Large-eddy simulations of the stable boundary layer (SBL) have been conducted for a range of meteorological forcing conditions in terms geostrophic wind speed and surface cooling. The SBL is of considerable interest because it is often the 'worst 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 subgrid-scale (SGS) turbulence model used in our large-eddy simulation (LES) approach allows the upscale transfer (backscatter) of energy and has made possible the simulation of episodes of enhanced turbulence in the SBL. The turbulent episode is associated with the breakdown of large-scale wave-like activity in the upper part of the SBL. Such episodes of enhanced turbulence have been observed in the SBL (i.e. Coulter, 1990).

We have analyzed the turbulent flux of heat as a surrogate for passive scalars in order to apply our LES results to problems of dispersion in the SBL. We have evaluated eddy diffusivities directly from our LES-generated fields during turbulent episodes and during undisturbed SBL periods. These are compared with eddy diffusivities from an algorithm developed for use in a practical dispersion model. The agreement is surprisingly good for the undisturbed SBL, but the algorithm cannot represent the complexity of eddy diffusion during the turbulent episode. However, if information is available about the vertical distribution of turbulence so that local scaling can be used, the estimates from the algorithm agree much better with LES-derived eddy diffusivities. Analysis of the LES temperature and velocity fields shows that traditional downgradient heat flux occurs in the layers near the ground during the episode, while, further up in the SBL, countergradient heat flux occurs due to overturning events.