Hyperspectral Satellite-Derived Bathymetry Considering Substrate Characteristics

This is the authors' abstract. We don't add key points for this paper.

In optically shallow waters, the spectral characteristics of the substrate contribute to the water’s reflected signal, influencing satellite-derived bathymetry (SDB). Previous studies have often focused on the spatial distribution of substrates while overlooking its spectral characteristics. In this study, we proposed a hyperspectral satellite-derived bathymetry method (SDB-SS) that incorporates both the spatial and spectral characteristics of the substrate. Spectral data from the substrate are extracted from the instantaneous waterline, and three common water depth inversion models are applied for comparison, using ZY1-02D hyperspectral images of Lingyang Reef and Qilianyu to derive shallow water depths. Validation with ICESat-2 data shows that the RF algorithm achieved an RMSE of 1.05 m for the Lingyang Reef and 1.12 m for the Qilianyu. Compared to methods that ignore substrate characteristics, the SDB-SS method improves RMSE by 18.25% and R2 by 6.66% in Lingyang Reef. Finally, the performance of different inversion algorithms utilizing the SDB-SS method is further discussed and analyzed.