High – Arctic aircraft measurements characterising black carbon vertical variability in spring and summer

The vertical distribution of black carbon (BC) particles in the Arctic atmosphere is one of the key parameters controlling its radiative forcing. Hence, this work investigates the presence and properties of BC over the high Canadian Arctic. Airborne campaigns were performed as part of the NETCARE project and provided insights into the variability of the vertical distributions of BC particles in summer 2014 and spring 2015. The observation periods covered evolutions of cyclonic disturbances 5 to the polar dome that caused and changed transport of air pollution into the High–Arctic, as otherwise the airmass boundary largely impedes entrainment of pollution from lower latitudes. A total of 48 vertical profiles of refractory BC (rBC) mass concentration and particle size, extending from 0.1 to 5.5 km altitude, were obtained with a Single–Particle Soot Photometer (SP2). Generally, the rBC mass concentration decreased from spring to summer by a factor 10. Such depletion was associated with 10 a decrease of the mean rBC particle diameter, from approximately 200 nm to 130 nm at low altitude. Due to the very low number fraction, rBC particles did not substantially contribute to the total aerosol population in summer. Profiles analysed with potential temperature as vertical coordinate revealed characteristic variability patterns due to different balances of supply and removal of rBC in specific levels of the stable atmosphere. Kinematic back–trajectories were used to investigate transport pathways into these levels. The lower polar dome was influenced by low–level transport from sources 15 within the cold central and marginal Arctic. During the spring campaign, a cold air outbreak over eastern Europe additionally caused northward transport of air from a corridor over western Russia to Central Asia that was affected by emissions from gas flaring, industrial activity and wildfires. This caused rBC concentrations between about 500 to 1800 m altitude to gradually increase from 32 to 49 ng m−3. The temporal development of transport to the level above, at around 2500 m, caused the initially low concentration to increase from <15 ng m−3 to 150 ng m−3. Despite the higher concentrations in the upper level, 20 significantly less rBC reached the High–Arctic relative to co–emitted CO. A shift in rBC mass–mean diameter, from above 200 nm in the low–level transport dominated lower polar dome to <190 nm at higher levels, indicates that rBC got affected 1 Atmos. Chem. Phys. Discuss., https://doi.org/10.5194/acp-2018-587 Manuscript under review for journal Atmos. Chem. Phys. Discussion started: 14 June 2018 c © Author(s) 2018. CC BY 4.0 License.


[…] The observation periods covered evolutions of cyclonic disturbances which favored the transport of air pollution into the High-Arctic, as otherwise the air mass boundary largely impedes entrainment of pollution from lower latitudes. […]
P3, L5 "spread of more than one order of magnitude" in what? rBC concentrations, deposition, or emissions? Please clarify what you are referring to here.
The statement was clarified, now it reads: […] the balance of these effects in the Arctic can only be estimated in models as accurately as vertical distributions of BC is known. However, profiles of BC concentration show a spread of more than one order of magnitude amongst different state-of-the-art models as well as between models and observations (AMAP, 2015). […] P5, Table 1: Please add a note to the table explaining that the "station" is the location the plane left from. Added.

P10, Figures 2 and 3: The black triangles are very hard to see against the blue and green colors. Consider making the triangles white instead?
Changed P11, Figure 4: The two blue shades for Alert and Eureka are very hard to tell apart. Can you make them easier to distinguish?
The colors were chosen to clearly distinguish the spring and summer measurements. In order to improve the readability of the lines, the color of the Alert profiles was changed to green. P12, L1-2: I'm not sure the ARCTAS and NETCARE observations discussed in this paragraph are enough to conclude that "wet removal becomes more efficient during summer within the polar dome, but as well already during northward transport outside the dome." Do you have other evidence supporting that the changes seen in both campaigns are due to more efficient wet removal?
Our interpretation of wet removal was found weak or inconsistent by other reviewers. In order to improve the interpretation of our observations, the frequency of liquid and ice cloud encounter by the air parcels during transport was calculated for each flight and discussed in Section 3.4. Even with this additional tool, it was difficult to properly estimate the effective impact of wet removal on BC concentration and its properties. Thus, the interpretation of wet scavenging based on the BC/CO ratio was substantially reduced. The mentioned statement was removed. The previous statement about the balance of rBC supply and removal is stressed by: […] This might be due to the fact that their observations were from a Sub-Arctic region (northern Alaska), where pollution supply and removal are not necessarily in the same balance as within the polar dome. The balance between supply and removal of rBC appears to have a pronounced seasonality, based on the ARCTAS and NETCARE observations. […] P12, L7-9: Are you saying that mixed Asian outflow is a source of the Arctic haze you observed in this campaign, or is it just an example to show that the haze concentration is usually lower than near the source?
The statement was meant to compare the mass concentration of BC found in the Arctic with other locations, inside and outside the Arctic. However, the comparison with Asian outflow conditions was removed because of its low relevance in the present context. The sentence was modified as following: […] The overall range of BC concentrations is similar to previous spring observations reported for the European Arctic  and comparable to measurements from the mixed boundary layer over Europe

