Mining and modeling for a metropolitan Atlanta ozone pollution decision-making framework

In this paper, we present a Decision-Making Framework (DMF) for reducing ozone pollution in the metropolitan Atlanta region. High ground-level concentrations of ozone continue to be a serious problem in several US cities, and Atlanta is one of the most serious of these cases. In contrast to the "trial and error" approach utilized by state government decision-makers, our DMF searches for dynamic and focused control strategies that require a lower total reduction of emissions than current control strategies. Our DMF utilizes a rigorous stochastic dynamic programming formulation and includes an Atmospheric Chemistry Module to represent how ozone concentrations change over time. This paper focuses on the procedures within the Atmospheric Chemistry Module. Using the US EPA's Urban Airshed Model for Atlanta, we use mining and metamodeling tools to develop a computationally efficient representation of the relevant ozone air chemistry. The proposed approach is able to effectively model changes in ozone concentrations over a 24-hour period.