauging sites, respectively. A reasonable good match between modeled and observed stream ows were obtained (u/s

Fig. 4: MIKE SHE Model simulated at daily time scale for calibration (2010) and validation (2011 and 2012) at d/s gauging site (Sijhora).
2 2

gauge; R = 0.84 and RSME = 9.73 whereas d/s gauge; R = 0.78 and RSME = 39.13) for the validation period. The higher R2 values between modeled and measured discharge during calibration and validation period for the u/s gauging site was mainly due to the fact that fewer ow events with high peak ows were underestimated compared to d/s gauging site and less modeling errors occurred. For the validation period, peak ow rates for a few more events were underestimated or overestimated and resulting in much higher simulation errors.

Water balance components were extracted for the whole watershed. Further, results were extracted and analysed for the forested and other land cover classes separately (Figure 5) to see the inter linkages between the water balance components
Fig. 5: Water balance components for the catchment and land cover classes

of forest regions with the neighbouring were extracted for the geological lenses
BULLETIN

areas. Also, water balance components

representing for springs located in the study area. Rainfall for all the three categories viz whole catchment, forest and other classes was of 4230 mm. Out of which surface water which goes to the river, base ow production, evaporation from ponded water, canopy water storage and transpiration were of 36.6%, 22.4%, 3.6%, 12.5% and 24.9%, respectively for the whole catchment. Forest area contribution was more for base ow (26.9%), transpiration (30.2%), canopy storage (16.6%) whereas less contribution was for evaporation from ponded water (0.8%) and surface water (25.5%) in comparison to other classes water balance components. Recharge to the aquifers beneath the spring locations which are basically geological lenses was of 34.5% whereas other components contributions were of 65.5% which includes interception, ET, overland water etc. This water seeps out in the at regions during the post monsoon period and contributes for the channel ow.

Water Volume Estimations

Percentage daily contributions to the main river (Halon) from small rivers / streams / nala were extracted and are presented in Figure 6. Volume of water contributed by various streams to the main river were also estimated and presented in Figure 7. Total stream flow water among the different
Fig. 6: Daily percentage of water among different streams and small rivers / nala

streams ranges from 18 Million Cubic Metre (MCM) (Stream- 9, lowest order stream) to 338 MCM (Kashmiri Nadi with highest order stream). Total volume of water within the river system considering surface water and base ow contributions is estimated of 1994 MCM. Out of that base ow water from the forest region (514 MCM) which ows through the river and recharged to the groundwater may be utilized within the forest and neighbouring areas for various applications.

Fig. 7: Volume of water produced in different streams and rivers

Conclusion
The objective of this study was to evaluate the ability of the distributed hydrologic-hydrodynamic model, MIKE SHE to simulate different water balance components through hydrological experiment 79

NNRMS

DECEMBER

2013

Water Balance Analysis

and integrating remote sensing inputs. Model calibration and validation suggested that the model could capture the dominant runoff process of the large watershed. The physically based model required calibration at appropriate scales. The model was useful for understanding the rainfall and its various partitioning parameters in the forest as well other land cover classes and their interlinking mechanism. However, more datasets with higher temporal resolution are needed to further apply the model for regional applications.

References
Abbott, M. B., Bathrust, J. C., Cunge, J. A., OConnell, P. E. and Rasmussen, J. (1986a). An introduction to the European Hydrological SystemSysteme Hydrologique Europeen, SHE. 1: History and philosophy of a physically-based, distributed modeling system. J. Hydrol., 87 (1), 4559. Abbott, M., Bathurst, J., Cunge, J., Connell, P. and Rasmussen, J. (1986b). An introduction to the european hydrological system systeme hydrologique europeen, SHE, 2: Structure of a physically-based, distributed modeling system, J. Hydrol., 87, 6177. Anderson, W.M.D. and K.G. Hoover (1976). Forest and Water effects of forest management on oods, sedimentation and water supply, USDA Forest Service, General Tech. Report. Dabral, B.G. and Subbarao, B.K. (1969). Interception Studies in Sal (Shoria Robusta) and Khair (Acacia-Catchu) Plantation New Forest, Dehradun, Ind. For., 96, pp. 313-323. DHI: MIKE SHE Technical Reference, Version (2005). DHI Water and Environment, Danish Hydraulic Institute, Denmark, 2005. Dhruvanarayana, V.V. and Sastry, G. (1983). Annual Report, CSWCRTI, Dehradun. Dutt, C.B.S.and Manikiam, B. (1987). Forest Ecology and Related Weather Inuences. NNRMS, ISRO Hq., Tech. Report ISRO-NNRMS-TR-66-87, Bangalore. Gupta, A., Thapliyal, PK., Pal, PK, Joshi, PC (2005). Impact of deforestation on Indian Monsoon- A GCM Sensisitivity study. J. Of Indian Geophysic. Union, 9(2), pp 97-104. Hibbert, A.R. (1965). Forest Treatment Effects on Water Yield, Proc. Int. Symp. on Forest Hydrology, Pennysylvania State Univ., pp. 527-543. Ice, G.G., and J. D. Stednick (eds.) (2004). A Century of Forest and Wildland Watershed Lessons. Bethesda, MD: Society of American Foresters. Kristensen, K. J., and Jensen, S. E. (1975). A model for estimating actual evapotranspiration from potential transpiration. Nord. Hydrol., 6(3), 7088. Lal, V.B. and S.K. Subba Rao (1981). Hydrological Inuences of Vegetation Cover in Water shed Management, paper presented at Nat. Workshop. on Watershed Management, Mathur. H.N. (1975). Research in Soil Conservation Forestry, in Soil & Water Conservation Research, 1956-71, ICAR, Publication. Mehar-Homji, V.M. (1986). Trends of Rainfall in Relation to Forest Cover, The French Institute, Pondichery, Memoir. PSW-18. Sahoo, G., Ray, C., and Carlo, E. (2006). Calibration and validation of a physically distributed hydro logical model, MIKE SHE, to predict streamow at high frequency in a ashy mountainous Hawaii stream, J. Hydrol., 327, 94109. Yan, J. and Smith, K. (1994): Simulation of integrated surface water and ground water systems Model formulation, Water Resour Bull., 30, 112. Zhang, Z., Wang, S., Sun, G., McNulty, S., Zhang, H., Li, J., Zhang, M., Klaghofer, E., and Strauss, P. (2008). Evaluation of the MIKE SHE model for application in the Loess Plateau, China, JAWRA, 44, 11081120.

Introduction
India is one of the most ood prone countries in the world. India, due to its geographical location, climate, topography and large population, has a greater impact of ood disasters. Twenty-three of the 35 states and union territories in the country are subject to oods. Around 40 million hectares (mha) or nearly 1/8 of Indian geographical area is ood prone and the countrys vast coastline of 5700 km out of 7500 km is exposed to tropical cyclones (National Flood Control Commission Report, 1980). The annual average area affected by oods is about 7.57 mha and the crop area affected is about 3.5 mha. The average loss is about Rs.13,000 million. On an average the human lives lost is about 1595 (Gopalakrishnan, 2002). Floods occur in almost all major river basins in India. The Indo-Gangetic and Brahmaputra river basins are the most chronic ood prone areas and are regarded as the worst ood affected region in the world (Agarwal and Sunita, 1991). Every year states like Assam located in Brahmaputra basin and Bihar, Uttar Pradesh and West Bengal located in Indo-Gangetic basin face severe ood problems. Nearly 75 per cent of the total Indian rainfall is concentrated over a short monsoon season of four months (June-September). As a result the rivers witness a heavy discharge during these months, leading to widespread oods in Uttar Pradesh, Bihar, West Bengal and Assam. The Himalayan Rivers also carry a large amount of sediment, causing erosion of the banks in the upper reaches and over-topping in the lower segments. Inadequate capacity of the rivers to contain within their banks the high ows brought down from the upper catchment areas following heavy rainfall, leads to ooding. The tendency to occupy the ood plains has been a serious concern over the years. Because of the varying rainfall distribution, many a time, areas which are not traditionally prone to oods also experience severe inundation. Areas with poor drainage facilities get ooded by accumulation of water from heavy rainfall. Excess irrigation water applied to command areas and increase in ground water levels due to seepage from canals and irrigated elds also are factors that accentuate the problem of water-logging. The problem is exacerbated by factors such as silting of the riverbeds, reduction in the carrying capacity of river channels, erosion of beds and banks leading to changes in river courses, obstructions to ow due to landslides, synchronisation of oods in the main and tributary rivers and retardation due to tidal effects (NDMA, 2008). Drainage problems also arise concurrently if oods are prolonged and the outfalls of major drainage arteries are blocked. One of the major reasons for the oods is the massive indiscriminate deforestation, which leads to large amounts of topsoil

NNRMS

BULLETIN

Srinivasa Rao G, Bhatt CM, Manjusree P, Sharma SVSP, Asiya Begum Remote Sensing Applications Area, National Remote Sensing Centre ISRO, Department of Space, Hyderabad - 500 037, India Email: srinivasarao_g@nrsc.gov.in

DECEMBER

FLOOD MONITORING AND MANAGEMENT USING REMOTE SENSING

2013

coming loose in the rains. Thus, the soil, instead of soaking up the rainfall, ows down into the river and in turn causes the riverbeds and its tributaries to rise.

Phases of Flood Management

Flood disaster management cycle has three main phases viz. ood preparedness (before ood occurs), ood response (during a ood) and the last phase called ood mitigation (after ood has occurred). Flood preparedness involves identication of chronically ood prone areas, identication of areas that are liable to be affected by a ood and planning of optimum evacuation plans. Flood response involves the immediate action taken once the ood disaster has occurred in terms of the identication of the region affected, spatial extent of inundation, ood damage statistics, ood progression and recession etc which can help in carrying out the relief and rescue operations on ground. Flood mitigation phase starts after the ood has occurred by identication of the changes in the river course due to ooding, status of ood control works, river bank erosion, drainage congestion, ood hazard and risk vulnerability assessment .

Space Inputs for Flood Disaster Management

The most important element in ood disaster management is the availability of timely information for taking decisions and actions by the authorities (Miranda et al., 1988, Okamoto et al., 1998). NRSC under the Disaster Management Support (DMS) Programme of ISRO provides support during the ood response phase by providing ood based products like ood inundation maps, damage statistics, ood progression, flood recession and flood persistence to state agencies. During the post disaster
Fig. 1: Different phases of flood disaster management being addressed using remote sensing

phase monitors the ood situation, status of the ood control works etc. Under the mitigation phase studies towards flood

hazard zonation, ood risk and vulnerability and bank erosion are also carried out. Apart from this ood forecasting and ood inundation simulation studies towards early warning are also initiated. Figure1 shows the different phases of ood disaster management being addressed using remote sensing.

Early Warning
The Godavari river system is one of the major river systems in the country and one of the most ood-prone in the Southern India. Heavy rains in the catchment of Godavari river during the rst week of August, 2006 caused heavy loss to lives and infrastructure in East Godavari, West Godavari and Khammam districts of Andhra Pradesh (India).

Flood Forecasting
Though Central Water Commission (CWC) is the main nodal agency having the mandate to provide the ood forecast, NRSC has collaborated with CWC for the development of medium-range Flood Forecast Model for the Godavari Basin using space inputs (LU/LC, Soil, SRTM-DEM) and hydro-meteorological data through semi-distributed modeling approach. Flood forecasting model is run by CWC during 2010 &11 using real time 3 hrs Hydro-meteorological data. The discharge data generated as output would be very useful input for inundation simulation for developing Spatial Early warning.

Flood inundation simulation studies for part of Sabari tributary from Konta to Kunavaram (35 km) stretch in Godavari River basin were carried out using discharge data, land use/land cover and ALTM DTM and the results were validated with satellite data. Figure 2 shows the owchart of the methodology used for simulation of ood inundation. The river geometry data extracted using HEC-GeoRAS, together with the discharge data and other parameters, spatial ood modeling has been carried out for peak ood situation using HEC-RAS, 1-D hydraulic model. Hourly water level data was used to generate different ood inundation patterns. Flood inundation is simulated for the year 2010 and the maximum ood inundation extent simulated by the model was compared with the corresponding actual inundated area obtained from Radarsat image. Figure 3 shows the comparison of the simulated and observed ood extent. The results obtained
Fig. 2: Flowchart of the methodology used for simulation of ood inundation

from the model for peak ood situation were in well accordance with ood inundation observed from corresponding satellite data.

Flood Response
Decision Support Centre (DSC) keeps a continuous watch on the ood situation in the country through different sources. Based on the cloud cover pattern observed from meteorological satellite KALPANA-1, rainfall pattern from Indian Meteorological Department (IMD), hydrological data from Central Water Commission (CWC), and flood related information from websites, news media and state departments satellite data from various satellite sensor systems is programmed. Presently emergency requests
Fig. 3: Comparison of the simulated and observed ood extent

are placed for programming of the Indian satellites (Resourcesat-1 & 2, Cartosat-1

& 2 and RISAT-1 & 2) and as well as the foreign satellites (RADARSAT-1 & 2) for ood mapping and monitoring and identication of embankment breaches. The extent of ood inundation is extracted from the satellite data and ood maps at various scales i.e. state, district and detailed levels are prepared for the ood affected states. The spatial inundation maps along with estimates on 83

NNRMS

BULLETIN

DECEMBER

2013

Flood Inundation Simulation

submergence are generated within 5-6 hours after receiving raw satellite data product and disseminated to the State and Central Governments and user departments through a VSAT based satellite communication network. In addition to the oodmaps, the ood inundation layer in GIS compatible format are also transmitted to the state remote sensing application centres for further value addition and dissemination as per the requirements of the user/line departments. Successful operational use of remote sensing technology for near real time ood mapping and monitoring was done for the Bihar oods of 2008, Andhra Pradesh oods of 2009, Uttar Pradesh oods of 2010, Orissa oods of 2011 and the for recent Assam oods of 2012.

Assam Floods 2012

During June, 2012, Assam State witnessed one of the devastating oods since 1988, due to high ood levels in Brahmaputra river and its tributaries causing huge loss of human lives, cattle and infrastructure. Decision Support Centre (DSC) kept a close watch on the ood situation. All efforts are made to acquire the satellite data over ood affected areas in Assam. In addition, DSC also activated International Charter Space and Major Disasters for frequent observations over the ood affected areas.The satellite datasets were analysed and ood inundation layer was extracted. Flood maps and ood inundated area statistics were generated and provided to State
Airport Settlement District Hq State boundary Railway National Highway Other Major Roads Normal river / Water bodies River bank Flood Inundation Area covered by satellite

& Central Government Departments.

Overview of Flood Situation

DSC has acquired and analysed satellite data of 27, 29 & 30 of June and 2 July 2012 and provided the overview of the flood situation to the concerned. Figure 4 shows the ood affected areas in various districts of Assam State during 27-Jun to 02-Jul, 2012. It was observed that about 4.65 lakh hectares of area was inundated. Major flood inundation was observed in the districts of Nowgong, Cachar, Marigaon, Sonitpur, Marigaon, Lakhimpur, Barpeta, Kamrup (Rural), Jorhat, Karimganj, Golaghat, Sibsagar, Dhemaji, Dibrugarh and Darrang districts.

Fig. 4: Flood inundated areas in Assam State during 27-Jun to 02-Jul, 2012

Affected Transport Network

Due to the severity of the oods, roads were damaged at several places. About 426 km length of major roads in Assam state was observed to be affected by these oods as on July 02, 2012. Signicant damages to railway
Fig. 5: Indian remote Sensing satellites, RISAT-1 & 2 showing the ood situation and the affected railway track in parts of Dibrugarh district

track due to these oods were reported. About

Figure 5 shows the affected railway track in parts of Dibrugarh district as observed by Indian microwave remote sensing satellites, RISAT-1 & 2.

Marooned Villages
The ood waters marooned several villages in the State and many villages were surrounded by ood waters for several days. About 3829 villages were observed to be affected by ooding as on July 02, 2012. In Nowgong district about 312 villages were affected by oods upto 02-July, 2012. It is observed that about 77 villages are marooned for more than 6 days, 85 villages for 4-5 days and 75 villages for 1-3 days. Recession in ood was also observed in 75 villages in Nowgong district.

Flood Situation at Kaziranga National Park

Kaziranga National Park is one of the most severely affected areas in Assam during these oods. Loss of life of several wild animals was reported. Figure 6 shows the temporal changes in ood inundation around Kaziranga National Park area during June 27-30, 2012 and detailed view of ood inundation around Kaziranga National Park area (Figure 7). Recession in flood
Fig. 6: Satellite images showing temporal changes in ood inundation around Kaziranga National Park area during June 27-30, 2012

inundation is observed in some parts during June 27-30, 2012.

Flood Mitigation
Towards ood mitigation DSC has carried out ood hazard zonation, ood risk and vulnerability and bank erosion studies. Flood Hazard Zonation (FHZ) is one of the most important non-structural measures, which facilitates appropriate regulation, and development of oodplains thereby reducing the ood impact. Flood hazard zonation maps for Assam state are prepared, whereas for Bihar state the ood hazard maps are being validated and the atlas is under preparation.
Fig. 7: Resourcesat-2 image showing the detailed view of ood inundation around Kaziranga National Park area as on July 01, 2012

Flood Hazard Zonation for Assam

A ood hazard zonation atlas is prepared based on the analysis of 93 satellite datasets acquired during 1998-2007 ood season over Assam region and the atlas has been released (Figure 8). The ood inundation layers generated from the analysis of the satellite data for different ood waves in a calendar year were integrated in GIS environment to generate the maximum ood inundation extent observed in that year. The maximum ood inundation layers corresponding to various years (1998-2007) were integrated for assessing the frequency of inundation and subsequent generation of ood hazard layer. 85

NNRMS

BULLETIN

DECEMBER

2013

88 km length of railway track in Assam state was observed to be affected by these oods as on July 02, 2012.

The ood hazard has been classied into ve categories based on frequency of inundation (Table 1). Very Low category indicates the areas which are inundated once or twice during the 10-year period. Similarly, Low indicates three to four times, Moderate indicates ve to six times, High indicates seven to eight times and Very High indicates the areas, which are regularly subjected to
Settlement
Very low Low Moderate High Very High River/Water bodies

inundation. Area under each category was estimated and ood hazard maps at state and district level were prepared. Further, cropped area (from land use) was also integrated with ood hazard layer to assess the impact. The statistics for cropped area affected under each hazard category was computed.

District Hq District boundary Taluk boundary Village boundary Railway National Highway District Road

Fig. 8: Flood hazard zonation map for Assam state

Table 1: Flood hazard area under various categories Sl No Hazard Severity Flood Hazard Area (ha) % Flood Hazard (with respect to State Geographic Area) 1.64 3.86 4.48 6.27 13.06 28.31 % Flood Hazard (with respect to Total Flood Hazard Area) 5.79 10.11 15.83 22.14 46.13 100.00 Crop Area Under Different Flood Hazard Categories (ha) 83488 168802 270558 351356 621367 14,95,571

1. 2. 3. 4. 5.

Very High High Moderate Low Very Low Total

1,28,687 2,24,629 3,51,667 4,91,761 10,24,584 22,21,328

It is observed from the analysis that about 22.21 lakh hectares of land constituting about 28.31% of Total Geographical Area (TGA) of Assam state is affected by ooding (Table 1). Out of the total ood affected area of 22.21 lakh hectares, about 1.28 lakh hectares of land falls under very high ood hazard category. This area is observed to be continuously under submergence during the last ten years period. About 2.24 lakh hectares of land falls under high ood hazard category, indicating that this area has been subjected to ooding for about 7-8 times during last ten years. Area falling under moderate ood hazard category (area subjected to inundation 5-6 times during last ten years period) is estimated to be about 3.51 lakh hectares, constituting about 4.48% of TGA of Assam state.

Development of Village-Wise Flood Risk Index Map

Village ood risk index map for Nagaon district has been generated using the ood hazard layer as primary input. About 50 satellite datasets, optical (IRS) and microwave (Radarsat), acquired during the last 10-year period (19982007) during the ood season have been analysed to extract ood inundation layer and generate composite ood hazard layer. Flood hazard layer is considered as the primary input and is integrated with land use/ land cover, infrastructure and population data and weightages are assigned to each class. Based on this, village ood risk index map for Nagaon district has been generated. Figure 9 shows the ow chart of the methodology adopted for generating village-wise ood risk index. The results of analyses indicate that about 267 villages are in the moderatehigh risk index zone. About 35,354 ha of the district is in high ood hazard zone and about 25,281 ha of crop area is affected annually.

Planning and acquisition of satellite data during foods

Village layer

Bank erosion is one of the most

Population vulnarability

Extraction of vector layer Land use/Land cover Agricultural land = 1 (kharif and double crop) Other land use =3

Infrastructure vulnarability

in the Brahmaputra basin. Rate of erosion in

Population Population > 3000 =1 Population 1001-3000 = 0.6 Population 1000 = 0.3

Pre-ood river and water bodies

Integration

Pre-ood river and water bodies

the Brahmaputra catchment, 953 ton per sq km per year, is the highest in any catchment system in the whole world (Venkatachary et.al., 2001). Resourcesat-1 LISS-III satellite

Flood inundation layer

Transport network National and state highways = 1 District roads = 0.6 Other roads = 0.3

Land use vulnarability index (LVI)

Infrastructure vulnarability index (IVI)

Population vulnarability index (PVI)

data having a spatial resolution of 23 m acquired during 2002 and 2010 were used for the bank erosion and deposition study in
BULLETIN

Flood layers 1998-2007

Integration

Vulnarability index (VI) =LVI + IVI + PVI

the Brahmaputra and Barak Rivers in Assam. The satellite datasets of 2002 and 2010 were geometrically rectied to the master base map for positional accuracy. The individual satellite data sets were mosaiced for the individual years of 2002 and 2010. The banklines of the rivers for the year 2002 and

Integration with village layer

Hazard index

Vulnarability index (VI)

Hazard index High road hazard Moderate road hazard Low road hazard

=1 = 0.6 = 0.3

Village-wise ood rish index (VFRI) = hazard index x Vulnarability index

2010 were delineated. Both the banklines were intersected to identify and estimate the amount of erosion and deposition at different pockets along the main Brahmaputra and Barak river stretch. Figure 10 shows the pockets where bank erosion and deposition has taken place from the comparison of the satellite images acquired during 2002 and 2010 for part of Assam.

Fig. 9: Flow chart of the methodology adopted for generating village-wise ood risk index

Conclusion
Space technology has emerged as the most powerful tool for decision making in ood disaster management. The technology can help in disaster identification, response prioritisation, damage assessment, inundation monitoring, river course changes and identication of vulnerable zones required for ood disaster management. The near realtime information provided during ood disaster can be of immense help to decision makers
Fig. 10: Bank erosion and deposition delineated from satellite images in parts of Assam

to evolve risk reduction strategies. Remote sensing technology due to its repetitive and synoptic coverage, cost-effectiveness prove

to be a viable tool for monitoring the river shifting, proximity of river course to the embankments and channel aggradation by the planners for taking measures required for channel stabilization and strengthening of embankments. 87

NNRMS

Flood layers 1998-2007

DECEMBER

Geo-rectication of satellite data

Land use vulnarability

commonly experienced problems especially

2013

Bank Erosion Studies

References
Agarwal, A. and Sunita, N. (1991). Floods, oodplains and environmental myths. State of Indian environment: A citizen report, Centre for Science and Environment, New Delhi. Gopalakrishnan, M. (2002). Central Water Commission. The International Seminar on Disaster Preparedness and Mitigation, 21 November 2002 New Delhi. Miranda, F.P., Fonseca, L. and Carr, J.R., (1988). Semivariogram textural classication of JERS-1 SAR data obtained over a ooded area of the Amazon rainforest. International Journal of Remote Sensing, 19, pp. 549556. NDMA, (2008). National Disaster Management Guidelines: Management of Floods National Flood Commission Report (Rashtriya Barh Ayog), 1980 Okamoto, K., Yamakawa, S. and Kawashima, H. (1998). Estimation of ood damage to rice production in North Korea in 1995. International Journal of Remote Sensing, 19, pp.365367. Venkatachary, K.V., Bandhypadhyay, K., Bhanumurthy, V., Rao, G.S., Sudhakar, S., Pal, D.K., Das, R.K., Sarma, U., Manikiran, B., Rani, H.C.M. and Srivastava, S.K. (2001). Dening a space-based disaster management system for oods: A case study for damage assessment due to 1998 Brahmaputra oods. Current Science. 80 3, 369-377

Introduction
Flood is a chronic disaster occurring almost every year in the state of Assam. Along with structural measures like the construction of embankment, non structural measures like ood forecasting, ood plain zonation and regulation etc in recent past has gained importance among researchers, technocrats and policy makers. In this context an attempt has been made by North Eastern Space Applications Centre (NESAC) on a request from Govt. of Assam to mitigate the ood damage by developing an operational ood warning system for the state by using geospatial technology coupled with established relationships among important hydro-meteorological parameters. Thus this exercise which was started with Lakhimpur district, in upper Assam, as a pilot study in 2009 has been able to cover 14 ood prone districts in Brahmaputra valley and three south Assam districts in Barak valley. Hydrological events require extensive datasets to understand its highly dynamic nature. However, the scarcity in gauge data has been a challenge to the scientic community. So far, this limitation has been addressed by various hydrological models attributing the hydrometry and morphometry of the catchment. The strength of hydro-meteorological modeling in geo-spatial domain for ood forecasting has been put to operational use covering such large study area for the rst time in any ood prone state of India.

Objectives
The study has been taken up to achieve following short and long term objectives (1) Issue of alert for possible ood situation in District/ Revenue Circle/ Village level with best possible lead time. (Operational). (2) Submission of annual periodic report on post-ood status of existing ood protection embankments in district level (Operational). (3) Development of optimum methodology for rainfall prediction from satellite based weather monitoring and numerical weather prediction models supported by insitu ground data (Research). (4) Development of river specic rainfall-runoff models for forecasting of ood (Research). (5) Development of inundation simulation for ood plain zonation (Research).

Methodology
The overall exercise of FLEWS (Flood Early Warning System) has got different multidisciplinary components under both operational as well as research objectives. Following

NNRMS

BULLETIN

Diganta Barman, Kundu SS, Jonali Goswami, Ranjit Das, Singh NGR, Arup Borgohain, Rekha Bharali Gogoi, Victor Saikhom, Suranjana B Bora and Sudhakar S North Eastern Space Applications Centre (NESAC) Umiam, Shillong, Meghalaya - 793 103, India Email: diganta_isro1@yahoo.co.in

DECEMBER

GEOSPATIAL AND HYDRO-MET APPROACH FOR FLOOD MANAGEMENT IN ASSAM

2013

ow chart explains the operation work ow leading to issue of early warning of ood. This work ow comprises of a series of analytical exercises involving hydro-meteorological modeling along with real time monitoring of critical hydro-meteorological observations as input for decision support in issuing ood warnings. The prominent components are briey explained in Figure 1.

The meteorological Component

This component of the FLEWS workflow basically deals with two subFig. 1: FLEWS operational block

components viz Near real time weather watch with daily collection insitu rainfall reading from various sources like Automatic Weather Stations, Automatic Rain Gauges etc and Running of Numerical weather prediction model such as Weather Research Forecast (WRF) with 3 hourly rainfall forecasts in grids of various spatial resolution such as 27 km x 27 km, 9 km x 9 km, 3 km x 3 km etc. that goes as the daily input to the run-off model. The MM5 and ARPS (Advanced Regional Prediction System) models are also being used to investigate their relative ability to simulate different events, particularly intense and localized rainfall events as shown in Figure 2. The High Performance Computers

Fig. 2: WRF derived daily rainfall chart over the study area

(HPC) at INCOIS, Hyderabad and at NARL, Gadanki have been used through remote access to operationally run the models during the monsoon season. Moreover the Synoptic weather analysis is being done to forecast rainfall for the basins under study. Since many of the basins under study is very small in size, rainfall forecast is given at synoptic scale. The forecasts are issued for four districts (for the districts in eastern Assam, western Assam, central Assam and Southern Assam) over the NER. The preparation of synoptic weather report involves studying the cloud properties over and around the area, the prevailing wind speed and direction at different

Fig. 3: Synoptic weather monitoring block diagram

heights (pressure levels), the condition of

preparation of synoptic scale rainfall forecast is shown in Figure 3.

The Hydrological Component

This component of the FLEWS workow basically deals with two sub-components viz preparation of all hydrology specic watershed layers such as basin boundary, landuse, drainage, slope, aspects, soil classication etc., leading to calculation of various watershed parameters discharge etc., and generation of discharge forecast with two approaches by using a lumped and a distributed hydrological model (Figure 4). Here the lumped approach delivers the peak discharge forecast and the distributed approach delivers a daily hydrograph with single or multiple peaks. The output of both the models together
Fig. 4: FLEWS rainfall to runoff conversion component

nally leads to the decision to issue a ood alert on the concerned river and the district.

The hydrological evaluations indicate that the average annual discharge of Ranganadi ranges from 72 to 163 cumecs (m3/sec) as based on observations (NEEPCO). The average monsoon discharge of Subansiri for the year 2009 is 5745 cumecs (CWC) as observed at Chouldhuwaghat GD site. The highest discharge of 18, 215 cumecs was recorded on 26th July. High resolution hydrologic forecasts can provide information on ash ood situations; however, an important question to consider in evaluating high resolution forecasts is whether or not larger simulation uncertainties at smaller scales will diminish the utility of these forecasts, and, if so, to what degree? As part of the FLEWS project, a statistical-distributed hydrologic modeling approach on a ne grid scale is being developed to simulate ash ooding on small basins and account for hydrologic modeling uncertainty. Hydrologic kinematic wave routing is adopted in parallel with the existing Muskingum / Muskingum-Cunge method (Rozalis et al., 2010) in the model approach with an objective to enhance our ability to predict the occurrence of ash ooding. The statistical-distributed approach requires running a distributed model using archived and WRF-derived precipitation grids at ner grid scales (less than 4 km) to derive ood probability characteristics of simulated ows for all sub-watersheds in each grid cell in the distributed model. When running the distributed model in forecast mode, the ooding ow threshold for each grid cell is dened in terms of a ood probability level rather than an absolute value of ow (Yatheendradas et al., 2008). A hydrological model used for ash ood modeling and prediction is inevitably an abstraction of reality.

