Redistribution of total reactive nitrogen in the lowermost Arctic stratosphere during the cold winter 2015/2016

During winter 2015/2016 the Arctic stratosphere was characterized by extraordinarily low temperatures in 20 connection with a very strong polar vortex and with the occurrence of extensive polar stratospheric clouds. From mid of December 2015 until mid of March 2016 the German research aircraft HALO (High Altitude and Long–Range Research Aircraft) was deployed to probe the lowermost stratosphere in the Arctic region within the POLSTRACC (Polar Stratosphere in a Changing Climate) mission. More than twenty flights have been conducted out of Kiruna/Sweden and Oberpfaffenhofen/Germany, covering the whole winter period. Besides total reactive nitrogen (NOy), observations of nitrous 25 oxide, nitric acid, ozone and water were used for this study. Total reactive nitrogen and its partitioning between gasand particle phase are key parameters for understanding processes controlling the ozone budget in the polar winter stratosphere. The vertical redistribution of total reactive nitrogen was evaluated by using tracer–tracer correlations (NOy-N2O and NOyO3). The trace gases are well correlated as long as the NOy distribution is controlled by its gas-phase production from N2O. Deviations of the observed NOy from this correlation indicate the influence of heterogeneous processes. In early winter no 30 such deviations have been observed. In January, however, air masses with extensive nitrification were encountered at altitudes between 12 and 15 km. The excess NOy amounted up to about 6 ppb. During several flights, along with gas–phase nitrification, indications for extensive occurrence of nitric acid containing particles at flight altitude were found. These observations support the assumption of sedimentation and subsequent evaporation of nitric acid containing particles leading to redistribution of total reactive nitrogen at lower altitudes. Remnants of nitrified air masses have been observed until mid 35 of March. Between end of February and mid of March also de–nitrified air masses have been observed in connection with high potential temperatures. This indicates the downward transport of air masses that have been denitrified during the earlier winter phase. Using tracer–tracer correlations, missing total reactive nitrogen was estimated to amount up to 6 ppb. Further, indications of transport and mixing of these processed air masses outside the vortex have been found, contributing to the chemical budget of the winter lower most stratosphere. Observations within POLSTRACC, at the bottom of the vortex, 40 reflect heterogeneous processes from the overlying Arctic winter stratosphere. The comparison of the observations with

February (Manney and Lawrence, 2016). In extended areas of the Arctic stratosphere, temperatures dropped below the existence temperature of ice . With the space-borne Lidar CALIOP on board of the CALIPSO satellite extended regions of polar stratospheric clouds have been observed between 15 and 26 km from December until end of 85 January Voigt et al., 2018). Extensive PSC abundance in the Arctic lead to the activation and repartitioning of chlorine species (Johansson et al., 2018;Johansson et al., 2019;. Trace gas measurements performed with the Aura Microwave Limb Sounder (MLS) showed remarkable denitrification in the polar vortex (Manney and Lawrence, 2016). A finding that was also illustrated by model simulations. With the EMAC atmospheric chemistry-climate model a strong denitrification of 4 to 8 ppb was simulated (Khosrawi et al., 2017). 90 Beginning of March, a major final warming began and the full breakdown of the vortex occurred at the beginning of April (Manney and Lawrence, 2016).
POLSTRACC has been an extensive measurement campaign using the German research aircraft HALO (High Altitude and Long-Range Research Aircraft) as platform for in situ and remote sensing instruments. POLSTRACC was part of a combined mission called PGS (POLSTRACC-GW-LCYCLE-SALSA) including also objectives with respect to gravity 95 waves and to stratosphere-troposphere exchange. One major goal was to study the interrelationship between climate and the polar stratosphere .
The investigation of the reactive nitrogen distribution at the bottom of the polar vortex and the search for nitrate containing particles was a key issue of this mission, which was pursued with in situ and remote sensing instruments .
With the AENEAS (Atmospheric Nitrogen Oxides Measuring System) in situ instrument the total reactive nitrogen 100 distribution was observed in the lowermost stratosphere. Observations of N2O and O3 have been used to interpret the nitrogen oxides measurements. Tracer-tracer correlations are an important tool for studying processes apart from transport and mere gas-phase chemistry.
The observation period provided an unique opportunity to study the lowermost stratosphere over an entire winter period. So, the following questions could be addressed: How does the distribution of reactive nitrogen evolve from late December to 105 mid-March? What influence do heterogeneous processes have on the distribution of reactive nitrogen compounds at the bottom of the polar vortex? This extremely cold winter favoured the formation of PSC particles and their sedimentation. So far, PSC particles containing reactive nitrogen compounds have been found almost exclusively at altitudes above 15 km. Therefore, another goal of these measurements was the search for PSC particles at flight altitude of HALO.
In addition, the observed reactive nitrogen is compared with CLaMS model simulations. It is investigated how the model-110 measurement comparison performs in the different phases of the winter and how well the model describes the different influences of heterogeneous processes on the nitrogen oxide distribution at the lowermost stratosphere.  During the three mission phases 21 science flights (some with intermediate landings) with more than 150 flight hours were performed. Most of the flights have been conducted out of Kiruna in Northern Sweden. More than 70 % of the data have been obtained north of about 60° N. Typical flight altitudes ranged between about 12.5 and 14.5 km in the lower stratosphere. A detailed description of the flights and the whole mission is given in . 125

