Sources, variability and parameterizations of intra-city factors obtained from dispersion-normalized multi-time resolution factor analyses of PM2.5 in an urban environment.

Ambient fine particulate matter (PM2.5) data of similar continuously monitored species at two air monitoring sites with different characteristics within the City of Toronto were used to gauge the intra-city variations in the PM composition over a largely concurrent period spanning two years. One location was <8 m from the side of a major highway while the other was an urban background location. For the first time, multi-time resolution factor analysis was applied to dispersion-normalized concentrations to identify and quantify source contributions while reducing the influence of local meteorology. These factors were particulate sulphate (pSO4), particulate nitrate (pNO3), secondary organic aerosols (SOA), crustal matter (CrM) that were common to both sites, a hydrocarbon-like organic matter (HOM) exclusive to the urban background site, three black carbon related factors (BC, BC-HOM at the highway site, and a brown carbon rich factor (BC-BrC) at the urban background site), biomass burning organic matter (BBOM) and brake dust (BD) factors exclusive to the highway site. The PM2.5 composition was different between these two locations, over only a 10 km distance. The sum of SOA, pSO4 and pNO3 at the urban background site averaged 57% of the PM2.5 mass while the same species represented 43% of the average PM2.5 mass at the highway site. Local or site-specific factors may be of greater interest for control policy design. Thus, regression analyses with potential explanatory, site-specific variables were performed for results from the highway site. Three model approaches were explored: multiple linear regression (MLR), regression with a generalized reduced gradient (GRG) algorithm, and a generalized additive model (GAM). GAM gave the largest fraction of variance for the locally-found factors at the highway site. Heavy-duty vehicles were most important for explaining the black carbon (BC and BC-HOM) factors. Light-duty vehicles were dominant for the brake dust (BD) factor. The auxiliary modelling for the local factors showed that the traffic-related factors likely originated along the main roadways at their respective sites while the more regional factors, - pSO4, pNO3, SOA, - had sources that were both regional and local in origin and with contributions that varied seasonally. These results will be useful in understanding ambient particulate matter sources on a city scale that will support air quality management planning.

Speciation of organic fractions does matter for aerosol source apportionment. Part 3: Combining off-line and on-line measurements.

The present study proposes an advanced methodology to refine the source apportionment of organic aerosol (OA). This methodology is based on the combination of offline and online datasets in a single Positive Matrix Factorization (PMF) analysis using the multilinear engine (ME-2) algorithm and a customized time synchronization procedure. It has been applied to data from measurements conducted in the Paris region (France) during a PM pollution event in March 2015. Measurements included OA ACSM (Aerosol Chemical Speciation Monitor) mass spectra and specific primary and secondary organic molecular markers from PM10 filters on their original time resolution (30 min for ACSM and 4 h for PM10 filters). Comparison with the conventional PMF analysis of the ACSM OA dataset (PMF-ACSM) showed very good agreement for the discrimination between primary and secondary OA fractions with about 75% of the OA mass of secondary origin. Furthermore, the use of the combined datasets allowed the deconvolution of 3 primary OA (POA) factors and 7 secondary OA (SOA) factors. A clear identification of the source/origin of 54% of the total SOA mass could be achieved thanks to specific molecular markers. Specifically, 28% of that fraction was linked to combustion sources (biomass burning and traffic emissions). A clear identification of primary traffic OA was also obtained using the PMF-combined analysis while PMF-ACSM only gave a proxy for this OA source in the form of total hydrocarbon-like OA (HOA) mass concentrations. In addition, the primary biomass burning-related OA source was explained by two OA factors, BBOA and OPOA-like BBOA. This new approach has showed undeniable advantages over the conventional approaches by providing valuable insights into the processes involved in SOA formation and their sources. However, the origins of highly oxidized SOA could not be fully identified due to the lack of specific molecular markers for such aged SOA.