On the use of Sentinel-2 satellites and lidar surveys for the change detection of shallow bathymetry: The case study of North Carolina inlets

Abstract

Mapping bathymetric change is a core task for a wide range of navigation, research, monitoring, and design applications. Satellite-derived bathymetry (SDB) can support this activity, particularly when using data from a platform, like the Sentinel-2A/B twin mission of the Copernicus programme, which provides routine and repetitive image acquisition at 10 m spatial resolution. As a result, SDB can expand on the temporal and spatial scope of more conventional mapping methods, such as lidar (or multi-beam). In this study, we use SDB, in comparison with high-resolution airborne lidar bathymetry (ALB), to quantify bathymetric changes at two inlets in North Carolina following the impacts of the devastating Hurricane Florence in September 2018. A recently developed multi-temporal SDB model that overcomes problems associated with turbidity and noise effects (whitecaps, ships, cloud shadows, etc.) is used with Sentinel-2 imagery. We identify bathymetric changes in shallow areas with navigation channels in two of the most dynamic inlets in the Outer Banks, Oregon and Hatteras. Multiple lidar surveys are used to validate the SDB method and for an assessment of accuracy and vertical uncertainty. The multi-temporal SDB products and ALB both show similar results in the erosion/accretion patterns. Comparing the change determined from the two methods, gives a median absolute error of ~0.5 m of SDB compared with ALB, with bias of ±0.2 m for depths ≤7 m; errors that are equivalent to those associated with the SDB estimated absolute depths. The composited SDB yields greater spatial coverage than lidar because it can retrieve data in areas where the lidar mission may have been constrained by turbidity or surf, and also because of the 300-km swath width. The Sentinel-2 constellation provides five-day revisit at the equator (more frequent at higher latitudes), allowing rapid construction of a bathymetric map as well as developing an archive for retrospective analysis. By implementing the multi-temporal turbidity correction, SDB based on Sentinel-2 may substantially enhance existing survey methods for change detection and support operational and recursive coastal monitoring on local to regional scales.