A HYBRID LAGRANGIAN DISPERSION MODEL FOR ELEVATED SOURCES IN THE CONVECTIVE BOUNDARY-LAYER

A hybrid Lagrangian dispersion model is presented for predicting the mean (C) and root-mean-square fluctuating (sigma(c)) concentrations from elevated sources in the convective boundary layer (CBL). The model is based on Gifford's (1959, Adv. Geophys. 6, 117-138) meandering plume concept and includes: (1) a Lagrangian stochastic model for treating the plume meander caused by the large-scale convective turbulence, and (2) an entrainment model to determine the plume rise and growth relative to the meander. The hybrid model is currently limited to weakly-to-moderately buoyant plumes as defined by F(*)less than or similar to 0.1, where F-* is the dimensionless buoyancy flux: F-*=F-b/U(*)(2)h, where F-b is the stack buoyancy flux, U is the mean wind speed, w(*) is the convective velocity scale, and h is the CBL depth. The modeled crosswind-integrated concentration (C-y) and sigma(c)/C frelds showed fair agreement with the laboratory convection tank measurements of Deardorff and Willis (1984, Atmospheric Environment 18, 1297-1309; 1988, Lectures on Air Pollution Modeling, pp. 357-384). The C-y fields were modeled both with and without relative dispersion (sigma(r)). The C-y fields with sigma(r) included showed better agreement with the measurements in the lower half of the CBL than those without sigma(r), and they demonstrated the importance of including entrainment due to ambient turbulence in the inertial subrange. Plume rise and buoyancy effects were most significant for dimensionless distances X = w(*)x/Uh less than or similar to 2-3, where the near-surface sigma(c)/C ranged from 1 to 7; x is the downwind distance. For the same X range, measurements in power plant plumes showed that sigma(c)/C less than or similar to 2 and hence smaller than the values in the laboratory. A possible reason for the difference is the small Reynolds number in the convection tank.