Local and urban scale modeling - Air Pollution Modeling and Its Application - page 15

surface during neutral to stable conditions, with a relative emission density, ρ e /ρ a ,
equal to 1.52. Both puffs or continuous plumes were performed. 84 fast-response
samplers were located on four downwind arrays (25, 50, 100 and 225 m). Wind
speeds and directions at 2 m levels, friction velocities and Monin-Obukhov lengths
were provided for each of the 52 trials. Moreover, two sets of artificial roughness
arrays were used: Uniform Roughness Array (URA) with a roughness length of
about 0.01-0.02 m and Equivalent Roughness Pattern (ERP) with a roughness
length of about 0.12-0.24 m.
A computation domain of 400 × 240 × 100 m was considered. MicroSwift had
horizontal grid spacing of 2 m and a stretched grid in the vertical. 10,000 particles
were released per second from the area source, whatever the duration of release.
Concentration at sampler locations was computed. Roughness lengths of 0.015 m
and 0.18 m were considered for the two sets of artificial arrays, that is to say
respectively URA and ERP. Maximum predicted concentrations and observations
have been compared, for a 20 s averaging times, at the different downwind distances.
Kit Fox trials can be split into four groups, that is to say URA - continuous (12
trials), URA - puff (21 trials), ERP - continuous (6 trials) and ERP - puff (13
trials).
At the moment, only comparisons related to URA - puff and ERP - puff have
been performed. Each of these two last groups has been statistically evaluated in
order to determine the accuracy of MicroSpray model predictions with the observed
data. Geometric mean bias (MG), geometric variance (VG), as well as factor of 2
(FA2) are presented in the Table 1.

Table 1. Statistical indexes related to maximum concentrations (20 s averaging time)

URA - puff ERP - puff

Kit Fox experiment

21 Experiments 13 Experiments

MG 1.11 1.11

VG 1.20 1.22

FAC2 92.9% 84.6%

These results are quite encouraging and, in particular, well agree with those
obtained by three different versions of the model HEGADAS (Hanna and Chang,
2001), and the CFD code FLACS (Hanna et al., 2004).

References

Carissimo, B., Dupont, E., Musson-Genon, L., and Marchand, O., 1997. Note de Principe du
Code MERCURE. Version 3.1, Electricité de France, EDF HE-33/97/001, EDF publications,
France
Hanna, S.R., Strimaitis, D.G., and Chang, J.C. (1991) Hazard response modeling uncertainty (a
quantitative method). Vol. 2, Evaluation of commonly used hazardous gas dispersion models.
Sigma Research Corporation for AFESC, Tyndall AFB, FL, and API, Report Nos. 4545,
4546, and 4547, 338 pp