Enhanced sulfur in the upper troposphere and lower stratosphere in spring 2020

. Sulfur compounds in the upper troposphere and lower stratosphere (UTLS) impact the atmosphere radiation budget, either directly as particles or indirectly as precursor gas for new particle formation. In situ measurements in the UTLS are rare but are important to better understand the impact of the sulfur budget on climate. The BLUESKY mission in May and June 2020 explored an unprecedented situation. (1) The UTLS experienced extraordinary dry conditions in spring 2020 over Europe, in comparison to previous years, and (2) the ﬁrst lockdown of the COVID-19 pandemic caused major emission reductions from industry, ground, and airborne transportation. With the two research aircraft HALO and Falcon , 20 ﬂights were conducted over central Europe and the North Atlantic to investigate the atmospheric composition with respect to trace gases, aerosol, and clouds. Here, we focus on measurements of sulfur dioxide (SO 2 ) and particulate sulfate (SO 2 − 4 ) in the altitude range of 8 to 14.5 km which show unexpectedly enhanced mixing ratios of SO 2 in the upper troposphere and of SO 2 − 4 in the lowermost stratosphere. In the UT, we ﬁnd SO 2 mixing ratios of (0 . 07 ± 0 . 01) ppb, caused by the remaining air trafﬁc, and reduced SO 2 sinks due to low OH and low cloud fractions and to a minor extent by uplift from boundary layer sources. Particulate sulfate showed elevated mixing ratios of up to 0.33 ppb in the LS. We suggest that the eruption of the volcano Raikoke in June 2019, which emitted about 1 Tg SO 2 into the stratosphere in northern midlatitudes, caused these enhancements, in addition to Siberian and Canadian wildﬁres and other minor volcanic eruptions. Our measurements can help to test models and lead to new insights in the distribution of sulfur compounds in the UTLS, their sources, and sinks. Moreover, these results can contribute to improving simulations of the radiation budget in the UTLS with respect to sulfur effects.


Introduction
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 (SO 2− 4 ) 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 nonvolcanic sulfur dioxide (SO 2 ), as well as tropospheric SO 2− 4 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 SO 2 (Sheng et al., 2015). The SO 2 chemistry and transport depends strongly on the ambient conditions. In the free troposphere and lower stratosphere, SO 2 reacts predominantly with hydroxyl (OH) to form sulfuric acid (H 2 SO 4 ) (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 H 2 SO 4 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 SO 2 to higher altitudes. Clouds could also limit the SO 2 lifetime to hours or days (Lelieveld, 1993), as the conversion of SO 2 to H 2 SO 4 is faster in cloud droplets than in the gas phase (Seinfeld and Pandis, 2006). Nevertheless, SO 2 can be transported from the planetary boundary layer (PBL) into the UTLS region via different pathways. Similar to OCS, SO 2 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 SO 2 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(Gottschaldt et al., , 2018Ploeger 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, SO 2 , 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 SO 2 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 SO 2 ) 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 SO 2 . 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 . Under these conditions, we found enhanced values of SO 2 in the upper troposphere (UT) and of particulate SO 2− 4 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 SO 2 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.

BLUESKY mission
The BLUESKY mission was led by the German Aerospace Center (DLR) and the Max Planck Institute for Chemistry, Mainz. Coordinated flights were performed from Oberpfaffenhofen with the High Altitude and Long Range Research Aircraft HALO and the DLR Falcon over Europe and the North Atlantic between 16 May and 9 June 2020. In total 20 flights were performed (Fig. 1). The period covered the first weeks of the COVID-19 lockdown in Europe and thus offered a unique opportunity to investigate an unprecedented situation of reduced anthropogenic emissions from industry, ground, and airborne transportation. The goal of BLUESKY was to explore the changes in the atmospheric composition and gain new insights on how anthropogenic emissions perturb chemical and physical processes in the atmosphere. Both aircraft were equipped with in situ instruments to investigate trace gases, aerosols, and cloud properties. The pay-load of both aircraft was complementary to obtain a comprehensive data set. In particular, during five days with coordinated flights over Germany (23,26,and 28 May) and over the North Atlantic (30 May, 2 June), the payload offered the opportunity to probe the air masses in more detail. An overview of the BLUESKY mission is given in Voigt et al. (2022), and further detailed studies are published by Schumann et al. (2021a, b), Reifenberg et al. (2022), Nussbaumer et al. (2022), Hamryszczak et al. (2022), and Krüger et al. (2022). In the present study we will focus on the sulfur compounds in the upper troposphere and lower stratosphere.

