Unveiling the Accuracy of Empirical Satellite-Derived Bathymetry on the Brazilian Equatorial Margin: Insights from the Rio Grande do Norte Shelf

Abstract Empirical satellite-derived bathymetry (SDB) encompasses methods that can predict depths in optically shallow water bodies by correlating seabed reflectance with in situ measurements. Over the past few decades, advances in these techniques have enabled the accurate, robust, and accessible mapping of diverse shallow-water environments worldwide. In this study, we evaluated the potential of SDB techniques in the optically shallow waters of the Brazilian Equatorial Margin. In total, 24 SDB models were tested using Sentinel-2 imagery, two mathematical modeling approaches, and six categories of in situ calibration data. The best-performing models accurately retrieved depths up to 20 m, thus achieving R 2 values between 0.83 and 0.85, with root mean square errors (RMSE) ranging from 1.45 to 2.02 m. Models extending to depths of up to 54 m also showed good correlation (R 2 = 0.82), but with higher RMSE, reaching up to 4.43 m. Deeper areas exhibited increased noise, thereby hindering the identification of seafloor features. Most errors were concentrated in incised valleys, submerged reefs, and waters deeper than 20 m. Considering the spatial distribution of errors, we refined the SDB models; for models limited to a 15 m depth range, R 2 significantly improved to 0.92, with an RMSE of 1.07 m. These results demonstrate that the SDB techniques perform well at the Equatorial Margin and can serve as valuable tools for local marine spatial planning. We recommend the development and implementation of new SDB configurations and approaches tailored specifically to the region to achieve optimal performance.