Measurement Techniques
The HALO research aircraft (https://www.dlr.de/content/en/missions/halo.html) is based on a Gulfstream G550 large business aircraft with a maximum ceiling of about 15 km and range of more than 8000 km. The payload of nearly three tons comprised a set of remote sensing and in situ instruments complementing each other. A detailed description can be found elsewhere . For the present analysis experimental data from several instruments have been used. 130 2.1.1 Total reactive nitrogen (gas-and particle-phase) Total reactive nitrogen (NOy) is the sum of all reactive nitrogen species in the atmosphere namely NO, NO2, HNO3, PAN, HNO2, HNO4, N2O5, ClONO2, and others. During POLSTRACC total reactive nitrogen was measured using the AENEAS (Atmospheric Nitrogen oxides measuring system) instrument. Since 2012 this measuring system has been regularly operated on HALO during several missions Wendisch et al., 2016;Voigt et al., 2017;Lelieveld et al., 2018). 135 The detection of total reactive nitrogen is based on a well-established technique, comprising catalytic conversion and chemiluminescence. NOy species are reduced catalytically on the surface of a heated gold tube to NO e.g. (Bollinger et al., 1983;Fahey et al., 1985). As reducing agent hydrogen is added. Subsequently, NO is detected by a chemiluminescence detector e.g. (Ridley et al., 1974;Kley, 1980;Drummond et al., 1985). At DLR this detector type has also been used for observations from other aircraft like the DLR Falcon (Ziereis et al., 2000b) and the DLR Dornier 228 (Reiner et al., 1999). A 140 modified instrument was used on the Russian research aircraft Geophysica Voigt et al., 2006;. For more than 15 years a NO/ NOy detector identical in construction has been operated on a commercial airliner in the framework of IAGOS-CARIBIC (In-service Aircraft for a Global Observing System: https://www.iagos.org/iagoscaribic/) (Stratmann et al., 2016).
During POLSTRACC, AENEAS was operated with two separate detector channels. Both channels of the instrument were 145 equipped with gold converters allowing the detection of NOy. The two separate NOy-channels were connected to a forwardand aft-facing inlet, respectively. With the backward facing inlet mainly gas-phase total reactive nitrogen is measured. The sampling of particles larger than about 1 µm is discriminated . The overall uncertainty of the total reactive nitrogen measurement depends on the actual ambient concentration. It is about 8 % for volume mixing ratios of 0.5 ppb and about 6.5 % for about 1 ppb (Stratmann et al., 2016;. 150 The forward-facing inlet oversamples particles. The oversampling is caused by the sub-isokinetic sampling of particles. Due to the high ratio between true air speed of the aircraft and the flow velocity inside the inlet line particles are sampled with enhanced efficiency relative to the gas-phase. This approach has been already used during earlier observations, for example from NASA ER-2 (Fahey et al., 2001;Northway et al., 2002), NASA DC-8 (Weinheimer et al., 1998) or the DLR Falcon Ziereis et al., 2004). It was used to investigate the nitrate content of PSC particles as well as that of cirrus 155 ice particles. The enhancement factor depends on the flow ratio as well as on ambient pressure, temperature and particle size.
The corresponding relation was derived by Belyaev and Levin (Belyaev and Levin, 1974) and was adapted for aircraft observations (Fahey et al., 1989;Feigl et al., 1999). The enhancement factor strongly increases with increasing particle diameter and yields a maximum value for diameters larger than about 10 µm. For larger particles the enhancement factor increases only slightly with diameter. 160 Whenever the expression "NOy" is used in this study, gas-phase NOy detected with the backward facing inlet is meant. Total reactive nitrogen observed with the forward-facing inlet comprises gas-phase NOy and enhanced particle NOy (NOy_P) as gas-phase equivalent. It is denoted as NOy_tot. NOy_tot is not corrected for oversampling. Apart from episodes when particles containing nitrogen oxides compounds where sampled with the forward-facing inlet, the NOy signal detected with the two channels agreed within about 7 %. 165 NOy _tot = EF* NOy _P + NOy (1)

EF=enhancement factor
NOy_net is defined as the difference between total NOy and gas-phase NOy. Particulate nitrate NOy_P as gas-phase equivalent can be derived from the difference between the signal obtained with the forward (NOy_tot) and aft-facing inlet (NOy) and the enhancement factor, respectively 170 NOy _P= NOy _net/EF (2)

Nitrous oxide
The measurement of nitrous oxide (N2O) is based on a quantum cascade laser infrared absorption spectrometer. The TRIHOP instrument was operated during POLSTRACC with a precision of 1.84 ppb and total uncertainty of 2.7 ppb for the measurement of N2O (Krause et al., 2018;. 175

Ozone
Ozone was measured by the Fast and Accurate In Situ Ozone Instrument (FAIRO). It combines UV-photometer and fast and precise chemiluminescence detector. The total uncertainty is 1.5 % and the typical precision is 0.5 % at 10 Hz .

Water vapour 180
Water vapour was measured by the Fast In Situ Stratospheric Hygrometer (FISH) based on a Lyman-alpha photometer fluorescence technique and achieved a precision and accuracy of 1 % and 5.6 % during POLSTRACC, respectively (Zöger et al., 1999;. The FISH instrument was connected to a forward-facing inlet and thus can detect ice particles (evaporated at the inlet walls) in addition to the gas-phase water vapor (Afchine et al., 2018) 2.1.5 Nitric acid 185 A separate measurement of gas phase nitric acid (HNO3) was performed by the Airborne chemical Ionization Mass Spectrometer (AIMS) using SF5as a reagent ion via another backward facing inlet. During POLSTRACC, the accuracy was 16 % with a precision of 10-15 % at a time resolution of 1.7 s (Jurkat et al., 2017;.

GLORIA
Vertical distributions of nitric acid were retrieved from limb-imaging observations by the GLORIA instrument (Gimballed 190 Limb Observer for Radiance Imaging of the Atmosphere; Friedl-Vallon et al., 2014;Riese et al., 2014). Characteristics of the GLORIA 2D chemistry mode and dynamics mode data presented here are discussed by Johansson et al. (2018) and Krasauskas et al. (2020). Note that GLORIA samples air volumes to the right-hand side of the aircraft and not along the vertical projection of the flight path. Furthermore, due to the limb viewing geometry, which enables detection of minor atmospheric constituents with low abundances, the individual GLORIA profiles constitute horizontally smoothed 195 representations of the atmospheric scenery perpendicular to the flight path. Thus, the GLORIA data at flight altitude does not exactly match the geolocations and sampling characteristics of the simultaneous in situ observations on board HALO. Therefore, in the presence of small-scale structures and horizontal trace gas gradients, differences between these different types of observations at flight altitudes are possible and are not necessarily indicative of instrument errors.

CLaMS 200
For the interpretation of the observations and the underlying processes, we employ simulations of the Chemical Lagrangian Model of the Stratosphere (CLaMS). The chemical transport model CLaMS is based on the Lagrangian transport concept and is described elsewhere (Grooß et al., 2014 and references therein). The Lagrangian concept is used in two ways. First the chemical composition of the air is simulated for so-called air parcels that follow the wind and are distributed irregularly in space. Second, also the particles are simulated by the Lagrangian principle. The simulations follow multiple representative 205 NAT particle parcels, which are transported by the wind and in addition are exposed to gravitational settling. As temperatures increase above TNAT, the sedimented particle parcels evaporate and cause nitrification of the surrounding air parcels. Similarly, the vertical redistribution of water vapour by ice particles is included . This simulation setup has shown to reproduce the observed denitrification and nitrification .
The CLaMS simulation is initialized on 1 November 2015 on the base of MLS satellite data, multi-annual CLaMS 210 simulations (Pommrich et al., 2014) and tracer-tracer correlations using a similar procedure as described by Grooß et al. (2014) and runs until March 2016. The simulation encompasses the Northern Hemisphere from the surface to 900 K potential temperature with a vertical resolution of 100 m. Results of this simulation have also been shown elsewhere Braun et al., 2019;Johansson et al., 2019).

