A comparison of radio-frequency interference detection based on AMSR-2 data

The influence of radio-frequency interference on microwave remote sensing has become increasingly serious. In order to better use the observed data, it is necessary to correctly identify and remove radio-frequency interference(RFI) contamination. To find out the applicability of various methods, the spectral difference method, multi-channel regression analysis method, double principal component analysis (DPCA) method and ratio method are used respectively to recognize RFI signals in the Advanced Microwave Scanning Radiometer 2 (AMSR-2) observations at 10.65 GHz and 7.3 GHz channels and the monthly results of above methods are compared. Research shows, the spectral difference method is influenced by the scattering of snow and misjudged the underlying surface of snow as RFI pollution in land during winter. Meanwhile, both the multi-channel regression analysis method and the DPCA method have stronger applicability and are equally effective in both land and ocean or winter and summer. When the RFI exists simultaneously in 7.3 GHz and 10.65 GHz channel observations, the ratio method will not recognize the RFI signal of the 10.65 GHz. On the ocean surface, the ratio method will misjudge the observation of the sea-land boundary and nearby field of view into RFI. In short, both the multi-channel regression analysis method and the DPCA method have a wide range of applications, but the spectral difference method and the ratio method have their own drawbacks.