Abstract: Remote sensing of marine suspended particulate matter concentration is important for monitoring and managing marine environments. Based on the theory of Cloude-Pottier decomposition and backscattered radiation mechanism, the scattering angle (α) and entropy (H), decomposed from full polarimetric SAR (synthetic aperture radar) data of Radarsat-2, were used to build a model for retrieving suspended particulate matter concentration. The results show good positive relationships between αand the concentration of marine suspended particulate matter and between Hand the concentration, with the coefficients of 0.79 and 0.88, respectively. Four categories of regions are constructed in the H-αspace, corresponding to the concentrations of suspended particulate matter of 20 mg·L^-1 and above, 10-20 mg·L^-1, 5-10 mg·L^-1, and 5 mg·L^-1 and below, respectively. The validation shows that the accuracy of the model was up to 95.6% for classifying different concentrations of marine suspended particulate matter, which would be well applied in the field of monitoring marine suspended particulate matter concentration.

Key words: Radarsat-2, marine suspended particulate matter, H-α, spatial classification