. […] P12, L15-16: I'm not clear what you mean by "could indicate a partitioning of rBC particle size within polluted layers." Can you please clarify what you mean by this sentence?
Anonymous referee #2 already reported this issue. With this sentence we meant to underline that the peaks of rBC mass concentration did not directly imply an enhancement of the rBC number fraction. This might involve different removal mechanisms for different aerosol types during transport. On the other hand, other processes, as different emission sources, might play a role. Due to its speculative character, the present statement was removed.
P12, L21-22: I think you need to be careful with the writing here. rBC could have a significant impact on solar light extinction (measured in W/m 2 ) even if the number fractions of rBC particles relative to total aerosol was low. However, the low ratio, combined with the low rBC mass concentration, means the impact in this case is negligible. Thus, I think you have to mention the low mass concentration of rBC here before concluding rBC has a negligible impact.
We agree with the referee's comment. The text was changed in accordance to other modifications made in Section 3.5.
[…] In contrast to the spring, the summer MMD showed a slight increase from the surface (129 nm) to about 600 hPa (140 nm; Fig. 4  , the mass absorption cross-section of BC particles depends, also, on the particle's diameter. As a consequence, the concentration of rBC mass in small particles could potentially contribute to the enhancement of the absorption coefficient of the total aerosol. Nevertheless, the low values of (average of 2 ng m -3 with IQR 0-12 ng m -3 throughout the column) and RnumTA (average of 0.75%) makes BC a minor contributor to the total aerosol light extinction. A more detailed description of seasonal and vertical variability of the BC core diameter will be provided in Section 3.

5[…]
P17, L31-32: How does this choice of only using trajectories that encountered above average M_rBC potentially impact your analysis and results? Is there a potential for bias from this choice?
Our choice of using a MrBC threshold was also questioned by other referees. Initially we wanted to focus on the most intense plumes, which most likely cause high but local forcing. We now understand the limits of our choice, which might systematically remove air parcels that originated in pristine regions or that experienced precipitation. For this reason, the trajectories presented and discussed in Section 3.4 include all points at which measurements were made.
P18, L1-2: Why did you not use the ECMWF boundary layer heights to determine the hatching instead?
The ECMWF data were used to run the LAGRANTO back-trajectories. A flag indicating whether the trajectory was traveling within the boundary layer was however not part of the available model output. Linking the trajectory position at each time step again with the boundary layer height information from datasets (e.g. ERAinterim) would be possible in post-processing, however only as a complicated approach relaying on several assumptions. A detailed evaluation of the trajectories' interactions with the boundary layer was beyond the scope of the maps in Fig. 7 and 8 and we believe that detailed information would have been blurred in these multiple day average maps. The text was changed and now stresses that the hatching highlights the presence of trajectories moving at atmospheric pressures >920 hPa in a grid cell in contrast to trajectories already lifted up from the lower atmosphere.
[…] A hatching highlights grids where trajectories travelled at atmospheric pressures >920 hPa, which is equal to less than about 0.5 km. Climatological boundary layer heights over Europe are typically <1 km during daytime [Seidel et al., 2012], thus pollution uptake from surface sources may be possible in a well-mixed atmosphere in the hatched areas in contrast to trajectories moving in the upper atmosphere or being lifted already due to vertical motion in synoptic scale systems (Sec. 3.1). […] P19, L1-2: I don't see the difference between Levels II and III discussed in this sentence in Figure 7. What should I be looking for in the figure?
The authors agree with the referee's comment. The interpretation of the Figure 7b and Figure 7c was not accurate. In fact, back trajectories suggested that the motion of airmasses from the Eurasian sector to Level II and Level III was quite similar. Insights on transport and origin of air parcels higher in MrBC are now more evident after extending the trajectory study to the entire dataset and focusing the discussion on MrBC instead of RCO. The entire Section 3.4 was substantially modified, we thus encourage the reviewer to consider the changes implemented in the entire section. • The location of the reference stations is now symbolized by a large black cross.
• RCO in Figure 7 and 8 was substituted with MrBC.
• Two color scales describing the overpass frequency and MrBC were added.
• A legend now indicates the two source types (wild fires and gas flaring) and the airmasses moving at low level. vertical variability in spring and summer We would like to thank the referees for their detailed and constructive comments, which helped us to improve our manuscript. While the referee comments are given in black bold, our answers are given below in blue letters. Additionally, we added the changes we made in the revised manuscript in blue bold letters.

