Using Factor Analysis to Attribute Health Impacts to Particulate Pollution Sources

Abstract

recently reported results of applying factor analysis to data taken in six cities from 1979 to 1988, identifying airborne particle sources potentially affecting daily mortality. These authors sought relationships between source groups and risk measures using source tracer elements, Se (coal combustion), Pb (light-duty motor vehicle sources), and Si (crustal—soil dispersion). Combined data analyses of this kind may overlook the complexity of source contributions, which have common tracer elements. In one of the cities, Boston, for example, the authors found coal combustion was an important source of mortality risk. For the city of Boston, the authors attribute coal combustion largely to distant upwind regional sources. The emphasis on coal combustion is confounded by the presence of major local sources of residual oil combustion, which contribute V, Se, and S (sulfur as sulfate) to the source apportionment. Evaluation of the source identification using single-element tracer analysis indicates that the detailed chemical composition or profile of major local sources needs to be taken into account in these investigations to minimize misclassification of airborne particle sources with potential adverse health effects.

This study was sponsored in part by the U.S. Department of Energy, Office of Biological and Environmental Research, grant DE-FG03-01ER63104. The opinions expressed in this article are solely those of the authors, and do not necessarily reflect the views of the Department of Energy. We acknowledge the valuable assistance of Dr. Roger McClellan in an early review of this article, and in acquisition of certain references.

Notes

* The assumption is made that the S/Se ratio remains approximately constant during transport of pollutants great distances. That is, deposition removal of the PM_2.5 components is the approximately the same.

* Laden et al. (2000) also consider a manganese (Mn) factor for defining residual oil combustion in Madison, WI, and St. Louis, MO. They derive a smaller coefficient using Mn as the source factor (5.6% for Mn, vs. 27.3% using V as the factor for residual oil use). However, there appears to be virtually no residual oil used in Madison based on ambient data (V concentration of 0.1 ng/m³compared with 23.2 ng V/m³ in Boston), and little in St. Louis (2.0 ng V/m³). Furthermore, there are several industrial facilities, including secondary metal processing, within 7 km of the sampling site in St. Louis (Spengler & Thurston, 1983), which confound the interpretation of the Mn and V factors.

* According to Energy Information Administration (EIA) data in 1980, about 94% of combined residual oil and coal energy used for electricity generation in Massachusetts came from residual oil. In 1985, after partial conversion to coal of two plants outside of Boston, about 67% of the combined energy expended was from residual oil.

* In fact, some of the secondary sulfate would be lower molecular weight ammonium bisulfate and sulfuric acid. We make the assumption that all of the sulfate would be ammonium sulfate only to calculate a maximum concentration of ambient sulfate, under the assumption of an S/Se ratio of about 1700.

* Between 1979 and 1986, the oil-burning Brayton Point power plant was partially converted to coal fuel (1980/1981). This plant is approximately 60 km south of Watertown. In addition, the Salem Harbor also was partially converted to coal in 1982. This plant is about 25 km north/northeast of Watertown. Both of these plants continued to burn residual oil during the sampling period.

† In the fall, the wind blows from the east about 5% of the time with half of the winds speeds of 3–5.1 m/s, with the rest between 1.5 and 8.2 m/s. For the spring months, east winds blow about 8% of the time with roughly half the wind speeds in the 3–5.1 m/s range and half in the 5.1–8.2 m/s range. During the summer, east winds blow about 7% of the time, with winds about 60% in the range of 3.1–5.1 m/s, and the remainder between 1.5 and 8.2 m/s. In winter, east winds occur about 2.5% of the time.