Radio frequency interferences investigation using the airborne L-band full polarimetric radiometer CAROLS

In the present paper, different methods are proposed for the detection and mitigation of the undesirable effects of radio frequency interference (RFI) in microwave radiometry. The first of these makes use of kurtosis to detect the presence of non-Gaussian signals, whereas the second imposes a threshold on the standard deviation of brightness temperatures, in order to distinguish natural emission variations from RFI. Finally, the third approach is based on the use of a threshold applied to the third and fourth Stokes parameters. All of these methods have been applied and tested, with a CAROLS radiometer operating in the L-band, on data acquired during airborne campaigns made in spring 2009 over the South West of France. The performance of each, or of two combined approaches is analyzed with our database. We thus show that the kurtosis method is well adapted to pulsed RFI, whereas the method based on the second moment is well adapted to continuous-wave RFI.