A Semi-empirical Model Relating MeteorologyAnd Ozone In The San Francisco Bay Area

Recent ozone trends in the San Francisco Bay Area have been volatile, with a total of 13 days exceeding the national 1-hour ozone standard of 124 ppb in 1995, 12 in 1996, and 8 in 1998 but only 3days in 1993, 2 in 1994, and 0 in 1997. The specific causes of this volatility are elusive due to the complex chemical reactions and variable meteorology that govern tropospheric ozone; these factors also make it difficult to determine the effects of reducing the emissions of ozone precursors. To help understand ozone trends, this study develops a semi-empirical model relating daily ozone to daily meteorological conditions in the Bay Area. Meteorological parameters used in the model include pressure and temperature gradients, mixing height, temperature (surface and aloft), cloud cover, and wind speed. An indicator for weekend days is also included. The model is constructed in two stages. First, synoptic-scale pressure gradients and meso-scale temperature gradients, summarized by principal components, are used to cluster all days in the May — October ozone seasons for 1988 — 1996. Second, within-cluster additive models use other variables to quantify daily basin-wide maximum ozone concentrations. The within-cluster models are calibrated using daily ozone data from 1993 through 1995 only. The overall model's performance is very encouraging, achieving an R^ of 83 percent for the 1993-1995 data. The modeling approach presented in this study uses meteorological data that are available routinely in many regions, suggesting that the method may be an effective tool in other regions. Transactions on Ecology and the Environment vol 29 © 1999 WIT Press, www.witpress.com, ISSN 1743-3541