Measurement-based modelling of bromine-induced oxidation of mercury above the Dead Sea

Atmospheric mercury depletion events (AMDEs) outside the polar region – driven by high levels of gaseous Br and BrO (i.e., BrOx) – were observed recently in the warm Dead Sea boundary layer. The efficient oxidation of gaseous elemental mercury (GEM) under temperate conditions by BrOx was unexpected considering that the thermal back dissociation reaction of HgBr is about 2.5 orders of magnitude higher under Dead Sea temperatures compared to polar temperatures, and hence was expected to significantly slow down GEM oxidation under warm temperatures. The goal of this modelling study was to improve understanding of the interaction of reactive bromine and mercury during Dead Sea AMDEs using numerical simulations based on a comprehensive measurement campaign in summer 2009. Our analysis is focused on daytime AMDE when chemical processes dominate concentration changes. Best agreements between simulations and observations were achieved using rate constants for kHg+Br and kHg+BrO of 2.7 × 10−13 cm3 molecule−1 s−1 and 1.5 × 10−13 cm3 molecule−1 s−1, respectively. Our model also predicted that a rate constant kHg+BrO of 5.0 × 10−14 cm3 molecule−1 s−1 may be considered as a minimum, which is higher than most reported values. These rate constants suggest that BrO could be a more efficient oxidant than Br in the troposphere as long as [Br]/[BrO] ratios are smaller than∼0.2 to 0.5. Under Dead Sea conditions, these kinetics demonstrate a high efficiency and central role of BrOx for AMDEs, with relative contributions to GEM depletion of more than∼90 %. Unexpectedly, BrO was found to be the dominant oxidant with relative contributions above 80 %. The strong contribution of BrO could explain why the efficiency of GEM oxidation at the Dead Sea does not critically depend on Br and, therefore, is comparable to that in cold polar regions. In order to confirm the suggested kinetics, additional studies, particularly for temperature-dependence of rate constants, are required.


Introduction
Atmospheric mercury is a persistent and toxic global pollutant, and the major contributions of mercury loadings in remote locations are due to atmospheric deposition (Lu et al., 2001;Driscoll et al., 2007).Most atmospheric mercury occurs as gaseous elemental mercury (GEM; >95 %) -a form which is relatively inert, has a low deposition velocity, and shows global lifetime of ∼1 yr (Lindberg et al., 2007).Field measurements in polar regions showed that GEM can be oxidized rapidly to reactive forms (reactive gaseous mercury: RGM and particulate Hg: Hg P ) after polar sunrise, followed by rapid deposition to the surface.Schroeder et al. (1998) were the first to show that during such events, termed Atmospheric Mercury Depletion Events (AMDEs), nearly complete depletion of GEM can occur within 24 h or less.There is a growing body of observational evidence indicating that rapid AMDEs, and subsequent scavenging by aerosols and deposition to the surface, are widespread in polar regions and the marine sub-Arctic and Antarctic (Lindberg et al., 2002;Ebinghaus et al., 2002;Berg et al., 2003;Skov et al., 2004;Poissant and Pilote, 2003;Steffen et al., 2008;Kim et al., 2010;Mao et al., 2010).The oxidation of GEM during AMDEs is considered to be caused by Br and BrO (i.e., BrO x = BrO + Br) as well as other reactive halogen species (RHS) (Lu et al., 2001;Lindberg et al., 2001;Ariya et al., 2004;Brooks et al., 2006).As a result, AMDEs generally correlate with ozone depletion events (ODEs) (Bottenheim et al., 1986).It is well known that reactive halogens also can exist outside of polar regions at temperate, mid-latitude locations such as over salt lakes and in the marine boundary layer (MBL) (Hebestreit et al., 1999;Saiz-lopez et al., 2004).Due to the important role that BrO x plays in GEM oxidation, several studies, including modelling, focused on interactions of BrO x with atmospheric mercury in the polar region (Calvert and Lindberg, 2003;Skov et al., 2004;Hedgecock et al., 2008;Xie et al., 2008;Durnford et al., 2011;Subir et al., 2011;Stephens et al., 2012), in the MBL at mid-latitudes (Hedgecock andPirrone, 2001, 2004;Hedgecock et al., 2003Hedgecock et al., , 2005Hedgecock et al., , 2008;;Holmes et al., 2006Holmes et al., , 2009)), and in urban areas (Shon et al., 2005).Holmes et al. (2006) calculated a global chemical lifetime of atmospheric mercury of 0.5-1.7 yr against oxidation by Br, implying that Br is a significant oxidant of GEM on a global scale, and Holmes et al. (2009) found that oxidation of GEM by Br accounts for 35-60 % of the MBL RGM source.
There are still significant uncertainties in regard to the reaction of BrO x with Hg (see Ariya et al., 2008;Calvert and Lindberg, 2004;and Xie et al., 2008).Reaction rates of Br and BrO with GEM are not well defined; it is not clear which is the dominant oxidant; and chemical pathways and products of RGM formation are unclear.Additional uncertainties exist regarding the role of other halogens such as I and IO (Raofie et al., 2008), ClO x (Stephens et al., 2012), O 3 and OH (Xie et al., 2008 and references within), and temperaturedependence of rate constants (Ariya et al. 2008;Xie et al., 2008).
Recently, it was observed that strong AMDEs are not limited to the polar region but also occur in the mid-latitudes under temperate conditions: near-complete (up to ∼90 %) conversion of GEM to RGM was observed in the Dead Sea atmosphere under temperatures of 40 • C and over (Peleg et al., 2007;Obrist et al., 2011).These pronounced Dead Sea AMDEs, which strongly correlated with high BrO x , were unexpected given that the dissociation of HgBr (formed by GEM oxidation by Br), a rate limiting step in GEM oxidation by Br, is considered strongly temperature dependent (Goodsite, 2004).
The objective of our study was to improve understanding of the BrO x -Hg interaction at the Dead Sea by combining results from a comprehensive measurement campaign with detailed modelling analysis.Due to the elevated [Br − ] and [Br − ]/[Cl − ] ratio of the seawater (Niemi et al., 1997), low pH (<6; Tas et al., 2005), and moderate anthropogenic pollution (Tas et al., 2005), extremely high levels of BrO x are formed almost daily at the Dead Sea (e.g., Tas et al., 2006).Due to regularly high BrO x levels, the Dead Sea provides excellent conditions for studying the interaction of BrO x with atmospheric mercury at mid-latitudes.We used the MECCA box model (Sander et al., 2005) heterogeneous mechanism, focusing on identifying reaction rates of BrO and Br with GEM, the relative importance of BrO x and other oxidants, and oxidation sensitivity to temperature.

