The NO x dependence of bromine chemistry in the Arctic atmospheric boundary layer

Introduction Conclusions References


Introduction
The episodic depletion of O 3 in the Arctic boundary layer following polar sunrise, referred to as ozone depletion events (ODEs), is attributed to a bromine gas phase reaction scheme, propagated by cycles such as Reactions (R1-R3; Simpson et al., 2007;McConnell et al., 1992).Ambient measurements at various Arctic sites have revealed maximum BrO mixing ratios between 30 to 40 pptv in the springtime (Liao et al., 2012;Pohler et al., 2010).However, modeling studies have shown that heterogeneous chemistry at the surface is needed to facilitate enhanced reactive halogen levels and drive ODEs (Toyota et al., 2014;Thomas et al., 2012;Michalowski et al., 2000), and recent field observations demonstrated that Br 2 is photochemically produced within the surface snowpack (Pratt et al., 2013;Foster et al., 2001).This heterogeneous chemistry mechanism, known as the "bromine explosion", is dependent on reactions involving HO x (Wennberg, 1999;Tang and McConnell, 1996;Vogt et al., 1996;Fan and Jacob, 1992) to produce hypohalous acids, which then oxidize halide ions at reactive surfaces (Huff and Abbatt, 2002;Abbatt, 1994).
drolyze on acidic surfaces to form HOBr, as shown in Reaction (R12) where X = Br (Hanson, 2003;Aguzzi and Rossi, 2002;Hanson et al., 1996).Thus, while reactions involving NO x can terminate the gas phase radical chain reaction, they can also generate products that contribute to the bromine explosion.Thus, it is not intuitively obvious what impact(s) NO x ultimately has on halogen chemistry.These halogen reaction pathways are summarized in Fig. 1 (Abbatt et al., 2012;Grannas et al., 2007;Simpson et al., 2007).As illustrated in Fig. 1, elevated levels of NO x can impact the halogen cycle through a variety of reactions.However, the sensitivity of the halogen radical chain reaction to NO x is currently not well understood.
The Arctic boundary layer typically has ambient background levels of NO x between 10 to 100 pptv resulting from its isolation from major anthropogenic sources, with its primary sources being photochemical production within the snowpack (Villena et al., 2011;Honrath et al., 2002;Ridley et al., 2000), and long range transport of photolyzable species such as organic nitrates (Muthuramu et al., 1994).Arctic field studies have led to observations of NO x fluxes from sun-lit snow surfaces (Grannas et al., 2007;Ridley and Orlando, 2003;Beine et al., 2002;Honrath et al., 1999Honrath et al., , 2002)), and lab studies have demonstrated that frozen solutions of nitrate and nitrite can release NO x when irradiated with UV light (Dubowski et al., 2002(Dubowski et al., , 2001;;Honrath et al., 2000).
With the possibility of increased anthropogenic sources throughout the Arctic, e.g. from coastal development, shipping, and oil and gas exploration, the impacts of increased NO x are likely to be seen (Peters et al., 2011;Corbett et al., 2010).However, the snow surface is an extremely complex matrix in which significant condensed phase photochemistry occurs, that we currently do not understand well and therefore can't properly simulate.Given the opportunity provided by the existing observational data base, we chose to investigate the effect of varying NO x mixing ratios on gas phase Arctic halogen radical chemistry, using a zero-dimensional (0-D) photochemical model.The model is constrained by recent observations of a wide variety of relevant precursors and intermediates, during the OASIS 2009 campaign conducted at Barrow, AK.To complement the model studies, observational evidence of the impact of NO x on BrO is examined from aircraft measurements near Prudhoe Bay, AK during the 2012 BROMEX.

