A circulation classification scheme applicable in GCM studies

The goal of the paper is to present and examine the method of classification of daily circulation patterns that allows (i) a fair comparison of groupings among different datasets (typically representing theobserved climate and that simulated by a general circulation model, GCM) and (ii) huge datasets, common in GCM studies based on daily values, to be classified. The circulation classification method is shownto be a useful tool in GCM validation and analysis of climate changeresponse, particularly in comparisons of (i) shapes of the mean typepatterns, (ii) the frequency and persistence of the types, (iii) theprobability of transitions from one type to another, and (iv) conditional surface temperature distributions. It is also shown that a simultaneous examination of multiple classifications is beneficial in eliminating subjectivity of any single classification and allowing a detailed inspection of differences between climates. The classification method is a modification of the T-mode principal component analysis (PCA). The T-mode refers to the input data matrix where gridpoint values are arranged in rows and daily patterns in columns. The classification procedure is applied to observed daily 500 hPa geopotential height patterns and those simulated by the control and $2\times {\rm CO}₂$ ECHAM3 GCM runs. [ABSTRACT FROM AUTHOR]