Field measurements
Model simulations are based on a comprehensive measurement campaign that took place at Ein Bokek, Israel (latitude 31.20 • N, longitude 35.37 • E) on the shore of the Dead Sea, between 29 June and 28 July 2009.Measurements are described in detail in Obrist et al. (2011), and additional details on methodology are found in Matveev et al. (2001) and Tas et al. (2005).In summary, measurements included speciated atmospheric mercury (GEM, RGM, and Hg P ) using Tekran Model 2537, Model 1130, and Model 1135 analyzers (Tekran Inc.; Toronto, Canada), O 3 (TEII, Model 49C), NO-NO x (TEII, Model 42i), SO 2 (TEII, Model 43C), CO (TEII, Model 48i), and particulate sulfates (TEII, Model 5020i, all Thermo Environment Instrument Incorporated, Waltham, MA, USA).Meteorological conditions including wind speed and direction, temperature, relative humidity (RH), barometric pressure, and solar radiation were measured using meteorological instruments (Met-One Instruments Inc. Grants Pass, Oregon, USA).Long-path differential optical absorption spectroscopy (LP-DOAS) technique (Model HMT DOAS Measuring System; Hoffmann Messtechnik, Rauensberg, Germany) was used to quantify BrO (detection limit as low as 3 pmol mol −1 ), O 3 (detection limit generally <7.3 nmol mol −1 ), and NO 2 (detection limit generally <0.5 nmol mol −1 ).The LP-DOAS system was placed in the top floor of an air-conditioned hotel room (18 m above water level and the same location as above instruments), and the DOAS reflector mirror was situated at the eastern side of an evaporation pond directly above the water surface, resulting in an 11.8 km light path traveling over the Dead Sea water.Details of the DOAS system and signal retrieval are described in Stutz and Platt (1996) and Peleg et al. (2007).

Model description
The model used in this analysis is the comprehensive heterogeneous MECCA box model (Sander et al., 2005).MECCA includes an explicit kinetic heterogeneous chemical mechanism, accounting for gas and aqueous phase reactions and heterogeneous reactions for two aerosol modes.In the gas phase, species are subjected to photochemical decomposition and dry deposition; aerosol processes include both scavenging and new particle formation.Gas-aerosol partitioning is performed based on Henry's law and kinetic limitations for coarse soluble and accumulation soluble aerosol modes.The O-H-C-N-S-Cl-Br-I chemical mechanism includes 186 gas  S1), aqueous (Supplement Table S2), aqueous equilibrium (Supplement Table S3), and heterogeneous reactions (Supplement Table S4) involving sea salt and sulphate aerosols.Photolysis rate coefficients are calculated using the method described in Landgraf and Crutzen (1998) and were updated at each time step (1 min) representing real time and location of the measurement site.
The initial sea salt aerosol composition was based on the Dead Sea water composition accounting for Br − and Cl − concentrations, as reported by Tas et al. (2005).In MECCA, the averaged mass transfer coefficient for each aerosol mode is based on integration over a lognormal shape distribution of the particle radius for each aerosol mode (Kerkweg et al., 2007).Dependence of average sea salt aerosol number concentrations on in situ measured wind speeds was taken into account using Eq. 4 in Gong et al. (1997) on five-minute time resolution.Average sulphate aerosol number concentrations were determined based on in situ measurements (Sect.2.1) using reported densities (Karg et al., 1995) and radius dependency on RH (Yue, 1979) and were updated in the model on time resolution of 5 min.

