Resolution of PM sources using highly time resolved data and discrete particle size bins

Abstract

The goal of this project was to explore the use of particle size distribution data for distinguishing between local vs distant and controlled vs uncontrolled combustion sources. A two day period in July was selected for an apportionment study from eleven months of highly time resolved data collected at the Baltimore Supersite. Number concentration vs particle size data was partitioned into four size bins as follows: a nucleation mode (<80>nm), fresh combustion (80-300 nm), secondary aerosol (0.3-1 µm), and coarse particles (1-2.5 µm).

Two multivariate receptor model studies were completed using positive matrix factorization. The first model included size binned particle counts, the second, included binned particle mass concentrations. Using this combination of models, we identified eight factors. Three sources were identified by both model results, while additional sources not identified by number concentrations were identified when we applied the mass moments to weight the size distribution data.