The Post Flood River Embankment Monitoring

This component of FLEWS project deals with analysis of high resolution satellite data on an annual basis after the completion of ood season in order to identify existing breaches in various embankments created during the completed ood season and to be reported to the concerned authority/ department to enable them to take corrective actions for the next season. An example of river embankment breach is shown in Figure 5. 91

NNRMS

BULLETIN

such as time of concentration, co-efcient of

DECEMBER

2013

the atmosphere in terms of its ability to allow or suppress convection, etc. The overall ow chart for

Results and Conclusion

This project has been taken up on the joint request of Assam State Disaster Management Authority (ASDMA), Government of Assam and the North Eastern Council (NEC), Ministry of DONER (Development of North Eastern Region), Government of India. Based on the initial success achieved so far, fresh request has come from Government of Assam for implementing this project in all the ood prone districts of Assam in phases. Based on the above described operational work components, ood warnings has been issued
Fig. 5: River embankment breach identied from Cartosat I data

to concerned district authorities several times during 2010, 2011 monsoon seasons in Assam with an average lead time of 12 to 18

hours. Majority of the signicant ood events during these two years have been successfully forecasted and encouraging feedback have been received from the user. By the year 2012, 14 severely ood prone districts of Assam covering 60 odd rivers, tributaries and rivulets have been brought under this project. At the behest of NDMA, New Delhi states like Bihar and West Bengal have also ofcially corresponded to NESAC for exploring avenues for joint collaboration for their respective states in the eld of ood management. Due to the gradually improving performance of FLEWS, in 2012 this project has been selected for professional documentation under Good governance initiative by Department of Administrative Reforms & Public Grievances under Ministry of Personnel, Public Grievances & Pensions, Govt. of India. Thus the FLEWS project has been able to start a holistic approach to look at the overall ood management scenario in North Eastern Region of India.

References
Rozalis, Morin, Yair, price. (2010). Flash ood prediction using an uncalibrated hydrological model and radar rainfall data in a Mediterranean watershed under changing hydrological condition Journal of Hydrology, 314 p. 245 255. Yatheendradas, Wagener, Gupta, Unkrich, Goodrich, Scaffner, Stewart (2008). Understanding uncertainty in distributed ash ood forecasting for semiarid regions Water Resources Research, AGU, Vol 44, Issue 5,

Introduction
Drought is a climatic anomaly characterized by decient supply of moisture resulting either from sub-normal rainfall, erratic rainfall distribution, higher water need or a combination of all the three factors. Agricultural drought is a situation when rainfall and soil moisture are inadequate during the crop season to support the timely cultural practices and healthy crop growth from sowing to harvesting. Drought results from adverse climatic conditions leading to deleterious impacts on various sectors of the economy. The immediate impact of drought is on crop area, crop production and farm employment. Less than normal crop sown area, delayed time of crop sowing, poor crop growth caused by inadequate soil moisture availability ultimately lead to decrease in crop yield or crop production. Impacts of droughts on crop productivity are more intense in low rainfall regions than in regions receiving higher quantum of rainfall. Droughts have a multiplier effect on agricultural production during the subsequent year also due to non-availability of quality seeds for sowing of crops, reduced use of inputs like fertilizers as the investment capacity of the farmers decline, non-availability of raw material in agro-based industries etc. Drought Occurrence in India On an average, severe drought occurs once in ve years in most of the tropical countries, though often they occur on successive years causing misery to human life and livestock. About two thirds of the geographic area of India receives low rainfall (<1000 mm), which is also characterized by uneven and erratic distributions. Out of net sown area of 140 mh about 68% is reported to be vulnerable to drought conditions and about 50% of such vulnerable area is classied as severe, where frequency of drought is almost regular. India experiences localized drought almost every year in some region or other. In the postindependence era, major droughts that affected more than 1/3rd of the country were reported during 1951, 1966-67, 1972, 1979, 1987-88 and 2002-03 (Subbaih, 2004). Drought Occurrence Global Scenario Nearly 50 per cent of the worlds most populated areas are highly vulnerable to drought (USDA 1994). In the worlds two largest agricultural producers, the United States and the former Soviet Union, drought occurs almost every year. More than 500 million people live in the drought prone areas of the world and 30% of the entire continental

NNRMS

BULLETIN

Sesha Sai MVR, Murthy CS, Chandrasekar K, Mohammed Ahamed J and Prabir Kumar Das Agricultural Sciences & Applications Group National Remote Sensing Centre ISRO, Department of Space, Hyderabad - 500 037, India Email: seshasai_mvr@nrsc.gov.in

DECEMBER

AGRICULTURAL DROUGHT: ASSESSMENT & MONITORING

2013

surface is affected by droughts or desertication process. In 2004, widespread drought in much of Asia resulted in loss of signicant agricultural production.

Meteorological Drought
Normal rainfall during south west monsoon season is 890 mm for the country as a whole and 611 mm in north-west India, 994 mm in central India, 723 mm in southern India and 1427 mm in north east India. Out of 890 mm of normal rainfall in the country, June month accounts for 162 mm followed by 293 mm in July, 262 mm in August and 175 mm in September (www.imd.gov.in). About 70% of the annual rainfall over India is contributed by south west monsoon which commences in the month of May from the southern tip. The normal duration is roughly 100 days and its withdrawal starts from Punjab and Rajasthan by mid of September. The onset of north east monsoon is not well dened and a general understanding is that the rainfall in the winter months from October to December represents the northeast monsoon. Tamil Nadu state receives signicant rainfall from north east monsoon, and it is about 463 mm representing 48% of annual rainfall of the state (www.imd.gov.in). Rainfall is the most important single factor inuencing the incidence of drought and practically all denitions use this variable either singly or in combination with other meteorological elements. Many studies have analysed the nature and frequency of droughts based on simple relation between actual and average rainfall. Based on rainfall, temperature, soil moisture and evaporation, various indicators of meteorological drought like Palmers index, Standardized Precipitation Index, Crop Moisture Index etc. have been developed. The India Meteorological Department (IMD) prepares rainfall maps on sub-divisional basis every week throughout the year. These maps show the rainfall received during a week and corresponding departures from normal. During monsoon season, these maps are indicative of development of drought. In addition, IMD also provides the information on weekly rainfall and its deviation from normal at district level for the entire country. This data is useful to identify the districts with decit/scanty rainfall and the prevailing meteorological drought. IMD also monitors drought using climatic water balance technique. The aridity index is calculated as a fraction of water decit/water need. The departure of aridity index from normal percentage terms is used to dene the drought severity. Anomaly upto 25% is attributed to mild drought, 26-50% to moderate drought and >50% to severe drought. IMD has been bringing out weekly aridity anomaly charts, which show the departures of actual aridity from normal aridity giving indication of the severity of water decit to water demand relationship on weekly basis. Monsoon and Tele-connections The variability of summer monsoon rainfall in the country has been found to be closely linked to the variations in Sea Surface Temperature (SST) over the equatorial Pacic and Indian Oceans (Gadgil et al., 2003). El-Nino Southern Oscillation Index (ENSO) a climatic phenomenon signifying temperature and pressure patterns in the Peruvian coast of South America and the central Pacic along the South American coast. Warm phase of ENSO called El-Nino is found to be associated with reduced summer monsoon rainfall over India. Rajeevan and Pai (2006) indicated that most of the severe droughts over India are associated with El-Nino events. But all the El-Nino years have not resulted in drought occurrence. In addition to El Nino, Southern Oscillation (ENSO), Equatorial Indian Ocean Oscillation (EQUINO) also inuences the summer monsoon rainfall in India. Severe droughts during 1958-2003 were associated with unfavorable phases of either ENSO or EQUINO phenomenon. Negative phases of ENSO and EQUINO were evident in 2002. Such a phenomenon was also observed in June 2009 but with lesser magnitude compared to 2002. The rainfall deciency in June 2009 was related to suppression of convection over Bay of Bengal with unfavourable SST gradient between Bay (more cooler) and Eastern Equatorial Indian Ocean (EEIO). A similar suppression had also occurred in 1995 resulting in a larger decit rainfall in the country.

recurrent, seasonally varying patterns of atmospheric circulation anomalies referred to as teleconnections. These tele-connections affect the precipitation regime over much of the Tropics. Observational studies and model experiments have also demonstrated a signicant link between Atlantic sea surface temperatures and precipitation over the drought-prone areas of the African Sahel and northeast Brazil.

Agricultural Drought
Agricultural drought results from the complex and nonlinear interactions between weather, continues to be a challenging task for researchers, drought managers and policy makers. Meteorological drought is measured by rainfall data recorded by weather stations and the hydrological drought is assessed by inows into the surface water bodies measured through gauging points. But, agricultural drought assessment could not be accomplished by such direct and quantitative measurements. It requires the quantitative information related to soil moisture, planting pattern and crop condition along with their inter-relations which is not adequately provided through manual surveys or sparsely located measuring stations. The intensity of agricultural drought is largely determined by the crop growing environment - diversity of crops grown in a given location, soil variability, rainfall variability etc. Since the water requirement of a crop varies during different phenol-phases of growth, the impact of drought depends on the crop growth stage. Central Arid Zone Research Institute (CAZRI) attempted classication of agricultural drought considering the values of Actual Evaporation/Potential Evaporation (AE/PE) during different phenol phases of crop growth seedling, vegetative and reproductive which was later improved with crop factor (Sastri et al., 1981). Agriculture departments of different states collect information on crop sown areas, crop development, pests and diseases occurrence etc. to assess the drought situation. A special task force known as Crop Weather Watch Group is constituted by Ministry of Agriculture, Government of India. This group reviews the progress of monsoons, crop situations, water levels in reservoirs/dams, availability of fertilizers etc. Agricultural Drought Assessment Geospatial Approach Dynamic nature of droughts with complex phenomenon having multiple effects is a major challenge in planning, monitoring, predicting, assessing impact and offering solutions to drought hit areas. Because of these complexities, high quality data and improved tools are needed to capture the spatial and temporal dimensions of drought intensity. Unlike point observations of ground data, satellite sensors provide direct spatial information on vegetation stress caused by drought conditions. Satellite remote sensing technology is widely used for monitoring crops and agricultural drought assessment. Over the last 20 years, coarse resolution satellite sensors are being used routinely to monitor vegetation and detect the impact of moisture stress on vegetation. The NOAA AVHRR NDVI has been extensively used for drought/vegetation monitoring, detection of drought and crop yield estimation (Batista et al., 1997, Beneditte and Rossini, 1993, Moulin et al., 1998 and Tucker et al., 1985). United States Agency for International Development has evolved a Famine Early Warning System NetWork (FEWSNET) by integrating the composite information on temperature, winds, humidity, soil and topography, observations on conict, civil interest, health, market prices, eld observations soil, crop and human actions and hence, the assessment of the intensity of agricultural drought

95

NNRMS

BULLETIN

DECEMBER

2013

The effect of these ocean variations is transmitted to remote areas of the globe through

on agriculture, satellite derived rainfall and NDVI. FEWSNET is operationally issuing monthly food security reports for decision makers in Africa and USA (http://ltpwww.gsfc. nasa.gov). The Drought Monitor of USA using NOAA-AVHRR data (www.cpc.ncep.noaa.gov), Golbal Information and Early Warning System (GIEWS) and Advanced Real Time Environmental Monitoring Information System (ARTEMIS) of FAO using Meteosat derived cold cloud duration and SPOT VGT derived decadal NDVI composites (Minamiguchi, 2005), International Water Management Institute (IWMI)s drought assessment in South west Asia using MODIS data (Thenkabail, 2004) and NADAMS drought monitoring in India with IRSWiFS/AWiFS and NOAAAVHRR (Murthy et al., 2007) data are the proven examples for successful application of satellite remote sensing for operational drought assessment Central Research Institute for Dry Land Agriculture (CRIDA), Hyderabad, provides information on drought conditions and their mitigations measures during the season. The weekly contingent crop plans for rain-fed regions over the country during the crop growing seasons and at fortnightly interval in the rest of the period are prepared from the inputs provided by the centers of both the coordinating projects and the general weather situations prevailed over the period. These are updated regularly in the website Crop-weather-outlook maintained by CRIDA. In addition, weather based agro-advisories prepared from the forecast of National Centre for Medium Range Weather Forecasting (NCMRWF) and crop condition are also updated regularly in the above mentioned website to meet any aberrant weather conditions. National Agricultural Drought Assessment and Monitoring System (NADAMS) In India, National Agricultural Drought Assessment and Monitoring System (NADAMS) was initiated towards the end of 1986, with the participation of National Remote Sensing Agency, Dept. of Space, Government of India, as nodal agency for execution, with the support of India Meteorological Department (IMD) and various state departments of agriculture. NADAMS was made operational in 1990 and has been providing agricultural drought information in terms of prevalence, severity and persistence at state, district and sub-district level. Over a period of time, NADAMS project has undergone many methodological improvements such as use of moderate resolution data for disaggregated level assessment, use of multiple indices for drought assessment, augmentation of ground data bases, achieving synergy between ground observations and satellite based interpretation, providing user friendly information, enhanced frequency of information etc. Monitoring of agricultural drought is limited to kharif season (June-Oct/November), since this season is agriculturally more important and rainfall dependent. Resourcesat-1 / Resourcesat-2 AWiFS based sub-district level agricultural drought assessment is carried-out in 4 states namely Andhra Pradesh, Karnataka, Maharashtra and Haryana. AVHRR/MODIS based district level assessment is carried out in 9 states namely Bihar, Chhattisgarh, Gujarat, Madhya Pradesh, Orissa, Rajasthan, Uttar Pradesh, Jharkhand and Tamil Nadu. Oceansat-2 OCM data was brought into operational use in kharif 2011 by generating fortnightly and monthly Rayleigh corrected TOA NDVI composites for all the 13 states. MODIS 250m derived NDVI and NDWI are also used for assessment. Coarse resolution products of soil moisture from passive microwave data, rainfall estimates and rainfall forecasts are also used in the assessment. The Area Favourable for Crop Sowing/crop sown area (AFCS), derived from Shortwave Angle Slope Index (SASI) and ancillary data is a recently added product in NADAMS project. This product is useful to monitor the sowing period drought conditions. Overview of NADAMS project is shown in Figure 1. From kharif 2012, NADAMS project is being carriedout by Mahalanobis National Crop Forecast Centre (MNCFC), Department of Agriculture and Cooperation, Ministry of Agriculture, with the technical support of NRSC. The satellite derived crop indices that are widely used in NADAMS project- NDVI, NDWI and SASI- are described in subsequent sections. Normalized Difference Vegetation Index (NDVI) Among the various vegetation indices, Normalized Difference Vegetation Index (NDVI) is widely used for operational drought assessment because of its simplicity in calculation, easy to interpret and its ability to partially

and Chowdhary 1987, Jhonson, et al., 1993). NDVI is a transformation of reected radiation in the visible and near infrared bands of a sensor system and is a function of green leaf area and biomass. Computation of NDVI is given by: NDVI = (NIR reectance Red reectance) / (NIR reectance + Red reectance) The severity of drought situation is assessed by the extent of NDVI deviation from its long term mean. Maps produced using relative greenness are quite useful to assess drought situation and hence this indicator is being used widely (Johnson et al., 1993). NDVI shows a lag correlation with rainfall and aridity anomaly. The lag time is about 2-4 weeks. However, the correlation is not unique either through the season or between the areas. The rainfall use efciency varies in time and space making direct satellite monitoring of vegetation development essential for reliable and objective monitoring of agricultural drought. However, conjunctive use of rainfall/aridity anomaly and VI provides greater reliability. The general crop growing period with regard to beginning, peak growth stage and senescence can be identied through the seasonal NDVI prole. However, NDVI can be an indicator of crop development/condition only after signicant spectral emergence of crops, which occurs at about 2-4 weeks after sowings/transplantations. Normalized Difference Water Index (NDWI) Normalised Difference Water Index (NDWI) is derived from two bands including a moisture sensitive SWIR band and insensitive NIR band (Gao 1996). In the beginning of the cropping season, when soil back ground is dominant, SWIR is sensitive to surface wetness of top soil. As the crop progresses, SWIR becomes sensitive to leaf moisture content. When the crop is grown up, the response in SWIR band is mostly from canopy and not from the underlying soil. Indices based on the reectance of Shortwave Infrared (SWIR) bands have been found to be sensitive to moisture available in soil as well as in crop canopy (Tucker and Choudhary 1987, Wang et al., 2008). SWIR band has got no penetrating capability. It provides only surface information. When the crop is grown-up, SWIR response is only from canopy and not from the underlying soil. NDWI using SWIR can complement NDVI for drought assessment particularly in the beginning of the season. Computation of NDWI is given by: NDWI= (NIR reectance SWIR reectance) / (NIR reectance + SWIR reectance) NDWI has been a popular index for crop stress detection and for monitoring moisture condition of crop/vegetation canopies over larger areas (Fensholt and Sandholt 2003, Jackson et al., 2004, Maki et al., 2004, Chen et al., 2005, Gu et al., 2007). The response of NDWI to moisture is instantaneous without any time lag. NDWI is more sensitive to both desiccation and wilting. NDVI is more sensitive to canopy chlorophyll changes and tend to saturate at high biomass levels. In view of the limitations associated with individual indicators either NDVI or NDWI, combination of both the indicators may provide a robust approach for drought monitoring. Combination approach would amplify the anomalies and become more responsive to the ground agricultural situation. Gu et al., 2007, combined NDVI and NDWI to form Normalised Difference Drought Index (NDDI), which was found to be more sensitive to drought conditions over grass lands. Combined use of NDVI and NDWI temporal anomalies has better delineated the rice transplanting areas in China (Xiao et al., 2002).

97

NNRMS

BULLETIN

DECEMBER

2013

compensate for the effects of atmosphere, illumination geometry etc. (Malingreau 1986, Tucker

Short-wave Angle Slope Index (SASI) A new index - Shortwave Angle Slope Index (SASI), adopted from Khanna et al., (2007), is being used in NADAMS project from kharif 2010 for generating a geo-spatial product called Area Favourable for Crop Sowing/crop sown area (AFCS). This product provides the estimates of the potential kharif area that is either favourable for sowing or already sown, progressively from June to August, at state level. Such crop sowing related information is a direct indicator of the impact of early-season agricultural drought. MODIS 500m 8-day composite images of NIR, SWIR 1 and SWIR 2 spectral bands are being used for computing the index. SASI is very sensitive to surface wetness. SASI is calculated as product of the angle at SWIR1 (SWIR1) and the inclination of the line that passes through the NIR and SWIR2 of a triangle with vertices at R (NIR), R (SWIR1) and R (SWIR2), where R is the reectance at broad bands. The slope of line, (RSWIR2RNIR) / (SWIR2NIR) can be approximated by the difference of the reectance at NIR and SWIR2 since the wavelength difference between the two vertices is constant. An advantage of angle indexes is that they are relatively insensitive to albedo differences for comparison between spectra. Prabir et al., (2011) demonstrated that weekly SASI proles at district level revealed seasonality and strong association with rainfall and sown area pattern. In the absence of operational procedures for nearreal-time soil moisture estimation, SASI can act as surrogate parameter to draw inferences on the commencement and progression of crop sowings and to characterize the agricultural situation in the early part of the season. Using multiple criteria SASI, soil texture, rice/non-rice areas, Soil Moisture Index derived spatial soil water balance, a procedure has been evolved to generate spatial product on crop sowing favourable area or already cropped area in the season. This product called Area Favourable for Crop Sowing (AFCS), has been generated on fortnightly or monthly scale from June-September. After validation with state level crop sown area statistics reported by respective agriculture departments, the product is generated and used operationally from kharif 2010. The cloud covered pixels in SASI images are resolved with the support of rainfall and water balance derived soil moisture index. Extending SASI for rabi season monitoring, Murthy et al., 2012, mapped spatial patterns of surface wetness in the transplantation period of rabi season of recent 10 years (2002-03 to 2011-12) in a rice dominant irrigated command area. OCM 2 NDVI Ocean Colour Monitor (OCM) sensor onboard Oceansat-2 satellite provides data in eight spectral bands located in blue, green, red and near Infrared regions of electromagnetic spectrum. The satellite covers entire India in two paths with each path data being available on alternate days. Images of large swath (1400 km) of either of the two paths are available every day. Although the satellite is mainly meant for Ocean applications, certain features of the data like wide swath, high temporal repitivity and availability of visible and infrared channels data make it equally useful for land applications particularly for agricultural drought monitoring. Keeping in view the advantages, OCM data is being used operationally in NADAMS project, from kharif 2011. The cloud free data from all the overpass scenes during the season have been procured. After geometric correction of each scene, the radiance data is corrected for Rayleigh scattering using the data of sun and sensor zenith and azimuth angles. Top-of-the-atmosphere (TOA) reectance is generated for each and cloud masking is done using the data of green, red and NIR bands. NDVI is generated with cloud masked TOA data of band 6 (red) and NIR (band 8) for each overpass date. Fortnightly and monthly time composites of Rayleigh corrected TOA NDVI are generated using

fortnightly/monthly scales. Narrow spectral band width of OCM images causes saturation problem, particularly of red reectance,well before the crop reaches the maximum greenness. As a result NDVI tends to saturate and reaches its maximum value by the end of August or rst fortnight of September. Therefore, OCM2 NDVI anomalies need to be interpreted with caution during September/October months. Soil Moisture Index (SMI) from Soil Water Balance (SWB) Model 30 cm prole soil moisture. This model considers the initial root depth of 30 cm throughout the season to capture the soil water scenario for crops sown and germinating during any part of the cropping season. The soil water balance in the upper layer is governed by daily values of rainfall, runoff, Evapo-Transpiration (ET) and drainage to the second layer. When the upper layer saturates in excess of Field Capacity (FC) due to rainfall, the excess water percolates to the lower passive root zone and are instantaneously redistributed in that zone. The excess soil water in the passive zone, moves out as deep percolation. Since the upper 30 cm is considered for the soil water assessment the lower limit of soil water is the residual water content of the soil as the upper layer is exposed to the atmosphere and subjected to upward ux due to the direct solar radiation. The climatic, soil and crop parameter are the main inputs for the SWB. The daily near real time TRMM 3B42RT spatial rainfall product and the daily global potential evapo-transpiration data are used as the rainfall and climatic input, respectively. The soil information was derived from the 1: 0.5 million scale NBSS&LUP soil map. Since this model does not take into account the irrigation applied from various sources, the results of the model should be considered over rainfed areas alone. The Soil Moisture Index (SMI) derived is dened as the proportion of the difference between the current soil moisture (SM) and the Permanent Wilting Point (PW) to the Field Capacity (FC) and the Permanent Wilting Point. The index values range from 0 to 100 with 0 as extreme dry condition and 100 as extreme wet condition. SMI = (SM-PW / FC-PW) x 100 Gridbased Vegetation Index (VI) Products Historically, coarser resolution VI products have been evolved for quick assessment of vegetation changes and drought detection over larger areas countries and continents. The LAC data of AVHRR has given rise to GAC data of 4 km resolution for global change detection studies. The GIMMS data of 8 km resolution has been widely used across the globe for various applications related to agriculture, hydrology and climate change. In NADAMS project, monthly composite 1 km NDVI images derived from NOAA AVHRR, 1 km NDWI images derived from Terra MODIS data and OCM2 NDVI were transformed to generate 5 km Grid images. Grid VI images were generated for each month from June to November for historic years (2009, 2010 and 2011) also along with current year. Monthly NDVI/NDWI anomalies for different months of kharif 2012 were also computed. These Grid images are useful for rapid assessment of agricultural situation during each month. These NDVI / NDWI deviation images are to be integrated with rainfall, soils, cropping pattern for effective interpretation of the agricultural situation. The grid images are also useful as inputs in the modeling tasks such hydrological, land surface and energy balance models. All the grid VI products, their anomalies and interpretation were disseminated to user community through Bhuvan portal.

99

NNRMS

BULLETIN

A simple book keeping bucket type water tight model was developed to derive the top

DECEMBER

2013

MVC approach to minimize cloud cover in the NDVI. NDVI anomaly maps have been generated on

Agricultural Drought Assessment in NADAMS Project The assessment of agricultural drought situation in each district / block / taluk takes in to consideration the following factors; (1) seasonal NDVI / NDWI progression i.e transformation of NDVI / NDWI from the beginning of the season, (2) comparison of NDVI / NDWI prole with previous normal years relative deviation and vegetation condition index, (3) weekly rainfall status compared to normal and (4) weekly progression of sown area. The relative deviation of NDVI / NDWI from that of normal and the rate of progression of NDVI / NDWI
Fig. 1: Overview of National Agricultural Drought Assessment and Monitoring System (NADAMS)

from month to month gives the rst level indication about the agricultural situation in the district. Ancillary data on rainfall,

soil moisture index, crop sown area, cropping pattern, irrigation support is anlaysed to attribute the VI anomalies to agricultural drought situation. The ground data from different states has been organized in to a data base along with satellite derived NDVI / NDWI data. During June to August, the extent of crop sowing favourable area/crop area against normal kharif area in each state is assessed using AFCS product derived from SASI, modeled soil moisture and other ancillary data. AFCS based crop sown area progression is useful to detect the intensity of crop-sowing period drought. During June to August, drought warning information is issued in terms of Watch, Alert and Normal categories. In case of Watch, external intervention is required, if similar drought like conditions persist during the successive month, while Alert warning calls for immediate external intervention, in terms of crop contingency plans. During September and October, based on NDVI anomalies corroborated by ground situation, drought declaration is done in terms of Mild, Moderate and Severe drought. The mandals/taluks of mild agricultural drought category are characterised by about 10-20% reduction in NDVI and NDWI persistently for more than a month. The agricultural situation of this class represents one or more of the attributes - slightly reduced crop sown area or slightly reduced vigour of crops leading to slight reduction (about 10%) in crop yield. For moderate agricultural drought category, mandals/taluks are characterised by more than 20-30% reduction in NDVI and NDWI persistently for more than a month. The agricultural situation of this class represents one or more of the following attributes more than a month delayed sowing time, more than 25-50% reduction in crop area, poor greenness/moisture levels of crop vegetation, signicant reduction in crop yield. In case of mandals/taluks under severe agricultural drought category, there would be more than 30% reduction in NDVI and NDWI persistently for more than a month. The agricultural situation of this class represents one or more of the following attributes more than a month delayed sowing time, more than 50% reduction in crop area, poor greenness/moisture levels of crop vegetation, signicant reduction in crop yield.

Departments of Agriculture and Relief of different states in addition to the Department of Agriculture and Cooperation, Ministry of Agriculture, Government of India. Whenever need arises, drought information is also disseminated on fortnightly basis, subject to availability of cloud free satellite data. Feedback received from the states indicates that the drought reports are being used as inputs in their review meetings on agricultural situation. The agricultural drought information of NADAMS reports are being used as inputs in the development of contingency plans and in relief management. It was also found that there is good correlation between NDVI images of NADAMS and aridity maps being provided by India Meteorological Department. NADAMS Assessment in Kharif 2012 During kharif 2012, drought assessment under NADAMS project was carried-out jointly by NRSC and MNCFC. Satellite data analysis and generation of different products was done by MNCFC. SASI and water balance derived geospatial product Area Favourable for Crop Sowing/ Crop sown area (AFCS) has progressively increased 44 mha in June, 87 mha in July, 98 mha in August and 106 mha in September. The normal kharif crop area in
Fig. 2: Agricultural drought assessment under NADAMS project, kharif 2012

the country is 109 mha. Thus 97% of normal area was covered in the season, although it was delayed in Andhra Pradesh, Karnataka,

Rajasthan, Gujarat and parts of Karnataka due to decit rainfall. The unfavourable area of 3 mha was mostly located in Karanataka, Gujarat and Rajasthan states. Seasonal proles of NDVI and NDWI from June to October indicate normal agricultural situation in many parts of the country - Chhattisgarh, Orissa, Madhya Pradesh, Jharkhand, Uttar Pradesh, Bihar, Tamil Nadu, Northern half of Andhra Pradesh, West coast of Karnataka, Vidarbha region of Maharashtra, Eastern Rajasthan and Eastern Gujarat. Less than normal NDVI and NDWI and reduced rate of transformation from June to October, signifying delayed crop sowing/reduced crop area/poor crop growth are observed in southern parts of Andhra Pradesh, northern and southern Karnataka, Marathwada and Madhya Maharashtra regions of Maharashtra, West Gujarat, Southern Haryana and Western Rajasthan. In many parts of Karnataka state, Marathwada and Madhya Maharashtra regions of Maharashtra, West Gujarat, Southern Haryana, Western Rajasthan and in parts of Andhra Pradesh, the extent of crop sown area and/or the vigour of already sown crops was signicantly less than that of kharif 2011 or kharif 2010 indicating agricultural drought situation. The trends of different crop indices were normal in many states indicating progressive improvement in agricultural situation from June. At the end of October month, Normal agricultural situation was evident in 316 districts followed by Mild agricultural drought in 43 districts and moderate agricultural drought in 51 districts (Figure 2). 101

NNRMS

BULLETIN

DECEMBER

2013

Monthly drought reports from June to November are being prepared and disseminated to the

Gujarat, Karnataka and Maharashtra were the states of concern, with signicant number of districts under Moderate drought, followed by Haryana, Rajasthan and Andhra Pradesh states. Sub-district level assessment in four states namely, Andhra Pradesh, Karnataka, Maharashtra and Haryana, agricultural drought assessment was done using Resourcesat AWiFS derived indices along with other indices, as shown in Table 1: Table 1 : Agricultural drought situation in kharif 2012 at sub-district unit level for 4 states State Andhra Pradesh Haryana Karnataka Maharashtra Sub-district unit Normal Mandal Blocks Taluk Taluk 687 52 55 223 No. of sub-district units affected Mild 201 30 53 59 Moderate 165 39 68 73

In Andhra Pradesh state, 165 mandals were categorised under Moderate drought class followed by 201 mandals in Mild drought class and 687 mandals in Normal class. In Haryana, 39 blocks were categorised under Moderate agricultural drought class followed by 30 blocks under Mild agricultural drought class and 52 blocks in Normal class. The Moderate agricultural drought class blocks mostly correspond to southern part of the state - Hisar, Bhiwani, Mahendragarh, Rewari, Jajjar and Mewat districts. In Karnataka, 68 taluks were categorised under Moderate agricultural drought class followed by 53 Taluks under Mild agricultural drought class and 55 Taluks under Normal class. The Taluks under Moderate drought class were distributed all over the state except in the west coast region. In Maharashtra, 73 taluks were categorised under Moderate agricultural drought class followed by 59 Taluks under Mild agricultural drought class and 223 Taluks under Normal class. The Taluks under Moderate drought were distributed mostly in Madhya Maharashtra and Marathwada regions of the state. Poor crop growth in parts of Karnataka, Gujarat, Maharashtra, Andhra Pradesh Rajasthan and Haryana may result in the reduction of crop yield during kharif 2012.

