Chemical composition of particulate matter

EC and OC were measured in the PM10 and PM2.5 size fractions at the Birkenes, Hurdal and Kårvatn sites, whereas the major inorganic anions (SO42-, NO3-, Cl^-) and cations (Ca²⁺, Mg²⁺, K⁺, Na⁺, NH4+) were obtained from open filter face samplers with a cut-off size exceeding 10 µm equivalent aerodynamic diameter (EAD). However, most of these species typically reside within the PM10 fraction.

Occasionally, sea salt aerosol larger than PM10 could be collected, i.e., during stormy weather conditions at Birkenes, at a southerly wind direction. The data obtained from the monitoring programme appear to be well suited for a mass closure of PM10, except that species representing mineral dust are not included. For 2014 - 2020, the PM10 filter samples collected at Birkenes were analyzed with respect to iron (Fe), which allows for calculating/estimating the mineral dust content.

Mass closure of PM2.5 and PM10-2.5 would include higher uncertainty than for PM10, as default assumptions would have to be made according to the size distribution of the inorganic species analyzed, of which the largest uncertainty would be associated with that of NO3-, thus this is not performed.

The annual mean chemical mass composition of PM10 is shown in Figure 5.3. The speciated mass explained 53 – 70% of the annual mean concentration of PM10 for the three sites, and 69 – 81% when allowing for other elements than carbon for OC and EC. The PM10 SIA fraction (and levels) was somewhat higher at Birkenes (25%) compared to Hurdal (17%) and Kårvatn (12%), reflecting the proximity of Birkenes to important SIA-precursor emission source regions in continental Europe. SO4

2-was the most abundant single (SIA) species at all sites, but only marginally higher than NO3- at Birkenes and Hurdal, whereas it was 1.5 times higher than sum of NO3- and NH4+ at Kårvatn.

The OM fraction (27-40%) (here: OM:OC = 1.7; See Yttri et al., 2007) was noticeably higher than for SIA (12-25%) at all sites, and by as much as a factor of two at Hurdal and a factor of three at Kårvatn.

The higher relative contribution of EC at Hurdal (2.0%) compared to Birkenes (1.5%) and Kårvatn (1.2%) was consistent with previous years, however, the level of EC in PM10 at Hurdal and at Birkenes was equally large for 2020, thus it was the lower mass concentration of PM10 which is explains the observed difference.

Situated approximately 20 km from the coastline, Birkenes experienced a substantial 25% sea salt aerosol contribution to PM10, which was substantially higher than at Kårvatn (14%) and at Hurdal (9.7%). Notably, 2020 was the first year where the sea salt fraction of PM10 was equally high (Birkenes) or higher (Kårvatn) than the SIA fraction.

Based on Fe measurements in PM10 at Birkenes and knowledge about mineral dust composition in Europe (Alastuey et al., 2016), an annual mean mineral dust concentration of 0.7 µg m^-3 was estimated for Birkenes for 2020, corresponding to 13% of PM10 annually. This was identical to the two previous years but noticeably higher than for 2014 - 2017 (7-10%) and allows for a more complete mass closure of PM10 of 94% for 2020; i.e., when using OM for mass closure. Mineral dust levels and its relative contribution to PM10 was elevated for most of the growing season, accounting for 8 – 18% of PM10 in April – September. Mineral dust explained 41% of PM10 in October caused mainly by a major long-range atmospheric transport episode in the first week of the month, bringing mineral dust from Central Asia to Northern Europe (see Chapter 7).

Statistically significant downward trends were observed for fractions of SO42- (-40%) and EC (-29%) at Birkenes for 2001 - 2020, whereas there was as substantial significant upward trend for the sea salt

At Hurdal, there was a statistically significant downward trend for the fraction of all SIA constituents to PM10, being somewhat more pronounced for NO3- and NH4+ (both -77%) than for SO42- (-54%). As for Birkenes, the hitherto lowest annual mean for SO42- was reported at Hurdal for 2020. A statistically significant downward trend in the EC fraction of PM10 of -41% was calculated. The OC fraction of PM10 (27%) and PM2.5 (33%) was significantly increased. We explain this by OC being dominated by natural sources not subject to abatement, whereas there at the same time has been a noticeable reduction in PM mainly driven by reduced SIA levels.

At Kårvatn, statistically significant changes were observed for the NO3- (-90%) and NH4+(-81%) fractions of PM10, with reductions comparable to that observed at Hurdal. Care should be made not to compare percentages calculated for Hurdal and Kårvatn with Birkenes, as trend calculations are made for 2001 – 2020 at Birkenes and for 2010 – 2020 at the two other sites.

Based on levoglucosan measurements (Table 4.2) we estimate that biomass burning emissions contributed 4-9% to PM annually, considering both PM10 and PM2.5. The fraction attributed to biomass burning in winter was 5-12%, whereas it was ≤1.5% in summer, considering both size fractions. Spring and fall are transition seasons with relative contributions typically lower than in winter and higher than in summer. Occasionally, high levels can be observed in these seasons, and for 2020 the highest relative contribution was calculated for fall, i.e., 6-17% when considering both size fractions. Biomass burning occasionally exceeded 30% on a weekly basis for PM10 and 40% for PM2.5, typically in the heating season, thus emissions are attributed to wood burning for residential heating. For the episode on the 2-3 of October see discussions in Chapter 7.

Table 5.3: Trends in aerosol particle species fractions of PM10 and PM2.5 mass concentration using Mann-Kendall test and Sen slope estimates, *significant level 0.05. Only statistically significant results are shown in the table.

Site Species/Fraction Year Change (%)

Birkenes SO42- to PM10 2001 – 2020 * -40%

a) Birkenes

b) Hurdal

c) Kårvatn