Answers of the authors to anonymous Reviewer#4
Anonymous Review of Manuscript acp-2018-587 GENERAL REMARKS This paper describes the results from the aircraft measurements of black carbon (BC) aerosols over the high arctic region. The vertical distribution of BC is one of the most important characteristics for assessing its radiative impact. Authors analyzed in detail the vertical distributions, their seasonal variations, and transport pathways of BC using the data sets from the aircraft observations which were performed in the summer of 2014 and the spring of 2015. The analyses of the vertical distribution of BC with potential temperature illustrated the fundamental feature of the transport of BC from the lower latitudinal region (i.e., Sub-Arctic). Single particle soot photometer (SP2) was deployed on the aircraft to reveal one of the microphysical parameters, size distributions, of BC. The changes in the size distributions of BC in the vertical coordinate indicated that the removal process of BC during the transport to the high-arctic region is related to precipitation. The results and discussion presented in this study meet the scope of ACP. The observed features, which are well illustrated in this study, will be really helpful for the research community of Arctic climate changes as well as I actually enjoyed reading this paper. What this paper does not present in detail is the analyses of wet removal process of BC during the transport and its impact on the abundance and microphysical parameters of BC-containing particles. The cloud processing and following precipitation during the transport in East Asia can significantly affect the microphysical parameters of BC-containing particles in the lower free troposphere ( There should be a difference in the actual wet removal process between East Asia and Arctic, because the scavenging of BC particles can be affected by cloud phase (e.g., Browse et al., 2012). Furthermore, we are interested in where BC-containing particles were removed and deposited in Arctic region in order to well understand the snow darkening induced by deposited BC. The more data analyses of precipitation during the transport (intensity of precipitation, where air masses were affected by precipitation, etc.) magnify the significance of the data sets used in this study.
The authors would like to point out that the referees raised questions concerning the interpretation of the BC/CO ratio as indicator for wet scavenging and encouraged us to verify the subsequent hypothesis and conclusions. Due to the high number of comments on this specific topic, we prefer to provide here a general and common answer to all reviewers. As a consequence of the above-mentioned reasons, Section 3.4 was substantially modified. The discussion now focusses on the importance of transport patterns on the observed BC concentration. Thus, Figure 7 and Figure 8 were modified. The discussion on potential impact of wet scavenging on BC and BC/CO ratio is now substantially reduced. However, additional analysis of back trajectories, including encounter with clouds, is now presented in the supplementary material.
Corrected P23, L11-13. The finding in  is that the average mass of non-BC materials on rBC-containing particles increased with increasing rBC core diameters. They just discussed shell to core (S/C) ratio of rBCcontaining particles. When we translate the relative enhancement of shell mass of non-BC materials into the S/C ratio, the similar tendency given in Kodros et al. (2018) will also be found in Moteki et al. Please modify this description and add appropriate discussion on this part.
The same issue was highlighted by anonymous referee#2. The text was modified in order to translate our core to shell diameter ratio into mass ratio. As matter of fact, our results are coherent with the findings of . The mass of coatings was calculated assuming a fixed density of 1 g cm -3  and quantified to be 4.4 fg and 9.7 fg for BC cores having diameters of 140 and 220 nm respectively. However, Section 3.5 was significantly modified and, based on other referees' comments, the statement mentioned by the anonymous reviwer#4 was removed.
Overview of changes to ACP 2018-587 Generally, the rBC mass concentration decreased from spring to summer by a factor :: of 10. Such depletion was associated with a decrease of the mean rBC particle diameter, from approximately 200 nm to 130 nm at low altitude. Due to the very low number fraction, rBC particles did not substantially contribute to the total aerosol population in summer.