Challenges in Drought Assessment

Because of the complexities of drought, no single index has been adequate to capture the intensity and severity of drought and its potential impacts. Different states are adopting different methodologies for drought assessment, preparation of drought memorandum, drought declaration and relief assessment in India. The criteria adopted in different states also vary depending on the rainfall and crops grown in the region. Keeping in view the need for rationalization of drought criteria and adoption of uniform and integrated approach, the Ministry of Agriculture, Government of India, broughtout a drought manual in 2010 after conducting wide range of discussions with different experts. The manual is available at www.agricoop.gov.in. The main challenges in the operational drought assessment are discussed in subsequent sections. Integrated Approach Development of a unied index for drought severity assessment by integrating the data from different sources is an important challenge. There is a need to arrive at a scientically acceptable indicator of drought for the country. The index should give appropriate weightages to the rainfall, soil moisture and crop condition and to make the criteria uniform irrespective of region or state. Standardized departures from normal of different indices need to be blended to characterize the drought intensity. Empirical models, process based models, satellite data and ground surveys with sampling techniques need to be explored in this context.

An early warning system should provide the information about the onset, progress and end of drought conditions to the decision makers at all levels. Early warning systems help in formulating drought intervention strategies that respond to the needs of the people and enables individuals/ community to face the risk with reduced damage. A good early warning system should have a composite database on meteorological conditions, agricultural situation, production estimates, availability of drinking water, fodder, price trends of food and feed etc. Vulnerability prole of the area should also form an important component of early warning methodology. Remote sensing data provides wide ranging inputs to drought early warning system. Time series data on different vegetation indices are extremely useful to detect the response patterns of agricultural crops to weather variations. Satellite meteorology provide inputs to all the three types rainfall prediction systems currently in operation in India (Roy et al., 2006). In the long term rainfall prediction, using global and regional atmosphere, land and ocean parameters, the remote sensing data from geo-stationary and polar orbiting weather satellites, such as, INSAT, NOAA etc is being ingested directly or through parameterization. In the medium range weather prediction, the NCMRWF uses satellite based SST, NDVI, snow cover area and depth, surface temperature, altitude, roughness, soil moisture at surface level and Tiros Operational Vertical Sounder (TOVS) and Radio sonde data on water vapour, pressure and temperature at vertical prole data in the T86/NMC Model. However, at present only global data with poor spatial resolution is being used. In the short range rainfall prediction also INSAT based visible and thermal data is being used.

Conclusion
Agricultural drought is a complex hydro-meteorological disaster. Occurrence of agricultural drought is determined by a number of parameters rainfall, soil moisture, cropping pattern, crop stage etc. State departments of Agriculture and Relief are adopting a number of parameters and indices for monitoring and management of drought situation. Geospatial approach for drought assessment involving different satellite derived indices, ground measured data and derived indices and their integration brought rationality, objectivity, spatial and temporal perspective to agricultural drought assessment. NADAMS project of NRSC successfully incorporated different satellite derived indices, geospatial products to represent soil moisture and crop status during the season, showcased the operational robustness of geospatial approach and strengthened the drought assessment mechanism in the country. Drought impact assessment and early warning are some of the yet-to-be operationalised issues to take the drought monitoring endeavor to next level and to achieve weather resilient rain-fed agriculture in India.

Acknowledgements
We express our sincere gratitude to Dr. V.K. Dadhwal, Director, National Remote Sensing Centre for his constant encouragement and guidance. Valuable suggestions offered by Dr. V. Raghavaswamy, Deputy Director (Training), NRSC are sincerely acknowledged.

References
Batista, T. T., Shimabukuro, Y.E. and Lawrence, W.T. (1997). The long term monitoring of vegetation cover in the Amazonian region of northern Brazil using NOAA-AVHRR data, International Journal of Remote Sensing, 18, pp.3195-3210. 103

NNRMS

BULLETIN

DECEMBER

2013

Early Warning Systems

Chen, D., Huang, J. and Jackson, T.J. (2005). Vegetation water content estimation for corn and soyabeans using spectral indices derived from MODIS near- and shrot-wave infrared bands, Remote Sensing of Environment, 98,pp.225-236 Fensholt, R. and Sandholt, I. (2003). Deviation of a shortwave infrared water stress index from MODIS near and short wave infrared data in a semi-arid environment, Remote Sensing of Environment, 87, pp.111-121. Gadgil, S., Vinayachandran, P.N. and Francis, P.A. (2003). Droughts of the Indian summer monsoon: Role of clouds over the Indian ocean, Current Science, 85, pp.1713-1719 Gao, B.C. (1996). NDWI A Normalised difference water index for remote sensing of vegetation liquid water from space, Remote Sensing of Environment, 58,pp.257-266 GuYingxin, Brown F Jesslyn, Verdin P James and Wardlow Brain. (2007). A ve year analysis of MODIS NDVI and NDWI for grass land drought assessment over central Great Plains of the United States, Geophysical Research Letters, 34, L06407, doi:10.1029/2006GL029127. http://ltpwww.gsfc. nasa.gov/bsb/hqvisit/16MAR2004/tucker.GIMMS_FEWS.pdf. Jackson, J. t., Chen, D., Cosh, M., Li, F., Anderson, M., Walthall, C., Doraisamy, P. and Hunt, E.R. (2004). Vegetation water content mapping using Landsat data derived from normalized difference water index for corn and soybeans, Remote Sensing of Environment, 92, pp.475-482. Jhonson, E., Gary., Rao Achutuni, V., Thiruvengadachari, S. and Kogan Felix, (1993). The role of NOAA satellite data in Drought early warning and monitoring: Selected studies, in Drought Assessment, Management and Planning: Theory and case studies, edited by Donald A. Wilhite, Kulwer Academic Publishers. Khanna, S., Palacios-Orueta, A., Whiting, M. L., Ustin, S. L., Riano, D. and Litago, J. (2007). Developement of angle indexs for soil moisture estimation, dry matter detection and land-cover discrimination. Remote Sens. Environ. 109, pp.154-165. Maki, M., Ishiahra, M. and Tamura, M. (2004). Estimation of leaf water status to monitor the risk of forest res by using the remotely sensed data, Remote Sensing of Environment, 90, pp.441-450. Malingreau, J.P. (1986). Global vegetation dynamics: Satellite observations over Asia, International Journal of Remote Sensing, 7, 1121-1146. Minamiguchi Naoki. (2005). The application of Geospatial and Disaster information for food insecurity and agricultural drought monitoring and assessment by the FAO GIEWS and Asia FIVIMS, Workshop on Reducing Food Insecurity Associated with natural Disasters in Asis and the Pacic, Bangkok, Thailand, 27-28 January 2005. Moulin, S.A., A. Bondeau and R. Delecolle. (1998). Combining agricultural crop models and satellite observations: from eld to regional scales, International Journal of Remote Sensing, 19, pp.1021-1036. Murthy, C.S., SeshaSai, M.V.R., BhanujaKumari. V. and Roy, P.S. (2007). Agricultural drought assessment at disaggregated level using AWiFS/WiFS data of Indian Remote Sensing satellites, Geocarto International, 22, pp.127-140 Murthy, C.S., Abid S.M., Prabir Kumar Das and SeshaSai M. V. R. (2012). Tracking surface wetness and rice transplantation using Shortwave Angle Slope Index (SASI), Proc. of ISRS Symposium, New Delhi 2012. Prabir Kumar Das., Murthy C.S., SeshaSai M.V.R. and G Behera. (2011). Analyzing seasonal dryness/wetness conditions during crop sowing period using Shortwave Angle Slope Index, Proc. of ISRS Symposium, Bhopal. Rajeevan, M. and Pai, D.S. (2006). On El Nino-India monsoon predictive relationships, Research Report no.4/2006, National Climate Centre, IMD, Pune. Roy, P.S., Joshi, P.C., Murthy, C.S. and Kishtawal C.M. (2006). Geoinformatics for Drought Assessment, in Drought Management Strategies in India (Eds. J.S. Samra, Gurbachan Singh & J.C. Dagar), pp. 23-60. ICAR, NEW Delhi, India.

of classication of agricultural droughts. Arch. Fur. Met.Geoph.And Biokl. Ser. B. 29: pp.293-297. Subbaiah, A.R., (2004). State of the Indian Farmer A Millenium study, Natural Disaster Management, vol. 21, Published by Department of Agriculture and Cooperation, Ministry of Agriculture, Government of India, New Delhi. P226. Thenkabail, P. S., Gamage, M.S.D.N., and Smakhtin, V.U. (2004). The use of remote sensing data for drought assessment and monitoring in Southwest Asia, International Water Management Institute (IWMI) Research Report no.85, Sri Lanka. Tucker, C.J., Townshend, J.R.G. and Goff, T.E. (1985). African land covers classication using satellite data, Science, 227, pp.369-375. Tucker, C.J., and Chowdhary, B.J. (1987). Satellite remote sensing of drought conditions, Remote Sensing of Environment, 23, pp.243-251 Wang, Lingli., Qu, John., J., Hao, Xianjun., and Zhu, Qingping. (2008). Sensitivity studies of the moisture effects on MODIS SWIR reectance and vegetation water studies, International Journal of Remote Sensing, 29, pp.7065-7075 Xiao, X., Boles, S., Frolking. S., Salas, W., Moore III., and Li, C. (2002). Observation of ooding and rice transplanting of paddy rice elds at the site to landscape scales in China using Vegetation sensor data, International Journal of Remote Sensing, 23, pp.3009-3022.

105

NNRMS

BULLETIN

DECEMBER

2013

Sastry, A.S.R.A.S., Ramakrishna, Y.S. and Ramana Rao, B.V. (1981). A new methods

Introduction
Irrigated agriculture accounts for more than 70 percent of water withdrawn and is the major consumer of fresh water resources. Global population continues to swell, increasing the demand for food and bre, demanding agricultural intensication, increasing water use, especially in arid and semi-arid regions. Arid and semi-arid regions, where in precipitation occurs over few months, irrigation support is essential for achieving optimal crop productivity. Irrigated agriculture is under severe scrutiny to produce more with fewer inputs, specically, the water. The other sectors of water utilization are posing great challenge and competition to irrigation to maximize its water use efciency. Anticipated climate change conditions are expected to alter water availability and demand conditions. Irrigated agriculture is projected to be one of the most vulnerable sectors, requiring adoptive technologies and management. Programs such as Accelerated Irrigation Benet Programme (AIBP) augment the irrigation potential creation. Programs such as Command Area Development (CAD), National Water Management Project (NWMP), Water Resources Consolidation Project (WRCP), National Pilot Project for Repair, Renovation and Restoration (NPRRR) were initiated for improving the performance of existing irrigation systems to bridge the gap between potential created and utilized and to improve overall water use efciency/productivity. Use of satellite remote sensing data for irrigation water management has been demonstrated through many studies addressing: base line inventory, performance assessment & monitoring, providing in-season inputs, monitoring physical progress of potential creation, generating inputs for feasibility assessment of new projects, environmental impacts such as water logging & soil salinity, reservoir management, etc. This would support the eld departments to cope up with water scarcity and augmenting the water use efciency through integration of geo-spatial information with their conventional practices.

Satellite Remote Sensing for Irrigated Command Area Management

Satellite Remote Sensing (SRS) & Geographic Information System (GIS) techniques have many roles to supplement and complement the data needs of irrigation sector to equip them for efcient management of existing schemes as well as for scientic planning of new schemes (Bastiaanssen et al., 2000).

NNRMS

BULLETIN

Raju PV, Abdul Hakeem K and Venkateswar Rao V Water Resource Group, National Remote Sensing Centre ISRO, Department of Space, Hyderabad - 500 037, India Email: raju_pv@nrsc.gov.in

DECEMBER

IRRIGATION COMMAND AREA MANAGEMENT USING REMOTE SENSING

2013

The high spatial resolution satellite data, of the order of few meters to sub-meter, (also equipped with stereo capability) is providing inputs for evaluating the feasibility of various alternatives while planning for new irrigation schemes. In ongoing projects, where new irrigation potential is under creation, high spatial resolution data is being used to capture the irrigation infrastructure, thus equipping the funding agencies to monitor the physical progress and accomplishment of the targets. Alternatively, such data is also found useful to generate geo-databases required for development of information & decision support systems. The intermediate spatial resolution satellite data is being extensively applied to existing irrigation systems to carry out baseline inventory, performance evaluation, impact assessment, near-real time monitoring, in-season estimation of irrigation water demand, surface water logging, soil salinity/alkalinity, etc. The large archives of historical satellite data provide valuable evidence of the responses and performance of irrigation system to varying water availability conditions. Near real time satellite data provides the current information on irrigated agriculture, which can be used to take appropriate in-season decisions in order to reduce the impact of water scarcity on agricultural production. These inputs are assisting irrigation department to improve and stabilize the performance of the existing irrigation systems. Satellite Remote Sensing data captures the information both command level and at different spatial units, thus providing the variability at sub-command units. This capability signicantly enhances the usefulness of satellite data for deriving decision variables at various hierarchal units. The repetitive coverage through time-series satellite data captures the temporal dynamics of agriculture, thus providing opportunity to monitor and derive decision variables during the season.

Baseline Inventory of Irrigation Systems

There has been a steep decline of new investments for expanding irrigation sector due to rapidly increasing capital investment and due to environmental problems associated with such infrastructure creation. In India, the funding for irrigation sector has sharply declined from 23% in mid 60-70s to a low of 9% in mid 90s. The focus has shifted towards improving the efciency of existing irrigation systems instead of creating new systems. In most of the existing irrigation systems there is a serious lack of systematic organizational structure providing/maintaining data pertaining to the system. Effective irrigation water management needs reliable, comprehensive and objective data base in a timely and cost-effective manner. Satellite data with its synoptic coverage coupled multi-spectral information and time series data sets are suitable for inventorying the irrigation systems (Thenkabail et al., 2009). Crops cultivated in irrigated command areas were identified and inventoried through analysis of multi-spectral optical remote sensing data and through digital image processing algorithms (Jonna & Chari, 1992; Nageswar Rao and Mohankumar, 1994) and also using microwave radar data (Saindranath et al., 2000). Baseline information on cropping pattern was generated using remote sensing data from command level to water course level. Multi-temporal optical and microwave (Ozdogan, et. al., 2010) data were used to identify multiple crops in irrigated agricultural system. Murthy et al., (2003) used advanced classiers like ANN back-propagation technique for classication of irrigated crops. Satellite data derived spectral indices have been used to evaluate crop condition. Some of the indices like Normalised Difference Vegetation Index (NDVI) were found to be directly related crop yield and thus were used for estimation of crop yield of cereal crops (Murthy, et al., 1996). NDVI was also used for ground sampling of crop cutting experiment in irrigation system (Murthy, et al., 1996). Satellite technology tools were applied to evaluate the schemes such as National Water Management Project (NWMP) and Water Resources Consolidation Project (WRCP).

Case Study Inventory of Bhakra Irrigation System, Haryana State

Multi-date satellite data of 1995-96 rabi season were analysed to assess irrigation system under Bhakra canal

low performing pockets, effectiveness and sustainability of improvement schemes, etc. Some of the performance indicators generated from satellite data are crop intensity, equivalent crop area intensity, principal crop intensity, proportionate crop intensity, crop condition, coefcient of variation in crop condition, tail-head ratio of cropping intensity, tail-head ratio of crop condition and sustainability in crop intensity. Remote sensing based performance indicators were used for evaluating the performance of various irrigation systems in the country (Thiruvengadachari et al 1994 and Raju et al., 1997). Bastiaanssen et. al., (1999) listed the performance indicators derived from RS algorithms supplemented by ground data. Ray et al., (2002) used RS data has to compute three indices namely, adequacy, equity and water use efciency for the evaluation of performance of distributaries in an irrigation system. Panigrahy et al., (2005) attempted to derive crop indices like Multiple Cropping Index, Area Diversity Index and Cultivated Land Utilization Index using satellite derived parameters such as cropping pattern, crop rotation, and crop calendar, crop type, acreage, rotation and crop duration. Command Area Development (CAD) scheme under 13 irrigation commands was evaluated using multi-year satellite data (Anonymous, 2005).

Case Study Bhadra Command Area, Karnataka State

Study carried out in Bhadra project command area in Karnataka State demonstrated the capability Satellite Remote Sensing techniques in providing spatial information on irrigated area, cropping pattern, crop productivity and crop water use efciency at micro level. Analysis of multi-date satellite data (IRS LISS I sensor) during 1992-93 rabi season along with field data has indicated 91 percent irrigation intensity with 66 percent coverage under paddy crop. Figures 2a and 2b depict canalwise paddy yield and paddy
Fig. 2a: Canal-wise Paddy yield during rabi (1992-93)

water use efciency respectively during rabi 1992-93. The total

3

depth of water application was 0.799 m with a water use efciency of 0.495 kg/m for paddy crop. Water distribution was found to be more inequitable in Bhadravathi division and was also having low paddy water use efciency of 0.394 kg/m 3. Malebennur division reorded higher performance with more uniform water distribution, high paddy productivity and high paddy water use efciency. Distributaries with low irrigation intensity, paddy yield and water use efciency could be identied. The study established the usefulness satellite data for assessing the performance of an irrigation system and in identifying the poorly performing pockets which require improvement measures (Anonymous, 1994).

Multi date satellite data of three rabi seasons (1997-98, 1990-91 and 198687) were used to assess the performance of Chambal irrigation command, Rajasthan state, at distributary/minor irrigation unit temporal information on cropping pattern, crop intensity and crop condition formed the basic inputs to develop the indicators of agricultural performance of the irrigation system. The total irrigated area during 1997-98 rabi season was higher than the irrigated area during 1990-91 and 1986-87 (Figures 3a, 3b, 3c and 3d). The increase was observed in both right main canal and left main canal. However, the irrigation intensity was observed low in Left Medium Canal
Fig. 2b: Canal-wise irrigation intensity

(LMC) indicating signicant gap between irrigation potential created and utilised. During 1997-98 rabi season, the water requirement is increased by 31.98 percent (in terms of Equivalent Wheat Area), mostly resulting from the signicant increase in wheat crop extent, when compared with 1990-91 rabi season. The supplies were increased by only 6.48 percent, resulting in highly variable wheat crop condition during 1997-98 rabi season (Anonymous, 2005)

In-season Inputs for Improved Irrigation Water Management

Indian irrigation systems have been traditionally designed and constructed with
Fig. 3a: Standard FCC of rabi (1998)

a minimal consideration of dynamic system operation and control capabilities. With the

growing competition for fresh water resources from other sectors, irrigation management is faced with increasing needs for more exible, reliable and efcient supply regime in order to achieve maximum efciency. To maintain control over the process of delivering water, real time information is to be obtained on various aspects, which control and inuence the supply & utilization regimes. 111

NNRMS

BULLETIN

level. The satellite derived spatial and

DECEMBER

Case Study Chambal Command Area, Rajasthan State

2013

Irrigation managers are constrained by the lack of real time information on - to what extent irrigated agriculture is confirming with their plans and the extent of deviations, if any. Such information, when provided during supply time, would equip the managers to make real-time decisions and to sensitize the release pattern in accordance with demand variability and its sensitivity. Field based studies have been developed to estimate or forecast irrigation water requirements (Li and Cui 1996, Pulido-Calvo et al., 2003) and to optimize water allocations (Vedula and Muzumdar 1992; Wardlaw and Barnes 1999; Westphal et al., 2003). In most of these studies the
Fig. 3b: Crop map of rabi 1998

mixed cropping was either predetermined or derived through optimization approaches and irrigation water allocations were optimized considering the system resources, operating principles and associated constraints. It is important to note that year to year variations in water availability and corresponding variations in distribution policies induce signicant variability in eld conditions and agricultural operations.

Near Real-Time Inputs from Remote Sensing Data

Satellite remote sensing data captures the information both at command level and at different spatial units, thus
Fig. 3c: Canal-wise Irrigation Intensity

providing the spatial variability at subcommand units. This capability signicantly

enhances the usefulness of satellite data for deriving decision variables at various hierarchal units. The repetitive coverage through time-series satellite data captures the temporal dynamics, thus providing opportunity to monitor and derive decision variables during the season. Time-series satellite data during the irrigation season can provide various sets of information: capturing the onset and extension of irrigation service; progression of cropped area; area under major crops/ crop-groups; crop and irrigation water requirements and crop condition/productivity. Monitoring the onset and extension of irrigation service : Irrigation supplies provided through a canal system, in general, get extended from head to tail reaches. In the initial time periods, supplies conne to main

Progressively the supplies get extended to the lateral canals and distribution system. The pattern in which the access to the irrigation supplies is taking place has a significant inuence on the amount of water usage and corresponding efficiency. Irrigation managers evolve a programmatic schedule irrigation to all parts of the command area. It is important and desirable to monitor the actual pattern of access to irrigation service and verify whether it is as per designed plans or not. Because, deviations, if any, results in
Fig. 3d: CV of wheat crop condition

inefcient water usage and would call for a revision allocation schedules.

Figure 4 shows the time series AWiFS data capturing the initiation of irrigation service and its extension. The time-series data depict False Colour Composite (FCC ) images of NIR-SWIR-Red over Hirakud command area, Orissa State during 2003-04 rabi season. The ability to characterize moisture status of a soil by multi-spectral information through remote sensing has many practical applications. The high frequency AWiFS data Resorcesat-1 visibly depicted the irrigation service initiation and progress across the command area. Satellite data of December shows fallow elds with no signicant residual soil moisture before the irrigation service and no rainfall during the preceding days. The irrigation supplies, which commenced in the last week of December, provided initial wetting of soil (blue to purple colour areas) indicating the access to irrigation service (Figure 4). This was mostly conned to a few areas within the vicinity of main canals. Subsequent dates satellite data depict the progressive increase in wetted area. The irrigation supplies were observed to extend to tail portions of the
Fig. 4: Onset & extension of irrigation service as captured by multi-date AWiFS data (FCC using NIR-SWIR-red)

command and into lateral distribution system by 10th Feb, The time-series AWiFS data clearly

captured pattern of irrigation service initiation and its extension to different parts of command area (Raju et al., 2008). The above information when generated on near-real time basis could assist the irrigation department to verify with their actual plans and identify the deviations. A comparison with planned

113

NNRMS

BULLETIN

through which they plan to provide access to

DECEMBER

2013

canal and its immediate surrounding areas.

irrigation schedules would help the managers to take necessary steps for minimizing the deviations from original plans. Estimation of Intra-seasonal Irrigation Water Requirements: Figure 5 shows the multi-date AWiFS data over Hirakud command area during 2003-04 rabi season. The temporal satellite data set clearly depicts the paddy crop area progression and its phenological cycle beginning from field preparation/ transplantation, active vegetative phase, heading, etc. From the progression of rice crop acreage, the variability in rice transplantation period was identied. Distinct classes representing different periods of rice transplantation are shown in Figure 6. The satellite data analysis facilitated statistics generation right up to distributary (tertiary) canal thus providing the spatial variability among irrigation units. This information was integrated with agro-meteorological data, to derive lateral canal-wise irrigation water demand and the critical periods were also identified. A comparison with the actual
Fig. 5: Multi-date AWiFS data over Hirakud command area

supply pattern (Figure 7) indicated poor correlation with the chronological variations

associated with crop water requirement, supplies were 15% excess during Dec-Jan and were 20.1% decit during later part of season (Feb to Apr). The study has demonstrated the usefulness of time series AWiFS data to generate the irrigation water requirement during the supply season and would support the irrigation managers to reschedule the irrigation water supplies to achieve better synchronization between requirement and supply leading to improved water use efciency.

Water Logging and Soil Salinity in Irrigation Systems

Water logging and subsequent salinization and alkalization are the major land degradation processes operating upon in the irrigation commands of the semiarid regions. The signicant occurrence of salt affected soils lies in the arid and semiarid regions reducing considerably (78%) the productive capacity of the land surface in the world. Due to improper management of soil and water resources in the command areas, the problems of salinity/ alkalinity and water logging are reported to be on the increase. Information on the nature, extent, spatial distribution and temporal behaviour of areas under water logging and salinity/alkalinity is essential for proper management of irrigated lands. Satellite data are being used regularly for mapping of salt affected soils (Singh & Dwivedi, 1989) and waterlogged areas (Sharma & Bhargava, 1987; Command Area Development (CAD) programme, the Ministry of Water Resources, Government of India, supported a programme to apply satellite remote sensing techniques to generate distributary-wise information

alkalinity periodically during selected years of operation in selected command areas. The information on nature, extent, and spatial distribution of waterlogged area and saltaffected soils was derived through systematic interpretation of satellite data. State-wise salt affected soils map of India on 1:250,000 scale were prepared using remote sensing data jointly with the Central Soil Salinity Research Institute (ICAR) and National Bureau of Soil Survey and Land Use Planning, Nagpur (NBSS & LUP). The database contains maps showing physiographic features, distribution and extent of salt affected soils supported by a base map and a descriptive dataset showing nature and degree of salinity/sodicity.

References
Anonymous, (1994). Performance evaluation of Bhadra command area, Karnataka State,
Fig. 6: : Variability in rice transplantation period

Project Report, NRSA, Hyderabad, India Anonymous, (1996). Remote Sensing Study of Bhakra Canal Command Area, Haryana State, Project Report, NRSA, Hyderabad, India Anonymous, (2005). Satellite remote sensing based evaluation study of Chambal irrigation command area, Rajasthan State, Project Report, NRSA, Hyderabad, India. Bastiaanssen, W.G.M. and Bos, M.G. (1999). Irrigation system performance indicators based on remotely sensed data: a review of literature. Irrigation Dranage Systems 13:291-311 Bastiaanssen, W.G.M., Molden, D.J. and

Fig. 7: Estimated irrigation requirement and actual supply under Attabira Canal command during Rabi 2003-04

Makin, I.W. (2000). Remote Sensing for irrigated agriculture: examples from research and possible applications. Agricultural Water Management 46, 137-155. Jonna, S. and Chari, S.T. (1992). Irrigated command area inventory using IRS 1A LISS I data, Proceedings of Interantional Space Year Conference on Remote Sensing and GIS, Feb 26-28, JNTU, Hyderabad, India. Li, Y.H. and Cui, Y.L. (1996). Real-time forecasting of irrigation water requirements of paddy elds. Agricultural Water Management 31, 185-193. Murthy, C.S., Thiruvengadachari, Raju, P.V. and Jonna, S. (1996). Improved ground sampling and crop yield estiamtion using satellite data, International Journal of Remote Sensing, Vol.17, No.5, pp. 945-956. 115

NNRMS

BULLETIN

DECEMBER

2013

on the status of water logging and salinity/

Murthy, C.S., Thiruvengadachari, Raju, P.V. and Badrinath, K.V.S. (2003). Classication of wheat crop with multi-temporal images: performance of maximum likelihood and articial neural networks., International Journal of Remote Sensing, Vol.24, No.23, pp. 48714890. Nageswar Rao, P.P. and Mohan Kumar, A. (1994). Crop land inventory in the command area of Krishnarajasagar project using satellite data, International Journal of Remote Sensing, 15, 1295-1305 Panigrahy, S., Manjunath, K.R. and Ray, S.S. (2005). Deriving cropping system performance indices using remote sensing and GIS. IJRS 26 (12): 2595-2606. Pulido-Calvo, I., Roldan, J., Lopez-Luque, R. and Gutierrez-Estrada, J.C. (2003). Demand forecasting for irrigation water distribution systems. Jl. of Irrigation and Drainage Engineering 129 (6), 422-431. Ozdogan Mutlu, Yang Yan, Allez George and Cervantes Chelsea. (2010). Remote Sensing of Irrigated Agriculture: Opportunities and Challenges,Remote Sensing 2010, 2, 2274-2304. Raju, P.V., Jonna, S., Murthy, C.S., Abdul Hakeem, K., Chari, S.T. (1997). Performance evaluation of a canal delivery system using satellite and eld data, Proc. of workshop on Remote Sensing and GIS Applications in Water Resources Engineering, Bangalore, India, II, 11-20. Raju P.V., Sesha Sai M.V.R. and Roy P.S. (2008). In-season time series analysis of Resourcesat-1 AWiFS data for estimating irrigation water requirement, International Journal of Applied Earth Observation and Geoinformation:10 (2008), 220228. Ray, S.S., Dadhwal, V.K. and Navalgund, R.R. (2002). Performance evaluation of an irrigation command area using remote sensing: A Case Study of Mahi Command, Gujarat, India, Agricultural Water Management, 56(2): 81-91. Saindranath, J., Narasimha Rao, P.V. and Thiruvengadachari, S. (2000). Radarsat data analysis for monitoring and evaluation of irrigation projects in the monsoon. International Journal of Remote Sensing, Vol. 21, No. 17,pp. 32193226. Sharma, R.C. and Bhargava, G.P. (1988). Landsat imagery for mapping saline soils and wet lands in north-west India, International Journal of Remote Sensing, 11, 39-44. Singh, A.N. and Dwivedi, R.S. (1989). Delineation of salt-affected soils through digital analysis of Landsat-MSS data. International Journal of Remote Sensing, 19(1), 83-92. Thenkabail, P.S., Biradar, C.M., Noojipady, P., Dheeravath, V., Li, Y.J., Velpuri, M., Gumma,M., Gangalakuntag, O.R.P., Turral, H., Cai, X.L., Vithanage, J., Schull, M.A., Dutta, R. (2009). Global irrigated area map (GIAM), derived from remote sensing, for the end of the last millennium. International Journal of Remote Sensing 30, 3679-3733. Thiruvengadachari S., Jonna S., Raju P.V., Murthy C.S. & Hakeem K.A. (1994). System performance evaluation and diagnostic analysis of canal irrigation projects. Asian Conference on Remote Sensing, Bangalore, India: B2-1 to B-2-6. Vedula, S. and Muzumdar, P.P. (1992). Optimal reservoir operation for irrigation of multiple crops. Water Resources Research 28(1), 1-9. Wardlaw, R. and Barnes, J. (1999). Optimal allocation of irrigation water supplies in real time. Jl. of Irrigation Drainage. 125(6), 345-354. Westphal, S.K., Vogei, R.M., Krishen, P. and Chapra, S.C. (2003). Decision support system for adaptive water supply management. Jl. of Water resources planning and management May/June, 165-177.

