Ocean wave extraction from RADARSAT synthetic aperture radar inter-look image cross-spectra

This study is concerned with the extraction of directional ocean wave spect ra from synthetic aperture radar (SAR) image spectra. The statistical estim ation problem underlying the wave-SAR inverse problem is examined in detail in order to properly quantify the wave information content of SAR, As a co ncrete focus, a data set is considered comprising six RADARSAT SAR images c o-located with a directional wave buoy off the east coast of Canada, These SAR data are transformed into inter-look image cross-spectra based on two l ooks at the same ocean scene separated by 0.4 s. The general problem of wav e extraction from SAR is cast in terms of a statistical estimation problem that includes the observed SAR spectra, the wave-SAR transform, and prior s pectral wave information, The central role of the weighting functions (inve rse of the error covariances) is demonstrated, as well as the consequence o f approximate (based on the quasilinear wave-SAR transform) versus exact li nearizations on the convergence properties of the algorithm. Error estimate s are derived and discussed. This statistical framework is applied to the e xtraction of spectral wave information from observed RADARSAT SAR image cro ss-spectra. A modified wave-SAR transform is used to account for case-speci fic geophysical and imaging effects. Analysis of the residual error of simu lated and observed SAR spectra motivates a canonical form for the SAR obser vation error covariance. Wave estimates are then extracted from the SAR spe ctra, including wavenumber dependent error estimates and explicit identific ation of spectral null spaces where the SAR contains no wave information, B and-limited SAR wave information is also combined with prior (buoy) spectra l wave estimates through parameterization of the wave spectral shape and us e of regularization.