The stratospheric aerosol layer changes over time. The aerosol layer becomes more pronounced after volcanic eruptions, with plume injection high into the stratosphere (Kremser et al., 2016). It plays a role in the radiative balance and thus impacts the climate (Solomon et al., 2011). An enhanced aerosol concentration leads to a larger albedo. The geoengineering community investigates the enhancement of the aerosol layer with injections of sulfur compounds into the stratosphere to partly counteract greenhouse gases related global warming (Crutzen, 2006; Schäfer et al., 2015). The stratospheric aerosol layer is often referred to as the “Junge layer” and can extend from the tropopause up to 25 km (Junge et al., 1961). The chemical composition of the stratospheric aerosol layer is dominated by sulfate (SO42-) particles, which consist mainly of pure sulfuric acid droplets, sulfuric acid with material from ablated meteoroids, or mixed organic–sulfate particles (Murphy et al., 2014; Cziczo et al., 2001; Schneider et al., 2021). During volcanic quiescent periods, precursor gases, like carbonyl sulfide (OCS), and non-volcanic sulfur dioxide (SO2), as well as tropospheric SO42- particles, preserve the stratospheric layer (Brock et al., 1995). Due to its long lifetime, OCS is vertically uplifted from the tropics into the stratosphere, and there it converts mostly through photodissociation to SO2 (Sheng et al., 2015). The SO2 chemistry and transport depends strongly on the ambient conditions. In the free troposphere and lower stratosphere, SO2 reacts predominantly with hydroxyl (OH) to form sulfuric acid (H2SO4) (English et al., 2011; Stockwell and Calvert, 1983); thus, the lifetime correlates with the OH concentration. At cold temperatures and in the presence of water vapor, the gaseous H2SO4 condenses quickly to particles (Almeida et al., 2013; Kirkby et al., 2011), thereby forming sulfate aerosol. In the boundary layer, pollution could significantly reduce the lifetime to hours (Lee et al., 2011) and, consequently, also the transport of SO2 to higher altitudes. Clouds could also limit the SO2 lifetime to hours or days (Lelieveld, 1993), as the conversion of SO2 to H2SO4 is faster in cloud droplets than in the gas phase (Seinfeld and Pandis, 2006). Nevertheless, SO2 can be transported from the planetary boundary layer (PBL) into the UTLS region via different pathways. Similar to OCS, SO2 can be vertically transported across the tropical tropopause layer (TTL) or by overshooting convection in the tropics (Fueglistaler et al., 2009) or by the transport of SO2 in a warm conveyor belt (WCB) in the midlatitudes (Arnold et al., 1997; Clarisse et al., 2011; Fiedler et al., 2009) or by transport processes connected with the Asian monsoon (Gottschaldt et al., 2017, 2018; Ploeger et al., 2017; Tomsche et al., 2019; Vogel et al., 2019; von Hobe et al., 2021).

An explosive volcanic eruption can inject huge amounts of ash, SO2, and other volcanic gases into the stratosphere and thus enhance the stratospheric aerosol layer (Kremser et al., 2016). In 1991, the volcano Mount Pinatubo (15∘ N) emitted approximately 20 Tg of SO2 and 30 Tg of aerosol (McCormick et al., 1995), which impacted the stratosphere globally. Volcanic eruptions in midlatitudes can also impact the stratosphere, e.g., Mount St. Helens (46∘ N, 0.8 Tg SO2) in 1980, but its impact vanished in about a year (Deshler et al., 2006). One recent midlatitude eruption of similar strength was the volcano Raikoke in June 2019 (48.28∘ N, Kloss et al., 2021; de Leeuw et al., 2021), which emitted approx. 1.4–1.6 Tg SO2. A further important source of stratospheric aerosol is intense wildfires, which can potentially develop pyrocumulonimbus (pyroCb) and thus transport biomass burning emissions into the UTLS (Fromm et al., 2005; Peterson et al., 2018). Moreover, air traffic is another source of particles and precursor gases in the UTLS (Lee et al., 2010; Voigt et al., 2010; Jurkat et al., 2011).

