A VGGNet-Based Method for Refined Bathymetry from Satellite Altimetry to Reduce Errors

Only approximately 20% of the global seafloor topography has been finely modeled. The rest either lacks data or its are not accurate enough to meet practical requirements. On one hand, satellite altimeter advantages large-scale and real-time observation. Therefore, it is widely used measure bathymetry, core topography. However, there often room improve precision. Multibeam sonar bathymetry more precise but generally limited a smaller coverage, so in complementary relationship with satellite-derived bathymetry. To combine altimetry-derived multibeam sonar-derived we apply deep learning perform sonar-based correction for altimetry data. Specifically, modify pretrained VGGNet neural network model train on three sets from West Pacific, Southern Ocean, East Pacific. Experiments show that correlation before after can reach high level, performance R2 being as 0.81, normalized root-mean-square deviation (NRMSE) improved by over 19% compared previous research. We then explore between water depth conclude varies at different depths. Thus, terrain specificity factor affects precision correction. Finally, difference evaluation find our method than 17% results better Hence, provide novel modeling