Improving the modeling of bathymetry in the Persian Gulf and the Oman Sea using data assimilation of geodetic observation data.

Accurate modeling of bathymetry has a paramount importance in various marine applications, including navigation, resource exploration, and environmental studies. In this study, we present an innovative approach to enhance bathymetry modeling in the Persian Gulf and the Sea of Oman by employing data assimilation techniques with geodetic observation data. Our approach involves the integration of satellite altimetry missions, the XGM2019e gravity model, and ship-borne marine gravity data to extract the gravity anomaly. By utilizing variance component estimation (VCE), we integrate these three data sources to estimate the final gravity anomaly. Comparing ship-borne gravity anomaly control profiles with altimetry, XGM2019e and the final gravity anomaly reveals the superior accuracy of the final gravity anomaly compared to altimetry and the XGM2019e gravity model. Next, we utilize the final gravity anomaly in the Parker physical model to estimate the bathymetry. In order to modification the estimated bathymetry and achieve local calibration, we employ the 3D variational (3DVAR) data assimilation method, assimilating echo sounder observations to improve the bathymetry estimation. The assimilated bathymetry is then validated by comparing it with control points derived from echo sounder observations. The results demonstrate that data assimilation has the potential to enhance the accuracy of bathymetry estimation derived from the physical model. Following the data assimilation process in the physical model, our focus shifts to modeling the residual error between the echo sounder observations and the assimilated bathymetry. To tackle this, we propose the utilization of a Multi-Layer Perceptron (MLP) algorithm to model the residual between the assimilated model and the echo sounder observations. The results indicate that employing the MLP algorithm for residual modeling leads to improved accuracy at the control points. [ABSTRACT FROM AUTHOR]