Large-eddy simulations of a neutral boundary layer: Sensitivity to grid spacing and implications for modeling the lower boundary condition

It has long been known that calculating the bottom boundary condition for large-eddy simulation (LES) using horizontal winds at the first interior grid point is problematic. The majority of turbulent kinetic energy near the surface is distributed on scales smaller than the grid spacing, violating the homogeneous and isotropic assumption of subgrid-scale turbulent motions. The unsatisfying performance of subgrid-scale model leads to inaccurate prediction of horizontal winds, not recommended as input for bottom boundary condition calculation. Therefore, the objective of this research is to investigate how far above the surface horizontal wind predictions are sufficiently reliable. The methodology is conducting LES of a neutral turbulent boundary layer, for which a theoretical prediction of a logarithmic layer is well established. We use the Cloud Model 1 (CM1, frequently used in thunderstorm studies), and conduct simulations with various vertical and horizontal resolution.The preliminary results show that horizontal winds are in good agreement with the logarithmic theoretical prediction above five vertical grid spacing. Recommendation about LES bottom boundary condition modeling will be provided based on comprehensive evaluation of turbulence statistics. Future research activities include revising the CM1 bottom boundary condition modeling, including situations in which a log law cannot be assumed to exist in the first place. Ultimately we will be investigating how improved lower boundary condition modeling improves convective storm simulations