Comparison between Conditional Probability Function and Nonparametric Regression for Fine Particle Source Directions

Abstract

The objective of this study is to examine the use of conditional probability function (CPF) and nonparametric regression (NPR) to identify directions of PM2.5 (particulate matter ⩽2.5 μm in aerodynamic diameter) sources using data collected from multiple monitoring sites across the US NPR has been used on cyclohexane data from Houston, TX and correctly showed the direction of the source. In recent source apportionment studies using positive matrix factorization (PMF), ambient PM2.5 compositional data sets from 24-h integrated samples including eight individual carbon fractions collected at four monitoring sites, Atlanta, GA, Washington, DC, Brigantine, NJ, and Seattle, WA, were analyzed identifying 10–11 sources. To analyze local point source impacts from various wind directions, CPF and NPR were calculated using the source contributions estimated from PMF coupled with wind direction measured on site. The comparison between CPF and NPR demonstrated that both methods agreed well with the locations of known local point sources. CPF was simpler and easier to calculate than NPR. In contrast, NPR provided PM2.5 concentrations and associated uncertainties. This study indicates that both methods can be utilized to enhance source apportionment study of ambient PM2.5.