PM2.5 source apportionment in Seoul, Korea: a comparison of PMF and SMP receptor modeling results

Abstract

Source apportionment results of particulate matter with an aerodynamic diameter smaller than 2.5 mu m (PM2.5) from two multivariate receptor models (positive matrix factorization, PMF; solver for mixture problem, SMP) were compared to assess the comparability of the results in view of model performance, source identification and source quantification. Two receptor models applied the same filter samples of PM2.5 collected between October 2012 and September 2013 in Seoul, Korea, not only reproduced well the measured PM2.5 mass concentrations, especially the SMP model (R-2 = 0.99, bias = - 0.02 mu g m(-3)), but also identified the four major PM2.5 contributors: biomass burning (PMF: 12.0%, SMP: 13.2%), vehicles (PMF: 18.9%, SMP: 10.2%), road dust (PMF: 5.0%, SMP: 11.2%) and secondary aerosol sources (PMF: 51%, SMP: 42.5%). The large sulfate fraction in the source profile is thought to be the high vehicle fraction in PMF. The major species constituting the secondary aerosols are OC (PMF: 11.7%, SMP: 16.2%), sulfate (PMF: 21.2%, SMP: 15.3%), nitrate (PMF: 29.4%, SMP: 44.1%), and ammonium (PMF: 16.7%, SMP: 16.9%). The difference in the contribution of nitrate and sulfate between the two models underscores that multiple receptor models may need to be applied to the same dataset to better understand their sources in polluted areas, particularly reducing secondary aerosols.