A method for retrieving the directional ocean wave spectra from synthetic aperture radar image

Abstract

Directional ocean wave spectra can describe the energy distribution of ocean wave and play an important role in oceanography. Synthetic aperture radar (SAR) can measure wave spectra based on a nonlinear mapping model between the image spectra (or cross spectra) and wave spectra. Due to the complexity of variable estimation and wave number cut off in azimuth direction in the mapping process, some scientists have been endeavoring to improve the retrieval of wave spectra. The paper proposes a method for the retrieval, which don’t depend on any external information except wind direction. It separates into two parts. In part 1, the spectra for real image or the cross spectra for complex image are calculated. And the modulation transfer function (MTF) is estimated including real aperture radar (RAR) MTF and velocity bunching. Then the wave spectra are retrieved by direct dividing based on the quasi-linear relation. The 180° ambiguity is removed from the imaginary part of cross spectra or the wind direction. The significant wave height (Hs), peak wave wavelength (L) and peak wave direction (D) are extracted from the retrieval. If the angle between the D and the azimuth direction is higher than 45°, the retrieval is finished. If the angle is lower than 45°, we go on retrieving. In part 2, the first guess spectra are needed to compensate the lost wave information, which can be parametrically constructed based on the retrieved wave spectra in part 1. Finally the wave spectra are iteratively retrieved from the first guess spectra based on the nonlinear relation. The Envisat ASAR images are used to validate the method. In case 1, the RMSE between this method and SARTool on D, L, Hs is 7.6°, 19.7 m, 0.18 m respectively. In case 2, the RMSE between this method and Jason-1 on Hs is 0.5 m.