Bathymetry Inversion Using Full Tensor Gravity Gradients: A Case Study in the Bay of Bengal

Conventional methods for bathymetry inversion based on gravity field data usually adopt gravity anomaly and vertical gravity gradient. Indeed, a gravity gradient tensor (GGT) has six components. Besides the commonly used vertical gravity gradient, the other five components can also contribute to bathymetry inversion, which is investigated in this article. The formulas for the inversion are derived based on the Parker&#x2019;s formula. In order to provide GGT data for the inversion, vertical deflections and gravity anomalies of Scripps Institution of Oceanography, including north_SWOT_02 (1&#x2019;), east_SWOT_02 (1&#x2019;) and grav_SWOT_02 (1&#x2019;) are firstly used to derive GGT in the study region, i.e., Bay of Bengal. And then, six bathymetric grid models are derived using the six components of the gravity gradients, respectively. The results show that within the study area, the <inline-formula><tex-math notation="LaTeX">${{{\bm{T}}}_{{\bm{yz}}}}$</tex-math></inline-formula> component exhibits the best performance among the bathymetric inversion results derived from the single components of GGT. Combining the inversion results of the six GGT components yields a new bathymetric model&#x2014;one with higher accuracy than that derived from the single <inline-formula><tex-math notation="LaTeX">${{{\bm{T}}}_{{\bm{yz}}}}$</tex-math></inline-formula> component. Evaluations using ship-borne depth data indicate that, at the test points, the root mean square error of the combined model is 5.24 meters less than that of the <inline-formula><tex-math notation="LaTeX">${{{\bm{T}}}_{{\bm{yz}}}}$</tex-math></inline-formula>-derived model. When GEBCO_2024 depths are treated as the reference (true values), the RMS error of the combined bathymetric model is 14.31 meters less than that of the <inline-formula><tex-math notation="LaTeX">${{{\bm{T}}}_{{\bm{yz}}}}$</tex-math></inline-formula>-derived model.