Semi-automated bathymetry using Sentinel-2 for coastal monitoring in the Western Mediterranean

Abstract

Nearshore bathymetry is one of the main parameters used in coastal studies. However, regular and continuous bathymetries are scarce due to the difficulty of continuous data collection using traditional surveys. Remote sensing offers an attractive alternative to conventional collection methods for acquiring freely available, reliable, and frequent high-resolution data from space missions, such as the European Union's Copernicus programme Sentinel-2A/B satellites, which are increasingly demonstrating their potential to derive bathymetric information. Consequently, satellite-derived bathymetry (SDB) is being used more regularly because of its low cost and high efficiency. In this study, we apply a log-transformed band ratio SDB model in the Western Mediterranean Sea (Mallorca Island, Spain). This model has been widely used in the Caribbean and along the U.S. coast, among other sites, with outstanding results in heterogeneous environments. The SDB model consists of an automatic multi-scene approach applied over several images corrected by a robust atmospheric correction model (ACOLITE), a switching model, and a procedure to remove pixels from optically deep waters. We retrieve depths up to 14 m with a mean bias of 0.02 m and a median absolute error of 0.71 m compared with multibeam echo-sounder (MBES) data. The outcomes of this study confirm the effectiveness of the multi-scene approach to automatically correct the imagery to derive accurate depths and characterize erosional and accretion patterns annually. Furthermore, it highlights the benefit of the switching model to take advantage of the spectral sensitivity of different multi-band ratio approaches. This study will provide the scientific community with substantial knowledge and improved SDB techniques to consolidate the understanding of nearshore processes in such a relevant ecosystem as the Mediterranean Sea.