Sea Ice Concerntration Extraction with Compact-Polarization SAR Images

Abstract

Synthetic Aperture Radar (SAR) has become a popular sensor and valuable tool for sea ice extraction because of its observation ability in all-day, all-weather and high-resolution features. With the development of SAR, there are three different polarization modes called by single-pol, quad-pol and compact polarization, which have been used for sea ice application. Only the intensity information is obtained by single-pol observation. As for the quad-pol, which can provide more information of subject and perform well in the classification of sea ice. However, the development of quad-pol is limited by its high cost and smaller coverage. Compact polarization was proposed, because it can obtain the similar information as quad-pol SAR with the wide coverage, which is very suitable for sea ice monitoring. In this study, sea ice information are extracted with k-means algorithm from compact polarization data. And the results are compared with those from quad-pol GaoFen-3(GF-3) data. The classification results from the two SAR modes are close. It indicates that compact polarization has a good potential in the application of sea ice monitoring.