Organic and inorganic bromine measurements around the extratropical tropopause and lowermost stratosphere: insights into the transport pathways and total bromine

. We report on measurements of total bromine (Br tot ) in the upper troposphere and lower stratosphere taken during 15 ﬂights with the German High Altitude and LOng range research aircraft (HALO). The research campaign WISE (Wave-driven ISentropic Exchange) included regions over the North Atlantic, Norwegian Sea, and northwestern Europe in fall 2017. Br tot is calculated from measured total organic bromine (Br org ) added to inorganic

Abstract. We report on measurements of total bromine (Br tot ) in the upper troposphere and lower stratosphere taken during 15 flights with the German High Altitude and LOng range research aircraft (HALO). The research campaign WISE (Wave-driven ISentropic Exchange) included regions over the North Atlantic, Norwegian Sea, and northwestern Europe in fall 2017. Br tot is calculated from measured total organic bromine (Br org ) added to inorganic bromine (Br inorg y ), evaluated from measured BrO and photochemical modeling. Combining these data, the weighted mean [Br tot ] is 19.2±1.2 ppt in the northern hemispheric lower stratosphere (LS), in agreement with expectations for Br tot in the middle stratosphere (Engel and Rigby et al., 2018). The data reflect the expected variability in Br tot in the LS due to variable influx of shorter lived brominated source and product gases from different regions of entry. A closer look into Br org and Br inorg y , as well as simultaneously measured transport tracers (CO and N 2 O) and an air mass lag time tracer (SF 6 ), suggests that bromine-rich air masses persistently protruded into the lowermost stratosphere (LMS) in boreal summer, creating a high bromine region (HBrR). A subsection, HBrR * , has a weighted average of [Br tot ] = 20.9 ± 0.8 ppt.
The most probable source region is air recently transported from the tropical upper troposphere and tropopause layer (UT/TTL) with a weighted mean of [Br tot ] = 21.6 ± 0.7 ppt. CLaMS Lagrangian transport modeling shows that the HBrR air mass consists of 51.2 % from the tropical troposphere, 27.1 % from the stratospheric background, and 6.4 % from the midlatitude troposphere (as well as contributions from other domains). The majority of the surface air reaching the HBrR is from the Asian monsoon and its adjacent tropical regions, which greatly influences trace gas transport into the LMS in boreal summer and fall. Tropical cyclones from Central America in addition to air associated with the Asian monsoon region contribute to the elevated Br tot observed in the UT/TTL. TOMCAT global 3-D model simulations of a concurrent increase of Br tot show an associated O 3 change of −2.6±0.7 % in the LS and −3.1±0.7 % near the tropopause. Our study of varying Br tot in the LS also emphasizes the need for more extensive monitoring of stratospheric Br tot globally and seasonally to fully understand its impact on LMS O 3 and its radiative forcing of climate, as well as in aged air in the middle stratosphere to elucidate the stratospheric trend in bromine.

Introduction
Bromine is thought to account for approximately one-third of the global stratospheric O 3 depletion, and it is particularly effective at destroying O 3 in the lowermost stratosphere (LMS) (e.g., Salawitch et al., 2005;Hossaini et al., 2015;WMO, 2018;Chipperfield et al., 2020). The LMS is defined as the atmospheric layer extending from the local tropopause (TP) to the potential temperature = 380 K isentrope, thereby excluding the tropics (Holton et al., 1995), and we define the lower stratosphere (LS) from the local TP up to the highest measurements of this study at ∼ 410 K. In the definition of the tropical tropopause layer (TTL), we follow Fueglistaler et al. (2009), i.e., for latitudes <∼27 • N and S and potential temperatures = 355-425 K.
There are four major contributions to total bromine (hereafter called Br tot ) in the troposphere, as suggested by previous research (e.g., Engel and Rigby et al., 2018): (1) CH 3 Br that has both natural and anthropogenic sources, with a contribution of 6.8 ppt to Br tot in 2016; (2) the four major halons consisting of H-1211 (CClBrF 2 ), H-1301 (CBrF 3 ), H-1202 (CBr 2 F 2 ), and H-2402 (CBrF 2 CBrF 2 ), all from anthropogenic activities and contributing 7.8 ppt to Br tot in 2016; (3) bromine compounds inferred from naturally emitted socalled very short-lived substances (Br VSLS ); and (4) inorganic bromine previously released from brominated VSLS or sea salt, of which a fraction of both is eventually also transported into the stratosphere. Total organic bromine (Br org ) is the sum of contributions (1)- (3), while inorganic bromine defined here as Br inorg y (often referred to as Br y ) consists of contribution (4).
The monitoring of CH 3 Br (e.g., Montzka et al., 2003) and the longer lived halons (e.g., Vollmer et al., 2016) is regularly performed by different observational networks, notably at NOAA (National Oceanic and Atmospheric Administration), UEA (University of East Anglia), and AGAGE (Advanced Global Atmospheric Gases Experiment; Prinn et al. (2018)), amongst others. Numerous observational and theoretical studies have addressed the variable contributions (3)-(4), which together are assessed to contribute 5 ± 2 ppt to stratospheric bromine (e.g., Engel and Rigby et al., 2018, and references therein). These include studies performed from the ground, ships, manned and unmanned aircraft, and highflying balloons at different locations/seasons and globally from satellites. In order to answer the question of how much bromine is in the stratosphere, detailed studies performed at the stratospheric entry in different regions and seasons are most suitable. However, additional studies addressing Br tot measurements in aged air of the middle and upper stratosphere at a given location and season are also crucial.
Many studies have occurred primarily in the tropics and the Asian monsoon during the convective season, where most of the stratospheric air is believed to enter (e.g., Holton et al., 1995;Fueglistaler et al., 2004 andPloeger et al., 2015). Observational and theoretical studies addressing bromine at the stratospheric entry level in the tropics started well over 2 decades ago. Numerous dedicated research activities have taken place to investigate the fate of atmospheric bromine (e.g., Jensen et al., 2013;Harris et al., 2017 andPan et al., 2017).
It is well known that contribution (3) to Br tot , the brominated VSLS summed up according to their Br atomicity (Br VSLS ), is predominantly from marine emissions of CH 2 Br 2 and CHBr 3 , with small contributions from other brominated hydrochlorobromocarbons (CH 2 ClBr, CHCl 2 Br, CHClBr 2 ) (Engel and Rigby et al., 2018). Early measurements of CH 2 Br 2 and CHBr 3 at the tropical TP by Schauffler et al. (1993Schauffler et al. ( , 1998Schauffler et al. ( , 1999 indicated that both species may contribute substantially to Br tot (e.g., a few parts per trillion). More recent studies reported varying contributions of brominated VSLS in the tropical upper troposphere (UT) and TTL. The Br VSLS observations range from ∼ 1−4 ppt in the tropical upper troposphere and tropopause layer (UT/TTL; e.g., Laube et al., 2008;Brinckmann et al., 2012;Sala et al., 2014;Wisher et al., 2014;Navarro et al., 2015;Wales et al., 2018;Adcock et al., 2021). A recent study by Keber et al. (2020) reported on aircraft measurements at the TP and in the LMS at middle and high northern latitudes during different seasons. They observed [Br VSLS ] = 3.99 ± 1.15 ppt in the 10 K below the TP in boreal late summer and fall 2012/2017 and [Br VSLS ] = 5.20 ± 1.25 ppt in December 2015-March 2016. Keber et al. (2020) ascribed the varying mixing ratios found for the major brominated VSLS at the local TP to the location and seasonality of their source strengths, transport pathways, and atmospheric lifetimes.
Br tot and reactive bromine are even less explored in the LMS in midlatitudes for all seasons and the high-latitude LS in boreal summer. This region is particularly interesting since it comprises about 40 % of the stratospheric background air, though dynamically somewhat different from the tropical LS and "overworld" circulation of the global middle stratosphere (e.g., Holton et al., 1995;Bönisch et al., 2009;Birner and Bönisch, 2011;Gettelman et al., 2011;Ploeger et al., 2015 andKonopka et al., 2015). It has recently attracted renewed interest since an apparent ongoing O 3 decrease is observed there, particularly in the Northern Hemisphere (NH) (e.g., Ball et al., 2018Ball et al., , 2019Ball et al., , 2020Chipperfield et al., 2018Chipperfield et al., , 2020. The continuing negative LS O 3 trends may partly be related to increases of anthropogenic and natural VSLS and their inorganic products (e.g., Ball et al., 2018;Barrera et al., 2020). Simulations by Barrera et al. (2020) show that brominated VSLS have the largest impact on O 3 destruction in spring, but the overall O 3 loss in the LMS is largest in winter. On the other hand, others suggested that the changing transport is the main driver for the variability and trends in LS O 3 (e.g., Chipperfield et al., 2018).
Further, it is well known that the following factors relevant for stratospheric Br tot may vary strongly with location and season: (a) the source strengths of VSLS and sea salt at the surface (e.g., Quack et al., 2004;Yokouchi et al., 2005;Warwick et al., 2006;Liang et al., 2010;Ordóñez et al., 2012;Ziska et al., 2013;Hossaini et al., 2012bHossaini et al., , 2013Hossaini et al., , 2016Wang et al., 2019, and many others); (b) the delivery of VSLS, their product gases, and eventually of halogens sourced from sea salt to the stratospheric entry level (e.g., Aschmann et al., 2009Aschmann et al., , 2011Aschmann and Sinnhuber, 2013;Liang et al., 2014;Navarro et al., 2015;Hossaini et al., 2016;Schmidt et al., 2016;Fiehn et al., 2018;Filus et al., 2020;Adcock et al., 2021, and others); (c) the efficacy of removal of the halogenated product gases from the upper troposphere and lower stratosphere (UTLS) (e.g., Sinnhuber and Folkins, 2006;Aschmann et al., 2011;Aschmann and Sinnhuber, 2013); and (d) the transport of air into LMS via the two major pathways, i.e., by stratosphere-troposphere exchange (STE) at midlatitudes and the isentropic transport of air from the lower TTL into the LMS (e.g., Holton et al., 1995;Hoor et al., 2005;Bönisch et al., 2009;Ploeger et al., 2015;Konopka et al., 2015;Inai et al., 2019, and others). Therefore, it cannot be assumed ad hoc that the sum nor the speciation of the source and product gases for each halogen type (here of bromine) is spatially and temporally constant in the air masses of each transport pathway and hence the whole LMS.
The transport of air masses associated with the Asian monsoon anticyclone has recently attracted interest since it delivers pollutants and eventually VSLS to the LMS (e.g., Levine et al., 2007;Gettelman et al., 2011;Vogel et al., 2014Vogel et al., , 2016Orbe et al., 2015;Müller et al., 2016;Rolf et al., 2018 andWetzel et al., 2021) but has not been studied extensively for bromine (Adcock et al., 2021). Studies such as those by Vogel et al. (2014) discuss the rapid (∼ 5-week) transport pathway of air masses from the southeastern Asia boundary layer to the LMS over northern Europe in September 2012 (spatially and seasonally similar to the current study). While bromine transport to the TTL has been observed in various studies, the continued transport to the LMS in boreal summer is still insufficiently investigated. In addition, it has been speculated that the varying source strengths, transport pathways, removal processes, and in particular the contributions from (3) and (4) to the LMS Br tot , may change with future climate change (e.g., Hossaini et al., 2012a;Falk et al., 2017 andZiska et al., 2017). However, direct observations of this varying composition, amount of bromine, and its consequence for LMS O 3 are still lacking.
The present study addresses the spatial variability of Br org and Br inorg y in the NH UTLS during fall 2017 as part of the WISE (Wave-driven and ISentropic Exchange) campaign. Together with measurements of transport tracers (CO and N 2 O) and air mass age tracer (SF 6 ), corroborated by Lagrangian chemistry transport model simulations, the major transport pathways of brominated VSLS and their product gases into the extratropical LS (Ex-LS), i.e., by STE at midlatitudes and isentropic transport from the tropical UT/TTL by planetary waves, are discussed. Our study of Br tot thus partly complements the recent study of Keber et al. (2020) on the contribution and origins of the different organic source gases to Br tot in the NH Ex-LS.
However, observational studies addressing Br tot around the UTLS may not directly relate to Br tot available in the middle stratosphere because of the variability and heterogeneity in the source strengths, the variable transformation and atmospheric transport of the brominated VSLS, and the amount of bromine inferred by the product gases and sea salt. In addition to this variability, the accuracies of the individual techniques (though they have been greatly improved over time) and the suspected small trend in stratospheric bromine from controlled substances (−0.15 ± 0.04 ppt yr −1 ; Engel and Rigby et al., 2018) make it difficult to infer a trend in Br tot from measurements in the UT/TTL and LMS. Therefore, a continuation of high-precision Br tot measurements in aged and well-mixed air of the middle stratosphere is necessary in the coming decades, both for monitoring the contribution of the controlled bromine substances as well as the expected climate-change-induced changing contributions of VSLS and inferred Br inorg y to stratospheric Br tot . The paper is organized as follows. The key instruments and methods along with some (necessary) data reductions are reported in Sect. 2. The WISE campaign measurements are described in Sect. 3. Inferred Br tot measurements are presented in Sect. 4.1, followed by their geographical distribution in the UTLS indicating their source regions in Sect. 4.2. Model results of transport pathways contributing to elevated Br tot air masses in the LMS are discussed in Sect. 4.3, and the consequences of this additional Br tot in the LMS on O 3 are discussed in Sect. 4.4. Finally, our conclusions are summarized in Sect. 5.

