Intense plumes of bromine monoxide (BrO) are regularly observed over sea ice during polar spring by satellite e.g. and ground-based instruments e.g.. Although subsidence of stratospheric air towards lower altitudes can substantially increase total column BrO , several studies have shown that the plumes are often of tropospheric origin and occur in conjunction with widespread ozone depletion . The latter is caused by an autocatalytic, heterogeneous chemical cycle, the so called “bromine explosion” , in which gas phase molecular bromine is photolysed in the presence of sunlight and oxidised subsequently by ozone to form BrO. The latter then reacts with HO2 to form HOBr which is eventually removed from the atmosphere by wet scavenging. The exact chemical reaction cycle as well as the substrate, from which bromine is initially released to the gas phase are still unclear . However, there is general agreement that the source must be rich in sea salts and specifically in Br- which reacts within the condensed phase substrate to form Br2 which is released to the atmosphere: HOBr+H++Cl-→BrCl+H2OBrCl+Br-⇌Br2Cl-Br2Cl-⇌Br2+Cl-HOBr+H++Br-→Br2+H2O

It is important to note that a pH lower than 6.5 is required for an efficient bromine activation cycle . The possible sequence of reactions involved in the bromine explosion is given in detail in numerous studies e.g..

In the past, it was widely believed that frost flowers are a primary source of bromine involved in ozone depletion due to their large surface areas and salinities of about three times higher than in sea water . combined sea ice coverage and tropospheric BrO from satellite remote sensors with regions potentially covered by frost flowers derived from a simple thermodynamic model. This was based on noting cold surface temperature and related conditions associated with source regions for BrO. They concluded that young ice regions potentially covered by frost flowers are the source of bromine in bromine explosion events (termed BEEs in the following). However, stated that the role of frost flowers for heterogeneous reactions should be reconsidered, as they measured the total surface area of frost flowers in the Arctic to be only 1.4 m2 m-2 of ice surface. investigated frost flowers in the lab and could not observe release of aerosols despite wind speeds in gusts up to 12 m s-1. Correlating BrO measurements with air mass histories from meteorological back trajectories, identified snow and ice contaminated with sea salt on first-year sea ice as a more likely bromine source compared to frost flowers at Barrow in Alaska. However, results by suggest that blowing snow could be salinated by frost flower contact and that, as a consequence, frost flowers and blowing snow in combination are a source of atmospheric bromine.

found a good agreement with satellite-derived tropospheric BrO when including sublimation of salty blowing snow as a bromine source in a chemical transport model. conducted snow chamber experiments on various types of snow and ice surfaces at Barrow, Alaska, and concluded that photochemical production of molecular bromine in surface snow may serve as a major bromine source. They found the most effective production rates of Br2 for tundra snow and the uppermost 1 cm thick layer of snow on top of first-year sea ice. As the snow chambers were located close to the coast, the inland tundra snow was most likely salinated by atmospheric processes. Using GOME satellite data, linked the development of boundary layer BrO plumes to locations of 1-year old sea ice.

Younger sea ice has gained much attention in studies on BEEs as it is much more salty than older ice so that snow lying on this ice can easily accumulate sea-salt . Moreover, liquid brine which forms on fresh ice during the freezing process is highly concentrated in sea-salts. Frost flowers growing on the ice or snow lying on top of it can get coated with the brine through capillary forces . The likelihood of production of atmospheric Br2 through heterogeneous reaction is enhanced, if the frost flowers, salty snow or sea salt aerosols are lifted up into the air by high wind speeds as will be described below. employed ground-based measurements from the International Polar Year together with satellite observations across sea ice sectors in Alaska and the Canadian Arctic. They showed that stronger BEEs occurred in 2009, when springtime perennial sea ice extent was reduced compared to the previous year. The authors concluded that the strength and frequency of BEEs may increase in the future, as perennial sea ice is replaced by younger, and hence saltier sea ice due to global warming.

investigated how atmospheric particles produced from alkaline seawater can trigger the acid-catalysed bromine activation cycle using a 1-dimensional atmospheric chemistry model. They concluded that below a temperature of 265 K most of the carbonate precipitates, which reduces the buffering capacity of sea water and hence facilitates its acidification. Moreover, at low temperatures, the equilibrium constant of Reaction (R2) shifts towards Br2Cl- which is then transformed to Br2. As ozone depletion events occur at a wide range of low temperatures , the role of low temperatures for ozone depletion events is still uncertain e.g.,.

