Nearshore Bathymetry From Fusion of Sentinel-2 and ICESat-2 Observations

TLDR

Nearshore bathymetry is essential for coastal process understanding, yet passive remote sensing methods typically require in situ depth measurements that are often unavailable, though ICESat‑2's ATLAS green laser can provide depth data up to 38 m in clear waters. The study investigates fusing ICESat‑2 ATLAS range data with Sentinel‑2 multispectral imagery to estimate nearshore bathymetry using two common inversion algorithms. The authors applied two standard bathymetric inversion algorithms to the fused ICESat‑2 and Sentinel‑2 data and evaluated the results against airborne CZMIL LiDAR collected in October–November 2018. In Destin, FL, the fused approach produced bathymetry with an RMSE of 0.35 m relative to CZMIL data in clear waters, demonstrating that ICESat‑2 depth observations combined with Sentinel‑2 imagery can generate seamless nearshore bathymetry.

Abstract

Nearshore estimates of bathymetry are crucial for understanding coastal processes. However, current passive remote sensing methods for estimating bathymetry require in situ depth measurements to train inversion models, which can be difficult or impossible to obtain in many areas. To address this issue, we investigated the fusion of range measurements from the advanced topographic laser altimeter system (ATLAS) aboard the NASA ICESat-2 satellite, and multispectral satellite imagery from European Space Agency (ESA) Sentinel-2 using two common bathymetric inversion algorithms. The active ranging capability of the ATLAS green (532-nm) laser has been shown to generate returns of up to 38-m depth in optically clear waters, providing depth measurements to constrain passive bathymetric inversion results. Data acquired in November 2018 over the nearshore in Destin, FL, USA, offer a proof of concept for this approach. The results of the bathymetric inversion were quantitatively assessed by comparison with airborne bathymetric LiDAR collected using the U.S. Army Corps Coastal Zone Mapping and Imaging LiDAR (CZMIL) system in October-November 2018. Overall, the results of the bathymetric inversion compared with the CZMIL data have a root mean square error (RMSE) of 0.35 m in waters with similar turbidity and bottom reflectivity, and demonstrate that a combination of ICESat-2 depth observations with Sentinel-2 multispectral imagery can estimate seamless nearshore bathymetry for optically clear coastal waters.