Tracer-tracer correlations
Tracer-tracer correlations of long-lived species are an established method to study transformation processes of trace gases in the lower stratosphere. The relation between chemical species in the stratosphere is linear and compact as long as their chemical life time is long compared to transport time scales (Plumb and Ko, 1992). The relation between total reactive nitrogen and nitrous oxide can be used in this sense because their lifetime is long compared to transport time-scales (Keim et 220 al., 1997). The lifetime of nitrous oxide is more than 100 years (Prather et al., 2015). The mean age of an air parcel can be understood as average time since the last contact with the troposphere (Ploeger et al, 2015). During POLSTRACC the mean age of the probed air masses ranged between about 1 and 5 years (Krause et al., 2018).
In the present study the redistribution of total reactive nitrogen in the lowermost stratosphere during POLSTRACC was analysed using tracer-tracer correlations, namely the correlation between total reactive nitrogen and nitrous oxide. Photolysis 225 and the reaction of nitrous oxide with O(1D) are the main sources of total reactive nitrogen in the stratosphere (Keim et al., 1997;Greenblatt and Ravishankara, 1990). These processes lead to the formation of nitric oxide. Subsequent reactions produce NO2 and finally nitric acid (HNO3) and other reactive nitrogen species. These processes are most effective in regions with high UV-radiation as in the tropical stratosphere (Murphy et al., 1993). The correlation between total reactive nitrogen and nitrous oxide is conserved during the transport of air masses from the tropics to the Polar Regions as long as no 230 sinks or sources for NOy are effective (Murphy et al., 1993). The observed negative slope of the relation between NOy and N2O can be understood as a kind of conversion efficiency. It is the portion of N2O that is converted into total reactive nitrogen by photolysis and subsequent reactions and lies in the order of 6-8 % e.g. (Fahey et al., 1990b;Loewenstein et al., 1993;Weinheimer et al., 1993;Fischer et al., 1997;Strahan, 1999). This correlation is linear over a wide range of N2O concentrations. Above about 25 to 30 km altitude the loss reaction of NO with N becomes increasingly important leading to 235 a deviation from the linear relationship for N2O values below about 100 ppb (Loewenstein et al., 1993). Observations within the lowermost stratosphere of the Northern Hemisphere suggest some seasonality of this slope (Hegglin et al., 2006). A stronger seasonality was observed in the tracer-tracer correlation between O3 and N2O (Hegglin and Shepherd, 2007;Bönisch et al., 2011).
An analysis for a large number of ER-2 flights showed that the NOy-N2O correlation is linear down to about 170 ppb 240 (Strahan, 1999). Based on these observations a semi-empirical quantity called NOy* can be derived, estimating the concentration of expected NOy arising from observed N2O in the stratosphere. Strahan (1999) formulated this relation-ship as follows.
NOy*=(N2O(ts) -N2O(obs))*f + NOy(ts) (3) N2O(obs) is the observed N2O concentration, N2O(ts) is the tropospheric concentration of N2O entering the stratosphere and 245 f is the conversion efficiency. An additional term was added accounting for the contribution of tropospheric NOy to the observed concentrations in the stratosphere.
As long as there are no additional processes, sources or sinks, in the lower stratosphere affecting the NOy concentration, observed NOy should be very close to NOy* (within the uncertainty range of observations). dNOy is defined as difference between calculated NOy* and observed gas-phase NOy. 250 dNOy = NOy -NOy * (4) In the lower winter polar stratosphere deviations of dNOy from zero can be indicative for processes like nitrification and denitrification, resulting from the formation of polar stratospheric cloud particles, their sedimentation and subsequent evaporation. This relationship was derived from observations in stratospheric air masses; it is not expected to hold in tropospheric air masses.  Similar NOy concentrations, with values between about 1 and 2 ppb, have also been found during earlier aircraft missions in the winter Arctic lower stratosphere (Hubler et al., 1990;Weinheimer et al., 1993;Arnold et al., 1998). During the POLSTAR I mission in January 1997 NOy values up to 4 ppb have been observed (Ziereis et al., 2000a). In the winter polar 270 stratosphere NOy is mainly comprised (>= 90 %) of nitric acid as was found by aircraft and balloon-borne observations (Wetzel et al., 2002;Schneider et al., 1999). For comparison, nitric acid observed with the AIMS mass spectrometer Jurkat et al., 2017; is also included in Figure 1. During most parts of the flight, nitric acid and total reactive nitrogen agree with each other within the uncertainty range of both instruments, which confirms that stratospheric NOy is mainly dominated by HNO3, while other NOy components may gain importance near the tropopause. 275 Figure 6a shows total reactive nitrogen plotted versus N2O for the flight on 21 December. Also included in this figure is the regression line resulting from a linear least squares fit (R 2 =0.87). The range of its uncertainty is indicated by dashed lines.
As expected for undisturbed conditions, NOy and N2O are anticorrelated. To exclude tropospheric values that would affect the correlation, only values obtained in the stratosphere have been used for this analysis. In 2016 the tropospheric N2O concentration amounted about 329 ppb (Combined Nitrous Oxide data from the NOAA Global Monitoring Laboratory). 280 Therefore, the analysis was performed only for N2O values smaller than 320 ppb.
The slope of the regression line, corresponding to the factor f given in Eq. (3), is about 0.064. This value agrees reasonably well with earlier observations performed with these instruments. In late summer 2012 the HALO mission TACTS (Transport and composition in the UT/LMS) (Müller et al., 2016) was performed at northern mid latitudes. Nitrification and denitrification could be excluded for this time of the year and region. A linear least squares fit between NOy and N2O for 285 stratospheric values (N2O < 320 ppb) obtained during the TACTS mission gave a slope of about 0.067. The derived slope is also comparable to findings during earlier observations in the winter Arctic region that were not affected by nitrification or denitrification. During the AASE missions in winter 1989 and 1991/1992 respectively, slopes between 0.064 and 0.078 have been observed (Fahey et al., 1990a;Fahey et al., 1990b;Weinheimer et al., 1993).
The equation describing the regression can be rewritten to take the form of Eq. (3). In this formulation, the following 290 calculations of NOy* were performed. The slope obtained during the midlatitude mission TACTS was chosen as conversion efficiency f. The deviation from the value determined during the flight PGS-flight 5 can serve as a measure for the uncertainty in determining this slope, it is about 4 %. A further uncertainty in the calculation of NOy* arises from the contribution of tropospheric NOy. For the POLSTRACC observations NOy(ts) was estimated to be about 0.78 ppb (derived from the regression curve for tropospheric N2O values). This value lies well within the range spanned by previous 295 observations. From the observations during the TACTS mission a tropospheric value of 0.65 ppb was derived. Strahan (1999) derived tropospheric NOy(ts) of 0.44+/-0.22 ppb by averaging measurements at the tropical tropopause. During STREAM-97 the observed mean NOy mixing ratios have been about 0.7 ppb in the upper troposphere (Fischer et al., 2000).
Observations with IAGOS-CARIBIC show a high variability of reactive nitrogen in the upper troposphere (Stratmann et al., 2016). Values depend on region and season where and when the observations were performed and range between about 0.4 300 and 1.4 ppb. The uncertainty in the estimation of NOy* resulting from the uncertainty of the tropospheric NOy contribution is highest directly at the tropopause where the relative contribution of tropospheric NOy to NOy* is largest. With decreasing N2O concentration and increasing stratospheric character of the air mass NOy arising from the photooxidation of N2O increases. At N2O values of 300 ppb and less this uncertainty amounts to about 10 % and less.
According to Eq. (3), the tropospheric concentration of N2O is included in the calculation of NOy*. The tropospheric N2O 305 value is steadily increasing over the years. Therefore, different air mass ages imply different tropospheric N2O concentrations. During POLSTRACC the age of the probed air masses ranged between about 1 and 5 years (Krause et al., 2018). Within five years the tropospheric N2O concentration increased by about 1.5 % (Combined Nitrous Oxide data from the NOAA Global Monitoring Laboratory). This corresponds to a difference in tropospheric N2O of about 5 ppb and a difference in NOy* (assuming a conversion efficiency of 0.067) of about 0.3 ppb. Although this contribution to NOy* is 310 comparatively small, the tropospheric N2O concentration in Eq. (3) was chosen according to its air mass age.
In Figure 1b measured NOy values are shown along with calculated NOy* values. Also shown is the uncertainty range of NOy*. During most of the time both curves agree well within the uncertainty range. A larger deviation at around 54000 s UTC was found at high N2O values close to the tropopause where this relation is not expected to be valid.
Another diagnostic tool for characterizing the lowermost stratosphere with respect to reactive nitrogen species is the NOy/O3 315 ratio (Murphy et al., 1993). NOy and O3 are mainly produced in the tropical stratosphere and have comparable lifetimes in the lower stratosphere. Their ratio is more constant than the concentration of the individual species itself. This can also be seen in Figure 1. NOy and O3 exhibit a substantial variability along the flight track while the NOy/O3 ratio does not change a lot. Typical values for this ratio in the lower stratosphere at high northern latitudes are around 0.003-0.004 for undisturbed conditions e.g. (Murphy et al., 1993;Fahey et al., 1996). Values of the same magnitude were observed during the 320 POLSTRACC flight in December (Figure 1). Here a median NOy/O3 of about 0.0038 was measured. For comparison, in winter 1997 aircraft observations within the STREAM campaign in Arctic found values between 0.003 and 0.006 for this ratio in undisturbed stratospheric air (Fischer et al., 2000).
In December 2015, the polar vortex had already reached temperatures that allowed the formation of PSC particles . With CALIOP large areas covered with PSC were observed between approximately 15 and 25 km altitude (Pitts et 325 al., 2018). Based on the present study of the NOy-N2O correlation and NOy/O3 ratio, no indications have been found that the lowermost Arctic stratosphere was already affected by redistribution of reactive nitrogen species at the beginning of the winter. However, considering that only one flight has been performed to the Arctic covering only a small part of the sub vortex region, redistribution in the lowermost stratosphere already in December cannot be ruled out by these observations.

