Simulation of an ozone episode during the Central California Ozone Study. Part 1: MM5 meteorological model simulations

The Central California Ozone Study (CCOS 2000) is a combined observational and modeling program designed to improve our understanding of the mechanisms of ozone formation and transport within California. CCOS 2000 was motivated by the fact that ozone concentrations frequently exceed the federal 1- and 8- hour (124 and 84 ppb) standards in central California. During the CCOS 2000 field program, extensive observations were collected in central California to document high ozone episodes and the meteorology that is associated with them. These observations included 25 wind profiling radars, 297 surface meteorological stations, 120 surface ozone monitors, 25 NOy monitors, and several instrumented aircraft. The CCOS field program operated from June 1 through October 2, 2000. During this period several moderately high ozone episodes occurred, of which the episode that occurred between July 30-August 2 will be examined in this study.

Meteorological phenomena in the central California region that are known to have a pronounced impact on ozone concentrations include 1) the sea-breeze, which can bring cooler, moister, and lower concentration air as it propagates inland; 2) flow through the San Francisco Bay area, which is the principle inflow to the central valley, and the split of this flow, which determines the relative inflow into the northern Sacramento and southern San Joaquin Valleys; 3) nocturnal low-level jets, which can rapidly transport boundary layer pollutants along the central valley; 4) mesoscale eddies (the Schultz, Fresno, and Bakersfield) which can re-circulate ozone and its precursors; and 5) slope flows, which result in transport in or out of the valleys, support boundary layer venting along mountain crests, and produce subsidence or ascending motion over the valleys. These flow features all potentially affect the transport of pollutants by the mean wind. In addition, the depth of the atmospheric boundary layer is of critical importance for air quality, as it determines the depth through which pollutants are vertically mixed.

To better understand the role of the above meteorological phenomena on ozone transport and mixing, a combined meteorological and chemical modeling system was used to simulate ozone concentrations. This system was comprised of the MM5 meteorological model, and the CAMx chemical model. In this paper we present the meteorological modeling results, while in Part II (Soong et al.) the emissions data base, chemical model, and ozone results will be presented.

Three MM5 simulations for this episode were run using a 36-12-4 km nested model domain. The 4 km domain encompasses the CCOS field study area, which extends from the Pacific Ocean in the west to the Sierra Nevada in the east, and from Redding in the north to the Mojave Dessert in the south. Boundary and initial conditions were prescribed using the 6-hourly 40 km NCEP Eta analysis. All three simulations used the Eta planetary boundary layer (PBL) scheme, and Grell convective parameterization scheme on the 36 and 12 km grids. No convective parameterization scheme was used on the 4 km grid. The runs differ in the land surface scheme and the use of the four dimensional data assimilation (FDDA). The specifications of the 3 runs are:

1) Run 1 uses a 5-layer soil model without FDDA. 2) Run 2 uses the Noah land surface model (LSM) without FDDA. 3) Run 3 uses the Noah land surface model and analysis nudging on the 36 km domain and observational nudging of the profiler and surface winds on the 4 km domain.

Although all 3 runs provided fairly realistic simulations of the observed meteorology, some significant differences were found. Specifically, the 5-layer soil model simulation was found to under-predict the maximum daytime surface and boundary layer temperatures within the central valley by about 2 C on average, and to overpredict surface dewpoints by about 5 C on average. In comparison, Run 2 using the Noah LSM provided more accurate surface temperatures and dewpoints, with the model surface temperature bias reduced by half, to a 1 C cold bias, and with no appreciable bias in dewpoint. As a consequence of the warmer inland temperatures produced by the model, the sea-breeze was also enhanced when using the Noah LSM.

The simulation using both analysis and observation nudging was found to improve the wind fields to the point that overall very small differences (typically less than 0.2 m/s on average over the entire domain) were found between the model and original surface observations. However, at some individual locations the sea-breeze was still over-predicted when using FDDA, perhaps because the local density of observations was lower. Finally, the PBL depths were found to be well simulated in the runs using the Noah LSM and Eta PBL scheme. Other PBL parameterization choices within MM5 were found to provide considerably less skill in simulating the PBL depth.