LAUSR

Abstract At kilometer-scale resolutions, the most energetic turbulent eddies associated with deep moist convection is partially resolved and partially subgrid scale (SGS). Therefore, SGS turbulent fluxes still play an important role in the transport of heat, momentum and other scalars. In contrast to its vertical counterpart, the horizontal SGS turbulent transport is often overlooked due to its relative insignificance at mesoscale grid spacings. This study investigates the effects of SGS horizontal turbulent mixing on a simulated real convective storm at kilometer-scale grid spacings. The SGS horizontal fluxes are parameterized by the widely-used 2D Smagorinsky model. Sensitivity studies are performed by varying the horizontal mixing length of the model, and are conducted on 3 km, 1 km and 333 m grids. Simulation results are evaluated against precipitation products and radar observations, and inter-compared with respect to both model resolution and SGS horizontal mixing. The comparison focuses on precipitation statistics as well as storm organization and morphology. The simulated precipitation is found to be most sensitive to SGS horizontal mixing at 3 km spacings, while the influence is much reduced at 333 m resolution. But for storm structure, SGS horizontal mixing remains important even at 333 m resolution. For practical numerical weather prediction purposes, this study suggests that in the kilometer-scale range, coarse resolution results could be made to resemble fine resolution results by optimizing the amount of SGS horizontal mixing. However, not all aspects of the coarse resolution simulations can be improved by tuning SGS horizontal mixing alone.