Model description
A 0-D photochemical model was developed using the modeling software FACSIMILE.The model has been described in detail by Thompson et al. (2015) and is described briefly here.The model includes known Arctic gas-phase chemistry with 189 gas-phase reactions (Table S1 in the Supplement) and 28 photolysis reactions (Table S2).The deposition of 19 gas-phase species to aerosols/snow surfaces (Table S3) and 16 aqueous-phase chemical reactions (Table S4) are also included, where the heterogeneous reactions are treated as aqueous reactions.The model is constrained to observations with time varying mixing ratios for a list of gas-phase species (Table S5), including halogen radical precursors (Cl 2 and Br 2 ) and a wide range of volatile organic compounds (C 2 H 2 , C 2 H 4 , C 2 H 6 , C 3 H 8 , C 3 H 6 , n−C 4 H 10 , i-C 4 H 10 , HCHO, CH 3 CHO, CH 3 COCH 3 , and methyl ethyl ketone), as well as calculated, time varying photolysis rates, from the field study OASIS (Ocean -Atmosphere -Sea Ice -Snowpack) in Barrow, AK.Mixing ratios of selected gasphase species along with the photolysis rates were called into the model every 10 min within the simulation.For this model study, the 10-day period from 24 March-2 April 2009 during OASIS 2009 was simulated.During this period, a 3-day ozone depletion event (O 3 < 5 ppbv) occurred, followed by a full ozone recovery (O 3 > 20 ppbv) that was due to vertical mixing (discussed in Sect.3.1), and thus, covers a full range of typical atmospheric and meteorological conditions.It should be noted that the constrained mixing ratio for Br 2 on 30 and 31 March in the model is based on the observed diurnal average of 29 March and 1 April.As discussed in Liao et al. (2012) atmospheric observations for Br 2 on 30 and 31 March were not available.
To investigate the role that atmospheric NO x plays in Arctic halogen chemistry, two different NO x simulation scenarios were performed.A "low NO x " case and a "high NO x " case were used to create the two different scenarios.This allowed us to isolate a single variable between the two simulations.However, this approach does not consider other chemical species (i.e.VOCs (volatile organic compounds)) that could also be elevated along with anthropogenic NO x .The two diurnal-cycle NO x profiles were derived from the actual observed NO x over the time period being simulated, as shown in Fig. 2. Representative average polluted (high) and clean (low) NO x diurnal cycles, which differ by about a factor of 15, were calculated based on observed local NO x mixing ratio data for the period (Fig. 2).The clean (low NO x ) and polluted (high NO x ) days during the 10 day (24 March to 2 April) period studied were selected based on the work of Villena et al. (2011), in which correlations with ambient CO enhancements were used to identify air masses influenced by local emissions, i.e. for conditions when [CO] > 160 ppbv.The non-influenced/background days were averaged together to calculate a "low NO x " diurnal average that ranged between 50 to 100 pptv.These values were in the range of previous observations of background NO x mixing ratios (Villena et al., 2011;Honrath et al., 2002;Ridley et al., 2000).The same was done for the days influenced by local anthropogenic emissions, to create the "high NO x " diurnal average characterized by NO x mixing ratios from 700 to 1600 pptv.Each diurnal average was fit to a sine wave curve to generate temporally smoothed profiles.During the 10-day simulation either the Low NO x or High NO x diurnal average was applied to each day of the simulation, to generate the two distinct model scenarios.This allowed us to evaluate the NO x dependence of the chemistry, since it was the only parameter altered between the two scenarios.
Molecular halogens respond to changes in the deposition rates and condensed-phase chemistry, the latter of which cannot be well simulated given the current state of knowledge of physical and chemical processes occurring in the snowpack (Domine et al., 2013).Therefore, to ensure proper representation of the gas phase halogen chemistry, the atmospheric molecular halogen (Br 2 and Cl 2 ) mixing ratios were constrained to observations.The observations for these species reflect the impact of NO x -dependent production of reactive reservoir species, and the impact of their deposition to and chemistry within the snowpack.While the model was constrained to observations for stable species (Table S5), it was used to calculate various radical species' (e.g., BrO and Br) mixing ratios given the sources and sinks of these radicals.In this way, we calculate the effective NO x -dependence of the gas-phase radical chemistry and the rate of ozone depletion, as discussed below.

