A reservoir storage estimation algorithm using digital elevation data and image classifications

Due to the limited spatial resolution of past and existing satellite altimeters, only about 15% of global reservoirs (by storage capacity) can be monitored from space. To estimate the storage variations of more reservoirs via satellite observations, a 3-step algorithm using Moderate Resolution Imaging Spectroradiometer (MODIS) images in combination with the Digital Elevation Model (DEM) from the Shuttle Radar Topography Mission (SRTM) is developed. First, water surface area is estimated from MODIS via an enhanced classification procedure. Second, a water surface area — height (A-H) relationship is generated using DEM information by regressing the cumulative area against the elevation values within the delineated reservoir region. Third, the reservoir storage variations are calculated by applying the water surface area estimations to the A-H relationship. Here, the new algorithm is validated (against gauge observations) over three reservoirs which are not covered by any existing spaceborne altimeters. The coefficients of determination from the validation are 0.84, 0.90, and to 0.90 for these reservoirs. By applying this algorithm to South Asian reservoirs, a total of 27 reservoirs representing 45% of the region's overall storage capacity can be monitored.