Instrumentation
In the present study several trace gas measurements on board Falcon and also trace gas and particle measurements on board HALO are used. On board Falcon, the atmospheric chemical ionization mass spectrometer AIMS measures gaseous SO 2 and nitric acid (HNO 3 ) among other compounds at mixing ratios relevant for the UTLS region by using SF − 5 as reagent ion. A more detailed description of the instrument can be found elsewhere Jurkat et al., 2016;Marsing et al., 2019). SO 2 is calibrated in-flight using an isotopically labeled calibration gas mixture of the isotope 34 SO 2 , which is heavier than the naturally dominant isotope 32 SO 2 but has the same chemical behavior . The natural isotopic ratio is 34 S/ 32 S = 0.0454, and the mass spectrometer can detect both isotopes separately as they differ in mass by 2 amu (atomic mass units). This has the advantage that the calibration gas can be continuously added to the sampling flow and the system is well conditioned for SO 2 . A drawback is that the background of the instrument is increased by 5 %, due to impurities of 32 SO 2 in the calibration gas. The SO 2 data are corrected for the instrumental background, which includes a moisture correction, as higher water vapor concentrations lead to cross sensitivities on m/z = 83 amu (FSO − 2 ; Jurkat et al., 2016). With increasing moisture in lower altitudes, a correction is more difficult and reduces the data quality. As the focus of the present study is the UTLS region, we limited our analysis on altitudes above 8 km and thus ensure the data quality. The other trace gas measured by AIMS is HNO 3 , which is in-flight calibrated using a permeation oven with a solution of HNO 3 in water . The data are background corrected including a moisture correction, which is necessary to account for cross sensitivities caused by water vapor . The AIMS measurements were performed with a 1.6 s time resolution and smoothed with a running mean of 20 s. The 1σ detection limit of SO 2 varied between 0.006-0.017 ppb. The total uncertainty is on average 22.7 % for SO 2 and included the uncertainty of the moisture correction. The 1σ detection limit for HNO 3 is in the range of 0.005-0.009 ppb. The HNO 3 total uncertainty is on average 16 % (Marsing, 2021). Further measurements on board Falcon included CO and O 3 . O 3 was measured using a UV pho-tometer (Schulte and Schlager, 1996;Ziereis et al., 2000), CO was measured by cavity ring-down spectroscopy (Klausner et al., 2020). The accuracies for CO and O 3 are 15 % and 5 %, respectively. Additionally, water vapor (H 2 O) was measured with the Lyman-alpha absorption instrument integrated in the meteorological sensor system.
On board HALO, the compact time-of-flight aerosol mass spectrometer (C-ToF-AMS) measured the aerosol composition (Drewnick et al., 2005;Schmale et al., 2010;Schulz et al., 2018). Aerosol particles of approximately 50 to 800 nm are analyzed, which then provides quantitative mass concentrations of organic matter, sulfate, nitrate, and ammonium. The instrument is equipped with a constant pressure inlet that ensures a steady mass flow and an operation pressure of the aerodynamic lens for stable in-flight operation (Molleker et al., 2020). Here, we focus on sulfate. For a better comparability of SO 2− 4 with SO 2 , we calculate mixing ratios (ppb) instead of using concentrations (µg m −3 ). Additionally, mixing ratios have the advantage of being pressure independent. We assumed that all SO 2− 4 would be evaporated and calculated the volumetric mixing ratio for SO 2− 4 . Above 8 km, the 1σ detection limit is (0.006 ± 0.001) ppb for SO 2− 4 , the accuracy is 30 %, and the precision on average is (0.002 ± 0.001) ppb (Schulz et al., 2018). Additionally, the trace gases CO and O 3 on board HALO are considered in the present study for altitudes above 8 km. CO was measured by the TRacer In Situ TDLAS for Atmospheric Research (TRISTAR; Schiller et al., 2008) with a total uncertainty of 3 % for tropospheric measurements (Nussbaumer et al., 2022). Note that due to a small nitrous oxide (N 2 O) interference the uncertainty in the lower stratosphere is higher (8.5 ± 3.9 ppbv). O 3 was measured by the Fast Airborne Ozone instrument FAIRO, which combines the technique of a UV photometer and a chemiluminescence detector , and the total uncertainty of O 3 was 2 %-2.5 %. Water vapor was measured with the tunable diode laser (TDL) hygrometer SHARC (Sophisticated Hygrometer for Atmospheric ResearCh).
