Source apportionment of ambient particles: Comparison of positive matrix factorization analysis applied to particle size distribution and chemical composition data

Positive matrix factorization (PMF) method was used to identify the sources of ambient particles (PM 10) in Augsburg in winter 2006/07. The analyses were carried out separately with particulate chemical composition (PCC) data at an urban traffic site and with particle size distribution (PSD) data at an urban background site on daily and hourly base, respectively. For PCC data, six factors are identified and associated with NaCl (6.7% of PM 10), secondary sulfate (13.0%), biomass burning (13.3%), secondary nitrate (30.5%), traffic emission (16.5%) and re-suspended dust (20.0%). For PSD data, seven factors are identified and are associated with fresh and aged traffic sources, secondary aerosols, stationary combustion, nucleation particles, re-suspended dust and long range transported dust. The two traffic factors were dominated by ultrafine particles (diameter < 100 nm), and accounted for 25% and 40% of total particle number concentration (NC). Stationary combustion factor, consisting of particles around 100 nm, accounted for 26% of total NC. Re-suspended dust was mainly composed of particles with diameters > 2.5 μm. The two different approaches (PCC and PSD data) led to comparable results with strong correlations for secondary nitrate and sulfate/secondary aerosols ( r = 0.92), which are considered to origin mainly from long range transport. Traffic emissions ( r = 0.52) and re-suspended dust ( r = 0.62) showed weaker correlation due to influences of local sources at the different sites. ► 6 PM 10 factors were characterized using chemical composition (PCC) data. ► 7 factors were characterized using particle size distribution (PSD) data. ► Secondary aerosols were in good agreement between two methods. ► Sources influenced by local emissions show weaker agreement. ► Two methods with their advantages and disadvantages were compared.