Evaluating the potential of unsupervised classifications for icefoot cartograpy using RADARSAT-2 high resolution imagery

Abstract

With the advent of a new generation of high resolution polarimetric SAR satellites, the extraction of ice structures as narrow as the strip of sea ice that forms the icefoot becomes potentially feasible. One RADARSAT-2 scene was acquired on February 15, 2011 and three processing chains were tested to accomplish this task. It was found that the Lee and Pottier [2, 4] method and more specifically the information imbedded in the class distribution offers a great potential. While some environments such as sea water or a city may be sorted out using a single class, the more diverse range of mechanisms involved into the backscattering processes in icefoot and peat bog areas make the class distribution a better indicator to automatically extract the icefoot.