A Probability Distribution Method for Detecting Radio-Frequency Interference in WindSat Observations

The detection of radio-frequency interference (RFI) continues to be an important problem for satellite-based microwave radiometers. This paper introduces two new probability-distribution-based techniques for computing the likelihood that a given brightness temperature observation contains an RFI signal. These methods extend the spectral difference method already being employed for the detection of RFI signals, and they allow for simultaneous observation-by-observation RFI detection of both land- and sea-based brightness temperature observations. This paper starts by laying out the theoretical groundwork for both techniques. It will then expand upon that groundwork to detail its practical application in analyzing WindSat brightness temperature observations. Finally, this paper compares the resulting probability indices from the detailed algorithms with spectral difference and principle component analysis indices computed from WindSat observations for various geographic regions. These comparisons will show effective RFI signal detection by these new techniques for RFI signal strengths as low as 4–5 K.