Abstract

Coupled land-atmosphere models exchange grid cell mean values of states and fluxes even as they separately simulate subgrid variability. The characteristics of atmospheric heterogeneity driven by land surface heterogeneity are examined in large eddy simulations. The degree of spatial variance in the atmosphere is highly correlated with the spatial variance of land surface fluxes, but the impacts are entirely at scales larger than 10-30 km. Between 1-10 km throughout most of the day there is no difference in power spectra over heterogeneous and homogeneous surfaces, as synoptic situations and internal dynamics associated with atmospheric turbulent transfer is insensitive to surface heterogeneity. To improve forecasts, the ability of land surface heterogeneity to organize mesoscale circulations affecting boundary layer evolution and convective development should be represented in non-cloud-resolving models, as modern convective parameterizations can account for such subgrid processes. This will require the exchange of subgrid information between land and atmosphere models.