Quantifying Source Impacts on Particulate Matter and Health Outcomes: Some Problems, Some Advances, A Ways Left to Go
A.G. RUSSELL1, S. BALACHANDRAN1, D. LEE1, J. PACHON1, T. ODMAN1,G. GOLDMAN1, J. MULHOLLAND1, Y. HU1, J. SARNAT2, S. SARNAT2, M. STRICKLAND2, P. TOLBERT2
(1) Georgia Institute of Technology, Atlanta, GA USA (correspondence to email@example.com) (2) Emory University, Atlanta, GA
Abstract Number: 447
Preference: Invited Plenary Speaker
Last modified: January 5, 2010
Working Group: sq2
An estimated 800,000 people die prematurely each year due to exposure to urban particulate matter (PM), and more suffer less adverse effects. Such estimates are derived largely from epidemiologic studies linking health records to observed concentrations of PM (typically PM2.5 or PM10). When the necessary data have been available, some of those studies have been able to go further to identify specific species that appear to be most responsible for the observed health outcomes. Such studies typically do not identify the sources of the PM that may have increased impacts. Recent studies using a range of source apportionment techniques have begun to contribute that information. However, the various methods used for source apportionment can lead to differing results. Further, recent studies have identified limitations in the use of air quality model results, including both chemical transport models and receptor modelling approaches.
Here, examination of the weaknesses of using different source apportionment techniques in health studies are presented, as well as how methods can be combined to alleviate limitations. Results from two new approaches, using ensembles and multi-method optimization, suggest improved performance. The ensemble method uses results of various source apportionment techniques to develop a weighted average result, which is then used to develop spatially and temporally applicable source profiles for application over longer periods. A second approach uses chemical transport modeling (CTM) and chemical mass balance (CMB) approaches together to provide a more detailed characterization of source impacts at a receptor.
While these new approaches appear to address some of the current limitations in using source apportionment methods, there is a need for more spatially and temporally accurate source impact fields for use in exposure assessment, health analysis and air quality planning. One area of interest is assessing method uncertainty, and both the ensemble and CTM-CMB methods provide some advances in our understanding of method accuracy. One question that arises is if uncertainties in the source apportionment methods are much larger than the observations, and if this is the case, will using source apportionment results increase overall uncertainty in the health analyses being conducted. As such, uncertainties in the source apportionment results are compared to the reliability of various measurement methods, measurement method uncertainties and potential errors in exposure assessments due to spatial variability not captured by monitoring.