![[logo] US EPA](https://webarchive.library.unt.edu/eot2008/20081107035840im_/http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
Final Report: Source Apportionment of Exposure to Toxic Volatile Organic Compounds
EPA Grant Number: R826788Title: Source Apportionment of Exposure to Toxic Volatile Organic Compounds
Investigators: Milford, Jana B. , Miller, Shelly
Institution: University of Colorado at Boulder
EPA Project Officer: Katz, Stacey
Project Period: October 1, 1998 through September 30, 2000 (Extended to June 30, 2001)
Project Amount: $129,695
RFA: Urban Air Toxics (1998)
Research Category: Air Quality and Air Toxics
Description:
Objective:The overall objective of this research was to estimate the contributions that various sources make to human exposure to toxic volatile organic compounds (VOCs) using receptor-oriented source apportionment techniques. The study sought to distinguish between indoor and urban area sources. The study also tested multivariate techniques for extracting source profiles from exposure data, along with mass balance methods for apportioning source contributions. An additional objective of the research was to demonstrate methods for evaluating the source apportionment results using simulated exposure data and questionnaire responses from personal exposure studies.
Summary/Accomplishments (Outputs/Outcomes):In this study, data from the Total Exposure Assessment Methodology (TEAM) studies conducted from 1980 to 1987 in New Jersey (NJ) and California (CA) and from the 1987-1990 California Indoor Exposure study were analyzed using four receptor-oriented source apportionment models. The models were applied to personal exposure concentrations for 14 VOCs reported for NJ participants and 17 VOCs reported for CA participants. The models used were chemical mass balance (CMB), principal components analysis/absolute principal component scores (PCA/APCS), positive matrix factorization (PMF), and graphical ratio analysis for composition estimates/source apportionment by factors with explicit restriction (GRACE/SAFER), incorporated in the UNMIX model. The CMB and PMF models also were applied to outdoor concentration data. A portion of the TEAM data were analyzed using the indoor air mass balance (IAMB) technique. Finally, to compare them more thoroughly, the CMB, PMF, PCA/APCS, and GRACE/SAFER models were applied to a simulated data set of personal exposure concentrations that was developed for this study.
To test the receptor models, simulated personal exposure data were generated using Monte Carlo sampling with known source contributions and profiles. The PMF model appeared best able to extract the major source profiles used to create the simulated data set. It was difficult, however, for PMF to extract sources that are similar?in this case, the aromatic sources auto exhaust, ETS, and gasoline vapors. It appears that PMF extracted one factor to represent the sum of these three sources. The CMB model similarly had problems with colinearity, tending to produce large, negative source contribution estimates if profiles for more than one of these three sources were included. PCA/APCS and UNMIX identified similar factors to each other, with the addition of a separate toluene factor identified with UNMIX. Both of these models were sensitive to correlations in the simulated data that were introduced by common meteorological effects on outdoor sources. Additionally, sources that contributed less than 5 percent to the total exposure were not identified as factors by any of the models.
Application of the four receptor models to the TEAM and California Indoor Exposure data suggested that in the late 1980s, important sources of personal exposure to toxic VOCs included an aromatics source resembling automobile exhaust, gasoline vapors, or ETS; a TCA source such as solvent or insecticide use; a PDB source such as mothballs; PRC from dry cleaning; and in CA only, higher alkanes and xylenes from building materials. Tap water also was tentatively identified as a source of exposure, but the corresponding profiles and source contribution estimates differed more across models than those for the other sources. As with the simulated data, the models were unable to distinguish between automobile exhaust, gasoline vapors, and ETS. The aromatics-dominated source appeared to be a major contributor to exposures, however, so it is important for measurements of personal exposure to try to include tracer compounds to help differentiate between these three sources.
The PMF and CMB models also were applied to outdoor concentration data for NJ and CA. The results suggest that an aromatics source and wastewater emissions were the major contributors to the sum of the outdoor concentrations for NJ. For CA, an aromatics source and either an urban air or office air profile accounted for the majority of the total outdoor concentrations. PMF was also used in its three-way mode to analyze the subset of study participants in both states for which both personal and corresponding outdoor concentrations were measured. Jointly analyzing the personal exposure and outdoor concentrations was helpful in delineating contributions from outdoor sources versus contributions from personal activities or indoor air.
Of the four receptor models compared here, PMF and UNMIX showed the closest agreement, extracting six similar factors from the NJ personal exposure data and four similar factors from the CA data. Between the two, PMF has the advantage of providing more complete information on model performance than the current version of UNMIX. The lack of a non-negativity constraint is a significant limitation of both PCA/APCS and CMB. Nevertheless, results from CMB were useful in helping to interpret the factors extracted by the other models.
Journal Articles on this Report: 3 Displayed | Download in RIS Format
| Other project views: | All 7 publications | 3 publications in selected types | All 3 journal articles |
| Type | Citation | ||
|---|---|---|---|
|
|
Anderson MJ, Daly EP, Miller SL, Milford JB. Source apportionment of exposures to volatile organic compounds: II. Application of receptor models to TEAM study data. Atmospheric Environment 2002;36(22):3643-3658. |
R826788 (Final) |
|
|
|
Larson TV, Gould T, Simpson CD, Liu L-JS, Claiborn C, Lewtas J. Source apportionment of indoor, outdoor, and personal PM2.5 in Seattle, Washington, using positive matrix factorization. Journal of the Air & Waste Management Association 2004;54(9):1175-1187. |
R826788 (2000) R826788 (Final) R826371C004 (Final) R827355 (2004) R827355 (Final) R827355C003 (Final) R827355C008 (2003) R827355C008 (Final) R827355C010 (Final) |
|
|
|
Miller SL, Anderson MJ, Daly EP, Milford JB. Source apportionment of exposures to volatile organic compounds. I. Evaluation of receptor models using simulated exposure data. Atmospheric Environment 2002;36(22):3629-3641. |
R826788 (Final) |
|
indoor air, exposure, VOCs, toxics. , Toxics, ENVIRONMENTAL MANAGEMENT, Air, Health, Risk Assessment, indoor air, Risk Assessments, HAPS, air toxics, area source program, Chloroform, exposure assessment, urban air pollutants, airborne urban contaminants, 1, 4-Dioxane (1, 4-Diethyleneoxide), Chlorobenzene, indoor air quality, Volatile Organic Compounds (VOCs), chemical composition, hazardous air pollutants (HAPs), human health risk, 1, 4-Dichlorobenzene(p), air pollutants, source apportionment, 1, 1, 2, 2-Tetrachloroethane, air pollution, Bromoform, carbon tetrachloride, Ethylene dibromide (Dibromoethane), modeling, o-Xylenes, Xylenes (isomers and mixture), positive matrix factorization, Benzene (including benzene from gasoline), human exposure
Progress and Final Reports:
2000 Progress Report
Original Abstract