Potentially predictable patterns of extratropical tropospheric circulation in an ensemble of climate simulations with the COLA AGCM

A variance decomposition approach to estimate the potentially predictable component of seasonal means has been further developed to separately diagnose the boundary‐forced component and the slowly varying internal dynamics component. This decomposition of the potentially predictable component has been applied to an ensemble of ten 50‐year simulations of the 500 hPa geopotential height field from the Center for Ocean‐Land‐Atmosphere Studies global atmospheric model forced by sea‐surface temperatures. The model performance is analysed for the winter season and compared with the National Centers for Environmental Prediction reanalysis. The majority of the hemispheric potentially predictable patterns, identified as the empirical orthogonal functions of the potentially predictable component of the seasonal mean fields, are captured by the model. These include all the patterns forced by the El Niño–Southern Oscillation in the Southern Hemisphere, and the dominant potentially predictable patterns in the Northern Hemisphere, i.e. the Pacific–North American pattern. However, the boreal winter Western Pacific Oscillation and Tropical Northern Hemisphere pattern, which are well simulated by some other general circulation models, are not well captured. Copyright © 2009 Royal Meteorological Society