Identification and Correction of Radio Frequency Interference of Fengyun-3 Microwave Radiation Imager Using a Machine-Learning Method

Abstract

Radio frequency signals can interfere with the radiation emanated from the earth atmospheres and affect the quality of the data received from spaceborne microwave instruments. For microwave radiation imager (MWRI) carried on China’s Fengyun-3 series satellites, the data contaminated by radio frequency interference (RFI) are usually identified and labeled as poor quality. In this study, using the high correlation between the observed brightness temperatures (TB) of MWRI channels, an RFI identification and correction method is developed through machine learning techniques. Compared with traditional methods, the new method can simultaneously identify and correct RFI affected data. Since it is trained with global MWRI data, the method works well for both land and oceans. Our analysis show that the MWRI data affected by RFI can be corrected to the quality level close to RFI-free regions.