Assessing the performance of the WRF model's large-eddy simulation capability in nested-domain simulations

Many contemporary atmospheric simulation models employ a grid nesting capability that permits large-eddy simulations (LES) to be conducted over subsets of larger bounding simulations, with the bounding-domain solutions providing lateral boundary conditions for the nested domains within. This approach can provide lateral boundary conditions that include mesoscale features and the effects of upstream topography and land cover for the LES. Such an approach is particularly applicable to wind power forecasting, especially over complex terrain, for which mesoscale flow conditions, local terrain effects and turbulence information can all be important.

While grid nesting has been successfully applied at GCM to mesoscale resolutions, such nesting behavior at higher resolutions, including those appropriate for LES, is less well understood. We investigate such grid nesting capabilities for conducting nested LES using the Weather Research and Forecasting (WRF) model. Comparisons among the velocity and stress profiles inside the nested domain relative to single-domain simulations indicate that errors contributed from the bounding domains persist within the nested-domain solution. We also examine the spatial scales required for flow structures to equilibrate to the finer mesh as flow enters a nest, and how equilibration depends on several parameters, including mesh resolution, surface roughness, terrain complexity, atmospheric stability, and the type of turbulence subfilter-scale stress model used. We find that the equilibration process of the flow within a nested domain is strongly governed by the discontinuity in the grid-dependent length scale at the nest interface, the application of a log-law surface boundary condition, and the advection of anomalously large turbulence structures contributed from coarsely-resolved LES on the bounding domain. We outline approaches for addressing each of these issues, and provide progress to date.

This work performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. UCRL# LLNL-ABS-427571