A chemometric approach to determine local and regional sources of PM10 and its geochemical composition in a coastal area

The present study aims at estimating the local and external sources of PM10 in a coastal semi-rural background site of Po Valley (Italy) using experimental data and different chemometric approaches. An improved application of the factor–cluster analysis (FCA) is here presented. A factor analysis on the elemental and water soluble inorganic ions composition of PM10 was first performed, to detect the most probable sources, then the application of the multi-linear regression analysis pointed out the most important: sea-spray contributed most to PM10 mass (24%), then secondary aerosol (21%), anthropogenic local pollution (13%) and combustions (8%). Samples with similar source profiles were then grouped using cluster analysis on the source contributions. Clusters interpreted on the basis of wind data and back-trajectory frequencies indicated the most probable origins, by discriminating local generation and long-range transport processes. An indirect estimation of the influence of local and external contributions to PM10 evidenced that PM10 regulatory limits are exceeded because of local weather conditions and external transport. A discriminant analysis was also carried out to check the performance of FCA in grouping samples and extracting further information on PM10 sources