REMOTE SENSING INPUTS FOR FEASIBILITY ASSESSMENT STUDIES OF PROPOSED WATER RESOURCES PROJECTS
Simhadri Rao B, Suresh Babu AV, Shanker M and Venkateswar Rao V Water Resource Group, National Remote Sensing Centre ISRO, Department of Space, Hyderabad - 500 037, India Email: simhadrirao_b@nrsc.gov.in

Introduction
The volume of freshwater resources in the world is about 35 M km3, which is 2.5% of total volume of water (1400 M km3). India has only 4% of worlds renewable water resources but with 17% of worlds population. Water resources has become scare element in the present day scenario with growing population and rising living standards of people. To utilise the water resources in efcient manner, water resources projects are developed. A number of water resources projects have been taken up to manage the available water resources for different purposes such as irrigation, hydropower generation, domestic and industrial consumption, navigation, ood control, recreation, etc. Before initiation of such massive and expensive water resources projects calls for feasibility studies as prerequisite. A feasibility study assesses the viability of the water resources project with an emphasis on identifying potential problems and addressing the needs of the project. Feasibility studies can be of technical and nancial in nature. This paper discusses only the technical feasibility of the projects. The aspects considered while assessing the feasibility is the quantum of water available, demand that has to be meet, structural requirements for storing the required water and availability of sufcient land area under the downstream of the project. Traditionally the tasks that are carried out during the feasibility study (Punmia, 1984) are (i) Hydrological Investigations, (ii) Engineering Surveys and (iii) Geological Investigations. Many studies were carried out for deriving the required information using remote sensing techniques. Yusof et al., (2000) studied development of criteria for locating optimum sites for reservoirs employing satellite remote sensing data in langkawi Island, Malaysia. Ravi Shankar and Mohan (2005) attempted to identify zones for augmentation of groundwater through IRS LISS-III data by deriving drainage pattern, drainage density, lineament density. Based on the derived information and adopting suitable criteria, zones are identied. Singh J.P. et al., (2009) conducted a study to identify suitable sites for water harvesting structures in a watershed in Punjab using IRS LISS-III satellite data and Geographic Information System(GIS). Ramakrishnan et al., (2009) used IRS LISS-III satellite data for identifying the potential water harvesting sites in a watershed by adopting SCS-Curve Number approach. Balachandar et al., (2010) used various satellite derived thematic maps and based on their weightages they had identied suitable sites for articial recharge.

NNRMS

BULLETIN

DECEMBER

2013

This paper presents a case study showcasing the abilities of remote sensing in deriving the necessary information for conducting the feasibility studies.

Case Study
Inter linking of rivers in India is aimed at reducing the regional imbalances in the availability of water by transferring water from surplus basins to decit basins. National Perspective Plan was envisaged for this purpose comprising of two components viz. Himalayan rivers development and Peninsular rivers development. Himalayan rivers development component consists of 14 river links and Peninsular river development component consists of 16 river links (www.nwda. gov.in). Ken-Betwa river link is one such river link under peninsular river development component, envisages transfer of surplus Ken river ows to Betwa river upstream of Parichha weir. About 659 M km3 of water is proposed to be transfered to Betwa river to be utilised through construction of 8 new irrigation projects.

Study Area
The study area covers the upper Betwa basin area and is located in Madhya Pradesh state. It is covered partly by Bhopal, Guna, Vidisha, Sagar and Raisen districts (Figure 1).

Ken and Betwa Rivers

Ken river is a major tributary of Yamuna river, flows in northern part of India and is an interstate river between Uttar Pradesh and Madhya Pradesh. The total catchment area of the basin is 28,058 km 2, out of which 24,472 km2 lies in Madhya Pradesh and remaining 3,584 km2 in Uttar Pradesh. The Betwa river is also an interstate river between Madhya Pradesh and Uttar Pradesh states which originates in Raisen district of Madhya Pradesh near Barkhera village at an elevation of about 576 m above mean sea level. The total catchment area of the basin is 43,895 km2., out of which 30,217 km2 lies in Madhya Pradesh and the remaining 13,768 km2 lies in Uttar Pradesh.

Ken-Betwa River Link project

Ken-Betwa Link Project envisages the diversion of 1,020 MCM of surplus water at Daudhan of Ken basin to Betwa basin. This surplus water is proposed to be utilised for irrigation and drinking water supply in Madhya Pradesh. Out of which, 659 MCM is transferred to Betwa river upstream of Parichha weir and 312 MCM is utilised in the enroute command. The Betwa command proposed to utilise the diverted water by constructing new projects. Eight new projects namely Barari, Neemkheda, Richhan, Makodia, Bebanai, Sindh, Tharr and Kesari were proposed in the upper reaches of the Betwa basin (Figure 1). An area of 1.27 lakh
Fig. 1: Study area of the proposed irrigation projects

ha in Raisen and Vidisha districts of Madhya

Pradesh will be beneted by utilising 659 MCM of water annually (NRSA Report, 2007).

Satellite Data Used

LISS IV MX data of IRS P6 was used to derive information on landuse/landcover and PAN data of Cartosat-1 was used for generation of Digital Elevation Model (DEM).

The methodology involves three major steps, the rst step is derivation of land use / land cover from IRS-P6 LISS IV MX satellite data and second step is generation of contours for selected proposed reservoir levels from Cartosat derived DEM. The third step involves carrying out reservoir submergence analysis and command area analysis using information derived from step 1 and step 2. Figure 2 shows the methodology used for the study. Suitable cloud free IRS-P6 LISS IV MX satellite data was selected for land use /
Fig. 2: : Methodology ow chart

land cover mapping. Ortho-rectied Landsat ETM+ data sets covering study area was downloaded from the Internet. The LISS IV

MX satellite images were geometrically rectied with reference to Land ETM+ data using 2nd order polynomial transformation with sub pixel accuracy using ERDAS Imagine image processing software. Initially, all the images in a particular path are rectied and then mosaiced together. Later all the pathwise mosaic images were mosaiced covering the entire study area. The ancillary maps collected from the project authorities were scanned at a suitable spatial resolution and georeferenced with satellite image of the study area. The boundary of the study area is digitized from
Legend Landuse - Land cover
Built Up-Urban Built Up-Rural Agriculture-Crop land Agriculture-Fallow land Forest-Dense/Closed Forest-Open Forest-Scrub Forest-Blank Guillied/Ravinous Sandy Area Barren Rocky River/Stream Lakes/Ponds Other scrub land

the georeferenced scanned map. The base layer information such as road network, railway network, settlements, drainage network, etc., land use / land cover classes were derived from LISS IV MX satellite image by visual image interpretation techniques. On screen digitisation was done at appropriate scale to generate ArcInfo coverage layers in ERDAS Imagine software. Two levels of land use / cover types were derived from the satellite data as given below: Level-I: Builtup, Agriculture, Forest, Waste land, Waterbodies, other

Fig. 3: Land Use / land cover of the study area

waste and scrub. Level-II: Builtup-Urban, Rural; Agriculture-Crop, Plantations,

Fallow; Forest-Dense, Open, Scrub, Blank, Wasteland-Gullied, Sandy area, Barren rock, Waterbodies, lakes and streams.

119

NNRMS

BULLETIN

DECEMBER

2013

Methodology

As the land use / land cover information was manually derived from satellite image ground truth eld visits were conducted to validate the information. Global Positing System (GPS) receivers are used to record the exact location information. Figure 3 shows land use / land cover derived from the IRS-P6 LISS-IV MX satellite data over the study area.

Cartosat DEM Generation and Derivation of Contours

The study area is covered by 40 Cartosat-1 stereo scenes which is about 22,000 km2. The accuracy of DEM derived from satellite data depends on the accuracy of Ground Control Points (GCP) used. Differential GPS survey was carried out for 110 GCPs within the study area. Levelling survey was conducted at all GCPs to find Mean Sea Level (MSL) levels using surveyed bench mark. These MSL heights were intended to be used during the DEM generation process instead of GPS heights. Figure 4 shows the sliced DEM derived from Cartosat satellite data covering the study area. Selection of dam height and reservoir levels play crucial role in optimum design of the dam. Change in every metre of dam height may vary the hydraulic particulars of the dam. Hence, reservoir submergence analysis, at
Fig. 4: Cartosat DEM of the Study area

every increment of 1 m contours of 1m interval was generated. Also the information derived is

basically used for feasibility assessment, reservoir submergence analysis for every 1 m for the proposed reservoirs.

Reservoir Submergence and Command Area Analysis

For each proposed dam site about 7 to 10 reservoir levels were chosen, out of which, one reservoir level has been taken as the Full Reservoir Level (FRL) depending on the reservoir submergence analysis. The required reservoir level contours were intersected with land use / land cover layer already prepared. Land use / land cover statistics within each reservoir level were extracted. Reservoir elevation-capacity curves were prepared using the generated contours. The reservoir capacity is estimated using contours generated from the DEM by Cone formula as given below.

where V = volume between two levels; h = contour interval; A1 & A2 = area at two successive contours The irrigation command area was delineated based on the Minimum Draw Down Level (MDDL) chosen for each of the dam site using the Cartosat-1 DEM. Contours were generated for each of the selected MDDL of all the dam sites. Land use / land cover statistics were extracted for MDDL by intersecting the derived contour and land use / land cover layer. Geographical Command Area (GCA) and Cultivable Command Area (CCA) within the command for each dam site were estimated.

The land use / land cover results obtained from the satellite data analysis for the entire study area are given below. For the 8 proposed dam sites namely Neemkheda, Barari barrage, Kesari barrage, Makodia, Bebanai, Tharr, Sindh and Richhan, the reservoir submergence analysis and command area analysis was carried out. Out of these dam sites only Bebanai results are produced in the following section.

Land Use / Land Cover Mapping

The study area is predominately rural in nature. The ground truth eld visits were undertaken that within the study area agricultural land (8,91,416 ha) followed by forest (2,94,684 ha) are the major land cover categories. The other land cover categories present within the study area are scrub (31,973 ha), built-up (30,094 ha), waste land (24,038 ha) and water bodies (10,778 ha). for validation of satellite image interpretation. The satellite derived land use / land cover statistics shows

Reservoir Submergence Analysis

The proposed Bebanai dam site is located on Bebanai nadi in Madhya Pradesh state. For this dam site reservoir levels of 438 m to 450 m (13 levels) above mean sea level were chosen for analysis. Figure 5 shows the contours that were generated from Cartosat-1 derived DEM. Table 1 provides land use / land cover area statistics derived from reservoir submergence analysis. It can be observed from the Table that agriculture followed by built-up types are affected by reservoir submergence.

Fig. 5: Cartosat-1 DEM derived contours for various reservoir levels of Bebanai project

Irrigation Command Area Analysis

The proposed irrigation command area was delineated based on the Minimum

Draw Down Level (MDDL) value of 431 m chosen for the Bebanai dam site. Land use / land cover statistics extracted with the MDDL contour shows that there are 3,548 ha. crop land and 6 ha. of fallow land. The total agricultural crop area that can be commanded is 3,554 ha. Hence, The GCA and CCA area estimated as 3,813 ha. and 3,554 ha. respectively.

Comparative Analysis
The reservoir submergence and command area statistics generated for the eight proposed dam sites provide the basis for project authorities to arrive at the feasibility of these projects. A comparative analysis of the submergence and command area statistics shows that out of 8 proposed sites, 4 sites with maximum proposed FRL was found to be feasible while other 4 sites the FRL can be decided based on the remaining available water. Sindh, Richhan, Barari and Kesari sites found to feasible with 121

NNRMS

BULLETIN

DECEMBER

2013

Results and Discussion

Table 1: Statistics of land use / land cover under submergence for different proposed reservoir levels Proposed Reservoir Level (m) Cumulative storage (MCM) 438 439 440 441 442 443 444 445 446 447 448 449 450 10.62 16.32 24.91 37.18 53.71 74.93 101.11 132.46 169.00 210.72 257.78 310.46 369.07 455 694 1,035 1,429 1,889 2,364 2,880 3,398 3,915 4,435 4,981 5,562 6,164 Rural 7 13 20 28 43 59 76 94 111 123 136 151 173 Cultivable 414 647 969 1,349 1,791 2,244 2,727 3,217 3,709 4,206 4,781 5,293 5,866 0 0 0 0 0 0.2 0.5 2.4 4.1 8.5 11.0 11.3 11.3 0 0 0 5 8 13 27 34 40 45 49 51 54 Fallow Scrub Barren rocky 0 0 0 0 0 0 0 0 0.2 0.6 2.1 4.2 7.1 33 34 46 46 46 46 46 46 46 46 46 46 46 0 0 0 0.7 1.3 1.7 2.3 3.6 4.8 5.8 5.8 5.8 5.8 4 7 8 11 13 14 17 19 19 19 22 24 28 Total Area (ha) Built-up (ha) Agriculture (ha) Forest (ha) Waste land (ha) Water bodies (ha) River Other scrub land (ha) No. of villages

maximum proposed FRL and Neemkheda, Tharr, Bebanai and Makodia sites with a lower FRL level than maximum proposed. The Sindh at maximum proposed FRL of 420 m has storage capacity of 14 MCM also has less area under submergence (1,400 ha of crop land) and 16,000 ha of cultivable land under its command. The Richhan site at maximum proposed FRL of 448 m has storage capacity of 1 MCM and has less area under submergence. Similarly, Barari and Kesari sites have 79 MCM and 98 MCM of storage capacity respectively at maximum proposed FRL and have extensive area under their command.

Conclusion
Water resources projects involve huge expenditure, vast resources of land and manpower. Hence, such projects are taken up for construction after feasibility assessment studies. In this regard, satellite remote sensing plays important role in providing necessary information for carrying out the feasibility assessment studies. This was demonstrated through case study of Ken-Betwa river link in which about 659 MCM of water is proposed to transfer to Betwa river to be utilised through construction of 8 new irrigation projects. From IRS-P6 LISS IV MX satellite image land use / land cover types and drainage network information was derived. Cartosat-1 stereo pairs are useful in deriving DEM from which contour information was generated. Using these inputs, reservoir submergence analysis and irrigation command analysis was carried out for all the proposed projects. The results of the analysis provide the decision makers to arrive at selection of suitable sites among the proposed project sites and the nal levels of the reservoir. This study amply demonstrates the important role played by satellite remote sensing in deriving the necessary inputs for the feasibility assessment studies.

Acknowledgement
The authors would like to acknowledge DD(RSAA), NRSC for encouragement and Former Group Director(WR) for guidance of the study. The authors would also like to acknowledge National Water Development Agency, Ministry of Water Resources, Govt. of India for sponsoring the study.

Balachandar, D., Alaguraja, P., Sundaraj, P., Rutharvelmurthy, K. and Kumaraswamy, K. (2010). Application of Remote sensing and GIS for Articial Recharge zone in Sivaganga district, Tamil Nadu, India, International Journal of Geomatics and Geosciences, Vol. 1, Issue.1, pp.84-97. Morris, Gregory, L. and Fan, Jiahua. (1998). Reservoir Sedimentation Handbook, McGraw-Hill Book Co., New York. NRSA Technical Report (2007). Satellite Data Based Inputs for Feasibility Assessment of proposed Irrigation Projects In Upper Betwa Basin as Part of Ken- Betwa River Link Project, Dept. of Space. Punmia, B.C., (1984 ). Irrigation and Water Power Engineering, Laxmi Publications (P) Limited, New Delhi 110 002. Ramakrihsnan, D., Bandyopadhyay, A. and Kusuma, K.N. (2009). SCS-CN and GIS based approach for identifying potential water harvesting sites in the Kali Watershed, Mahi River Basin, India, Journal of Earth System science, Vol. 118, No.4 pp.355-368. Ravi Shankar, M.N. and Mohan, G. (2005 ). A GIS based hydrogeomorphic approach for identication of site-specic articial recharge techniques in the Deccan Volcanic Province, Journal of Earth System Science, 114, No.5, pp.505-514. Singh, J.P., Darshdeep Singh and Litoria.P.K. (2009). Selection of suitable sites for water harvesting structures in Soankhad watershed, Punjab using remote sensing and GIS approach - A case study, Journal of ISRS, 37:21-35. Yusof, K.W., Serwan, M. and Baban, J. (2000 ). Identifying optimum sites for locating Reservoirs employing Remotely Sensed data and Geographical Information System, Proceeding of 21st Asian Conference on Remote Sensing, 4-8, Taipei, Taiwan.

123

NNRMS

BULLETIN

DECEMBER

2013

References

MODELING THE IMPACT OF LAND USE/ COVER CHANGE ON THE RUNOFF WATER AVAILABILITY: CASE STUDY FOR THE NARMADA RIVER BASIN
Gupta PK, Punalekar S1, Singh RP, Panigrahy S and Parihar JS Space Applications Centre, Ahmedabad-380015, Gujarat, India 1 PhD. Student, Department of Geography and Environmental Science University of Reading, UK Email: pkgupta@sac.isro.gov.in

Introduction
Land use change is an important characteristic in the runoff process that affects inltration, erosion, and evapotranspiration. Due to rapid development, land cover is subjected to changes causing decrease in the soil inltration rate and consequently increase in the amount of runoff. Deforestation, urbanization, and other land-use activities can signicantly alter the seasonal and annual distribution of runoff (Dunne 1978). The conventional methods of detecting land use changes are costly and low in accuracy. Remote sensing technique, because of its capability of synoptic viewing and repetitive coverage provides useful information on land use dynamics. The changes in land use due to natural and human activities can be observed using current and archived remotely sensed data. There is a need to investigate the relationship between the landuse change and the runoff water availability. With advances in computational power and the growing availability of spatial data, it is possible to accurately describe watershed characteristics when determining runoff response to rainfall input (kite and piteroniro, 1996; Singh and Woolhiser 2002). With the development of Geographic Information System (GIS) and remote sensing techniques, it is possible to enumerate various interactive hydrological processes considering spatial heterogeneity (Mohan and Shrestha 2000). Soil Conservation Service Curve Number method (USDA 1985) is widely used conceptual rainfall-runoff model as the major input parameters are dened in terms of land use, DEM, rainfall and soil types. The advantage of this method is that the user can experiment with changes in land use and assess their impacts.

Study Area
Narmada river basin has undergone various changes in land cover due to establishment of various dam projects. It has also been under controversies due to probable changes in forest cover and submergence of agricultural areas under big multi-purpose projects like Sardar Sarovar, Indrasagar etc. Hence, it is important to study the land cover dynamics in the area. Satellite images and image analysis techniques have made it easier to undertake these kinds of land use/cover (LULC) dynamics studies. Moreover, rainfall- runoff models can be used to help further understand and predict changes in runoff water availability. In this paper, an integrated approach using RS, GIS and hydrological modeling is presented to delineate and analyze the past and present land cover conditions of a basin and to estimate the impacts of the detected changes on surface runoff. Study was undertaken to estimate the change in LULC over Narmada river basin from 1970s to 2004. Satellite imageries of the basin during the decade of 1970 were available from Landsat mission. Whereas, for the recent scenario images obtained from AWiFS sensor from IRS P6 was used which has similar spectral resolution.

Landsat 1 MSS (multi spectral scanner) images were used for the LULC mapping of 1972. As limited data was available, different date images were used to obtain a complete mosaic of the area. However, the images of single season i.e. November to January were preferred. Original spatial resolution of the images by MSS was 79 m but resampled product for 57m was available. Dates used were 16th December 1972 (2 images), 17th December 1972, 30th November 1972, 1st December 1972 (2 images), 8th November1977, 27th November1975, 29th January1977. AWiFS images were used to prepare LULC map for the year 2004. Two images of same date 23rd November 2004 were geo-registered and mosaicked to cover the study area.

Methodology
LULC map and change detection: Landsat Images were mosaiced with feathering at cutline option in ERDAS. River basin area was then clipped out of it. Unsupervised classication was done with ISODATA algorithm. At rst, 50 classes were
Fig. 1: False color composite of mosaiced image obtained from Landsat MSS (1972)

produced after 6 iterations. The classes were then merged into four major classes agriculture, forest, water and other classes. The other class category was a broad class which covers barren areas, fallow land and dried river coasts, etc. as, mosaic of the area was obtained from multi date images, there were some miss-classication observed. These areas were then identied using toposheets.

Fig. 2: False color composite of mosaiced image obtained from AWiFS (2004)

The Areas of Interests (AOI) were selected which showed some haziness in the original

images. AOIs were again reclassied using same method. The classes in the classied AOIs were then recoded to merge them in the proper class as per the toposheets and Crop region map. These classied AOIs were then merged with the rest of the map. It was observed that settlements were also very small and inseparable from surrounding agricultural tone. Because the agriculture in this area appears to be kharif type and hence, it had lost the entire canopy till November-January. So, the river channel and major tributaries and lakes and settlements were digitized separately. This vector le was then converted into raster and then integrated with the LULC map. Similar procedure was adopted for the classication of AWiFS data to prepare LULC map for the year 2004. FCC images for 1972 and 2004 are presented in Figures 1 and 2, respectively. Methodology for the LULC map generation is presented in Figure 3. Accuracy Assessment: In both the maps accuracy assessment was done using ERDAS postclassication assessment tool. 130 random points were added using equally distributed (w.r.t. land cover classes) option in 23rd Nov, 2004 map. About 160 points were added in 1970s land cover map to have sufcient number of points in settlement class which covered very small area. The 2004 map was compared with the available LULC map (from NRDB library) as reference. In case of 1970s map assessment was done through visual interpretation of original satellite imagery, toposheets and crop region map of India. Error matrix was produced by this method. 125

NNRMS

BULLETIN

DECEMBER

2013

Data Used

Runoff Estimation: Modified curve number model (Mishra et al., 2005 b; Gupta et. al., 2012; Sharpley and Williams 1990) was used for the estimation of runoff in the Narmada river basin. Major inputs used in runoff modeling were soil texture map (NRDB), DEM (SRTM), rainfall (CPC NOAA), LULC maps. Runoff model was evaluated for the year 2006, due to non availability of observed runoff data for the year 2004. Evaluated model was used to study the impact of LULC change during 1972 to 2004 on the runoff production. Normal rainfall was considered for both the years (1972 and 2004) runoff estimations.

Result and Discussion

Land cover analysis: The overall classication accuracy for the LULC map of 2004 was 92% and for 1970s map it was 91.7%. The kappa statistics was 0.9245 for 1972 map and 0.9313 for 2004 map. The
Fig. 3: Methodology ow chart for LULC map generation

producer and user accuracy for all the classes for both the maps were signicantly high.

The area statistics shows that the agriculture area has increased by 2.22% (130176.6 ha) whereas area under forest has declined by -1.67% (39211.3 ha) (Table 1). The other class which includes barren, fallow and dried river channels showed decline of -71.9% (209743.2 ha). Agriculture covers the largest area among all classes followed by forest. Area under settlements greatly increased by 228.7% (18185.4 ha) and as expected water bodies area also show signicant increase of 124.9% (100592.5 ha). Classied land cover maps for 1972 and 2004 are presented in Figures 4 and 5, respectively. Table 1: Area under different classes and overall change between 1972 and 2004 Class name Agriculture Forest Other classes Water bodies Settlements Total Area Area covered in 1970s (ha) 5855967.3 2353372.9 291372.3 80522.1 7949.1 8589183.8 Area covered in 2004 (ha) 5986143.9 2314161.7 81629.1 181114.7 26134.5 8589183.8 Total change 130176.6 -39211.3 -209743.2 100592.5 18185.4 -

Table 2 shows net gain or loss of each LULC class to other classes. The statistics was generated by considering both area increase and decrease between pair of classes from matrix shown in Table 2. The positive sign shows that the area gained from the respective class and negative sign shows area lost to it. There was net loss of 62598 ha (78.7%) from agriculture land under water bodies and 16930.6 ha (21.3%) for settlement expansion. In spite of these losses there was overall increase (130176.6 ha) in the agricultural land. Forest on the other hand showed overall loss of area. Though there was some degree of increase in the forest cover, especially over barren and uncultivated areas, area of

transformed into agriculture, 33316.2 ha (64.4%) into water bodies and 267.2 (0.6%) into settlements. It mostly consists of land surrounding large reservoirs. The area under other category showed mostly net loss to the agricultural one. Settlements have expanded
Fig. 4: Land cover map of Narmada river basin (1972)

over agricultural land by 16930.6 ha (94.53%), and other classes by 980.9 ha (5.47%) and marginally over forested areas. The area under other classes has been undertaken for agricultural development, settlements and water bodies too. Settlements show signicant rise over surrounding agriculture and other land cover classes.

Fig. 5: Land cover map of Narmada river basin (2004)

The area statistics shows the overall increasing trend of agricultural and urban

development in the river basin. These changes were mainly due to enhanced availability of water for agriculture which was made available from many large reservoirs. This increase has been achieved at the cost of loss of some forested area, but on the positive side, it was evident that mostly the area which was not under any kind of utilization has been brought under some development. The multipurpose projects developed in the basins have signicantly proven useful to improve agriculture over the area. This increase obviously has also been supported by increase in the water availability. Table 2: Change detection matrix showing net gain and loss of area between all pairs of land cover classes Class name Agriculture Agriculture Forest Other Class Water bodies Settlements Net change 18121.7 191583.6 -62598 -16930.6 130176.6 Area gained or lost to classes (ha) Forest -18121.7 12493.8 -33316.2 -267.2 -39211.3 Other classes -191583.6 -12493.8 -4684.9 -980.9 -209743.2 Water bodies 62598 33316.2 4684.9 -6.6 100592.5 Settlements 16930.6 267.2 980.9 6.6 18185.4

-ve sign indicates loss of area to respective class mentioned in column title

Runoff Analysis
Model calibration was done for 2006 by comparing the observed and modelled runoff. Scatter plot between observed and modelled runoff is presented in Figure 6. A reasonably good match was obtained for the calibration period with coefcient of determination 0.875 Evaluated model was used to see the impact of LULC change on the runoff during 1972 to 2004. There was overall increase in runoff by 2.2% (816 Mm3) due to change in the land cover during 1972 to 2004. Signicant increase in runoff production in agriculture (1138 Mm3) and settlement (84 Mm3) has been observed whereas there was decline in runoff for the forest (131 Mm3) and other 127

NNRMS

BULLETIN

DECEMBER

2013

18121.7 ha (35%) under forest has

land cover classes (938 Mm3). These changes are in parity with alterations in their extent. Runoff water pattern for different land cover classes during 1972 and 2004 is presented in Figure 7. Significant increase in runoff in agriculture (1138 MCM) and settlement (84 MCM) has been observed, whereas, there was decline in runoff for the forest (131 MCM) and other land cover classes (938 MCM). These changes are in parity with alterations in their extent. Decrease in runoff over upstream sub-basins (Manot, Sandia)
Fig. 6: Relationship between observed and modeled runoff

whereas increase in runoff for the basins falling down stream. This may be due to gain and loss of forest cover in the upstream and downstream, respectively. There was 7% increase in the runoff over Rajghat (264 MCM) and Parts of Garudeshwar (392 MCM) whereas 6% decline in runoff over Manot (95mcm) was estimated (Table 3). Runoff change pattern for major sub-basins of Narmada river is presented in Figure 8.

Agriculture Forest Other class Water body Settlement

Fig. 7: Runoff for different land cover classes (%) in the Narmada basin for the year (a) 1972 and (b) 2004

Upstream sub-basin such as Manot showed increase in runoff contribution from forest areas whereas reverse trend has been obtained for the downstream sub-basins such as Garudeswar. Rajghat sub-basin has shown significant increase in runoff contribution from agriculture during 1972 to 2004. Great reduction in runoff from other land cover class has been observed for Sandia and Rajghat sub-basins. Runoff contribution from settlement has increased considerably in Mandleswar sub-basin.