20
Climate change in the Arctic is more rapid than on global scale and a significant loss of the summertime sea-ice extent has been observed over the past decades (e.g. Lindsay et al., 2009). The fast progression of change is largely a result of the ice-albedo and temperature feedback (Screen and Simmonds, 2010;Pithan and Mauritsen, 2014). The driving agents of Arctic warming, however, still remain unclear. Recent studies suggest that next to CO 2 , short-lived climate forcers contribute significantly to the observed warming, but their complex interactions with the Arctic climate system cause high uncertainties 25 (Quinn et al., 2008;Shindell et al., 2012;Yang et al., 2014;AMAP, 2015;Sand et al., 2015). Black carbon (BC) particles, emitted during incomplete combustion of fossil fuels and biomass, are the major light absorbing component of atmospheric aerosol. Bond et al. (2013) concluded that atmospheric BC's interaction with solar radiation induces a global radiative forcing of +0.71 W m 2 , which has ::: with : an uncertainty range of +0.08 up to +1.27 W m 2 . BC may also affect the distribution, lifetime, and microphysical properties of clouds when particles act as cloud condensation nuclei (e.g. Chen et al., 2010) or, in 30 BC loaded atmospheric layers, cloud properties can change as adjustment to increased temperature and stability (e.g. Lohmann and Feichter, 2001). The aerosol cloud interaction is suspected to significantly impact climate (IPCC 2013), but the overall level of scientific understanding is still low (Bond et al., 2013). The aerosol interactions with solar radiation and clouds not only depend on concentrations, but also on microphysical properties, namely the size distribution and mixing state, of BC particles (Kodros et al., 2018).
Model studies of the Arctic climate system by Flanner (2013) and Samset et al. (2013) emphasise that Arctic surface temperatures have different sensitivities to BC's radiative forcing : , depending on the altitude where the absorbing aerosol layers :: at ::::: which :::::::: absorbing ::::::: aerosols : are distributed. When absorption and scattering through aerosols occur higher in the atmosphere, it 5 has a dimming effect on the solar radiation reaching the surface. The energy absorbed at higher levels is inefficiently mixed downward and atmospheric stability is even increased, thus BC containing aerosol can cause in a net cooling effect at the surface (MacCracken et al., 1986). On the other hand, thermal radiation from absorbed solar light in the lower parts of the atmosphere can actually contribute efficiently to surface warming. Reflections from the bright, high albedo ice and snow surfaces in the Arctic increase the amount of energy absorbed by aerosol :::::: aerosols : like BC. Aerosol particles are removed from the 10 atmosphere by sedimentation as well as nucleation, impaction and below cloud scavenging (e.g. Kondo et al., 2016). The result within the Arctic is likely a deposition of BC in ice and snow that can darken the otherwise highly reflective surfaces Tuzet et al., 2017, and references therein). Studies (e.g. Hansen and Nazarenko, 2004;Flanner et al., 2007) suggest that albedo decrease due to deposition can offset the cooling effect through dimming by higher atmospheric aerosol, but the balance of these effects in the Arctic can only be estimated in models as accurately as the distribution :::::: vertical ::::::::::: distributions 15 of BC is known. However, there is :::::: profiles ::: of ::: BC ::::::::::: concentration ::::: show a spread of more than one order of magnitude amongst different state-of-the-art models as well as between models and observations (AMAP, 2015).
Consequently, in order to provide accurate radiative forcing estimation in the Arctic region, it is necessary to understand what controls the vertical distribution of BC particles in the Arctic atmosphere. Import of polluted air from lower latitudes is controlled by the cold airmass that lies over the Arctic like a dome ::: with ::::::: sloping ::::::::: isentropes, ::: the :::::: isolines :: of :::::::: potential :::::::::: temperature 20 ::::::::::::: (Barrie, 1986;?). The polar dome's vertical temperature structure forces warmer air from the mid-latitudes :::: lower :::::::: latitudes to ascend along :::: those isentropic surfaces when transported into the ::::: colder :::: high :::::: latitude : polar region, reaching it in layers in the mid and upper troposphere (Stohl, 2006). The polar front between the cold polar and the warmer mid-latitude airmasses, a strong horizontal temperature gradient, acts as transport barrier that is controlling the intrusion of polluted air from southern source regions ::: into ::: the :::::: Arctic :::::::::::::::::::: (see . Emissions from sources within the cold polar dome :: air are transported through 25 the Arctic at lower altitudes. BC emitted from continental areas in the northern hemisphere is mainly carried poleward by midlatitude low-pressure systems and is eventually mixed across the polar front in the systems' warm and cold fronts. These frontal systems, with life-times :::::: lifetime : of 1-2 weeks, are frequently generated and poleward mass transport is continuously induced (Stohl, 2006;Sato et al., 2016). The polar dome :::: front retreats northward in the summer and leaves many pollution sources south of the polar front ::: this ::::::: transport :::::: barrier. Increased wet removal (scavenging) of aerosol particles is thought to help maintaining 30 much more pristine conditions throughout the Arctic in summer, compared to winter and spring (Barrie, 1986;Shaw, 1995;Garrett et al., 2011;Tunved et al., 2013;Raut et al., 2017). This pronounced seasonal variability of BC concentration was observed at ground based High-Arctic measurement sites (e.g. Eleftheriadis et al., 2009;Massling et al., 2015;Sharma et al., 2017), however the near surface air is decoupled from the mid and upper troposphere due to the high stability of the atmosphere (Brock et al., 2011) and these measurements cannot represent variability in the vertical (Stohl, 2006). The concentrations of 35 3 BC particles in the lower atmosphere might be affected by increasing numbers of local emissions. In fact, as the sea ice retreat makes the Arctic region more accessible, commercial activities in the marginal Arctic (the sea-ice boundary and boreal forest region), associated with flaring of gas in connection with oil production :::::::: extraction (Stohl et al., 2013;Evans et al., 2017) and shipping (e.g. Corbett et al., 2010;Aliabadi et al., 2015), are increasingand the : . :::: The possible consequences are an area of current research demand . Models aiming to assess the radiative forcing impact of Arctic aerosol 5 have shortcomings in the representation of concentrations and size distributions as well as their vertical variability, which was partly attributed to incorrect treatment of scavenging processes in parametrisations (Schwarz et al., 2010b;Liu et al., 2011).
Therefore, vertically resolved observations of aerosol mass and size distributions are an important benchmark for chemical transport models.
Nevertheless :::::: Despite ::::: their :::::::: important :::::::::: implications, measurements of the vertical distribution of BC and its variability in the 10 Arctic atmosphere are very sparse (AMAP, 2015). Aircraft campaigns like ARCPAC, ARCTAS and POLARCAT (Spackman et al., 2010;Brock et al., 2011;Kondo et al., 2011a;Matsui et al., 2011;Wang et al., 2011), PAMARCMIP (Stone et al., 2010), HIPPO (Schwarz et al., 2010b and ACCESS Raut et al., 2017) delivered limited numbers of BC vertical profiles from within the cold polar :::::::::: High-Arctic : airmass. To increase the validity and reduce biases of vertical profile measurements, which are :: the ::::::: vertical :::::::::: information :: on ::: the :::::::: presence :: of ::: BC :: in ::: the ::::::: Arctic, ::::: which :: is the basis to improved 15 ::: our system understanding, high-latitude observations at better spatial and temporal resolution are required. Such observations may resolve the internal variability due to weather changes as well as regional characteristics due to the prevailing atmospheric transport pathways with respect to differences between the seasons. This paper will discuss a set of measurements from the spring and summer aircraft campaigns in the NETCARE project (Network on Climate and Aerosols: Addressing Key Uncertainties in Remote Canadian Environments, http://www.netcare-20 project.ca). Motivated by the high sensitivity of the mechanisms of BC's radiative forcing in the Arctic climate system on its vertical distribution, the main goal of this study is to characterise the vertical variability of BC concentrations and particle properties in the polar dome, contrasting spring and summer. The campaigns yielded a unique and detailed dataset from within the polar dome at high latitudes, and covered time scales that give insight in the variability of aerosol distributions due to changes in the meteorological conditions and transport pathways for air pollution in spring and summer.   10 was used to detect BC particles. The operation principle and evaluations of the method are given by Stephens et al. (2003); Schwarz et al. (2006); Moteki and Kondo (2010). Briefly, the SP2 is based on the laser-induced-incandescence method: a concentric-nozzle jet system directs the aerosol sample flow through a high-intensity continuous-wave intra-cavity laser beam at a wavelength of 1064 nm, in which highly absorbing particles, such as BC, are heated to their vaporisation temperature and emit thermal radiation (incandescence). Particles containing a sufficient amount of BC (⇠0.5 fg) can absorb enough energy 15 to reach incandescence, which excludes sensitivity to other, less absorptive, material such as organic carbon, brown carbon or inorganic aerosol components. The peak intensity of the emitted thermal radiation, which occurs when the boiling point Table 1. Overview :: An :::::::: overview of all measurement flights of the NETCARE summer campaign 2014 and and the spring campaign 2015 that are evaluated in this study. :: The :::::: stations ::: are ::: the ::::: Arctic :::::: airfields :: the ::::: plane ::::: started :::: from ::: and :::::: returned :: to :::: (see ::: Fig. :: 1). :::: Two ::::: flights ::: took ::::: place :: on :: 08 :::: April ::::: 2015, ::::: which :: are :::::: refered :: to ::: with ::: the :::::::: shorthand ::::::: notations :: F1 ::: and ::: F2. temperature of BC is reached, is proportional to the BC mass contained in the particle. Following the terminology defined by Petzold et al. (2013), the refractory, essentially pure carbon, material detected with the SP2 is hereafter referred to as refractory black carbon (rBC). All other particulates that may be internally mixed with a BC core evaporate at temperatures below the boiling point of BC (⇠ 4000 C) such that no interference occurs in the quantification of the rBC mass (Moteki and Kondo, 2007).

