Statistical Variability of Dispersion in the Convective Boundary Layer: Ensembles of Simulations and Observations

A Lagrangian particle dispersion model (LPDM) driven by velocity fields from large-eddy simulations (LESs) is used to determine the mean and variability of plume dispersion in a highly-convective planetary boundary layer (PBL). The total velocity of a ``particle" is divided into resolved and unresolved or random (subfilter scale, SFS) velocities with the resolved component obtained from the LES and the SFS velocity from a Lagrangian stochastic model. This LPDM-LES model (Weil et al., 2004) is used to obtain an ensemble of dispersion realizations for calculating the mean, root-mean-square (rms) deviation, and fluctuating fields of dispersion quantities. An ensemble of 30 realizations is generated for each of three source heights: surface, near-surface, and elevated. We compare the LPDM calculations with convection tank experiments and field observations (the CONDORS and Prairie Grass experiments) to assess the realism of the results.

The overall conclusion is that the LPDM-LES model produces a realistic range of dispersion realizations and statistical variability (i.e., rms deviations) that match observations in this highly convective PBL, while also matching the ensemble-mean properties. This is true for the plume height or trajectory, vertical dispersion, and the surface values of the crosswind-integrated concentration (CWIC), and their dependence on downstream distance. One exception is the crosswind dispersion for an elevated source, which is underestimated by the model. Other analyses highlighted important LPDM results on: 1) the plume meander and CWIC fluctuation intensity at the surface, and 2) the applicability of a similarity theory (Yaglom, 1972) for plume height from a surface source to only the very strong updraft plumes---not the mean height.