Journal of the Atmospheric Sciences · 1988 · 375 citations · 0 references
Natural SciencesEarth SciencesMeteorologyAtmospheric ScienceBoundary Layer MeteorologyComputational Fluid DynamicsBoundary-layer DynamicsBoundary-layer StructureLarge-eddy SimulationPlanetary Boundary LayerTurbulenceTurbulent FlowDissipation RateEddy ViscosityEddy Viscosity ModelsLarge-eddy Simulation Of Boundary LayersLarge-eddy Simulation Of ConvectionSubgrid-scale ModelingSubgrid-scale ParameterizationTurbulent Flow Modeling
The study examines the fidelity of turbulent flow in large‑eddy simulations by analyzing their spectra. The authors derived inertial‑range spectra expressions for a filtered wind field, compared them with theoretical predictions, and analyzed spectra from 96×96×96 large‑eddy‑simulation results to assess fidelity. The study concludes that a wave‑cutoff filter outperforms a Gaussian filter, that only a specific combination of subgrid‑scale dissipation and eddy‑viscosity parameters yields accurate inertial‑range spectra, and that higher‑order moment statistics remain robust to moderate parameter changes.
To examine the fidelity of the simulated turbulent flow, we analyzed the spectra of 96×&96×96 large-eddy-simulation results. We derived expressions for the inertial-range spectra of a filtered wind field, compared our computed spectra with the theoretical predictions, and drew two main conclusions. First, the wave cutoff filter is more appropriate for our large-eddy-simulation model than the Gaussian filter. Second, only a certain combination of the subgrid-scale parameters for the dissipation rate and eddy viscosity provides good inertial-range spectra. We offer an explanation for the unreasonably large sugrid temperature and moisture variances reported in Deardorff's 1974 lame-eddy-simulation results, and show that moment statistics up to third order are not sensitive to moderate changes in the subgrid-scale parameters.