Bromine chain length
The chain length for a radical chain reaction is the rate of propagation divided by the rate of termination (or initiation).A chain length of < 1 means that most of the radicals terminate after production.Here we calculate the bromine radical chain length ( ) for the interconversion of Br and BrO radicals using Eq.(1; Thompson et al., 2015) for the 10-day simulation, where k corresponds to the reaction rate of the associated species.
Termination reactions include those that form non-radical brominated species (e.g., HBr, HOBr, BrONO 2 ), with photochemical lifetimes substantially longer than that of BrO or Br.It should be noted that although production of certain species represents a termination of the gas phase Br chain reaction (e.g.HOBr and BrONO 2 ), they can also play a crucial role in producing and increasing the Br radicals available for reacting with ozone, either through photolysis or heterogeneous reactions.This emphasizes the complexity of the BrO x cycle that takes place in the Arctic.In Eq. ( 1) the Br + NO 2 reaction is multiplied by the branching ratio (0.15) for the production of BrNO 2 (Orlando and Burkholder, 2000).Orlando and Burkholder (2000) observed that the dominant product for NO 2 reaction with a Br atom is BrONO, and while at lower temperatures isomerization to BrNO 2 is possible, the overall yield would still be minor.The production of BrONO is not considered a sink for BrO x because of its rapid thermal decomposition, photolysis and reaction with Br radicals, regenerating BrO x (Burkholder and Orlando, 2000;Orlando and Burkholder, 2000).The bromine chain length was only calculated during daylight hours (10:00 to 18:00 AKST) because the bromine radical chain is photochemically initiated via the photolysis of Br 2 (Reaction R1).The bromine radical chain length was calculated throughout the entire 10day simulation for both the low and high NO x simulations, as shown in Fig. 3.Both the low NO x and high NO x simulations show a chain length dependence on ozone mixing ratio (Fig. 3), with generally higher chain lengths at high O 3 , due to Reaction (R2).When the O 3 mixing ratio is > 5 ppbv, the average low NO x bromine chain length was 1.72 (±0.70), while the average high NO x bromine chain length was 1.81 (±0.35).The shorter simulated bromine chain lengths (< 1.0) on 26, 27, 28, and 31 March can be explained by the low O 3 mixing ratio (< 5 ppbv).One might hypothesize that the high NO x simulation would yield a shorter bromine chain length because NO x acts as a sink for BrO x .Indeed, the model simulation shows that NO x , on a percentage basis, is a more dominant sink for BrO x during the high NO x simulation compared to the low NO x simulation (Fig. 4).However, during the low NO x simulation the HO 2 +BrO reaction significantly decreases the bromine chain length, occurring more frequently, by a factor of 170, compared to the high NO x simulation.This is due to the suppression of HO 2 in the high NO x simulation via Reactions (R15) and (R16  Thompson et al. (2015), and indicated in Eq. (1), the observed chain length increases with BrO mixing ratio.