SO 2 was sampled on board Falcon, and SO 2− 4 was probed on board HALO. Nevertheless, campaign averaged CO and O 3 profiles from both aircraft agree and motivate the combined interpretation of the SO 2 and SO 2− 4 distributions during spring 2020 (see Sect. 3.2).

Trajectory calculations
Back-trajectory calculations were performed using the HYS-PLIT atmospheric transport and dispersion model (Stein et al., 2015;Rolph et al., 2017) with the GDAS (Global Data Assimilation System) meteorological data set (Kanamitsu, 1989). For selected cases with either elevated SO 2 or SO 2− 4 mixing ratios, 360 h back-trajectory ensembles were calculated. One ensemble consists of 27 single trajectories, which are offset by one meteorological grid point in the horizontal and 0.01σ in the vertical coordinate. With the help of the back trajectories, air mass origins and transport pathways in the atmosphere could be identified.

Trace gases along flight track
On 2 June 2020, HALO and Falcon took off in Oberpfaffenhofen and headed towards the North Atlantic, west of Ireland, with similar flight tracks, altitude, and similar takeoff time. With a shorter flight range, Falcon landed in Shannon, Ireland, to refuel, while HALO continued its flight. In Fig. 2, flight altitude and in situ measurements are plotted against longitude for comparability from both aircraft for the time between 07:00 and 10:00 UTC. O 3 and CO were measured aboard both aircraft. Both trace gases show similar patterns independent of the aircraft. The highest O 3 mixing ratios with maxima of 475 ppb (Falcon) and 423 ppb (HALO) were approximately between 6 • W and 3 • E, while CO had the lowest mixing ratios of 18 ppb (Falcon) and 28 ppb (HALO). Along the longitudes outside of the pervious mentioned range, the trace gases showed an opposite behaviour: when CO mixing ratios were enhanced with maxima of 110 ppb (Falcon) and 109 ppb (HALO), O 3 mixing ratios were low (Falcon: 74 ppb; HALO: 62 ppb). Because CO and O 3 from both platforms agree within their uncertainties and reflect the same trends, we assume that both aircraft probed the same air mass. Between 6 • W and 3 • E, O 3 and HNO 3 are positively correlated, as expected, and HNO 3 mixing ratios increase up to 1.6 ppb. A similar trend can be observed in the particulate compound SO 2− 4 , which also is enhanced when O 3 , as a stratospheric tracer, is enhanced. SO 2− 4 ranges from the detection limit to 0.21 ppb. SO 2 mixing ratios show a larger variability, but in general they follow CO, which is a tropospheric tracer (Fischer et al., 2000;Hoor et al., 2002). SO 2 ranges from the detection limit to 0.15 ppb and experiences one short peak with 0.15 ppb around 6.5 • W at an altitude of 11.9 km. Other trace gases on board Falcon measured 266 ppb O 3 , 70 ppb CO, and 0.8 ppb HNO 3 during this event, but beside HNO 3 , which increased slightly, CO and O 3 showed no perturbations around this location. An in-depth analysis and discussion on potential explanations for this or similar features is presented in Sect. 4.
