Comparison of PM10 Sources Profiles at 15 French Sites Using a Harmonized Constrained Positive Matrix Factorization Approach

Receptor-oriented models, including positive matrix factorization (PMF) analyses, are now commonly used to elaborate and/or evaluate action plans to improve air quality. In this context, the SOURCES project has been set-up to gather and investigate in a harmonized way 15 datasets of chemical compounds from PM10 collected for PMF studies during a five-year period (2012–2016) in France. The present paper aims at giving an overview of the results obtained within this project, notably illustrating the behavior of key primary sources as well as focusing on their statistical robustness and representativeness. Overall, wood burning for residential heating as well as road transport were confirmed to be the two main primary sources strongly influencing PM10 loadings across the country. While wood burning profiles, as well as those dominated by secondary inorganic aerosols, present a rather good homogeneity among the sites investigated, some significant variabilities were observed for primary traffic factors, illustrating the need to better characterize the diversity of the various vehicle exhaust and non-exhaust emissions. Finally, natural sources, such as sea salts (widely observed in internal mixing with anthropogenic compounds), primary biogenic aerosols and/or terrigenous particles, were also found as non-negligible PM10 components at every investigated site.