Discrimination of Ice and Water using PolSAR Parameter-based Clustering

Abstract

We have provided a simple unsupervised method to discriminate ice and water from quad polarimetric SAR imagery. The measure of asymmetry and degree of purity along with total power are crucial to discriminating ice and water. Future methods using other polarimetric modes should use this information to discriminate ice and water. In this respect, the work is also relevant for the Extreme Earth project (ICT-Horizon 2020) which will utilize the ESA Copernicus Big data.