3.2 SO 2 and SO 2− 4 median profiles Following the case mentioned above, we broaden our analysis to all flights of the whole campaign. In Fig. 3, median, 25th percentile, and 75th percentile profiles of the trace gases and the particulate compounds are displayed with the potential temperature as a vertical axis. The medians, 25th percentiles, and 75th percentiles are calculated for 5 K potential temperature bins from 310 to 355 K for Falcon flights and up to 385 K for HALO flights. The Falcon profile is limited in height due to the maximum flight altitude of approx. 12.5 km in comparison to HALO with a ceiling altitude of 14.5 km. First, we compare the trace gases measured on both aircraft, Falcon and HALO. The stratospheric tracer O 3 behaves similarly for both aircraft within the 25th and 75th percentiles with a step around 340 K, while the spread between the percentiles starts to increase already around 330 K. Below this altitude, median O 3 mixing ratios reach minima of 41 ppb (Falcon) and 62 ppb (HALO), and above they rise up to 420 ppb (Falcon) and 642 ppb (HALO). The chemical tropopause is marked by strong gradients in several tracers. O 3 as a common indicator shows a kink in its median profile at around 140 ppb and 340 K potential temperature, which is within the limits given by Thouret et al. (2006). The dynamical tropopause, displayed as the 2 PVU based on ECMWF/ERA5 data along the Falcon flight tracks (Fig. 3), is around 335 K and thus has a similar height to the chemical tropopause. The tropospheric tracer CO from both platforms has similar profiles with median mixing ratios of 37-112 ppb (Falcon) and 12-96 ppb (HALO). Both CO profiles show a decrease with height and are thus anticorrelated to O 3 , but they reflect a significant change in the mixing ratios at 340 K, similar to O 3 . The profiles of the tropospheric tracer H 2 O also decrease with height, following the CO profile, but with a less pronounced step around the chemical tropopause. The H 2 O mixing ratios are 26-268 ppm (Falcon) and 3-153 ppm (HALO). Additionally, the median HNO 3 profile follows the trend of O 3 with low mixing ratios down to 0.3 ppb followed by a steep increase around 340 K and reaches a maximum of 1.4 ppb. Between 330 and 350 K, the stratospheric tracers O 3 and HNO 3 as well as the tropospheric tracer CO show larger variations between the percentiles, which indicates the mixing layer described by Hoor et al. (2002) over the course of the mission. As shown, the profiles for O 3 and CO for HALO and Falcon flight tracks agree well within their 25th and 75th percentiles. Both aircraft probed the atmosphere above Europe and the North Atlantic within the mission period and thus during similar meteorological conditions, even though the days and routes partly differ. This gives us confidence to further investigate and compare measurements of sulfur compounds sampled on both aircraft: SO 2 was sampled on Falcon, while SO 2− 4 was probed on HALO. Profiles of SO 2 and SO 2− 4 are shown in Fig. 3a and b, respectively. The median SO 2 profile decreases with height and median values range from 0.05 to 0.08 ppb, with the lowest mixing ratios above 330 K. The opposite behavior can be observed in the median SO 2− 4 profile, which increases with height. The mixing ratios are lowest at 0.02 ppb and rise up to 0.33 ppb. The profile shows a similar trend in comparison to O 3 with a stepwise increase around 340 K. The enhanced SO 2− 4 mixing ratios above the chemical tropopause can be associated with stratospheric sulfate aerosol.
In Fig. 3c the profile of the ratio of SO 2− 4 /(SO 2 + SO 2− 4 ) is plotted. The ratio is a measure of the relative contribution from the precursor SO 2 to the total sulfur budget. Below the chemical tropopause most of the SO 2 is still present as SO 2 , while above the tropopause SO 2− 4 dominates the ratio. The sum of SO 2 and SO 2− 4 for the median profiles is rather stable around 0.10 ppb between 310 and 340 K; above the chemical tropopause the sum increases up to on average 0.20 ppb and is dominated by the enhancement of SO 2− 4 in the range 340-355 K. In the next section we will discuss potential sources explaining the distribution of sulfur compounds in the UTLS.

