Source apportionment of ambient PM2.5 at five Spanish centres of the European Community Respiratory Health Survey (ECRHS II)

Fine particulate matter (PM2.5) was sampled at 5 Spanish locations during the European Community Respiratory Health Survey II (ECRHS II). In an attempt to identify and quantify PM2.5 sources, source contribution analysis by principal component analysis (PCA) was performed on five datasets containing elemental composition of PM2.5 analysed by ED-XRF. A total of 4-5 factors were identified at each site, three of them being common to all sites (interpreted as traffic, mineral and secondary aerosols) whereas industrial sources were site-specific. Sea-salt was identified as independent source at all coastal locations except for Barcelona (where it was clustered with secondary aerosols). Despite their typically dominant coarse grain-size distribution, mineral and marine aerosols were clearly observed in PM2.5. Multi-linear regression analysis (MLRA) was applied to the data, showing that traffic was the main source of PM2.5 at the five sites (39-53% of PM2.5, 5.1-12.0 is a subset of gm-3), while regional-scale secondary aerosols accounted for 14-34% of PM2.5 (2.6-4.5 is a subset of gm-3), mineral matter for 13-31% (2.4-4.6 is a subset of gm-3) and sea-salt made up 3-7% of the PM2.5 mass (0.4-1.3 is a subset of gm-3). Consequently, despite regional and climatic variability throughout Spain, the same four main PM2.5 emission sources were identified at all the study sites and the differences between the relative contributions of each of these sources varied at most 20%. This would corroborate PM2.5 as a useful parameter for health studies and environmental policy-making, owing to the fact that it is not as subject to the influence of micro-sitting as other parameters such as PM10. African dust inputs were observed in the mineral source, adding on average 4-11 is a subset of gm-3 to the PM2.5 daily mean during dust outbreaks. On average, levels of Al, Si, Ti and Fe during African episodes were higher by a factor of 2-8 with respect to non-African days, whereas levels of local pollutants (absorption coefficient, S, Pb, Cl) showed smaller variations (factor of 0.5-2).