Air pollutants in the particulate (PM2.5 species, PAHs) and gaseous phases (VOCs) collected between 2013 and 2019 once every three or six days for a period of 24 hours in an industrialized city in Ontario were analyzed to apportion their common sources. The consequences of using these species jointly for receptor modelling were assessed via combined-phase source apportionment that used the data as is, and in a protocol that factored in the potential for photochemical losses of gas-phase species. Thus, photochemically corrected initial concentrations (PIC) were calculated. Analyses of the inputs followed either with positive matrix factorization or its dispersion-normalized variant (DN-PMF). Comparisons of applying PMF to the originally observed input data (BASE) and DN-PMF on data with PIC corrections were made. When the inputs consisted only of VOCs, three factors were resolved with BASE PMF: natural gas, vehicular emissions, and industrial emissions co-emitted with summertime gasoline evaporation. A fourth factor was obtained, representing reactive VOCs when DN-PIC PMF was used. When the combined phase input data were analyzed, nine factors were resolved for both BASE and DN-PIC PMF. These factors in order of diminishing average PM mass contributions were: particulate sulphate, secondary organic aerosol (SOA), particulate nitrate (pNO3), biomass burning with natural gas, crustal matter, winter blend of gasoline, coking/coal combustion, steelmaking, and summer blend/light duty vehicular emissions. When BASE and DN-PIC PMF results are compared, the average PM mass contribution of the summer gasoline fuel factor increased from 2% in BASE case to 5%, suggesting severe underestimation of this source's contributions without DN-PIC. Also, substantial increases of reactive VOCs in the SOA factor, and PAHs with ≥four rings in the pNO3 and steelmaking factors were observed with DN-PIC PMF compared to the BASE PMF case, indicating that for SOA, reactive VOCs at this location contributed to SOA sources.
Read full abstract