Stratospherically and tropospherically influenced air masses
In order to obtain an overview on the distribution of sulfur compounds with respect to the chemical tropopause, Fig. 4 shows tracer-tracer correlations of O 3 , CO, and HNO 3 , comparable to previous studies investigating the cross-tropopause exchange and the chemical composition of the tropopause (Fischer et al., 2000;Hoor et al., 2002). In contrast to the median profiles in the previous section, here all available data are plotted, either for HALO (considering SO 2− 4 ) or for Falcon (considering SO 2 ). In the correlation plot between O 3 and CO color-coded with SO 2− 4 ( Fig. 4a) from HALO for the whole altitude range (0-14 km), the stratospheric branch is visible with low CO and high O 3 values, while the tropospheric branch is characterized by low O 3 and high CO values, similar to, e.g., Fischer et al. (2000). The transition layer is clearly visible; the layer is a mixing layer, influenced by air masses with stratospheric and tropospheric origin (Hoor et al., 2002). Without the exchange processes across the tropopause, we would expect an L-shape profile (Fischer at al., 2000). The mixing layer almost extends over the whole O 3 range from 150 to 400 ppb, similar to other mixing layers in the same latitude and season (Hoor et al., 2002;Pan et al., 2004), and is thicker in comparison to a winter polar mixing layer (Fischer et al., 2000). The higher SO 2− 4 mixing ratios are either in the (unmixed) stratospheric branch or partly mixed into in the upper part of the transition layer. With respect to the chemical tropopause, the elevated SO 2− 4 mixing ratios appear only in the stratosphere (O 3 ≥ 120 ppb; Thouret et al., 2006). In Fig. 4b the correlation between O 3 and CO with and without color-coded SO 2 on board Falcon is displayed. A subset of the Falcon flights is missing there due to missing O 3 data in the beginning of the campaign. As the Falcon mainly operated up to 12 km, the pure stratospheric branch is hard to identify, while the tropospheric branch is clearly identifiable with the black dots (without SO 2 ). However, within the mixing layer, the stratospheric and tropospheric influences still differ, which is reflected in the SO 2 mixing ratios. In order to cover all Falcon flights, we use here HNO 3 as a stratospheric tracer. In Fig. 4c, the HNO 3 to CO correlation for the measurements on board Falcon is plotted with color-coded SO 2 , and Fig. 4c includes all Falcon flights; Fig. 4b and c show similar patterns for SO 2 , with higher mixing ratios towards more tropospheric influence and lower mixing ratios when the stratospheric influence dominates. One SO 2 outlier with higher mixing ratios at enhanced HNO 3 and reduced CO can be identified ( Fig. 4c; grey circle). A possible explanation could be (aged) aircraft plume encounters (as observed, e.g., in Jurkat et al., 2011), as HYSPLIT back-trajectory calculations tend towards longrange transport in the UTLS region for this case (Fig. 5a/b). Generally, the trajectories do not indicate transport from local PBL sources for cases with elevated SO 2 . In Fig. 5a-h, examples for cases with elevated SO 2 are shown, including the case on 2 June 2020 described in Sect. 3.1 (Fig. 5g/h). Additionally, Hamryszczak et al. (2022) found that hydrogen peroxides were scavenged by clouds during BLUESKY in the lower and middle troposphere (0-7 km). As SO 2 can easily be scavenged by clouds (Seinfeld and Pandis, 2006), the potential for SO 2 being transported from the local PBL to the UT seems unlikely. The tracer-tracer correlation and the relative distribution of SO 2 along the transition layer show no direct link to the SO 2− 4 distribution, as they are in different regimes; thus, we assume that they originate from different sources. The back trajectories of SO 2 and SO 2− 4 cases support the assumption that the origins differ, especially concerning the different altitude ranges of the trajectories (Fig. 5).

Enhanced SO 2 in the UT
In Fig. 6a, the median SO 2 profile with shaded 25th and 75th percentiles is plotted against the flight altitude, in order to compare the data with literature values. The SO 2 percentiles are calculated for 500 m bins. Similar to Fig. 3b, the median SO 2 decreases with height from around 0.08 to 0.04 ppb. Previous in situ SO 2 measurements at similar flight altitudes are included in Fig. 6a. Overall, the BLUESKY SO 2 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 SO 2 values for the upper-tropospheric back-   Speidel et al. (2007). Above 12 km the BLUESKY SO 2 mixing ratios agree well with the stratospheric background from  and the tropospheric background from Speidel et al. (2007). Surprisingly, the BLUESKY SO 2 profile slightly exceeds the previous measurements below these altitudes in the upper troposphere. The upper-tropospheric SO 2 profile compares better to SO 2 mixing ratios, which were associated with the SO 2 background in the North Atlantic flight corridor in 1997 or 2010 (Arnold et al., 1997;Jurkat et al., Figure 5. HYSPLIT 360 h back trajectories calculated for cases with elevated SO 2 (a-h; Falcon) and SO 2− 4 (i-p; HALO) mixing ratios. The release points started in the vicinity of these events. Left column: the flight tracks are in black, and the potential temperature along the trajectories to indicate the transport altitude is color-coded. The right column represents the trajectories as hours before release vs. altitude, also with color-coded potential temperature. The cases of enhanced SO 2 were on 19 May 2020 In 2020, a 72 % reduction of the air traffic above Europe has been reported due to the COVID-19 lockdown in compar-ison to the same time period in 2019 (Schumann et al., 2021a, b), hence providing a lower aviation SO 2 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 SO 2 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 SO 2 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 SO 2 budget in the upper troposphere. To analyze the origin of air masses with elevated SO 2 , HYSPLIT back trajectories are calculated and representative examples are plotted in Fig. 5a-h. SO 2 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 SO 2 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(Vogel et al., , 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 SO 2 (Fig. 5a-h) show lower potential temperatures in comparison to the trajectories calculated for elevated SO 2− 4 ( Fig. 5i-p). Hence, long-range transport of SO 2 -enriched PBL air masses could have contributed to the observed BLUESKY SO 2 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 SO 2 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 SO 2 might have already been transformed to SO 2− 4 before reaching the UTLS and does not contribute to the elevated SO 2 in the UT. Moreover, the decrease of SO 2 in the LS, as expected, does not support transport of SO 2 beyond the UT into the LS either via convection or via warm conveyor belts. This can be confirmed by the trajectories for SO 2− 4 as they indicate long-range transport at high altitudes with negligible influence from lower altitudes. The trajectories helped identify potential source regions of SO 2 .
