The stable boundary layer (SBL) in the atmosphere is of considerable interest because it is often the worse case scenario for air pollution studies and health effect assessments associated with the accidental release of toxic material. Traditional modeling approaches used in such studies do not simulate the non-steady character of the velocity field, and hence often overpredict concentrations while underpredicting spatial coverage of potentially harmful concentrations of airborne material. The challenge for LES is to be able to resolve the rather small energy-containing eddies of the SBL while still maintaining an adequate domain size. This requires that the subgrid-scale (SGS) parameterization of turbulence incorporate an adequate representation of turbulent energy transfer. Recent studies have shown that both upscale and downscale energy transfer can occur simultaneously, but that overall the net transfer is downscale. Including the upscale transfer of turbulent energy (energy backscatter) is particularly important near the ground and under stably-stratified conditions. The goal of this research is to improve the ability to realistically simulate the SBL. The large-eddy simulation (LES) approach with its subgrid-scale (SGS) turbulence model does a better job of capturing the temporally and spatially varying features of the SBL than do Reynolds-averaging models. The scientific objectives of this research are: (1) to characterize features of the evolving SBL structure for a range of meteorological conditions (wind speed and surface cooling), (2) to simulate realistically the transfer of energy between resolved and subgrid scales, and (3) to apply results to improve simulation of dispersion in the SBL.

Large-eddy simulation (LES) of the evolving stable boundary layer (SBL) provides unique data sets for assessing the effects of stable stratification on transport and dispersion. The simulations include the initial development of the convective boundary layer (CBL) in the afternoon, followed by the development of an SBL after sunset with a strong, surface-based temperature inversion. The structure of the turbulence is modified significantly by negative buoyancy associated with the temperature inversion. The magnitude of velocity variances is reduced by an order of magnitude compared to that in the CBL, and the vertical velocity variance is damped further as the static stability preferentially damps vertical motions. The advanced subgrid-scale turbulence model allows simulation of intermittently enhanced periods of turbulence in the SBL that am often observed. During these turbulent episodes, mixing is increased within the SBL. Air pollution models that account only for the long-term mean structure of the SBL do not include the effects of these episodes. In contrast, our LES results imply that material released near the surface and mixed to higher elevations would be transported by stronger winds and in different directions, due to the vertical shear of horizontal wind speed and direction. Material released at altitude in the SBL will tend to be mixed downward toward the surface during these turbulent episodes in a fumigation-like scenario at night.

The stable boundary layer (SBL) has received less attention in atmospheric field studies, laboratory experiments, and numerical modeling than other states of the atmospheric boundary layer. The low intensity and potential intermittency of turbulence in the SBL make it difficult to measure and characterize its structure. Large-eddy simulation (LES) offers an approach for simulating the SBL and, in particular, its evolution from the onset of surface cooling. Traditional approaches that involve Reynolds-averaged models of turbulence are not able to simulate the stochastic nature of the intermittent turbulence that is associated with the SBL. LES shows promise in this area through its explicit calculation of turbulent eddies at resolved scales. In the LES approach, the Navier-Stokes equations governing the flow are averaged (filtered) over some small interval, such as one or more cells of the computational grid. The grid size is small enough so that large eddies, which carry most of the turbulent energy, are explicitly calculated. The turbulence associated with the subgrid-scale (SGS) eddies is modeled. In the Reynolds-averaging approach, on the other hand, the turbulence model must account for all scales of turbulence. Thus the advantage of LES is that the choice of turbulence parameterization for the SGS turbulence is not nearly as critical as in the Reynolds-averaged approach. Complications faced by turbulence models, such as anisotropy and pressure-strain correlations, are associated mainly with large, energy-containing eddies. LES offers the potential for more realistic simulations since the more complicated features of turbulence are calculated explicitly. The ability of LES to simulate the stochastic behavior of turbulence makes this approach suitable for developing and testing stochastic models of turbulent diffusion. One of the goals of the present work is to provide stochastic datasets to be used in such studies.

Research was performed using a turbulence boundary layer model to study the behavior of cold, dense flows in regions of complex terrain. Results show that flows develop a balance between turbulent entrainment of warm ambient air and dense, cold air created by surface cooling. Flow depth and strength is a function of downslope distance, slope angle and angle changes, and the ambient air temperature.

First concise textbook on Large-Eddy Simulation, a very important method in scientific computing and engineering From the foreword to the third edition written by Charles Meneveau: "... this meticulously assembled and significantly enlarged description of the many aspects of LES will be a most welcome addition to the bookshelves of scientists and engineers in fluid mechanics, LES practitioners, and students of turbulence in general."