The seasonal variation of cloud parameters over Central Europe: a fuzzy approach for the analysis of NOAA-APT data

A method for the retrieval and analysis of cloud parameters from NOAA satellite data is presented. The cloud classification method (FLOP) is a threshold algorithm with fuzzy class boundaries. The class definitions are not based on training data, but on the radiative properties of the cloud types. Three cloud layers are separated: low clouds, middle clouds and high clouds. The clouds in each layer are separated into layered and convective clouds by means of spatial variance analysis in the visible and thermal infrared band. The classification results are verified by comparison to ground observations. FLOP is applied to a time series of NOAA-11 data in APT format which has been recorded and processed daily at the University of Basel between October 1990 and December 1991. The major improvement of FLOP compared to traditional classification methods is the better handling of the uncertainty connected with the classification of clouds with variable radiative properties, such as Cumulus, semi-transparent clouds, and cloud edges. This is achieved by the fuzzification of the cloud classes. The degree of class membership for each pixel is derived in each of the four input bands separately (visible, thermal infrared, visible variance, thermal infrared variance). A fuzzy operator is then applied to obtain the overall degree of class membership for each cloud class in each NOAA scene. For each cloud class, one image per day is stored, the pixel values corresponding to the degree of class membership in percent. From these daily images, mean images for months and seasons are obtained and converted to sharp results. The climatological interpretation of the results shows that the cloud cover patterns over Central Europe are closely related to the general circulation. The temperature gradients between land and ocean surfaces as well as the dynamic and thermal pressure systems have a significant influence not only on cloud cover, but also on cloud fragmentation and top height. In summer and winter, when the meteorological conditions tend to be relatively stable over longer periods, certain weather types are characterized by typical patterns of cloud cover.