PLUME DISPERSION IN THE CONVECTIVE BOUNDARY-LAYER .2. ANALYSES OF CONDORS FIELD EXPERIMENT DATA

Extensive analyses are performed on data from the CONDORS (convective diffusion observed with remote sensors) field experiment, described in detail by Eberhard et al. Convective scaling is used to facilitate co mparisons with laboratory and numerical simulations and to give the re sults generality. Near-surface chi/Q from lidar-detected oil fog are g enerally in excellent agreement with nearby samples of coreleased SF6, considering the large spatial gradients found in the oil fog. Extrapo lations to the surface of integral chidy/Q of radar-detected ''chaff'' agree reasonably well with most oil fog values after a mathematical c ompensation for the chaff's settling speed. Measured wind direction di stributions compare favorably with integral chidy and integral chidz d istributions of both tracers. The directly measured bulk variables sig ma(y) and sigma(z) show little effect of source height or of tracer ty pe except that surface-released sigma(y) are enhanced by up to 60% at X < 0.3 X = (x/U)w/z(i), where U is mean wind speed, w* is the conve ctive scale velocity, and z(i) is mixing depth!. Generally, sigma(y) a lmost-equal-to 0.6z(i)X = 0.6wx/U at X < 1, sigma(y) almost-equal-to 0.6z(i)X2/3 at X > 1, and sigma(z) almost-equal-to 0.6z(i)X until limi ted by reflections; sigma(y) systematically reduces for averaging time s divided by z(i)/U less than 6. Also, for surface releases at small X , oil fog surface-extrapolated integral chidy/Q gives good agreement w ith Nieuwstadt; with the Gaussian assumption, this implies sigma(z) al most-equal-to 0.9z(i)X3/2. Composite patterns of C(y) = Uz(i) integral chidy/Q versus X and z/z(i) for surface releases are substantially in agreement with Willis and Deardorff's laboratory and Lamb's LES simul ations. In the aggregate, the oil and chaff measurements support a (1 + 2z(s)/z(i)) enhancement factor in maximum surface values of integral chidy over Gaussian plume model predictions, with these maxima occurr ing mostly near x = 2z(s)U/w, Where z(s) is source height. This agree s with most simulation results. Attention is focused on the effect of wBAR anomalies and persistence of C(y) patterns.