Extended Abstract (376K)2.2
Simulations of an ozone episode during the Central California Ozone Study, Part II: CAMx air quality model simulations
Su-Tzai Soong, Bay Area Air Quality Management District, San Francisco, CA; and S. Tanrikulu, J. M. Wilczak, J. W. Bao, P. Martien, and S. A. Michelson
Ozone concentrations frequently exceed the federal 1- and 8-hour (124 ppb and 84 ppb, respectively) ozone standards in northern and central California. The Central California Ozone Study (CCOS) field program was conducted in 2000 to study the causes of ozone formation in the region. Several major cities are located in the study domain: San Francisco, Sacramento, Fresno, and Bakersfield. During a typical ozone episode, the Eastern Pacific high-pressure system dominates the study area. This high-pressure system fosters low-level divergence, subsidence, and a shallow mixing depth. These meteorological conditions and emission characteristics of the region can cause ozone concentrations to exceed the ozone standards in the San Francisco Bay Area (SFBA), and areas near or downwind of Sacramento, Fresno and Bakersfield.
Continuous meteorological and air quality measurements were made during the CCOS field program from June 26 to October 2, 2000. Additional meteorological and air quality data were collected during ozone episodes, including aircraft measurement, hydrocarbon and carbonyl measurements as well as rawinsonde/ozonesonde measurements. Several ozone episodes were captured. The July 30 – August 2 episode was selected for simulation because, during this episode, ozone exceeded the federal 1-hour standard in all major metropolitan areas of the study region.
Three MM5 simulations were made for the period from July 29 to August 2. Details of these runs are reported in Part I of this paper:
· Run 1 uses the 5-layer soil model without FDDA. · Run 2 uses the Noah land-surface model without FDDA. · Run 3 uses the Noah land-surface model and the analysis nudging on the 36 km domain and observational nudging on the 4 km domain.
Using these meteorological model outputs, CAMx model simulations were conducted to investigate the response of CAMx to the meteorological simulations above. These CAMx runs use the day-specific emissions prepared by the California Air Resource Board. The initial and boundary conditions were adapted from those used in the SARMAP Air Quality Model. All runs started on 04:00 PST July 29, 2000, to provide a two-day spin up time. In this abstract we emphasize work completed for the SFBA; however, the formation of ozone in all areas (SFBA, Sacramento, Fresno and Bakersfield) will be reported in the paper.
During this episode, there was only one exceedence of the federal 1-hour ozone standard in the SFBA, which occurred at Livermore station (126 ppb) on July 31, 2000. In the simulation using MM5 Run 1 outputs, ozone greater than 80 ppb started to form at 14:00 PST along a north-south line through Livermore. During the next two hours, this line of high ozone moved slowly eastward to about 10 km east of Livermore while the maximum ozone increases to 98 ppb. Compared to the observed location of ozone at Livermore and Patterson Pass (an elevated CCOS site about 10 km east of Livermore), the CAMx-simulated location of maximum ozone using MM5 Run 1 output is quite acceptable. The CAMx runs using the MM5 Run 2 and 3 outputs also produced a north-south line of high ozone. The magnitudes of maximum ozone on these lines were between 95 and 100 pbb, comparable to the values in Run 1. The locations of the lines, however, were further east, about 20-25 km east of Livermore.
The locations of the convergence line and the locations of simulated high ozone are closely related. The overall surface-wind patterns in the San Francisco Bay Area are similar in the 3 MM5 runs, but there are subtle differences in the wind pattern among the runs in and near the Livermore Valley. At 14:00 PST, two hours before the observed Livermore ozone maximum, Run 1 has a well defined convergence line along the hills about 12 km east of Livermore. This convergence is likely caused by mountain heating. Another convergence area occurs at the sea breeze front, located about 10 km west of Livermore at this time. The use of Noah LSM in Run 2 and 3 increases the westerly wind throughout the Livermore Valley. The two convergence lines simulated in Run 1 due to sea breeze and mountain heating, respectively, have merged together in Run 2 and 3. The stronger westerly wind in these two runs also pushed this convergence line 5-10 km further from Livermore to the east.
In all simulations of this episode, ozone is underestimated. The degree of underestimation increases from about 30 ppb near Livermore to about 50 ppb in the San Joaquin Valley. We have conducted numerous experiments with CAMx to identify the causes of underestimated ozone. These experiments include modifications to surface temperature, grid size, horizontal and vertical diffusivity, and mixing depth. In addition, we have examined the effect of applying different parameterizations for predicting mixing depth within MM5.
In general, the MM5-CAMx couple produced reasonable predictions of locations and timing of peak ozone. However, improvements are needed to minimize the underestimation of peak ozone. Also, a high degree of accuracy in simulating the location and timing of the sea breeze is needed in the SFBA. A misplaced center of maximum ozone by 25 km there can move the location of simulated peak ozone from one district to another. Additional analyses of the meteorological patterns are being conducted to help understand the causes of underestimation of peak ozone and will be used to improve air quality model performance throughout the modeling region.
Session 2, Air Quality Forecasting - Case Studies
Monday, 23 August 2004, 10:30 AM-12:00 PM
Previous paper Next paper
Recorded presentation