5
The incandescence light detector, an avalanche photo-diode : a ::::::::::::: photomultiplyer :::: tube : with a 350 to 800 nm band-pass filter, used two gain stages. It was calibrated with a Fullerene Soot standard from Alfa Aesar (stock #40971, lot #FS12S011) by selecting a narrow size distribution of particles with a differential mobility analyser upstream of the SP2 (following Schwarz et al., 2010a;Laborde et al., 2012). The mass of these mono-disperse particles was empirically calculated using the relationship between mobility diameter and the effective density of Fullerene Soot (Gysel et al., 2011). The Fullerene Soot calibrations 10 used for the datasets of the two NETCARE campaigns agreed to within ±10% with each other, ensuring a good degree of comparability between the two campaigns in agreement with the reproducability :::::::::::: reproducibility : of SP2 rBC mass measurements evaluated by Laborde et al. (2012). After calibration, the SP2 allowed 100% detection efficiency of particles with mass in the range 0.6 to 328.8 fg, which is equal to 85 to 704 nm mass equivalent diameter (D rBC ), assuming a void free bulk material density of 1.8 g cm 3 . The SP2 was prepared for the research flights following the recommendations given in Laborde 15 et al. (2012). Stability of the optical system and laser power was confirmed during the campaign by measuring mono-disperse polystyrene latex spheres (PSL). An estimated total uncertainty of rBC mass concentrations is 15%, including reproducibility and calibration uncertainty (Laborde et al., 2012) and uncertainties of airborne in situ measurements (e.g. precision of the sample flow measurement). The SP2 was used to obtain rBC mass concentrations (M rBC ) , rBC number-size distribution weighted by particle mass (mass-size distributions, : MSD) and mass-mean diameters (MMD) of rBC particles. 20 The measured M rBC were not corrected for the mass of particles outside the detection range, and are thus only valid for the range 85 to 704 nm. The contribution of small Aitken mode particles as well as particles larger than 704 nm to the total PM 1 rBC mass (mass of particles smaller than 1000 nm) may be significant and the measurements presented here can underestimate the total PM 1 mass by variable degrees. Approaches as used by Sharma et al. (2017), to estimate the total PM 1 rBC mass by fitting a lognormal distribution to a measured particle MSD, cannot be applied to aircraft measurements since MSD vary 5 with location and altitude and statistics are insufficient to derive multivariate correction factors. The underestimation of the total PM 1 mass due to the contribution of particles smaller than 85 nm were estimated :::::::: calculated for selected cases to be an additional 4.5% (between 2 and 7%) rBC mass in the summer polar dome, 7.5% (between 4.5 and 8.5%) in the lower spring polar dome and up to 10% (7.8 to 12%) within high concentration pollution plumes. Assuming the SP2 was likely able to count (but not size) all particles between 700 and 1000 nm, an infrequent (< 30 particles/flight) underestimation of the PM 1 mass due 10 to large particles occurred in spring in high concentration plumes as well as in the lower atmosphere. No influence of particles larger than 700 nm was apparent for summer conditions.
The particle number-size distributions and number concentration of the total aerosol (TA) were measured with a DMT Ultra-High Sensitivity Aerosol Spectrometer (UHSAS). As described in Cai et al. (2008), the UHSAS measures the scattered light intensity of individual particles crossing an intra-cavity solid-state laser (Nd 3+ :Y LiF 4 ), operating at a wavelenght of 15 1054 nm, to evaluate the particle size (under the assumption of the refractive index of PSL particles and spherical shape).

