Subgrid-scale modeling for the atmospheric boundary layer: Experiments and Simulations

F. Porte-Agel

Ph.D. Thesis, The Johns Hopkins University, 2000, Baltimore MD

ABSTRACT: Large-eddy simulation (LES) is a critical tool for the study of turbulent transport in the atmospheric boundary layer (ABL). A key factor in LES of the ABL, or more generally any turbulent flow, is the performance of the subgrid-scale (SGS) model that accounts for the dynamics of the non-resolved scales. Experiments and simulations are performed to address open questions in SGS modeling for the ABL. Three field experiments are carried out to obtain high-resolution spatial distributions of wind velocity and temperature using arrays of 3D sonic anemometers. These data allow us to study in detail the characteristics of the SGS heat flux and dissipation of temperature variance. Large-scale coherent structures, associated with the sweep/ejection nature of the flow, are found to have a very important contribution to the SGS variables. The field data are also used to test different SGS model formulations. The widely used eddy-diffusion model yields SGS fluxes and dissipation whose statistics (mean values, pdfs) are significantly different from the statistics of the measured SGS variables. Also, this model is not able to provide backscatter (negative SGS dissipation), which is found in the measurements. Models based on scale-similarity arguments (e.g. similarity and non-linear models) are better able to reproduce the SGS observations. In particular, they capture backscatter and yield SGS fluxes and dissipation that are in good agreement with the measurements. A critical issue in LES of high-Reynolds-number wall-bounded flows (such as the ABL) is the specification of the SGS model coefficient(s) at different positions in the flow. The dynamic model, which computes the value of the model coefficient based on scale similarity at the smallest resolved scales and assuming scale invariance of the coefficient, is implemented in simulations of a neutral ABL over homogeneous terrain. Results from the dynamic model demonstrate a clear dependence of the model coefficient on scale. A new scale dependent dynamic model is proposed that allows for the coefficient to change with scale. Applications of the scale-dependent model to LES of the ABL show improved dissipative properties, leading to more realistic spectra and mean velocity profiles.