Mid-Winter Phase 330
The second phase of the POLSTRACC mission started with the transfer flight of HALO from Oberpfaffenhofen to Kiruna in northern Sweden on 12 January 2016. Seven flights from Kiruna were completed by 2 February 2016. More than 90 % of these flight routes were lying north of 60° N, with more than 85 % of the total flight time in the lower stratosphere with potential vorticity values of more than 2 PVU. The height of the dynamical tropopause is commonly attributed to the level where the potential vorticity equals this value. 335 During the midwinter phase the observed relation between NOy and N2O differs significantly from the situation in the early winter phase or from the observations at midlatitudes as during TACTS in 2012. This deviation was already observed during the first local flight out of Kiruna on 18 January. The flight went along the east side of Sweden and then along the coast of Norway back to Kiruna. The main scope of this mission flight was the probing of filamented stratospheric air . At altitudes above 12 km, significantly higher NOy concentrations were measured, with values up to about 10 ppb, 340 than during the December flight, with maximum values up to 3.4 ppb (Fig. 2). In parallel, the N2O concentration was as low as 280 ppb. At the beginning and end of the flight NOy and NOy* were nearly identical. NOy and NOy* also agreed between about 41600 and 44000 UTC seconds when HALO flew close to the tropopause. Outside these periods observed NOy was significantly higher than calculated NOy*. In Figure 6b NOy is shown versus N2O along with NOy*. For stratospheric N2O concentrations, observed NOy levels were much higher than NOy*, by up to about 6 ppb. This excess NOy cannot be 345 explained by any unknown additional tropospheric source as the deviation from NOy* increases with decreasing N2O. The excess NOy amounts up to about 50 % of the whole gas-phase total reactive nitrogen and is also reflected in the highly variable NOy/O3 ratio during this flight. High ratios have been found along with high values for dNOy. Values of the NOy/O3 ratio changed from around 0.004 to values up to about 0.01. Observations with the GLORIA instrument on HALO complement the in situ observations down to upper troposphere and also show nitrification of the lowermost stratosphere 350 . Figure 2d shows a section through the atmosphere below HALO's flight altitude for the flight on 18 January. For this flight typical vertical resolution of ~500-1000 m was diagnosed for the chemistry mode data. A typical total uncertainty (1 σ) in the order of 10-20 % are estimated for this particular flight. Both, vertical resolution and total errors thereby depend on altitude and observed scenery.
In Figure 2b nitric acid obtained in the high spectra resolution "chemistry mode" of GLORIA for this flight is shown along 355 with in situ observed total reactive nitrogen. Although airmasses probed with GLORIA and AENEAS are not identical, a high agreement between both measurements was found. Concurrently, ozone and total reactive nitrogen increased (not shown). During this period observed NOy values have been up to more than twice as high than estimated NOy*. dNOy ranged between about two and more than 6 ppb. Again, higher NOy concentrations are also reflected in a higher NOy/O3 ratio. These findings suggest a substantial redistribution of total reactive nitrogen the lower Arctic stratosphere. Particles containing total reactive nitrogen can sediment down to HALO flight 370 altitudes within a few days. The fall speed for particles with diameter of 10 µm and above is more than 1 km per day (Fahey et al., 2001). Here, evaporation of the particles leads to increased NOy volume mixing ratios. The interpretation of this observation as particle-based redistribution is supported by LIDAR observations from space and from HALO. CALIOP detected PSCs in the Arctic stratosphere between December and late January at altitudes between 15 and 26 km . 375 Lidar observations onboard of HALO also show extensive regions of polar stratospheric clouds. On 22 January, a large ice PSC cloud was detected with a horizontal expanse of about 1400 km and thickness of up to 6 km between 18 and 24 km altitude . Observations of nitrification of the lower stratosphere up to more than 10 ppb by evaporating 380 particles have also been made during previous airborne missions to the Arctic and Antarctic (Hubler et al., 1990;Fischer et al., 1997;Arnold et al., 1998;Dibb et al., 2006;.  Kangerlussuaq to Oberpfaffenhofen in Germany (not shown), the potential temperature did not exceed 380 K and N2O values remained above 260 ppb. During this flight no deviation of NOy from NOy* was found. As mentioned earlier, besides the flight from Kangerlussuaq to Oberpfaffenhofen most of the flights stayed north of about 60° N. One exemption was the flight on 16 March to the Canary Islands. Within the accuracy of the measurement no significant deviation of observed NOy from NOy* was found (not shown). In situ observations from HALO revealing denitrified air masses fit well in the overall 425 picture of this winter. They complete the picture arising from satellite observations and model simulations for this winter.
Denitrified air masses have been observed with the Aura microwave limb sounder (MLS) (Manney and Lawrence, 2016).
Simulations with the EMAC model showed both denitrified zones in the middle stratosphere and regions with enhanced total reactive nitrogen below for December to February. By mid of March the denitrified zone stretches down to pressures higher than 100 hPa (Khosrawi et al., 2017). 430 Figure 7a summarizes all concurrent NOy-N2O observations during POLSTRACC. The data are grouped in sub-datasets according to the three mission phases. The NOy-N2O data field seems to be split into two major branches divided by the line representing NOy*. During the early winter phase (points in black) the NOy-N2O data pairs are essentially grouped around the NOy*-curve. The mid-winter period (points in red) is mostly characterized by values lying above NOy*, for N2O 440 concentrations between about 250 and 300 ppb. The late winter (points in blue) shows values above NOy* as well as values below. Values below NOy* are associated with N2O values below about 250 ppb. As pointed out earlier, this cannot be attributed to the deviation of NOy from the linear correlation at low N2O concentrations which is only expected for N2O concentrations below about 100 ppb (Loewenstein et al., 1993). In Figure 7b, dNOy versus N2O is presented for the three mission phases. The values range between about +6 ppb and -6 ppb. In both figures, high values of NOy and dNOy for N2O 445 values close to 320 ppb indicate tropospheric airmasses where the relation between NOy and N2O is not valid. The influence of downward transport on nitrification or denitrification is also reflected in Figure 8. Here dNOy is shown versus potential temperature. Positive values of dNOy are predominantly found between 340 and 370 K during the flights in January. The largest denitrification was observed in air masses with potential temperatures between 390 and 410 K in late 455 winter. The highest values of nitrification or denitrification, respectively were each found at the highest flight altitude of approximately 14 km. During previous observations evidence for denitrification has mainly been found at higher altitudes in the stratosphere. E.g., in February 1995, measurements with the MIPAS-B balloon-instrument found a 50 % reduction in NOy at altitudes between 16 and 22 km (Waibel et al., 1999). Denitrification was also detected by satellite observations in winter 2009/2010 in the Arctic between about 475 and 525 K (Khosrawi et al., 2011). With the CLaMS model denitrification 460 at about 500 K and nitrification at about 400 K was simulated for this winter . The POLSTRACC mission covered the whole winter season from December to mid of March providing the unique opportunity to probe the lowermost stratosphere under different conditions. The distribution of reactive nitrogen changed 465 from undisturbed condition in early winter to a condition with elevated concentrations (nitrification) and finally to a condition with lowered concentrations (denitrification) in late winter. This transformation of the reactive nitrogen distribution in the lower Arctic stratosphere is shown in Figure 6. In this figure the NOy-N2O relation for five selected flights is depicted. The sequence of these figures illustrates the temporal evolution of NOy at the bottom of the vortex. In early winter (6a) NOy and N2O are well correlated reflecting that the distribution is controlled by the gas-phase production of NOy 470 from N2O. In mid-winter (6b and 6c) the observed NOy exceeds the calculated NOy* by several ppb indicating the influence of the evaporation of sedimenting particles containing nitric acid. In late winter the distribution of NOy is controlled by the downward transport of air masses that have undergone removal of nitric acid by heterogeneous processes (Figure 6e). Figure   6d shows a flight typical for the transition between the mid-and late winter distribution of NOy. The observations at HALO flight altitude reflect the processes in the stratosphere above. Denitrification in the middle stratosphere by heterogeneous 475 processes and removal by sedimenting particles happened at nearly the same time as nitrification by evaporating was observed at the lowermost stratosphere. However, descending denitrified air masses at flight altitudes were first observed with a time lag of several weeks in late winter. In a certain way, the vertical distribution of NOy in the winter stratosphere was mapped to a temporal variation at HALO flight altitude. equivalent latitude and theta (potential temperature). Equivalent latitude takes advantage of the adiabatically quasi-conserved nature of potential vorticity. It therefore removes the variability in trace gas distributions that originates from reversible deviations from the climatological mean due to Rossby and smaller scale waves (see e.g. Hegglin et al., 2006). The earlywinter period shows a relatively undisturbed distribution of reactive nitrogen, the dNOy values are close to zero.
The mid-winter period is mostly characterized by positive dNOy values, particularly above 340 K and polewards of 50° N 490 equivalent latitude. The late winter period shows a nitrified region at the same location, but with weaker nitrification than in phase II. A denitrified region is located above, predominantly at potential temperature over 380 K and equivalent latitudes over 50° N. However, weak denitrification with losses up to 1 ppb is also observed throughout the whole latitude range above 360 K, even outside the vortex. Similarly, at lower isentropes slightly positive values of dNOy at lower equivalent latitudes are consistent with export of former vortex air to lower latitudes (Hoor et al., 2004, Krause et al., 2018. These 495 findings indicate transport and mixing of vortex processed air masses to the mid-latitude lowermost stratosphere in late winter and early spring.
As was pointed out earlier the NOy/O3 ratio is a further diagnostic tool for studying processes in the lower stratosphere.
Evaporating particles lead to a nitrification of the lowermost stratosphere and this process finds its echo in the increase of the NOy/O3 ratio as can be seen in Figure 2. The influence of the denitrification in the late winter phase does not form such an 500 obvious signature in the NOy/O3 ratio. For example, Figure 5c shows the NOy/O3 ratio versus time for the flight on 13 March. Before about 37000 UTC s NOy and NOy* agree reasonably well. There are no indications for denitrification during this part of the flight. In the further course of the flight the difference between NOy and NOy* increased along with decreasing N2O concentrations, a clear signature of denitrification. The NOy/O3 ratio, however, does not decrease during this period as one might expect but it increased from about 3.9*10 -3 to about 4.8*10 -3 . The obvious explanation for these 505 observations is that not only NOy was removed from the air masses but also ozone. In sense of the NOy/O3 ratio the decrease in NOy by denitrification is counterbalanced by the decrease in ozone.
As a plausibility check, with a rough "back of the envelope" calculation the missing ozone can be estimated for this flight.
Between about 37000 and 42200 UTC s, the denitrification amounted up to about 5 ppb, the NOy/O3 ratio increased to about 0.0048. Based on these observations and assuming an "undisturbed" NOy/O3 ratio of 0.004, on average in the order of about 510 1 ppm ozone were missing during this flight period. This is at least roughly in accordance with a much more detailed simulation using the CLaMS model. For this POLSTRACC flight a chemical ozone depletion of more than 1 ppm or about 50 % was estimated. .
The temporal evolution of the subvortex region is also visible in an ozone-nitrous oxide coordinate system. Figure 10 shows the correlation between these two trace gases for the same flights as in Figure 6. In addition, the regression line resulting 515 from a linear least squares fit analysis of the December flight is plotted. Similar criteria apply to this tracer relationship as for the NOy-N2O correlation (Hegglin et al., 2006;Hegglin and Shepherd, 2007;Bönisch, 2011). A slope of -18.5 was derived from the December 2015 flight. This compares quite well with the mid-latitude slope derived from the September data during the TACTS flights (see above) of -19.2. From late January through the end of the mission, the observed ozone concentrations deviate more and more from the regression line to lower values. The difference increases with decreasing 520 nitrous oxide concentrations, parallel to the evolution of the NOy-N2O correlation in Figure 6. Thus, the denitrification observed in the descending air masses is also reflected in an ozone decrease.