Instruments and methods
The present study uses four instruments assembled on the German High Altitude and LOng range research aircraft (HALO). They are the mini-DOAS instrument for remote sensing measurements of O 3 , NO 2 , and BrO (Hüneke et al., 2017), FAIRO (Fast Airborne Ozone monitor) for the measurements of in situ O 3 , GhOST (Gas chromatograph for the Observational Studies using Tracers) for the in situ measurements of the organic brominated source gases (Sala et al., 2014) and SF 6 mean air mass lag times (Bönisch et al., 2009), and UMAQS (University of Mainz Airborne QCL (quantum cascade laser) Spectrometer) for in situ measurements of air mass transport tracers N 2 O and CO Kunkel et al., 2019). For the interpretation of our measurements regarding the chemistry and the transport of air masses, we use simulations by the Lagrangian chemistry transport model (CTM) CLaMS (Chemical Lagrangian Model of the Stratosphere) (e.g., McKenna et al., 2002b;Pommrich et al., 2014). In addition, the O 3 loss caused by the transport of bromine-containing substances into the NH UTLS is simulated by the TOMCAT/SLIMCAT model (Chipperfield, 1999(Chipperfield, , 2006. Since all methods and tools have previously been described in detail in dedicated studies, only their major features are briefly summarized here.

Mini-DOAS measurements of O 3 , NO 2 , and BrO
The mini-DOAS instrument, operated by the University of Heidelberg on the HALO aircraft, is a remote sensing instrument consisting of UV-Vis-NIR six-channel optical spectrometers. The scattered skylight received from scanning limb and nadir directions can be analyzed for O 3 , NO 2 , and BrO (as well as a variety of other trace gases). In this study only the UV and visible channels (FOV (field of view) is 0.5 • vertical, 3.15 • horizontal) in the limb orientation are used. Therefore, the probed air masses are located perpendicular to the aircraft main axis on the starboard side, and the telescopes are continuously adjusted in flight to compensate for changes in the roll angle of the HALO aircraft for constant alignment with the flight altitude. Data collected from misaligned telescopes are excluded in the analysis. For extensive details of the HALO mini-DOAS instru-ment, see Hüneke (2016) and Hüneke et al. (2017), and references therein. For additional mini-DOAS measurements, see Werner et al. (2017) and Stutz et al. (2017), and for more details of the mini-DOAS measurements from HALO, see Kluge et al. (2020).
The post-flight spectral analysis of the collected data for the detection of O 3 , NO 2 , and BrO is based on the DOAS method (differential optical absorption spectroscopy; Platt and Stutz, 2008). For the UV spectrometer retrieval of BrO differential slant column densities (dSCD), given in molecules per square centimeter, a fit window of 346−359 nm is used as suggested by Aliwell et al. (2002), i.e., a two-band fit of the vibrational transitions 4.0 and 5.0 of the A 2 3/2 ← X 2 3/2 electronic transition (Wilmouth et al., 1999). For O 3 -UV dSCDs, used as the corresponding scaling gas, a fit window of 338−357 nm is applied (Puk ,ī te and Wagner, 2016). NO 2 dSCDs from the visible channel spectrometer are evaluated between 432−490 nm (adjusted lower limit compared to Hüneke et al., 2017 because the wavelength range of the updated spectrometer changed, and therefore the lower wavelengths are removed to avoid instabilities). And O 3 -Vis dSCDs are evaluated between 450−500 nm (Hüneke et al., 2017). A summary of the trace gases and details of their absorption cross sections used for the DOAS analysis can be found in Tables 1 and 2. The dSCDs of each gas are calculated using an in-flight Fraunhofer reference spectrum, which is the same for the scaling and target gas (e.g., O 3 -UV and BrO, respectively). A running mean of five spectra (on average 158 s per running mean spectrum and corresponds to ∼ 31.6 km with a 1σ variability of 55 s) is used for BrO (and O 3 -UV) dSCD analysis to improve the signal to noise due to its low volume mixing ratios (VMRs), while for NO 2 (and O 3 -Vis) each spectrum is evaluated individually (on average 12 s per spectrum corresponding to ∼ 2.4 km with a 1σ variability of 5 s). Outliers of selected individual dSCD data points that deviate by more than 5 × 10 13 , 1 × 10 15 , or 2×10 18 molecules cm −2 (∼ 10 %-25 % jump) from the ninepoint running mean for BrO, NO 2 , or O 3 , respectively, are averaged to remove unstable spectra. Typical dSCD errors of BrO, NO 2 , O 3 -UV, and O 3 -Vis can be found in the final column of Table 1. Additional settings for the DOAS analysis are described in detail in Hüneke et al. (2017).
The summation of the dSCD and a reference slant column density (SCD ref ), which accounts for the absorption in the Fraunhofer reference chosen from the flight, results in the total SCD (see Eq. (1) in Hüneke et al., 2017). The O 3 -UV SCD ref is found by evaluating the Fraunhofer reference using an ozone-free external high-resolution absorption-free Sun spectrum from Chance and Kurucz (2010). The BrO SCD ref is calculated using the quasi-linear relationship between BrO dSCDs and the absorption-free spectra (Chance and Kurucz, 2010) SCDs of O 3 -UV; any data points from the initial aircraft ascent and final descent as well as when the solar zenith angle (SZA) is > 89 • are removed. When inspecting the O 3 vs. BrO relationship, an offset of BrO when   (Hüneke et al., 2017;Stutz et al., 2017). The solar I 0 correction used is suggested in Aliwell et al. (2002). For the retrieval of VMRs at flight altitude, the recently developed O 3 scaling method is used, the details of which are discussed in Stutz et al. (2017), Hüneke et al. (2017), and Werner et al. (2017). For this method, the radiative transfer of each limb measurement is modeled using McArtim  and considers the atmospheric, instrumental, and celestial (e.g., Sun position) parameters and geolocation as described in Hüneke et al. (2017). In the RT model, a combination of climatological profiles of the aerosol concentration are used, including EARLINET data (European Aerosol Research Lidar Network; Pappalardo et al., 2014) and SAGE III-ISS data (Stratospheric Aerosol and Gas Experiment III on the International Space Station; NASA/LARC/SD/ASDC, 2017). The a priori profile shapes of the scaling and target gases are taken from simulated trace gas VMR curtains along the flight track modeled by CLaMS (see Sect. 2.5 below). To remove any discrepancies between the CLaMS model and the real conditions, particu-larly when crossing a strong O 3 gradient, the CLaMS results for all gases are vertically adjusted (up or down) to match the CLaMS O 3 at flight altitude with the in situ O 3 observed by FAIRO (see Sect. 2.2 below). This was done previously in Werner et al. (2017) and Stutz et al. (2017), who shifted the TOMCAT/SLIMCAT (see Sect. 2.6 below) model results to match the measured O 3 . The vertical shifting of CLaMS is on average ∼ 1 km, which corresponds to approximately one pressure-altitude grid box of the original CLaMS resolution and has a maximum shift of 2.5 km. These adjusted profiles are used by the RT model to calculate the trace gas absorption in the line-of-sight to the total absorption, i.e., the so-called α factors (see Stutz et al., 2017, or Hüneke et al., 2017, for a derivation).
Lastly, to retrieve the trace gas VMRs ([X] j ) in an atmospheric layer j (i.e., at flight altitude for limb measurements), the scaling method equation is applied (see Stutz et al., 2017, or Hüneke et al., 2017: where X represents the target gas (BrO or NO 2 in this study), and P is the scaling gas (O 3 in this study). The scaling method virtually compensates for any effects of the light path length modifications (e.g., scattering by aerosol and clouds) observed in the SCDs and includes instrumental, celestial, and climatological data mentioned above in the α factor. The α X j of the target gas and α P j of the scaling gas are wavelength-dependent, and therefore the target and scaling gas should have similar or largely overlapping wavelength fit windows (Table 1). However, for a given wavelength the α factor ratio is insensitive to scattering by aerosols and clouds due to the equivalence theorem (Irvine, 1964) as shown in simulations performed by Knecht (2015) or in the Supplement of Hüneke et al. (2017). The SCD X and SCD P are the total slant column densities of the target gas (BrO or NO 2 ) and scaling gas (O 3 ), respectively, evaluated from the mini-DOAS data.
[P] j is the scaling gas VMR, which in this study is the in situ O 3 as measured by the FAIRO instrument aboard HALO (see Sect. 2.2). The VMR of the target gas, [X] j , for these limb UV-Vis measurements is an average over a given atmospheric volume based on the distance the aircraft has traveled during a single measurement, over the FOV of the telescope to the right of the aircraft, and the light path length in the current conditions (  . The CL sensor detects light emission from the reaction of O 3 with an organic dye, coumarin (e.g., Ermel et al., 2013). The CL response is calibrated by the UV photometer, thus combining the accuracy of a UV photometer with the high measurement frequency of a CL detector. Typical 1σ measurement precision is 0.08 ppb for the UV spectrometer measuring at 0.25 Hz and 0.5 % for the CL detector measuring at 10 Hz. Accuracy for the UV spectrometer measurements is ∼ 1.5 % (mostly determined by the uncertainty of the O 3 cross section). Throughout this study, the O 3 from the CL method is used with the exception of one flight (on 9 October 2017 -RF11; refer to Sect. 3) for which the UV photometer method O 3 is used as only partial CL O 3 data were available.

