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

Final Report — September 2004
Overall, then, except for wind speed aloft, the quality of the 36-km meteorological
fields approximated the ad hoc benchmarks within the area of the 12-km domain, which
means that the wind direction, temperature, and mixing ratio fields approximate the state of
the art of meteorological modeling. The error in 36-km wind speed aloft was greater than its
benchmarks and its effect on regional and interregional transport assessments must be
considered. (See Section 9.1.2 below for additional evaluation of wind performance aloft.)
9.1.2 Comparison of MM5, EDAS, and FNL Wind Fields to Radar Wind Profiler
Measurements
Wind fields generated by the MM5 and FNL models and the EDAS analysis system
(see Section 8.2 for descriptions) were used with several source and receptor models applied
in the BRAVO study. REMSAD and CMAQ modeling used MM5 fields. The receptor
models used wind fields from either MM5 or a combination of EDAS (July-September) and
FNL (October). For some receptor models, both MM5 and the EDAS/FNL combination
were used, with somewhat differing attribution results. Here we compare the various wind
fields and evaluate them against wind profiler measurements. A more comprehensive
discussion of this comparison is provided in an appendix of the CIRA/NPS report on the
BRAVO Study (Schichtel et al., 2004). The CIRA/NPS report is included in the Appendix of
this report.
For this evaluation, wind measurements by radar wind profilers at Big Bend, Eagle
Pass, Llano, and Brownsville were compared to wind predictions by MM5, EDAS, and FNL
four times per day (6 am, noon, 6 pm, and midnight Central Standard Time). (See Figure 3-4
for map of profiler locations.) The radar wind profiler provided data for 60-m thick layers up
to about 2000-2500 m AGL, then every 100 m up to about 3500-4000 m AGL. The single
radar wind profile level closest to each model layer height was used for comparison. Since
FNL has few layers in the lower atmosphere, the number of comparisons was less than for
the other two methods.
Several metrics were used to compare model and radar wind profiler (RWP) winds:
• Average model wind and average RWP wind speed;
• Magnitude of vector difference between modeled and RWP winds;
• Average difference in wind direction (degrees);
• Average absolute value of wind direction difference;
• Percent of periods where model and RWP wind directions were within 20 degrees
and 30 degrees; and
• Percent of model and RWP wind directions from 8 general directions (N, NE, E,
SE, S, SW, W, NW) by layers -- 0-500 m, 500-1500 m, >1500 m AGL.
Results for some of these metrics are described here. The others are addressed in the
CIRA/NPS report in the Appendix.
Because the EDAS fields were not available for October 1999, the summary of MM5
fields was done separately for the July- September and October periods. This facilitates the
9-5