In late spring 2020, the UTLS region was probed over Europe during the BLUESKY mission, the period covered the first weeks of the coronavirus disease 2019 (COVID-19) lockdown in Europe, which caused reductions in emissions from industry, ground, and especially airborne transportation (Voigt et al., 2022). Under these conditions, we found enhanced values of SO2 in the upper troposphere (UT) and of particulate SO42- in the lowermost stratosphere (LS), which motivated us to investigate these sulfur compounds with respect to their sources and sinks.

In the following, we first present the BLUESKY mission in Sect. 2. In Sect. 3.1 we introduce the airborne measurements and in Sect. 3.2 the trace gas and particulate profiles obtained during BLUESKY and also show tropospherically and stratospherically influenced compounds in tracer–tracer correlations (Sect. 3.3). Afterwards, we focus on the SO2 profile in the UT (Sect. 4) and continue in Sect. 5 with the stratospheric sulfate aerosol. Finally, we summarize our results and give an outlook in Sect. 6.

In Fig. 6a, the median SO2 profile with shaded 25th and 75th percentiles is plotted against the flight altitude, in order to compare the data with literature values. The SO2 percentiles are calculated for 500 m bins. Similar to Fig. 3b, the median SO2 decreases with height from around 0.08 to 0.04 ppb. Previous in situ SO2 measurements at similar flight altitudes are included in Fig. 6a. Overall, the BLUESKY SO2 measurements are within the range of previous airborne studies, even though all studies are snapshots of the atmosphere of different locations on the Northern Hemisphere, different seasons, and different meteorological situations. Keeping this in mind, we will have a closer look. Williamson et al. (2021) reported low mixing ratios for the remote northern-hemispheric background over the Pacific for the upper troposphere as well as for the lowermost stratosphere during the Atmospheric Tomography Mission (ATom; 2016–2018). Speidel et al. (2007) reported higher SO2 values for the upper-tropospheric background over Europe and the eastern Atlantic in summer 2004. Jurkat et al. (2010) measured the stratospheric background over Europe in autumn 2008 in a similar range to Speidel et al. (2007). Above 12 km the BLUESKY SO2 mixing ratios agree well with the stratospheric background from Jurkat et al. (2010) and the tropospheric background from Speidel et al. (2007). Surprisingly, the BLUESKY SO2 profile slightly exceeds the previous measurements below these altitudes in the upper troposphere. The upper-tropospheric SO2 profile compares better to SO2 mixing ratios, which were associated with the SO2 background in the North Atlantic flight corridor in 1997 or 2010 (Arnold et al., 1997; Jurkat et al., 2010). Arnold et al. (1997) reported SO2 mixing ratios in the range of 0.03–0.3 ppb in October 1993 during POLINAT (Pollution from aircraft emissions in the North Atlantic flight corridor) and Jurkat et al. (2010) measured 0.09 ppb of SO2 during CONCERT in autumn 2008. Due to implementation of SO2 emission control policies, the global SO2 emissions decreased since 1980 (Hoesly et al., 2018; Aas et al., 2019); nevertheless, the sulfur content in kerosene remained unchanged (Lee et al., 2021; Miller et al., 2009); thus, the aviation-based SO2 emissions depend on the air traffic.

In 2020, a 72 % reduction of the air traffic above Europe has been reported due to the COVID-19 lockdown in comparison to the same time period in 2019 (Schumann et al., 2021a, b), hence providing a lower aviation SO2 source with respect to 2019. Compared to 2010, the 3.5 % increase in air traffic per year (Lee et al., 2011) promotes an increase by a factor of 1.5 of the 2010 air traffic for a scenario without COVID-19 restrictions in 2020 and consequently a theoretical increase of 50 % in aviation SO2 emissions for 2020, given that the sulfur content of the kerosene is unchanged (Lee et al., 2021; Miller et al., 2009). In 2020, the remaining air traffic of 28 % (in comparison to 2019) corresponds to roughly 40 % of the 2010 air traffic and might hence in part explain the SO2 mixing rations detected in the upper troposphere, which are still higher than remote background measurements (Williamson et al., 2021).

