Statistical analysis of atmospheric polychlorinated biphenyl concentrations at two urban locations

Atmospheric deposition of persistent organic pollutants (POPs) such as polychlorinated biphenyls (PCBs) is an important source of contamination to many U.S. water bodies, including the Delaware River and Lake Michigan. In this thesis, data on concentrations of PCBs in gas and aerosol phases in Chicago, IL and Camden, NJ were analyzed to identify factors that affect atmospheric PCB concentrations in urban areas. In order to investigate factors driving atmospheric PCB concentrations in Camden, NJ, concentrations of gas and particle-phase PCBs, meteorological parameters (temperature, barometric pressure, wind speed, wind direction, relative humidity, and solar radiation) and three criteria pollutants (ozone, carbon monoxide, and nitrogen oxides) measured during 2000-2003 were analyzed by multi-parameter linear regression. The well-known temperature dependence of gas-phase PCBs was evident (average R2 = 0.62; p < 0.05). Gas phase PCB concentrations displayed weak but significant relationships with ozone, wind speed, relative humidity and solar radiation (p < 0.05). When paired with temperature, only wind speed, relative humidity and/or solar radiation produced significant relationships (p < 0.05). Solar radiation and humidity are themselves negatively correlated (p < 0.05), with relative humidity seeming to be the stronger of the two predictor variables. An advanced factor analysis method, Positive Matrix Factorization (PMF), was used to identify the dominant source types of polychlorinated biphenyls (PCBs) in the atmosphere of Chicago, IL. PMF identified 6 factors. When compared to Aroclor compositions, only factor 1 (which comprised 23% of the PCB mass in the data set) was clearly identifiable as Aroclor 1242. Although PCB levels in most environmental compartments are declining gradually, with half lives on the order of 5 to 20 years, atmospheric PCB concentrations in Chicago displayed a marked increase starting around 2005. Results demonstrate this increase was due to factor 4 (which appears to be a mixture of Aroclors 1248/1254, R2 = 0.674), and factor 5, which resembled Aroclor 1254 (R2 = 0.731). Factors 4 and 5 comprised 24% and 16% of the total PCB mass in the data set, respectively. Factors 3 and 6 displayed an exponential decay with half lives of 3.0 and 5.8 years, respectively.M.S.Includes bibliographical referencesby Dawn M. Caci