Toward Automated Generation of Reservoir Water Elevation Changes From Satellite Radar Altimetry.

Abstract:

Until now, processing satellite radar altimetry data over inland water bodies on a large scale has been a cumbersome task primarily due to contaminated measurements from their surrounding topography. It becomes more challenging if the size of the water body is small and thus the number of available high-rate measurements from the water surface is limited. A manual removal of outliers is time consuming which limits a global generation of reservoir elevation profiles. This has limited a global study of lakes and reservoir elevation profiles for monitoring storage changes and hydrologic modeling.

We have proposed a new method to automatically generate a time-series information from raw satellite radar altimetry without user intervention. With this method, scientist with little knowledge of altimetry can now independently process radar altimetry for diverse purposes. The method is based on K-means clustering, backscatter coefficient and statistical analysis of the dataset for outlier detection. The result of this method will be validated using in-situ gauges from US, Indus and Bangladesh reservoirs. In addition, a sensitivity analysis will be done to ascertain the limitations of this algorithm based on the surrounding topography, and the length of altimetry track overlap with the lake/reservoir. ­­

Finally, a reservoir storage change will be estimated on the study sites using MODIS and Landsat water classification for estimating the area of reservoir and the height will be estimated using Jason-2 and SARAL/Altika satellites.