Biases in wind-retrieval by empirical ERS-1/2 scatterometer models due to the presence of long ocean waves

The determination of the surface wind vector from the scatterometer of the European remote sen sing satellites (ERS-1/2) is based on three normalized radar cross sections (RCS), viewing a surface element from different azimuth angles. The scatterometer operates at 5.3 GHz with vertical polarisation (VV). Retrieval algorithms make use of empirical scatterometer models, which describe the RCS in terms of wind speed and direction. By making use of the composite surface scattering theory, the empirical RCS is decomposed into portions originating from small ripples (Bragg scattering) and their modulation by underlying long waves. The decomposition of the RCS allows the investigation of two questions. (1) Are RCS-models derived for the scatterometer applicable for the synthetic aperture radar (SAR)? On ERS-1/2 both measurements are performed by the same instrument. However, the illuminated areas differ considerably. This can lead to an underestimation of wind speed by 10%, when applying a scatterometer RCS-model to SAR data. (2) May the presence of swell distort the retrieval of wind direction? Empirical RCS-models account for the wind sea but not for swell, which is an occasional phenomenon with directions independent of the local wind. Under certain conditions, errors in wind direction due to the presence of swell of more than 20/spl deg/ are estimated.