Abstract
The characteristics of atmospheric low-frequency variability and midlatitude SST variability as simulated by the National Center for Atmospheric Research’s Climate System Model are analyzed in the vicinity of the North Pacific and North Atlantic basins. The simulated spatial patterns of variability correspond quite well to those seen in observational datasets, although there are some differences in the amplitudes of variability. Companion uncoupled integrations using the atmospheric component of the coupled model are also analyzed to identify the mechanisms of midlatitude SST variability on interannual timescales. These integrations are subject to a hierarchy of SST boundary conditions, ranging from the climatological annual cycle to global monthly mean observed SST. Even uncoupled atmospheric model integrations forced by climatological SST boundary conditions are capable of simulating the spatial patterns of atmospheric variability fairly well, although coupling to an interactive ocean does produce some improvements in the spatial patterns. However, the presence of realistic SST variability, especially in the Tropics, is necessary to obtain the right variance amplitudes for the different modes of variability. It appears that coupling to an interactive ocean essentially reorders, rather than reshapes, the dominant modes of atmospheric low-frequency variability. The results indicate that the dominant modes of SST variability in each ocean basin are forced by the respective dominant modes of atmospheric low-frequency variability in the vicinity of the ocean basin. The relationship between atmospheric variability and the surface heat flux is also analyzed. Evidence is found for a local thermal feedback in the coupled integration, associated with the finite heat capacity of the ocean, that acts to damp surface heat flux variability. It is also shown that the relationship between midlatitude SST anomalies and the surface heat flux in the Atmospheric Model Intercomparison Project–type of atmospheric model integrations is quite different from that in the coupled model integration.