Reactive bromine plays a key role in oxidising gaseous elemental mercury to reactive gaseous mercury . This increases deposition of mercury to the snow and ice which is harmful to the environment . BEEs therefore not only cause ozone depletion events, but also atmospheric mercury depletion events during polar spring.

Satellite images show that many BrO plumes observed over polar sea-ice regions are spiral or comma-shaped and resemble high-latitude cyclones in appearance. However, not much is known on the role these weather systems play for the formation, duration and transport of BEEs. In the past, it was widely believed that all BEEs form within a stable boundary layer and are accompanied by low near-surface wind speeds. Subsequent transport of the bromine plumes from their genesis regions explained why large concentrations of bromine were also observed at high wind speeds e.g.. In contrast to this, showed that a stable boundary layer acting as a “closed reaction chamber” is not a prerequisite for the development of BEEs. developed a qualitative model showing that the likelihood of ozone depletion is strongly enhanced at very calm weather conditions and at wind speeds larger than 10 m s-1. They argued that both weather situations increase the number of reactants in the air and facilitate contact between the gaseous and condensed phase, thereby favouring bromine explosions. Their findings were supported by observations of an Antarctic BEE, for which high wind speeds caused by a cyclone and saline blowing snow were reported. investigated the vertical structure of ozone depletion events based on ozone measurements from two Antarctic field campaigns and found that those events which occurred at wind speeds below 7 m s-1 did not exceed 40 m in the vertical, while those observed at altitudes above 1 km were accompanied by high wind speeds caused by low pressure systems. They concluded that high wind speeds and rising motions within cyclones cause uptake of snow to the air, which in turn caused the observed bromine explosions. conducted a case study of a BrO plume in the Arctic, which was seen on GOME-2 satellite images for at least 5 days and was transported by a cyclone over a large distance. They concluded that recycling of BrO on wind blown snow or aerosol surfaces enhanced the lifetime of the plume substantially.

The present study aims to improve knowledge of the role of high-latitude cyclones in BEEs. We present GOME-2 satellite observations of a tropospheric BrO plume which developed in late March 2011 over the Beaufort Sea to the north of Alaska. As will be shown below, the evolution of this BEE is closely linked to weather conditions and transport within a polar cyclone and it is therefore termed “bromine cyclone transport event” or BCTE in the following. Here, GOME-2 retrievals of tropospheric BrO are regarded as an indicator of activated bromine species (such as Br, Br2, HOBr and BrCl) in general, although activated bromine species may also be present in the absence of BrO. The regional Weather Research and Forecasting (WRF) model is used to investigate meteorological conditions during the BCTE. As only columns of tropospheric BrO, i.e. no information on vertical distribution, are available from GOME-2, runs with the Lagrangian FLEXible PARTicle dispersion model (FLEXPART) are carried out to derive information on the altitude of the BrO plume. Moreover, conclusions on the location of the plume in relation to airflows within the polar cyclone and on possible BrO sources are derived from additional satellite data. In contrast to previous studies on BCTEs using satellite data, we show that the dry conveyor belt as a potentially bromine-rich stratospheric airstream is spatially separated from the BrO plume for the case investigated.

Satellite data used in the present study will be described in Sect. . Details on WRF and FLEXPART model set-ups are given in Sect. , followed by results in Sect. . The paper ends with a summary and conclusions (Sect. ).

