Long-term PM2.5 source analyses in New York City from the perspective of dispersion normalized PMF

Receptor models like positive matric factorization (PMF) have been widely used in atmospheric studies for source identification and quantification. However, one of the issues with ambient air pollution data used in those analyses is that the concentrations result from a combination of emission/formation rates and meteorological dispersion. Dispersion normalized PMF (DN-PMF) was recently developed to reduce the influence of dilution on source resolution. Studies have shown its capability to better reflect actual emission rates and yield more interpretable results for hourly integrated ambient data for particulate matter and particle number size distributions. In this study, conventional and DN-PMF were used to deconvolute long-term 24-h integrated speciated PM2.5 data sets, for the first time, for three urban sites in the NYC metropolitan area. DN-PMF using 24-h integrated data does not provide much improvement over DN-PMF using hourly integrated data since averaging over 24 h loses the strong diel dispersion patterns. However, it provides improved seasonal patterns for different sources that are lost with conventional PMF results. DN-PMF implements necessary corrections to the conventional PMF approach, and makes better use of the chemical monitoring data that generally have lower time resolution.