Conclusion
In this study, change pattern in the land use/ cover during 1972 to 2004 were delineated and its impact on the runoff water availability in the Narmada river basin has been modeled and analysed. Signicant change in the LULC pattern has been observed. This is because
Fig. 8: Runoff change (%) pattern for 2004 w.r.t. 1972 for different land cover classes under major sub-basins of Narmada river

of construction of several new dams in the Narmada river. Settlement (228.7%) and

Sub-basin Manot Sandia Hoshangabad Mandleshwar Rajghat Garudeshwar

1599 5731 3773 13029 3596 5327

1504 5590 3884 13242 3859 5719

-6 -2 3 2 7 7

Water bodies (124.9%) area have increased signicantly with a slight increase in the agriculture area (2.2%). Forest and other class have been reduced by 71.9% and 1.7%, respectively. Increased water availability due to large dam projects has acted as one of the encouraging factor for agricultural and urban development. Overall loss in forest area to agriculture was 39211 ha. Expansion of settlements was mainly over surrounding agriculture area. Runoff analysis show changes in contribution of different land cover classes in total runoff in parity with increase or decrease in their extent. Signicant increase in runoff was obtained over settlements. There was overall increase in runoff was of 2.2% (816 MCM).

References
Dunne, T. and Leopold. (1978). L.B. Water in Environmental Planning, W.H. Freeman& Co., New York, NY, pp 818. Gupta, P.K., Punalekar S., Panigrahy S., Sonakia, A. and Parihar, J.S. (2012). Runoff Modeling in an Agro-Forested Watershed using Remote Sensing and GIS. Journal of Hydrologic Engineering, American Society of Civil Engineers Vol. 17, No. 11, November 1, 2012. ASCE, ISSN 1084-0699/2012/111255-1267. Kite, G. W. and Piteroniro, A. (1996). Remote sensing applications in hydrological modeling. Hydrol. Sci. J., 41 (4), 561591. Mishra, S.K., Jain, M.K., Pandey, R. P. and Singh, V.P. (2005 b). Catchment area-based evaluation of the AMC-dependent SCSCN- based rainfallrunoff models. Hydrol. Processes, 19(14), 27012718. Mohan and Madhav Narayan Shrestha (2000). A GIS based Integrated Model for Assessment of Hydrological change due to Land use modications, proceeding of symposium on Restoration of Lakes and Wetlands, Indian Institute of Science, November 27-29, 2000, Banglore, India. SCS (1972). SCS National Engineering handbook, Section 4. Hydrology, Soil Conservation service, USDA, Washington, DC. Sharpley, A.N. and Williams J.R. (1990). EPICErosion/Productivity Impact Calculator: 1. Model Documentation. US Department of Agriculture Technical Bulletin No. 1768. US Government Printing Ofce: Washington, DC. Singh, V.P. and Woolhiser, D.A. (2002). Mathematical modeling of watershed hydrology. J. Hydrol. Eng., 7 (4), 270292. USDA, Soil Conservation Service (1985). National Engineering Handbook, Section 4, Hydrology. U.S. Government Printing Ofce, Washington, DC. 129

NNRMS

BULLETIN

DECEMBER

Runoff vol. ( Mm3) Year-1972

Runoff vol. ( Mm3) Year- 2004

Change (%)

2013

Table 3: Runoff change during 1972 to 2004 for major sub-basins of Narmada river

Hydrological Modeling Approach for Annual Water Resources Assessment - A Pilot study in the Godavari and Brahmani-Baitarani Basins, India
Durga Rao KHV, Raju PV, Simhadri Rao B, Venkateshwar Rao V, and Sharma JR National Remote Sensing Centre, ISRO, Department of Space, Hyderabad - 500 037, India Email: khvdurgarao@yahoo.com

Introduction
Proper assessment of the availability of water resources is the cornerstone for proper planning, development and management. Natural ow in the river basin is reckoned as water resources of a basin. The mean ow of a basin is normally obtained on pro-rata basis from the average annual ow at the terminal site. However, at any point of time, the water resources in a river basin has already been developed and utilized to some extent through construction of major or medium storage dams and development of hydro-power, irrigation and other water supply systems. A large number of diversion schemes and pumped storage schemes may have also been in operation. Assessment of natural ow has, therefore, become complex in view of the upstream utilizations, reservoir storages, re-generated ows and return ows. The natural ow at the location of any site is obtained by summing up the observed ow, upstream utilization for irrigation, domestic and industrial uses both from surface and ground water sources, increase in storage of reservoirs (surface and sub-surface) and evaporation losses in reservoirs, and deducting return ows from different uses from surface and ground water sources. The rst ever attempt to assess the average annual ow of all the river systems in India was made by the Irrigation Commission of 1901-1903. The major constraint at that time was that while records in respect of rainfall were available, data in respect of river ows were not available even for most of the important river systems. The Commission, therefore, resorted to estimation of river ows by adopting coefcients of runoff. The average annual ow of all the river systems in India was assessed as 1443 BCM (Billion Cubic Meter). Dr. A.N Khosla developed an empirical relationship between mean temperature (as an expression for mean evaporation loss) and mean runoff by studying the ows of Sutlej, Mahanadi and other river systems. According to these studies, the total annual ow of all the systems worked out to 1673 BCM. Central Water and Power Commission (1954-66), estimated the surface water resources of different basins during the period 1952 to 1966. The study was mostly based on the statistical analysis of the ow data wherever available and rainfall-runoff relationships wherever data was meagre. The country was divided into 23 basins / sub-basins. According to these studies, the water resources of various basins amounted to 1881 BCM (Rao, K L, 1979). National Commission on Agriculture (1976) has estimated total annual water resources of the country as 1850 BCM (1800 BCM available in an average year) based on water balance approach taking into account rainfall, percolation of water in soils, evaporation and evapotranspiration. Central Water Commission (1988) has made estimates of water resources using lumped approach. As per the report Water Resources of India, the natural run-off of a basin could be computed

the basin, the net increase of the surface water storage, additional evapotranspiration caused by use or storage of surface water, direct ground water ow from the river basin below or along the terminal site, the net export of ground water out of the basin, the net increase in ground water storage, soil moisture storage and the additional evapotranspiration caused by use or storage of ground water. Basin wise reassessment of water resources potential in the country was carried out by Central Water Commission (CWC) in 1999 and presented in the report entitled Reassessment of Water Resources Potential of India (CWC, 1999). The water resources potential of the country was reassessed as 1869 BCM against an earlier assessment of 1880 BCM done by CWC in 1988. While assessing the water resources of India, the country has been divided into 20 river basins comprising 12 major basins and 8 composite river basins. The natural ow at any location on a river is obtained by summing up the observed ow, upstream utilization for irrigation, domestic and industrial uses both from surface and ground water sources, increase in storage of reservoirs (both surface and sub-surface) and evaporation losses in reservoirs, and deducting return ows from different uses from surface and ground water sources. The observed ows at terminal sites of the basins were corrected for upstream abstractions to arrive at the natural ows. Based on this methodology, CWC assessed the average annual water resources potential of the country as 1869 BCM in the year 1993. The broad limitations of the CWCs study are (1) Water resources assessment was done based on CWCs gauge data only, no rainfall input is taken into consideration. (2) There is no cross check mechanism in the studies (3) Data from major and medium reservoirs only taken into consideration for computation of irrigation utilisation (4) Study was fully lumped approach having the estimates at terminal sites of the basins only. In this perspective, assessment of water resources in a comprehensive way is a very important aspect in water resources development and management. Remote satellite data may be used in studying the land use dynamics and its affect on hydrology. Keeping in view of these, a joint research project has been executed by National Remote Sensing Centre (NRSC) and CWC for re-assessment of water resources in the Godavari Basin, Brahmani-Baitarani Basin, India using hydrological modeling approach. The main objectives of the study are to compute water resources in the basin during the last 20 years (1988-89 to 2007-08), mean annual water resources and the availability of water resources during extreme wet and dry rainfall conditions through distributed hydrological modelling approach using space inputs.

Study Basins
In the present pilot study Godavari and Brahmani-Baitarani Basins are taken up as study basins to demonstrate the methodology. Godavari Basin extends over an area of 312,800 km2, covering nearly 9.5 percent of the total geographical area of India. The Godavari River is perennial and is the second largest river of the India and joins Bay of Bengal after owing through a distance of 1470 km (CWC, 1999). It ows through the Eastern Ghats and emerges into the plains after passing Polavaram. Pranahita, Sabari and Indravati 131

NNRMS

BULLETIN

DECEMBER

2013

by adding to the surface ow measured at the terminal site, the net export of the surface water out of

are the main tributaries of the Godavari River. The basin receives the major part of its rainfall during the Southwest monsoon period. More than 85 percent of the rainfall takes place from July to September. Annual rainfall of the basin varies from 880 to 1395 mm and the average annual rainfall is 1110 mm. The combined Brahmani-Baitarani river basin extends over a geographical area of 50,768 km2 and the basin is bounded on the north by the Chhotanagpur Plateau, on the west and south by the ridge separating it from Mahanadi basin and on the east by the Bay of Bengal. The drainage area of the basin lies in the States of Orissa (33,923 km2), Jharkhand (15,479 km2) and Chhattisgarh (1,367 km2). Geographic locations of the study basins are shown in the Figure 1.

Spatial and Non-spatial Data Used

In this study, various spatial and nonFig. 1: Geographic location of the study basins

spatial database such as landuse, soils, digital elevation model, command area boundary, hydro-meteorological data, and groundwater data were used in the study.

Land Use/land Cover: The rainfall-runoff relationship is one of the most complex hydrologic phenomena to comprehend due to the tremendous spatial and temporal variability of watershed characteristics, precipitation patterns and the number of variables involved in the physical processes. Rate of evapotranspiration mainly depends upon the landuse/landcover pattern. Landuse/landcover maps of the years 2004-05, 2005-06, 2006-07, 2007-08 prepared using AWiFS sensor data of IRS-P6 satellite (Source NRC project: ISRO) were used for runoff calculations in the study. Soil Texture: In the hydrological cycle, inltration is another major component after the evapotranspiration. Inltration at a given time depends upon the soil texture and the existing soil moisture. Soil maps (1:250,000 scale) prepared by NBSS&LUP, India were used in the study. These soil maps were reclassied into soil textural grids and were used in computing soil monthly moisture availability subsequently. Digital Elevation Model (DEM): DEM is one of the main inputs for hydrological modeling studies. SRTM DEM of 90 m resolution was used to delineate the watershed and sub-watershed boundaries of the study basins. Using the DEM, ow direction for each cell was assigned based on the direction of the steepest slope from among the eight possible directions to the adjacent cells. Based on the ow direction, ow accumulation towards the outlet of the watershed was calculated. Godavari Basin and Brahmani-Baitarani Basin were divided into 23 and 8 sub-basins respectively based on the drainage pattern and corrected with the satellite data. These sub-basins were used in aggregating the nal results at the identied gauge sites.

area to rain-fed areas. It was assumed that irrigation supplies are provided for all the agricultural areas within the command boundaries. Kharif crop outside of the command area was considered as rainfed crop and rest is assumed as fully irrigated. Command area map of the study basins were obtained (source: India WRIS project, ISRO) and used in the evapotranspiration computations and subsequently in runoff computations. Rainfall and Temperature Grids: Daily rainfall grids at 0.5 degree and temperature grids at 1 degree resolution of the mentioned 20 years period (Rajeevan and Jyote, 2008) were obtained converting them into monthly grids in Geographic Information System (GIS) domain. Groundwater and Reservoir Flux Data: Ground water level data of wells spread across both the basin and specic yield map were collected from the Central Ground Water Board (CGWB), India. Annual ground water ux (recharge or withdrawal) for each observation was arrived at through arithmetical difference between June / May months observations of the two consecutive years. A point map was created and spatial interpolation was done in GIS environment to create groundwater ux grids. Annual ground water recharge or withdrawal in each year was computed by integrating groundwater ux grids with specic yield grid. CWC monitors 11 major and medium reservoirs in the Godavari basin and 3 reservoirs in Brahmani-Baitarani Basin. Water level data of these reservoirs was collected and reservoir ux in each hydrological year (June-May) was computed. Domestic, Livestock and Industrial Water Consumptive Use: Census data of 1991 and 2001 (www.Censusindia.gov.in) was used for estimating domestic and industrial water use. Survey of India village administrative information was integrated with population attribute data to prepare spatial population maps for each year by interpolating the above census data. For estimating domestic requirements, it was assumed that, 70 litres per capita per day (lpcd) and 140 lpcd for rural and urban consumption respectively (NCIWRD, 1999). Industrial demand was assumed being equivalent to the domestic demand. Domestic & Livestock consumptive use is taken as 15% of its demand, and industrial consumptive use is taken as 50% of the demand (NCIWRD). As per NCIWRD, average water requirement by the livestock is about 30 litres/ livestock/day and its population is 50% to the human population. River Discharge Data: Monthly observed river discharge data for 20 years of various gauge stations in the study basins was collected from CWC and used in the model validation and calibration. from the India Meteorological Department (IMD) and used in the model for runoff computations after

Modeling Framework and Methodology

The modeling frame work for the present study (Figure 2) involves integration
Fig. 2: Modeling framework

of spatial data sets (DEM, LULC, soil texture,

133

NNRMS

BULLETIN

DECEMBER

2013

Command Area Map: Estimation of actual evapotranspiration (AET) varies from irrigated

village census) with hydro-meteorological data sets (Rainfall, temperature, GW ux, reservoir ux, river discharge) in GIS environment to carry out water balance computations at hydrological response unit level. The model development and calibration was carried out using the four years data sets (2004-2008) and the calibrated model was extended for all the remaining years. The 20 years water balance outputs were averaged to arrive at Long-term Mean Annual Basin level Water Resources. Broad modeling framework is shown in the Figure 2.

Thornthwaite & Mather Method

After examining various hydrological models, Thornthwaite and Mathor Model (Thornthwaite and Mathor, 1957) was selected for estimating runoff in the present study (NRSC, 2009; Ray, 2001; Sokolov and Champan, 1994; X U CY and Singh, 1998). The water balance has been used for computing seasonal and geographic patterns of irrigation demand, the soil moisture stresses under which crop and natural vegetation can survive. Water table calculated for a single soil prole or for an entire catchment, refers to the balance between incoming of water by precipitation and outow of water by evapotranspiration, ground water recharge and stream ow. Among the several possible methods of calculation, the one introduced by Thornthwaite and Mather (1957) generally has been accepted (Jianbiao, 2005). This technique uses long term average monthly rainfall, long term average potential evapotranspiration (ET), and soil & vegetation characteristics. The last two factors are combined in the water capacity of the root zone. Computation of ET in this method is mainly based on temperature and day length factors. Day length factor grids were prepared in GIS environment. Potential Evapotranspiration (PET) was calculated using the temperature and day length grids through spatial modeling technique. These spatial PET maps of the country were prepared and subsequently PET grids of Godavari, Brahmani - Baitarani were extracted for further use. Vegetation Factors: The Thornthwaite method doesnt account for vegetative effect which is most useful parameter in water balance estimations (Peter E. Black). Monthly landuse factors have been derived for both the river basins using satellite remote sensing data and integrated with PET to account for vegetation effect on PET. Thus the PET revised has been calculated using the equation 1 PET revised = PET * vegetation factor Eq. 1

The crop coefcient integrates the effect of characteristics that distinguish a typical eld crop from the grass reference, which has a constant appearance and a complete ground cover. Consequently, different crops will have different crop coefcients (Kc) The Kc values primarily depend on crop type, crop growth stage, soil evaporation. Uniform vegetation coefcient during all the months has been considered for the vegetations like forest, scrub land etc.(Canadell et. al., 1996; Descheemaeker et al., 2011). Whereas for agricultural lands, variable coefcients taken in different months according to the crop growth stage and type of crop (FAO; Mohan and Arumugam, 1994). These vegetation coefcients are further calibrated using the eld discharge data. After the calculation of PET revised, water loss in each month and accumulated water loss (La) were computed based on PET revised and rainfall data. Next, the water storage capacity (SM), which depends upon the soil texture type, rooting depth of vegetation and land use, in the root zone of the soil must be determined. Thus, from the readily available tables, graphs and by using the Mathers empirical formula soil moisture and the change in soil moisture in each month were calculated. Actual evapotranspiration (AET) represents the actual transfer of moisture from the soil and vegetation to the atmosphere. When the rainfall (P) exceeds PET revised, it was assumed that sufcient moisture exists in the soil within the root depth to meet the climatic demands, in such case Actual Evapotranspiration (AET) will be equal to PET revised. In the condition when P < PET revised, AET demand will be met from the P and change in soil moisture. In irrigated agricultural land (canal and well irrigation) it is assumed that full irrigation support is provided to meet the AET requirements. Irrigation support (P- PET revised) is added to rainfall to make AET equal to PET revised. Kharif crop outside the command area boundaries

The added irrigation support was subsequently adjusted while computing runoff. Monthly moisture surplus & decit and runoff were calculated based on P, PET revised, AET and SM.

Model Calibration and Validation

If any unknown variable exists in the model, it can be calibrated using the observed/eld data during the calibration process. The vegetation coefcients are the main variable to be calibrated. Basically the calibration process is a hit and trial method. Runoff computations were repeated by changing

the model, the runoff calculations have to be revised using the calibrated (revised) coefcients. In the present study the model has been calibrated during 2004-05, and 2006-07 since these two are dry and wet years respectively. Once the model is calibrated perfectly, it has to be validated with other set of eld observations to check the calibrated parameters. In the present study, model has been validated with the hydro-meteorological data of 2005-06 and 2007-08. Calibration of the Model is done using the equation 2. R Calibrated/computed = (R Model - F GW - F R - F DIL ) R o R Calibrated/computed = Calibrated/computed runoff R Model FR FGW F DIL R0 = Model estimated runoff (output from Thornthwaite Mather Model) = Reservoir Flux ( - ve sign for drawdown) = Ground water Flux (- ve sign for drawdown) = Domestic, Industrial and Livestock consumption = Observed runoff at gauge sites (CWCs observed data is taken) Eq. 2

All water balance components are in volumetric units (BCM/MCM)

Water Resources Availability (WRA)

Water resources of the basin comprises of runoff in the river at nal outlet, upstream effective utilisations for irrigation, domestic and Industrial, groundwater ux, and surface water ux. Thus, it can be expressed as ; WRA = R Calibrated/computed + IS + E +F DIL + FGW + F R Where, E = evaporation from the reservoirs (computed) IS = Estimated Consumptive Irrigation Input Provided (computed) Annual water resources availability during the 20 years (1988-89 to 2007-08) has been computed for both the pilot study basins. Mean annual water resources have been further calculated. Rainfall during the last 135 Eq. 3

NNRMS

BULLETIN

the landuse coefcients till the computed runoff matches with the observed runoff. After calibrating

DECEMBER

2013

is considered as rain-fed and the rest is considered as irrigated either by canal or well irrigation system.

1400

35 years was analysed and the water resources availability during the extreme minimum and

1300

maximum rainfall years was analysed further. It is noticed that these extreme events in both the basins are falling in the study period of 1988-89 to 2007-08.

1200

Rainfall (mm)

1100

1000

900

800

Results and Discussions Godavari Basin

From the analysis of meteorological
73-74 74-75 75-76 76-77 77-78 78-79 79-80 80-81 81-82 82-83 83-84 84-85 85-86 86-87 87-88 88-89 89-90 90-91 91-92 92-93 93-94 94-95 95-96 96-97 97-98 98-99 99-00 00-01 01-02 02-03 03-04 04-05 05-06 06-07 07-08

700

data, it is found that during the last 35 years (1973-74 to 2007-08) maximum rainfall was recorded in the Godavari Basin as 1393 mm in 1994-95 and minimum as 881 mm in 2002-03 as shown in the Figure 3. Hence these two are considered as meteorologically wet and dry periods respectively during this 35 years span. Average monthly temperature varies from 20oC to 35oC in a year which causes lot of monthly variations in the potential evapotranspiration in the basin. From the landuse/landcover map (2006-07), it is found that nearly 16 landuse classes exist in the study area. Agriculture land is the predominant class in the Godavari Basin having more than 50% (including current fallow) of the basin area; this extent varies slightly from year to year. Next dominating

Year

Fig. 3: Annual rain fall of the Godavari basin

Fig. 4: Landuse/landcover of Godavari Basin derived from IRS P6-AWiFS (2004-05)(Courtesy: NRC Project, NRSC)

class in the study area is forest. Paddy, cotton and pulses are the main crops in the basin. Land use land cover derived from IRS P6 data

of the year 2004-05 is shown in the Figure 4. Clay, loam, loamy skeletal, clayey skeletal, sand and rock outcrop are found to be main soil textural classes in the study basin. Predominant soil textures in the study area are clayey and loamy which has the property of low inltration rate and more runoff. Annual groundwater ux in the basin is found to vary from + 10m to -10m. In some pockets these uctuations are more, otherwise stable ux is noticed. Specic yield of the basin varies from 1.5 to 16%, with maximum part of the basin having the specic yield of 1.5%. The mean annual groundwater ux in the basin is estimated at 0.67 BCM (drawdown). Reservoir uxes in individual sub-basins are aggregated. It is noticed that many reservoirs maintains sustainable ux (less annual ux). The mean annual reservoir ux of all the 11 major and medium reservoirs is estimated at 0.01 BCM (drawdown). The mean annual domestic, livestock and industrial consumption ux is estimated at 0.99 BCM in the basin. Landuse coefcients calibrated through trial and process are found to be ranging from 0.5 for bare soil to 1.2 for paddy during peak crop stage. Landuse factors for scrubland, grassland and forest lands are found to be 0.65, 0.7 and 0.9 respectively.

observed runoff was validated and calibrated at 5 prominent CWC gauge stations namely; Polavaram, Asthi, Bamini, Patagudem and Tekra. Polavaram is the nal gauge station in the Godavari Basin that represents the hydrology of the complete Basin. It is found that at Polavaram computed runoff is very well matching with the observed runoff. Deviation between average computed runoff and average observed runoff is found to be 5.58 % only. Mean water resources of the basin (1990-91 to 2007-08) computed is found to be 113.09 BCM. In this, evaporation from only 11 reservoirs (data provided by CWC) is considered. Mean water resources availability in the Godavari Basin and its components are shown in the Figure 5. Mean water resources of the basin during 1967-68 to 1984-85 as assessed by CWC was 110.54 BCM. From the rainfall data analysis, it is found that nearly 8 BCM of rainfall has increased from the period 1973-1985 to 1988-2008; this may be one of the reasons for increase in the WRA of the basin during 1990 91 to 2007-08. From the rainfall data of 1973 to 2008 (35 years data) it has been inferred that 1994-95 and 2002-03 were maximum and minimum rainfall years having rainfall of 435.1 and 275.3 BCM respectively. Hence, WRA during these two years were analysed separately. It is found that water resources availability in maximum and minimum rainfall years are 178.7 BCM and 72.63 BCM respectively.

Brahmani-Baitarani Basin
Rainfall varies both spatially and temporally in the Brahmani-Baitarani basin. Figure 6 shows the annual rainfall of 37 years (1971-72 to 2007-08) in the basin. Among these 37 years, the lowest annual rainfall is 802 mm (2004-05) and highest annual rainfall is 2022 mm (1994-95). The analysis of rainfall during the study period of 1988-89 to 2007-08 (20 years) indicated that the average
Fig. 5: Water Resources Availability in the Godavari Basin

annual rainfall is 1467 mm. From the landuse/landcover map of 2004-05, it is found that 17 landuse classes

2100

exist in the Brahmani-Baitarani Basin as shown in the Figure 7. Forest cover forms the major constituent (31.9%), followed by crop area (29.15%) and current fallow (28.25%). The remaining 10.7% of basin area is covered by built up land, plantation, littoral swamp, grassland, gullied land, scrubland, other waste land and water bodies. The crop area
71-72 72-73 73-74 74-75 75-76 76-77 77-78 78-79 79-80 80-81 81-82 82-83 83-84 84-85 85-86 86-87 87-88 88-89 89-90 90-91 91-92 92-93 93-94 94-95 95-96 96-97 97-98 98-99 99-00 00-01 01-02 02-03 03-04 04-05 05-06 06-07 07-08

1900

1700

Rainfall (mm)

1500

1300

1100

900

700

Year

is further categorised as Kharif only (23.64%), Rabi only (0.9%), Zaid only (0.26%) and Double/Triple (4.35%) classes. 137

Fig. 6: Annual Rainfall of Brahmani-Baitarani basin from 1971-72 to 2007-08

NNRMS

BULLETIN

DECEMBER

2013

Runoff in each sub-basin during all the years was aggregated separately. Computed runoff and

The main soil types found in the basin are red and yellow soils, red sandy and loamy soils, mixed red and black soils and coastal alluvium. The coastal plains consist of fertile delta area highly suited for intensive cultivation. The soils are classied based on the soil textural information as sandy, loamy, clayey, loamy skeletal, clay skeletal. Annual groundwater flux in the basin is found to vary from + 7m to -7m. In some pockets these uctuations are more, otherwise sustainable ux is noticed. Specic yield of the basin was assumed as 3% as the eld data is not available. The mean annual groundwater ux in the basin is estimated at 14 MCM (recharge). It is noticed that many reservoirs maintain sustainable flux (less annual ux). The mean annual reservoir ux of all the 3 major and medium reservoirs is estimated at 11 MCM (drawdown). The mean annual domestic and industrial consumption
Fig. 7: LandUse/Landcover Map of Brahmani-Baitarani basin (2004-05)

ux is estimated at 53 MCM in the basin. On Brahmani River Tilga, Jaraikela, Panposh, Gomlai and Jenapur discharge sites are located and the model estimated runoff is calibrated against the observed discharge at all the locations. Similarly, on Baitarani River, Champua and Anandpur discharge sites are located and the model estimated runoff is calibrated against the observed discharge at these 2 locations. For the combined Brahmani-Baitarani deltaic region observed discharge data is not available. Hence, for entire Brahmani-Baitarani basin, runoff is computed by adding model estimated runoff at Jenapur(entire upstream), Anandpur (entire

Fig. 8: Mean Water Resources Availability in the Brahmani-Baitarani Basin

upstream) and Delta region (exclusive) and computing calibrated runoff.

The average annual computed runoff of the basin is about 35,129 MCM. The maximum annual computed runoff is 60,429 MCM during 1994-95 which is wettest year in the 20 years. The minimum annual computed runoff is 12,003 MCM during 2004-05 which is the driest year in the 20 years. The maximum annual water resources available in the basin is 62,020 MCM during 1994-95 and the minimum annual water resources availability is 13,922 MCM during 2004-05. The average annual available basin water resources is 35,129 MCM. The average available water resources of Brahmani-

and its components are shown in the Figure 8. The mean annual rainfall during 1988-2008 is 6.6% more than the mean annual rainfall during the period 1971-85 which was considered for 1993 estimate. In 1993 estimates, the river discharge data was available only at Jenapur on Brahmani river and at Birdi on Baitarani river. Discharge data at the outlet of the composite basin was not available. Hence, area proportionate approach was adopted to estimate composite delta water resources. While estimated from present study is 7,887 km2 (which is based on geo-spatial data sets). As a result of this, the water resources estimate of composite delta was 2,050 MCM during 1993 as against present estimate of 4,780 MCM. Following the successful completion of the present pilot project in the Godavari and BrahmaniBaitarani Basins, it is proposed to extend the study to other river basins of the country as a joint project by National Remote Sensing Centre and Central Water Commission, India. using this approach, the composite delta area was estimated at 3,595 km2 as against the delta area

Conclusion
In this study, a procedure was developed for realistic assessment of water resources at basin scale using a simple Thornthwaite and Mather method by incorporating landuse coefcients derived from the remote sensing data. This approach requires limited meteorological data and can be up-scaled to other river basins in India. This study emphasises quantifying Indias water wealth by transformation from presently adapted basin terminal gauge site runoff aggregation to meteorological based water budgeting exercise through hydrological modeling approach. The spatial modeling approach can help in quantifying water resources availability in any major tributary of the basin also. This modeling approach can help in studying impact of future climate change in water resources of the basin. Different components in the water balance such as; evapotranspiration from agriculture, forest area and other landuses can be computed in spatial environment using this spatial modeling approach. Accuracy of runoff computations are found to be more than 90% compared to the eld observations.

Acknowledgments
The authors sincerely thank Dr. V K Dadhwal, Director, NRSC for providing constant support, encouragement and guidance during the project. Thanks are due to Dr Raghava Swamy, DD, training, NRSC and former Deputy Directors of RS&GIS AA, NRSC for providing continuous encouragement and guidance during the project. The authors sincerely acknowledge the support and guidance provided by Sri. M. E. Haque, Member (WP&P), Sri. Shankar Mahto, then CE (BPMO) and Sri. Rishi Sreevastava, CWC, New Delhi. The authors deeply acknowledges the support provided by Sri. V. N. Wakpanjar and Sri. M. Raghuram, CWC, Hyderabad for providing the eld data. The authors sincerely acknowledge Ashis Benarjee and Lalit Kumar, CWC, New Delhi for their contribution in executing the project and participating in the discussions. The authors thank Smt. R. Jyothsna, JRF, NRSC and Smt. K Anusree for providing technical support during the project.