In this section, results on links between the polar cyclone and the associated bromine explosion will be presented. In principal there are three conceivable explanations for the occurrence of polar tropospheric BrO plumes in GOME-2 satellite images: (1) the plume is a result of light path enhancement within clouds possibly combined with shielding of BrO below optically thick clouds at other locations, (2) the plume is due to subsidence of stratospheric, bromine rich air towards lower altitudes, which strongly enhances total column BrO and the stratospheric correction method failed to remove this air mass from tropospheric BrO VCDs, (3) the plume developed due to the bromine explosion chemical reaction in the boundary layer with possible subsequent transport towards higher altitudes in the troposphere. In Sect. , it was shown that the first explanation is not valid for the BCTE investigated in this paper. We will show in the following that the second point is also not valid, whereas plume development due to the bromine explosion mechanism in the boundary layer can reasonably explain the occurrence of the observed tropospheric BrO plume.

Figure  shows satellite observations, together with parameters used for GOME-2 tropospheric BrO retrieval, covering different development stages of the BCTE. The first GOME-2 observation of the tropospheric BrO plume was made on 31 March at approximately 22:00 UTC (not shown). At this time, the plume was already shaped like a comma, resembling the clouds of a polar cyclone in appearance. As the bromine explosion chemical reaction cycle requires daylight, the BCTE most likely developed between sunrise (approximately at 15:40 UTC) and 22:00 UTC on 31 March. It was generated to the north of Alaska over the Beaufort Sea, which was covered by sea-ice at this time of the year. The location as well as spatial pattern of the plume did not change from the first to the second GOME-2 observation (see next paragraph). Moreover, similar meteorological conditions were present at plume location for both satellite observations.

At 23:30 UTC on 31 March (Fig. a, left panel), absolute values of tropospheric BrO VCDs had increased by about 4×1013 molec cm-2 inside the plume. This may indicate that the BCTE had intensified, but may also result from differing satellite viewing conditions. Figure  shows meteorological conditions from the WRF simulation corresponding to GOME-2 observation times shown in Fig. . Comparison of both figures shows that the plume was located to the west of a low pressure system at an occluded front (shown by 2 m temperature patterns in Fig. c, left panel), which means that northerly wind directions prevailed during the development of the event. Wind speeds reached about 10 m s-1 at the plume location. Note that fronts indicate vertical lifting. This is in agreement with the, relative to the surrounding areas, high planetary boundary layer height (Fig. e) values at plume and front location. The latter is true for all development stages of the BrO plume.

On 1 April at 21:30 (Fig. a, middle panel), the BrO plume had fully developed. WRF simulations indicate that the plume was transported cyclonically eastwards around the low pressure system, which had also deepened at this time of development, reaching minimum mean sea level pressure values of approximately 983 hPa (see Fig. a, middle panel). South-westerly winds of up to 15 m s-1 prevailed at the plume location. The wind speed maximum was located around 160∘ W, 76∘ N which coincides with convergent wind directions, indicating strong north-eastwards directed uplift at the southern end of the BrO plume, which is in agreement with the planetary boundary layer height pattern shown in Fig. e.

On 2 April at 19:30 UTC (Fig. a, right panel), the plume had moved further north-eastwards with the low pressure system and reached the Canadian High Arctic Archipelago. Values inside the plume had decreased and the plume had lost its comma-shape indicating the dissolving stage of the BCTE. Likewise, the low pressure system weakened and the cyclone lost frontal structure (see Fig. , right panels). However, wind speeds were still quite high (up to about 13 m s-1) and blowing from southerly directions at plume location.

On 3 April (not shown), parts of the diluted plume were observed at the Polar Environment Atmospheric Research Laboratory (PEARL) on Ellesmere Island by ground-based Multi Axis-DOAS. The BCTE arrived at PEARL late on 3 April. Investigation of NCEP/NCAR Reanalysis data (not shown) reveals that by 5 April, the weather system associated with the BCTE had joined another, more southerly low-pressure system. This resulted in low wind speed weather conditions different from those discussed in this study. The measurements at PEARL, documenting the arrival of the plume during blowing snow weather conditions followed by local recycling of BrO under stable shallow boundary layer conditions, are described in detail by . Overall, the observed lifetime of the high wind speed BCTE is about 4 days according to GOME-2 observations, covering the onset (evening of 31 March), mature stage (evening of 1 April) and dissolving stage (evening of 2 April).

