A GCM Simulation of Global Climate Interannual Variability: 1950–1988

Abstract

Three long-term climate simulations have been performed with an atmospheric general circulation model using monthly global SSTs for the period 1950?1988. EOF analysis is used to study the ensemble-mean results for seasonal-mean fields as a means of evaluating the capability to simulate interannual variability. The analysis reveals a strong ENSO-related signal in the major fields of mean sea level pressure, rainfall, cloud cover, and zonal winds. The leading EOFs are compared, where possible, with observed ENSO-related patterns. The EOF for surface pressure closely resembles the Southern Oscillation pattern but, although significantly correlated with the Southern Oscillation index, cannot explain a large proportion of the observed variance. This is evident in the simulation of the 1982/83 warm event, which appears far less distinguished than observed. EOFs for both rainfall and low-level zonal winds also resemble, and are significantly correlated with, ENSO-related patterns deduced from observations. However, there are a number of important differences between the model rainfall pattern and known ENSO-related rainfall anomaly patterns that may limit any potential predictability given accurate SST forecasts. Similarly, differences between model zonal winds and observed zonal-wind stresses in the tropical Pacific would be expected to limit the performance of any coupled model comprising this particular low-resolution atmospheric GCM.