Shallow water bathymetry with multi-spectral satellite ocean color sensors: Leveraging temporal variation in image data Remote Sensing of Environment

Polar-orbiting ocean color satellites such as Landsat-8, Suomi National Polar-orbiting Partnership (SNPP), and Sentinel-3 offer valuable image data for the derivation of water bathymetry in optically shallow environments. Because of the multi-spectral limitation, however, it is challenging to derive bathymetry over global shallow waters without reliable mechanistic algorithms. In this contribution, we present and test a physics-based algorithm for improved retrieval of bathymetry with multi-spectral sensors. The algorithm leverages the temporal variation of water-column optical properties in two satellite measurements. By incorporating two remote sensing reflectance spectra in an optimization procedure, it enhances the spectral constraining condition for the optimization, thus leading to improved retrieval accuracy. This scheme is evaluated using synthetic multi-spectral data. It is shown that the new approach can provide accurate estimation of water depths over 0–30 m range with three types of benthic substrates (corals, seagrass, and sand) and for a wide range of water column optical properties. Based on the degree of improvement, Landsat-8 appears to be benefited the most, followed by SNPP, and then Sentinel-3. The application of the new approach is demonstrated with satellite images over shallow waters (0–30 m) dominated with coral reefs, seagrass, and sand, respectively. This proof-of-concept study confirms the promise of multi-spectral satellite sensors for accurate water depth retrieval by accounting for the temporal characteristics in multiple measurements, suggesting a path forward for the derivation of bathymetry from the existing satellites over global shallow waters.