An Appraisal of Atmospheric Correction and Inversion Algorithms for Mapping High-Resolution Bathymetry Over Coral Reef Waters

Bathymetric inversions constitute a key preliminary step when characterizing coral topography. The Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) is poised to provide true values for satellite-derived bathymetry (SDB) in the absence of measured data. Nevertheless, the influences of the different available algorithms and atmospheric correction (AC) processes applied to construct SDB over coral reef waters are still poorly understood. In this article, three empirical-based bathymetric algorithms [the linear ratio model (LRM), polynomial ratio model (PRM), and exponential ratio model (ERM)] and three ACs [AC for the operational land imager (OLI) lite (ACOLITE), sea-viewing wide field-of-view sensor (SeaWiFS) data analysis system (SeaDAS), and sentinel 2 (atmospheric) correction (Sen2Cor)] are evaluated with Sentinel-2A/B multispectral imagery to achieve fine-spatial-resolution bathymetric information. The bathymetry estimates produced by fusing the Sentinel-2A/B and ICESat-2 datasets using the three models exhibit sufficient accuracies, with low percent errors and mean absolute errors (MAEs) between 0.44 and 0.74 m and root-mean-square errors (RMSEs) between 0.61 and 0.96 m. In all cases, the PRM outperforms the ERM and LRM. The AC method comparisons show that the performance of ACOLITE is better than those of Sen2Cor and SeaDAS. The obtained depth estimates are less sensitive to the impacts of the AC methods in relatively shallow waters than in relatively deep waters, and ACOLITE performs more stably. Overall, the PRM coupled with the ACOLITE method performs best. With depths extracted from ICESat-2, high-spatial-resolution SDB data capable of supporting the long-range goal of coral topography analyses are attainable even without in situ datasets.