Trace gases
Carbon monoxide (CO) was measured with an Aerolaser ultra fast CO monitor model AL 5002 based on VUV-fluorimetry, An inlet flow of approximately 12 L min 1 was continuously monitored.
Atmospheric BC and CO are often co-emitted from the same combustion sources (Streets et al., 2003), but the relative ratio of the species depends on the combustion type, i.e. fuel types such as biomass or fossil fuel, and combustion conditions, such as flaming, smouldering or (engine) internal. Other than :: As ::::::: opposed ::: to aerosol, which is affected by dry and wet removal 15 mechanisms, CO can be used as a nearly inert combustion tracer within timescales of a few weeks(. ::::: This :::::::::: assumption :: is neglecting possible sources of CO due to biogenic production or by means of oxidation of other trace gases (Gaubert et al., 2016)). The ratio of rBC to CO relative to their background levels (R CO ) can be used as an indicator for when rBC particles were depleted by removal processes Stohl et al., 2013). Based on the measured rBC and CO concentrations, the ratio is calculated as R CO = rBC/ CO = M rBC / CO (with units ng m 3 ppbv 1 ). A background completely depleted 20 of rBC is assumed throughout the column as in previous studies (e.g. Kondo et al., 2016) due to the short atmospheric lifetime of BC in the order of days to weeks (Bond et al., 2013). The CO background value is altitude dependent ( Fig. ?? ::: S1) and hence defined as the fifth percentile value of all CO mixing ratios observed within defined altitude intervals, following Kondo et al. (2016). R CO is only calculated when CO exceeded the measurement uncertainty.

Meteorological parameters 25
The meteorological state parameters pressure, humidity and temperature were recorded at 1 Hz resolution with the basic meteorological sensor suite and data acquisition of Polar 6. The ambient air temperature was measured with a PT100 type sensor mounted to the aircraft fuselage in a Goodrich/Rosemount 102 EK 1BB housings :::::: housing : with deicing facility. Corrections for the deicing heat and adiabatic temperature increase due to pressurisation of the airflow inside the sensor housing (RAM raise and recovery factor) were applied to the temperature readings (following Stickney et al., 1994). The relative humidity (RH) 30 was measured with a Vaisala humidity sensor HMT333 mounted inside a Rosemount housing Model 102 BX, which is also deiced and similar in its flow characteristics to the housing of the temperature probe. The saturation vapour pressure and RH are corrected with the actual ambient temperature from corrected PT100 readings. The potential temperature was calculated from ambient temperature and the ambient pressure from a static pressure probe.
A Forward Scattering Spectrometer Probe (FSSP), model 100, by Particle Measuring Systems (PMS Inc., Boulder, CO) was used for the measurement of cloud particles. Data from the probe, which was mounted in a canister on a wing pylon, were analysed in more detail in Leaitch et al. (2016)and : . :: It contributes to the following analysis as indicator for visible and invisible 5 clouds by an empirically chosen threshold above instrument noise level to the measured cloud particle concentrations (droplets and ice crystals). Aerosol data has been masked when the aircraft was in clouds.

Model weather data and transport pathway analysis
The ERA-Interim re-analysis data (Dee et al., 2011) from the European Centre of Medium-Range Weather Forecasts (ECMWF) is analysed at certain pressure levels in the form of classical weather maps in order to understand the meteorological situation 10 in the Arctic during the period of our measurement flights. ECMWF operational data is further used to drive the Lagrangian analysis tool (LAGRANTO: Wernli and Davies, 1997;Sprenger and Wernli, 2015) and its kinematic back-trajectories are analysed to estimate the regions of origin for polluted air encountered during the research flights. The model's input data has a horizontal grid spacing of 0.5 with 137 hybrid sigma-pressure levels in the vertical from the surface up to 0.01 hPa. Trajectories were initialized every 10 seconds from coordinates along the research flight tracks and calculated 10 days back in time. 15 The time series of trajectories along the track of the aircraft were correlated with in-situ measurement values, in particular with rBC and CO concentrations, in order to relate individual features in the vertical profiles to an ensemble of trajectories by the means of threshold filtering. Trajectories fulfilling these criteria are displayed on maps in Sec. 3.4, that also show spatial data ::: (see :::: Sec. :::: 3.4). :::: Due :: to ::: the :::::::: potential ::::::: influence ::: of :::: wild :::: fires ::: and ::: gas :::::: flaring ::: on ::: BC ::::::: presence :: in ::: the :::::: Arctic :::::: region, ::: the :::::: spatial ::::::::: distribution : of gas flaring sites from the ECLIPSE (Evaluating the Climate and Air Quality Impacts of Short-Lived Pollutants) 20 emission inventory (Stohl et al., 2015;Klimont et al., 2017) and active fires from the MODIS level 2 satellite product (Giglio et al., 2003) in order to mark potential rBC sources :::: were ::: also :::::::::: considered :: for ::: the :::::::::::: interpretation :: of :::::::: trajectory :::::::: pathways.

