Quantitative assessment of the variability in chemical profiles from source apportionment analysis of PM10 and PM2.5 at different sites within a large metropolitan area

Abstract

The study aims to assess the differences between the chemical profiles of the major anthropogenic and natural PM sources in two areas with different levels of urbanization and traffic density within the same urban agglomeration. A traffic site and an urban background site in the Athens Metropolitan Area have been selected for this comparison. For both sites, eight sources were identified, with seven of them being common for the two sites (Mineral Dust, non-Exhaust Emissions, Exhaust Emissions, Heavy Oil Combustion, Sulfates & Organics, Sea Salt and Biomass Burning) and one, site-specific (Nitrates for the traffic site and Aged Sea Salt for the urban background site). The similarity between the source profiles was quantified using two statistical analysis tools, Pearson correlation (PC) and Standardized Identity Distance (SID). According to Pearson coefficients five out of the eight source profiles present high (PC > 0.8) correlation (Mineral Dust, Biomass Burning, Sea Salt, Sulfates and Heavy Oil Combustion), one presented moderate (0.8 > PC > 0.6) correlation (Exhaust) and two low/no (PC < 0.6) correlation (non-Exhaust, Nitrates/Aged Sea Salt). The source profiles that appear to be more correlated are those of sources that are not expected to have high spatial variability because there are either natural/secondary and thus have a regional character or are emitted outside the urban agglomeration and are transported to both sites. According to SID four out of the eight sources have high statistical correlation (SID < 1) in the two sites (Mineral Dust, Sea salt, Sulfates, Heavy Oil Combustion). Biomass Burning was found to be the source that yielded different results from the two methodologies. The careful examination of the source profile of that source revealed the reason for this discrepancy. SID takes all the species of the profile equally into account, while PC might be disproportionally affected by a few numbers of species with very high concentrations. It is suggested, based on the findings of this work, that the combined use of both tools can lead the users to a thorough evaluation of the similarity of source profiles. This work is, to the best of our knowledge, the first time a study is focused on the quantitative comparison of the source profiles for sites inside the same urban agglomeration using statistical indicators.