Digital Bathymetric Model Fusion Using Vertical Gravity Gradient: A Bayesian Approach

Exploring ocean depth is crucial for understanding oceanic dynamic processes, deep-sea resource exploration, underwater navigation, and rescue operations. As a systematic outcome of ocean depth exploration, global digital bathymetric models provide convenient data support and tools for the in-depth study and application of seafloor topography. To our knowledge, five major digital bathymetric models are publicly available. However, significant differences can be found between these digit models at some places through a careful comparison, as they are constructed on methods and data originating from different sources. Confusion may exist when choosing which one of them to use in practical applications. A natural question that arises is how to obtain a more reliable fused model based on the existing models. To this end, we develop a novel model fusion method within the Bayesian framework. Its most outstanding feature is the ability to incorporate the marine gravity field as auxiliary data to guide the fusion of digital bathymetric models. It turns out that the fused model achieves an average accuracy improvement of 9.63%, 7.83%, 9.46%, 7.87%, and 15.49% over DTU18, ETOPO2022, GEBCO2024, SRTM15$+$v2.6, and Topo27.1, respectively, based on the empirical studies.