An Investigation into the Capability of Compact Polarized SAR to Classify Multi-Sea-Surface Characteristics

This paper assesses the capability of Compact Polarized (CP) Synthetic Aperture Radar (SAR) to classify multi-sea-surface characteristics (defined as typical sea surface suspended sediments, such as oil spills algal look-alikes and algal blooms). CP reconstruction is carried out on two C-band RADARSAT-2 quad-polarized (QP) SAR images using Nord's algorithm. The accuracies of the reconstructed polarized signatures of the multi-sea-surface characteristics of open water, oil spills and algal blooms derived from QP original quad-polarized data (quad-pol) and CP reconstructed approximate quad-polarized data (namely pseudo quad-pol) are then compared and analyzed. In addition, 10 types of polarization parameters are chosen in order to investigate the accuracy of the reconstruction for open water, oil spills and algal blooms. A comparison of the results shows that, of the chosen parameters, the reconstruction accuracies for sigma(0)(HH), sigma(0)(VV), , l(1), l(2) and are the highest whereas the accuracies obtained for sigma(0)(HV), l(3), H and A are unsatisfactory. The influence of reconstructed CP polarization parameters on the classification of multi-sea-surface characteristics is also assessed. Based on an assessment of this classification, it is concluded that the hybrid polarization (CTLR) mode should be adopted for the reconstruction of sigma(0)(HH), sigma(0)(VV), l(1), H and whereas the /4 mode should be used for the reconstruction of sigma(0)(HV), , l(2) and l(3).