The location of the BrO plume observed by GOME-2 broadly coincides with regions of low temperatures around 350 geopotential metres (gpm) (see Fig. d, the difference between geopotential heights and altitudes above ground is assumed to be negligible) simulated by WRF, although the relation is less clear during the development of the event compared to later stages. The BrO plume location also broadly coincides with regions of low temperatures at higher altitudes up to roughly 500 gpm during the development stage and roughly 1000 gpm for the mature and dissolving stage of the BCTE. This is in agreement with the results by (see Sect. ), who found that recycling of BrO on aerosol surfaces is most efficient at low temperatures.

Figure  shows sea ice thickness retrieved by SMOS for 1 April. This date is chosen as a proxy of sea ice thickness conditions for other days during the BCTE (potential bromine sources deduced from SMOS images do not change significantly from late March to early April 2011). SMOS shows reduced sea ice thicknesses in the area around 170∘ W, 77.5∘ N and 158∘ W, 74∘ N (these regions are indicated by black arrows in Fig. ). Comparing GOME-2 observations of the BrO plume to SMOS retrievals and considering wind directions simulated by WRF, we infer that the former identified region may have acted as a bromine emission source during the onset of the BCTE, while the latter region may have been a source of bromine during the mature stage of the event.

To identify the location of the BrO plume with respect to cyclonic air flows, ozone VCDs [DU] from GOME-2 as well as MODIS false colour images close to GOME-2 observation times are shown in Fig. c and e, respectively. Note that further inspection of all MODIS observations of the BCTE available before and after each GOME-2 observation indicates that MODIS orbits shown in Fig.  are to a good approximation representative of cloud and/or blowing snow conditions at GOME-2 observation times. The dry conveyor belt is a low moisture, ozone-rich air stream within an extra-tropical cyclone, descending from the lower stratosphere towards tropospheric altitudes. On satellite images, a dry intrusion can be identified as a nearly cloud-free region sandwiched between a high-topped cloud band associated with the cold front and an often lower cloud head . The location of this air stream, coinciding with high ozone VCDs observed by GOME-2, is indicated by red arrows plotted on top of MODIS false colour images in Fig. e. Apart from enhanced ozone VCDs near the cloud head, where the dry conveyor belt most likely overlaps with tropospheric air flows, the dry conveyor belt is clearly separated from the BrO plume. Moreover, the plume pattern is not significantly correlated with low WRF tropopause heights shown in Fig. d. In this sense, this BEE differs from previous case studies for which the dry conveyor belt complicated the interpretation of tropospheric BrO or total column BrO from satellite retrievals e.g.,, so that the contribution of stratospheric air to the observed BrO plumes remained uncertain. Note that the high ozone values coincide with high GOME-2 stratospheric BrO VCDs (Fig. b), which is in agreement with conclusions drawn in this paragraph.

The MODIS images shown in Fig.  reveal that the BrO plume moved with the occluded front and generally coincided with cloudy areas. As clouds indicate vertical mixing, it is likely that the plume development is closely linked to regions of vertical uplift and high wind speeds near the cyclone centre. In agreement with the study by , these weather conditions are favourable for blowing brine wetted snow production, which may have acted as a bromine source during the BCTE. This conclusion is strengthened by mesoscale features shown in Fig.  (left panel), which shows a zoom of the MODIS image for 1 April at 22:20 UTC in Fig. . There are yellow-brown, parallel, stripy features visible in the image. Blowing snow detection from MODIS is difficult as clouds and blowing snow particles both stand out in the mid-infrared and therefore appear yellow in Fig. . It is therefore not clear if the stripy features are snow billows or just cloud streets, or a mixture of both. Nevertheless, as shown by WRF simulations (see above), the stripes occur in an area of high wind speeds and convergence, which most likely causes vertical lifting near the ground. Moreover, SMOS satellite observations (Fig. , right panel) show reduced sea ice thicknesses in the area of the stripy features observed by MODIS (see above). A reduced sea ice thickness indicates younger and saltier sea ice. Snow lying on top of younger sea ice, is covered in brine and more salty itself. This possibly favoured the bromine explosion chemical chain reaction together with weather conditions in the area of stripy features observed by MODIS.

