Determining the movement of sea ice in the Arctic Ocean using high-resolution satellite images

Abstract

This study was carry out determining the velocity of sea ice in the Arctic ocean using high resolution satellite images for better understand the rapidly changing marine environment and sea ice characteristics of the Arctic Ocean due to the continuous decreasing of sea ice. The high-resolution satellite data for calculating the sea ice velocity used Sentinel-1 EW GRD SAR images collected by the European Space Agency (ESA). For distinguishing the open sea and sea ice from satellite images, regression analysis and inflection detection were applied to the backscattering value of satellite images. The velocity of sea ice was calculated only in the area where sea ice was present by masking the area where the water was exposed. The sea ice movement retrieved from SAR images was acquired by Maximum CrossCorrelation (MCC) and Robust Optical Flow (ROF) methods using feature tracking of the paired images. Since the accuracy and the time required for calculating the velocity of sea ice are influenced by the size of the window set in the image, the most efficient window size was determined by the various window sizes. The sea ice movement based on the SAR image was verified with the drift buoy data provided by the International Arctic Buoy Program (IABP) and the medium-resolution sea ice motion data from the National Sea Ice Data Center (NSIDC). The difference of speed between drift buoys and sea ice motion by MCC and ROF method are 0.0318 cm/s and 0.0971 cm/s averagely. When compared with the NSIDC's sea ice motion, difference of both methods are 0.8235 cm/s and 0.9505 cm/s, respectively. In addition, some NSIDC sea ice motion showed different moving pattern compared to drift buoys and high-resolution sea ice motion in this study. It is believed to cause the low-spatial resolution. The highresolution sea ice motion was more similar moving pattern of the drift buoy than that of the NSIDC.