Answers are grounded in this paper and cite their sources.
Numerical simulations of the atmosphere are routinely carried out on various scales for purposes ranging from weather forecasts for local areas a few hours ahead to forecasts of climate change over periods of hundreds of years. Almost without exception, these forecasts are made with space/time-averaged versions of the governing Navier-Stokes equations and laws of thermodynamics, together with additional terms representing internal and boundary forcing. The calculations are a form of large eddy modelling, because the subgrid-scale processes have to be modelled. In the global atmospheric models used for long-term predictions, the primary method is implicit large eddy modelling, using discretization to perform the averaging, supplemented by specialized subgrid models, where there is organized small-scale activity, such as in the lower boundary layer and near active convection. Smaller scale models used for local or short-range forecasts can use a much smaller averaging scale. This allows some of the specialized subgrid models to be dropped in favour of direct simulations. In research mode, the same models can be run as a conventional large eddy simulation only a few orders of magnitude away from a direct simulation. These simulations can then be used in the development of the subgrid models for coarser resolution models.
No takes yet. Share an insight, caveat, or question.
Cullen et al. (2009) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: