Search the repository

A+ | A- | | SLO | ENG

Abstract: Recent studies show that combining data with different time resolutions results in more detailed and likely more accurate source apportionment (SA) (Hopke et al., 2020; Manousakas et al., 2022; Via et al., 2023). Despite the advantages, SA studies, combining 24-hour filters and highly time-resolved analyses, are limited. This research aims to compare on- and off-line SA results in a pre-Alpine valley with a variety of sources. A comprehensive SA study of the local and regional PM sources and their size-segregated elemental fraction will be presented. PM10 was sampled daily on quartz filters from November 2020 to November 2021 and analyzed for a large array of chemical species. Equivalent black carbon (eBC) measurements were taken with the Aethalometer AE33. Hourly elemental PM10 and PM2.5 composition was measured in parallel with two Xact 625i from February until May 2021. Positive Matrix Factorization (PMF) was used to determine the sources of PM10 (off-line PMF) with 24-hour time resolution, and of the elemental fraction of PM (on-line PMF) with 1-hour time resolution for the overlapping time period. For off-line PMF, chemical species together with the source-specific eBC (Sandradewi et al., 2008) were used. For on-line PMF, two analyses were performed on the elemental composition: using PM10, and using a combination of PM2.5 and PMcoarse (PM10-PM2.5). The final off-line PM10 PMF results were compared to on-line elemental PM PMF and assessed accordingly. A combination of the different PMF models with various instrument data resulted in improved SA in terms of the number of identified sources and their uncertainties. The off-line PMF resolved ten PM10 sources (Fig. 1, left), while the on-line elemental PM10 and the combined elemental PM2.5+coarse PMF recognized 6 and 7 factors, respectively (Fig. 1, right). Due to large number of samples, we were able to separate between the sources with strong seasonality and sources featuring stability throughout the year. The study disclosed two rarely encountered factors, i.e., chloride-rich (chlorine-rich in case of on-line PMF) and cement dust. Introduction of source-specific eBC in the off-line PMF model resulted in more stable factor solutions.The high-resolution on-line PMF enabled us to distinguish between regional and local sources (see mineral dust factors). Furthermore, the size-segregated on-line PMF provided more speciated sources (traffic separation into heavy-duty and light-duty vehicles). A very good regression (R2 > 0.7, slopes = 0.05–0.35, p < 0.001) between the compared off-line PM10 and on-line elemental PM10 (daily averages) factors’ concentrations confirmed the adequacy of the SAs. Further work involves a direct comparison of fractions of the elements in the sources’ chemical profiles. The study clearly shows the advantages and limitations of the different PMF models. A combined use of various SA approaches appears to be a promising way towards a comprehensive analysis of the PM sources in complex environments.
Keywords: positive matrix factorization, PM10 composition, metals, Xact, HVS digitel
Published in RUNG: 24.01.2025; Views: 327; Downloads: 7
Link to file
This document has many files! More...

Keywords: air pollution, extensive PM chemical characterization, positive matrix factorization (PMF), new source identification, health metric
Published in RUNG: 03.06.2024; Views: 1368; Downloads: 16
Full text (3,58 MB)
This document has many files! More...