Net O 3 loss rate
Although the bromine chain length is relatively unchanged between the two scenarios, high levels of NO x decrease the rate of net O 3 loss within the model.The net O 3 loss rate was calculated from Eq. ( 2) as the sum of all of the rates of the chemical reactions that destroy ozone minus the sum of the rates of the chemical reactions that produce ozone (Thompson et al., 2015).It should be noted that the reactions in Eq. ( 2) that produce ozone (the 6th through 9th terms of the equation) are included as offsets for the depleted ozone destruction rate, which includes XO photolysis and NO oxidation to NO 2 .
(2) Reaction counters were utilized for all the HO 2 /RO 2 + NO reactions for both NO x scenarios to determine the importance of those reactions towards O 3 production.The two different NO x cases yielded equal numerical values when the counters were summed, indicating that NO x did not have a large impact on these small O 3 production terms.The calculated net O 3 loss rate for both NO x scenarios is shown in Fig. 5. On average, the net O 3 loss rate is a factor of 2 times slower for the high NO x simulation compared with the low NO x simulation.This can be explained by decreased BrO x during the high NO x simulation compared to the low NO x simulation in significant part due to the enhancement in the rate of Reaction (R10), as shown in Fig. 4. To study this effect further, we calculated the net O 3 destruction rate as a function of NO x mixing ratio by conducting simulations with a wide range of NO x mixing ratios.The various NO x diurnal cycles were calculated by starting with the low NO x diurnal profile and step-wise increasing the NO x by 250 pptv until the diurnal maximum of 2000 pptv was reached.The results, for the period 11:00-13:00 on 30 March are shown in Fig. 6. 30 March was selected because it fell into the "clean day" category during OASIS and during a time when ozone was decreasing from 30 to 5 ppbv.As shown in Fig. 6, the net O 3 loss rate decreases steeply, by a factor of 1.5 (from 0.80 to 0.52 ppbv hr −1 ), during the increase of NO x mixing ratios of ∼ 100-500 pptv, clearly expressing the strong NO xdependence of the chain reaction.We note, however, that this model experiment is directly testing the gas-phase component of this sensitivity on NO x , and not the NO x dependence of BrONO 2 deposition.However, as discussed later, when NO x increases, BrO decreases and thus BrONO 2 is not sensitive to the NO x mixing ratio.This does not negate the potential importance of BrONO 2 during ODEs.Cao et al. (2014) found, from a modeling study, that BrONO 2 production increased the rate of ozone depletion through the production of HOBr from its hydrolysis (Reaction R12) on snow/aerosol surfaces.HOBr is a main component of the bromine explosion and leads to an increased production rate of Br 2 (Reactions R5-R6).However, regression of the observed [Br 2 ] vs. observed [NO x ] reveals that the highest mixing ratios of Br 2 do in fact occur when NO x is below 300 pptv (Fig. S1 in the Supplement).This is likely due to the fact that HO 2 is high when NO x is low, contributing to greater HOBr at low NO x (see Fig. 7b for HOBr on this day).This observed Br 2 clearly supports that the O 3 loss rate is minimized when NO x mixing ratios are elevated, as the maximum Br atom production rates will occur at low [NO x ].The net ozone loss is, of course, also extremely low for days when the observed mixing ratios of O 3 were very small (< 5 ppbv), as the rate of ozone destruction will approach zero as O 3 is nearly completely removed.Based on the results from Figs. 5 and 6, the influence that elevated NO x mixing ratios have on decreasing the net ozone loss rate could be a factor in the Arctic as NO x point sources continue to increase in remote Arctic locations.
The O 3 loss rate for 30 March from 11:00 to 13:00, based on the observations, was approximately 3.4 ppbv hr −1 .This is much larger than the calculated net O 3 loss rate for the low NO x simulation of 0.80 ppbv hr −1 .The low net O 3 loss rate for the low NO x simulation could be a result of the constrained model Br 2 mixing ratio not being representative of the ambient air in Barrow at that time, as mentioned previously.If we constrained Br 2 to an estimated concentration lower than what was actually present during that time, we would expect a decreased depletion rate.This belief is supported by a comparison of the BrO data for this time period (discussed later), which is significantly under-simulated by the model.However, this apparent dramatic ozone loss rate could be the observation of an ozone-depleted air mass being transported to the measurement site (Halfacre et al., 2013).