Observations of Particulate Nitrate 530
The POLSTRACC payload mainly comprised in situ gas-phase and remote sensing instruments. It did not include instruments for specific measurements of particle parameters. However, the NOy detector offers an indirect method to observe particles containing reactive nitrogen compounds. For these measurements the oversampling characteristic of the forward-facing inlet was used. With the backward facing inlet, only gas-phase total reactive nitrogen is measured (see Section 2.1.1). Particles evaporating within the inlet-system release nitrate molecules that are detected by the NOy 535 instrument. During one episode during the flight on 18 January, the resulting chemiluminescence signal was even so high that it could not be processed by the detection electronics. During the midwinter phase of POLSTRACC particulate nitrate has been observed during four flights using this measurement approach. The flight on 20 January went from Kiruna north towards Spitzbergen, then westwards towards Iceland and back to Kiruna. One mission objective was to probe forecasted nitrification of the lowermost stratosphere. Gasphase NOy ranged between about 1 and 8 ppb at altitudes between 12.5 and 14.3 km (see Figure 11). During four periods the 545 signal obtained with the forward-facing inlet was up to a factor of 50 higher than the gas-phase signal obtained with backward facing inlet. In Figure 11b total NOy measured with the forward-facing inlet is shown along with gas-phase NOy.
NOy-tot is not corrected for enhancement and shows values up to 100 ppb. In total, the areas in which particle NOy was observed during this flight extended over about 2000 km. Outside these episodes NOy obtained with the forward and aft facing inlet, respectively, agreed within about 7 % percent. 550 Particulate nitrate was observed in situ on HALO only during the mid-winter phase. In addition to the 20 January flight, particulate nitrate was also observed during the flights on 18, 25 and 31 January. The large-scale particulate events were observed over distances of approximately 800 to 1400 km. Widespread polar stratospheric clouds of different compositions have also been observed with the WALES lidar instrument (Water Vapour Lidar Experiment in Space -airborne demonstrator) on board the HALO aircraft above flight altitude . Between December and end of January 555 PSC particles have been detected with space-borne LIDAR between 15 and 26 km Voigt et al., 2018).
The measurement approach, using the oversampling characteristic of inlets has also been used during earlier aircraft missions in the Arctic. High NOy-net values indicating particulate nitrate have been observed during the SOLVE mission in winter 1999/2000 from the ER-2. Large fields of PSC particles were found between about 15 and 21 km (Northway et al., 2002). Also balloon-borne measurements confirmed the presence of NAT (Voigt et al., 2000) in that winter while liquid 560 ternary solution particles (Schreiner et al., 1999) were observed in the THESEO winter before. During the Vintersol/Euplex mission in February 2003 single NAT particles have been observed at altitudes between about 18 and 20 km onboard the Geophysica . Compared to previous observations, during POLSTRACC particulate nitrate was found at lower altitudes between about 10 and 14.5 km. This corresponds roughly to potential temperatures between about 310 and 370 K (Figure 12). 565 Figure 12. NOy-net versus altitude (a) and NOy-net versus potential temperature (b) for all flights with the occurrence of particulate nitrate (18, 20, 25, 31 January 2016).