References
Canadell, J., Jackson, R.B., Ehleringer, J.R., Mooney, H.A., Sala, O.E. and Schulze, E.D., (1996). Maximum rooting depth of vegetation types at the global scale. Oecologia. 108, 583-595. 139

NNRMS

BULLETIN

DECEMBER

2013

Baitarani basin accounts about 47% of mean annual rainfall. Mean annual water resources availability

CWC: (1999) Reassessment of Water Resources Potential of India. CWC, New Delhi. Descheemaeker, K. , Raes, D., Allen, R., Nyssen, J., Poesen, J., Muys, B., Haile, M. and Deckers, J. (2011). Two rapid appraisals of FAO-56 crop coefcients for semiarid natural vegetation of the northern Ethiopian highlands. Jouranal of Arid Environments. 75, 353-359. FAO: Crop evapotranspiration guidelines for computing crop water requirements. FAO CorporateDocument Repository. www.fao.org.docrep/x0490e/x0490e0h.htm Jianbiao Lu, Ge Sun, Steven G. McNulty and Devendra M. Amatya, (2005 ). A comparison of six potential evapotranspiration methods for regional use in the Southeastern United States, Journal of the american Water Resources Association, 41(3):621-633. Mohan, S. and Arumugam, N., (1994). Crop coefcients of major crops in South India. Agricultural Water Management. 26, 67-80. NCIWRD, (1999). Integrated Water Resources Development, A plan for the action, Report of the national commission for Integrated Water resources development, Volume 1, Ministry of Water Resources, Government of India. NRSC, (2009). Water Resources Assessment the National Perspective- A Technical Guide for Research and Practice, NRSCRSGIS AA-WRG-WRD-Oct2009-TR98. Peter E Black, Revising the Thornthwaite and Mather Water Balance, http://www.watershedhydrology.com/pdf/T&M%20 Revisited.pdf Rajeevan M. and Jyote Bhate, (2008). A high resolution daily gridded rainfall data set (1971-2005) for mesoscale meteorological studies, National Climate Centre Report, India Meteorological Department, Pune. Rao, K.L., (1979). Indias Water Wealth, Its Assessment, Uses and Projections, Pub. Orient Longman Limited, New Delhi. Ray, S. S. and Dadhwal, V. K., (2001). Estimation of crop evapotranspiration of irrigation command area using remote sensing and GIS. Agricultural Water Management, 49, 239-249. Thronthwaite, C.W. and Mather, J.R. (1957). Instructions and tables for computing potentialevapotranspiration and water balance, Laboratory of Climatology, Publication No. 10, Centerton, NJ. Sokolov, A.A. and Champan, T.G., (edited 1994). Methods for water balance computations an international guide for research and practice. The Unesco Press, Paris X U CY and Singh V P. (1998). A Review on Monthly Water Balance Models for Water Resources Investigations. Water Resources Management 12: pp. 3150.

Introduction
More than half of humanity relies on the freshwater that accumulates in mountains. The Himalaya mountains located in northern part of India holds one of the largest resources of fresh water in the form of snow and ice and abode of many glaciers and glacial lakes. A signicant portion of the low ow contribution of Himalayan rivers during the dry season is from snow and glaciers melt. The runoff supplies communities with water for drinking, irrigation and industry, and is also vital for maintaining river and riparian habitat. In Himalayas, the snow melt runoff occurring mostly during April, May and June months constitutes a substantial part of the water resources of the major perennial rivers of Northern and Eastern India, namely the Indus, Ganga, Brahmaputra and their tributaries. However, the data availability situation in the Himalayas is quite critical. The snowcover accumulated in winter months melts in lean summer months, which is vital during the period of high demand for water and power. A timely forecast of snow melt water is very useful in planning and managing the multi-purpose project. To forecast such runoff, information on terrain, snow pack and meteorological parameters are required. But collection of such information is not always possible by ground based systems due to rugged, hazardous and, in most cases, inaccessible terrain and sometimes due to the areas extending across international boundaries. Satellite remote sensing is therefore best suited for collecting information on the snow-cover under such situations.

Snowmelt Runoff Forecast in Sutlej Basin Based on Depletion Concept

For more than a decade, National Remote Sensing Centre (NRSC) has been giving long-term (during April-May-June) snow melt runoff forecast from Sutlej basin into Bhakra reservoir (NRSC, 2010). The Sutlej river is one of the main tributaries of Indus and has its origin near Manasarovar and Rakas lakes in Tibetan plateau. It travels about 300 km in Tibetan plateau in North-Westerly direction and changes direction towards South-West and covers another 320 km up to Bhakra gorge where 225 m high straight gravity dam has been constructed. This western Himalayan basin is highly rugged terrain with abundant natural water resource in the form of snow pack. It cover and area of about 51,000 km2. Characteristics of the basin and inaccessibility of the major part of it make remote sensing application ideal for hydrologists to monitor the snow cover information of the region and assess the water resource.

NNRMS

BULLETIN

Siva Sankar E, Abdul Hakeem K, Simhadri Rao B National Remote Sensing Centre ISRO, Department of Space, Hyderabad - 500 037, India Email: sivasankar_e@nrsc.gov.in

DECEMBER

SNOW MELT RUNOFF MODELING IN HIMALAYAN RIVER BASINS

2013

The contribution of snow melt is quite signicant in the Sutlej river. The snow melt exceeds the rainfall component after middle of March leading to onset of depletion of snow pack in the basin. The main snow accumulation period extends from October to March. Sometimes, intermittent snowfall events occur in April and May for short durations.

Snow Cover Area Mapping

The satellite data from AVHRR sensor onboard NOAA satellite has been used to compute snow cover area in the basin during the snow melt periods. Field data on discharge, snowfall and temperature were collected from stations established by Bhakra Beas Management Board (BBMB), all located inside the Indian territory of the basin. Cloud free images during accumulation and depletion period (generally between November to June every year) were used to derive Snow Cover Area (SCA) for the basin. Snow cover area statistics were derived for three regions of the basin namely, Spiti, Tibet and lower Sutlej.

Snow Cover Accumulation and Depletion

Snow cover depletion curve is generated with SCA as a percentage of the total basin area versus the time scale. This depletion curve graphically depicts the gradual change in the snow cover area over a time period. In physical terms, it indicates the change in the Snow Water Equivalent (SWE) of the snow pack over time representing the rate at which the snow pack and the embedded SWE change during the melting period, namely, April - May - June. This curve also indicates the occurrence of temporary snowfall events during the melting period by way of sudden increase in the SCA values. In the forecast situation, the satellite data is processed in near-real time as soon as the satellite overpasses the study area. The percentage SCA as on 1 April is interpolated from depletion curve and it is used as input for forecasting.

Initial Snow Melt Runoff Forecast

The hypothesis of depletion analysis is that the thick snow pack having high water equivalent depletes later and slower compared to a thin snow pack having low water equivalent which depletes early and faster. The SCA, the rate of depletion of snow cover at the beginning of the melt period and the commencement of depletion help in characterizing the current snow pack vis-a-vis the previous years qualitatively. The depletion analysis helps to assess the type of snow pack existing in the basin and the likely trend it may follow. It also indicates the relative/ qualitative status of the SWE with reference to the previous years. An empirical relationship has been developed by regression analysis. The maximum snow cover in the basin does not necessarily yield maximum runoff. Also, same snow cover area in the basin does not necessarily yield same runoff in two different years. In view of the complex topography and varied hydro-meteorological settings within the basin, the spatial variation in snow cover needs to be addressed to understand the snow melt process. The spatial distribution of snow cover in the 3 regions namely the Spiti, Tibet and lower Sutlej regions and their respective contributions to snow melt runoff need to be modeled to explain the spatial variability. The historic discharge data of the three regions have been studied and weightage factors have been arrived at to represent the contributions of the 3 regions to the total runoff. The weighted snow cover area index has been computed with above factors. The inows volume is regressed with weighted snow cover area index.

Revised Snow Melt Runoff Forecast

The monitoring of snow cover in the Sutlej basin was continued during the summer months i.e. April, May, June. The snow cover area in the regions of the Sutlej basin was computed. The snow cover depletion curves, the rate of depletion of snow pack were analysed. Any deviation in the actual depletion pattern from the assumed depletion pattern at the time of initial forecast is found through the monitoring of satellite derived snow cover and other meteorological observations, or whenever any new information on the ground situation is obtained, and the initial forecast needs to be revised.

pack melts during the subsequent summer months while the intermittent snowfall occurring during April-May-June is less signicant. Hence, the snowfall accumulated in the basin till 31st March of the year is considered for the purpose of computing the snow melt runoff. During the forecast period, the snow cover in the basin depletes meaning the water embedded in the snow pack is depleting with energy input. Therefore, the product of snow cover area on a particular day and the snow fall index derived from ground measured snowfall data cumulated till 31st March works as an index of the water volume embedded in the snow pack as on that day. This product does have a relation to the snow melt runoff occurring from that day to 30th June of that season which is the potential snow melting season. The actual ground measured snowfall data and temperature data are usually received from BBMB in the middle of May. The snowfall measured at 21 gauge stations within Indian portion of Sutlej basin does not explain the complete spatial variability within the basin. The snowfall measured at Namgia is assumed to represent the Tibet region while the snowfall measured at other stations represent Spiti and lower Sutlej regions based on their physical location. A model has been developed using regression analysis. The snow melt runoff occurring between the date of satellite overpass and 30th June is taken as dependent variable. The sum of products of SCA and the average snowfall in that region is taken as independent variable. SCA values derived from satellite data of multiple dates during the melting season of the previous years have been used in the regression analysis along with the corresponding snowfall data. Q= c*A*(C1S1W1+C2S2W2+C3S3W3) Q - Inows between the date of SCA computed and 30th June in cusec-days C1, C2, and C3 - Runoff coefcients for three regions S1 - SCA in Spiti region as % of the region area S2 - SCA in Tibet region as % of the region area S3 - SCA in Lower Sutlej as % of the region area W1 - Average snowfall in Spiti region measured till 31 March W2 - Average snowfall in Tibet region measured till 31 March W3 - Average snowfall in Lower Sutlej basin measured till 31 March A - Basin area c - Unit conversion factor NRSC provides initial forecast in the 1st week of April and a revised forecast in the last week of May during every year. The forecast period was specic 91 day period i.e. from 1st April to 30th June. The actual measured daily inows into Bhakra reservoir were provided by BBMB and the cumulated inows up to 30th June were computed in lakh cusec-days. The deviation in the forecasts were calculated with reference to actual measured inows. The summary of forecast results given for last three years is shown in Table 1.

Snowmelt Runoff Modeling Based on Energy Balance Approach

Based on the experience gained in Sutlej basin, Central Water Commission requested NRSC for developing seasonal and short term snow melt runoff models for Sutlej, Beas, Yamuna, Chenab and Ganga basins. 143

NNRMS

BULLETIN

DECEMBER

2013

The seasonal snowfall accumulated in the basin during winter months in the form of snow

Table 1: Forecast results (in lakh cusec-days) Year 2007 2008 2009 Initial Revised Initial Revised Initial Revised Forecast 16 +/- 10% 14 +/- 10% 12 +/- 10% 12 +/- 10% 12 +/- 10% 12 +/- 10% 11.14 16.95 Actual 12.48 Deviation +28.0% +12.1% -29.0% -29.0% +7.7% +7.7 %

The development of snowmelt runoff models is carried out for the following basins. 1. 2. 3. 4. 5. a. b. Beas up to Bhuntar Sutlej basin up to Bhakra Chenab up to Premnagar Yamuna up to Hatnikund In the case of Ganga basin, model shall be developed for Bhagirathi up to Uttarkashi and Alaknanda up to Rudraprayag The location map of study area is shown in Figure 1. The drainage area up to the basin outlet for each basin is given in Table 2. Table 2: Drainage area for each basin Basin Beas up to Bhuntar Sutlej up to Bhakra Chenab up to Premnagar Yamuna up to Hatnikund Alaknanda up to Rudraprayag Bhagirathi up to Uttarkashi Area (km2) 3,160 51,451 17,273 11,323 10,201 4,527

Data Used
The snowmelt runoff model has been developed using satellite data based inputs as the availability of eld measured hydrometeorological data is less in most of the basins. The hydrometeorological data such as snowfall, temperature were received from CWC and is available at fewer stations. The elevation range of the study area is between 500 m and 7500 m in Himalayan region whereas the majority of hydromet stations are located below 2500 m and very few between 2500 m and 4000 m. However transient snow line is generally
Fig. 1: Location map of study area

above 2000 m.

data at few stations cannot be accurately interpolated spatially for the entire basin. Therefore it is observed that the data measured at these stations is not representing the entire basin and its spatial variability. The data used in the present study are: Snow Cover Area (SCA), Land Surface Temperature (LST), emissivity Snow Albedo and Snow Persistence Index (SPI) were derived from MODIS satellite data, while Glacier Cover Area (GCA) and classied Land Cover were derived from AWiFS satellite data. The other data included digital and discharge Elevation Model, Slope, Aspect from SRTM DEM data, Hydrometeorological data such as rainfall

Methodology
The snow melt runoff is modeled either by lumped approach using degree day index or by empirical models with skeleton data sets or by energy balance approach wherein the spatial distributed modeling is adopted. The present study proposes to use energy balance based approach for modeling the snow melt runoff process and thus for forecasting the snowmelt runoff. The runoff at the basin outlet comprises of snowmelt, glacier melt, runoff due to rainfall and base ow components. The details of energy balance principles are mentioned in the following sections.

Energy Balance Principles

The exchange of energy between the snowpack and its environment ultimately determines the rate of snowpack water losses due to melting and evaporation / sublimation. Energy exchange primarily occurs at the snowpack surface through exchange of shortwave and longwave radiation and turbulent or convective transfer of latent heat due to vapour exchange and sensible heat due to difference in temperature between the air and the snow. These linkages between the different energy components responsible for snowmelt are summarized in Figure 2. The sources of energy that cause snowmelt include both shortwave (Qsn) and long-wave (Qln) net radiation, convection from the air (sensible energy, Qh), vapor condensation (latent energy, Qe), and conduction from the ground (Qg), as well as the energy contained in rainfall (Qp) as shown in the upper left of Figure 2. These uxes are usually measured as energy per time per unit area of snow. The energy budget equation that describes the energy available for snowmelt is given in Equation 1 below. The total energy available for snowmelt is Qm. Qm = Qsn + Qln + Qh + Qe + Qg + Qp - Qi ------- (1) where Qi is the rate of change in the internal energy stored in the snow per unit area of snowpack. This term is composed of the energy to melt the ice portion of the
Fig. 2: Schematic diagram showing snowmelt processes (After Price, Hendrie, and Dunne (1979)

snowpack, freeze the liquid water in the snow, and change the temperature of the snow.

145

NNRMS

BULLETIN

DECEMBER

2013

In the present study, spatially distributed modeling approach is adopted. The hydromet point

Whenever sufcient energy is available, some snow (ice) will melt and form liquid water. Since the physical structure of the snowpack is a porous matrix, this snowmelt will be held as liquid water (provided it does not refreeze) in the interstices between the snow grains and will increase snow density and snow water content. The snowpack is commonly called ripe when it is isothermal at 0 C and saturated. Whenever the capacity of the snowpack interstices to hold the liquid water is exceeded, some of the snowmelt will begin to move down-gradient to become a portion of the snowmelt runoff. Some of the snowmelt may inltrate into the ground depending on inherent soil characteristics, the soil moisture content, as well as whether or not the ground surface is frozen. The inltrated snowmelt later re-emerges as interow into stream channels, or it percolates into deeper groundwater storage.

Energy Balance of the Snow Cover

Evaluating snowmelt theoretically is a problem of heat transfer involving radiation, convection, condensation, and conduction. The relative importance of each of these heat transfer processes is highly variable, depending upon conditions of weather and the local environment. Gray and Prowse (1992) tabulate selected results of the relative contributions of each heat transfer process as a function of site environment. The basic equations and coefcients that describe snowmelt at a point have been derived primarily from various laboratory and eld experiments. Equation 1 given above summarized the energy sources available to melt snow. The summation of all sources of energy represents the total amount of energy available for melting the snowpack (Qm). The amount of snowmelt at a point may be expressed by the general formula given as Eqn. 2 Qm M = ---------------LwB Where M = snowmelt, m of water equivalent Qm= algebraic sum of all heat components, kJ/m2 B = thermal quality of the snow (e.g., ratio of heat required to melt a unit weight of the snow to that of ice at 0 C) L = latent heat of fusion of ice, 334.9 kJ/kg w = density of water, kg/m3 A melting snowpack consists of a mixture of snow (ice) and a small quantity of free (liquid) water trapped in the interstices between the snow grains. The relative proportion of a snowpack that consists of ice determines the thermal quality (B) of the snowpack. A snowpack that contains no free water has a thermal quality of 1.0. However, after melt has begun, there is some free water held within the snow matrix, yielding a thermal quality of less than 1.0. The heat energy required to release 1 g of water is somewhat less than the latent heat of fusion of water (that is the energy required to change state from ice to water; 334.9 kJ/kg or 80 cal/g for pure ice). For a melting snowpack, after free drainage by gravity for several hours, the thermal quality normally averages between 0.95 and 0.97, corresponding to 3 to 5 percent liquid water in the snow. The thermal quality of snow may be far lower for ripe snows and in extreme cases where the water cannot drain freely from the snowpack. ----------------(2)

Radiational Energy Exchange

Radiational energy is the prime source of energy at the Earths surface. Some of this energy is classed as solar or shortwave radiation (radiation having wavelengths () between 0.2 and 2.2 m) and terrestrial or long-wave radiation (wavelengths between 6.8 and 100 m). The rst two terms of Equation 1 are sometimes referred to as net radiation Qn, the sum of net shortwave Qsn and net long-wave Qln energy uxes. As the net long-wave exchange is often a loss from the snow surface, Qn is expressed as Qn = Qsn - Qln

snowpack. The net amount of radiant energy that is available to melt snow depends on how much of the radiation is either reected from or absorbed by the snowpack. The amount of heat transferred to the snowpack by solar radiation varies with latitude, season, time of day, atmospheric conditions, forest cover, and reectivity of the snow (albedo). The amount of energy available for snowmelt from the absorption of shortwave radiation (Qsn) is given in equation for Qsn = (1-) Ii Where= albedo (expressed as a decimal fraction) Ii= daily incident solar radiation (kJ/m2 per day) The short wave energy is corrected for the inuence of forest canopy using landuse and landcover information. Longwave Radiation : Some of the energy absorbed by the snowpack from solar radiation is radiated from the snowpack to the atmosphere as long-wave radiation. Snow is nearly a perfect blackbody, with respect to long-wave (terrestrial) radiation, absorbing all such radiation incident upon it and emitting the maximum possible radiation in accordance with the Stefan-Boltzman law (Equation 5). Qln = Ts4 Where Qln= total longwave energy emitted by the snow, kJ/m2 per second = emissivity, 0.99 for clean snow = Stefan-Boltzman constant, 5.735 10-11 kJ/m2s K4 Ts= blackbody temperature in Kelvin (K) (temperature of the snow surface) The long wave energy is corrected for the inuence of forest canopy using landuse and landcover information. ------------(5) ------------(4)

Turbulent Transfer
Energy is also exchanged between the snow pack and atmosphere through the processes of convection and condensation. Depending on the climatological and local weather conditions, the relative importance of these processes differs widely. Turbulent exchange involves the transfer of sensible heat from warm air advected over the snoweld (convection), and also the latent heat of condensation of water vapor from the atmosphere condensed on the snow surfaces. Qh = Dhuz(Ta-Ts) Qe = Eeuz(eu-es) Where Dh = bulk transfer coefcient for sensible heat transfer, kJ/m30C Ee = bulk transfer coefcient for latent heat transfer, kJ/m3 Pa 147 ------------(7) ------------(8)

NNRMS

BULLETIN

DECEMBER

2013

Shortwave Radiation : Shortwave radiation is the most important source of energy to the

 uz = wind speed at a chosen height above the snow surface, m/s eu = vapour pressure of the air, Pa es = vapour pressure of the snow surface, Pa

Ground Heat
Heat entering the snow from the ground (Qg) by solid conduction is a very small component to the overall energy budget, especially compared with the radiational and turbulent exchange at the air/snow interface. This ground heat component can be neglected over short periods of time (less than 1 week). Although the daily melt caused by ground heat is small, it can amount to a signicant quantity of water over an entire snow season. Most lumped, conceptual models use constant daily values in the range of 0-5 J/m per second. Ground heat ow can also be estimated using soil temperature gradients measured near the surface in an equation for steady-state, one-dimensional heat ow by conduction (equation 9): Qg = k dTg /dz where k = thermal conductivity of the soil Tg = Soil temperature dTg /dz = temperature gradient from soil to snow ------------(9)

Heat Convected by Rain

The heat convected from the snow by rainfall is (equation 10) Qp = CpwPr(Tr-Ts)/1000 WhereCp = specic heat of rain, kJ/kg C w = density of water, kg/m3 Pr = rain quantity, mm/unit time Tr = temperature of the rain, C Ts = snow temperature, C T h e t e m p e r a t u r e o f t h e r a i n i s a s s u m e d t o b e e q u a l t o t h e a i r t e m p e r a t u re o r, i f available, the wet-bulb temperature. The specific heat C p is equal to 4.20 kJ/ (kg C) for rainfall and 2.09 kJ/(kg C) for snowfall. ------------(10)

Internal Energy
By denition, if the cold content or heat decit of the snowpack is positive, the snowpacks temperature is below freezing. The internal energy Qi can be changed and the heat decit reduced by the heat released when melt or rainwater freezes within the snow cover. This phenomenon is prominent during diurnal temperature cycles with refreezing at night because of radiational cooling. Melt and rainwater will continue to freeze within the snow cover until the total heat decit reaches zero. When the total heat decit reaches zero, the snow cover will become isothermal at 0 C. This internal energy is calculated by the following expression (Gray and Prowse 1992) equation 11: Qi = ds(iCpi + lCpl + vCpv) Tm ------------(11)

ds = depth of snow = density, i = 922 kg/m3 , l = 1000 kg/m3 Cp = specic heat, Cpi 2.1 kJ/kg C; Cpl 4.2 kJ/kg C Tm = mean snow temperature, C The subscripts i, l, and v refer to the ice, liquid, and vapor phases, respectively. The contribution of the vapor phase is assumed negligible.

Based on the energy balance approach, the snow melt is calculated spatially using the remote sensing based inputs from various sources. The overview of the methodology is shown in the ow chart given in Figure 3. Initially the basin boundaries have been delineated using SRTM digital elevation model and AWiFS satellite data considering the basin outlet points at
Fig. 3: Overview of the methodology followed in snow melt runoff modeling

which the runoff forecast is proposed to be made. The AWiFS satellite data has been used to map the land use and land cover

in the study area. The land cover is categorized primarily into open areas, forest, deciduous forest and water bodies. The snow cover present in a basin can be mapped using appropriate satellite data. As the study necessitates frequent temporal observations, it was decided to use snow cover products available as 8-day time composites on MODIS web site for free download. The 8-Day time composites are for specic Julian 8-Day periods commencing from 1st Jan of a year. The maximum possible snow cover during the 8-Day time period is mapped in this snow cover product. The data from 30-March to 30June were used for this purpose. The snow cover products are reprojected and subsets are prepared for the study area. Snow cover in each basin is computed using basin mask. The total net energy ux is primarily contributed by incoming solar radiation and outgoing longwave radiation. The incoming solar radiation is a function of location of a pixel (latitude, longitude), elevation, Julian day and time. The longwave radiation is a function of surface temperature, air temperature, and emissivity. In addition, to account for snow depth in an indirect manner, the snow cover depletion concept is used. The snow cover depletion concept states that a thicker snowpack depletes slower and later than a thinner snowpack, which depletes faster and earlier. Based on this concept, to account for snow depth in an indirect manner, snow persistence index is computed for each year depending on the snow residency period at each pixel. The index is scaled between 0.1 and 1 where 1 representing a pixel containing snow for full 91 day period and 0.1 representing a pixel containing the snow for minimum period. This persistence index is used to qualify the snowpack in runoff estimation. 149

NNRMS

BULLETIN

Snowmelt Runoff

DECEMBER

2013

where -

The snow melt runoff in a basin is computed using the total energy available for melt during that 8-day time period and snow persistence index. The snow melt runoff for all the 12 time periods of 8-days between 1st April and 30th June are computed individually with corresponding data inputs and then integrated to arrive at seasonal snow melt runoff.

Glacier Melt Runoff

The presence of glaciers in a watershed or a basin signicantly affects runoff volume. The partial glacierization of a basin by as little as a few percent of cover can cause moderate to extreme variations in peak runoff magnitude and frequency over days, year, and decades. Runoff is not directly related to precipitation within a glacierized basin, so it is, at present, difcult to predict because of lack of glacio-hydrologic data and limited, rather rudimentary knowledge of the glacio-hydrologic process controlling runoff. Runoff from glacierized areas of a basin is generally greater than that from nonglacierized areas with similar precipitation. Majority of runoff from ice-covered areas comes during the melt season, generally from mid-May to mid-September. The glacier melt runoff during summer months is a signicant component of total runoff. The glacier melt runoff depends on the extent of glacier area within a basin and the level of exposure of the glaciers and duration of exposure during the progress of the snow melt season. The level of exposure of glaciers in the summer months depends on the presence of seasonal snow cover. The probability of glacier melt and its magnitude is more when the seasonal snow cover is less. The glacier melt runoff increases as the snow melt season progresses. For the purpose of estimation of glacier melt runoff, it is necessary to map the glacier in each basin using satellite data of the September - October months. The glacier map prepared from satellite data is used to estimate glacier melt volume during summer period considering the net energy available for melt. The glacier melt component during the 3 months period is computed with energy available for melt and the level of exposure of glacier in each basin. Qg = Cg * Glacier Area

Rainfall Runoff
The runoff due to rainfall constitutes signicant proportion of the total runoff during the snow melt season in Himalayan region. The rainfall in a region varies spatially and temporally and needs to be measured in systematic and well representative manner. The accuracy of runoff estimation due to rainfall depends on the number of rain gauge stations available and their distribution within the basin. Generally, in Himalayan mountainous region where elevation is more than 4500 m only snowfall occurs and rainfall does not occur. Hence, in these study areas with elevation below 4500 m is considered as areas where rainfall contributes for generating runoff. In each basin, areas with elevation lower than 4500 m is extracted using DEM. Mean rainfall of a basin is estimated depending on the number of rain gauge stations available within a basin. In basins where only one rain gauge station is available the rainfall at this station is assumed to be representative of the entire basin. In basins where multiple rain gauge stations are available theissen polygon method is used to estimate the mean rainfall of the basin. The rainfall runoff is estimated using suitable runoff coefcient. In the present study, the scope is restricted to runoff due to rainfall during snow melt period i.e. Apr-May-Jun months. An appropriate runoff coefcient is assumed for each basin which varies temporally during the season. Basin level runoff from rainfall is computed by multiplying the average rainfall with rainfall contribution area and runoff coefcient.

Base Flow
Base ow as dened by Hall (1968) is the portion of ow that comes from groundwater or other delayed sources. During snow melt season, the base ow is a part of total runoff. When snow melts, a signicant portion of it percolates in the underlying ground and joins the stream within the basin, sometimes with a time delay. In the initial part of snow melt season, depending on the soil conditions, the snow melt water percolates until the ground is saturated before it becomes direct runoff. The base ow varies marginally depending on the snow cover conditions prevailing in the basin. may be similar to that of winter months or it can vary marginally. In the present study, the base ow is empirically estimated as a function of seasonal Snow Cover Available in each basin. Qb = Cb * SCA The base ow during winter months is generally constant. The base ow during summer months

Total Runoff
The total runoff during snow melt season (Apr-May-Jun months) is the sum of snow melt runoff, glacier melt runoff, rainfall runoff and baseow components. The runoff measured in the eld at the outlet point represents total runoff and the same is compared with computed runoff. The details are shown in the ow chart (Figure 4). Q = Qs + Qg + Qr + Qb

Conclusion
The seasonal snow melt runoff model for all the basins has been developed and the experimental forecast has been provided to CWC. The results are encouraging. Some more ner modications are being made to the model based on the experimental forecast
Fig. 4: Total runoff estimation

and its verication.

References
NRSC (2009). Inventory of glacial lakes and water bodies in Tawang river basin, Technical Report, National Remote Sensing Centre. NRSC (2010). Snow melt runoff forecasting in Sutlej basin using remote sensing inputs, Technical Report, National Remote Sensing Centre. Gray, D.M., and Prowse, T. D. (1992).Snow and oating ice, in Hand book of Hydrology, D.R. Maidment.,ed., McGraw-Hill, Inc., Newyork:7.1-7.58. Hall, F.R. (1968). Base ow recessions--a review. Water Resources Research., Vol.4 Issue 5: 973-983.

151

NNRMS

BULLETIN

DECEMBER

2013

Qr = Cr * Mean Rainfall * Rainfed Area

REMOTE SENSING IN GROUNDWATER MODELING

Sudhir Kumar and Sanjay Kumar Jain National Institute of Hydrology, Roorkee 247667, India Email: sudhir.nih@gmail.com

Introduction
Groundwater is an important natural resource. Availability and utilisation of groundwater plays a leading role in social and economic development of the nation. For scientic exploitation of groundwater, proper and good quality data is required to be collected and well documented. Geological and hydrological data are required to be collected before initiating studies on groundwater problems. This includes information on surface and subsurface geology, aquifer type, aquifer parameters, water tables, precipitation, evapotranspiration, stream ows, land use, vegetative cover on the surface, extraction from study wells, aquifer boundaries, irrigation, aquifer characteristics etc. These data are important not only for exploitation but also for planning, design and operation of the groundwater structures. Since groundwater is a dynamic source, the accuracy and reliability of acquired data usually increases with the time available for observation and interpretation. Extensive groundwater exploitation during past two decades for meeting irrigation, water supply and industrial needs, has drawn the attention of groundwater hydrologists towards maintaining sufcient groundwater supply throughout the year. The assessment and management of groundwater resources are more important in the sense that groundwater is more protected against pollution, has natural system for its storage, does not have the problem of huge evaporation losses. In order to arrive at the management alternatives, the assessment of groundwater potential and the behaviour of groundwater system under various stress conditions are required besides the alternatives for the creation of additional resources. All these require the collection of reliable data and their compilation and processing in an intelligent manner.

