Shallow water bathymetry based on seafloor classification and active–passive data

High-accuracy bathymetric data are crucial for navigation safety, as well as for research on coastal ecosystems and related fields. In this regard, optical passive bathymetric inversion is limited by its in-situ measurement data requirement and that it fails to consider the inhomogeneity of seafloor substrate. To address this issue, this study proposes an active–passive fusion bathymetric inversion technique with seafloor classification. First, we partition the seafloor substrate into bottom-reflectance-related optical zones using Sentinel-2 optical imagery, the depth invariant index and the k-means clustering algorithm. Second, based on the seafloor classification, the active–passive technique is employed to fuse the ICESat-2 and Sentinel-2 satellite-based data to obtain the bathymetric maps of Yongle Atoll. The results indicate that the proposed method can weaken the influence of complex substrates and improve the bathymetric accuracy: the root mean square error is 1.64 m and the R2 is 0.86 for Ganquan Island, and the root mean square error is 0.99 m and the R2 is 0.88 for Shiyu Island.