570
The vapour pressure of nitric acid over NAT can be derived from measured NOy, water vapour, ambient pressure and temperature using the expression given by Hanson and Mauersberger (1988). For estimating the saturation ratio one can assume that observed NOy mainly consists of nitric acid in the stratosphere as confirmed by simultaneous measurements of nitric acid (Figure 1). In Figure 13 the saturation ratio of nitric acid is plotted along with the gas-phase equivalent of 575 particulate nitrate. In general, a good agreement between the occurrence of particulate nitrate and high NAT saturation ratios was found. During the flight on 18 January, the saturation ratio was up to about 0.25 indicating ongoing evaporation of the particles. During the particle observations of the other flights the air masses have been supersaturated with respect to NAT.
To convert total NOy to an equivalent gas-phase NOy concentration, the particle size dependent enhancement factor has to be known. As mentioned earlier, the HALO payload did not include instruments for the independent measurement of the 580 particle size. The enhancement factor therefore can only be estimated with some uncertainty from the signal obtained with the forward-facing inlet. As pointed out earlier, nitrate containing particles evaporate completely in the inlet and converter and the released nitrate molecules are detected. From this signal the diameter of the particles can be derived. This method was described in detail by Northway et al. (2002). Due to wall effects particles do not evaporate within the sampling frequency of the instrument (1 second). The complete evaporation takes up to 20 s. Therefore, for estimating the diameter of 585 the particle, the signal has to be integrated over the complete time of evaporation. Assuming NAT density and for the specific parameters of the AENEAS NOy-detector the particle diameter can be estimated. The pre-factor of the formula combines density and specific instrument parameters. The measured concentration enters the formula with the units mol/mol: D[m]=4.28*10 -3 (NOy) 1/3 (5) 590 Figure 13. NOy-net, ambient temperature, saturation ratio and TNAT for all flights where particulate nitrate was observed. During the second half of the flight PGS8 the NAT saturation temperature TNAT could not be calculated because of missing data.

