(BRAVO) Study: Final Report. - Desert Research Institute

Final Report — September 2004
7
6
Other, 1.56
5
PM 2.5 = 4.66 µg/m 3
Concentration (µg/m 3 )
4
3
2
OC, 1.27
EC, 0.12
NH 4 + , 0.82
NO 3 – , 0.20
PM 2.5 = 6.49 µg/m 3 SO 4 2– , 2.63
Other, 0.80
OC, 0.44
EC, 0.06
NH + 4 , 0.68
NO – 3 , 0.04
1
SO 4 2– , 2.52
0
Observed
Predicted
Figure 9-28. Compositions of observed and simulated PM 2.5 at K-Bar, averaged over the
duration of the BRAVO Study.
than the observations (r 2 =0.61). In addition, the model failed to simulate peak PM 2.5 mass
events at times when sulfate was only a minor contributor. Finally, linear regression lines
through the observed and modeled data in plots of PM 2.5 concentration versus sulfate
concentration yield similar slopes, but the line representing the simulated results is offset
1.96 µg/m 3 below the one for the measurements. (This separation is consistent with the mean
bias of -1.83 µg/m 3 shown for PM 2.5 in Table 9-22.)
Rather, it turns out that error in simulating the organic mass concentrations is the
principal contributor to the underestimation of total fine particulate mass. The average mean
simulated organic mass is approximately one-third of the observed organic mass. Even as a
proportion of the simulated PM 2.5 mass concentration, organic mass only accounted for 9.5%
of the total predicted PM 2.5 mass whereas organic material corresponded to 19.5% of the
mass in the ambient data.
The contribution of organic material to the PM 2,5 mass concentration can be
significant. On seventeen days, observed organic mass concentrations were greater than
those of sulfate. In contrast, the model only predicted fine organic mass greater than half of
the predicted fine sulfate mass on four occasions. The tendency to underestimate the
contribution of organic mass was greatest at higher PM concentrations.
9-62