Extraction of bathymetric features using multiple SAR images produced by Sentinel-1

Bathymetry, as a key parameter for evaluation of available resources of renewable marine energy, can be extracted from high resolution Synthetic Aperture Radar images acquired by Sentinel-1. In this paper, the high repetitiveness of these acquisitions is used to improve the accuracy of bathymetric maps extracted from SAR images. A method to extract wavelengths from the free surface elevations was developed, through experiments made using a phase-resolving wave propagation model (Boussinesq Ocean & Surf Zone model). It makes use of several different swell conditions propagating over the same bathymetry. The retrieved bathymetry is established by averaging the different bathymetric maps, obtained from different swell conditions. It allows obtaining a Root Mean Square Error of order 4.5 m for depths 10 m to 50 m. Twenty SAR images on a particular region have been processed using this methodology. The extraction of wavelengths and peak periods on each image led to a first estimation of the bathymetric data of this specific region compared to an actual bathymetric map with a 20 m spatial resolution. All these maps have been averaged to create a final bathymetric map with a 50 m spatial resolution and a vertical accuracy given by the RMSE over the entire domain of study of about 2.6 m. These preliminary results are encouraging. Future tracks are proposed for improving these results.