595
A similar approach was also chosen for analysing particles from observations from the Geophysica . As was shown by Northway et al. (2002) this method can only be applied as long as the evaporation periods of the individual particles do not overlap. During POLSTRACC the frequency of evaporating particles was that high that for most time this approach could not be used to estimate the particle diameter. At the edge of some particle episodes, however, individual evaporation events could be identified. Using the above given formula, particle diameters between about 9 and 18 µm have 600 been derived. The particle diameters derived from the POLSTRACC observations are roughly in the same order of magnitude as those measured during previous aircraft missions to the Arctic. Particle diameters, between 5 and 20 µm, have been derived from measurements with the ER-2 during the SOLVE mission at altitudes between 15 and 21 km (Northway et al., 2002;Fahey et al., 2001). During RECONCILE in January 2010, particle diameters between 10 and 24 µm have been concluded from observations on board of the research aircraft Geophysica . Particles with diameter 605 smaller than 6 µm have been observed from the Geophysica in February 2003 .
As discussed in Section 2.1.1, the enhancement factor reaches a maximum for particles larger than about 10 µm and is then independent of diameter (Belyaev and Levin, 1974). It also increases strongly with increasing ambient pressure. The maximum enhancement factor ranged between about 50 and 85 for the instrumental setting during this mission at the altitudes where particles were detected. For comparison, an enhancement factor between about 13 and 22 was applied for 610 measurements on board of the ER-2 at lower ambient pressures during AAOE in the Antarctic (Fahey et al., 1989). For observations during POLSTAR at altitudes up to 13 km a maximum enhancement factor of 140 was derived for particles larger than 10-20 µm . 615 Figure 14. NOy-partitioning for PGS-flight 8. Gas-phase NOy, dNOy = NOy -NOy * (calculated) and enhancement corrected particulate nitrate are shown. (a) absolute values, (b) dNOy/NOy ratio, NOy_P/NOy ratio -1 s values and as running average (30s).
Following the above sketched assumptions, the gas-phase equivalent of the observed particulate nitrate can be estimated as is presented for the observations during the flight on 20 January (Figure 14). This allows to establish a kind of NOypartitioning. The observed total reactive nitrogen in the lower stratosphere comprises three contributions: a) gas-phase NOy 620 arising from the photo-oxidation of N2O -"undisturbed NOy*", b) NOy from already evaporated nitrate particles and c) particulate nitrate. For the flight on 20 January observed NOy exceeds calculated NOy* by up to 6 ppb. During the last part of this flight up to about 60 % of observed NOy can be attributed to evaporated particles. Averaging over the whole flight, the median ratio between excess dNOy and NOy was about 20 %. The highest values for particulate nitrate amounted about 1.5 ppb. The highest peak ratio between particulate nitrate and gas-phase nitrate was up to more than 0.7. Averaged over a 625 longer period (50800 to 53500 s UTC) particulate nitrate amounted around 0.2 ppb. This corresponds to a ratio between particulate and gas-phase nitrate of about 0.14. High dNOy values were not necessarily observed at the same time as high particulate nitrate, indicating that the particles have been already completely evaporated. With the exception of the peak values, more reactive nitrogen was found in the gas-phase than in the particulate phase.
Besides the above sketched dependences, the estimation of the enhancement factor might be hampered by the fact that the 630 aircraft itself has an influence on the sampling characteristic of the inlets on the top of the fuselage. This was shown in a publication studying the influence of the inlet position on the quantitative determination of the ice water content. (Afchine et al., 2018). Depending on size of the ice particles, overestimation as well as underestimation of the ice water content was found for the specific inlet position of the FISH instrument. Due to the different positions of the water and reactive nitrogen inlet and the different particle size distributions of the NAT PSC and cirrus ice crystals these finding cannot easily be 635 transferred to the present measurements. However, an influence of the aircraft on the particle sampling cannot be excluded although this effect might be smaller for the smaller NAT particles compared to ice. This might affect the enhancement factor used for the estimation of the gas-phase equivalent of particulate nitrate and therefore might affect the quantitative determination of the ratio between particulate and gas-phase nitrate. However, this uncertainty does neither affect the fact that nitrate particles have been observed nor does it affect the estimate of the particle diameter. It also has no influence on 640 the gas-phase measurements of nitrified and denitrified air masses.

