ceptor models can adopt more sophisticated statistical techniques to derive a more complex (and possibly more accurate) description of the relationship. They have been used most successfully and most widely in estimating the relative contributions of various emission sources to measured ambient air particulate matter (PM) composition at a monitoring site. The estimates are made by relating the measured elemental and organic tracer composition of ambient PM to the known elemental composition of the sources of PM in the region. Reviews of these methods can be found in Brook et al. (2003) and the references therein.
Although receptor models do not require emission inventories, some of them (such as chemical mass balance models) require source profiles, which define the detailed elemental or organic composition of all sources. In contrast, statistical factor analysis models do not require source composition data but use variations and covariations among species over time to identify sources. If the source profiles vary among sources in the same category, space, or time, a large measurement project is required to obtain profiles that adequately characterize all the source classifications that are to be used in the model. These models have a long history of applications and have in general provided valuable information for the design of primary PM-control strategies. Recent extensions of these models allow the identification of sources of primary organic PM as well as sources of inorganic particles. Their greatest weakness is the estimation of the contribution of secondary PM. To date, a combination of receptor models has been the most successful approach for PM nonattainment areas.
Emissions-based models for O3 and CO are important tools used today in AQM to evaluate alternative emission-control strategies and estimate the amount of emission reductions required to meet a specific air quality goal or standard. The models use a mathematical representation of the relevant physical and chemical processes and then solve the governing equations (usually numerically) in time and space to determine the relationships between pollutant emissions and pollutant concentrations. Because these models require the input of pollutant emission rates, they are sometimes referred to as emissions-based models. One of the major advantages of these models for AQM is their predictive capability. Because they calculate pollutant concentrations as a function of pollutant emissions, they can be run in a prognostic mode to predict the air quality response to any hypothetical change in pollutant emissions.
Emissions-based air quality models can vary in complexity. The simplest are box models, which simulate the evolution of pollutant concentrations in an idealized well-mixed parcel of air. Dispersion models, such as