Meteorological overview
With the focus on the polar dome and the vertical distribution of rBC therein, subsets of the flights in spring 2015 and summer 25 2014 were selected for this analysis, which . :::: The :::::: subset :::::::: selections : are based on the variability of the polar dome's position and southern boarder ::::: border. The structure and extent of the polar dome in both seasons has been evaluated by , who defined the polar dome based on trace gas gradients measured during the NETCARE campaigns. They found that the polar dome boundary was, on average over the course of the campaigns, located between 61.5 :: 66 N and 64.5 :::
The meteorological situation in April 2015 was dominated by a pool of very cold air centred over the Canadian Arctic Archipelago that surrounded the stations Alert and Eureka on Ellesmere Island. The cyclonic flow surrounding the cold air stabilised this system by blocking perturbations of low-pressure systems (Fig. 2). The polar vortex was in a weak state and not well defined. Cold airmasses in the Russian Arctic were cut off from the dome over the Canadian Arctic Archipelago. Near the 10 beginning of the measurement period, a strong low-pressure system caused an outbreak of cold air over Eastern Europe, while warm mid-latitude air moved poleward further west. This synoptic feature affected Alert strongest on 8 April, and its influence was diminishing during the measurements around Eureka 11 to 13 April (Fig. 2). Conditions during all flights were low wind speeds and clear sky with only few, mostly thin clouds .
The NETCARE summer campaign 2014 operated in an area of the high Canadian Arctic that was situated within the summer 15 polar dome. The first half of the campaign (4 to 12 July) was characterised by a northern influence (Fig. 3). The atmosphere featured a low boundary layer height capped by a distinctive temperature inversion leading to a very stable stratification of the lower troposphere. Prevailing conditions for the research flights were a clear sky, only few or scattered clouds and low wind speeds (Burkart et al., 2017). These conditions gave the opportunity to characterise the summer polar dome in undisturbed conditions, when 6 flights with a total of 28 vertical profiles were conducted in the study area around Resolute Bay on Lancaster Sound (see map in Fig. 1). Starting from 13 July 2014, the weather pattern changed and Resolute Bay got into the transition zone between polar and mid-latitudinal air, as a consequence of a low-pressure system coming from the north-western Beaufort Sea and passing south of Lancaster Sound. Bad visibility due to fog, clouds and precipitation impeded flights until 17 July, on which day the study area shifted back into the cold airmass and a westerly air movement (Fig. 3).

Seasonal characteristics of rBC vertical distribution in the polar dome
In this section, the vertical distribution of rBC is examined focusing on changes from spring to summer. For each ascent or descent of the flights listed in Tab. 1, data points within fixed pressure-altitude intervals were averaged. These profiles were then successively averaged to mean flight profiles and mean regional profiles, thus avoiding sampling biases due to varying sampling times in each altitude interval. As shown in Fig. 4 , ::::: shows :::: that there are substantial differences between the average 10 spring and summer profiles of rBC mass concentration (M rBC ), mass-mean diameter (MMD) and rBC to TA number ratio (R numTA ).
The absolute and relative presence of rBC was generally reduced during summer. Ground-based observations at High-Arctic sites like Alert show a pronounced seasonal cycle in rBC concentrations (e.g. Leaitch et al., 2013;Stone et al., 2014;Sharma et al., 2017), which is well matched by the difference in mean M rBC of observations in the lower part of the atmosphere 15 (>920 hPa) presented here. During spring, averaged M rBC of 31.5 and 30.1 ng m 3 were present in the Alert and Eureka region, respectively, while summertime observations showed one order of magnitude lower mean M rBC of 1.4 ng m 3 . Figure 4a shows that this seasonal difference is ::: was : present throughout the vertical extent of the polar dome. A one order of magnitude difference in M rBC between the seasons was also found during the ARCTAS spring and summer campaigns in 2008 reported by Matsui et al. (2011). However, their observed M rBC were a factor of two higher compared to the NETCARE observations. This might be due to the fact that their observations were from a Sub-Arctic region (northern Alaska), where pollution supply and removal not necessarily are in the same balance as within the polar dome. These two observations of pronounced difference in M rBC between the seasons highlights how wet removal becomes more efficient during summer within the polar dome, but as well 5 already during northward transport outside the dome ::: The :::::: balance :::::::: between ::::: supply :::: and ::::::: removal ::::: (under :::: way :: or :::::: within ::: the :::::: Arctic) :: of :::: rBC :: at :::: high :::::::: latitudes :::::: appears :: to ::::: have : a :::::::::: pronounced ::::::::: seasonality.
During spring, mean profiles from the Alert and Eureka regions showed a similar M rBC range, however, the M rBC vertical trend :::::: vertical ::::: trends :: in ::::: M rBC showed certain differences between the two regions. At Eureka, the maxima of the mean profiles occurred between 900 and 800 hPa with averaged M rBC of 55 ng m 3 and an interquartile range (IQR) of 11-120 ng m 3 .

Vertical distribution of rBC relative to potential temperature
In order to fully understand the vertical variability of the aerosol distribution, it is important to consider the vertical structure of the polar dome with its core of cold, dense air at the ground and successive dome shaped layers of warmer air above.

Vertical distribution of rBC in the spring polar dome
The spring mean flight profiles from Alert and Eureka averaged over intervals of potential temperature are shown in Fig. 5.
The potential temperature range 277 to 285 K (level IV) is ::: was in the transition zone to the airmass above the dome . Within a temperature gradient zone marking the upper boundary of the dome, all profiles peak before sharply decreasing in the airmass above. This transition is also apparent in a gradient of trace gas concentrations , and occurs at slightly varying temperature. The maximum M rBC on 8 April were comparable to that in the level below, but high R CO around 6 ng m 3 ppbv 1 suggest a different, more efficient transport to this level. M rBC on the higher end of the IQR in level IV (145 ng m 3 ) were encountered by one out of three profile flights on 11 April. The other two profiles included in the mean of that day encountered air depleted in rBC, where low MMD as well as R CO suggest substantial removal of rBC from 5 the airmass by precipitation. However, rBC reaches ::::::: reached its maximum contribution to the TA by number :::: with :::: mean ::::::: R numTA of 6.2% : in ::::: level :: IV.  defined the region of potential temperatures higher than about 285 ::: 287 K (level V) to be outside the polar dome due to : in ::: the :::::::: transition :::: zone :: to ::: the :::::::::: troposphere ::::: above ::: the ::::: polar ::::: dome :::: with a strong negative gradient in CO concentrations and stronger connection of transport trajectories to mid-latitudes. At the highest altitudes of the profiling flights, low combined with a low R CO (0.7 ng m 3 ppbv 1 ) suggest that polluted air was transported to this level, but scavenged of much of the BC during lifting of the air parcels.