Groundwater Modeling
Groundwater models are tools that help in understanding the physical, chemical, and biochemical processes taking place in groundwater systems. They also help in understanding the intricate interactions between these processes, and they can provide the information needed in order to manage these processes benecially and without harm to the environment. Numerical models are now used in virtually all areas of groundwater hydrology. A series of mathematical formulas are linked together to explain the working of a particular phenomenon. A good model has the ability to predict the outcome of a set of inputs, as they would affect the real world. A simulation model is used to analyze the known information about a data feature. This could be a stretch of stream, a point-source of pollution, or a census tract. A simulation model uses

and creates a new result based on the information from one or more variables. These models deal with extracting more information from what is already known about a feature. A predictive model, on the other hand, is used to predict how a change in a variable will affect other conditions. Once again, this can be applied to any data feature, but it applies more to the attributes of a feature than the feature itself. For instance, such a model might be used to predict the effect of any stress (pumping, discharge of pollutants etc.) on the groundwater conditions. These models deal with predictive models deal with changes that will occur if certain variables of a feature change. Groundwater models are broadly grouped into two categories: water quantity and water quality. In the water quantity category, models are used in aquifer management, well eld design, groundwater recharge enhancement, determination of optimum aquifer yield, well interference studies, studies of groundwater-streamow interactions, and similar problems. Models of this type have been well proven in many years of use. In the water quality category, groundwater models are used to study the consequences of groundwater contamination, the measures that need be taken to prevent contamination, and the design of remediation measures. Natural groundwater systems are often highly complex and a complete description of the physical characteristics in most cases is impossible. Instead, the physical properties of the system are generally described either in terms of averages or in the form of statistical distributions. Also, physical/ chemical/biochemical processes can interact in complex ways and some of these interactions (i.e. reaction kinetics) are not yet fully understood. In view of these complexities, it is risky to use a model to predict in absolute terms, for example, the extent of a contamination plume, or the status of a clean-up project. Such absolute predictions are justied only in very simple scenarios. On the other hand, groundwater models are appropriately and meaningfully used for purposes of (i) obtaining insight into complex processes, (ii) assessing the relative importance of the various processes occurring in a given situation by means of sensitivity analyses, (iii) analyzing worst-case or what if situations and (iv) making probabilistic predictions.

Constructing a Numerical Model

To construct a numerical model of a groundwater system, the geological units and their hydrogeologic properties within the domain of interest are dened. The domain should be selected such that conditions along its boundaries can be dened unambiguously. Next, the processes that may play a role in the transport of contaminants are dened. These processes will include; groundwater ow, advective transport, dispersive/diffusive transport, chemical interactions, radioactive or biological decay, gravity forces, thermal processes and capillarity. The processes taking place in a groundwater system are subject to physical laws such as; constitutive laws (Darcys law, Ficks Law), conservation laws for uid mass, solute mass, thermal energy

153

NNRMS

BULLETIN

DECEMBER

2013

information from the data table associated with the object, plugs that information into a formula,

and force equilibrium laws. These physical laws can be expressed mathematically in terms of governing equations, which are usually partial differential equations. In cases of linear processes with simple geometry, these equations can generally be solved by analytical methods. In general cases involving nonlinear processes or complex geometry, the equations are solved numerically.

Conceptualization of Groundwater Model

The mathematical description of the relevant physical laws, together with the description of the hydrogeology and the denition of the boundaries, constitute the conceptual model of the system. In order to solve the mathematical equations numerically, the system is discretized and the partial differential equations are approximated by algebraic equations at a nite number of points in the domain. The numerical solution then solves the resulting sets of algebraic equations. Typically, several thousands of simultaneous equations are solved repeatedly in contaminant transport problems. The entire solution process is formally expressed in a computer code. Although often the computer code is seen as The Model, the most important component is actually the conceptual model which has been selected to represent the real physical system and the processes taking place within this system. Thus great care must be taken in dening the conceptual model. Before the model can be applied to a real system, the models should be calibrated and vaidated. Once the model is properly calibrated and validated, it can be applied to simulate the processes occurring in a real system, to perform sensitivity analysis, and to investigate worst-case situations. At this stage, the model should be used to obtain the best possible insight into the system, and to use this insight in developing appropriate strategies for achieving the overall objectives of the study.

Numerical Methods for Groundwater Models

The types of numerical methods that are most often used in groundwater modeling are Finite Difference, Finite Volume, Finite Element, and Particle Tracking Methods. The rst three of these approaches share the basic formulation of the solution as a boundary value problem. The nite difference method in its elementary form, which is the oldest of the above methods, has the advantage of being conceptually simple and easy to understand. The nite volume method can be considered a variant of nite difference method and it has certain advantages with respect to the maintenance of local mass balance. The nite element method is highly exible and versatile in representing domains with irregular geometry or anisotropic and heterogeneous media; it is somewhat more complicated than nite differences, but it provides a very powerful and versatile tool to the user once the basic principles are mastered. In terms of overall accuracy, these methods are essentially equivalent. Any of these methods can be applied to the solution of ow and transport problems, which allows for the incorporation of chemical-biochemical interactions. Particle Tracking method differs from the above methods as it does not solve a boundary value problem, but instead considers the fate of tracer particles as they migrate through the system. It is therefore applicable to the solution of transport problems only. Particle tracking models are usually coupled with nite difference models for determining the ow eld. The main advantage of this technique is a simple and understandable concept. Its disadvantage is that chemical/biochemical interactions cannot be easily incorporated in a general way (although some advances in this area are being made). The method is particularly useful for the simulation of hydraulic remediation measures and the denition of well head protection zones. The user of numerical models should be aware of certain pitfalls which may affect the quality of numerical simulations. One of these pitfalls is numerical dispersion, which may arise in the solution of the transport equation by

dispersion). This effect can be deceiving to the novice because modeling results may look perfectly reasonable. Techniques for the control of numerical dispersion have been developed and are easy to implement. Excellent texts on groundwater modeling at the introductory level are Kinzelbach (1986), and Wang and Anderson (1982). Huyakorn and Pinder (1983) give an advanced treatment of numerical methods with a strong mathematical basis. For developing a groundwater model, following basic steps are involved: Develop a conceptual model of the system. Design a grid to represent the model area with as much detail as needed. Assign hydrologic properties to each grid cell. Select an appropriate time step for transient models. Set up boundary conditions. Select a solver. Run the model and compare with measured data to evaluate the effectiveness of the model. If necessary, revise the model. Plotting the data, and Tracking down errors.

Numerical Groundwater Models

The three dimensional unsteady movement of groundwater of constant density through porous earth material in a heterogeneous anisotropic medium can be described by the following partial differential equation: Where, x,y,z = Cartesian coordinates aligned along the major axes of conductivity Kxx, Kyy, Kzz h W Ss t = potentiometric head [L], = volumetric ux per unit volume and represents sources and/or sinks of water [T-1], = specic storage of the porous material [L-1] and, = time [T].

In general, Ss, Kxx, Kyy and Kzz are functions of space, for example; Ss = Ss(x,y,z), Kxx = Kxx(x,y,z), etc. whereas W and h are functions of space and time i.e W = W(x,y,z) and h = h(x,y,z). Equation (1) together with specication of ow conditions at the boundaries of an aquifer system and specication of initial head conditions, constitutes a mathematical model of ground water ow. Except for very simple systems, analytical solutions of equation (1) are rarely possible. Therefore, various numerical methods are employed to obtain an approximate solution of the above equation. One such approach is the nite-difference method. The continuous system described by equation (1) is replaced by a nite set of discrete elements in space and time, and the set of nite difference equations are solved numerically which yield values of head at specic points and times. These values constitute an approximation to the time-varying head distribution that would be given by an analytical solution of the ow equation. 155

NNRMS

BULLETIN

DECEMBER

2013

either nite differences or nite elements (the particle tracking method is not affected by numerical

Role of Remote Sensing Data for Use in Groundwater Modeling

Groundwater resources assessment, modeling and management are hampered considerably by the lack of data. Classical hydrological measurements provide only a limited number of point data, for example, at a weather station, a gauging station, or a borehole. One of the main problems in hydrological research to transfer from point information to regional distributed information. Regional hydrological models require spatial and temporal distributions of input and calibration data. If such data are not available, models cannot play their proper role in decision support as they are underdetermined and uncertain. Recent developments in remote sensing have opened new sources for distributed spatial data. Remote sensing offers a possibility to do this for certain parameters required in groundwater modeling. A very good account of use of remote sensing in groundwater modeling is given by Brunner et al., (2007). In general, only limited information on the spatial distribution of these input parameters (K, S and W) is available. Yet, a model computation needs a complete set of parameters. There are different ways to determine or estimate those. In traditional model calibration, the aquifer is divided into a limited number of zones. Within these zones, aquifer properties are assumed to be constant. This means a strong reduction in degrees of freedom. The zonation should be such that the parameters are expected to show little spatial variation within the dened zones. Remote sensing can play a role in the denition of these zones. For subsurface features, structural elements as seen in aerial geophysical surveys together with point data from drillings and pumping tests allow zoning. So the rst main use of remote-sensing data is seen in the spatial modulation and interpolation of input data, where otherwise a homogeneous value or a purely mathematical interpolation function would have to be used. During the process of model calibration, updated estimates of the missing parameters such as hydraulic conductivity (for the dened zones), are obtained such that a historical record of head and/or ux observations can be reproduced. This process is non-unique. Piezometric head data do not reduce the uncertainty of the estimated parameters of storativity, hydraulic conductivity and recharge, in case those parameters are only known within large error intervals. If, however, the spatially distributed input data can be constrained, the calibration problem stabilizes. If the spatial distribution of relative recharge can be estimated from land use and soil type and the yearly regional variation can be estimated from local lysimeter data, then the total function of recharge in space and time R(x,t) could be reconstructed as the product of the temporalspatial average of recharge Rav, i.e. R (x,t) = Ravf(x) g(t) Where, f(x) = relative values on areas with different land use and a weighting factor and g(t) = relative proportion of recharge in a certain time interval If f(x) can be obtained from remote-sensing data and g(t) can be determined from point data at a few lysimeters there is only one unknown parameter left and the large number of degrees of freedom residing in a temporal-spatial distribution collapses into one single number, the temporal-spatial average value Rav. Alternatively, remote sensing information on properties such as recharge could also be introduced in the traditional model calibration in the form of prior knowledge. Carrera and Neuman (1986) have shown that disadvantages of the model calibration can be mitigated by prior knowledge about the parameters to be estimated. Remote sensing can even be introduced as a kind of soft (not exact) information into the traditional zone based model calibration strategy. Remote sensing information intrinsically contains uncertainty because the correlation between remote-sensing patterns and ground truth will not be perfect. The stochastic modeling approach is able to use this type of information. The remote-sensing-based data and ground truth can be used to generate a series of equally likely images of the variable

simulation of exhaustive data on a regular grid is very useful for this purpose (Almeida and Frykman 1994). This algorithm simplies the relation between the remote-sensing data and the variable of interest to a linear correlation coefcient (Markov assumption) (e.g. Goovaerts 1997). The generated images are conditioned to the two different data sources, and take into account their estimated errors, a (linear) correlation between them, and a variogram estimated from the spatially distributed remotesensing data. These equally likely images of the variable of interest sample the high dimensional space of possible spatial distributions of the variable of interest within the uncertainty bounds mainly given by the mismatch between remote-sensing data and ground truth. In a case where the resolution of the hydrological model coincides with that of the remote sensing raster, and further, if the remotesensing data are perfect (perfect correlation between variable of interest and the measured signal), this space of possible spatial distributions would be reduced to one deterministic truth. As this truth will never be known, the stochastic calibration of a groundwater ow model consists of the selection of an ensemble of realizations of input data (combined from stochastic and deterministic information), which reproduce hydraulic head, ux and possibly tracer data to a predened degree.

Application of Remote Sensing Data in Groundwater Modeling

Airborne geophysical surveys allow for the identication of faults and dykes, changes in lithology and the depth of magnetic features. This information is helpful in constructing more realistic conceptual models of aquifers. An aquifer that is compartmentalized by dykes and faults will behave differently from an aquifer without such ow guides. Lineaments on the surface have been identied early as conduits for groundwater ow in fractured aquifers and hence targeted for locating production wells. Their use in geology is already widespread. The overlaying of lineaments mapped from conventional remote-sensing techniques (aerial photographs and satellite images) and those derived from airborne geophysical methods can be implemented using geographical information systems (GIS) at both local and regional scales. Some lineaments detectable by airborne geophysics may be due to deep-seated sources (up to several tens of kilometers) and hence have no effect on groundwater ow in aquifers of interest, which are mostly within a few hundreds of meters below the ground surface. Therefore, the depths to magnetic sources must be estimated in order to retain only lineaments that are deemed relevant to groundwater ow. On the other hand, lineaments identied with conventional methods give only information on structures with surface expression and no information on depth and vertical continuity of the structures. Space-borne gravitational surveys such as the Gravity Recovery and Climate Experiment (GRACE) mission can be used to detect temporal changes in the total water storage (surface water, soil water and groundwater). A 2 cm thick, innitely extended layer of pure water located at any depth below a gravimeter generates an incremental gravitational acceleration of 1108 m/s2 or 1 Gal (microgal). The temporal change in total water storage (TWS) in the Earth system is therefore directly proportional to the temporal change in the measured gravitational acceleration. The GRACE twin satellites have dramatically improved the accuracy and resolution of regional observations. This satellite mission delivers an accuracy of 0.4 Gal or 1 cm of groundwater on spatial scales larger than 1,300 km (Andersen and Hinderer 2005; Andersen et al., 2005) and delivers reliable observations of the regional part of the global hydrological cycle. Although the spatial resolution is still less than the size of typical groundwater systems, the prospects of this method for future use in verication of models, especially for the determination of the storage coefcient, are bright. 157

NNRMS

BULLETIN

DECEMBER

2013

of interest within the uncertainty of the correlation. The geostatistical algorithm developed for co-

For a phreatic aquifer, the surface of the terrain is also the upper boundary of the aquifer and constrains the groundwater levels. Surface elevations can be determined by various remote-sensing techniques, from airborne platforms (e.g. light detection and ranging LIDAR (Bufton et al. 1991), interpretation of stereo orthophotos, or satellite platforms using, for example, radar interferometry (Slater et al. 2006). In the latter case, the phase differences in pixels seen from different points in orbit allow a translation into differences in elevation. To obtain absolute elevation data and to verify their relative distribution, accurate elevation data at ground control points are required. These can be obtained, for example, with differential GPS (Global Positioning System). In many applications, the depth to groundwater is of importance for environmental reasons, including water supply to vegetation or salinization by phreatic evaporation. This distance is the difference between the surface elevation given by the Digital Elevation Model (DEM) and the groundwater level. High-precision measurements of the surface elevation changes can reveal regional subsidence caused by piezometric depression around well elds (e.g. Hoffmann et al., 2001) or seasonal variations of the groundwater level (Chang et al., 2004). Once a relation is given between subsidence and drawdown, a spatial distribution of drawdown can be obtained from the amount of surface subsidence observed. Differential GPS can also serve the purpose of determining temporal variations in the ground level related to groundwater pumping or recharge. Also, river and lake levels can be determined by using radar satellites (e.g. Berry et al., 2005; Jekeli and Dumrongchai 2003). Such data are available close to real time and can be of relevance for subsurface hydrology if they are indicative of groundwater levels (ESA, 2005). The bulk of remote-sensing data relevant for groundwater modeling are data that allow for quantication of the distribution of recharge or discharge. Recharge is one of the most important quantities for sustainable ground- water management. In dry regions, its estimation has been, up to today, a challenge, as it may occur only sporadically at intervals of several years. It may also be spatially very heterogeneous due to the distribution of precipitation, soil properties, water use by plants or runoff processes. One of the earliest applications of remote-sensing relevant in hydrology was the characterization of vegetation type, density and its status (e.g. Fensholt et al., 2006). This information is also of interest as a proxy for evapotranspiration (e.g. Loukas et al., 2005). Vegetation may be an indicator for the presence of water and the depth to groundwater level. For flat terrain, the groundwater recharge potential over long time intervals is the long-term average residual between precipitation (P) and evapotranspiration (ET). Both quantities can be estimated from remote-sensing data. The precipitation can be estimated from cloud temperature data (e.g. Herman et al., 1997) in combination with precipitation data from meteorological stations on the ground. Evapotranspiration can be derived from multispectral satellite data via a surface energy balance. For example, a dry pixel will heat up to higher temperatures than a pixel which has a large amount of water available for evaporative cooling. In this sense, radiation data can be related to evapotranspiration. The fraction of net radiation energy consumed by evaporating water can be estimated with different methods like, SEBAL (surface energy balance algorithm for land; Bastiaanssen et al., 1998a,b) in which the energy uxes in the surface energy balance are calculated explicitly, while in a simplied method described by Roerink et al., (2000), this fraction is determined from a pixel-wise plot of surface temperature versus albedo. Other methods use different dimensions of the feature space instead, e.g. the Normalized Difference Vegetation Index (NDVI), which is a measure of the vigor of vegetation growth (Sandholt and Andersen 1993). Though at present, both ET and P obtained from remote sensing may not be very accurate and the difference in PET may lead to error propagation, especially when both quantities are of similar magnitude. This is often the case in semiarid and arid areas. Still, the spatial patterns of P - ET may be of help in regionalization of traditional point measurements of recharge, e.g. obtained with the chloride method (Brunner et al., 2004).

precipitation is homogeneous. If water collects and inltrates in depressions, those may dominate the total recharge of an area. This process has been documented in Niger (Leduc et al., 2001). Water surfaces forming in the landscape and their temporal behavior can be identied by remote sensing, e.g. via radar or multi-spectral characterization (e.g. McCarthy et al., 2003; Roshier and Rumbachs 2004). Their density and distribution are indicative of the spatial distribution of recharge. In wetlands, the interaction between surface water and groundwater is crucial for the over time is, in this case, a valuable data set for model calibration (Bauer et al., 2006a,b). Groundwater recharge from rivers, streams and wetlands, under certain circumstances, can also be inferred from remote sensing through anomalies in temperature or electrical conductivity. In arid environments, evaporation is mostly through plants in the form of transpiration. This increases salinity in groundwater and hence electrical conductivity. The freshly inltrated water beneath a stream, in contrast, has a low electrical conductivity. The varying electrical conductivity of the underground can be detected by airborne electromagnetic methods (Paine and Collins 2003). In arid and semi-arid areas, the discharge of groundwater via direct evaporation from the water table and evapotranspiration by trees may account for most of the discharge of an aquifer. Discharge via a draining stream, as in a humid zone, rarely occurs. The estimation of discharge via trees has been the subject of remote-sensing studies looking both at ET derived from energy balance calculations as well as single tree counts according to species and canopy size and combining this remote-sensing information with information on the single tree, e.g. obtained from sap ow measurements (Lubczynski and Gurwin 2005). Salt crusts indicate high water tables with phreatic evaporation. They can be mapped by multispectral satellite data and used as an indicator for phreatic uxes and depth to groundwater (Metternicht and Zinck 2003). Soil-water balance calculations as a function of time require data in addition to average ET and P to account for water storage in the soil. A soil-water balance model requires some information on the eld capacity of a soil which could be estimated on the basis of the soil type. Here, hyperspectral satellite information can help (Chabrillat et al., 2002; Leone and Escadafal 2001; Shepherd and Walsh 2002; Ben-Dor et al., 2004) as well as gamma radiation counts from airborne platforms (Cook et al., 1996) indicating clay content (Rainey et al., 2003). Soil moisture itself and its temporal variation may in the future be accessible from passive and active microwave sensors. A mission, SMOS (Soil Moisture and Ocean Salinity) has been launched in 2009 by France to observe soil moisture over the Earths land mass with a repetitive coverage of 3 days. It is pertinent to emphasis that the moisture seen relates only to the top centimeters and the use of this data type requires substantial modeling. The vegetation vigor derived from multi-spectral satellite data can be used as an indicator for irrigation and can, therefore, be employed as a relevant parameter in monitoring the irrigated areas and for timing of irrigation (Droogers and Bastiaanssen 2002).

159

NNRMS

BULLETIN

understanding of the wetland behavior. The development of water surfaces and ooding patterns

DECEMBER

2013

The spatial distribution of recharge may be very heterogeneous even if the distribution of

Conclusion
The potential of remote sensing for improving groundwater models is considerable and still underutilised. The range of applications is substantial in model conceptualisation and model calibration. With all justied optimism, expectations for the easy use of remote-sensing data in groundwater modeling should not be exaggerated. The defaults of any single method can be counteracted by combining several methods. The remotely sensed data unfold the usefulness usually in combination with a model in which even noisy or correlated data can be used for conditioning. Finally, it should be remembered that the largest and most costly effort in applying remote sensing data to groundwater models lies in the eld work necessary to obtain a sufcient data base of ground truth.

References
Almeida, A.S., Frykman, P. (1994). Geostatistical modeling of Chalk properties in the Dan Field, Danish North Sea. In: JM Yarus, RL Chambers (eds) Stochastic modeling and geostatistics: AAPG computer applications in geology, no. 3. AAPG, Tulsa, OK. Andersen, O.B., Hinderer, J. (2005). Global inter-annual gravity changes from GRACE: early results. Geophys Res Lett 32: L01402. Andersen, O.B., Seneviratne, S.I., Hinderer, J., Viterbo, P. (2005). GRACE-derived terrestrial water storage depletion associated with the 2003 European heat wave. Geophys Res Lett 32: L18405. Bastiaanssen, W., Menenti, M., Feddes, R.A., Holtslag, A.A.M. (1998a). A remote sensing surface energy balance algorithm for land (SEBAL). 1. Formulation. J Hydrol 213(14):198212. Bastiaanssen, W., Pelgrum, H.J., Wang Ma, Y., Moreno, J.F., Roerink, G.J., van der Wal, T. (1998b). A remote sensing surface energy balance algorithm for land (SEBAL). 2. Validation. J Hydrol 213(14):213229. Bauer, P., Gumbricht, T., Kinzelbach, W. (2006a). A regional coupled surface water/ground water model of the Okavango Delta, Botswana. Water Res Res 42:W04403. Bauer, P., Held, R., Zimmermann, S., Linn, F., Kinzelbach, W. (2006b). Coupled ow and salinity transport modeling in semi-arid environments: the Shashe River Valley, Botswana. J Hydrol 316(14):163183. Ben-Dor, E., Goldshleger, N., Braun. O., Kindel, B., Goetz, A.F.H., Bonl, D., Margalit, N., Binaymini, Y., Karnieli, A., Agassi, M. (2004). Monitoring inltration rates in semiarid soils using airborne hyperspectral technology. Int J Remote Sens 25(13):26072624. Berry PAM., Garlick, J.D., Freeman, J.A., Mathers, E.L. (2005). Global inland water monitoring from multi-mission altimetry. Geophys Res Lett 32 (16):L16401. Brunner, P., Bauer, P., Eugster, M., Kinzelbach, W. (2004). Using remote sensing to regionalize local precipitation recharge rates obtained from the chloride method. J Hydrol 294(4):241250. Brunner, P., Hendricks Franssen, H.J., Kgotlhang, L., BauerGottwein, P. and Kinzelbach, W. (2007). How can remote sensing contribute in groundwater modeling? Hydrogeology Journal, 15 (1): 5-18. Bufton, J.L., Garvin, J.B., Cavanaugh, J.F., Ramosizquierro, L., Clem, T.D., Krabill, W.B. (1991). Airborne LIDAR for proling of surface topography. Opt Eng 30(1):7278. Carrera, J., Neuman, S.P. (1986). Estimation of aquifer parameters under transient and steady state conditions. 2. Uniqueness, stability, and solution algorithms. Water Resour Res 22(2):211227.

identication and mapping of expansive clay soils and the role of spatial resolution. Remote Sens Environ 82(23):431445. Chang, C.P., Chang, T.Y., Wang, C.T., Kuo, C.H., Chen, K.S. (2004). Land surface deformation corresponding to seasonal ground-water uctuation, determined by SAR interferometry in SW Taiwan. Math Comput Simul 67(45):351359. Cook, S.E., Corner, R.J., Groves, P.R., Grealish, G.J. (1996). Use of airborne gamma radiometric data for soil mapping. Aus J Soil Res 34(1):183194. European Space Agency (2005). River and Lake. http://www.earth.esa.int/riverandlake/. Fensholt, R., Sandholt, I., Stisen, S., Tucker, C. (2006). Analysing NDVI for the African continent using the geostationary meteosat second generation SEVIRI sensor. Remote Sens Environ 101(2):212229 FEWS NET (2006) Famine Early Warning Systems Network, http://www.fews.net/. Goovaerts, P. (1997). Geostatistics for natural resources evaluation. Oxford University Press, Oxford. Herman, A., Kumar, V.B., Arkin, P.A., Kousky, J.V. (1997). Objectively determined 10-day African rainfall estimates created for famine early warning systems. Int J Remote Sens 18(10): 21472159. Hoffmann, J., Zebker, H.A., Galloway, D.L., Amelung, F. (2001). Seasonal subsidence and rebound in Las Vegas Valley, Nevada, observed by synthetic aperture radar interferometry. Water Resource Res 37(6):15511566. Huyakorn, P.S., Pinder, G.F. (1983). Computational Methods in Subsurface Flow. Academic Press, New York: 473 pp. Jekeli, C., Dumrongchai, P. (2003). On monitoring a vertical datum with satellite altimetry and waterlevel gauge data on large lakes. J Geod 77(78):447453. Leduc, C., Favreau, G., Schroeter, P. (2001). Long-term rise in a Sahelian water-table: the Continental Terminal in South-West Niger. J Hydrol 243(12):4354. Leone, A.P., Escadafal, R. (2001). Statistical analysis of soil colour and spectroradiometric data for hyperspectral remote sensing of soil properties (example in a southern Italy Mediterranean ecosystem). Int J Remote Sens 22(12):23112328. Loukas, A., Vasiliades, L., Domenikiotis, C., Dalezios, N.R. (2005). Basin-wide actual evapotranspiration estimation using NOAA/AVHRR satellite data. Phys Chem Earth 30(13):6979. Lubczynski, M.W., Gurwin, J. (2005). Integration of various data sources for transient groundwater modeling with spatiotemporally variable uxes: Sardon study case, Spain. J Hydrol 306(14):7196. McCarthy, J., Gumbricht, T., McCarthy, T.S., Frost, P.E., Wessels, K., Seidel, F. (2003). Flooding patterns of the Okavango Wetland in Botswana between 1972 and 2000. Ambio 32(7):453457. Metternicht, G.I., Zinck, J.A. (2003). Remote sensing of soil salinity: potentials and constraints. Remote Sens Environ 85(1):120. Paine, J.G., Collins, E.W. (2003). Applying AEM induction in groundwater salinization and resource studies, west Texas.SAGEEP, Denver, CO, pp 722738

161

NNRMS

BULLETIN

DECEMBER

2013

Chabrillat, S., Goetz, A.F.H., Krosley, L., Olsen, H.W. (2002). Use of hyperspectral images in the

Rainey, M.P., Tyler, A.N., Gilvear, D.J., Bryant, R.G., McDonald, P. (2003). Mapping intertidal estuarine sediment grain size distributions through airbor ne remote sensing. Remote Sens Environ 86(4): 480490. Rodriguez, E., Morris, C.S., Belz, J.E., Chapin, E.C., Martin, J.M., Daffer, W., Hensley, S. (2005). An assessment of the SRTM topographic products. Technical Report JPL D-31639, Jet Propulsion Laboratory, Pasadena, CA. Roshier, D.A., Rumbachs, R.M. (2004). Broad-scale mapping of temporary wetlands in arid Australia. J Arid Environ 56(2):249263. Sandholt, I., Andersen, H.S. (1993). Derivation of actual evapotranspiration in the Senegalese Sahel, using NOAA-AVHRR data during the 1987 growing season. Remote Sens Environ 46(2):164172. Shepherd, K.D., Walsh, M.G. (2002). Development of reectance spectral libraries for characterization of soil properties. Soil Sci Soc Am J 66(3):988998. Slater, J.A., Garvey, G., Johnston, C., Haase, J., Heady, B., Kroenung, G., Little, J. (2006). The SRTM data nishing process and products. Photogramm Eng Remote Sens 72(3):237247. Wang, J.F., Anderson, M.P. (1982). Introduction to Groundwater Modeling. Freeman, San Francisco, CA: 237 pp.

