Antarctic ice sheets play an important role in global sea-levels and climate changes. Snowmelt information with high spatial resolution on the surface of ice sheets is of great significance to the study of global climate change. Currently, the spatial resolution of snowmelt detection results based on low-frequency data from microwave radiometers is low, and accurate freeze-thaw changes cannot be obtained. The spatial resolution of Advanced Microwave Scanning Radiometer-2(AMSR-2)89GHz data is at least twice that of other microwave channels, but vulnerable to atmospheric water vapour. This paper proposes a method to detect snowmelt on the Antarctic ice sheet based on 89GHz data. First, according to the stable relationship of the polarization ratio(PR)of 89GHz and 36GHz data under clear, cloudless weather, the affected 89GHz data were selected. Then, the functional relationship between 36 GHz data and the unaffected 89 GHz data was obtained. Finally, the modified 89 GHz data were applied to snowmelt detection on the Antarctic ice sheet. The average detection accuracy of this method in six automatic weather stations was 91%, while the average detection accuracy of the cross-polarized gradient ratio algorithm(XPGR)was 74%. The experimental results show that the modified 89GHz data have a high accuracy in detecting snowmelt in Antarctic ice sheets.