| Air pollution is characterized by transboundary properties, dynamic processes, non-linear behavior, and long-range transport. Because of these characteristics, it is difficult to model air pollution behavior using the basic equations that represent the underlying physical transport and chemical transformation processes. As an alternative, a modeling approach is proposed in the particular case of sulfur dioxide (SO2), based on a circular and sectoral spatial representation of an airshed, and combining concepts derived from physical diffusion modeling and statistical regression modeling. An important feature of this approach is that it provides a means to construct a regionally scalable air quality model. While using data collected locally (at the airshed level) for estimation purposes, the spatial interconnection of individual airsheds provides the means to model larger-scale air pollution processes, including the national or continental scale.; The basic statistical model estimates the relationship between (1) background concentrations and pollution emissions from point and area sources and (2) the resulting concentration at a receptor site located at the center of the airshed, while accounting for pollution decay and uptake, meteorological conditions, and land cover characteristics. To empirically capture this spatial relationship, a considerable database, with extensive use of a Geographical Information System (GIS), is developed, connecting pollution emission sources, air quality monitoring stations, meteorological stations, and land uses. Because of the complex and non-linear structure of the model, an interactive grid search procedure has been designed to estimate the model, based on Ordinary Least Squares regression. The final estimated model explains about 56% of the variations in SO2 concentrations. An air quality optimization model, based on this statistical model and using linear programming, is introduced, and a case study, focusing on the state of Pennsylvania and larger-scale emitters (more than 10,000 tons of SO2 per year), is presented. The solution outlines how emission patterns change with the ambient pollution standard. The applicability of this airshed-based modeling approach for policy analysis is discussed, including (1) air quality forecasting, and (2) air quality planning. Dynamic extensions of the approach and potential data sources are also discussed. Finally, areas for further research are delineated. |