In addition, further sources could have contributed to the SO2 budget in the upper troposphere. To analyze the origin of air masses with elevated SO2, HYSPLIT back trajectories are calculated and representative examples are plotted in Fig. 5a–h. SO2 emissions from anthropogenic and natural sources in the PBL can be lifted to the UT via convection or via warm conveyor belts and transported to the measurement region. Arnold et al. (1997) reported an extended layer of enhanced SO2 with maxima of up to 3 ppb in the northeast Atlantic, which was an air mass uplifted and transported from the polluted PBL from the eastern United States. A few cases show trajectories with similar pathways, like the example in Fig. 5e/f. Nevertheless, the PBL contacts are also over the Pacific and East Asia. The latter one suits better to the findings of Fiedler et al. (2009), who observed the uplift of polluted air masses from East Asia via warm conveyor belts and upper-tropospheric long-range transport towards Europe. Further, the Asian monsoon also serves as a vertical transport pathway for emissions from the PBL up to high altitudes, where the air mass can enter the LS and horizontally be transported either eastwards (Vogel et al., 2014, 2016) or can be horizontally transported in the UT (Tomsche et al., 2019) and finally reach Europe. Similar trajectory pathways can be found for the cases in Fig. 5a/b and 5g/h. These trajectories indicate long-range transport in the UT and could have been impacted by the Asian monsoon. Generally, the trajectories with elevated SO2 (Fig. 5a–h) show lower potential temperatures in comparison to the trajectories calculated for elevated SO42- (Fig. 5i–p). Hence, long-range transport of SO2-enriched PBL air masses could have contributed to the observed BLUESKY SO2 mixing ratios in the UT. In contrast, the trajectories do not indicate local transport from the central European PBL to the UT; hence, the transport of SO2 from wildfires in Germany in May 2020 (European Commission et al., 2021) to the UT seems negligible. Even if the transport of the smoke was via self-lofting (Ohneiser et al., 2021), i.e., absorption of sunlight leads to warming of the ambient air and thus lifting of the smoke, the transport is slow and so SO2 might have already been transformed to SO42- before reaching the UTLS and does not contribute to the elevated SO2 in the UT. Moreover, the decrease of SO2 in the LS, as expected, does not support transport of SO2 beyond the UT into the LS either via convection or via warm conveyor belts. This can be confirmed by the trajectories for SO42- as they indicate long-range transport at high altitudes with negligible influence from lower altitudes. The trajectories helped identify potential source regions of SO2.

Beside the sources, also sinks of SO2 can alter the SO2 concentrations in the UTLS. Beside the conversion to H2SO4, leading to sulfate particles, SO2 is removed from the atmosphere by wet and dry deposition. SO2 can be scavenged by clouds, which lead to a significant reduction of the SO2 lifetime (Lelieveld, 1993). Van Heerwaarden et al. (2021) investigated the meteorological situation in spring 2020 and found that a stable high-pressure system over Europe led to a lower cloud fraction in comparison to the mean 2010–2019 period over Europe. This would lead to less cloud processing and reduce SO2 sinks. Furthermore, elevated humidity favors the faster conversion of SO2 to SO3 and sulfate, as water vapor enhances the potential for elevated OH concentrations (Seinfeld and Pandis, 2006). As reported by Schumann et al. (2021a, b) the UTLS was drier in spring 2020 in Europe in comparison to previous years. The median H2O profiles during BLUESKY reach only mixing ratios of up to 268 ppm in the upper troposphere (Fig. 3g), which is still within the range of typical springtime H2O mixing ratios in the upper troposphere (e.g., Hegglin et al., 2009; Kaufmann et al., 2018). But van Heerwaarden et al. (2021) found with respect to humidity and cloud cover that spring 2020 was amongst the springs with the lowest values. Thus, the lower available H2O led to lower OH concentrations during the BLUESKY period, which implies less chemical processing and hence a reduction of SO2 sinks. Less SO2 sinks could lead to an enhanced SO2 lifetime in the UTLS and thus higher SO2 mixing ratios.

In sum, the enhanced SO2 mixing ratios at cruise levels in Europe in spring 2020 can possibly be explained by a non-negligible aviation SO2 contribution, WCB, or convective transport from the boundary layer, followed by long-range transport and the prolonged SO2 lifetime caused by the unusually dry UTLS conditions. Neither the sources nor the sinks could separately explain the SO2 mixing ratios in the UTLS. Beyond that, we are not able to analyze in more detail the different amounts of the aforementioned factors and how they contribute to single flights.

