A Hybrid Physical-Statistical Algorithm for SAR Wave Spectra Quality Assessment

Abstract

A new approach for assessing the quality of Synthetic Aperture Radar (SAR) wave spectra is presented here. The algorithm addresses two specific issues, related to the 180° directional ambiguity in the propagation direction inherent to SAR measurements, and the removal of noise. In spite of several and progressive advances in the mapping and retrieval of SAR wave spectra, these issues are persistent in the existing official databases and hinder the use of these data for practical uses. This new algorithm is based on a recently developed database of long-term global wave spectral characteristics, which allows estimating the occurrence probability of every individual wave system found in the observed spectra, and that of its ambiguous pair. In addition, assuring the spatial consistency of the wave systems along track, helps obtaining more robust results. In this sense, wave spectra within a track are processed in several steps, so that wave systems with more likelihood of being correct are processed first, and consequently used to evaluate the more challenging components. Accordingly, a specific quality label is assigned to every wave system depending on the certainty achieved. An Envisat track over a challenging area with multiple crossing swells is used to illustrate the performance of the algorithm.