Role of Earth Observation for Grass root level Water Resources Planning- Technology Demonstration for a Cluster of Villages in Semi-arid region of Rajasthan
Rama Subramoniam S, Manoj Joseph, Bera AK and Sharma JR Regional Remote Sensing Centre (West), NRSC, Jodhpur, India Email: ramasubramoniam@gmail.com

Introduction
Sustainable development aims at optimal use of natural resources, protection and conservation of ecological systems, and improving economic efciency. Food & Agriculture Organization (FAO) dened sustainable development as the management and conservation of natural resources base and the orientation of technological and institutional changes in such a manner as to ensure the attainment and continued satisfaction of human needs for present and future generations (FAO, 2013). Making sustainable development a living reality requires integrated assessment of ecosystem and physical environment with the help of Earth Observation data to know the potential and constraints of the area, socio economic situation analysis for need assessment and long-term planning and development of area specic enduser technologies, which are affordable and adoptable by the people. Approach towards sustainable development should consider regional issues like environmental, physical factors and social & cultural practices followed in the area. The most important characteristics of semi-arid areas, which limit availability of adequate soil moisture for plant growth, include high temperatures, low humidity, intense sunlight and high winds. These factors encourage very high rates of potential evapotranspiration to the extent that rainfall amounts exceed potential evapotranspiration only in very few and scattered days. The region often witnesses dry spell which occurs during the growing season. These dry spells occur with signicant variation from season to season in the same place and from place to place within the same season. In most cases therefore, good amount of rain occurs at the wrong time and/or place. Because of large spatial and temporal variability in rainfall distribution, rain-fed agriculture is very susceptible to water shortage. In general total annual rainfall is sufcient for crop production but the highly variable distribution in time and space frequently threatens crop production and contributes to food insecurity (Jennie Barron, 2009). Also due to unproductive soil evaporation, the measured rainfall is not available to the crops and results in low crop yields. In the western semi-arid region of India, there is a rich cultural practice to collect water in a village pond as runoff from the adjacent village common land/ grassland locally known as gochar or oran land and use it for domestic requirements, which include drinking water. With the increase in population and shrinkage in the runoff generating area, today the water is not sufcient and also collected water is highly contaminated due to various reasons and hence a large number of people are affected by water borne diseases in the area. On the

NNRMS

BULLETIN

DECEMBER

2013

agricultural front, though farmers are fully dependent on monsoon rain for crops, there is no practice of collecting and storing water individually or collectively for agricultural use/irrigation purpose (Sharma et al., 2011). Hence in-eld rain water harvesting can have an important impact and is a technique through which farmers can easily collect and store about 8- 10% of the water in their own eld for later use. The technology is low cost, highly decentralized empowering individuals and communities to manage their water (Critchley and Siegert, 1991). It has been used to improve access to water and sanitation at the local level. In agriculture, rainwater harvesting has demonstrated the potential of doubling food production by 100% compared to the 10% increase from irrigation. Rainwater harvesting is the viable adaptation strategy for people living with high rainfall variability, both for domestic supply and to enhance crop, livestock and other forms of agriculture (Prinz et.al.1998 and WRC, 2008). Advances in space based Earth Observation (EO) technology and its applications have great potential for water resources planning. High resolution spatial and temporal satellite data from Indian satellites like RESOURCESAT and CARTOSAT are available for creating various resource layers and GIS database on various resources on 1:10,000 scale. This scale sufces the requirement of decentralized planning at grass root level and along with overlay of cadastral boundaries will help in generating information in greater detail. Earth observations play an important role in understanding the terrain and micro-hydrology of the region along with natural resource status. This technique provides accurate, up to date and time series information that can be used for site selection of suitable Rain Water Harvesting (RWH) structures at regional as well as micro scales for operational decision making. The location suitability for RWH depend on amount of rainfall, soil characteristics (including texture, inltration and runoff rates) terrain features (including topography and slope), current land use/ land cover, and hydrological features. Earth observation is also useful for monitoring the status of RWH implementation and regional impacts of harvested water. This paper is based on technology demonstration for smart rainwater harvesting with the help of EO data. A pilot study has been carried out to identify the potential sites for rain water harvesting and also to assess the volume of harvestable rain water with the help of remote sensing and GIS techniques. The high resolution satellite data were used for natural resource characterisation. The Digital Elevation Model (DEM) is used to analyse the topography, slope and also to delineate potential drains. The integration of above thematic layers results in identication of potential rainwater harvesting sites at village level and also in calculating the volume
Fig. 1: Location map of the study area

of rain water than can be harvested from a

for food security, safe drinking water, introduction of agro horticulture, agro forestry practices and high value crops for livelihood and achieving long term sustainability of local and regional populations.

Study Area
Study area covers an area of 584.58 km2 consist of 50 villages in Jodhpur and Barmer districts of Rajasthan state (Figure1). It lies between 250 56 45N to 260 12 41N latitude and between 720 34 38E to 720 53 10E longitude. Land utilization pattern includes mixture of moderately cultivated land (mostly rain fed agriculture), open scrub, dense scrub, waterbodies and settlement during south west monsoon and this is main sources of water for kharif crops. Soils are loamy sand in texture followed by loam, sandy loam and ne sand. Out of the entire cadastral in 50 villages, 2.51% are of 0-1ha category and 22.96%, 18.97%, 47.1% are of area 1-3 ha, 3-5 ha and >5 ha class respectively. areas. The mean annual rainfall in this region is 275 mm. About 86.14 % of annual rainfall is received

Data & Methodology

Cartosat stereo data has been used to generate high resolution DEM for the study area. Potential drains have been generated from Carto DEM. Land use/land cover map of the study area has been generated from LISS-IV-Carto merged product. India Meteorological Department (IMD) daily grid data (1901-2004) has been used to analyze the rainfall pattern of the study area. Soil texture map from NRDB database has been used. Cadastral boundaries and other ancillary data has been collected form department of land and settlement, Rajasthan. Identification of the Potential Sites/fields for Rainwater Harvesting (RWH) The major steps involved in the process of identifying RWH sites are as follows. Baseline survey and creation of a geo-spatial database using GIS and Earth Observation data sets at cadastral level for current land use, soil resources and rainfall for the cluster of villages. Generation of high resolution DEM from Cartosat stereo data. Generation of slope map from DEM Delineation of Potential drains from DEM RWH sites identication analysis has been based on potential drains & its order and Multicriteria analysis using various parameters.

Multi-criteria Evaluation
Multi-Criteria Evaluation (MCE) is used to identify the suitability of each grid cell for water harvesting and storage. To generate multi-criteria based rankings, Weighted Overlay Process feature of GIS is used. From the literature review and information obtained from eld survey supported by expert judgment, ve criteria were selected for the identication of potential areas for in-eld RWH viz., soil texture (clay percentage), rainfall, slope, land use/cover and ow accumulation. Since all the criteria are not equally important for the identication of potential RWH areas, different weights (Rainfall and Flow accumulation (35%), LULC (15%), Texture (10%) and Slope (5%) were assigned to the criteria. Also high suitability rank was given for areas with large rainfall surplus as it ensures the availability of runoff to be harvested. High suitability rank has been given to low slope areas. Among the land use category, higher rank has been given to open lands, agriculture lands and dense scrubs. In the case of soil texture, ranks are assigned in such a way that areas with higher clay percentage are having 165

NNRMS

BULLETIN

DECEMBER

2013

given eld. The studies point to the importance of smart ineld rain water harvesting in semi-arid regions

higher rating as the run-off is more for clay sand. Block diagram showing the detailed methodology is given in Figure 2.

Assessment of Volume of Harvestable Rainwater

To assess the volume of water that can be harvested in a given eld, a methodology has been developed. The criteria considered different factors like soil texture, LU/LC, slope and rainfall of the given eld. Two methods have been proposed based on annual rainfall and event rainfall.
Fig. 2: Block diagram of multi-criteria evaluation for RWH site

Based on Annual Rainfall

In this study annual rainfall of the study area has been considered. During the

rainfall in the initial phase some of the water goes as soil moisture & seepage and after that runoff begins depending upon the intensity of rain. This runoff depends on various factors like type & size of catchment, slope, land use, texture of soil depth etc. Based on these factors run off coefcient is decided as per eld observations in an area. This runoff coefcient is a dimensionless gure and can be estimated from the individual eld data (Murthy, 2003 and CGWB, 2007). These coefcients have been adjusted for the Agricultural Land (no crop, early crop and late crop) and considering the soil factors like compaction, moisture and Inltration. This coefcient is multiplied with the annual rainfall (mm) and the drainage area (sq.m), which gives the estimate of the amount of water (L) that can be harvested in that particular area. It is assumed that around 20 % of water losses due to various factors like evaporation, seepage etc. The runoff coefcient depends on different factors like type of vegetation, slope and soil texture. Run off coefcients were derived based on two years experience at the pilot site, the intensity of rain observed and last ten years daily rainfall data of the area (Table 1). The formula for the calculation of total amount of rainwater available that can be harvested is: Rainwater Harvested (L) = Annual rainfall (mm) x 0.8 x Area (sq.m) x Runoff coefcient

Based on Event Rainfall

This concept takes the last 48 hrs rainfall and particular day intensity of rainfall. The following Table 2 describes R factor of amount of rainfall occurrence. The amount of water that can be harvested in a particular day is: Rainwater Harvested (L) = Rainfall (mm) x R factor x Area (sq.m) x Runoff coefcient

Results and Discussion

Natural resource characterization of the study area including rainfall, land use/land cover and soil texture has been carried out. Annual rainfall of the study area for the period from 1901-2004 has been analysed. IMD daily
Fig. 3: Annual rainfall and coefcient of variation of the study area

grid data of 10 m x 10 m has been used for

Dense scrub

0-2 2-5 5-10

0.1 0.1 0.16 0.10 0.10 0.25 0.28 0.30 0.40 0.17 0.20 0.30 0.24 0.25 0.35

0.3 0.3 0.35 0.30 0.30 0.36 0.47 0.50 0.60 0.31 0.33 0.39 0.38 0.42 0.48

0.4 0.4 0.40 0.40 0.55 0.58 0.60 0.70 0.33 0.37 0.40 0.40 0.45 0.50 0.50

Open scrub/Open land

0-2 2-5 5-10 0-2 2-5 5-10 0-2 2-5 5-10 0-2 2-5 5-10

Agricultural Land (early crop Stage)

Agricultural land (late crop stage)

Table 2: Amount of Rainfall and R factor Amount of rainfall (mm) R factor No wetness (Last 48 hrs < 20mm rainfall) <20 20-30 >30 0.10 0.40 0.50 Wetness (Last 48 hrs >20mm rainfall) 0.20 0.60 0.80 the analysis. The coefcient of variation of weekly rainfall during crop growing season over 100 years (Figure 3) shows high amount of variation and imply a high uncertainty in crop yields. Land use and land cover map of the study area is given in Figure 4. Land use pattern is predominantly agriculture (91%), followed by open scrub (5.6%), dense scrub (1.2%), settlement (1.92%) and waterbody (0.28%). The soil texture variation includes mostly loamy sand soils followed by loam, sandy loam and ne sand (Figure 5). Coarse texture indicates low water accumulation and ne
Fig. 4: Land use/land cover map

167

texture indicates high water accumulation.

NNRMS

Agricultural Land (No crop)

BULLETIN

DECEMBER

Type of Vegetation

Slope Range (%)

Runoff Coefficient Coarse (Sand, Loamy Sand, Sandy Loam) Medium (Silty Fine (Sandy Loam, Loam, Clay, Silty Silt, Sandy Clay Clay, Clay) Loam, Clay Loam, Silty Clay Loam)

2013

Table 1: Values of Runoff Coefcient Factor for different Soil Conditions

Soils are low in organic matter and poor in nutrients (available N is low, available P and K status is medium). Micronutrients such as Mn and Cu contents are adequate whereas Zn and Fe are decient. As soils have low clay and silt content, nutrient adsorption and retention by these soils are very low. Crop production is limited due to low soil fertility.

Identification of RWH sites

Potential drains are derived using Carto DEM. The existence of potential drains on agriculture elds gives high weightage for considering that eld for RWH. Analysis carried out by overlaying the potential drains on cadastral elds shows that 1480 elds (9.7%) having maximum potential to harvest rain water (Figure 6). Existence of potential
Fig. 5: Soil Texture map

drains during rainy season has been veried in the test eld. Multi-criteria analysis has been carried out based on layers viz. rainfall, land use, soil texture, flow accumulation and slope. Areas with good rainfall, higher slope and flow accumulation are having high suitability (Figure 7). A rainwater harvesting calculator has been developed to assess the volume of harvestable rainwater based on annual and event based rainfall (Figure 8). The tool will help the farmer to identify his eld and calculate the volume of water than can be harvested and also calculate the dimension of the water harvesting structures.

Conclusion
Role of EO data for identifying RWH sites in semi-arid region has been investigated. A technology demonstration has been carried out for a cluster of villages. The results obtained were validated in a test plot. The study underlines the convergence of science and technology with traditional wisdom.
Fig. 6: Identied potential RWH sites

The studies shows the importance of smart

ensuring for food security, safe drinking water, promoting agro horticulture, agro forestry practices and high value crops for achieving long term sustainable livelihood for local populations.

References
Central Ground Water Board (CGWB), Ground water.Ministry of Water Resources, Government of India. Newdelhi. Critchley, W. and Siegert, C. (1991). Water Harvesting Manual.FAO Paper AGL/ MISC/17/91, FAO, Rome.
Fig. 7: RWH site suitability based on multi-criteria analysis

FAO. (2013). World inventory of sheries. Conditions for sustainable development. Issues Fact Sheets. Ed. Rebecca Metzner and Serge M. Garcia. In: FAO Fisheries and Aquaculture Department. Rome. Jennie Barron, (2009). Rainwater harvesting: A lifeline for human well-being. United Nations Environment Programme, Stockholm Environment Institute, Sweden. Murthy, V.V.N. (2003). Land and water management engineering.Kalyani Publishers, New Delhi. Prinz, D., Oweis, T. and Oberle, A. (1998).

Fig. 8: In Field Smart Rainwater Harvesting Calculator

Rainwater Harvesting for Dry Land Agriculture - Developing a Methodology Based on

Remote Sensing and GIS. In Proceedings of XIII International Congress Agricultural Engineering, ANAFID, Rabat, Morocco. P. 2-6. Sharma, J.R., Jay Pearlman and Chilka Sharma. (2011). Developing Earth Observation based End user technology for making Sustainable development a living reality in semi-arid areas - Nurturing through convergence of technologies at grass root level. IEEE Global Humanitarian Technology Conference, DOI 10.1109/GHTC.2011.67 WRC, (2008). In-eld Rainwater Harvesting (IRWH) adoption on small farm plots. www.wrc.org.za.

169

NNRMS

BULLETIN

(2007). Manual on Artificial Recharge of

DECEMBER

2013

rain water harvesting in semi-arid regions

INDIA-WRIS WEBGIS DESIGN AND DEVELOPMENT OF WEB ENABLED WATER RESOURCES INFORMATION SYSTEM OF INDIA
Sharma JR and Project Team Project Director (India-WRIS) & Chief General Manager, RCs National Remote Sensing Centre ISRO, Department of Space, Loknayak Bhawan, New Delhi, India Email: jrsharma50@gmail.com

Introduction
In the emerging knowledge society and wide spread use to Information Technology, up-todate information on water resources is vital to support economic development, improve the quality of life as well as to conserve the nature and the environment and hence; an operational water resources information system at national level is essential for planning and development of the country. Looking at this CWC and ISRO are jointly executing since January 2009; the project Generation of database and implementation of web enabled water resources information system of India short named as India-WRIS WebGIS. India-WRIS WebGIS aims as a Single Window solution for comprehensive, authoritative and consistent data & information of Indias water resources in a standardized national GIS framework for planning, development and management of water resources. The current version 3.0 of the WebGIS portal is designed and developed keeping in view multiusers from all sections of society, varied and multi-source data input, current map policy, requirement of regular updates and near real time data accessibility, data security domains, scale of information and level of access of the portal as well as download of different GIS maps, data and value added products along with tool kit for further analysis and value addition. The information system has four key elements besides other facilities namely: 1. Data input/entry/collection system 2. Data storage, analysis, and transformation into user friendly information 3. Interactive system for geo-visualization and temporal analysis and 4. Information dissemination system in public domain as downloads and further processing tools

Water Information in Public Domain

Dissemination of data in public domain constitute the most important aspect of the water resources management being multi stakeholders involvement, peoples participation and information sharing to increase transparency, public awareness, elevating the importance of water information and enlighten public involvement in water management. The thrust has been towards development of an open source user end web enabled information system. It provides adequate and contemporary information on the state of water resources which are must for planning and water resources management strategy. This, in turn, will ensure increase in public awareness about the crucial

participation in management, planning and development of water resources of the nation leading towards the holistic goal of water security.

Water Wealth of India

On an average, India receives annual precipitation (including snowfall) of about spatial and temporal variations in the distribution of rainfall and hence in availability of water in time and space across the country. It is estimated that out of the 4000 km3 water, 1869 km3 is average annual potential ow in rivers available as water resource. Out of this total available water resource, only 1123 km3 is utilizable (690 km3 from surface water resources and 433 km3 from ground water resources). 4000 km3. However, there exist considerable

Water Resources India at a Glance

Area of the country as % of World Area Population as % of World Population Water as % of World Water Rank in per capita availability Rank in water quality Average annual rainfall Range of distribution Range Rainy days Range PET Per capita water availability (2010) Scope of the Portal Based on the requirements and data availability, comprehensive information have been collected, thought fully categorized and arranged in GIS environment under 12 major and 30 sub information systems besides base sub information system having large number of attributes data of last 5 100 years. 2.4% 17.1% 4% 132 122 1160 mm ( world average 1110 mm) 150 -11690 mm 5-150, most rain 15 days in 100 hrs. 1500-3500 mm 1588 m3

India-WRIS WebGIS Application Architecture (Technologies & Tools)

The three components India-WRIS Web GIS Application are: Database Design & Generation: The database for India-WRIS is highly complex with numerous sources involved. Much of the data is spatial in nature but the amount of associated data is very large and also having time series, and will further increase exponentially with the passage of time. The creation and management of such data is a colossal feat in itself and requires state of the art tools. The database standards and relationship have been developed for all type of data. The database generation software used have the capabilities of creating maps, viewing or exploring data, editing data, storing, conation (integrating datasets from different sources), transforming (into different coordinates systems, different representations, re-sampling, resulting in new representation/format of the same data), querying, analyzing etc. 171

NNRMS

BULLETIN

DECEMBER

2013

issues related with water and attract their

Water Resource at a glance Annual precipitation (Including snowfall) Precipitation during monsoon Evaporation + Soil water Average annual potential ow in rivers Estimated utilizable water resources

Quantity - BCM 4000 3000 2131 1869 1123

Percentage 100 75 53.3 46.7 28.1

Source: Water Resources at a Glance 2011, CWC, New Delhi, (http://www.cwc.nic.in) Basin Code 1 2a 2b 2c 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Basin Name Indus (Up to border) Basin Ganga Basin Brahmaputra Basin Barak and others Basin Godavari Basin Krishna Basin Cauvery Basin Subernarekha Basin Brahmani and Baitarni Basin Mahanadi Basin Pennar Basin Mahi Basin Sabarmati Basin Narmada Basin Tapi Basin West owing rivers South of Tapi Basin East owing rivers between Mahanadi and Godavari Basin East owing rivers between Godavari and Krishna Basin East owing rivers between Krishna and Pennar Basin East owing rivers between Pennar and Cauvery Basin East owing rivers South of Cauvery Basin West owing rivers of Kutch and Saurashtra including Luni Basin Minor rivers draining into Bangladesh Basin Minor rivers draining into Myanmar Basin Area of North Ladakh not draining into Indus Basin Drainage Area of Andaman and Nicobar Islands Basin Drainage Area of Lakshadweep Islands Basin Area (Sq. km) 453931.87 808334.44 186421.60 45622.41 302063.93 254743.31 85624.44 25792.16 51893.68 139659.15 54243.43 38336.80 30678.59 92670.51 63922.91 111643.87 46243.06 10345.16 23335.82 63646.21 38646.11 184441.06 5453.23 24731.08 29238.78 6918.20 462.59

Web Application & User Interface technology: The major user requirement from the web portal is data dissemination; hence advance GIS data processing systems at the back end augmented with the best database connectivity over the internet is used so that the user is able to get intuitive and real time information. User has the facility for data visualization, analysis on the client side and use further to create customized reports. Adobe Flex is able to deliver Rich Internet Application (RIAs) across the enterprise and over the web efciently. Using the Flex Application Programme Interface (API), India-WRIS combines GIS based Web services from ArcGIS Server

in simple, dynamic mapping applications over the Web. All the published map services are compliant with OGC standards and the services can be accessed using WMS, WFS, WCS and KML standard formats. India-WRIS system is using Oracle 11g, Relational Data Base Management System (RDBMS) which as oracle together used to handle geospatial data and to createmultiuser geo-database. Database Storage & Hosting: In order to ensure reliable and secure, 24 x 7 availability of the WebGIS, a robust hosting architecture has been designed. The same has been replicated at three places namely, RRSC (West) - Jodhpur, the data generation and s/w development as lead centre; NRSC - Hyderabad for web hosting and CWC - New Delhi for intranet users and data validation & updation. supports multi user system. ArcSDE as well

Public Outreach and Designing t h e I n t e r f a c e , To o l s a n d Facilities in India-WRIS WebGIS

Considering large number of factors as; type and volume of data, large number of varied users, ease of handling, varied nature of internet connectivity in the country, information requirement by the users and available technologies. The user interface of the portal has been designed carefully. The home page is divided into three sections: Main Menu Toolbar The main menu has six modules namely, WRIS Info Discovery, WRIS Explorer, WRIS Connect, Input Data Builder, Share Success Story and Create Your WRIS. This is the heart of India-WRIS information system, where all the major links to the various WebGIS modules are provided in a rich Graphical User Interface (GUI) assisted format for easy access and use. WRIS Info Discovery This module provides the user in discovering information contained in India-WRIS of a particular geographic area. The user can select area of interest based on the Administrative units, Hydrological units and Constituency wise and is presented with a condensed list of all the information available in India-WRIS for the area. 173

NNRMS

BULLETIN

DECEMBER

2013

with other Web content, which are displayed

WRIS Explorer This is the core module of IndiaWRIS WebGIS, where all the data can be explored and viewed using the various tools available for the purpose. Geo-Visualization - This section provides basic facility to visualize all the layers together in any combination by turning layers on and off as per the requirements. Sub-Information Systems - There are 12 major information systems namely, base data, surface water, ground water, hydroWeb Application Architecture

met, water quality, snow cover/glacier, inland Tasks Software / Technologies 2D 3D Mobile Meta Data Adobe Flex, HTML, PHP ArcExplorer, .NET, ArcGlobe .NET(Windows) / Flex (Android) / JAVA (Symbian) Visual Basic

Tasks and Software/ Technologies Used for India WRIS WebGIS Front end

Data Generation (Digital Image Processing / GIS Mapping) Publishing / Web Geodata Services Geodatabase /Back end

ERDAS Imagine, ENVI, ArcMap, ArcCatalog, ArcSDE, AutoCAD 3D, IGIS, Map Window Library, GeodatabaseXML ArcGIS Server Oracle 11g, MySQL navigation waterways, inter-basin transfer links, hydro-met extremes, land resources, water tourism and socio-economic. These have been further divided into 35 subinformation systems. Each sub-information system is based on a particular theme. It contains relevant layers and specially created tools to make the best use of the data. Temporal Analyst - A large amount of water resources and related data regarding hydrological, meteorological, pollution etc. are temporal in nature. In order to represent

Web hosting architecture

these datasets, a separate module has been created, where facilities are provided to

represent the time series data using suitable charts, animations and to compare the data across stations or years. Climate Trend Analysis - The aim of this section is to provide the users with a facility to analyze the changes in climate over the course of past 104 years (1900-2004). This analysis can be performed on various units viz. River Basin,

tools such as tabular, statistical and graphical analysis. WRIS Connect The sub-modules in WRIS Connect are: Live Telemetry Data - This system provides real time information such as Water Level, Rainfall, Air Temperature, Wind measured at 466 telemetry stations across the country. Water level data for the past
Use friendly GUI - India-WRIS Version 3.0

72 hours is displayed where as the other parameters of the past hour are presented on a dashboard. Query Interface - User can have a lot of queries answered directly through WRIS Explorer and associated available tools. To explore more details, user can place queries through Query Interface that contains set of fixed queries on various

2D-3D Linked View

hydrological parameters. The answers are generated through different permutations and combinations of these xed queries. The result of a query is displayed in spatial as well as non-spatial formats. Report Generation - This section has the utility to automatically generate report of the user dened area / region containing the all data into tables and maps and allows Save As and Download in .pdf format.

Linked View with Bhuvan Poral

Data Download - Apart from viewing the available data, the user may also wish to take the data and perform analysis / add value. This link allows the download of GIS layers and associated attributes. Automatic Map Generation - This module provides the user with a highly useful facility of generating high-quality, multi layered, theme based maps in GeoPDF format. The latest GeoPDF documents are
3D Viewing

175

highly versatile PDF documents that have

NNRMS

BULLETIN

Speed, Wind Direction and Solar Radiation

DECEMBER

2013

State, District, and Village through various

the features of a mini Geographic Information System complete with tools such as layer visibility toggling and attribute data viewing. Input Data Builder This module aims at keeping the data content of the various layers of India-WRIS up to date by providing facilities to the data providing sources to ingest the current attribute data directly into the relevant layers. The authorized users can enter the respective spatial and non-spatial data in the specied format into the information system through this facility. The three sub-modules of Input Data Builder are Spatial, Non-spatial and Metadata Input Builder. Share Success Story The objective of this module is to connect people for water resources planning and management by providing platform to upload the success stories, so that, others can view, interact and practice. Water Resources Planning & Management Create Your WRIS - This module provides facilities to the user to have further analysis of the downloaded data, adding new datasets using available general hydrology tools and generate report of the area. 2D-3D Linked View - Through this highly interactive feature, 3D y-through simulations can be generated along any linear feature like rivers, roads or user drawn features. The high resolution satellite imagery makes it a very realistic ying experience. Collaborative Planning In the water resources sector, need is often felt to share and work collaboratively for understanding and decision making. This tool provides the platform for this purpose wherein the users can share their screen views, drawing & actions while communicating through text and voice messages. General Information Toolbar This toolbar is available in the lower section of the home page and provides links to general information about India-WRIS as Visitor Number, Disclaimer, Sitemap, Links, Contact Us and Last Updated.

Conclusion
The hydrological processes are continuous as well as complex and therefore, an updated comprehensive, reliable and easily accessible Information System having time series data of the hydrological and meteorological observations is pre-requisite for effective management of water resources.

Acknowledgements
We are grateful to Dr. V.K. Dadhwal, Director, NRSC/ISRO for guidance, keen interest in the project and constant encouragement. Project Team wishes to acknowledge the help and support received from CWC ofcials and Irrigation Departments / Water Resources Departments of all the states of the country for making the large volume of time series data available for the project. We gratefully thank all the individuals for fruitful discussions, sharing their expertise and knowledge.

Universal Toolbar First section

Universal toolbar is present at the top of the home page and has two sections. The toolbar at the top of the page contains the links to popup window having information required by the user at any point and toolbar is visible at all times. The links in this toolbar are:

About WRIS

This page contains a brief overview of India-WRIS project including its history, scope, vision, goals, deliverables and time-frame for completion. This provides information for navigating through India-WRIS like screen resolution, keyboard shortcuts for easy navigation etc.

Tools Metadata

Numerous tools along with symbols for easy access are described in this part. This link leads the user to the Metadata Explorer which provides comprehensive information of the source of spatial and non-spatial data.

WRIS Wiki

Comprehensive information for the water resources assets and projects of the country is made available through WRIS Wiki application.

Help

A comprehensive and universal help is documented in this section assisted with diagrams, screenshots and short videos.

Search Second section

Consolidated search into the complete information system is provided. The advanced information toolbar is available right below the banner. It contains links to pages containing detailed information that a user requires when visiting the home page but may not require while exploring the other sections of the information system. The links available in this toolbar are:

Home Publications

This link leads to the main page of India-WRIS Portal. Various documents generated for India-WRIS are made available and reports being generated would be available to the users through links.

Gallery

This section presents the user with an image gallery of events related to development of India-WRIS project.

WRIS Mobile FAQ

A precise version of India-WRIS is being developed for mobile and handheld devices. This section contains answer to common questions and queries about the project and outcome.

Feedback Sign In / Register

Provides interface to post user suggestions and feedback. This section provides provision for new user to register and get connected to India-WRIS portal for updates. For downloads and providing data inputs, login is provided based on user categories and with a password. 177

NNRMS

BULLETIN

Accessibility

DECEMBER

2013

Six Categories of Tools

I. Navigation Tools Zoom In: It zoom into a particular area on map that is selected by the user. Zoom Out: Is zooms-out the map to come out of the detailing. Pan: It allows user to Pan around the whole map. Full Extent: It allows viewing the map at the full extent. Previous Extent: It allows going to previous extent when the map extent is changed. Next Extent: It allows returning from the previous extent. Map Overview: Provides location of current view in context with larger map area. Go To: Zoom to an area based on specied latitude and longitude. Select Area Zoom/Rubber Zoom: Smooth Zooming into a selected area. II. Display Tools Swipe: It swipe the selected layer in the map to reveal underlying layers. Spotlight: It removes overlaid layer from the selected portion for better visualization. Magnier: To view the zoom in layer details of selected portion only. Get Feature Info: On selecting a particular feature it displays summary info. Identify: identify the details of all the visible layers including the elevation details of the point. III. Search and Query Tools Query Builder: create user dened queries. Share a Link: Share the current view of map with another user through mail. iFrame: Sharing frame of India-WRIS in other applications. Links on twitter/Facebook/Google+. Surface Prole: Generate the surface terrain height graph of selected points on map. Network Analysis/Route Tool: It species the dened route of road, rail and river network. Tools to calculate parameters based on location and user input. Linked View: Can View multiple view of different information in a single window. VI. Sharing Tools Bookmark: Allows bookmarking a specic location on the map for future reference. Pin Mark: User can pin mark his location of interest and type his comment on same. V. Advanced Tools Draw: Allows to draw a shape/line/point on map. Print: To print the current viewing area in landscape or portrait mode. Save as Image. Search by Proximity. IV. Personalization Tools

Biswas, A. (1981). Integrated water management: Some international dimensions, J. of Hydrology, 51, 1-4, pp. 369-379. Bouwer, H. (2000). Integrated water management: Emerging issues and challenges, Agricultural Water Management, 45, 3, pp. 217-228. Ganapathy, C. and Ernest, A.N.S. (2004). Environ. Inf. Arch., 2, pp. 938945. Kaden, S., Becker, A. and Gnauck, A. (1989). Decision-support systems for water management. IAHS

Kumar, R., Singh, R.D. and Sharma, K.D. (2005). Water resources of India, Current Science, Vol. 89, No. 5, 10 SEPTEMBER 2005, pp. 794-811. Lal, M. (2001).Climate change Implications for Indias water resources. J. India Water Res. Soc., 21, pp. 101119. Loucks, D. (1995). Developing and implementing decision support systems: A critique and a challenge, Water Resource. Bulletin, 31, 4, pp.571-582. Maidment, D. (1997). Opportunities for the development of a global water information system. In : Land and Water Resources Information Systems - FAO Land and Water Bulletin 7 (Proceedings of a Technical Consultation Rome, Italy, 15-17 December 1997), pp. 115-120. OHagan, R.G., Robinson, B., Swan, G. and Finny, D. (2008). Web-based visualisation of water information: an overview. CSIRO: Water for a Healthy Country,National Research Flagship. Zalewski, M. (2002). Ecohydrology- the use of ecological and hydrological processes for sustainable management of water resources, Hydrological Sciences Journal, 47, 5, pp. 823.

179

NNRMS

BULLETIN

Publ. No. 180, pp. 11-21.

DECEMBER

2013

References