Remote sensing monitoring on river water level based on multi-source satellite collaboration

Water level is a critical indicator of river hydrological conditions.High-precision monitoring of river water level is essential for effective water resource management, disaster mitigation, and ecological conservation.This study investigates the feasibility of remote sensing on river water level by integrating data from the Chinese GaoFen-7 (GF-7) stereo mapping satellite with other optical satellites, including GF-1/2/6, Landsat-8/9, and Sentinel-2.The research focuses on the Shigu and Benzilan reaches in the upper Jinsha River.High-resolution topographic data of the riverbanks were derived from GF-7 satellite stereo images, which were subsequently used to develop water area-level relationship models for different river reaches.Dynamic remote sensing of river water level was achieved by incorporating time-series data of the river water surface area into these models.The accuracy of the remote sensing-derived water levels was validated against in-situ measured water levels.Results show that the mean absolute error (MAE) of remote sensing-derived water levels for the two river reaches ranges from 0.27 to 0.31 m when using a combination of high-spatial-resolution satellites such as GF-1/2/6.The root mean square error (RMSE) of remote sensing-derived water level variations in the two reaches ranges from 0.280 to 0.369 m.For water level fluctuations exceeding 1 meter, the accuracy of remote sensing-derived water level variations is approximately 84% for the Shigu reach and 90% for the Benzilan reach.When utilizing medium-to-high spatial resolution satellites such as Landsat-8/9 and Sentinel-2, the MAE of remote sensing-derived water levels for the two reaches ranges from 0.35 to 0.56 m.This methodology demonstrates significant potential for capturing river hydrological dynamics, providing a valuable tool for dynamic remote sensing of river water level in ungauged regions.