A sensitivity study is performed to examine the potential effect of spatial variations in sea surface temperature (SST) that typically are not resolved in general climate models (GCMs). The study uses a single-column atmospheric model, representing a grid box of a GCM, that overlies a surface domain divided into many subgrid cells. The model is driven by boundary conditions representative of the Gulf Stream off the mid-Atlantic coast of the United States, for the year 1987. A heterogeneous simulation, which includes subgrid spatial variability in SST, is contrasted with a homogeneous simulation, which assigns spatial mean SST to all cells.

In summer, the presence of both stable and unstable surface layers in the heterogeneous domain causes heterogeneous–homogeneous differences in monthly, spatially averaged surface latent-heat flux of up to 47%. In contrast, in winter, the surface layer is unstable everywhere and heterogeneous–homogeneous differences in latent heat flux are smaller. Spatially averaged, surface sensible heat flux shows less influence of SST heterogeneity because this flux during summer is small. Further simulation suggests that a GCM can capture the effect of spatially varying boundary layer stability by resolving it just at the surface. The SST heterogeneity is also capable of driving sea-breeze-type circulations. Scale analysis suggests that typical resolution of contemporary climate GCMs will generally be insufficient to resolve these circulations.