Water Depth Inversion From a Single SPOT-5 Dataset

Knowing bathymetry at intermediate depth, over large areas, and at a reasonable cost is a key issue. Spaceborne remote sensing techniques must play an essential role in retrieving such bathymetry. In this paper, a method is proposed that aims to address this issue without any in situ measurements by exploiting the characteristics of the SPOT-5 satellite dataset. The proposed method is designed to provide bathymetry from two optical SPOT-5 satellite images separated by a time lag DT of 2.04 s. It relies on the estimation of several clouds of wave celerity and wavelength pairs using wavelet and cross-correlation techniques and on the linear wave dispersion relation. This method has been applied to two SPOT-5 images on a test site characterized by complex bathymetry (Saint-Pierre, La Réunion Island). A comparison of the retrieved bathymetry with in situ bathymetric measurements reveals good morphological agreement. The mean relative error is less than 30% in the 3-80-m water depth range. The methodological choices made during method development are discussed based on additional computations, and guidelines for using the proposed method on other images at other sites are provided.