Atmospheric Particulate Organic Matter

Abstract

Multivariate regression models have been developed for apportioning the contributions of emission sources to airborne particulate organic matter. Weekly samples of respirable (<3.5 µm A.D., 50% cut) suspended particulate matter were collected in New York City from January, 1978 through August, 1979. The samples were analyzed for trace metals and sulfate as well as for three fractions of particulate organic matter (POM) using sequential extraction with cyclohexane (CYC), dichloromethane (DCM) and acetone (ACE). Factor analysis was used to identify the principal types of emission sources and select source tracers. Using the selected source tracers, models were developed of the form POM = a(V) + b(Pb) + - - -, where a and b are regression coefficients determined from ambient data adjusted to constant dispersion conditions. The models for CYC and ACE together, which constitute 90% of the POM, indicate that 40% (3.0 µg/m3) of the mass was associated with oil-burning, 19% (1.4 µg/m3) was from automotive and related sources and 15% (1.1 µg/m3) was associated with soil-like particles. Comparison of the coefficients from the multiple regression analysis with available source emission data supports the validity of the models.