Dispersion in the planetary boundary layer (PBL) is a random process driven by the statistical nature of turbulence. The variabililty of dispersion and concentration fields is probably greatest for instantaneous sources or puffs and is important in a number of problems. These include estimating the peak concentration of toxic and hazardous materials from instantaneous releases, either planned or unplanned.

This paper focuses on the statistical variability of puff dispersion using concentration fields generated by a Lagrangian particle dispersion model (LPDM) driven by large-eddy simulations (LESs) of the PBL. In this case, we study the convective boundary layer (CBL), which typically exhibits the most variability of all PBLs. Puff dispersion is a problem in relative dispersion (about the wandering puff centerline) and can be simulated using a ``two-particle" dispersion model. Thus, we adopt a two-particle model in the LPDM. The coupled LPDM-LES approach with a "one-particle" model was used previously to simulate the mean concentration fields from continuous sources (i.e., plumes) in the CBL (Weil et al., 2004) and found to agree well with laboratory data and field observations. In this paper, we extend the approach to the mean and variability of concentrations due to an instantaneous source. For each source, 30 realizations or samples of the concentration field are obtained for the same average LES flow field, which has much spatial and temporal variability. From the realizations, we determine the overall statistical properties---the mean height, the root-mean-square (rms) puff meander, the mean and rms fluctuating concentrations, etc.

Results of the simulations reveal the broad variability in the puff dispersion properties for several source heights. The results are compared with those generated for plumes and show the greater variability and rms values in the case of puffs. In conjunction with observations and experiments, the results may be used to evaluate simpler models of puff dispersion.