Enhancing Seafloor Topography Inversion Based on Marine Gravity Data Using Robust Weighted Total Least Squares

作者：Fengshun Zhu, Rumeng Guo, Jinbo Li, Jianqiao Xu, Jiangcun Zhou, Jinyun Guo, Shuai Zhou, Yang Li, H. Sun · 发表于：IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · 年份：2025 · DOI：10.1109/jstars.2025.3649032 · 被引用次数：2 · 研究领域：Geophysics and Gravity Measurements、Underwater Acoustics Research、Oceanographic and Atmospheric Processes

The seafloor topography (ST) is essential for oceanographic research and geophysical applications, and its modeling relies on satellite altimetry-derived gravity data. The ST inversion commonly applies linear regression method that account primarily for shipborne bathymetry errors, while often neglecting the potential uncertainties in altimetry-derived gravity data. To address this question, we draw on the idea of total least squares, which has been widely applied in numerical analysis. In this study, we propose an improved linear regression framework for ST inversion: we use high-precision multibeam shipborne bathymetry data (MSB1) as a constraint to iteratively determine the optimal weight ratio between bathymetry and gravity data, and subsequently construct initial weight matrices; then the Robust Weighted Total Least Squares (RWTLS) method is applied to estimate regression parameters, and the corresponding ST model is constructed. A case study is conducted in a region of the South China Sea (113°∼118°E, 13°∼19°N). Comparisons with the check data (MSB2) and existing models (topo_27.1, ETOPO2022, and SDUST2023BCO) indicate that the overall accuracy of the constructed ST model is approximately 150 m. And compared with traditional method, the application of RWTLS improves ST inversion accuracy, with power spectral density analysis further revealing a significant enhancement in topographic energy within the 12∼30 km wavelengths. Moreover, we find that the improvement in ST acc...