Data assimilation and air quality - Air Pollution Modeling and Its Application - page 262

is still not the ideal situation of hour-centered fields, but rather staggered late by
half an hour from the time stamps I/O API sees. This half-hour lagged hourly
average Kz is used to drive turbulent mixing of tracer gases in CMAQ.
3. August 2-3, 2006 Episode and Verifications
A surface elevated O 3 concentration episode on August 2-3, 2006 is chosen. This
episode provides opportunity to analyze the characteristics of the schemes at
different locations. The episode is chosen to co-incide with the TexAQS campaign
where a good set of ozonesonde data are available. All runs are daily 12 UTC
cycle out-to 48 h forecast initialized by outputs from a previous cycle. A regionalized
second-day daily 8 h maximum forecast verification dividing the CONUS into
6 regions: Pacific Coast, Rocky Mountains, Lower Midwest, Upper Midwest,
Southeastern and Northeaster US, in terms of correlation and Root Mean Square
Error (RMSE) for July 30-August 4, 2006, showed that there was no obvious
dominant winner among the three schemes. Dependent on the region and the time
of focus, all three schemes performed comparable in all six regions during the
analyzed 6 days period. It is arguable that Cases b and c are superior to Case a for
Lower Midwest (LM) and Northeastern (NE) regions; especially Case c has been
consistently superior among all three schemes throughout the period.
4. Structures of K z , TKE, O 3 and RH on August 2, 2006
The vertical structure of the O 3 field is further examined through using ozonesonde
data made available through the ITEX-B campaign (Thompson et al., 2008). Four
sites have been chosen: (1) Table Mountain, CA (34.4°N, 117.7°W) at elevation
2,285 m; Boulder, CO (40.3°N, 105.2°W) at elevation 1,745 m; Huntsville, AL
(34.7°N, 86.5°W) at elevation 196 m, and Beltsville, MD (39.04°N, 76.52°W) at
elevation 24 m. Figure 1a shows the Kz profiles predicted by the three schemes
and then NAM predicted TKE profile at launching time of 20:52 UTC, on August
2, 2006 at the Table Mountain, CA site. Figure 1b shows ozonesonde data overlaid
with the model-predicted O 3 concentrations by all three cases over the same site at
20:52 UTC. A measured moisture profile inferred PBL height at 1,200 m AGL is
also indicated with overlaid of measured and NAM-predicted relative humidity (RH).
Figures 2 -4 repeated similar analyses from the above paragraph to Boulder, CO
with ozonesonde launch time at 19:28 UTC on August 2, 2006; Huntsville,
AL with ozonesonde launch time at 17:36 UTC on August 2, 2006; and Beltsville,
MD with ozonesonde launch time at 19:18 UTC on August 2, 2006; respectively.
Over Boulder, CO, all three cases captured the PBL height, inferred at 1,750 m
AGL, rather well with a relatively small overshot ( Fig. 2). It is noted that both
measured and model-predicted O 3 concentration profiles show a rather uniform