COMPARISON OF ALGORITHMS AND INPUT VECTORS FOR SEA-ICE CLASSIFICATION WITH L-BAND POLSAR DATA

Two unsupervised methods, fuzzy c-means (FCM) and k-means, as well as three supervised methods, support vector machine (SVM), neural network (NN), and convolutional neural network (CNN), are applied to classify sea-ice type of first-year ice (FYI), multi-year ice (MYI) and open water, by using L-band polarimetric synthetic aperture radar (PolSAR) images in winter and advanced-melt phases, respectively. Different input vectors, pending on different scenarios, are also proposed to increase the accuracy rate. The efficacy of different algorithms in conjunction with different input vectors are analyzed and related to the underlying physical mechanisms.