A cloud detection algorithm for high-spectral-resolution infrared sounders

A method for detecting cloud contamination in radiances measured by high-spectral-resolution infrared sounders is presented. It seeks to identify clear channels within a measured spectrum, rather than the locations of completely clear spectra. Applied to simulated cloudy spectra, the scheme is able to detect clear channels with residual cloud contamination better than (i.e. less than) 0.2 K in many channels, and is thus considered sufficiently stringent for numerical weather-prediction applications. The scheme has been applied to spectra measured by the Advanced InfraRed Sounder and, whilst a quantitative validation is more difficult with real data (without true clear radiances), it is found to perform well compared with coincident imagery.