Shallow Water Bathymetry Recovery of South Penghu Marine National Park, Taiwan

Abstract: The South Penghu Marine National Park is one of a few pristine environments with low pollution and minimal development in the surrounding ocean of Taiwan. However, the coasts in this region are susceptible to changes in appearance due to natural forces such as wind, waves, and currents. The traditional way of mapping and monitoring intertidal zones and nearshore shallow water topography have been high-cost tasks. Hence, this study aims to use space-born LiDAR on ICESat-2 launched by NASA for depth measurements, coupled with the optical satellite to map the coastal and seafloor topography of the South Penghu Marine National Park more promptly and accurately. To map the shallow water bathymetry in this area, we first filter and extract elevation profiles from ICESat-2 ATL03 photon data, which can detect depths reaching up to 40 meters in clear waters. Next, we adopt the green and blue bands from Sentinel-2 images and apply the modified log-ratio model for water depth estimation. In our preliminary tests, the ICESat-2 seafloor elevation achieves an accuracy of ~50 cm as compared with airborne LiDAR data. Also, the accuracy of water depth derived from Sentinel-2 is about 10% of the depth. It is summarized that an integration of ICESat-2 and Sentinel-2 can provide more timely observation of changes compared to traditional methods. It also overcomes the challenges in the areas inaccessible by vessels due to coral reef terrain and unsafe under keel clearance. Key Words: Satellite Derived Bathymetry, Log-Ratio Model, Taiwan South Penghu Marine National Park