Left panels in Fig.  show CALIOP vertical feature mask giving insight into vertical cloud and aerosol distributions inside the BrO plume for all development stages of the BCTE. CALIPSO footprints corresponding to these CALIOP observations are given by red and black dotted lines plotted on top of MODIS false colour images and GOME-2 tropospheric BrO VCDs in Fig. e and a, respectively. Again, further inspection of all MODIS observations of the BCTE available before and after each CALIOP observation indicates that MODIS observations shown in Fig.  are to a good approximation representative of cloud conditions at CALIOP observation times. Comparing the vertical feature masks with GOME-2 tropospheric BrO VCD along corresponding CALIPSO footprints (Fig. , right panels) shows that at the plume location, clouds and aerosols were restricted to about 3 km height in the vertical (with the exception of the onset of the event, for which some parts of the plume occurred in an area of higher cloud tops). Note that approximate BrO plume locations are indicated by red dashed boxes in Fig. . Cloud tops indicate boundaries regarding vertical mixing. Hence, it is likely that vertical transport of tropospheric bromine from the ground was also limited to 3 km height along CALIPSO footprints. However, the maximum time difference between CALIOP and GOME-2 observations is about 1.5 h so that cloud and aerosol conditions shown in Fig.  may differ in the vertical from the one at GOME-2 observation time.

The WRF simulations indicate that the planetary boundary layer height (Fig. e) did not exceed 1 km in the vertical at plume location. This means that the BrO plume must have been transported out of the planetary boundary layer into the free troposphere, given that transport of BrO was most likely limited to 3 km height in the vertical.

FLEXPART simulations from FS1 for the mature and dissolving stage of the BCTE are displayed by Fig.  together with corresponding GOME-2 observations of tropospheric BrO VCD for reference. For 1 April, the best agreement between FLEXPART and GOME-2 is achieved by assuming that the BrO plume was located within a 1 km thick layer at the surface at time of initialisation (1 April at 00:00 UTC). The magnitude of BrO observations within the plume is reproduced well by FLEXPART. The model underestimates background values outside the BrO plume. The latter is most likely due to the fact that only satellite retrievals within a specific area around the BrO plume and above a BrO threshold value were used for model initialisation (see Sect. ). This is also the case for all other FLEXPART results shown in this paper. The agreement between satellite retrievals and model output is also rather good regarding spatial distribution of the plume, for runs initialised by plumes between 1 and 7 km altitude (see Fig. c). Note that 1–3 and 5–7 km plume results broadly resemble those for 3–5 km runs and are therefore not shown in Fig. . However, the magnitude of BrO tropospheric VCDs is considerably lower than those of the satellite observations and the plume slightly turns anticlockwise with height, so that the spatial agreement is not as good as for the surface plume run. Simulations substantially lose resemblance to GOME-2 retrievals for the 7–9 and 9–11 km runs (Fig. d and e, respectively), suggesting that a stratospheric origin of the BrO plume is rather unlikely.

