Representing large-scale dynamics in local column models of the tropical and subtropical atmosphere

Reliable large-eddy simulations (LES) can now be carried out with resolutions of order 1m in domains that are comparable in size to a typical GCM grid box. This opens up the possibility to develop improved single column models for parameterization schemes in GCMs and to resolve longstanding questions about how clouds change with climate. Yet LES and other cloud-resolving simulations can only be expected to deliver climatically relevant answers to the extent that the large-scale processes driving such simulations are represented in a physically realistic manner. In particular, this means that for simulations of clouds in statistically steady states, it is desirable that the large-scale processes satisfy steady-state energy and moisture balances. The surface energy balance, for example, is crucial for how the near-surface relative humidity responds to climate changes; it is generally not satisfied in simulations that, as is commonly done, prescribe surface temperatures and/or surface fluxes. Moreover, it turns out that only representing the mean effect of large-scale dynamics in local column models does not suffice to obtain realistic column dynamics; fluctuations about the mean must also be represented. Here we present a framework for forcing local column models with large-scale processes that respect large-scale balances and include fluctuating components. Radiative energy fluxes are modeled explicitly, albeit in an idealized manner, surface and top-of-atmosphere energy balances are satisfied in climatically relevant ways, and large-scale eddy fluxes of sensible heat and moisture are represented parametrically. We argue that such a representation of large-scale processes is necessary to obtain realistic responses of clouds to climate changes and show preliminary results of large-eddy simulations of subtropical low clouds.