Abstract
The use of high time resolution datasets of aerosol organic and inorganic species as input for receptor models poses a number of challenges. The estimation of uncertainties differ between different analytical methods and the number of chemical species may considerably vary among the different techniques. In this study, an approach to harmonise the uncertainties of different online datasets for their combined use in source apportionment with positive matrix factorization (PMF) is presented. The concentration of inorganic ions, organic fragments and trace elements were measured in a Po Valley background site using offline and online methods. Six-hour PM 2.5 samples were collected on filters and chemical analyses were carried out offline. Parallel hourly online measurements were made using the Xact 625 (CES LLC) XRF analyser and the Q-ACSM (Aerodyne Research Inc.) spectrometer. Online and offline methods produced comparable results for the major chemical component and some trace elements, while others (Ba, Ni, As and Se) showed limited comparability between the two methods. To ensure the consistency of the final PMF results, a multistep approach was adopted. In the first step PMF was run with only the offline dataset, in the second step only the online organic data were used and in a third step the PMF run was executed using only the online inorganic species. In the first three steps running PMF with homogeneous data made it possible to identify the main sources and produce chemical profiles to be used as internal reference for the final fourth step in which all the online species (major inorganic ions, m/z of organic fragments and trace elements) were combined. The sources of the final solution were developed using internally consistent chemical profiles and those from the literature and were validated by analysing the source diurnal variations and by comparison with external tracers. The sources identified were: biomass burning, aged biomass burning, secondary ammonium nitrate and ammonium sulphate, traffic, steel industry and waste thermal treatment. The source profiles with a large set of organic and inorganic species (87) and associated source diurnal variations resulting from this study are expected to serve as reference for future studies.
Highlights
Ambient air particulate matter (PM) impacts both human health (Lelieveld et al, 2015) and the global climate (Baker and Peter, 2008; Leibensperger et al, 2012; Myhre et al, 2013)
The objective of this study is to develop a methodology for the combined use of hourly resolved organic and inorganic datasets with heterogeneous uncertainties for their use in source apportionment, in order to find an approach to deal with such issues
The online and offline elemental analyses were carried with two different techniques particle induced X-ray emission (PIXE)/particle induced gamma-ray emission (PIGE) and XRF, respectively
Summary
Ambient air particulate matter (PM) impacts both human health (Lelieveld et al, 2015) and the global climate (Baker and Peter, 2008; Leibensperger et al, 2012; Myhre et al, 2013). Particles with aerodynamic diameter smaller than 2.5 μm (PM2.5) are associated with “all-cause” lung cancer and cardiopulmonary mortality (Pope et al, 2002). Ambient PM originates from several sources that are either directly emitted into the atmosphere or formed from precursor gases by different mechanisms, such as nucleation and condensation (Kulmala et al, 2004). In order to help establishing mitigation strategies concerning PM, sources of pollution should be identified and apportioned. Chemical markers may help with the identification of certain sources, such as biomass burning, sea spray and dust particles (Guieu et al, 2002; Ovadnevaite et al, 2012; Simoneit et al, 1999), the general case is that each source has a distinct chemical profile and time trend that cannot be directly distinguished from co-emitted sources. The most common RMs are based on factor analysis of observed variables weighted by their uncertainty (Hopke, 2009; Watson et al, 2008)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
