The critical role of cross-polarized backscatter in understanding L-band PolSAR data in forested and urban environments

Abstract

Ambiguity exists in interpreting polarimetric synthetic aperture radar (PolSAR) L-band backscatter data from forested and urban environments. Two types of ambiguity are studied. The first is between flooded forests and urban buildings with small azimuth angles, and the other is between upland forests and urban buildings with large azimuth angles. To resolve the ambiguity, we developed an algorithm based on the PolSAR decomposition and backscatter model construction. First, the PolSAR backscatter data, including four co−/cross-polarized (XP) backscatter elements (i.e., the elements were not assumed to be zero) in a 3 × 3 covariance matrix, were decomposed into three components. Next, the decomposed XP backscatter determined each component's scattering mechanism of the single, double, or volume scattering. Then, with backscatter models, we resolved the first ambiguity by considering the volume and single scattering mechanisms and the second ambiguity by studying the volume and double scattering mechanisms. Thus, the volume scattering, primarily from the XP backscatter, was used in both ambiguity removals. The algorithm was evaluated using multiple L-band PolSAR data. The average resolving rate of the two types of ambiguity was 74.3%. Also, the relative powers of the single and double scattering mechanisms of radar targets decreased slightly (≤ 30.6%) in absolute value, while the volume scattering increased significantly (≥ 91.9%). The small change in the single or double scattering but the significant increase in the volume scattering quantitatively revealed the XP backscatter's criticality. The volume scattering component is vital for understanding the PolSAR backscatter data in forested and urban environments.