Evaluation of Four Global Bathymetry Models by Shipborne Depths Data

Abstract

More and more global digital bathymetric models (DBMs) have been developed using different kinds of data and have played a great role in navigation, national military defenses, tsunami predictions, marine resource developments, and so on. In order to select appropriate DBMs for usage in different ocean areas, it is necessary to verify their accuracies. In this study, the accuracy of four recent global DBMs, i.e., DTU18BAT, ETOPO1, GEBCO_2020Grid, and SRTM15 + V2.0, are evaluated by shipborne bathymetry data. By setting the shipborne depths data as true values, the error statistics and spatial distribution of the models are firstly analyzed both in spatial and frequency domains. The results show that SRTM15 + V2.0 has the highest accuracy. After removing gross errors, the error standard deviations of SRTM15 + V2.0 are smaller than 50 m, and 90% of the evaluation points have errors smaller than 100 m except in the Indian Ocean. The four models were fused together to obtain a new DBM with higher accuracy. This was done using a weighted combination algorithm based on iterative search to determine the combination parameters. The results showed that compared with the initial four models, the new model has an improved standard deviation of 2.816 m. Also, in a 10°×10° area in the Indian Ocean where initial mean error and error standard deviation were, respectively, −10.688 and 94.041 m, the average error has decreased by 8.527 m, and the error standard deviation has decreased by 13.528 m. The results of this study can provide a reference for the selection and optimization of the seabed topography model.