GhOST measurements of brominated source gases, transport tracer SF 6 , and air mass lag time
The Gas chromatograph for the Observational Studies using Tracers (GhOST) instrument, operated by the University of Frankfurt, detects stable atmospheric trace gases. It has two gas chromatographic channels, one which is coupled to a mass spectrometer (MS; Agilent MSD 5975) and the other one to an electron capture detector (ECD) (e.g., Sala et al., 2014;Obersteiner et al., 2016;Keber et al., 2020). SF 6 and CFC-12 (chlorofluorocarbon) are measured on the ECD channel with a time resolution of 60 s and a high precision of typically about 0.6 % for SF 6 and 0.2 % for CFC-12, in a similar setup as used in Bönisch et al. (2009). The mean age of air can be inferred from SF 6 measurements only if a single entry point can be assumed. As air masses observed during this study, i.e., the WISE campaign, are strongly influenced by transport of air from the NH midlatitudes and tropics, a direct derivation of mean age is not possible. Instead we use so-called lag time here, i.e., the time lag between the occurrence of a certain SF 6 VMR in the UTLS with respect to the time when this mixing ratio was observed in the global mean.
In addition, SF 6 data are smoothed using a local correlation with the CFC-12 data (an atmospheric transport tracer) which has a higher precision. A total of 10 data points surrounding the measurement on both sides are used to determine the local correlation for each data point. This procedure retains the local information without introducing an offset to SF 6 mixing ratios and removes some instrumental scatter. The lag time derived in this way has an overall precision better than 0.3 years and an estimated accuracy of 0.6 years. The global mean SF 6 VMR used as a reference is lower than the NH VMRs. In particular, close to the TP, observed mixing ratios are often higher than those in the reference time series, resulting in negative lag times. The second channel measures bromocarbons and other halogenated species with cryogenic enrichment (Obersteiner et al., 2016), gas chromatographic separation, and mass spectrometric detection with a time resolution of 4 min (Sala et al., 2014). Atmospheric mixing ratios of CH 3 Br and the halons H-1301, H-1211, and H-2402, as well as the short-lived bromocarbons (CH 2 Br 2 and CHBr 3 ), are reported on the scales used in the AGAGE network (Prinn et al., 2018). Mixed bromochlorocarbons (CH 2 BrCl, CHBr 2 Cl, and CHBrCl 2 ) and H-1202 are calibrated based on intercalibration experiments with the UEA. Due to a decrease of CH 3 Br in the calibration gas and poor signal-to-noise ratios for this compound, a correlation between H-1211 and CH 3 Br measured during the PGS (three-part mission: POL-STRACC (Polar Stratosphere in a Changing Climate), GW-LCYCLE (Investigation of the Life cycle of gravity waves), and SALSA (Seasonality of Air mass transport and origin in the Lowermost Stratosphere)) campaign has been used. This correlation leads to a scaling of 12.5 % so as to arrive at typical tropospheric mixing ratios of CH 3 Br observed in the NH in 2017, and then these are applied to the data of the current study. Typical detection limits for the different substances during WISE are given in Keber et al. (2020). The 1σ scatter (the standard deviation) of Br org measurements in a 10 K potential temperature interval directly below the TP is ±0.81 ppt (4 %), which includes the precision of the measurement, as well as the atmospheric variability (Sala et al., 2014). Table 3. Bromine boundary conditions near the TP from various research studies for the Ex-TP and the tropical UT (Trop-UT) used to initialize the TOMCAT/SLIMCAT model. The base run uses the Ex-TP bromine boundary conditions along the entire TP for all latitudes. There are two further runs with elevated TP bromine initializations in the tropics (30 • S-30 • N, run 1 and run 2), while the extratropics are still set to the Ex-TP bromine boundary conditions as in the base run.

UMAQS measurements of N 2 O and CO transport tracers
The University of Mainz Airborne QCL (quantum cascade laser) Spectrometer (UMAQS) is operated by the University of Mainz. It simultaneously measures nitrous oxide (N 2 O) and carbon monoxide (CO) from the HALO aircraft. The instrument is in situ-calibrated against secondary standards, which are compared to NOAA primary standards before and after the campaign. It uses direct absorption of infrared radiation at 2200 cm −1 using a Herriott cell with a 76 m light path at 50 hPa. Measurements are recorded with a frequency of 1 Hz, and the total uncertainty during this study was determined to be 0.18 ppb for N 2 O and 0.9 ppb for CO. These tracers are used for the interpretation of air mass transport around the UTLS during this study. For more details, refer to Müller et al. (2015) and Kunkel et al. (2019).

Lagrangian modeling by CLaMS
The Chemical Lagrangian Model of the Stratosphere (CLaMS) is developed at Forschungszentrum Jülich in Germany. It is a Lagrangian CTM (McKenna et al., 2002a, b;Pommrich et al., 2014, and references therein), which utilizes wind and temperature data of the ECMWF (European Centre for Medium-Range Weather Forecasts) from ERA-Interim reanalysis data (Dee et al., 2011). CLaMS is particularly good at portraying the gradients of trace gases and the transport of air parcels, especially in the region of transport barriers, for example near the extratropical tropopause (Ex-TP) due to its Lagrangian nature (e.g., Vogel et al., 2011Vogel et al., , 2016. The simulations are initialized with Aura-MLS (Microwave Limb Sounder) measurements of O 3 , CO, HCl, and N 2 O, along with ACE-FTS (Atmospheric Chemistry Experiment -Fourier Transform Spectrometer) satellite data tracer-tracer correlations of N 2 O and CH 4 , CFC-11, CFC-12, and NO y . NO y is the sum of the gas-phase NO, NO 2 , 2×N 2 O 5 , HNO 3 , ClONO 2 , and HNO 4 . Further details of the initialization procedure are described in Grooß et al. (2014). The simulation used here starts on 1 May 2017 and has a horizontal resolution of 100 km. The bromine chemistry in the simulations is set up to optimally reproduce reactions in the stratosphere but lacks accuracy for photochemical processes in the troposphere as the heterogeneous reactions of brominated species on ice and liquid water are not yet implemented . Also, only long-lived Br inorg y is part of the simulation, and bromine compounds from VSLS are considered to already be converted to Br inorg y . This may cause offsets in Br inorg y , but nevertheless, ratios like BrO/Br inorg y are reasonably determined by this method.
Trace gas curtains along the flight tracks have been derived for each flight of the WISE campaign. These curtains are defined by the latitude, longitude, and time of the flight track for many altitudes. CLaMS model information for these points was achieved using back-trajectories to the model output time of the previous day at 12:00 UTC. The chemical composition from the 3-D CLaMS simulation is then interpolated to the trajectory endpoints; subsequently the CLaMS chemistry is calculated forward in time along these trajectories to determine the chemical composition along the given curtain. The CLaMS curtains has a 1 min temporal resolution (corresponding to ∼ 13 km) and 41 pressure levels between 10−1000 hPa. For the present study, CLaMS curtains are used as input for the RT simulations for the mini-DOAS data analysis (see Sect. 2.1). These simulated trace gas mixing ratios along the flight track are used for direct comparisons with the measured data of O 3 , NO 2 , and BrO, and the simulated BrO/Br inorg y partitioning coefficient is used to infer measured Br inorg y from mini-DOAS BrO data. In previous studies, different kinds of artificial tracers have been implemented in CLaMS to identify the transport pathways and to quantify the transport of air masses from defined domains/regions to other locations of the atmosphere considering advection and mixing (e.g., Vogel et al., 2011Vogel et al., , 2019. To compare the WISE measurements with the CLaMS artificial tracers, the simulations are interpolated in space and time to the position of the HALO flight track using CLaMS tra-jectories (started every second along the flight track). In this study we use two different kinds of artificial tracers. The first kind is an artificial air mass origin tracer initialized on 1 May 2017 approximately 5 months prior to the WISE campaign. The air mass origin tracers mark nine distinct 3-D domains in the entire model atmosphere: the tropical troposphere, upper and lower TTL, tropical pipe, midlatitudinal troposphere and LMS, polar troposphere and LMS, and lastly the stratospheric background. These are used to study vertical and horizontal transport of these air masses to the LMS in the NH within 5 months. Divisions of the air mass origin domains are described in further detail in Table 1 and Fig. 4 of Vogel et al. (2011).
The second kind is an artificial tracer that marks specific regions in the CLaMS model boundary layer referred to as surface emission tracers, and these are used here to study the transport of young air masses into the LMS in the NH. In contrast to the air mass origin tracers discussed above which are initialized once on 1 May 2017, the different surface emission tracers are continuously released (every 24 h) in the model boundary layer (2−3 km above the surface following orography) from the start of the simulation on 1 May 2017 (for more details, see Vogel et al., 2015Vogel et al., , 2019. The surface emission tracers are divided into 24 surface regions (see Fig. 9 below) defining relatively small regions in southeastern Asia to study in particular the influence of air masses from the Asian monsoon and its adjacent tropical regions on transport into the LMS. Moreover, in this study, the entire Earth's surface is covered by surface emission tracers. For some regions, coastlines or orography (both provided by ECMWF) are used as criteria to define the boundaries between different regions. The same setup for the surface emission tracers was already used for WISE measurements in Wetzel et al. (2021).

