Improving Turbulence Parameterization for an Idealized Supercell at Kilometer-Scale Resolution

The spatial resolution of numerical weather prediction models is approaching kilometer-scale, which is close to the integral scale of deep cloud systems. A significant portion of heat, moisture and other scalar fluxes in deep convective clouds are carried by motions on the scale of several kilometers, hence only partially resolvable on kilometer-scale grids. Therefore subgrid-scale (SGS) turbulence parameterization is essential in modeling deep clouds. Conventional eddy-viscosity-based turbulence schemes are unable to represent non-local mixing in the convective clouds, thus incapable of reproducing locally counter-gradient fluxes. To improve the SGS representation, a scale similarity model (i.e., the modified Clark model) introduced by Moeng et al. 2010, is implemented in a community atmospheric model. The new model parameterizes SGS vertical fluxes in terms of horizontal gradients of resolved variables, and is able to reproduce counter-gradient fluxes. The scale similarity model can also be paired with an eddy viscosity model to form a mixed model. The new schemes and convectional eddy-viscosity schemes are used in simulations of an idealized supercell at horizontal resolutions of 0.5, 1, 2 and 4 km. Using a 50 m high-resolution large-eddy simulation as a benchmark, effects and improvements of the new scheme is investigated focusing on the representation of fluxes, precipitation, and overall structure of the supercell.