Model-simulated species vs. OASIS 2009 observations
To further understand both how NO x decreases the rate of net O 3 loss and affects the bromine chain length, several species that play a crucial role in the bromine cycle were examined.Molecular bromine and its precursors are produced from Reactions (R6) and (R12).The prominent gas-phase recycling reaction is BrO self-reaction (Reaction R3) while the forma-  tion and subsequent deposition of HOBr and BrONO 2 leads to heterogeneous reactions that can form Br 2 (Reactions R5-R12).Here we simulated the effect of the low and high NO x scenarios on BrO, HOBr and BrONO 2 , and compared the results with the observed mixing ratios during OASIS (Fig. 7ac).For the majority of the days, simulated BrO for the low NO x case is close to that observed during OASIS, as expected since low NO x conditions were typically observed.However, this is not the case for 30 and 31 March, for which the simulated BrO is lower than what was observed during OASIS.This is likely a result of the fact that atmospheric observations for Br 2 on 30 and 31 March were not available, as discussed earlier.For those days the constrained Br 2 was a diurnal average of 29 March and 1 April Br 2 observations.The high NO x simulation results in a highly suppressed BrO mixing ratio (Fig. 7a), compared to the low NO x simulation, by more than an order of magnitude throughout the time period.For the high NO x case, BrO never exceeds 3 pptv, whereas for the low NO x case, BrO ranges from 2.5 to 25 pptv.For high NO x days, the BrO mixing ratios are low because of Reactions (R8) and (R10), in which BrO and Br radicals are scavenged by NO x and converted to temporary termination products.
HOBr responds in similar fashion to changes in NO x as does BrO, with the low NO x simulated HOBr being within ±5 pptv of the OASIS observations while the high NO x simulated HOBr is much lower.For the most part the low NO x simulation HOBr is slightly elevated compared to the observed HOBr (26 and 29 March) because the low NO x simulation has a greater HO 2 mixing ratio compared to the observations.
Figure 7b shows that the high NO x condition completely suppresses HOBr, and that is what is observed for the high NO x mixing ratios day, 24 March.However, in contrast to the case for BrO and HOBr, BrONO 2 is not suppressed by high NO x mixing ratios (Fig. 7c), since while increased NO x suppresses BrO, the rate of R8 is compensated by the increase in NO x mixing ratios, therefore R8 is largely unchanged.This is shown in Fig. 7C as the BrONO 2 mixing ratio is similar for both the high and low NO x simulations.Additionally, when the model simulated BrONO 2 mixing ratio for both NO x simulations is plotted against the BrONO 2 production rate (k BrO+NO 2 [BrO][NO 2 ]), it affirms that the BrONO 2 mixing ratios follow k BrO+NO 2 [BrO][NO 2 ] (Fig. S2).

BrO x and NO x sinks
It is clear from the discussions above that NO x influences BrO x partitioning.The sinks of BrO x and NO x were quantified to evaluate their NO x dependence by including reaction counters on the relevant reactions in the model that convert BrO x and NO x to reservoir species.Over the 10-day simulation period for both low and high NO x cases, NO x is a significant sink for BrO x radicals (> 27 %), although for the high NO x case it contributes more than 50 % (Fig. 4).However, both products of R8 and R10 result in species that can regenerate Br 2 .As expected, CH 3 CHO plays a major role as a BrO x sink as well (Shepson et al., 1996), contributing to more than > 35 % in both simulations.Though anthropogenic emissions are known sources of acetaldehyde, observed CH 3 CHO mixing ratios were unaffected by Barrow emissions throughout the 10-day simulation period.Thus it is likely that the snowpack is the dominant CH 3 CHO source in this study (Grannas et al., 2002).For NO x , while reactions with HO 2 are important (∼ 20 %), reactions of NO 2 with Br and BrO still represent significant NO x sinks (∼ 30 %).Peroxynitrates result from an important sink pathway of NO 2 , e.g.via Reactions (R17) and (R18), below.Reaction (R17) will be the subject of a separate study.

