Use of Sentinel-2 optical imagery for the reconstruction of coastal bathymetry: a case study in Mozambique

Bathymetric surveys are essential for ensuring safe navigation and supporting environmental studies in coastal regions. Collecting bathymetry data in the field using acoustic sonars allows us to accurately reproduce underwater topography but is limited by high operational costs and challenges that may arise when operating in shallow waters or unsafe areas due to political instability. In recent decades, remote sensing techniques have become more prominent, particularly those leveraging open data sources. The study presents a feasible approach for regular monitoring of bathymetry in shallow areas (0–5 m). The well-established ratio-transform algorithm by Stumpf et al. (2003) and Sentinel-2 open data of the Palma Bay area, in Northern Mozambique, are used. The analysis of images characterised by different environmental and atmospheric conditions, and the presence of non-uniform seabed and sometimes turbid water makes it possible to highlight critical aspects affecting the accuracy of the method used, providing a realistic view of the method’s capabilities. Attention to these critical aspects, through the application of appropriate pre-processing techniques and the classification of the seabed, allows good results (RMSE values of 0.2 m and 0.5 m in sandy and mudflat areas respectively), aligned with the reviewed literature.