(BRAVO) Study: Final Report. - Desert Research Institute
(BRAVO) Study: Final Report. - Desert Research Institute
(BRAVO) Study: Final Report. - Desert Research Institute
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
<strong>Final</strong> <strong>Report</strong> — September 2004<br />
Table 9-22. CMAQ-MADRID performance for 24-hr concentrations at the K-Bar site over the duration<br />
of the <strong>BRAVO</strong> study. 6<br />
Statistic units SO4 2– NO3 – NH4 + EC OM Other PM2.5<br />
Observed mean µg/m 3 2.52 0.20 0.82 0.12 1.27 1.56 6.49<br />
Simulated mean µg/m 3 2.63 0.04 0.68 0.06 0.44 0.80 4.66<br />
Coeff. of determination (r 2 ) 0.53 0.03 0.50 0.00 0.10 0.02 0.23<br />
Mean bias µg/m 3 0.11 -0.16 -0.13 -0.06 -0.82 -0.76 -1.83<br />
Mean error µg/m 3 1.08 0.20 0.34 0.09 0.84 1.47 3.13<br />
Mean Normalized Bias % 19 26 6 8 -60 -45 -13<br />
Mean Normalized Error % 54 194 51 89 6 160 51<br />
Fractional Bias % -1 -154 -13 -13 -116 163 -31<br />
Fractional Error % 46 169 48 19 121 282 87<br />
RMS error µg/m 3 1.51 0.24 0.45 0.14 1.10 2.37 4.11<br />
particle mass were measured at the K-Bar site. The “Other” component represents the<br />
difference between the measured PM 2.5 mass concentration and the sum of the major<br />
components; this “other” component is likely to contain dust, crustal material, metal oxides,<br />
and sea-salt and may also represent water remaining in the sample at the time of weighing.<br />
In addition, since some components of PM 2.5 may volatilize during sample handling and<br />
measurement, the “other” component could also reflect negative contributions. <strong>Final</strong>ly,<br />
“other” bears the burden of absorbing all of the measurement uncertainties of the other<br />
components.<br />
The stacked bar charts in Figure 9-28 illustrate the composition information in Table<br />
9-22. Over the analysis period, the mean PM 2.5 mass is underestimated by a factor of 1.4,<br />
equivalent to an underestimate of 1.83 µg/m 3 . On the other hand, the average simulated<br />
absolute concentrations of sulfate and ammonium agree quite well with the measured<br />
concentrations. However, due to the underestimation of all non-sulfate components, the<br />
model estimates a higher contribution of fine sulfate to PM 2.5 mass. The simulated<br />
proportional contributions of the various other components of PM 2.5 are also quite different<br />
from the measured proportions, except for ammonium (NH 4 + ) and elemental carbon.<br />
The difference between model estimates and observations of total fine particle mass<br />
cannot be explained solely by discrepancies in model predictions for sulfate. CMAQ-<br />
MADRID predicted a much tighter relation between fine sulfate and PM 2.5 mass (r 2 = 0.91)<br />
6 Statistics in Table 9-22 were computed without the lower cutoff value of 0.1 µg/m 3 used in previous<br />
tables in this section, and thus these statistics test model performance at lower concentration levels<br />
than those that apply for Table 9-17. Also, only those days with valid measurements for all<br />
components, and thus with the ability to derive a value for “Other”, were used.<br />
9-61