Chemical transport modeling by TOMCAT/SLIMCAT
The TOMCAT/SLIMCAT (hereafter TOMCAT) 3-D offline global CTM (Chipperfield, 1999(Chipperfield, , 2006 is used here for the interpretation of the transport of trace gases from the tropical UT/TTL and Ex-TP regions into the LMS and to assess the corresponding O 3 loss. Similarly to CLaMS, the TOM-CAT model is driven by meteorology (e.g., horizontal winds, temperature, surface pressure, and convective mass fluxes) from ECMWF ERA-Interim reanalysis (Dee et al., 2011), with no feedback of the chemistry on the dynamics. In the setup used here, the model employs a hybrid vertical coordinate with terrain-following sigma levels near the surface and pressure levels at higher altitudes. Large-scale vertical motion is calculated from the divergence of the horizontal winds. In our simulations, the model uses a detailed stratospheric chemistry scheme TOMCAT (e.g., see Chipperfield et al., 2018) with photochemical information from JPL-2015 (Sander et al., 2011). The O 3 loss reactions are described in Werner et al. (2017) Appendix A. The simulations used here were initialized from a multidecadal control simulation which started in 1977 (Dhomse et al., 2019). Output from this run on 1 January 2016 was used to initialize a series of sensitivity runs. These simulations calculated the chemistry and dynamics in 15 min time increments over a high horizontal resolution of 1.2 • × 1.2 • with 36 unevenly spaced vertical levels with an altitude range from the surface to around 63 km. The simulations continued until the end of the WISE campaign in late October 2017, i.e., 22 months' integration time.
Initial constraints for the brominated species in the UT are given in Table 3. We have performed three different TOM-CAT sensitivity runs from January 2016: a base run followed by two runs with elevated bromine constraints. The base run initializes the model with constant bromine contributions along the entire TP with the Ex-TP boundary conditions of [Br tot ] = 19.6 ppt. The additional two runs continue to use the Ex-TP conditions in the extratropics, while elevated bromine in the form of Br VSLS and Br inorg y is initialized in the tropical UT for two differing scenarios. Run 1 uses Br VSLS constraints from Navarro et al. (2015) and Br VMRs from the current study in the tropical UT/TTL, resulting in [Br tot ] = 21.7 ppt. These runs, with differing bromine VMRs in the tropical UT but with constant Ex-TP constraints, simulate different air masses being transported into the LMS. Comparison of the model runs allows for the diagnosis of the impact of the differing bromine levels on the O 3 levels in this region.

Measurements: WISE campaign
For the present study, we analyze simultaneous measurements of bromine collected during 15 flights on the German HALO research aircraft during the WISE campaign based from Shannon, Ireland. Specific topics of interest during WISE include the transport pathways, timescales, and mixing processes affecting the late summer and fall in the UTLS. A particular focus was the role of the transport of air from the TTL and Asian monsoon region in H 2 O and trace gas budgets, including VSLS and their product gases in the LMS, as well as small-scale processes leading to irreversible exchange across the Ex-TP by Rossby wave-breaking events (Homeyer and Bowman, 2013).
The research flights in this study, RF02-RF16, were conducted from 13 September until 21 October 2017 with the majority of the ∼ 140 flight hours probing the extratropical UTLS in midlatitudes to high latitudes over the North Atlantic, Norwegian Sea, and northwestern Europe. The flight tracks are shown in Fig. 1, with a few research flights emphasized by dashed lines, to be discussed in more detail below. The flights were mainly during daytime, but three take-offs  (RF01) is not included in this study. The first research flight, RF02, was conducted locally from the home base in Oberpfaffenhofen, Germany, and all subsequent flights were based from Shannon, Ireland, aside from the transfer flights. The flights illustrated by the dashed lines, i.e., RF04 (blue), RF08 (yellow), RF14 (magenta), and RF15 (cyan), are discussed in the text. and six landings were near sunrise/sunset or in the dark. The trace gas measurements of O 3 , N 2 O, CO, and SF 6 are available for all research flights. NO 2 , BrO, and Br inorg y VMRs are also available for all flights, however only at SZA < 89 • . Br org VMRs are not available during four flights (RF07, RF09, RF14, and RF15), and during three other research flights (RF02, RF06, and RF11) Br org VMRs are only available for a portion of the flight time. As such, inferred Br tot measurements are available for 11 of the 15 WISE research flights.
The measurement region ranged from longitudes of 48.4 • W-15.7 • E and latitudes of 37.9-74.4 • N. In equivalent latitudes (Butchart and Remsberg, 1986), the investigated air masses ranged from those being recently transported from the tropics (∼ 6 • S) up to 86 • N. The equivalent latitude describes an enclosed area relative to the area of the globe of specific potential vorticity on a given potential temperature contour and is a useful quasi-Lagrangian coordinate for the interpretation of tracers in the stratosphere (Pan et al., 2012).
The majority of the flight time was spent at altitudes between 8.5−14.5 km, allowing for measurements in the UTLS and crossing the TP frequently. The maximum potential temperature reached was = 410 K, allowing for observations throughout the LS.
Two flights are chosen as examples. One has typical Br tot mixing ratios near the campaign LS average, along with a few segments observing elevated Br tot in the LMS (RF04; Fig. 1, dashed blue line). The other has elevated Br tot from former tropical UT air masses (RF08; Fig. 1 dashed yellow line).
The first sample flight, RF04, on 20 September 2017, consisted of two different length, but overlapping, rectangular shuttles northeast of Ireland up the Scandinavian coast, over the Norwegian Sea, and only briefly crossing land masses over the United Kingdom and Ireland (see Figs. 1 and 2). The flight track reached a maximum latitude of 69.4 • N and longitude of 11.2 • E. The final stretch back to Shannon was flown at the highest altitude during this flight of 14.35 km and reaching a maximum potential temperature of = 399 K,  Fig. 2a. The TP was crossed a few times, but the majority of the flight time was spent in the LMS. The mean lag time of the air masses ranged from −1.08 to 1.85 years (Fig. 2f). The negative mean air mass lag times are due to higher SF 6 VMRs near the NH source regions (com-pared to the global mean) and in particular in the upper levels of the tropical TP region. This flight is chosen as a sample flight of Br tot mixing ratios near the LS campaign average, as well as flight segments within the LMS region with elevated bromine, indicated by the dark gray shading discussed in Sect. 4.2.
The second sample flight, RF08, on 1 October 2017, probed the remnants of Hurricane Maria. This flight consisted of two overlapping rectangular shuttles southwest of Ireland over the Celtic Sea and North Atlantic Ocean (see  Fig. 1). Panels (a-g) are as given in Fig. 2. Light gray shading indicates flight sections in the tropical high -PV low region ( > 355 K and PV < 2 PVU, i.e., former tropical UT/TTL air), and the dark gray shading indicates flight sections in the HBrR as discussed in the text. At SZA > 89 • , mini-DOAS measurements (NO 2 , BrO, and Br inorg y ) are unavailable due to the lack of skylight. Figs. 1 and 3). The latitudes ranged from 45.8−63.0 • N, while the longitudes of the rectangular patterns ranged between 15.0−8.6 • W. The highest altitude during this flight was 14.36 km. The maximum potential temperature reached was = 398 K, which also corresponds to the furthest distance above the WMO TP with = 84 K. As during all WISE flights, the TP was crossed several times shown in Fig. 3a, with the mean lag time ranging from −1.10 to 1.96 years (Fig. 3f). Large sections of this flight observed subtropical as well as former tropical UT/TTL air masses. These air masses transported from the tropics with high potential temperature ( > 355 K) and low potential vorticity units (< 2 PVU) are from here on called tropical high -PV low air, indicated by the light gray shading. The different mixing regimes observed between air masses during this flight are discussed in detail by Schäfler et al. (2021). This flight is chosen representing the variety of air masses ob-served during WISE, resulting in elevated Br tot mixing ratios, largely due to the more abundant VSLS in the tropical high -PV low air. Figures 2 and 3 provide overviews of the O 3 measured by FAIRO, CO measured by UMAQS, the organic bromine measured by GhOST-MS, and NO 2 and BrO measured by mini-DOAS. The modeled O 3 , NO 2 , and BrO from the CLaMS collectively vertically shifted and thereby adjusted to match in situ-measured O 3 (refer to Sect. 2.1) curtains show excellent agreement to measurements, particularly for the first two gases. Additionally, the CLaMS-calculated ratios of BrO/Br  flected in Br tot fluctuations, but also in the range of air mass lag times, which vary between −1.05 and 2.33 years. Complementary measurements and insight for a suite of sources gases and studies of the dynamics, potential vorticity, baroclinic waves, and isentropic mixing of the region have been reported elsewhere for flights during the WISE campaign (e.g., Fix et al., 2019;Kaluza et al., 2019;Kunkel et al., 2019;Hauck et al., 2020;Keber et al., 2020;Wetzel et al., 2021;and Schäfler et al., 2021). More studies from the WISE campaign addressing wave-driven isentropic exchange in the NH UTLS are published in the ACP/AMT/WCD interjournal special issue (see https://acp.copernicus.org/articles/ special_issue1061.html, last access: 1 October 2021).

Results and discussion
4.1 Total bromine as a function of distance from the tropopause Here we report on inferred total bromine, Br tot , in the UTLS for all available data during the WISE campaign in the fall of 2017. Figure 4 shows the additive breakdown of the organic (CH 3 Br -dark blue, halons -purple, and Br VSLS -light blue) and inorganic (BrO measurements scaled by the modeled partitioning coefficient Br inorg y /BrO -red points) bromine components with the potential temperature distance from the WMO TP in kelvin, defined as = − Tropopause . Data points where the SZA > 89 • may introduce additional photochemistry-related uncertainties in the Br inorg y measurements (e.g., Kreycy et al., 2013) and therefore are excluded from the following discussion. The weighted mean Br tot in the UTLS (all measurements, i.e., between = −50 and 88 K) is 19.4±1.3 ppt (the uncertainty being the 1σ variabil- ity of the measurements). Excluding the tropospheric measurements, the LS ( = 0-88 K) has a weighted average of [Br tot ] = 19.2 ± 1.2 ppt, designated by the solid black line in Fig. 4. A more rigorous criterion with which to exclude data such as SZA > 85 • , as well as excluding data based on the modeled partitioning coefficient BrO/Br inorg y < 55 %, only impacts the Br tot weighted mean by an ∼ 0.1 ppt decrease. Br tot ranges from 16.2−23.4 ppt (with the exception of two larger outliers: one from large VSLS VMRs near the surface and another at approximately 35 K above the TP because of an instability of the Br inorg y retrieval). The variability is partly due to the different origins of the air masses and various inferred mean air mass lag times which have a spread of more than 3 years. Inferred Br tot measurements are up to = 88 K above the WMO TP, with the measurements mostly in the LMS during the WISE campaign.
The LS mean of [Br tot ] = 19.2 ± 1.2 ppt within the given uncertainties is in good agreement with previous data (Engel and Rigby et al., 2018). The expected [Br tot ] = 19.6 ppt for 2016 is primarily based on balloon soundings in the middle stratosphere and the expected trend in stratospheric bromine since the last balloon soundings in 2011. The trend is suspected to now be decreasing by −0.15±0.04 ppt Br yr −1 between 2012 and 2016 (see Fig. 1-15 in Engel and Rigby et al., 2018). Furthermore, our results fit well into the expectation based on the total amount of stratospheric Br tot , published earlier by our group from different measurements (e.g., Dorf et al., 2006a, and others, e.g., Hendrick et al., 2008. Within = ±5 K of the TP, the weighted average Br org (light blue points in Fig. 4) is 17.8 ± 1.2 ppt, with bromine mixing ratios several parts per trillion larger below the TP. Within the upper LS region ( = 78-88 K), referred to as LS top , Br org quickly decreases to 13.5 ± 1.2 ppt as Br org is being converted to Br inorg y species. The breakdown of the following bromine mixing ratios for Br org , CH 3 Br, halons, Br VSLS , and Br inorg y at the TP and higher up in the LS top layer is indicated in Table 4. CH 3 Br (contribution (1) to Br tot ) is destroyed with increasing distance into the stratosphere at a slow rate from 7.1 ± 0.2 ppt at the TP to 6.1 ± 0.5 ppt in the LS top layer (refer to Table 4). The tropospheric CH 3 Br (7.3 ± 0.2 ppt) is slightly larger than the 2016 global annually averaged 6.8 ppt as reported in Engel and Rigby et al., 2018) but agrees well with the NH measurements of ∼ 6.75−7.75 ppt from NOAA measurements, which show that NH CH 3 Br is always larger than in the Southern Hemisphere (SH) (Fig. 1-7 of Engel and Rigby et al., 2018). Contribution (2) to Br tot , the four different halons (H-1301, H-1211, H-2402, and H-1202), has a mean VMR of 7.5±0.2 ppt at the TP (in good agreement with the annually averaged tropospheric 7.8 ppt as from 2016 reported by WMO (2018)) and also decreases slowly to 6.5 ± 0.5 ppt by the LS top layer. The larger part of the Br org decrease in the LS is by contribution (3), the VSLS (CH 2 Br 2 , CHBr 3 , CH 2 BrCl, CHBrCl 2 , = 78-88 K 6.1 ± 0.5 ppt 6.5 ± 0.5 ppt 1.0 ± 0.3 ppt 13.5 ± 1.2 ppt 5.8 ± 1.8 ppt TP: = ±5 K 7.1 ± 0.2 ppt 7.5 ± 0.2 ppt 3.1 ± 0.9 ppt 17.8 ± 1.2 ppt 1.5 ± 0.6 ppt and CHBr 2 Cl), due to their short global lifetime of less than 6 months. At the TP, the Br VSLS VMR is 3.1 ± 0.9 ppt (up to 7.2 ppt) while in the LS top layer the Br VSLS VMR decreases to 1.0 ± 0.3 ppt. As such, the accumulated decrease of Br org (contributions (1)- (3) to Br tot ) throughout the LS between the TP and = 88 K above the TP is ∼ 4.3 ppt. Although further into the stratosphere the Br org is destroyed, bromine is converted to Br inorg y and is compensated for in the Br tot budget. Contribution (4) to Br tot , the Br inorg y , near the TP is around 1.5 ± 0.6 ppt (with a range of 0.2−3.3 ppt) and increases to 5.8 ± 1.8 ppt (2.9−10.1 ppt) within the LS top layer (refer to Table 4). The gradual conversion of ∼ 4.3 ppt Br org to Br inorg y by the LS top layer consequently results in minimal (or essentially no) changes of the Br tot budget, which is therefore approximately constant for the entirety of the air masses probed during the WISE campaign. In the LS the weighted mean Br tot ranges from 18.7 ± 1.2 ppt (in the = 5-15 K layer) to 19.9 ± 1.2 ppt (in the = 35-45 K layer). These small changes from the LS Br tot weighted mean of 19.2 ± 1.2 ppt do not result in an obvious trend between the WMO TP and = 88 K above it but rather hint at air masses of differing Br tot , which is discussed in the next sections.

