Colloquium

There are two general classes of particulate matter source apportionment methods, one using receptor-based and the other using emissions-based models. Their strengths and weaknesses are complimentary. This has two implications. First, if one can develop hybrid methods (taking the best of both, let's hope), one can make a major step towards developing source apportionments with greater confidence. Second, if results of the two can be compared and reconciled, the results should also be more robust. Here, emissions-based modeling will be the focus, emphasizing the current state of the models, recent performance evaluations, and source apportionment methods. Analyses of recent studies suggest that the performance of emissions-based PM models are improving significantly. However, significant uncertainties still remain due to emissions and meteorological inputs. A second aspect will be comparison of emissions-based and receptor modeling source apportionments, and the implications. In this regard, CMAQ, PMF and CMB (with and without using molecular markers) have been applied to receptors in Atlanta using detailed data from the Atlanta Supersite, SEARCH and ASACA. The comparisons of the results suggest that there are significant uncertainties left to resolve. Future source apportionment studies should concentrate on understanding and reconciling the differences. As part of this, more uncertainty analysis is needed for the various methods.