| Popis: |
Identification of dominant airborne Particulate Matter (PM) sources is essential for maintaining high air quality standards and thus ensuring a good public health. In this study, different approaches were applied for source apportionment of three PM fractions (PM1, PM2.5 and PM10) at the outdoor of 14 schools of a coastal city with a significant land use interweaving such as Tarragona (Spain). PM were collected in 24h-quartz microfiber filters in two seasonal campaigns (cold and warm), together with nine local potential sources, so a total of 84 samples were chemically, mineralogically, and isotopically characterised. Source apportionment was assessed by (i) main chemical components, (ii) Principal Component Analysis (PCA), (iii) dual C and N isotope approach, and (iv) a Bayesian isotope mixing model. When chemical concentrations were grouped into marine, crustal, secondary inorganic aerosols and organic matter + elemental carbon categories, the unaccounted component reached 45% of PM mass. The PCA allowed to identify also traffic and industrial contributions, reducing the unaccounted mass to about 25%. Adding δ13C and δ15N values, secondary organic aerosol could be estimated and a continuous contribution of diesel combustion was identified together with a remarkable use of natural gas in winter. Isotopic values were better understood when considering air masses back trajectories and a possible long-distance contribution from coal-fired electric generating units (EGUs). Finally, using Bayesian dual isotope mixing models, the unaccounted PM mass was reduced up to 5% when adding these EGUs to marine-carbonate related, road traffic, domestic heating, waste incinerator and livestock waste contributions. The added value of the dual isotope approach combined with a Bayesian isotope mixing model, in comparison with conventional chemical approaches, was thus demonstrated for PM source apportionment in an urban and industrial site where many sources and processes converge and can then be applied to other complex cities. |
Full Text from ScienceDirect