Spectral analysis of non-stationary ocean SAR images using the Wigner-Ville transform

Physical events, such as swell or internal waves, have a strong signature on ocean SAR (synthetic aperture radar) images. The spatial periodicity of these phenomena justifies the use of spectral analysis tools for their detection, localization and pertinent parameter estimation (wavelength, direction of propagation, energy, ...). However, these spectral phenomena are highly non-stationary, especially near the coast where the bathymetric conditions greatly influences the propagation. As the study of these variations are important near the coast, because of human activity, spectral analysis tools robust in non-stationary situations have to be considered. For that purpose, investigations were carried out to evaluate the efficiency of the Wigner-Ville distribution. This representation exhibits very good properties and has been quite extensively studied, even if the numerical implementation still presents some difficulties.