Shallow water depth (≤5 m) estimation based on single-beam echo sounder and optical satellites

This study combined single-beam echo sounder data with PlanetScope multispectral data to invert shallow water depth (≤5 m) around Weizhou Island, analyzing how each band's reflectance varies with depth by establishing their quantitative relationship and building nine statistical regression and machine learning models. In the inversion of water depths less than 5 m, the correlation R2 between the blue and green bands and water depth was less than 0.1, while the R2 between the red-edge band and water depth was 0.73. In addition, after classification of the sediment type, water depth inversion improved the correlation between the water depth and reflectance. The random forest (RF) and support vector regression (SVR) models demonstrated the highest accuracy in terms of water depth inversion, with R2 = 0.87, RMSE = 0.28 m and MAE = 0.18 m.