Receptor model based identification of PM2.5 sources in Canadian cities

Source apportionment of 24-hour integrated PM2.5 chemical speciation data, collected at five Canadian urban sites, Windsor, Toronto, Montreal, Halifax, and Edmonton was performed using the receptor model, Positive Matrix Factorization (PMF). In order to determine the influences of local and regional sources, in-depth wind direction and back trajectory analyses were performed using the conditional probability function (CPF) and the potential source contribution function (PSCF). The highest PM2.5 levels were observed in Windsor followed by Toronto and Montreal. Secondary sulfate and nitrate were the major factors contributing to the PM2.5 mass, accounting for 41% - 61% in the five sites. These secondary factors were associated with trans-boundary emissions from Ohio, Pennsylvania, and New York. An elemental carbon (EC)-rich factor was identified in Windsor, Toronto, and Montreal, characterized by distinct EC and organic carbon (OC) profiles. The EC-rich factor accounted for 6% - 19% of the total PM2.5 mass in summer and also appeared to be related to trans-boundary pollutants. The combined contributions of traffic and road dust ranged from 14% to 19%, with a portion of the nitrate factor also coming from vehicles. In Halifax, sea salt was the second strongest source, contributing 18% of the PM2.5. In Edmonton, strong correlation of volatile organic compounds with the major PM2.5 factors suggested that local industrial sources were significant sources of secondary aerosol. Further, biomass burning contributed 12% of the PM2.5 mass in Edmonton. Both local and regional sources were found to contribute at all sites. Thus, PM2.5 can be reduced at all the sites through local controls. However given the significant contribution of trans-boundary contributions to the PM2.5 mass, a substantial reduction of PM2.5 in four of the cities will also require agreements to limit the production and transport of trans-boundary pollutants. (C) Author(s) 2011. This work is distributed under the Creative Commons Attribution 3.0 License.