Vertical distribution of rBC in the summer polar dome
As for the spring case, the ::: The : variability of aerosol properties :::: was :::::::::: investigated ::: also ::: for ::::::: summer : as function of the potential temperature was investigated for summer within the polar dome over the area of Resolute Bay (Fig. 6). As already introduced 20 in Sec. 3.2, the general concentration of rBC particles was almost one order of magnitude lower and the variability in the distributions had a lower absolute magnitude than the spring observations. Two strong temperature gradients ( Fig. ?? ::: S1b) structured the atmosphere below 5 km into three levels in which similar vertical tendencies of rBC concentration and mixing ratios were observed.
Close to the surface, within air at potential temperatures between 273-284 K (level I), the 75 th percentile M rBC did not exceed recorded at low altitude, where mean R numTA was 0.5%. The R CO well below 1 ng m 3 ppbv 1 suggested, combined with the low particle diameter (average MMD of 125 nm), that particles in the summer polar dome were subject to strong wet removal.
The MMD values show a larger variance amongst the different profiles due to few particles in the statistics.

30
A weakly stable to neutral atmospheric level was present above the stable near-surface level and up to a strong temperature gradient aloft (level II), in which . : M rBC was relatively constant within the lower part : of ::::: level :: II, but increased within the temperature gradient zone in the :: its upper part. This zone lay around 288 to 294 K ( Fig. ?? ::: S1b) in the period before the weather change (4-12 July) and lower, around 284 to 290 K, on 17 July after the perturbation of the polar dome by a low pressure system (see Sec. 3). The altitude of the gradient zone was changing amongst individual profiles flown in different regions and was likely affected by orography and the variable sea-ice cover (see map in Fig. 1). High humidity was frequently observed in the neutral or weakly stable zone below the temperature gradient. Highest values of ::: The ::::::: highest M rBC ::::: values : up to 12 ng m 3 were encountered around 286 K on 17 July, while in the earlier ::::::::::: measurement : period, the mean profiles of M rBC peak at only around 4 ng m 3 , which is however still a factor 2 increase over the concentrations within the less stable lower 5 part of this level. Also the relative presence of rBC showed a significant difference between the two parts. R numTA reached a mean of 1.3% (1.7%) within the concentrations peaks in the first (second) period, while the background in the lower part was around 0.6%. Similarly, R CO was 0.2 ng m 3 ppbv 1 in the background and mean profiles reached 0.6 ng m 3 ppbv 1 within concentrations ::::::::::: concentration : peaks in the first period. Although the highest rBC concentrations were encountered on 17 July, R CO of 0.0 to 0.3 ng m 3 ppbv 1 indicate that rBC aerosol was depleted compared to the ::::::: strongly ::::::: depleted ::::::: relative :: to 10 co-emitted CO, which featured elevated concentrations throughout the column compared to the first weather period of stable northern influence (Fig. ?? :::: S1b).

rBC source areas for the spring polar dome
The aerosol over Alert and Eureka in the period 7 to 13 April was influenced by air transport from Eastern Europeand Central Asia : , :::::: Central :::: Asia :::: and :::::: Siberia as well as North America (Fig. 7). In those regions lobes of cold polar air reached south due to cyclonic perturbations of the polar front (Fig. 2).
While a shift in R CO can indicate that a different type of combustion source influenced the airmass (as discussed in Sec. 3.3), it 10 is not apparent from the back-trajectories that a clear shift from one source type to another happened between the two levels.

15
As a likely consequence of the air being lifted up above the polar dome , rBC was mostly found to be depleted relative to CO, possibly due to precipitation events. Due to the heterogeneity of these events, the variability amongst observations, which are available from three days, is large. .
The mean MMD showed a general trend towards smaller particles with increasing altitude in spring and variability amongst individual profiles indicated sensitivity of the MMD to different atmospheric processing of the aerosol during transport (Sec. 3.3.1). MSD from higher levels (III and above) had a narrower mode with slightly smaller peak diameter. Furthermore, the 10 distribution was clearly shifted towards smaller particles with decreased contribution of rBC cores with diameters larger than about 300 nm, which were only high in number within the lower levels.
Averaged rBC mass size-distributions normalised with the total mass concentration under the curve for levels I-V from the spring polar dome observations around Alert (A) and Eureka (E), as well as for level II from the summer polar dome 30 observations around Resolute Bay.

4 Summary and conclusions
Two aircraft campaigns within the NETCARE project during spring and summer made it possible to observe the vertically variable :::::: allowed ::::::::: observing ::: the :::::: vertical distribution of black carbon aerosol over the high-latitude Canadian Arctic .
Vertical profiles of pressure (lines in grey shading) and CO mixing ratio (lines in yellow shading) versus potential temperature for spring (a) and summer (b). Hatch patterns indicate the extent of the atmospheric levels defined in Sec. ??. The thick pink line marks the fifth percentile of measured CO mixing ratios which is assumed to represent the background concentrations in 20 each potential temperature interval.
Author contributions. HS wrote the manuscript, with significant conceptual input from MZ, WRL, ABH, and critical feedback from all co-authors. HS, HB, MDW, JB, and WRL operated instruments in the field and analysed resulting data. HB, DK and PMH ran LAGRANTO simulations and HS analysed the resulting data with input from HB. WRL, JPDA and ABH designed the field experiment.