Fumigants
135
was compensated by running the model twice: once using the meteorology
near the source (north) and again using the meteorology at the receptors
(south). More detailed meteorological considerations, such as rotation and
stability index, are described in the article (Honaganahalli and Seiber, 2000).
Concentration patterns, highs and lows, and trends at a receptor exhibited
diurnal differences somewhat similar to measured concentration patterns.
Just as in the measured data, the diurnal meteorological patterns prevailing in the valley strongly infuenced the distribution of MeBr in the valley
airshed. In agreement with measured concentrations during daytime, the
ISCST3 model predicted higher concentrations on the windward slopes of
the Gabilan range than on the valley foor (i.e., CC); at night, the model predicted higher concentrations on the valley foor in the northern part (SN, SS).
Statistical analysis showed moderate correlation considering 24-hour
time-weighted average concentrations (R 2 ≈ 0.70) and nighttime concentrations
(R 2 ≈ 0.7), but poor daytime correlation (R 2 ≈ 0.2). We suspected the reasons
for this poor daytime performance of the model was due to the relatively
unstable atmospheric conditions throughout the valley, and that using different degrees of “rotation” for daytime and nighttime would result in higher
daytime correlations. For additional details regarding the analysis of MeBr
feld experiment, see Honaganahalli and Seiber (2000).
8.3.2.3 Flux from Multiple Fields: CALPUFF Model Performance
The CALPUFF model, using 3D meteorology provided by CALMET, was successful both physically and statistically in predicting the nighttime concentrations but was less successful in predicting concentrations at all stations
during daytime. During daytime, the CALPUFF model predicted concentrations only at four sites in the northern part, two of which were on the valley
foor (SN and SS). At all other sites, the model predicted zero or insignifcant
concentrations. During nighttime, CALPUFF predicted concentrations at all
receptors.
The correlation between measured and CALPUFF-predicted 24-hour
time-weighted average concentrations (R 2 = 0.55 and 0.82 for Seiber et al.’s
(1996) and Yates et al.’s. (1996) source inputs, respectively) and nighttime concentrations (R 2 = 0.89 and 0.90) suggest that the CALPUFF model was moderately successful in predicting the measured concentrations. For daytime,
the correlation between measured and model predicted values was poor
(R 2 = 0.22 and 0.24). This mixed result points at the inherent limitations of the
CALMET model to provide precise turbulence information needed for accurate estimation of dispersion parameters in a highly complex terrain.
For the CALMET model, topographic information of high resolution, which
could have resulted in estimation of better wind fow patterns, was not available. Also, all surface meteorological stations in the domain were clustered
in and around the city of Salinas. Among the available meteorological stations, all required elements were not available at all stations. For all of these
135
was compensated by running the model twice: once using the meteorology
near the source (north) and again using the meteorology at the receptors
(south). More detailed meteorological considerations, such as rotation and
stability index, are described in the article (Honaganahalli and Seiber, 2000).
Concentration patterns, highs and lows, and trends at a receptor exhibited
diurnal differences somewhat similar to measured concentration patterns.
Just as in the measured data, the diurnal meteorological patterns prevailing in the valley strongly infuenced the distribution of MeBr in the valley
airshed. In agreement with measured concentrations during daytime, the
ISCST3 model predicted higher concentrations on the windward slopes of
the Gabilan range than on the valley foor (i.e., CC); at night, the model predicted higher concentrations on the valley foor in the northern part (SN, SS).
Statistical analysis showed moderate correlation considering 24-hour
time-weighted average concentrations (R 2 ≈ 0.70) and nighttime concentrations
(R 2 ≈ 0.7), but poor daytime correlation (R 2 ≈ 0.2). We suspected the reasons
for this poor daytime performance of the model was due to the relatively
unstable atmospheric conditions throughout the valley, and that using different degrees of “rotation” for daytime and nighttime would result in higher
daytime correlations. For additional details regarding the analysis of MeBr
feld experiment, see Honaganahalli and Seiber (2000).
8.3.2.3 Flux from Multiple Fields: CALPUFF Model Performance
The CALPUFF model, using 3D meteorology provided by CALMET, was successful both physically and statistically in predicting the nighttime concentrations but was less successful in predicting concentrations at all stations
during daytime. During daytime, the CALPUFF model predicted concentrations only at four sites in the northern part, two of which were on the valley
foor (SN and SS). At all other sites, the model predicted zero or insignifcant
concentrations. During nighttime, CALPUFF predicted concentrations at all
receptors.
The correlation between measured and CALPUFF-predicted 24-hour
time-weighted average concentrations (R 2 = 0.55 and 0.82 for Seiber et al.’s
(1996) and Yates et al.’s. (1996) source inputs, respectively) and nighttime concentrations (R 2 = 0.89 and 0.90) suggest that the CALPUFF model was moderately successful in predicting the measured concentrations. For daytime,
the correlation between measured and model predicted values was poor
(R 2 = 0.22 and 0.24). This mixed result points at the inherent limitations of the
CALMET model to provide precise turbulence information needed for accurate estimation of dispersion parameters in a highly complex terrain.
For the CALMET model, topographic information of high resolution, which
could have resulted in estimation of better wind fow patterns, was not available. Also, all surface meteorological stations in the domain were clustered
in and around the city of Salinas. Among the available meteorological stations, all required elements were not available at all stations. For all of these