In contrast to GOME-2 observations, FLEXPART still simulates a comma-shaped plume at the dissolving stage of the BCTE for the 0–1 and 3–5 km simulations. Moreover, the plume is located northwards of where it actually occurred. The lowest elevation run overestimates satellite retrieved tropospheric VCDs of BrO. This may be due to the fact that no removal processes of BrO are included in the FLEXPART set-up. The higher elevation runs do not show a comma-shaped plume but again, the simulated plume is located further northwards of where it actually occurred. This is most likely due to the fact that emission sources are fixed to a specific point in time (1 April at 00:00 UTC for FS1) for FLEXPART simulations presented here. However, the shape of the BrO plume observed by GOME-2 most likely reflects the continuous change of emission sources associated with the passage of the front of the polar low pressure system.

FLEXPART simulations from FS2 and FS3 for the dissolving stage of the BCTE (together with the corresponding GOME-2 tropospheric BrO observation) are given in Fig. . Results for runs with emissions between 1 and 3 km and between 5 and 7 km look quite similar as results from the 3–5 km run and are therefore not shown in this figure. Like FS1, FS2 shows comma-shaped plumes for runs up to initialisation altitudes of 9–11 km. FS2 runs predict large values of tropospheric BrO vertical column density in the same area as the satellite observations, but also further northwards of the satellite observed plume. FS2 overestimates the magnitude of values reached inside the plume. In contrast to FS1 and FS2, FS3 results agree well with satellite observations at the dissolving stage of the BCTE. The simulated plume has largely lost its comma-shape for FS3. The best agreement between satellite retrievals and FS3 runs is achieved when assuming that the plume was located between 0 and 1 km altitude at time of initialisation (2 April 00:00 UTC). The fact that FS3 results compare much better with satellite data than FS2 and FS1 shows that emission sources around 150∘ W, 75∘ N contributed to the long observed lifetime (about 4 days) of the BrO plume. WRF simulations show that high wind speeds, convergent air flow and hence uplift occurred in this region (see above). This, together with SMOS and MODIS images suggests that recycling of bromine on salty blowing snow most likely caused the long observed lifetime of the BCTE. Note that as for FS1, differences between satellite retrieved tropospheric BrO VCDs and results from FS2 and FS3 are most likely due to the fact that the continuous change of emission sources associated with the passage of the front of the polar cyclone is not reflected by the FLEXPART simulations.

Overall, FLEXPART runs show that the plume was located in the troposphere with largest concentrations close to the surface, confirming that GOME-2 observed a tropospheric feature. Further investigation of FLEXPART runs shows that the plume resided between 0 and 3 km altitude during the whole simulation time. Provided that cloud top heights are representative of the upper limit of convection, FLEXPART simulations agree well with cloud top heights observed by CALIOP, further indicating that the BrO plume most likely occurred in the lowest three kilometres of the troposphere.

An intense BCTE which developed on 31 March 2011 over the Beaufort Sea has been investigated based on a combined use of satellite observations and numerical models. Despite the short atmospheric lifetime of BrO, the high wind speed BCTE was observed for about 4 days in GOME-2 satellite images. Comparison of GOME-2 satellite retrievals to FLEXPART and WRF model results reveals that the BrO plume moved eastwards with a polar low-pressure system. To our knowledge, this is the first study on a BEE that documents not only a link between plume occurrence and high boundary layer wind speeds, but also to frontal lifting near the cyclone centre as well as different development stages of the weather system. Our findings support who stated that most of the ozone destruction during BEEs occurs in fronts. The BCTE intensified on 1 April in the evening, when the low pressure system deepened and wind speeds increased. The BCTE reached its dissolving stage on 2 April in the evening, when the low pressure system weakened and the cyclone lost frontal structure. High values of tropospheric BrO VCDs in regions of low wind speed which occurred near the cloud head at the onset and mature stage of the event can be explained by production of gas phase bromine in high wind speed regions and subsequent anti-clockwise transport around the cyclone centre towards the cloud head.

