A Descalloping Postprocessor for ScanSAR Images of Ocean Scenes

Abstract

Due to its specific way of recording signals from multiple adjacent swaths in an alternating manner, a scanning synthetic aperture radar (SAR) (ScanSAR) cannot sample Doppler histories continuously like a SAR in stripmap mode. This can cause an effect known as azimuth scalloping, a wavelike modulation of the image intensity in near-azimuth direction. In theory, azimuth scalloping can be straightened out by using appropriate beam pattern corrections and multilooking techniques in the SAR processor. This works well over land, but lower signal-to-noise ratios and less accurate Doppler centroid estimates over water cause significant residual scalloping in many ScanSAR images of ocean scenes. The scalloping patterns hamper a correct interpretation of signatures of wind streaks, waves, and other phenomena. To overcome this problem once and for all, we have developed an algorithm that can eliminate scalloping patterns from existing ScanSAR images by postprocessing. Our algorithm detects the dominant scalloping pattern in an image automatically and eliminates most of it with very small side effects. We treat the scalloping pattern as a multiplicative effect, i.e., the amplitude spectrum of an affected image is assumed to be the convolution of the amplitude spectra of the unscalloped image and of the scalloping pattern. The proposed descalloping technique works partly in the spatial and partly in the spectral domain to approximate an exact deconvolution. We give a detailed technical description, show example results, and perform a quality analysis. We demonstrate the positive effects of the proposed descalloping treatment with a wind field retrieval example.