Detection of Cirrus Clouds Using Infrared Radiometry

The detection of low-level clouds from ground-based infrared (IR) radiometry is usually based on the IR brightness temperature (IRBT) contrast between the warm clouds and the cold clear-sky background. This method works as long as the brightness temperature contrast subsists. It is not the case for cirrus clouds, which are usually optically thin and exhibit a low brightness temperature, comparable to that of clear sky in similar atmospheric conditions. In this paper, we propose a detection scheme to discriminate between clear sky and cirrus sky based on the fluctuations of the IRBT rather than on the absolute IRBT values. For this, we use the detrended fluctuation analysis (DFA) method on IRBT data when no liquid clouds were present. We compute the exponent α for nine IRBT time series covering clear- and cirrus-sky situations. We find that, for the fluctuation range of 1-5 min, α ≃ 0.1 for clear sky, α ≃ 0.5 for stratiform cirrus layers, and α ≃ 1 for broken cirrus layers. We suggest a threshold value of α ≃ 0.25 for the discrimination between clear sky and cirrus sky. We also examine the relationship between DFA and the classical spectral analysis and find that spectral analysis could be an alternative to the DFA method in detecting the presence of cirrus clouds.