Classification of meteorological patterns

Abstract

A classification system which organizes in a restricted number typical meteorological situations, is useful in order to find a relationship between meteorological patterns present in the forecasts and derive local weather parameters. The conventional classifiers based on climatology commonly adopted by weather centres suffer from subjective construction and application which weak the analogy method. This paper deals with the construction of an artificial neural network aimed to classify meteorological patterns. Solutions to meteorological requirements are presented together with operational applications.