Monitoring water level fluctuations of reservoirs in the krishna river basin using sentinel-3 and icesat-2 altimetry data

Abstract

The traditional approach to water level monitoring, using sensor devices to automatically measure levels in tanks or reservoirs, is costly and requires regular maintenance. This makes satellite altimetry crucial for reducing monitoring expenses and enhancing the observation of inland water bodies. This study employs Sentinel-3 and ICESat-2 sensors to examine six major reservoirs in the Krishna River basin in India, each intersected by at least two altimetry tracks. A two-way retrieval method for Sentinel analysis, including individual wet tropospheric correction and back-substitution, determines water levels. A novel hydro-flatness test removes outliers, increasing efficiency and saving time, though it is limited when fewer than ten high-distortion footprints are present. The 2018-2022 time series was effectively built for all reservoirs after assessing sensor performance against Central Water Commission data. The study successfully generates water level time series for the reservoirs, with correlation coefficients above 0.96 and Mean Absolute Percentage Errors below 1%. The Root Mean Square Error (RMSE) remains under 0.45m for all reservoirs except Ujjani, monitored by Sentinel-3A, which has an RMSE of 0.53 and a correlation of 0.96. These results can be applied to flood forecasting, reservoir operations, bathymetry retrieval, river modeling, and long-Term water level-discharge curves. © 2024 The Authors.