Identification of rice fields in a complex land-use region using RADARSAT-2 data

In the studied watershed, the land-use is complex, the cloud cover is frequent, and the medium resolution of optical images (LANDSAT, SPOT) make rice detection a challenge. To solve these problems, polarimetric radar data was used from RADARSAT-2 (C-Band). This paper addresses a classification scheme to extract rice fields based on the analysis of the temporal variation of backscattering coefficients in horizontally (HH) and cross-polarized (HV) polarization and the Support Vector Machine (SVM) classification algorithm. The results confirm that HH is better than HV polarization for rice detection because the rice backscattering coefficient in HV is not significantly different from the coefficients from other vegetation types. Preliminary results also show the capabilities of RADARSAT-2 C-band polarimetric images to identify rice fields.