Total bromine as a function of equivalent latitude distribution in the NH
To identify potential source regions and transport pathways of Br tot into the LS, the various gases measured during WISE are displayed in Fig. 5 with respect to potential temperature and equivalent latitude (Butchart and Remsberg, 1986). The VMRs of each trace gas are averaged horizontally in 5 • equivalent latitude and vertically in 5 K grids to remove short-term atmospheric variability. The campaignmean WMO TP varies from ∼ 316−374 K at high to low equivalent latitudes, respectively. The NH distributions of Br org , Br inorg y , and Br tot are shown in Fig. 5a-c, respectively. Only 11 flights have Br org measurements available, resulting in a sparser distribution. Near the Ex-TP Br org , mixing ratios are ∼ 18 ppt, while in tropical high -PV low air, Br org has higher mean VMRs of up to 20.5 ppt, in agreement with previous studies (e.g., Navarro et al., 2015;Werner et al., 2017;Wales et al., 2018). The maximum mean Br inorg y VMR is 8.3 ppt at ∼ 400 K and ∼ 60 • N equivalent latitude, in contrast to the Br org behav-ior. In the troposphere, Br inorg y ranges from ∼ 0−3 ppt. These results agree well with previous studies, such  who reported [Br inorg y ] = 7.66 ± 2.95 ppt between 390−400 K in the subtropical LS and reaching down to 0.5 ppt in the tropical troposphere, as well as others (e.g., Wang et al., 2015;Koenig et al., 2017). A modeling study by Schmidt et al. (2016) reports Br inorg y mixing ratios at the lower boundary of the TTL and Ex-TP of 3−4 ppt and daytime BrO mixing ratios of 1−2 ppt, respectively.
Older air (see Fig. 5d) has generally more Br inorg y and less Br org present, in large part due to the short lifetime of the VSLS. Averaged mean air mass lag times reach above 2 years around ∼ 400 K between 40−80 • N equivalent latitudes and largely coincide with high Br inorg y VMRs, with the exception of directly above the TP at ∼ 85 • N equivalent latitudes. There, a small pocket of older air (with lag times larger than 1 year) has low Br org and Br inorg y . Within ±5 K of the TP, the mean lag time is −0.16 ± 0.37 yrs, with negative averaged lag times down to −0.5 years in the troposphere. Here, the relative change of the mean air mass lag times greater than 3 years is more relevant in the context of our study.
The average Br tot mainly varies between 18−20.5 ppt both in the troposphere and stratosphere (see Fig. 5c); however there are noticeable regions with higher bromine. Of particular interest is the region in the LMS at middle to high equivalent latitudes where Br tot local grid means of up to 21.5 ppt are observed. For the following discussion, this high bromine region (HBrR) is defined between potential temperatures of = 350-385 K and equivalent latitudes of 55−80 • N (designated by the black boxes in Fig. 5). A total of 12 out of the 15 research flights include flight segments within the HBrR, varying from minutes to several hours, with Br tot measurements available during eight flights (RF03-05, RF08, RF11-13, and RF16) over 8 h. A frequency distribution of inferred Br tot is displayed in Fig. 6a showing Br tot in the UTLS (blue) or only in the LS (brown). The distribution of the Br tot in the HBrR (brown) as compared to the rest of the LS excluding the HBrR (blue) shows the characteristically significant increase of bromine in this region (Fig. 6b). The HBrR has a weighted mean of [Br tot ] = 19.9 ± 1.2 ppt. The HBrR measurements result in the bromine bulge noticeable in Fig. 4 around =∼ 35 K and are depicted by the right-pointing (gray-and black-edged) triangles. Inferred Br tot measurements within the HBrR that are larger than the LS weighted mean by at least 1σ are designated by HBrR *  . Bromine-rich former tropical UT/TTL air masses at > 355 K and PV < 2 PVU are labeled high -PV low (shaded red box). The Ex-TP (shaded blue box) is defined here as ±5 K from the local TP between 35−55 • N. The transport of bromine from these two regions contributes to the high bromine region (HBrR; shaded light purple box) in the LMS, located between 55−80 • N and = 350-385 K. The Br tot -inferred measurements larger than the LS mean by more than 1σ within the HBrR are indicated by HBrR * (shaded dark purple oval). The dark red arrows show regions of strong convection transporting surface air masses (from bromine source regions) to the UT. The two main transport pathways of bromine are from the tropical UT/TTL by isentropic transport (light red arrows) and less by STE across the TP in the extratropics (blue arrow). Br tot VMRs are from the WISE campaign as discussed in the text.
(i.e., only the black-edged right-pointing triangles in Fig. 4 and thicker black outlined bars in Fig. 6b). The HBrR * has a Br tot weighted mean of 20.9 ± 0.8 ppt. Here we note that for individual measurements the estimated Br tot in the HBrR air is compatible with the LS average when considering the 1σ variability of the measurements. However, the clustering of the independently measured data, in particular of the HBrR * sample, suggests that the co-added error from the organic and inorganic measurements is actually lower than the stated 1σ variability of ±1.2 ppt, and thus the HBrR is a real and persistent feature over the course of the WISE campaign. HBrR * then represents the elevated bromine transported to the LMS, which is mixed with air of lower bromine content from the rest of the LMS and stratospheric background. The question is therefore as follows: what are the causes of this HBrR * air? Potential sources include other regions of elevated bromine larger than the LS average, such as from the tropical and extratropical troposphere. These regions have younger (tropospheric) air masses which correspond well with the relatively younger air masses observed within the HBrR (local grid mean lag times ranging from 0.4−1.2 years) compared to the surrounding air (see Fig. 5d). Elevated bromine in tropical high -PV low air is observed with a weighted mean of [Br tot ] = 21.6 ± 0.7 ppt, largely in agreement with previous experimental and modeling studies (Wang et al., 2015;Schmidt et al., 2016;Koenig et al., 2017;Werner et al., 2017). The bromine measurements in tropical high -PV low air are from one flight, RF08 (see Fig. 3), over the course of almost 4 h. One other flight (RF14) observed tropical high -PV low air, although Br tot data are missing. Along with the elevated bromine in the tropical high -PV low air mass, measurements of the transport tracers ( Fig. 5e-h) suggest that these air masses were eventually entrained by isentropic transport from the tropical high -PV low region via the shallow branch of the Brewer-Dobson circulation (Levine et al., 2007;Birner and Bönisch, 2011). Near the Ex-TP (35−55 • N and ±5 K from the TP), Br tot has a weighted mean of 19.3 ± 0.9 ppt, indicating a possible second pathway of the air masses entraining bromine into the LMS by STE across the Ex-TP (e.g., Holton et al., 1995;Bowman, 2013 andKunkel et al., 2019). A schematic summarizing these regions of elevated bromine and the transport pathways from the UT into the LMS is shown in Fig. 7. Both possible transport pathways are investigated in the following with observational data from other measured gases (Fig. 5) and in Sect. 4.3 by transport modeling. The observational findings are as follows: a. Elevated Br tot in tropical high -PV low air masses has also been previously observed over the Pacific in 2013/2014, where inferred Br tot was measured to be up to 2 ppt larger in the TTL than expected for the middle stratosphere (Navarro et al., 2015;Werner et al., 2017;Wales et al., 2018). Studies such as those by Fueglistaler et al. (2004), Levine et al. (2007), Gettel-man et al. (2011), andPloeger et al. (2013) discuss the transport of air masses into the LMS that primarily stem from the tropics, in our study initiated by transport of the Asian monsoon (see Sect. 4.3). By the time tropical high -PV low air reaches the LMS in the midlatitudes to high latitudes with transport times of ∼ 0.5−1.5 years, significant portions of Br org have converted to Br c. N 2 O is a good transport tracer due to its long lifetime of 114 years (e.g., Ko et al., 2013). The wellmixed tropospheric air with high N 2 O VMRs (measured by UMAQS) from the tropical high -PV low region penetrates into the LMS above the TP extending to 350−380 K and 50−80 • N (Fig. 5g) (2021) further concluded that the main outflow of the "fast convective chimney" from the Asian monsoon anticyclonic region in the tropics was below 370 K with slower ascent rates above this as well as a smaller outflow from the chimney above 380 K, i.e., in direct agreement with our results of Fig. 5g. d. The SF 6 measurements are also used for analysis of the transport as it constantly increases, and its atmospheric lifetime is in excess of 850 years (e.g., Kovács et al., 2017 andRay et al., 2017). The dominant transport from the tropical UT/TTL into the LMS has been previously investigated with SF 6 and CO 2 measurements (Bönisch et al., 2009). As seen in Fig. 5h, SF 6 VMRs range from ∼ 9.6 ppt below the TP to ∼ 8.8 ppt in the LMS. Similarly to N 2 O, higher SF 6 VMRs (local grid means up to 9.3 ppt) protrude into the LMS at the tropical TP extending to 355−385 K and 50−80 • N, which overlaps with the HBrR.
e. At the same time, CO (from UMAQS shown in Fig. 5f) with a shorter lifetime of 2−3 months (Novelli et al., 1998) has an average of [CO] = 37±5 ppb in the HBrR. The surrounding LS below and above the HBrR (i.e., from the TP to = 350 K as well as > 385 K and between equivalent latitudes 55−80 • N) has an average of 36 ± 12 ppb (influenced by higher CO near the TP) and VMRs reaching down to 22 ppb. The CO mixing ratios seen in the HBrR are considerably lower than in the tropical high -PV low air (average of [CO] = 76±5 ppb) and Ex-TP (average of 66 ± 19 ppb). This indicates that some of the HBrR air might not have recently entered the stratosphere (within the past few weeks) but some months (2−3) prior to the measurements. However, the dominant transport of tropical high -PV low air to the LMS using CO as a tracer was demonstrated by Hoor et al. (2005), the first of its kind. They concluded that transport of air masses to the LMS over Europe in summer and fall is dominated by transport from the tropical UT/TTL up to 55 % above > 25 K and a slightly smaller contribution from the stratospheric background aged air of up to 45 %. The influence from the extratropical troposphere is dominant at the TP up to < 25 K. The study by Bönisch et al. (2009) had similar conclusions.
The persistence of the HBrR observed in many flights during WISE and the already months-long presence and respective mixing generally rules out an origin of this air by some event-like (short duration) recent transport such as by highreaching mesoscale convective systems or another form of STE process. The wide range of lifetimes of these tracers, the distinct change in VMRs across the Ex-TP, and the high VMRs of mainly the two longer lived transport tracers extending from the tropical high -PV low region to the HBrR, as well as the low O 3 observed within the HBrR, suggest that the dominant contribution of Br tot to the LMS is via the isentropic transport from the tropical UT/TTL rather than by extratropical STE (e.g., Holton et al., 1995;Levine et al., 2007;Fueglistaler et al., 2009;Bönisch et al., 2009;Ploeger et al., 2015). Studies such as by Pan et al. (2016) and Garny and Randel (2016) discuss the transport of Asian monsoon air that is isentropically entrained into the LMS by the shedding of small-scale eddies. Vogel et al. (2016) provides more evidence that portions of the Asian monsoon anticyclone air mass are transported horizontally to the southwest (via eddy shedding) entering the TTL and further into the LMS in late boreal summer. A second horizontal transport pathway can occur through filamentation by Rossby wave breaking, east of the anticyclonic event in the extratropics . Specific Rossby wave-breaking events producing filaments irreversibly mix young tropospheric air into older air Table 5. The air mass contributions of the CLaMS surface emission tracers (defined in Fig. 9) in the HBrR and the LS below and above the HBrR (shown in Fig. 11c-d respectively). The HBrR is between equivalent latitudes 55−80 • N and = 350-385 K, while the LS below and above is between equivalent latitudes 55−80 • N from the TP to = 350 K and > 385 K.  Homeyer and Bowman, 2013;Müller et al., 2016 andKunkel et al., 2019). The relative contributions of these transport pathways are investigated in the next section.

