Bathysent - A Method to Retrieve Coastal Bathymetry from Sentinel-2

This paper presents a method for deriving shallow to intermediate (1m to 50m) coastal bathymetry from space-borne multispectral data taking advantage of the short time-lag between sensors’ bands. The idea is to quantify local waves’ characteristics (wavelengths and celerities) that are related to the water depths using optical data: local spectral analysis can provide the significant wavelengths and inter-band offset-tracking and the corresponding celerities (knowing the inter-band time-lag). Such an approach was firstly described in [1]. However, for an application to extended areas and using large data sets (as possible with the Sentinel-2 archive), a faster technique is required: the ability of processing large areas and data acquired at different dates is required for actual operational uses. The approach we propose here is based on Fast Fourier Transform analysis in order to simultaneously extract the wavelengths and celerities.