SST-Forced and Internal Variability of the Atmosphere in an Ensemble GCM Simulation

Abstract

A 20-member ensemble simulation has been conducted with an atmospheric general circulation model to in- vestigate the time-space characteristics of the leading modes of internal and SST-forced variability in the winter Northern Hemisphere. Each of the 20 integrations is forced by the identical global sea surface temperature (SST) and sea ice history observed for 1959-1998, and diers only in the initial conditions. A variance analysis is per- formed to quantify the relative importance of the SST-forced and internal variability. In the extratropics, the SST-forced signals are much higher in the North Pacific than in the North Atlantic where month-to-month and interannual variability is dominated by internal chaos. The leading empirical orthogonal function mode resembles the Arctic Oscillation (AO) for both the internal and ensemble-mean variability but with significant dierences in correlation between the Pacific and Atlantic cen- ters of action. The Pacific-Atlantic correlation is significantly higher for the ensemble mean than for all but one member integration. The correlation between the Arctic and midlatitude North Atlantic, by contrast, is higher in individual member runs than in the ensemble-mean. These results suggest that SST-forced variability is organized into a hemispheric AO pattern while internal variability is more confined in the North Atlantic sector. Seasonal air-sea interactions in the North Pacific and Atlantic are also discussed.