Computation of ocean wind vectors using RADARSAT ScanSAR and ERS SAR images

The high resolution together with the large coverage of satellite based synthetic aperture radars (SAR) offer an unique opportunity to derive mesoscale wind fields over the ocean surface. In this paper different methods for deriving wind fields from SAR data are introduced and compared to each other. Wind directions are extracted from wind induced streaks visible in most SAR images. Wind speeds are computed with C-band models for both VV- and HH-polarization from the calibrated normalized radar cross section (NRCS). Another method is based on sea surface movements due to the local wind which allows the width of the peak of the autocorrelation function (ACF) to be a measure of wind speed. The algorithms are applied to ScanSAR wide swath images acquired by RADARSAT and to SAR images from the European remote sensing satellites (ERS-1 and ERS-2). Comparisons to in situ measurements and results of an atmospheric model emphasize the benefits of each method. Sources of errors in wind retrieval are discussed for different algorithms. Errors in wind speed due to speckle, uncertainty in wind direction, and accuracy of NRCS are estimated for different wind speeds, wind directions and polarization. ERS SAR derived ocean surface winds are applied to investigate the spatial variation of the wind on different scales resulting in a decrease of spectral density according to the power law k^5/3 predicted by two dimensional isotropic turbulence