Beside the sources, also sinks of SO 2 can alter the SO 2 concentrations in the UTLS. Beside the conversion to H 2 SO 4 , leading to sulfate particles, SO 2 is removed from the atmosphere by wet and dry deposition. SO 2 can be scavenged by clouds, which lead to a significant reduction of the SO 2 lifetime (Lelieveld, 1993). Van Heerwaarden et al. (2021) in-vestigated 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 SO 2 sinks. Furthermore, elevated humidity favors the faster conversion of SO 2 to SO 3 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 H 2 O 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 H 2 O 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 H 2 O led to lower OH concentrations during the BLUESKY period, which implies less chemical processing and hence a reduction of SO 2 sinks. Less SO 2 sinks could lead to an enhanced SO 2 lifetime in the UTLS and thus higher SO 2 mixing ratios.
In sum, the enhanced SO 2 mixing ratios at cruise levels in Europe in spring 2020 can possibly be explained by a nonnegligible aviation SO 2 contribution, WCB, or convective transport from the boundary layer, followed by long-range transport and the prolonged SO 2 lifetime caused by the unusually dry UTLS conditions. Neither the sources nor the sinks could separately explain the SO 2 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.

Stratospheric sulfate aerosol
As mentioned in Sect. 3.2, SO 2− 4 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 O 3 mixing ratios are stable, and above O 3 increases. The O 3 mixing ratio above 120 ppb indicates stratospheric air masses as mentioned above; thus, the higher SO 2− 4 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 SO 2− 4 correlates well with O 3 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. (2001Martinsson et al. ( , 2005 observed significantly lower SO 2− 4 mixing ratios in the European upper troposphere and tropopause region. The SO 2− 4 profile (Morgan et al., 2009) obtained from April 2005 to September 2006 over the UK shows higher SO 2− 4 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 . The SO 2− 4 mixing ratio was on average (0.04±0.01) ppb and compares well to the BLUESKY SO 2− 4 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 SO 2− 4 mean was (0.03 ± 0.01) ppb, similar to the BLUESKY value. Above 11 km in the lower stratosphere, SO 2− 4 rises to (0.09 ± 0.03) ppb. Considering heights above the tropopause, i.e., with enhanced SO 2− 4 mixing ratios, the stratospheric BLUESKY SO 2− 4 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 SO 2− 4 mixing ratios during BLUESKY. As mentioned above, we calculated HYSPLIT back trajectories for cases of elevated SO 2− 4 . In Fig. 5i-p, represen-tative 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 SO 2− 4 is already in the stratosphere 360 h before sampling, and hence influence from the troposphere is negligible. One major source of SO 2− 4 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 SO 2 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 SO 2 into the atmosphere and simulated also that approximately 0.9-1.1 Tg SO 2 thereof was injected into the stratosphere.  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 SO 2− 4 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 SO 2 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 . Ohneiser et al. (2021) discussed selflofting 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 SO 2 , it is transformed to SO 2− 4 . Finally, the SO 2− 4 could have contributed to the enhanced SO 2− 4 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 SO 2− 4 in the UTLS. In comparison to the Raikoke eruption, these biomass burning contributions are of lower magnitude.  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 SO 2 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 SO 2− 4 in the LS during BLUESKY.