As mentioned in Sect. 3.2, SO42- has a distinct profile with a steep increase at a potential temperature of 340 K, which refers here to around 11 km with respect to altitude (Fig. 6b). This altitude reflects also the thermal tropopause height during BLUESKY (Fig. 6b). Up to this altitude, the mixing ratio is rather constant, and then it increases. Between 8–11 km O3 mixing ratios are stable, and above O3 increases. The O3 mixing ratio above 120 ppb indicates stratospheric air masses as mentioned above; thus, the higher SO42- mixing ratios above 11 km can be attributed to the stratosphere and are hence associated with stratospheric aerosol. The layer between 11 and 13.5 km can be influenced from the stratosphere as well as the troposphere, as the data are averaged over a few weeks and varying meteorological conditions, which lead to a broadening of the 25th to 75th percentile range. In Fig. 4a, this layer represents the mixed layer. The SO42- correlates well with O3 for all flights, similar to the flight on 2 June 2020, presented in Sect. 3.1 (Fig. 2).

Previous studies investigated the sulfate aerosol in the UTLS region in northern-hemispheric midlatitudes (Fig. 6b). The BLUESKY mixing ratios in the UT agree well with the observations by Schneider et al. (2006) during May 2003, which were partly influenced by aircraft exhaust plumes. The BLUESKY mixing ratios are lower than the UT background reported by Schmale et al. (2010). Martinsson et al. (2001, 2005) observed significantly lower SO42- mixing ratios in the European upper troposphere and tropopause region. The SO42- profile (Morgan et al., 2009) obtained from April 2005 to September 2006 over the UK shows higher SO42- concentrations compared to the BLUESKY measurements. Morgan et al. (2009) suggest that the elevated mixing ratios in the UT are the result of regional uplift of polluted air masses during stagnant meteorological conditions over the UK.

Sulfate was measured in two previous HALO missions with the aerosol mass spectrometer C-ToF-AMS in a similar altitude range and region. During EMeRGe-EU, seven research flights were conducted in June/July 2017 above Europe at altitudes up to 10 km (Andrés Hernández et al., 2022). The SO42- mixing ratio was on average (0.04±0.01) ppb and compares well to the BLUESKY SO42- mean in the same altitude range below 10 km. The second HALO mission was CAFE-Africa in summer 2018, which reached altitudes up to 14 km. Here, only data obtained over Europe (38–57∘ N and 14∘ W–16∘ E) are used for the comparison, which include two test flights and the ferry flights (27 July, 1 August, 7 August, and 7 September 2018; Voigt et al., 2022). For the CAFE-Africa subset the thermal tropopause was slightly higher than during BLUESKY (Fig. 6b). For the altitude range 8–11 km, the SO42- mean was (0.03±0.01) ppb, similar to the BLUESKY value. Above 11 km in the lower stratosphere, SO42- rises to (0.09±0.03) ppb. Considering heights above the tropopause, i.e., with enhanced SO42- mixing ratios, the stratospheric BLUESKY SO42- concentrations are a factor of 2 to 3 higher than the observations in summer 2018.

