The State of the Art of Receptor Models Relating Ambient Suspended Particulate Matter to Sources

Abstract

Receptor models are used to determine the source contributions to ambient particulate matter loadings at a sampling site based on common properties between source and receptor. This is in contrast to a source model which starts with emission rates and meteorological measurements to predict an ambient concentration. Three generic types of receptor model have been identified, chemical mass balance, multivariate, and microscopical identification. Each one has certain requirements for input data to provide a specified output. An approach which combines receptor and source models, source/receptor model hybridization, has also been proposed, but it needs further study. The input to receptor models is obtained from ambient sampling, source sampling, and sample analysis. The design of the experiment is important to obtain the most information for least cost. Sampling schedule, sample duration and particle sizing are part of the ambient sampling design. Analysis for elements, ions, carbon, organic and inorganic compounds are included in the sample analysis design. Which sources to sample and how to sample them are part of the source sampling design. In order for receptor modeling to become a useful tool, it must be developed in four major areas: 1. General Theory - The generic types of receptor model are related to each other, but that relationship has not been generally established. A general theory of receptor models, including the input data uncertainties, needs to be constructed. 2. Validation - Simulated data sets created from known source contributors and perturbed by random error should be presented to models, and their source contribution predictions should be compared to the known contributions. Several models should be applied to the same data set and their results compared. 3. Standardization- Validated models need to be placed in standard form for easy implementation and use. 4. Documentation and Education - Users need to be informed of the powers and limitations of these models and instructed in their use.