Document Control Sheet 1. Security Classification For Official Use only 2. Distribution NRSC / ISRO, Online Distribution Through Bhuvan Portal 3. Report / Document version (a) Issue no. 01 4. Report / Document Type Technical Document 5. Document Control Number NRSC /February, 2016/TR-796 (b) Revision & Date 6. Title Water Bodies Information System (WBIS)-Technical Document 7. Particulars of collation Pages 11 Figures 3 Tables 1 References 5 8. Project Team /Authors Algorithm Development and Updation, Development of Water Bodies Information System (WBIS) Software Development and Productization in IMGEOS from DPSD&DBG/SDAPSA Support for hosting on Bhuvan Dr. A.V. Suresh Babu, S/E-SF Shri. S. Karthik Reddy, S/E-SC Water Informatics and Quality Division Water Resources Group, RSAA, NRSC Core Implementation :Ms. Adiba F, S/E-SC, Ms. N. Sai Kumari, S/E-SE Scheduler : Mr. Uzair Mujeeb, S/E-SD Dr. Naresh Kumar, S/E-SE UOPS& Utilities : Ms. Richa Goyal, S/E-SD Mr. Amit Kumar, S/E-SD Sri. M. Arul Raj, S/E-SE Ms. Sonal Aggarwal, S/E-SD 9. Affiliation of authors NRSC,Hyderabad 10. Scrutiny mechanism 11. Originating unit NRSC 12. Sponsor(s) / Name and Address Guidance / Reviewed by : Shri. K. Abdul Hakeem, Head, WIQD,WRG,RSAA Ms. Manju Sarma, S/E-SG, Group Head, Dr. V. Venkateshwar Rao, Group Director, WRG,RSAA Dr. Vinod M Bothale, Group Director-G &WGS,DPPA&WAA Approved /Controlled by : Dr. P.G. Diwakar, DD-RSAA Dr. B. Gopalakrishna, DD-DPPA & WAA In-house R & D 13. Date of Initiation 01.04.2012 14. Date of Publication 01.02.2016 1 P a g e
15 Abstract Automated water bodies extraction algorithms were developed using Resourcesat-1 AWiFS (56m) / RISAT-1 MRS (18m) Sensor datasets. Spectral knowledge based hierarchical automated algorithm was developed for quick processing of satellite data and to extract water bodies information in near real time. As there are number of inland water bodies that are useful for agriculture and drinking water purposes exist, it is necessary to monitor them on a regular basis. NRSC had made efforts to process the satellite data at every 15 day interval and host the water bodies information on Bhuvan portal. Water Bodies Information System (WBIS) has been developed for integral view of dynamic satellite derived water spread area information to understand the variations within a water body over pre-identified Water bodies (~2,00,000 No. which are greater than 2 Ha) and also at regional level such as River basin, River sub basin, District, and State level. This facilitates visualization along with analysis tools for understanding the water bodies dynamics. This document describes the technical procedures used for processing of the satellite data and presentation of the same in Water Bodies Information System (WBIS). The accuracy, limitations, possible applications are summarized in this document. 2 P a g e
Water Bodies Information System (WBIS) 1. Introduction Satellite sensors can capture the scenario of surface features at frequent intervals facilitating mapping and monitoring of surface water bodies at regular intervals. Satellite data utilization for inventory, mapping and monitoring water bodies over larger areas is a popular technique for generation of dynamic databases. There are several water bodies in India in the form of major/ medium artificial reservoirs, irrigation tanks, lakes, ponds, which are primarily used for agriculture and drinking water purposes. Variation in water spread occurs as a function of rainfall amounts, intensity of rainfall, etc. over season / year. Indian Remote Sensing Missions like Resourcesat-2 is provided with AWiFS sensor (56m, 5 days revisit) & LISS III sensor (24m, 24 Days repetivity) and RISAT-1 (microwave satellite) is with MRS Sensor (18m, 23 Days repetivity) and CRS Sensor (36m, programmable coverage) can provide frequent coverage of India. This is useful for the generation of water bodies information at fortnightly / monthly time interval. Currently, these datasets are being utilized for the generation of fortnightly / monthly water bodies information. In view of huge datasets that are to be processed on daily basis to cover entire India, NRSC has developed automated water bodies extraction algorithms for quick processing and automation of satellite data processing in IMGEOS as soon as it is acquired. Thus the water bodies layers derived are integrated into a common geospatial database. View of information updated in the geospatial dynamic water bodies database is presented over pre-identified water bodies ( for about ~2,00,000 No (>2 Ha)) is presented in the form of Water Bodies Information System (WBIS)for better visualization and analysis of water spread over India. The brief methodology for satellite data processing and development of water bodies information system is described in this document. 2. Automated algorithms for extraction of surface water bodies Water bodies database has been generated with a combined use of Reourcesat-2 AWiFS (optical data) and RISAT-1 MRS (microwave data) for which automated algorithms have been developed & operationally implemented in IMGEOS in near real time. 3 P a g e
2.1. Processing of Resourcesat-2 AWiFS/LISS III data Space-borne multi-spectral satellite images are analyzed using supervised/unsupervised classification techniques available in commercial /open source image processing softwares. The parameters for image classification are provided manually for each of the scene and the process is repeated for all. This process is time consuming and manual interpretation is required for identification of thresholds and assigning clusters to water feature class. Automatic algorithm developed by NRSC (S. Subramaniam, A.V. Suresh Babu and P.S. Roy 2011) for quick processing of satellite data was used for generation of a spatial water layers. Automated algorithm is a spectral knowledge driven, developed after spectral characterization and transformation to mathematical relationships viz. band ratios, band thresholds & other combination of relationships across the spectral data of four bands 0.52-0.59 µm (Green), 0.62-0.68 µm (Red), 0.77-0.86 µm (NIR) and 1.55-1.70 µm (SWIR) and hierarchical, multi-logic algorithm is developed. The few logics used are :Water has lower reflectance in the Visible-SWIR spectral region compared to other surface cover features ;Very high contrast between Green & NIR (except turbid water pixels) and Green & SWIR due to the absorption of NIR & SWIR radiation by water ; Reflectance of water in NIR is lower than that of Red band. (Ref.4) Algorithm is well trained over larger datasets across the country and is scene independent, which is operationally used for extracting water feature. A typical satellite derived water bodies layer is shown in Figure.1. Figure.1. Example showing Resourcesat-2 AWiFS image and the corresponding water layer 4 P a g e
2.2. Processing of RISAT-1 MRS / CRS Data RISAT-1, a microwave satellite data has been used in addition Resourcesat-2 AWiFS for taking of advantage of microwave imaging in case of cloud cover and also to have additional frequency of coverage for generation of fortnightly / monthly water layer. RISAT-1 MRS sensor (18m and 128km swath) has systematic coverage over India with C Band, dual polarization (HH, HV). Back scatter coefficient (σ o ) was estimated for each of HH, HV images and this information was used for the generation of knowledge base, and to develop an algorithm for identifying water pixels. Methodology consists of characterization of microwave response to various types of water features, observations on behavior of back scatter coefficient (σ o ) from these features through generation knowledge base, and identification of thresholds for back scattering coefficients to enable water pixel detection. Legacy database on water bodies was also used for improving the thematic accuracy. (Ref.5). A typical satellite derived water bodies layer is shown in Figure.2. Figure.2. Example showing RISAT-1 MRS image and water layer 3. Productization in IMGEOS IMGEOS is a ground segment infrastructure designed as a multi layer architecture having User Services, Data reception & Processing systems and Storage along with Security and Monitoring components to meet the increasing demand of varied types of satellite data products acquired from all IRS Earth Observation missions. In IMGEOS the data product generation chain (Figure.3) is triggered by work-orders from User order processing system (UOPS). The chain comprising of dedicated processing systems controlled by Data product work-flow manager (DPWFM), generates products as per the defined processing level. ( Ref.2, Ref.3) 5 P a g e
The Information product generation system (IPGS ) is a newly added work-flow in IMGEOS to automatically generate water layers from standard products of Resourcesat-2 AWIFS sensor in near real time. Every day 1-2 passes of AWIFS sensor data covering Indian region are processed in near real time to generate scene based water layers. These scene based layers are used to generate fifteen day composite mosaic of Indian region. Figure.3. Flow chart IPGS chain The major elements under IPGS production chain are: Product generation Initiation through Work Order generation: IPGS Data processing Chain is initiated with IPGS Work Order generated at User Ordering system. DPWFM routes this work-order to DPGS work-centre to generate terrain corrected products Information Product Generation System : IPGS Scheduler at IPGS work-centre processes the scene wise work-orders, invokes and manages the IPGS Layer Generation Software to generate scene wise water layer. Once all scene based water layers of the path are generated the path wise Mosaic generation 6 P a g e
