A Multi-satellite Remote Sensing Product of Reservoir Storage, Elevation, and Surface Area in South Asia

Monday, 15 December 2014: 9:00 AM
Shuai Zhang1, Huilin Gao1 and Bibi S Naz2, (1)Texas A & M University, Zachry Department of Civil Engineering, College Station, TX, United States, (2)Oak Ridge National Lab, Oak Ridge, TN, United States
Reservoir storage information is essential for accurate flood monitoring and prediction. South Asia, however, is dominated by international river basins where communications among neighboring countries about reservoir storage and management are extremely limited. A suite of satellite observations were combined to create a high quality reservoir storage product in South Asia from 2000 to 2012. The approach used water surface area estimations from the Moderate Resolution Imaging Spectroradiometer (MODIS) vegetation indices product and the area-elevation relationship to estimate reservoir storage. The surface elevation measurements were from the Geoscience Laser Altimeter System (GLAS) on board the Ice, Cloud and land Elevation Satellite (ICESat). In order to improve the accuracy of water surface area estimations for relatively small reservoirs, a novel classification algorithm was developed. This remotely sensed product contains time series of reservoir elevation, area, and storage for a total of 21 reservoirs, which represents 28% of the integrated reservoir capacity in South Asia. The remotely sensed results were validated comprehensively over five reservoirs through two steps. First, the MODIS surface water classification images were compared with Landsat high resolution (30 m) classifications. Second, the reservoir elevation and storage dataset from remote sensing was evaluated using gauge observations. . The storage estimates were highly correlated with observed values (i.e., the correlation coefficients were all larger than 0.9), with normalized root mean square error (NRMSE) ranging from 9.51% to 25.20%. Uncertainty analysis was also conducted for the remotely sensed storage estimations. For the parameterization uncertainty associated with surface area retrieval, the storage mean relative uncertainty was 3.90%. With regard to the uncertainty introduced by ICESat/GLAS elevation measurements, the storage mean absolute uncertainty was 0.67%.