GCM-Forced Large Eddy Simulations and Single Column Models: A Numerical Laboratory For Reducing Uncertainties in Climate Predictions

Much of the spread in predictions of climate sensitivity arises from uncertainties in the parameterization of boundary layer clouds and turbulence in climate models. Here we discuss a framework in which large eddy simulations (LES) and an eddy diffusivity mass flux (EDMF) single column model (SCM) are driven by dynamically and thermodynamically consistent large-scale forcing from a grid column of a general circulation model (GCM). This framework provides a methodology for systematically optimizing GCM parameterizations.

The forcing framework partitions GCM grid-column water and energy budgets into terms explicitly resolved by the GCM, like horizontal advection and large-scale subsidence, and those determined by GCM parameterizations, like boundary layer turbulence, convection, and radiative transfer. The effects of GCM resolved processes are applied directly to the LES and SCM while the parameterized processes are not. Instead, the LES and SCM are allowed to individually represent the unresolved GCM processes subject to dynamically consistent large-scale forcing. The LES and SCM are both coupled to a slab-ocean model and run toward statistically quasi-steady states, thus satisfying physically realizable surface energy balances.

The framework provides a unique experimental tool for designing and improving GCM parameterizations. For example, if we assume the LES provides a high fidelity representation of the unresolved GCM processes, then an ideal SCM should provide representations of unresolved GCM processes that closely match those produced by the LES. Should the LES and SCM solutions not agree, the LES provides data useful for improving the SCM.

Here we discuss the use of this framework to explore the behavior of the SCM relative to the LES in simulations of a transect across a Walker circulation subject to climate perturbations.