Vertical dispersion of a passive scalar in the convective boundary layer: new laboratory results

The laboratory results of Willis and Deardorff (1976, 1978, 1981) on the vertical dispersion of passive tracers in the CBL continue to be widely used to validate and ‘tune’ dispersion models. However, there remain significant unresolved discrepancies between these data and results from a range of models including LES, such as differences of up to 50% in the downwind location of ground-level maxima. The CONDORS experiment collected field data to compare with the modelling results. While general agreement was obtained, sampling problems prevented detailed resolution of the discrepancies.

As part of a study to obtain a detailed set of CBL turbulence and dispersion data from our laboratory convection tank, a series of experiments has been undertaken with passive releases in the CBL at dimensionless heights of 0.1, 0.25, 0.5 and 0.9. The water tank simulation of penetrative convection employs saline rather than thermal convection to model a no-wind atmospheric CBL. Typical values of the scaling parameters in these experiments were mixing depth zi~25 cm, convective velocity w*~1 cm/s, and convective timescale t*~25 s. The total length of the tank was 12zi and the width 6zi.

Experimental difficulties of introducing an instantaneous line source were overcome by modifying the analysis technique to allow the line to be laid down more slowly (at about 10 cm/s). The tracer dye was released from a backward-facing tube towed along the long axis of the tank at a speed equal to the mean efflux velocity of the dye. Using five separate realisations for each release height gave an effective sampling time of about 40t*. Crosswind-integrated concentrations were calculated from analysis of video images of the dispersing plume to produce Z-T contour plots. The analysis procedure also allowed separation of the relative and meander components of the total dispersion, as well as computation of the variance of the crosswind-integrated concentration.

A range of these results will be presented and compared with earlier experimental data and model results. Preliminary analysis has shown that for release heights of 0.25 and 0.5zi, peak ground-level concentrations generally occurred slightly further downwind than in the Willis and Deardorff experiments. A brief video of the some of the laboratory dispersion realisations will also be shown.