Electronic Theses and Dissertations

HUMAN EXPOSURE TO AIR TOXICS IN URBAN ENVIRONMENTS: HEALTH RISKS, SOCIODEMOGRAPHIC DISPARITIES, AND MIXTURE PROFILES

Zhuqing Xue

Data is provided by the student.

Abstract

Exposure to air toxics in urban environments may be of significant health concern because populations and emission sources are concentrated in the same geographic area. The overall objective of this study is to characterize the sources, variations, and mixture profiles of ambient air toxics in urban environments, and examine the sociodemographic disparities in exposures to air toxics in a typical U.S. metropolitan area.A model-to-monitor comparison was performed to evaluate the validity of modeling air toxics data using national datasets. Modeled concentrations in the 2011 National-scale Air Toxics Assessment (NATA) moderately agreed with monitoring measurements, and a sizable portion showed underestimation. Results warranted the need for actual monitoring data to conduct air toxics exposure assessment.Air toxics samples were collected in 106 census tracts in the Memphis area in 2014, and samples were analyzed for 71 volatile organic compounds (VOCs). Ambient VOC levels in Memphis were generally higher than the national averages in urban settings, but were mostly below the reference concentrations (RfCs). Factor analysis identified 5 major sources: manufacturing processes, vehicle exhaust, industrial solvents, refrigerants, and gasoline additives. The major non-cancer risks were from neurological, respiratory, and reproductive/developmental effects. The cumulative cancer risk was 5.9 3.3 10-4, with naphthalene and benzyl chloride as risk drivers.Sociodemographic disparities in cancer risks were examined by regressing cancer risks against socioeconomic, racial, and spatial parameters at the census tract level. We conducted separate disparity analyses using modeling data from 2011 NATA and our air toxics monitoring data. Analysis using modeling data showed strong sociodemographic disparities but that using monitoring data did not show. The discrepancy brought cautions for use of modeling air pollution data in environmental disparity research.We further assessed exposure to VOCs mixtures in five typical microenvironments (MEs): home, office, vehicle cabin, gas station, and community outdoors. The multivariate analysis of variance and pairwise analysis showed VOC profiles were distinguishingly different among MEs. The classification of profiles was achieved using the random forest. We anticipate wide applications of exposure profiles in epidemiologic research of exposure to air toxic mixtures.