Major air mass origins and transport pathways
The transport of air masses from different atmospheric domains into the NH LMS during summer and fall 2017 is simulated by CLaMS using artificial air mass origin tracers released on 1 May 2017, approximately 5 months prior to the WISE campaign (for details, see Sect. 2.5). The advection and mixing processes of the air parcels over the integration time result in the diverse compositions of the LMS air from the various domains of the atmosphere. The fractions of air originating in four of the atmospheric domains most relevant to our discussion are shown in Fig. 8. In the HBrR (marked by the black box), the fraction of air from the tropical troposphere is 51.2 ± 8.3 %, while only 27.1 ± 8.1 % is from the stratospheric background, and smaller contributions are from the lower TTL (7.5 ± 1.4 %), midlatitudinal LMS (7.0 ± 0.6 %), the midlatitudinal troposphere (6.4 ± 1.4 %) and < 1 % from the other domains (see also Fig. 11a). In contrast, the air masses in the LS below and above the HBrR (ranging from equivalent latitudes 55−80 • N between the TP and = 350 and > 385 K) consist of 42.6±9.7 % air from the tropical troposphere, 36.1±10.3 % from the stratospheric background, 8.3 ± 1.4 % from the lower TTL, 7.0 ± 0.6 % from the midlatitudinal LMS, 5.2 ± 2.4 % from the midlatitudinal troposphere, and < 1 % from the other domains (see Fig. 11b). The uncertainties stated (1σ standard deviation) indicate the variety of air masses mixed together in each domain.
Over the 5 months of CLaMS simulation prior to the HALO measurements, the influx of tropospheric air to the HBrR is almost 10 % more than the surrounding LS be-  . CLaMS artificial surface emission tracers by region based on previous simulations described in Vogel et al. (2015Vogel et al. ( , 2019, with higher resolution regions at coastal boundaries mainly in southeastern Asia (INO, IND, NIN, BoB, Neast, and ECH), as well as in other regions (TIB, NEA, NAF, TAO, NAM, NEP, and NAO). Adapted from Wetzel et al. (2021), where these defined regions were also used for the analysis during the WISE campaign. low and above the HBrR. Although this difference is near the 1σ variability, the distributions of air mass origins from the HBrR compared to the LS below and above have distinct shifts, especially for the tropical and midlatitudinal troposphere and stratospheric background domains (not shown). In particular, this increase of tropical tropospheric air (+8.6 %) and to a lesser degree extratropical tropospheric air (+1.2 %) in the HBrR results in additional transport of air masses from these source domains with elevated bromine compared to the surrounding LS (as shown in Fig. 7). Fig-ure 7 provides a schematic of the relative contributions of tropical high -PV low air vs. the Ex-TP bromine transported to the LMS. The elevated bromine observed in tropical high -PV low air has air fractions of up to 92 % (with a mean of 84 ± 2 %) originating from the tropical troposphere (Fig. 8b). As expected, air from the tropical troposphere is transported into the tropical high -PV low region during the last 5 months caused by the large-scale circulation. Therefore, the elevated bromine measured in tropical high -PV low air during WISE can be categorized as air masses largely originating in the tropical troposphere as defined by the CLaMS origin tracer. Yan et al. (2019) also showed that similar contributions of air from the 350−360 and 370−380 K layers of 18 % and 22 %, respectively, are transported into the NH LS, with only small air mass fractions continuing vertically into the tropical pipe. Their findings are in agreement with those of von Hobe et al. (2021).
Large bromine sources, particularly the VSLS, are from tropical marine regions, mainly in coastal areas (Engel and Rigby et al., 2018). Therefore, surface emission tracers (for details see Sect. 2.5) from CLaMS simulations are also used to infer the contribution of young air masses (< 5 months) from different surface regions to the UTLS (see Fig. 9). The surface emission regions with the largest contributions to the HBrR during the WISE campaign flights are shown in Fig. 10. The sum of the surface emission tracers (i.e., air masses younger than ∼ 5 months) contributing to the HBrR accounts for 37.7 % of the air, while the LS regions below and above the HBrR both receive less surface air, with an average of 30.3 % (see Fig. 11c-d, respectively, as well as Table 5). Therefore, the larger tropical tropospheric air contribution to the HBrR compared to the surrounding LS as seen in the air mass origin tracers is largely from recent surface emissions rather than from the free troposphere.
The CLaMS simulations show that surface emissions from the Asian monsoon region (AMR; consisting of NIN, IND, INO, TIB, ECH, and BoB; refer to Fig. 9) and from its tropical adjacent region (TAR; consisting of Wpool, SEA, TWP, TEP, NAF, and Neast) have the strongest contributions of the various surface emission tracers to the LMS in late boreal summer and early fall. Maximum contributions of the AMR and TAR surface emissions are delivered to equivalent latitudes between 25−45 • N and from = 325-390 K (see Fig. 10a-b). The grouped emission tracer contributions to the HBrR and surrounding LS can be found in Table 5, and for a breakdown of all surface emission tracers, see Fig. 11c-d. The AMR accounts for 14.8 % of the air in the HBrR, while air originating from the TAR contributes a further 15.2 %. Together, the AMR and TAR make up 30.0 % of the air reaching the HBrR, which is 79.8 % of all surface air masses. The LS air masses below and above the HBrR receive ∼ 1/4 less surface (younger) air compared to the HBrR, mainly because the AMR and its TAR contribute less. This is in general agreement with results of CLaMS simulations using a previous version of surface emission tracers for 2012 and 2008 that found a flooding of the northern Ex-LS in boreal summer with young air masses from the region of the Asian monsoon and from the tropical Pacific (Vogel et al., , 2019. Similarly, the Asian monsoon air masses impacting the LMS was observed by Müller et al. (2016). Furthermore, this feature is in general agreement with findings of Orbe et al. (2015), who showed that in boreal summer and fall, ∼ 20 %-30 % of the midlatitudinal LMS air at 100 mb originated from the Asian planetary boundary layer (slightly further north than defined here). The outflow of the AMR into the LMS (see Fig. 10a) has also been recently shown in a study by von Hobe et al. (2021); i.e., the main outflow is below 370 K with weaker outflow up to 380 K. Levine et al. (2007) previously also concluded that vertical transport from the tropical boundary layer into the TTL is strongest over the Indian Ocean, Indonesia, and western Pacific Ocean through convection. They further concluded that the main outflow (∼ 95 %) of air from the TTL is transported below 380 K into the LMS, while only ∼ 5 % is transported further upwards into the stratospheric "overworld".
A subgroup of regions from the AMR and its TAR belonging to the tropical marine environment (TME; defined by SEA, Wpool, BoB, INO, TWP, and TEP) in southeastern Asia alone contribute 17.7 % to the HBrR, which is 47.0 % of the air masses from the surface. The maximum TME air mass contribution is in the subtropics to midlatitudes between 25−55 • N from = 320-390 K, but air masses with elevated TME contributions extend to the tropical UT/TTL and the HBrR (see Fig. 10c). This suggests entrainment of TME air masses crossing the TP around 30 • N with further transport to the tropical UT/TTL and LMS, in agreement with findings of Orbe et al. (2015). The surrounding LS air masses below and above the HBrR receive 13.8 % from the TME, i.e., a 3.9 % decrease compared to the HBrR. The variations of TME contribution per flight in the UTLS range from 17.3 %-30.1 % at flight altitude (examples are shown in Figs. 2f, 3f, and 12c-d).
The increased contribution of younger surface air to the HBrR compared to the surrounding air masses below and above has a significant influence on the trace gas composition, in particular, considering the majority of the surface air masses are transported from regions with large biological activity and known bromine sources (Engel and Rigby et al., 2018), i.e., the tropical marine environment below the Asian monsoon anticyclone and its tropical adjacent regions. This highlights the influence of surface emissions from southeastern Asia on pollutants and in particular bromine in the northern LMS during boreal fall. Adcock et al. (2021) observed enhanced CH 3 Br VMRs (up to 9 ppt) as well as CH 2 Br 2 and H-1211 VMRs (up to about 0.75 and 3.5 ppt, respectively; comparable to Engel and Rigby et al., 2018) at the TP in the Asian monsoon region in 2016/2017.
Using ERA-Interim reanalysis from the years 2007/2008, Vogel et al. (2019) showed that surface air masses can reach 360 K in a few days and reach 360−460 K within a few The gray outline marks the tropical high -PV low region, and the solid black line is the average WMO TP. months in the tropics and subtropics due to the strong vertical spiral uplift associated with the Asian monsoon anticyclone including its tropical adjacent air masses. Using data from the ERA-Interim/ERA5 reanalysis, Bucci et al. (2020) showed that air from Nepal in July/August 2017 could rise through convection in the Asian monsoon anticyclone region with transport times up to the TP > 17 km to be ∼ 20 d. Similarly, Ashfold et al. (2012) reported a transport time of less than 15 d of air masses from the western Pacific boundary layer (< 1 km) up to the lower TTL (∼ 15 km). The fast transport (on the order of days) from the Asian tropical planetary boundary layer to the TTL allows even the short-lived trace substances from the southeastern Asian region to reach the UTLS (Orbe et al., 2015).
The impact of maritime boundary sources on the chemical composition of the Asian monsoon anticyclone by convective uplift in tropical cyclones in the western Pacific is discussed by, e.g., Vogel et al. (2014) and Li et al. (2017Li et al. ( , 2021. A study by Liang et al. (2014) concludes that deep convection transports ∼ 7.8−8.4 ppt brominated VSLS from the marine boundary layer of the tropical Indian Ocean, tropical western Pacific warm pool, and the Pacific coast of Mexico to the LMS. These brominated VSLS are half in the form of source gases, and the other half consist of product gases such as BrO, and only ∼ 10 % is lost due to wet scavenging after transport to the TP. Hamer et al. (2020) similarly found that inside the convective system, CHBr 3 (one of the VSLS) is the most efficient bromine gas to reach the UT as > 85 % of surface CHBr 3 is transported to the tropical UT. In addition, they simulated the presence of 1 ppt of inorganic bromine outside the convective system near Borneo, i.e., in the tropical adjacent regions, while inside the convection system inorganic bromine (< 0.1 ppt) is almost completely depleted due to washout. Therefore, the solubility of the gases and local surrounding conditions may also impact the delivery to the TTL and finally the Ex-LS. This is partially when tropospheric O 3 is low and prevents the soluble forms of bromine from reacting and converting to insoluble inorganic bromine compounds.