In the following, we investigate the origin of the elevated stratospheric SO42- mixing ratios during BLUESKY. As mentioned above, we calculated HYSPLIT back trajectories for cases of elevated SO42-. In Fig. 5i–p, representative examples of trajectories are displayed. The majority indicates long-range transport at high altitudes with potential temperatures between 343 and 465 K. While the lower range of the potential temperature is associated with midlatitude tropopause height, the upper values are clearly associated with the stratosphere. As only few trajectories indicate lower potential temperatures, we assume that the majority of elevated SO42- is already in the stratosphere 360 h before sampling, and hence influence from the troposphere is negligible. One major source of SO42- in the stratosphere is volcanic eruptions. One year before BLUESKY, the volcano Raikoke on the Kuril Islands (Russia, 48.29∘ N, 153.25∘ E) in the western Pacific started to erupt on 21 June 2019 and continued for some days; it was categorized as volcanic explosivity index (VEI) ≥ 4. Several explosive eruptions emitted a dense ash and SO2 plume, which rose up to 19 and 20 km on consecutive days (Hedelt et al., 2019), thus also injecting into the stratosphere. Based on TROPOspheric Monitoring Instrument (TROPOMI) analysis, de Leeuw et al. (2021) reported that the eruption released 1.4–1.6 Tg SO2 into the atmosphere and simulated also that approximately 0.9–1.1 Tg SO2 thereof was injected into the stratosphere. Kloss et al. (2021) used satellite-based OMPS-LP (Ozone Mapping and Profiler Suite Limb Profiler) stratospheric aerosol optical depth (sAOD) to investigate the temporal evolution from before the Raikoke eruption until May 2020. Almost 1 year later, the sAOD was still higher than prior to the eruption. This suggests that elevated SO42- measured in the stratosphere during BLUESKY was partly caused by the Raikoke eruption a year earlier. The eruption of Mount St. Helens in 1980 was of comparable size, midlatitude location, and SO2 emissions (Deshler et al., 2006), and its impact was also visible for almost a year. Still, other sources cannot be completely ruled out. For example, severe wildfires in Alberta, Canada, developed pyrocumulus clouds in June 2019. The biomass burning emissions were uplifted into the lower stratosphere (Osborne et al., 2022). In July 2019, also severe fires in Siberia, Russia, impacted the OMPS-LP sAOD (Kloss et al., 2021). Ohneiser et al. (2021) discussed self-lofting as a potential transport pathway in the UTLS for these Siberian fires in the absence of strong vertical motion in July 2019. The smoke plume could rise and reach the UT and further ascend into the LS. During the slow ascent, the emissions are chemically altered; in the case of SO2, it is transformed to SO42-. Finally, the SO42- could have contributed to the enhanced SO42- in the LS. Further, wildfires in central Europe in May 2020 (European Commission et al., 2021) could also have undergone this self-lofting process as the trajectories do not indicate uplift over Europe and thus might additionally have contributed to elevated SO42- in the UTLS. In comparison to the Raikoke eruption, these biomass burning contributions are of lower magnitude. Reifenberg et al. (2022) suggest that other small- and medium-sized volcanic eruptions from tropical latitudes could have reached the stratosphere and thus impacted the stratospheric aerosol over Europe. One example is the volcano Taal in the Philippines (14.00∘ N, 120.99∘ E), which erupted on 12 January 2020, and its ash and gas plume rose up to around 10–15 km height (Krippner and Venzke, 2020). The VEI was estimated to 3 and the SO2 emissions estimated to 0.019 Tg (Liu et al., 2020). According to simulations of Reifenberg et al. (2022), the Taal eruption led partly to an increase of SO42- in the LS during BLUESKY.

The measured SO42- mixing ratios in the LS agree with other volcano-related in situ studies. The highest mixing ratios are reported by Schmale et al. (2010). They probed layers with enhanced SO42- in October/November 2008, roughly 3 months after the eruption of Mount Kasatochi (erupted 8 August 2008; 52.18∘ N, 175.51∘ W; 1.5 Tg SO2), with an injection height reaching into the stratosphere, and additionally Mount Okmok (53.40∘ N, 168.17∘ W) erupted on 12 July 2008 (0.2 Tg SO2, Carn et al., 2008). Jurkat et al. (2010) also measured enhanced SO2 concentrations in the stratosphere in the 3-month-old Kasatochi plume during the CONCERT campaign (Voigt et al., 2010). Martinsson et al. (2009) reported that particulate sulfur concentrations shortly after the Kasatochi eruption were 10 times higher than before the eruption, and even 3–4 months after the eruption they were enhanced by a factor of 3. In contrast, during volcanic quiescent periods, like the period between 1997 and 2008 (Deshler, 2008), the SO42- reduced mixing ratios in the stratosphere, and Martinsson et al. (2005) reported SO42- mixing ratios of 0.01 ppb for the lower stratosphere for the years 1999–2002.

