An assessment of circulation type classifications for precipitation distribution in Norway

Abstract

This paper presents a climatological validation of atmospheric circulation type catalogues included in the COST 733 dataset (COST733CAT). The circulation type classifications ability to describe patterns of the Norwegian precipitation climate is analysed. For all 73 classifications included in the COST733CAT several statistical measures are calculated for each individual circulation type. These statistics include both frequencies of precipitation occurrence and statistical moments used to fit γ-distributions for each circulation type at precipitation stations. In this paper results from four stations in different precipitation regions and classifications with ∼9 types are emphasized. The circulation types employed represent the COST 733 domain 02 (57–72°N, 6°W–25°E). Precipitation occurrence frequency statistics and probability density functions are used to analyse if the types within the circulation type classifications can describe different precipitation conditions. Simple measures are developed in order to compare classifications. The similarity of probability density functions of type specific precipitation within classifications are analysed applying Kolmogorov–Smirnov tests. The type-specific distribution functions are also compared with the probability function of all precipitation events. The results reveal that some classifications, particularly those associated with the neural network methods does not discriminate precipitation distributions very well. On the other side many classifications performs well and get high scores. Evaluating all 73 catalogues the results indicate best results for classifications with 27–30 types.