An approach to classify polarimetric P-band SAR images for land use and land cover mapping in the

In this paper the potentiality of polarimetric P-band SAR data for Amazon tropical forest land cover mapping is assessed. The classifying approach is based on the Iterative Conditional Mode (ICM) algorithm, taking into account several specific distributions to SAR data. Distinct land cover classes are modeled considering different distributions. The results show that the P-band data is not capable to discriminate the nine classes initially used. However this capability improves significantly when classes having similar vegetation structure are grouped. The HV image is effective in differentiating primary and very old regeneration forest areas from other land cover classes, while W image increases the classification of bare soil and crop/pasture areas. The results show the importance of polarimetric information for the classification of several land use classes.