Environmental Engineering Reference
In-Depth Information
to ozone closer to the source region. In general, the spatial patterns are very
similar between the various implementations.
The CAMx and CMAQ estimated PM2.5 contribution from each source region
to the Milwaukee/Waukesha area is consistent. The total PM2.5 contribution can
also be examined at each receptor location by chemical composition. This type of
evaluation suggests that the contribution from more distance source regions tends
to be from secondarily formed species like nitrate and sulfate and local contri-
bution tends to be dominated by primarily emitted species.
CMAQ and CAMx show the best agreement for species dominated by primary
emissions like elemental carbon and primary organic carbon. The poorest agreement
is for PM2.5 nitrate ion, which is likely due to differences in model formulation
for nitrate aerosol formation and exacerbated by differences in advection and
deposition as it tends to be secondarily formed specie that is more likely to trans-
port in from nearby source regions.
Source apportionment also tracks the contribution from initial conditions,
which allows for an evaluation about how long initial concentrations remain an
important contribution to model estimates. The domain peak ozone contribution
from initial conditions decreases to approximately 1 ppb by the seventh day of the
modeling episode. Total 24-h averaged PM2.5 concentrations from initial con-
ditions reduce below 1.0 μg/m
3
after four simulation days in each quarter. This
suggests that four to five simulation days would be necessary to remove the
influence of start-up conditions on model estimates of PM2.5 and a week to
remove ozone initial concentrations.
4. Conclusions
Both models tend to under-predict ozone greater than 75 ppb. The relationship
between estimated ozone contribution over all receptors from all source regions is
strong between CAMx and CMAQ. Most of the variability in the error is due to
differences in contribution estimates from the boundary and the group consisting
of all other emissions in the modeling domain that were not tagged. CAMx and
CMAQ are fairly consistent with absolute model estimates of PM2.5 species, with
the exception of nitrate ion, of which CMAQ tends to predict higher concentrations.
The modeling systems also tend to have strong relationships in predicted source
contributions. Again, nitrate ion source contribution between models has the
weakest relationship of the species examined in this study. The largest contri-
buting species on the top 10% of modeled days include primarily emitted species
including primary organic carbon and elemental carbon. Contributions from source
areas furthest away from the Milwaukee/Waukesha area tend to be dominated by
secondarily formed species such as nitrate ion and sulfate ion. Despite differences
in model formulations, the source contributions estimated by each modeling
system compare well with each other. This increases the confidence that each is
appropriately implemented and suitable for the estimation of source contribution.
