Daily time series measurements of elements or compounds are widely used to apportion the contribution of specific sources of particulate matter concentration in the atmosphere. We present results obtained for the urban area of Genoa (Italy) based on several hundred of PM10, PM2.5 and PM1 daily samples collected in sites with different geo-morphological and urbanization characteristics. Elemental concentrations of Na to Pb were obtained through Energy Dispersive X-Ray Fluorescence (ED-XRF), and the contributions of specific sources of particulate matter (PM) concentration were apportioned through Positive Matrix Factorization (PMF). By sampling at different sites we were able to obtain, in each PM fraction, the average and stable values for the tracers of specific sources, in particular traffic (Cu, Zn, Pb) and heavy oil combustion (V, Ni). We could also identify and quote the contamination of anthropogenic PM in "natural" sources (sea, soil dust). Sampling at several sites in the same urban area allowed us to resolve local characteristics as well as to quote average values.
IntroductionIn recent years, atmospheric aerosols have been studied extensively (Charlson et al., 1992;Harrison et al., 2001;Satheesh and Moorthy, 2005 , 2002a,b;Stieb et al., 2002;Fernandez et al., 2003).Up to now, PM concentrations have been routinely monitored. However, this level of monitoring is insufficient and a measurement of the elemental and chemical composition of PM is recommended in order to achieve a more complete picture. Indeed concentration limits have been set in Europe for some toxic elements (Pb, Ni, Cd, Hg; see the recent European Directive 2004/107/CE). Element and/or compound measurements can also help to trace specific emission patterns. Thus, the knowledge of the chemical composition of particulate matter can be used to evaluate the impacts of the various pollution sources on air quality. Several "source apportionment" strategies have been developed; receptor models (Gordon, 1988) are presently considered the most effective approach. These models usually provide three pieces of information: the number of (major) sources of particulate matter, the source profiles and the mass contribution of each source to total PM. These models single out groups of elements with correlated concentration trends, which