Model simulations
The distribution of reactive nitrogen in the lowermost winter Arctic stratosphere was also simulated with the Chemical Lagrangian Model of the Stratosphere (CLaMS). CLaMS was developed at the Forschungszentrum Jülich as modular chemistry transport model. With CLaMS, denitrification in the Arctic stratosphere in winter 2002/2003 and 2009/2010 was 645 simulated and compared to in situ measurements from the Geophysica (Grooß et al., 2005;Grooß et al., 2014).
The simulation of the impact of vertical settling of NAT particles is challenging as it depends critically on temperature and also the heterogeneous nucleation (Tritscher et al., 2021). For example, nitrification occurs when the sedimentation of NAT particles fall into altitudes with temperatures above TNAT. This results in a horizontally and vertically filamentary small-scale structure in the NOy distribution, seen both in the observations during the campaign but also in the simulation. As time 650 proceeds, these NOy-rich structures are diluted and mixed with ambient air. Due to the limited knowledge of NAT formation nuclei, small-scale temperature distributions and also due to the resolution of the model, an exact and detailed simulation of the nitrification filaments is not possible. In the simulation, all of the HNO3 released from NAT particle evaporation is collected in the air parcels that have about 100 km distance. In some cases, the interpolation of the resulting simulated NOy to a specific location as the flight path may result in unrealistically high values. 655 First, we compare in situ NOy measurements with model values simulated along the flight paths. In Figure 15   generally around 1 ppb with a few exceptions. Around 11:20 UTC (40800 s), the simulation indicates a strong nitrification peak of about 14 ppb NOy that is not present in the observations. A detailed investigation revealed that this nitrification filament in the simulation belongs to one short episode on this flight west of the coast from Greenland (lon 303.4, lat 71.7, theta 404.1K) where the flight path is very close to one specific air model air parcel (distance 7 km). In this air parcel all evaporating HNO3 is collected over the range given by the horizontal model resolution of about 100 km. Small uncertainties 695 in the temperature of the reanalyses could cause a small displacement of such a filament. It may be that there is a similar nitrification filament close to the flight path but this cannot be clarified here.
Besides these comparisons along individual flight paths, we used the CLaMS simulation to investigate if the vertical NOy redistribution on a vortex-wide scale can be understood. This is done by examining the correlation on NOy and dNOy with the inert tracer N2O. The simulations were used to put the observations in a broader context. Especially they can contribute, 700 to what extent the observations represent the global development within the polar vortex. To do that, the simulations were  The comparison between observations and model simulations is a multi-step process. First, we were able to show, through 720 the comparisons along the flight paths, that the processes underlying the model simulations are so well understood that the observations can be reproduced. In a second step, we showed that these simulations along the flight path are representative for the vortex-wide NOy distribution. This, in turn, suggests that the observations, although limited in time and space to individual flights, provide a good description of the distribution of nitrogen oxides in the subpolar region during this winter.

Summary 725
During the course of the extremely cold winter 2015/2016 aircraft-based measurements with the German research aircraft HALO have been performed in the lowermost stratosphere of the Arctic region within the POLSTRACC mission. The observation period covered the whole winter-spring season from December to March. This extended observational period offered the unique opportunity to study the changing distribution of total reactive nitrogen in the lowermost stratosphere with time. 730 Tracer-tracer correlations, the relation between NOy and N2O and O3, respectively have been used as tool to study and interpret the observed temporal evolution of the UTLS composition. In December, the distribution of reactive nitrogen did not show any indications for deviations from undisturbed conditions controlled by gas-phase chemistry and transport. This changed during the second mission phase in January and beginning of February. During several flights enhanced NOy values have been observed. Using NOy-N2O tracer correlations, nitrification could be clearly identified. Observed NOy values have 735 been up to 6 ppb higher than expected without redistribution of nitrogen species. This is also reflected in the NOy/O3 ratio that was up to more than a factor of two higher in January than in December. This could be interpreted in terms that the subvortex region was affected by heterogeneous processes taking place in the stratosphere above. Particles falling down from the PSC regions evaporate and released gas-phase NOy leading to a nitrification of the lowermost stratosphere. This means, that during some periods more than 60 % of the observed NOy was caused by evaporating particles. 740 Along with enhanced gas-phase values particulate nitrate was observed during mid-winter at flight altitude between about 10 and 14 km. The occurrence of PSC particles at such altitudes is rare. Particulate nitrate was observed over wide regions during four flights out of Kiruna. The diameter of these PSC particles ranged between about 9 and 18 µm. The occurrence of particulate nitrate at flight altitude and the nitrification of the lowermost stratosphere fit into the picture of other observations made in this winter. Extended PSC coverage has been already observed during December by CALIOP . 745 Particulate nitrate formed in the middle stratosphere was also observed by LIDAR on board of HALO and simulated by models Khosrawi et al., 2017). However, the uncertainty of the sampling characteristic of the inlet makes a quantitative determination of the ratio between particulate and gas-phase nitrate difficult.
Nitrified regions at the lowermost stratosphere have also been found in the late winter phase. Along with nitrified regions subsidence of air masses from the polar vortex controlled more and more the distribution of reactive nitrogen at flight 750 altitudes. Using tracer-tracer correlations substantial denitrification could be derived in subsiding air masses with minimum values of down to about -5 ppb. This means that up to about 50 % of the undisturbed NOy was missing.
Nitrification of the lowermost stratosphere in mid-winter and denitrification in late winter are linked together by heterogeneous processes in the above lying stratosphere. While nitrification caused by sedimenting particles was already observed in mid-winter at flight altitude, the result of the denitrification at higher altitudes was not observed at the bottom of 755 the vortex before end of February. Concurrently with dentrification, lower ozone concentrations were observed in the sinking air masses indicating ozone destruction at higher altitudes.
In situ observed total reactive nitrogen has been compared with the results of simulations with the Chemical Lagrangian Thus, the present measurements provide a comprehensive picture of the temporal evolution of the reactive nitrogen distribution during a whole winter period. They allowed to observe the transition of the lowermost sub-vortex region from the undisturbed to the nitrified and finally denitrified state. 765 Data availability: The observational data obtained during the HALO flights are available at the HALO database: https://halodb.pa.op.dlr.de.
Author contributions: HZ, GS, PS and ML were responsible for the NOy measurements. HZ for the further data analysis 770 and the writing of the manuscript. PH and JK performed the N2O measurements. VB performed part of the N2O-analysis.
JUG performed the CLaMS model simulations. AZ was responsible for the Ozone measurements. AA and CR carried out the water observations. Airmass age was provided by AE, HNO3-measurements with the AIMS mass-spectrometer were performed by AM and CV. HNO3 observations by GLORIA were provided by WW, MB, JU, and the GLORIA Team. HO and BMS were responsible for the scientific flight planning and GLORIA observations. All co-authors were involved in 775 review and editing of the paper.
Competing interests: The authors declare that they have no conflict of interest.
Acknowledgements. We would like to thank all colleagues of the DLR Flight Department for their excellent support, which 780 made this successful flight campaign possible. We thank the entire POLSTRACC-GW-LCYCLE-SALSA team for the very productive and pleasant collaboration. We thank all colleagues who participated in the meteorological flight planning. The