The Weather Research and Forecasting Model permits finescale large-eddy simulations (LES) to be nested within coarser simulations, an approach that can generatemore accurate turbulence statistics and improve other aspects of simulated flows.However, errors are introduced into the finer domain fromthe nestingmethodology. Comparing nested domain, flat-terrain simulations of the neutral atmospheric boundary layer with singledomain simulations using the same mesh, but instead using periodic lateral boundary conditions, reveals the errors contributed to the nested solution from the parent domain and nest interfaces. Comparison of velocity spectra shows good agreement among higher frequencies, but greater power predicted on the nested domain at lower frequencies. Profiles of meanwind speed show significant near-surface deficits near the inflowboundaries, but equilibrate to improved values with distance. Profiles of the vertical flux of x momentum show significant underprediction by the nested domain close to the surface and near the inlet boundaries. While these underpredictions of the stresses, which cause the near-surface velocity deficits, attenuate with distance within the nested domains, significant errors remain throughout. Profiles of the resolved turbulence kinetic energy show considerable deviations from their single-domain values throughout the nested domains. The authors examine the accuracy of these parameters and their sensitivities to the turbulence subfilter stress model, mesh resolution, and grid aspect ratio, and provide guidance to practitioners of nested LES.
All Science Journal Classification (ASJC) codes
- Atmospheric Science
- Large eddy simulations
- Model evaluation/performance
- Subgrid-scale processes