BROMEX 2012 field observations
To further examine the interactions between NO x and reactive bromine, NO 2 and BrO were measured within and around a large combustion plume in the coastal Arctic atmosphere.During the spring 2012 Bromine, Ozone, and Mercury Experiment (BROMEX) field campaign in Barrow, Alaska, airborne measurements of BrO and NO 2 column density (from ∼ 700 m to the surface) were conducted using an aircraft-mounted MAX-DOAS (General et al., 2014) in nadir view.The results for the derived BrO and NO 2 concentrations (from the vertical column density data, assuming a 600 m layer thickness), the average effective mixing ratio from the aircraft to the surface, for a section of a flight on 30 March near Prudhoe Bay, AK, are shown in Fig. 8. Prudhoe Bay is the largest oil field on the North Slope, located ∼ 330 km southeast of Barrow, AK, and produces significant CO 2 , NO x , and CH 4 (Brooks et al., 1997;Jaffe et al., 1995).Anthropogenic emission plumes can easily be observed from the flight by increased NO 2 concentration.This is a good example of the type of point source that is increasing in prevalence in the Arctic (Roiger et al., 2014;Harsem et al., 2011).It can be seen from Fig. 8 that an increase in the concentration of atmospheric NO 2 , corresponding to a plume near Prudhoe Bay, coincided with a decrease in the concentration of atmospheric BrO.The BrO mixing ratio suppression observed during the airborne BROMEX measurement is compared to our model simulations of the OASIS 2009 field observations, in the Fig. 8 insert.To examine the sensitivity in the model and observation of BrO to NO 2 , the Br 2 concentration was held constant at 5 pptv throughout the simulation, while NO 2 was varied, for mid-day conditions.The Br 2 concentration was fixed based on the average observed midday (12:00-14:00) concentration during the 10-day simulation time period.NO and NO 2 concentrations were held constant during the entire simulation at a constant ratio of 0.60 for NO 2 : NO, but the absolute magnitude of each concentration was varied between simulations.The data points for both the model and OASIS ground based observations represent data from the time period of 12:00-14:00 AKST.
The model captures the observed distinct sensitivity of BrO to NO 2 , decreasing rapidly with increasing NO 2 in the 0-500 pptv NO 2 range.For the BROMEX aircraft data at 600m altitude, while BrO decreases with NO 2 , there is more apparent BrO aloft for a given [NO 2 ].The relative suppression of BrO at the surface implies a greater BrO x surface sink, compared to that at 600m.This can be explained from the observations of the BrO x surface layer sinks expressed in Fig. 4, where a significant sink for BrO x is Br atom reaction with aldehydes.Given the known snowpack emission source of aldehydes (Grannas et al., 2002;Sumner and Shepson, 1999) and the surface-based character of chlorine chemistry that also produced aldehydes (Tackett et al., 2007), we expect the BrO x sinks to be enhanced at the surface.The BrO depletion observed during BROMEX is mainly due to Reactions (R8) and (R10) and supports our findings that BrO is suppressed by elevated levels of NO x (Fig. 7a).Thus, while BrONO 2 and BrNO 2 can be recycled on surfaces to re-emit Br 2 , it seems clear that the net effect of high NO x mixing ratios is to slow down the overall halogen chain chemistry, as demonstrated in Figures 8 and S1.It should be noted that in very large NO x plumes O 3 can be removed by the reaction of NO with O 3 , as shown in Reaction (R19), although in day light, a photo-steady state will develop from NO 2 photolysis.
If no ozone is present the BrO x can repartition from BrO towards Br, via photolysis of BrO.During the 30 March 2012 flight the air mass depleted of BrO also had lower ambient O 3 mixing ratios which could explain the BrO depletion.
In Prudhoe Bay, the sources of NO x are buoyant plumes, located aloft of the surface (smoke stacks), where natural aerosol extinction would be less compared to the surface (Breider et al., 2014).This would limit the availability of surfaces on which BrONO 2 could react to re-emit Br 2 .However, these NO x sources also produce aerosols (data not shown; Peters et al., 2011); if these sources continue to increase in number, a shift in the aerosol extinction could occur, providing increased available aerosol surfaces for reaction of BrONO 2 .