The enhanced SO2 in the UT as described in Sect. 4 and the longer SO2 lifetime could possibly have a minor effect on the stratospheric sulfate aerosol. In these conditions, the SO2 had more time to be transported into the LS and finally be transformed to SO42-, adding to the SO42- mixing ratios. But it seems unlikely, as the trajectories for elevated SO42- (Fig. 5i–p) stay at high altitudes (and high potential temperatures), and thus mixing from the troposphere into the stratosphere is negligible. A few trajectories indicate a downward transport from higher altitudes and thus an origin deeper in the stratosphere (Fig. 5k/l), which could be a hint for OCS as an SO42- precursor. OCS is transported within the Brewer–Dobson circulation from the upper stratosphere to the lower stratosphere and is transformed via SO2 to H2SO4. According to Brühl et al. (2012), it is the most important source for maintaining the stratospheric aerosol layer in volcanic quiescent periods, and also for BLUESKY, OCS oxidation adds to the stratospheric SO42- background to some extent. Further, the Junge layer might also influence the SO42- mixing ratios in the LS. Even though the Junge layer is most pronounced at higher altitudes (Junge et al., 1961), it could extend further down, or due to the downward transport mentioned above (Fig. 5k/l) SO42- could be transported downward from the Junge layer into our measurement altitudes and thus contribute to the elevated SO42-.

We find elevated SO42- mixing ratios in stratospheric air masses and enhanced SO2 mixing ratios in tropospheric air masses over central Europe and the North Atlantic in spring 2020. The elevated SO2 of 0.06 ppb in the UT agrees with SO2 mixing ratios performed in the background of the North Atlantic flight corridor in 2008 (Jurkat et al., 2010) despite lower air traffic due to COVID-19 restrictions in 2020. The 3.5 % increase in air traffic since 2010 in part compensates for the air traffic reduction in 2020. In addition, exceptional dry weather conditions leading to a low cloud fraction and low OH concentrations in the UTLS in May 2020 (Schumann et al., 2021a, b; Van Heerwaarden et al., 2021) reduced SO2 sinks and increased SO2 lifetime. Back trajectories provided indications of other boundary layer SO2 sources from convective or WCB transport and further long-range transport in the UT, which could have contributed to a small extent.

In the LS, enhanced SO42- mixing ratios were observed. In comparison to previous studies, the SO42- mixing ratios were clearly above SO42- mixing ratios reported during volcanic quiescent periods (e.g., Martinsson et al., 2001, 2005) and agreed with SO42- mixing ratios in volcanically impacted air masses (e.g., Schmale et al., 2010) measured at altitudes below 12 km. Compared to 2018, stratospheric SO42- was significantly enhanced in 2020. The eruption of the volcano Raikoke injected 0.9–1.1 Tg of SO2 into the stratosphere (de Leeuw et al., 2021) in June 2019. In May 2020, an enhanced sAOD still was observed by Kloss et al. (2021) in the Northern Hemisphere caused by the Raikoke eruption and to a smaller extent by severe biomass burning events from June/July 2019 in Siberia and Canada. Further, Reifenberg et al. (2022) found that the eruption of the tropical volcano Taal in January 2020 contributed to the enhanced SO42- in the LS. We suggest these to be the primary sources of the enhanced stratospheric SO42- concentrations measured during BLUESKY, because back trajectories mainly showed long-range transport in the lower stratosphere.

Overall, the unprecedented BLUESKY mission was conducted during exceptional meteorological conditions and also reduced air traffic; both impacted the SO2 mixing ratios in the UT due to changes in the emissions and also sinks. The enhanced stratospheric sulfate aerosol, which was observed, was likely impacted by the volcano Raikoke and smaller sources.

Together with the observations of other sulfur compounds such as gaseous H2SO4 on HALO, which are still under evaluation, the unique and comprehensive data set of sulfur compounds allows us to test our understanding of the sulfur chemistry in global models (Reifenberg et al., 2022).

In a broader context, the present results give new insights in the sulfur chemistry in the UTLS region with respect to limited sources and sinks. They help to better understand (a) the sensitivity of SO2 to missing sinks and (b) the stratospheric aerosol and its dependence on perturbations and their lasting impacts. Both aspects are important to improve models, especially with respect to simulations of the Earth's radiation budget, because changes in the radiation balance in the UTLS impact feedback processes in the global climate.