Aerosol source apportionment uncertainty linked to the choice of input chemical components

• DOI: 10.1016/j.envint.2024.108441
• Publication Date: 2024-01-12T15:42:53Z
• Authors: F. Amato; B.L. van Drooge; J.L. Jaffrezo; O. Favez; C. Colombi; E. Cuccia; C. Reche; F. Ippolito; S. Ridolfo; R. Lara; G. Uzu; T.V.D. Ngoc; P. Dominutti; S. Darfeuil; A. Albinet; D. Srivastava; A. Karanasiou; G. Lanzani; A. Alastuey; X. Querol
• Keywords: Apportionment; Levoglucosan

For a Positive Matrix Factorization (PMF) aerosol source apportionment (SA) studies there is no standard procedure to select the most appropriate chemical components be included in input dataset for given site typology, nor specific recommendations this direction. However, these choices are crucial final SA outputs not only terms of number sources identified but also, and consequently, contributions estimates. In fact, PMF tends reproduce PM mass measured independently introduced as total variable data, regardless percentage which has been chemically characterized, so that lack some tracers (e.g. levoglucosan) can potentially affect results whole study. The present study elaborates further on same concept, evaluating quantitatively impact lacking sources' apportionment, both contributions. This work aims provide first suitable critical analyses order reduce output uncertainty much possible, better represent commons observed many sites Western countries. To aim, we performed three sensitivity different datasets across EU, including extended sets organic tracers, cover types urban conditions (Mediterranean, Continental, Alpine) , types, fractions. Our findings reveal vehicle exhaust resulted less sensitive choice analytes, although estimates deviate significantly up 44%. On other hand, detection non-exhaust one clearly necessary analyze inorganic elements. analysing non-polar organics likely causes loss separation factors, thus obtaining unique road traffic source, provokes significant bias contribution. Levoglucosan was, cases, identify biomass burning Milan Barcelona, spite presence PAHs while case Grenoble, even discarding levoglucosan, allowed identifying BB factor. Modifying rest analytes provoke systematic underestimation SIA factors generally overestimated with respect base analysis, also ions were analysis. Trace elements shipping emissions (V Ni) industrial (Pb, Ni, Br, Zn, Mn, Cd As). When changing variables, was narrow large processes. Major trace mineral/soil factor at all cities. Biogenic SOA Anthropogenic their molecular since availability OC alone unable separate Arabitol sorbitol detecting fungal spores odd higher alkanes (C27 C31) plant debris.