Contributions to UTLS bromine from recent tropical cyclones
Interestingly, emissions from Central America (CAM) also contributed to elevated Br tot in the tropical high -PV low air masses with maximum CAM air mass fractions around ∼ 345−370 K during WISE (see Fig. 10d). There are two flights with particularly large air mass fractions from CAM emissions observed during the campaign, i.e., RF08 and RF14 (discussed below), with flight segments in tropical high -PV low air. During both flights, remnant air masses of two mesoscale convective systems in the Central American tropics were probed. For reference for the following discussion, the average contribution of emissions from CAM (excluding RF08 and RF14) per flight ranges from 4.3 %-8.9 % at flight altitude. Generally, the largest CAM emissions during the WISE campaign are seen in tropical high -PV low air with a strong decrease in the LS. The average CAM emissions in the LS range from 2.7 %-6.2 %, with the exception of air masses during RF14 and RF15 (also discussed below). Although larger CAM air mass fractions contribute to elevated bromine in tropical high -PV low air, the HBrR only has 3.3 % of the air fraction from CAM, similarly to the LMS, which has mostly low CAM air mass contributions.
The first flight with large segments of elevated contributions of air masses from CAM was RF08 on 1 October 2017 (Fig. 3f). This flight, off the coast of Ireland over the Atlantic Ocean (up to 15.0 • W), encountered the remnants of Hurricane Maria, which had transitioned into an extratropical cyclone and later dissipated on 2 October 2017 (Pasch et al., 2019). The average CAM surface air fraction along the flight track is 18.0 ± 13.0 %, with the maximum contribution reaching 60.6 % at 20:35 UTC at = 361 K and = −11 K (Fig. 3a, f). However, sections with increased CAM air mass fractions only reached the UT, while the two flight sections breaching into the LS have low CAM emissions. This low-pressure system transported surface air with increased CAM air mass fractions (25 %-45 %), CO, and N 2 O to the UT, bringing with it more brominated VSLS up to [Br VSLS ] = 5.8 ppt (at 17:29 UTC in Fig. 3g), coinciding with flight sections in tropical high -PV low air (light gray  Fig. 3g). Only a short flight segment reached the HBrR, which is indicated by the dark gray shading in Fig. 3, where the CAM air mass fraction is low (∼ 1 %), while the TME air mass fraction is still ∼ 11 %. Further insights into the flight segment between 18:40−20:00 UTC are analyzed in the study by Schäfler et al. (2021). They use H 2 O and O 3 lidar data as tropospheric and stratospheric tracers to discuss three mixing regimes near the tropopause fold indicating the presence of different air masses.
A second tropical cyclone encountered during WISE was Hurricane Ophelia. It made landfall over Ireland on the 16 October 2017, dissipated over Norway 2 d later, and rapidly transported tropical air from Central America to Europe (Stewart, 2018). This is verified from data collected during RF14 on the 15 October 2017 (Fig. 12 left column). A large influence of surface emission tracers from CAM is present, with a flight average of 23.3 ± 11.8 % (see Fig. 12a, c, e). The initial flight section between ∼ 13:20-15:00 UTC just below = 355 K indicates the transport of recently tropical air masses (low equivalent latitudes) with large CAM air mass fractions. Unlike RF08, this low-pressure system transported larger CAM air masses into the LMS. The average influence from CAM air masses in the LMS flight segments is 20.8 ± 5.4 %. These are the largest influxes of CAM emissions into the LMS detected during the WISE campaign. Although this flight does not have organic bromine data available, the variability of air masses probed by the aircraft in the LMS is noticeable in Br  Table 3). Panel (b) shows the relative percentage O 3 between run 1 with elevated Br tot in the tropical UT and the base run. Panel (c) displays the relative O 3 between run 2 and the base run. The solid black line is the campaign-average WMO TP, the black box in the LMS is the HBrR, and the gray outline marks the tropical high -PV low region.
was observed, with high CAM (22.3% ± 4.1 %) and TME (30.4 ± 2.5 %) air fractions. Shortly after, between 17:04-17:46 UTC when the aircraft cruised in the UT (around ∼ −10 K), elevated Br inorg y was observed and reached up to 7.8 ppt. During this event low CO and N 2 O mixing ratios, low CAM emissions (6.2 ± 2.0 %), only slightly lower TME air mass fractions (24.2±3.6 %), and large air mass lag times (up to 1 year old) were observed in the UT. This event characterizes an intrusion of stratospheric air into the UT with Br inorg y > 4 ppt (indicated by the light blue shading). A third flight, RF15, on 19 October 2017 ( Fig. 12 right column), around Ireland and the UK likely still captured former air masses from Hurricane Ophelia. The air of this flight is also characterized by a slightly elevated CAM influence in the LS, mainly with a surface emission of 8.9 ± 4.1 % at extratropical equivalent latitudes (Fig. 12c-d). Only partial organic bromine data are available with slightly elevated VMRs (Fig. 12f). CH 3 Br VMRs up to 7.8 ppt and bromine in halons up to 8.1 ppt are observed in the LMS ( = 22 K), with elevated CAM emissions around 14 % (and TME air mass fractions ∼ 27 %). A similar event is again detected during the initial ascent between 09:25-09:38 UTC (light blue shading) after passing through a small stratospheric intrusion. Elevated Br inorg y > 4 ppt and up to 9.2 ppt was observed in the UT ( = −33.2±3.6 K), while air mass fractions from CAM emissions were low (7.4±1.9 %), and TME emissions were around 25.0 ± 1.8 %.
In summary, the synoptic-weather-system-related contributions from the CAM region evidently may directly impact observed UTLS bromine. The CLaMS simulations indicate the variety of different air masses influencing our observations. Near the tropical TP, high CAM air mass fractions (through rapid vertical transport by hurricanes), as well as high TME air mass fractions (uplifted by local convection associated with the Asian summer monsoon), result in significantly elevated Br tot observed in the former tropical high -PV low air (RF08). However, in the LMS there are lower air mass fractions from CAM but still relatively high TME contributions, which also result in elevated Br tot , i.e., the HBrR during eight flights. Additionally, the two events with higher Br inorg y observed in the UT (shaded light blue sections in Fig. 12), with comparatively lower CAM emissions and low CO and N 2 O mixing ratios, are of particular interest. These events may indicate that air from the LMS intruded into the UT by a recent STE event, which is discussed in a recent study by Kunkel et al. (2019).

