Gravity-Geological Method for Seafloor Topography Inversion Considering Spatial Heterogeneity

Seafloor topography is highly important in marine science as a crucial physical background field. Seafloor topography inversion mainly relies on marine gravity data to supplement missing bathymetry data in areas lacking ship measurements. However, existing inversion algorithms often overlook the geophysical properties and spatial differences of seafloor topography. As an improvement, we develop a new seafloor topography inversion algorithm based on topographic heterogeneity partitioning where different parts having different densities. It consists of three steps: First, by quantifying the slope of observed vertical gravity (VG), a 94% change rate as the threshold for partitioning topographic heterogeneity, and a quadtree spatial indexing algorithm is used to segment the observed sea area into sub-regions. Secondly, the constrained ship soundings data and the optimal reference density contrast for each subregion, derived by minimizing the misfit between observed and calculated ship soundings data, are used as inputs for the VG inversion. Finally, a geographically weighted linear equation system is derived from the observed gravity data, and the least squares method is employed to solve for the inversion of seafloor topography. A case study in the western Pacific Ocean shows that the RMSE between the inversion results and the bathymetry data is 83.6 m, representing an 11.5% reduction compared to the traditional GGM that does not consider topographic segmentation. Compared to the GEBCO_2024 model, this algorithm reduces the overall RMSE by 38%.