Validation and application of water levels derived from Sentinel-3A for the Brahmaputra River

For the first time ever, this study aims at applying Sentinel-3A to the Great Brahmaputra River (GBR) and validating water levels derived from this newly-launched altimetry satellite mission. The GBR is divided into three primary parts: (1) a large section of the Yarlung Zangbo River in China, also termed the Upper Brahmaputra River (UBR) in this study, featured by high elevation, complex terrain, narrow river widths (from less than 100 to 400 m), and limited in situ measurements; (2) the Middle Brahmaputra River (MBR) with widths varying from similar to 400 m to similar to 1 km; (3) the Lower Brahmaputra River (LBR), dominated by braided channels with wide river channels (up to several kilometers). For the altimetry data, both waveform retracking and hooking effect correction were applied to mitigate the influence caused by land contamination and to improve the accuracy of water level retrievals. Water levels were derived from 41 virtual stations (VSs) across the GBR and different retracking algorithms were compared with in situ data from two gauging stations in the UBR. Time series of altimetry-based water levels were categorized into three types based on the quality: no data (type 1), degraded (type 2), and good (type 3). Results showed that the VSs (type 1) only existed in the mountainous regions, accounting for similar to a half of the total in the UBR. Validation with the gauged water levels showed that the TIC algorithm performed best among all of the retrackers applied, followed by the Ice-1 algorithm. The standard deviation of the difference between the gauged and TIC-derived water levels ranged from 0.41 to 0.76 m among four different VSs (type 3). In addition, the quality of VSs in the LBR was best, followed by the MBR. Our study has demonstrated the capability of Sentinel-3A in monitoring water levels in the GBR, thereby paving the way for future applications such as discharge estimation and hydrologic/hydrodynamic applications.