Impact on LMS O 3
The isentropic transport of bromine-enriched air from tropical UT/TTL into the LMS likely has a significant impact on O 3 . In the HBrR, the average O 3 measured is 301 ± 69 ppb (with a range of 153−524 ppb), while the LS below and above has an average of 404 ± 169 ppb (range 73−791 ppb), largely due to the significant increase of O 3 further into the stratosphere that extends into the upper area of the HBrR. However, in the HBrR, lower O 3 VMRs are measured with a slight increase below the HBrR (see Fig. 5e).
In order to investigate the influence of elevated bromine on the Ex-LS O 3 , we conducted three simulations using the 3-D global TOMCAT model with 22 months' integration time (see Sect. 2.6). The base run was initialized with constant bromine (19.6 ppt from the Ex-TP) VMRs along the entire TP. Run 1 and 2 each had elevated bromine boundary conditions in the tropical UT in addition to the Ex-TP boundary conditions from the base run (see Table 3). Run 1 and 2 are initialized with an additional [Br tot ] of 1.2 and 2.1 ppt, respectively, in the tropical UT. The increase in run 1 is mainly due to increased Br inorg y , while the increase of Br tot in run 2 is largely due to additional Br VSLS .
These increases of bromine in the tropical UT lead to additional destruction of O 3 in the LMS. The simulations can be used as a measure of the O 3 sensitivity due to transport of bromine-enriched air from the tropical high -PV low region into the LMS by the lower branch of the Brewer-Dobson circulation. The TOMCAT O 3 VMR distribution over the NH from the base run is shown in Fig. 13a and is comparable to the FAIRO-measured O 3 VMRs (Fig. 5e). The relative changes of TOMCAT O 3 between the base run and either run 1 or 2 are displayed in Fig. 13b-c, respectively. The impact of increasing tropical UT bromine (run 1) leads to 5.0±0.4 % more Br tot in the HBrR compared to the base run, which results in an O 3 change of −2.5±0.2 % (panel b). Run 2 leads to 8.7 ± 0.7 % more Br tot in the HBrR compared to the base run, which results in an O 3 change of −2.8 ± 0.2 % (panel c). In the HBrR, the absolute O 3 mixing ratios change by −6.6 ± 1.1 ppb (run 1) and −7.4 ± 1.5 ppb (run 2) (not shown). The decrease of O 3 is not strictly linear with increasing Br tot , partially due to the nature of the increased bromine (Br inorg y vs. Br VSLS ) differing between run 1 and 2, as well as other feedback loops such as via lower stratospheric HO x chemistry.
The average relative O 3 loss throughout the whole LS is −2.6 ± 0.6 % and −2.7 ± 0.4 % for run 1 and 2, respectively. However, the largest relative O 3 decreases are observed along the tropical TP where there is the largest bromine enhancement due to the initialization. Further significant decreases of O 3 extend to the midlatitudes along the TP. Run 1 has the strongest O 3 decreases up to −5.4 % along the TP in the tropics to 40 • N equivalent latitude (in comparison to the base run). As the bromine increase in run 1 is mainly in the form of Br inorg y , this shows the efficiency and immediate destruction of O 3 by Br inorg y . Run 2 has a maximum O 3 decrease of −3.9 %; however the strongest decreases are not near the tropical TP but rather along the midlatitude TP. As the bromine increase of run 2 is largely in the form of Br VSLS , O 3 is not immediately destroyed because of the time delay for the conversion of Br org to Br inorg y . Although the O 3 destruction is not as strong near the source of increased bromine in run 2 compared to run 1, the O 3 decrease extends further towards the pole just above the TP. The O 3 decrease along the TP up to < 20 K results in −3.2 ± 0.6 % and −3.0±0.4 % for run 1 and 2, respectively (available data only between equivalent latitudes of 22−88 • N). A study by Hossaini et al. (2015) analyzed the impact of increasing VSLS (bromine, chlorine and iodine) on O 3 in the global LS. The model results showed LS O 3 decreases of 4 %-12 %, where the brominated VSLS account for ∼ 85 % of the O 3 destruction. In TTL, the O 3 impact just below 100 hPa was ∼ −8 %, with stronger decreases towards the TP and surface (Hossaini et al. (2015) Supplement  Fig. S1). At 200 hPa at the North Pole, the relative O 3 change was ∼ −5 %, also with stronger decreases at lower altitudes. These are comparable relative O 3 percentage changes to our study when considering only the bromine VSLS impact as well as a significantly larger difference of the VSLS bromine loading (up to 8 ppt) in their model runs. If the transport of source gases into the stratosphere in the tropics increases as climate change progresses, the strongest decrease in O 3 can be expected near the cold-point TP in agreement with the simulations here (e.g., Hossaini et al., 2012a;Falk et al., 2017 andZiska et al., 2017).

Conclusions
We report on simultaneous measurements of O 3 , NO 2 , BrO, Br org , Br inorg y , and Br tot , as well as transport tracers (CO and N 2 O) and a tracer for the mean age of air (SF 6 ), from the HALO aircraft during the WISE campaign of fall 2017 in the Northern Hemisphere (NH) upper troposphere and lower stratosphere (UTLS). Additionally, CLaMS simulations of O 3 , NO 2 , and BrO are reported with generally good agreement to the measurements. The majority of the observations vary between 8.5−14.5 km, reaching a potential temperature of ∼ 410 K and focusing on the lowermost stratosphere (LMS). The observations include air masses of the entire NH in equivalent latitudes, indicating geographical distributions as well as allowing for transport interpretation.
Br tot is assessed from simultaneous measurements of all relevant Br org source gases and inferred Br inorg y . The Br org source gases include CH 3 Br, the four halons, and the five major brominated, very short-lived substances. Br inorg y is inferred from remotely measured BrO scaled by the modeled BrO/Br inorg y partitioning. The resulting lower stratosphere (LS) weighted average is [Br tot ] = 19.2 ± 1.2 ppt. Our observations largely agree with previous measurements of Br tot in the UTLS (e.g., Engel and Rigby et al., 2018) but emphasize some of the causes for the variability in Br tot in the extratropical lower stratosphere (Ex-LS). Air mass lag times inferred from SF 6 measurements vary by over 3 years.
Major insights into the causes of the Br tot variability have been gained from studying the geographical distribution of the brominated gases as well as transport tracers as a function of equivalent latitude, both from observations (Fig. 5) and transport modeling (Figs. 8 and 10). The key findings of our study are summarized in Fig. 7. Elevated Br tot is observed in a persistent region extending into the LMS, i.e., a subsection of the high bromine region (HBrR * ) with [Br tot ] = 20.9 ± 0.8 ppt. The air mass origin tracers modeled by CLaMS agree reasonably well with our observations. They indicate that the HBrR in the LMS has 51.2 % of air originating from the tropical troposphere and 6.4 % from the midlatitude troposphere, which are mixed with only 27.1 % stratospheric background air (as well as contributions from other domains). The LS air masses below and above the HBrR show a decreasing tropospheric contribution while an approximately equivalent increasing stratospheric background contribution. This increase of approximately 10 % young (tropical and extratropical) tropospheric air in the HBrR compared to the LS below and above it is reflected in the bromine-rich air masses of the HBrR.
Additional CLaMS surface emission simulations indicate the majority of surface contributions to the HBrR are from the Asian monsoon region (AMR) and its tropical adjacent region (TAR) and account for 30.0 % of the air mass in the HBrR. Only a small fraction of air (3.3 %) in the HBrR originates from Central America (CAM). The LS regions below and above the HBrR only receive 23.2 % from the AMR and TAR. This shows the efficiency of the vertical uplift of the AMR rapidly transporting surface air from the TAR and specifically the tropical marine environment (TME) into the tropical upper troposphere and tropopause layer (UT/TTL) and further via the lower branch of the Brewer-Dobson circulation into the Ex-LS. This transport pathway in boreal late summer and fall has a direct impact on the trace gas composition and specifically of bromine in the northern LMS. Our study complements previous studies on the importance of the transport of air masses via the lower branch of the Brewer-Dobson circulation (e.g., Hoor et al., 2005;Levine et al., 2007;Bönisch et al., 2009;Vogel et al., 2016 andHauck et al., 2020).
However, during WISE we observed additional causes for varying bromine in the UT. Notably, air masses of high bromine were transported by two hurricanes from CAM into the tropical UT, i.e., the high -PV low region, of which the remnants were later observed to the west of the British Isles on 1 October 2017 (RF08). These large CAM (and TME) contributions led to the bromine-richest air masses observed during WISE, with [Br tot ] = 21.6 ± 0.7 ppt in the high -PV low region. Additionally, our flights captured two stratosphere-troposphere exchange (STE) events where stratospheric air of elevated Br inorg y was recently transported into the UT caused by breaking Rossby waves in the extratropics.
Transport of younger bromine-rich tropospheric air masses into the LMS directly impacts LS O 3 . The consequences of more Br tot in the LS on O 3 are analyzed with TOMCAT model simulations. An enhancement of ∼ 1−2 ppt Br tot in the tropical UT and further in the LS through isentropic transport results in the LS O 3 change of −2.6 ± 0.7 % (average of two elevated bromine scenarios), conveying its sensitivity to bromine. The O 3 change in the 20 K layer directly above the tropopause (TP) is −3.1 ± 0.7 %.
The continuous nature of our measurements covering a large range of air mass lag times, even for individual flights, offers several advantages over previous mainly balloon and some aircraft-related studies on Br tot . This includes an improved space and time resolution as well as a higher precision and accuracy of inferred Br tot . Another major advantage of the aircraft measurements is the ability to display the widespread behavior of Br org , Br inorg y , and Br tot in the investigated air masses. The multi-pathway transport of bromine (as well as of other gases) into the northern LMS via the major branches may well be very specific for the region and season (midlatitudes in boreal fall) investigated in the present study. However, it lends evidence that bromine as well as other O 3 -harming gases is not simply injected into the stratosphere from a single source region but that a manifold of different pathways may act (e.g., Holton et al., 1995;Homeyer and Bowman, 2013;Ploeger et al., 2015;Kunkel et al., 2019). In this respect, further probing the TP region and LS for Br tot and other O 3 -depleting species is required. Ultimately, the detected variability of Br tot in the UTLS may not give a firm answer on Br tot and its trend in the middle and upper stratosphere. Therefore, probing the middle and upper stratosphere for Br tot (and other O 3 -depleting substances) by ground-based, balloon-based instrumentation, as well as satellites, is still required in the future.
Author contributions. MKR, TH, FK, BS, and KP operated the mini-DOAS instrument; AE, TK, and TS provided the GhOST measurements; AZ and his group provided FAIRO O 3 data; PH and VB provided the UMAQS data; JUG and BV provided the CLaMS simulations; and MPC provided the TOMCAT model results. MKR performed the data analysis and wrote the paper with contributions from KP.
Competing interests. The authors declare that they have no conflict of interest.
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 "WISE: Wave-driven isentropic exchange in the extratropical upper troposphere and lower stratosphere (ACP/AMT/WCD inter-journal SI)". It is not associated with a conference. are given to the scientific coordinators of the WISE campaign, Peter Hoor (University of Mainz) and Martin Riese (FZJ). For more information about the WISE campaign, visit https://www.wise2017.de/ (last access: 1 October 2021). The authors gratefully acknowledge the computing time for the CLaMS curtain simulations granted on the supercomputer JURECA at Jülich Supercomputing Centre (JSC) under the VSR project ID JICG11. The TOMCAT/SLIMCAT simulations were performed on the Leeds ARC and ARCHER national HPC machines. The authors acknowledge EARLINET for providing aerosol LIDAR profiles available at https://data.earlinet. org/ (last access: 10 September 2018). The research leading to these results has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement no. 654109 and previously from the European Union Seventh Framework Programme (FP7/2007(FP7/ -2013 under grant agreement no. 262254. Additionally, the authors acknowledge the use of SAGE III -ISS aerosol data, which were obtained from the NASA Langley Research Center Atmospheric Science Data Center. Financial support. This research has been supported by the Deutsche Forschungsgemeinschaft (grant nos. HALO-SPP 1294, PF 384/16, PF 384/17, PF 384/19, EN 367/11, EN 367/13, EN 367/13-1, EN 367/14-1, HO 4225/7-1, HO 4225/8-1, and VO 1276/5-1), the Bundesministerium für Bildung und Forschung (grant nos. 01LG1908D, 01LG1908B, and 01LG1908C), and the National Centre for Earth Observation (grant no. NE/R001782/1).
Review statement. This paper was edited by Mathias Palm and reviewed by Qing Liang and one anonymous referee.