Sources and temporal variations of constrained PMF factors obtained from multiple-year receptor modeling of ambient PM2.5 data from five speciation sites in Ontario, Canada

•The coefficient of variation of imposed constraints was used to select the right post-PMF analyses.•Post-PMF work used trajectory ensembles & surface winds conditional probabilities.•Temporal trends of PMF factors showed time-based anthropogenic & natural influences.

The Canadian National Air Pollution Surveillance (NAPS) network operates five fine particulate matter (PM2.5) speciation sites in Ontario. Data from 2005 to 2010 from these sites were subjected to constrained positive matrix factorization (PMF) to obtain factors of particulate matter pollution. Eight factors were found to be common across the sites. These constrained factors were particulate sulfate and nitrate, aged combustion aerosols, biomass/wood combustion, non-ferrous metal smelting, crustal matter, steel works, and road salt. Other factors found at these sites were largely related to local sources in their respective neighbourhoods. Diagnostic results from constraining the PMF factors could be used to select the appropriate post-PMF analyses as they reveal the factors that were expected to have regional or local sources. Interpretation of the spatial trends of factors was guided by the extent of similarity of target ratios across the five sites as determined by their coefficient of variation (C.V.). On average, particulate sulfate had PM2.5 mass contributions that ranged from 7% (Wallaceburg) to 36% (Simcoe) and nitrate species ranged from 7% (Simcoe) to 23% (Windsor). A clear low rural – high urban divide for particulate nitrate was also observed. These non-uniform impacts across the sites indicate that a single representative location is not suitable for studying PM2.5 in Ontario. Temporal variations showed greater abundance of particulate sulfate in the summer and particulate nitrate in the winter respectively. Yearly median values of the sulfate factor that is largely energy-related have been declining for most sites since peaking in 2007. A trajectory ensemble model, simplified quantitative transport bias analysis (sQTBA) was used to identify potential source regions of the constrained sulfate and nitrate factors while radial plots of the conditional probability function (CPF) were used to explore the more locally-sourced factors. The US Midwest, east coast and Canada's west were dominant regional sources.