Atmospheric implications
It is clear that elevated levels of atmospheric NO x have a significant effect on the atmospheric chemistry that occurs in the Arctic relating to halogen species.With the possibility of more anthropogenic sources appearing in the Arctic as sea ice area continues to decrease, due to gas and oil exploration and increased shipping traffic, a shift in the atmospheric oxidation capacity and chemical pathways could occur.The main springtime atmospheric oxidizers could change from halogens in the surface layer to a greater contribution from OH via O 3 photolysis, via R20, along with a potential decrease in the frequency of ODEs.Further, open water pro-duces convective mixing, bringing O 3 from aloft down to the surface (Moore et al., 2014).However, the climate change impacts on the Arctic atmosphere and associated chemistry are complex.As multiyear ice retreats, the fraction of first year ice is increasing, leading to more saline ice surfaces.However, the findings of Pratt et al. (2013) indicate that the surface snowpack on the sea ice needs to be acidified for halogen activation to occur, and this likely depends on the depth of the snowpack, which is impacted by snowfall rates and timing, as well as the presence of sea ice.It is now known that snowpack depths in the Beaufort and Chukchi seas have been decreasing at a significant rate (Webster et al., 2014).The acidification of the surface snow also depends on long range transport of acidic species, such as SO 2 and HNO 3 (Worthy et al., 1994), which are likely to change.Moreover, increasing latent heat fluxes (Serreze and Barry, 2011) resulting from more open water will increase the OH production rate in the Arctic because of the humidity dependence of ozone photolysis as shown in Reactions (R20-R22), although a significant increase in humidity would be needed for Reactions (R20-R22) to have an impact on Arctic OH production.
Thus, there are multiple competing variables that influence Arctic atmospheric chemistry, and it is surely the case that we will need to continue to integrate simulations with observations to understand the coupling of physical and chemical processes, as the Arctic continues to warm and undergo change at the surface.Finally, we note that the deposition rates and aqueous phase chemistry for the products from the reaction between atmospheric NO x and halogen radicals are not well known, and fluxes of molecular halogens from Arctic surface snow have not been measured to date.Thus further efforts in this area are necessary to gain a greater understanding of these chemical processes to improve models.
The Supplement related to this article is available online at doi:10.5194/acp-15-10799-2015-supplement.

Figure 2 .
Figure 2. Observed (blue) NO x mixing ratios for the 10-day OA-SIS period in Barrow, AK, as well as model scenario diurnal NO x mixing ratios for low (black) and high (red) cases.

Figure 3 .
Figure 3. Calculated bromine chain length for the low NO x simulation and the high NO x simulation along with observed O 3 and radiation.

Figure 4 .
Figure 4. Fractional contributions of BrO x and NO x sink reactions from the low and high NO x simulation cases (Note that Br+NO 2 only represents BrNO 2 production).

Figure 5 .
Figure 5. Calculated net O 3 loss rate for the low NO x and high NO x simulations, along with the observed O 3 mixing ratios.

Figure 6 .
Figure 6.Net O 3 loss rate as a function of the NO x mixing ratio, for 30 March mid-day (11:00 to 13:00 AKST) conditions.

Figure 7 .
Figure 7. (a) Simulated BrO mixing ratio (low NO x & high NO x cases) and the observations during the study period.(b) HOBr levels from the model simulations (low NO x & high NO x cases) and the observations during the simulation dates.(c) BrONO 2 mixing ratio from the two simulation cases.

Figure 8 .
Figure 8. BrO and NO 2 measured mixing ratios via MAX-DOAS during the BROMEX field campaign, near Prudhoe Bay (70 • N, 149 • W), AK at 700 m above the surface on 30 March 2012.The insert of the NO 2 versus BrO shows the anti-correlation between the two species. ).