The high wind speeds and vertical lifting associated with the front of the polar cyclone are consistent with weather conditions which cause production of blowing snow reported in other studies e.g.,. The MODIS false colour image from the mature stage of the BCTE shows mesoscale features resembling snow billows or cloud streets at the occluded front of the low in an area of high wind speeds and vertical uplift simulated by WRF. The latter coincides with reduced sea ice thicknesses retrieved from SMOS which means that any snow lying on sea ice in this area would have been more salty. Therefore, recycling of bromine on blowing snow is a reasonable explanation for the long observed lifetime of the plume and also for its development during the onset of the event. Moreover, the WRF simulations indicate that low temperatures in the area of the BrO plume in the lowest kilometre of the troposphere may have favoured the bromine activation cycle as proposed by . Our results are consistent with who observed the arrival of the BrO plume together with blowing snow at PEARL on 4 April with a MAX-DOAS instrument and Millimetre Cloud Radar data. Investigation of NCEP Reanalysis data (not shown) confirms that the BCTE initially developed when the low pressure system moved northwards towards the Beaufort Sea and reached potential source regions for salty blowing snow production observed by SMOS. Results presented in this paper document that weather conditions associated with fronts within polar cyclones are favourable not only for development of BEEs, but also to sustain high values of tropospheric BrO through continuous release of bromine over the course of the low pressure system, thereby extending plume lifetime substantially.

GOME-2 satellite observations of tropospheric BrO and total column ozone together with MODIS false colour images show that the plume was spatially separated from the dry conveyor belt associated with the polar cyclone. In this sense, this BEE differs from previous case studies for which the dry conveyor belt as a potentially bromine-rich stratospheric airstream complicated the interpretation of tropospheric BrO or total column BrO from satellite retrievals e.g.,. Moreover, FLEXPART simulations suggest that the BrO plume developed in a 1 km thick layer near the surface and was then transported up to 3 km altitude. This combination of model results and satellite observations shows that the BrO plume observed by GOME-2 most likely resided in the lowest parts of the troposphere over the entire lifetime of the BCTE. Our findings are consistent with who found that ozone depletion events which extend above 1 km in the vertical are usually associated with high wind speed conditions. Our results demonstrate that the close proximity of fronts and dry conveyor belts needs to be considered when deciding whether cyclone-like shaped plumes observed from satellite are of tropospheric or stratospheric origin, as both would be expected to show a comma- or spiral-shaped BrO pattern. This issue can be solved by using meteorological model data in combination with satellite observations.

The BrO plume occurred at the same location as low-level clouds observed by MODIS and CALIOP. As described in Sect. , light path enhancement and shielding of boundary layer BrO from the satellite sensors view cannot account for the plume pattern observed by GOME-2. Assuming that clouds are representative of vertical boundaries regarding convection, cloud and aerosol top heights observed by CALIOP agree well with FLEXPART results indicating that plume transport was limited to the lowest 3 km of the atmosphere.

In recent years, global climate models have been extended to successfully reproduce tropospheric BrO and BEEs observed by satellite e.g.,. However, our results suggest that a mesoscale model like WRF/Chem may be better suited for incorporating the bromine explosion chemical mechanism in a chemical transport model, as mesoscale snow billows produced by high wind speeds will not be resolved by global models.

As the present paper is based on one event only, more studies characterising links between polar cyclones and BCTEs on a climatological basis are required. Especially links to fronts should be investigated further. Moreover, possible links between strength and frequency of BCTEs and climate change need to be investigated in future research. Our results suggest that global warming potentially not only affects strength and frequency of BEEs by replacing perennial by younger sea ice , but also through effects on cyclone strength and frequency. According to , the frequency of extreme Arctic cyclones is expected to increase as a result of global warming. This, in addition to an increased area of younger sea ice, may lead to more frequent BCTEs in the future. This in turn impacts on the oxidative capacity of high-latitudes, the depletion of tropospheric O3 and deposition of mercury.

In summary, this manuscript has demonstrated the important role of frontal systems in generating tropospheric BrO in the lower troposphere. Further studies are required to quantify the relative importance of surface production of BrO and brine coated snow and ice lifted by frontal systems, future changes of BCTEs as well as their impact on tropospheric chemistry.