Abstract:

The Greenland Ice Sheet is of extreme importance in relation to global climate, and the freeze-thaw
characteristics of the ice sheet reflect Arctic climate change. Using Microwave Radiation Imager (MWRI)
data acquired by the Chinese FY-3 satellite, the Cross-Polarized Gradient Ratio (XPGR) method is introduced,
which increases the otherness of dry and wet snow. The Support Vector Machine (SVM) hyperplane,
which has better detection capability compared with the existing threshold method, is used to detect
freeze–thaw information on the Greenland Ice Sheet. Comparison of the results with Special Sensor Microwave
Imager (SSM/I) data (with a threshold of −0.025) showed the proposed method (i.e., XPGR combined
with SVM) is feasible for the detection of surface freeze–thaw information on the Greenland Ice Sheet.

Key words: FY-3, Greenland, XPGR algorithm, SVM algorithm