Reproducing a Subjective Classi cation Scheme for Atmospheric Circulation Patterns over the United Kingdom using a Neural Network

Atmospheric circulation patterns are currently classi ed manually according to subjective schemes, such as the Lamb catalogue of circulation patterns centred on the United Kingdom. However, the sheer volume of data produced by General Circulation Models, used to investigate the e ects of climatic change, makes this approach impractical for classifying predictions of the future climate. Furthermore, classi cation extending over long periods of time may require numerous authors, possibly introducing unwelcome discontinuities in the classi cation. This paper describes a neural classi er designed to reproduce the Lamb catalogue. Initial results indicate the neural classi er is able to out-perform the currently used rule-based system by a modest, but signi cant amount.