Eulerian dispersion modeling with WRF-LES of plume impingement in neutrally and stably stratified turbulent boundary layers

The vast range of space-time scales associated with turbulent flow adjacent to rugged terrain is especially problematic to predictive dispersion modeling in atmospheric boundary layers (ABLs) partly due to the presence of non-linear flow features (e.g., recirculation zones, diffusion enhancement, etc.). It has been suggested that in such ABLs, explicitly modeling large turbulent eddies, through large-eddy simulation (LES), may help to curtail predicted concentration errors. In this work, passive scalars were introduced into the Weather Research and Forecasting (WRF) LES model for the purpose of simulating scalar plume interaction with an isolated terrain feature. Using measurements from the Cinder Cone Butte (CCB) field campaign, we evaluate the ability of WRF-LES to realistically simulate the impingement of Sulfur Hexafluoride (SF6) plumes onto CCB in both neutrally and stably stratified environments. Simulations reveal relatively accurate scalar trajectories with respect to thermal stability, including complex patterns such as plume splitting below the hill dividing streamline. Statistical accuracy varied with case study, but for the neutral case we recorded greater than 50% of predicted 1 h averaged surface concentrations within a factor of 2 of the observations. This metric, along with several others, indicates a performance accuracy similar to, or slightly better than, alternative Reynolds Averaged Navier–Stokes models. For the stably stratified case, the spatial distribution of surface concentrations was captured well; however, a positive concentration bias was observed which degraded quantitative accuracy scores. The variable accuracy of the WRF-LES model with respect to thermal stability is similar to what has been observed in regulatory analytical models (i.e., concentration under predictions in neutral environments and concentration over predictions in stable environments). Possible sources of error and uncertainty included the omission of mesoscale wind meandering (i.e., realistic boundary conditions) and sub-grid turbulence parameterization.