The measured SO 2− 4 mixing ratios in the LS agree with other volcano-related in situ studies. The highest mixing ratios are reported by Schmale et al. (2010) Carn et al., 2008).  also measured enhanced SO 2 concentrations in the stratosphere in the 3-month-old Kasatochi plume during the CONCERT campaign . 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 SO 2− 4 reduced mixing ratios in the stratosphere, and Martinsson et al. (2005) reported SO 2− 4 mixing ratios of 0.01 ppb for the lower stratosphere for the years 1999-2002.
The enhanced SO 2 in the UT as described in Sect. 4 and the longer SO 2 lifetime could possibly have a minor effect on the stratospheric sulfate aerosol. In these conditions, the SO 2 had more time to be transported into the LS and finally be transformed to SO 2− 4 , adding to the SO 2− 4 mixing ratios. But it seems unlikely, as the trajectories for elevated SO 2− 4 ( 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 SO 2− 4 precursor. OCS is transported within the Brewer-Dobson circulation from the upper stratosphere to the lower stratosphere and is transformed via SO 2 to H 2 SO 4 . 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 SO 2− 4 background to some extent. Further, the Junge layer might also influence the SO 2− 4 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) SO 2− 4 could be transported downward from the Junge layer into our measurement altitudes and thus contribute to the elevated SO 2− 4 .

Conclusion and outlook
We find elevated SO 2− 4 mixing ratios in stratospheric air masses and enhanced SO 2 mixing ratios in tropospheric air masses over central Europe and the North Atlantic in spring 2020. The elevated SO 2 of 0.06 ppb in the UT agrees with SO 2 mixing ratios performed in the background of the North Atlantic flight corridor in 2008  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 SO 2 sinks and increased SO 2 lifetime. Back trajectories provided indications of other boundary layer SO 2 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 SO 2− 4 mixing ratios were observed. In comparison to previous studies, the SO 2− 4 mixing ratios were clearly above SO 2− 4 mixing ratios reported during volcanic quiescent periods (e.g., Martinsson et al., 2001Martinsson et al., , 2005 and agreed with SO 2− 4 mixing ratios in volcanically impacted air masses (e.g., Schmale et al., 2010) measured at altitudes below 12 km. Compared to 2018, stratospheric SO 2− 4 was significantly enhanced in 2020. The eruption of the volcano Raikoke injected 0.9-1.1 Tg of SO 2 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 SO 2− 4 in the LS. We suggest these to be the primary sources of the enhanced stratospheric SO 2− 4 concentrations measured during BLUESKY, because back trajectories mainly showed longrange transport in the lower stratosphere.
Overall, the unprecedented BLUESKY mission was conducted during exceptional meteorological conditions and also reduced air traffic; both impacted the SO 2 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 H 2 SO 4 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 .
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 SO 2 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.
Author contributions. CV, TJW, HS, and JL planned the flight experiment. AM, JL, KK, JS, MS, and LR performed the in-flight measurements. KK and JS provided evaluated particulate data and previous campaign data and supported the analysis. AM provided supporting evaluation and assisted the analysis. FO, AZ, HF, LR, and MZ provided supporting evaluation. SK supported the analysis. LT evaluated and analyzed the data and prepared the manuscript with contributions from CV, AM, TJW, JS, KK, HF, and SK. All co-authors commented on the manuscript.

Competing interests.
The contact author has declared that none of the authors has any competing interests.
Disclaimer. Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Special issue statement.
This article is part of the special issue "BLUESKY atmospheric composition measurements by aircraft during the COVID-19 lockdown in spring 2020". It is not associated with a conference.
Acknowledgements. The HALO flights during the BLUESKY mission were funded by the Max Planck Society. The authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY website (https://www.ready.noaa.gov, last access: 22 November 2022) used in this publication. The authors thank Andreas Giez, Christian Mallaun, and Vladyslav Nenakhov for providing the water vapor measurement on board HALO and Falcon, as well as Benedikt Steil for providing the ECMWF/ERA5 reanalysis data.
Financial support. Laura Tomsche is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), grant no. TRR 301 and project ID 428312742.
The article processing charges for this open-access publication were covered by the German Aerospace Center (DLR). Review statement. This paper was edited by Joshua Fu and reviewed by Daniele Visioni, Cheng-En Yang, and one anonymous referee.