A parametric scheme for ocean wave spectra retrieval from complex SAR data using prior information

A parametric inversion scheme for the retrieval of two dimensional ocean wave spectra from look cross spectra (LCS) acquired by spaceborne synthetic aperture radar (SAR) is presented. The scheme takes information about the spectral shape of different wave systems from a prior wave spectrum, while estimates for wavelength, waveheight and wave propagation direction are extracted from SAR cross spectra. The Partition Rescaling and Shift Algorithm (PARSA) is based on a partioning of a prior wave spectrum. e.g. taken from ocean wave models. For each ocean wave system a stochastic model is set up, which defines the probabilitity that the propagation direction, the wavelength or the energy of the different wave systems deviate from the prior knowledge. The prescribed probabilities thereby quantify the confidence into the prior wave spectrum. Based on the probability models for the prior wave spectrum and the measured cross spectrum an optimal ocean wave spectrum is estimated using a maximum a posteriori approach. To solve the corresponding minimization problem the prior model is approximated with a multivariate Gaussian model. The optimization problem is solved with a Gauss Newton method. The scheme is tested using both simulated cross spectra and reprocessed wave mode data acquired by the ERS2 SAR. The reprocessed data are similar to the products, which will be available from the ENVISAT satellite launched in March 2002.