11.07.2014 Views

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

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

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

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>Final</strong> <strong>Report</strong> — September 2004<br />

include HYSPLIT with EDAS/FNL input, the CAPITA Monte Carlo model with MM5,<br />

ATAD with MM5, and ATAD with raw sounding data. In general, choice of 5, 7, or 10-day<br />

back trajectories made little difference for this evaluation and the results were usually the<br />

same within the standard error of the regression coefficients, irrespective of the trajectory<br />

length. (Note that the results given here depend somewhat on the geometries of the tracer<br />

release regions shown in Figure 9-7. Different boundaries for the regions surrounding each<br />

tracer source may produce different perceptions of performance.)<br />

The worst performance is by HYSPLIT with MM5 input, a combination that<br />

dramatically overstates the Eagle Pass tracer contribution and understates that from San<br />

Antonio. Eagle Pass is approximately 250 km from K-Bar, while San Antonio is almost<br />

twice as distant at approximately 450 km. This may be an indication that HYSPLIT with<br />

MM5 has too many back-trajectory endpoints close to the receptor at the expense of too few<br />

farther away.<br />

Other problem combinations were ATAD with EDAS/FNL input and the Monte<br />

Carlo model with EDAS/FNL input, both of which overestimated tracer arriving from<br />

Houston and underestimated the contribution from San Antonio. Houston is the most distant<br />

of the three modeled release sites at approximately 750 km from K-Bar, so these latter<br />

combinations are overestimating the most distance source area.<br />

Results of this evaluation suggest that the best model/wind field combination for<br />

tracer attributions, and thus for accurate back trajectory placement within south Texas, is<br />

HYSPLIT with EDAS/FNL input. The combinations of CAPITA Monte Carlo or ATAD<br />

with MM5 input and ATAD with raw sounding input also were able to re-create the known<br />

tracer attributions. Because HYSPLIT with MM5 input performed poorly in both the tracer<br />

test described in this section and the simulated sulfate tests described in Section 9.6, this<br />

combination was not used for <strong>BRAVO</strong> sulfate source attribution. Attribution modeling using<br />

back trajectories from either the Monte Carlo or ATAD models with EDAS/FNL input is also<br />

suspect.<br />

9.4 Evaluation of the CAPITA Monte Carlo Model Using Perfluorocarbon Tracer<br />

Measurements<br />

The CAPITA Monte Carlo particle dispersion model was tested against the<br />

perfluorocarbon tracer measurements, in order to evaluate that model's ability to simulate<br />

near field and synoptic scale transport and to further explore the effects of different wind<br />

fields. The approach used for this evaluation and the results it produced are described here;<br />

details can be found in the CIRA/NPS <strong>BRAVO</strong> report (Schichtel et al., 2004), which is<br />

included in the Appendix.<br />

As described in Section 8.4.1, the CAPITA Monte Carlo model is a long-range<br />

transport model that simulates air mass transport and diffusion by tracking the movement of<br />

multiple particles released from a source. A modeled wind field is used to advect the<br />

particles in three-dimensional space. The vertical mixing that takes place within the<br />

atmospheric boundary layer is simulated using a Monte Carlo technique that evenly<br />

distributes the particles between the surface and the mixing height.<br />

9-20

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!