Grid-dependent Convection in WRF-LES

Friday, 19 December 2014
Jason S Simon, University of California Berkeley, Berkeley, CA, United States, Bowen Zhou, Nanjing University, Nanjing, China and Fotini K Chow, UC Berkeley, Berkeley, CA, United States
Traditional numerical weather prediction (NWP) models parameterize boundary layer turbulence with planetary boundary layer (PBL) schemes, which assume a coarse [O(10 km)] grid resolution. Newer NWP models also have the ability to be large-eddy simulation (LES) models, which use a grid resolution that is sufficiently fine to resolve energy-containing eddies. The range in resolution-space between the maximum appropriate resolution for an LES closure and the minimum appropriate resolution for a PBL scheme is the turbulent gray zone, or the terra incognita. PBL schemes are designed for grid spacings that are much larger than the energy-containing eddies, typically considered to be of the same scale as the PBL depth [O(1 km)], to be contained in the sub-grid scale (SGS). LES closures are designed for a resolution that explicitly resolves the most energetic eddies, leaving the SGS turbulence approximately isotropic. The resolution limit for LES remains largely unexamined for atmospheric flows despite its dynamical significance and the increasing use of atmospheric LES models.

Here we examine the grid-dependence of the Weather Research and Forecasting model in LES mode (WRF-LES). We attempt to identify the symptoms of approaching the turbulent gray zone with WRF-LES under primarily convective conditions using the Wangara Day 33 case. Grid-dependence is evaluated by considering the development of the stability profile, the onset of resolved convection, higher-order statistical profiles, and turbulence spectra. Also considered are the effects of isotropic mixing length-scales, domain extent and spatially heterogeneous surface fluxes. We find that resolved convection is often too coarse to be physically realistic when the resolution falls within the gray zone.