An algorithm for estimating the base height of individual levels in multilayer clouds from passive satellite sensor data based on fuzzy logic methods is presented. The procedure for retrieving the cloud-base height is considered as a special case of solving the classification problem. The classes are narrow ranges of the target parameter; the classification features are cloud parameters retrieved from passive satellite sounding. Methods of fuzzy logic allow one to include one object in several classes at once but with different grades of membership, which enables estimating the base height for several simultaneously observed levels of clouds. The classifier is trained based on synchronous CALIOP (Cloud-Aerosol Lidar with Orthogonal Polarization) and MODIS (Moderate Resolution Imaging Spectroradiometer) data received over Western Siberia in summer in 2013–2018. The cloud base height is estimated only from passive satellite sensor data. Two fuzzy self-organizing methods (Fuzzy C-means and Gustafson–Kessel) are considered. The second approach has been found to be more effective; it provides a deviation of −0.5 km for the retrieved base heights of clouds with an optical thickness less than 10 as compared to reference ones at a standard deviation of 1.5 km for the overlying cloud layer and −0.1 km at a standard deviation of 2.1 km for the underlying layer.
А. В. Скороходов (Wed,) studied this question.