A stepwise procedure is used to fit a full compliment of potential predictors, including surface and boundary layer conditions as well as descriptive measures of the sea-breeze and large-scale weather patterns, to the background and locally contributed 8-hr ozone observed in the EPA non-attainment region surrounding Houston, Texas.

In this context the “background” is considered the lowest observed peak 8-hr ozone at a select number of candidate background stations, while the “local contribution” is considered the difference between the highest observed peak 8-hr ozone in the region and the background. The treatment of background and local contribution separately is due to the expectation that a somewhat different set of meteorological predictors will be selected to best explain each component. Different long term trends may also be noted in background versus locally contributed ozone.

The resulting models will be utilized to control for meteorological factors and identify trends in background and locally contributed ozone over the past 10 years. In addition to this primary result, we also hope to validate our reasoning and methodology for separating peak ozone into background and locally contributed components, and to demonstrate some predictive skill for model use in forecasting day-to-day exceedances and ozone action days.