Computing and Informatics

This paper describes our contribution to the research of parametrized models and methods for detection and prediction of significant meteorological phenomena, especially fog and low cloud cover. The project covered methods for integration of distributed meteorological data necessary for running the prediction models, training models and then mining the data in order to be able to efficiently and quickly predict even sparsely occurring phenomena. The detection and prediction methods are based on knowledge discovery -- data mining of meteorological data using neural networks and decision trees. The mined data were mainly METAR aerodrome messages, meteorological data from specialized stations and cloud data from special airport sensors -- laser ceilometers.