Intensive optical parameters of pollution sources identified by the positive matrix factorization technique

Abstract A new methodology based on optical parameters from integrating nephelometer measurements and chemically speciated PM10 mass concentrations, to associate intensive optical parameters with pollution sources identified by the Positive Matrix Factorization (PMF) technique, is presented. PM10 samplings and integrating nephelometer measurements at 450, 525, and 635 nm, co-located in space and time, were performed from November 2011 to November 2012. The PM10 samples were chemically characterized for 16 species, including ions (Na+, NH4+, K+, Mg2+, Ca2+, Cl−, NO3−, and SO42−), metals (Al, Cd, Cu, Fe, Mn, and Ti), OC, and EC. The scattering σs and backscattering βs coefficients at 450, 525, and 635 nm, and the PM10 chemically speciated data were used as input of the PMF model. Traffic (TRA, 28.3%), Biomass Burning and Nitrates (BBN, 27.4%), Soil Dust (SDU, 14.7%), ammonium Sulphate (SUL, 17.0%), and Aged Sea-salt (ASS, 12.6%) were the identified pollution sources, according to the PM10 mass apportionment, which did not show any significant difference in terms of source assignment and contribution, with respect to the solution without optical variables. The possibility of retrieving intensive optical parameters associated with the pollution sources from the related spectrally resolved σs and βs values is the main feature of the proposed approach. The mass scattering efficiency (ΣPM10), the scattering Angstrom exponent (A), the spectral curvature of the scattering Angstrom exponent (ΔA), and the asymmetry parameter (g) were the main intensive parameters calculated at different wavelengths or wavelength pairs to characterize the identified pollution sources. ΣPM10 and g at 450 nm, A(450, 635 nm) and ΔA were equal to 3.4 m2 g−1, 0.57, 0.96, and 0.54 for the TRA-source, to 5.0 m2 g−1, 0.58, 1.57, and − 0.06 for the BBN-source, to 5.0 m2 g−1, 0.67, 1.54, and 0.24 for the SUL-source, and to 0.6 m2 g−1, 0.33, −0.65, and 0.12 for the ASS-source, respectively. The analysis of monitoring days with a prevailing pollution source and the comparison of the paper's results with literature values have demonstrated the reliability of the used methodology.