Identification of PM sources by principal component analysis (PCA) coupled with wind direction data

The effectiveness of combining principal component analysis (PCA) with multi-linear regression (MLRA) and wind direction data was demonstrated in this study. PM data from three grain-size fractions from a highly industrialised area in Northern Spain were analysed. Seven independent PM sources were identified by PCA: steel (Pb, Zn, Cd, Mn) and pigment (Cr, Mo, Ni) manufacture, road dust (Fe, Ba, Cd), traffic exhaust (P, OC + EC), regional-scale transport (NH4+, SO42-, V), crustal contributions (Al2O3, Sr, K) and sea spray (Na, Cl). The spatial distribution of the sources was obtained by coupling PCA with wind direction data, which helped identify regional drainage flows as the main source of crustal material. The same analysis showed that the contribution of motorway traffic to PM10 levels is 4–5 μg m−3 higher than that of local traffic. The coupling of PCA-MLRA with wind direction data proved thus to be useful in extracting further information on source contributions and locations. Correct identification and characterisation of PM sources is essential for the design and application of effective abatement strategies.