is initiated. Once in five days all the path wise mosaics are used to generate Indian region mosaic. The three, five day Indian region mosaics are time composited to generate the fifteen day mosaic product. Product Dissemination to FTP: The generated water mosaic layers are then transferred to FTP Server via Data Exchange (DEG) system and are downloadable from FTP Server through a FTP user-account. Chain Monitoring and Routing through DPWFM and EMC: DPWFM generates status messages of processing done at each work-centre in the chain for the use by EMC software to allow graphical monitoring of progress done in IPGS Chain. IPGS System and Software architecture: User access Tier: work-order generation, User access for finished Products on FTP Control tier: Routing and work-order management (DPWFM), IPGS Scheduler Production tier: Level-0 processing subsystem, Data Product generation. Information Product generation system(ipgs) (IPGS), Mosaic generation Monitoring tier: Production Monitoring, & Reports (EMC) Data tier: (SAN Storage) Work-order Data base, Reference Masks, Ancillary Meta database, Satellite Products Water Layers IPGS software is a new component in IMGEOS consisting of Information Layers, Mosaic generation and Scheduler software. The layer and Mosaic generation software is developed and optimized in IMGEOS framework with standardized interfaces for automation and error handling in production chain using open source tools. The synchronization of Mosaic generation and time composition are achieved using system level daemon services. The IPGS scheduler is capable of distributed scheduling to take advantage of available processing nodes. 7 P a g e
3. Availability of water layers The availability of water layers is provided in the following table.1. Table.1. List of time series water layer information hosted on Bhuvan 2016 2015 2014 2013 2012 January February March April May June July August September October November December Data available Data to be processed 4. Water Bodies Information System (WBIS) WBIS is developed using latest front end and backend open source web technologies. The front end is developed on top of Angular JS. Leaflet along with a few plugins, Angular material framework, d3 and its derivative libraries are utilized in developing the user interface, data visualization and map elements. Raster data is automatically pooled from time-series water layers into a Relational database and processed to generate individual 8 P a g e
water body wise, regional (River Basin, River Sub-basin, State, District) spread area statistics using a software written in C++. A REST API for this data is hosted using node.js. Components of WBIS : 1. Home Page: Interface for selection of water body/region. Link for layer selection, map facilities: Layer manager, search by location, selection of time series spatial water layer view 2. Water Spread area visualization widgets 9 P a g e
3. Region wise statistics 5. Accuracy assessment and Limitations Accuracy assessment of AWiFS derived water layer was carried out with random grid sampling (75 x 75pixels) representing different geographical regions spread across the country, for different time periods. The overall accuracy is found to be 93.5% with kappa coefficient of 0.821. Cloud cover will limit the satellite data acquisition and hence, partial data will be available. Cloud free regions were only represented in Water Bodies Information System. The accuracy of water pixel identification is influenced by hazy cloud, cloud shadows and also land water boundary pixels (mixed area). There is possibility of over estimation in smaller water bodies : 10-25 Ha as there are more number of land water boundary pixels and moist pixels around which contiguous to the main water body. In case of microwave data processing, the vegetation over smaller water bodies is influencing roughness on the surface resulting in higher back scatter coefficient and hence, accuracy of detection of water pixel is affected in some areas. Glacial lakes are not part of this database. There could be errors due to cloud or non detection of water pixels from the knowledge based automated algorithm and such inaccuracies observed will be considered for updating the knowledge base and improve the automated water bodies extraction algorithms. 6. Possible Applications Facilitates study of water surface dynamics at regional/ river basin/river sub basin level for trend analysis and generate scenario of inter / intra annual / seasonal dynamics. 10 P a g e
Analysis of relationships between regional rainfall scenarios vs. water spread in major, medium reservoirs and status of smaller irrigation tanks. This will help in understanding water storages and utilization across the time periods. Inventory, mapping and monitoring of surface water area at frequent interval. 7. References [1] S. Subramaniam, A.V. Suresh Babu and P.S. Roy, 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, 2011. [2] IPGS Design Document,NRSC-DPSD&DBS-RS2-Aug2013. [3] IMGEOS Design document : NRSC-DPS-IMGEOS-Sept 10-TR187. [4] NRSC Document. Satellite derived Information on Water Bodies (WBA) and Water Bodies Fraction (WBF). NRSC SDAPSA-RSAA-Feb 2014-TR-580. [5] NRSC Document. Near Real Time Extraction of Water bodies through Automated Algorithm from RISAT-1 MRS Datasets in IMGEOSNRSC RSA /SDAPSA Jun 2014. Acronyms IMGEOS - Integrated Multi-mission Ground Segment for Earth Observation Satellites RISAT-1 Radar Imaging SATellite -1 MRS Medium Resolution Sensor IPGS-Information Product Generation System UOPS User order Processing System SAN Storage Area Network SWFM Station Work Flow Manager EMC Enterprise Monitor and Controller DPGS Data Product Generation System SWFM Station Work Flow Manager AOI- Area of Interest DEG-Data Exchange Gateway. 11 P a g e