Model simulations
Initial model runs showed that reactive bromine production via aerosols is limited to RH higher than 28 %.Therefore, aerosol chemistry in our simulations was activated only for RH above 28 %.Smoydzin and von Glasow (2009) found that due to low pH, direct release from the seawater is likely an additional source for BrO x at the Dead Sea.A BrO x source representing direct Br 2 release from the seawater via the bromine explosion mechanism (Vogt et al., 1996) was parameterized in the model with a first order rate constant of 1.0 × 10 −3 s −1 , as described in Obrist et al. (2011).
Simulation days showed dominant daytime local wind direction from the east and south (i.e., directly from the Dead Sea) so that no significant sources from land were expected during mid-day AMDEs where we focused our investigation; however, during nighttime, morning, and evening advection and possibly other sources occur as explained below.Simulations were updated on a 5-min.time resolution using the following real-time measured parameters: temperature, RH, wind speeds, and mixing ratios for NO, NO 2 , SO 2 , and 13 different hydrocarbon species (based on HC measurements; see Tas et al., 2006).Table 1 presents typical measured values of chemical and meteorological parameters used during our simulations.Boundary layer height was determined using a 1-D meteorological model (McNider and Pielke, 1981) and was continuously updated.Photolysis rate coefficients were calculated by the MECCA model, representing clear sky conditions for the Dead Sea.Using the above in the model enabled to obtain relatively good agreements with measurements for O 3 and BrO diurnal profiles (Sect.3.1).
Our analysis is focused on daytime-AMDEs, during which GEM concentrations are mainly controlled by chemical processes, rather than other source or sink processes (such as advection or emission) which could not be well described by the used box model.Our measurements indicate that once AMDE/ODEs are terminated, both GEM and O 3 are quickly advected back as indicated by sharp decrease in BrO and sharp rebounds of O 3 and GEM to background levels.To account for this, O 3 and GEM fluxes were added for 30 min at the end of AMDE/ODEs to simulate these rebounds.For Julian day (JD) 197, GEM fluxes were added instead between 07:30-08:00, and during JD = 188 fluxes were also added between 07:30 and 13:30 (see Sect. 3.2).It is important to note that no fluxes were added during the simulation of AMDEs which is the main focus of our study.Hence, our model simulations focused on the chemical and kinetic processes during AMDEs, as opposed to processes such as advection, surface emissions, or pollution inputs -all of which affect Hg and O 3 concentrations during non-AMDEs periods.
Additional detailed simulations were performed for Julian days 188 and 201 (Table 2).Simulation "NOBROX" is identical to the "BASE" simulation, except that all direct reactions of Hg with BrO x were switched off.Simulations "ONLYBROX" and "ONLYBR" are identical to the "BASE" simulation except that they include only BrO x and Br as GEM oxidants, respectively.All above simulations used real-time in situ measured summer Dead Sea temperatures (which averaged ∼310 K during the summer).In addition, "WINTER" and "POLAR" simulations are identical to the "BASE" simulation, except that typical winter Dead Sea temperatures (294 K) and polar temperatures (240 K) were used instead of real-time summer temperatures.

Model verification
The primary, non-halogen, gas-phase oxidants of GEM considered in the urban and lower troposphere boundary layer are O 3 , H 2 O 2 , and OH (Hedgecock et al., 2005;Holmes et al., 2009;Sillman et al., 2007;Spivakovsky et al., 2000).In the marine boundary layer and upper troposphere, halogens are presumed to be the dominant oxidants (Lin et al., 2006).Atomic bromine is a major global sink for GEM, especially in the middle and upper troposphere where relatively cold temperatures suppress thermal back-decomposition of the HgBr intermediate (Holmes et al., 2006;Lin et al., 2006).During AMDEs in the polar regions, RHS, particularly BrO x , dominate the rate of GEM depletion (e.g., Xie et al., 2008).At the Dead Sea, AMDEs are strongly linked to BrO x levels as well (Obrist et al., 2011), primarily due to the very high concentrations of reactive bromine species compared to reactive chlorine species which are not expected to play a significant role at the Dead Sea (Tas et al., 2006).In addition, iodine concentrations are generally low when bromine species are active (Zingler and Platt, 2005).Still, our model was configured to include a detailed account of multiphase chemical reactions involving all halogens (reactive bromine, chlorine, and iodine; based on the actual Dead Sea water composition Sect.2.2) plus non-halogen oxidants.
Figure 1 shows good agreements between simulated and measured O 3 concentrations during AMDEs for JD = 188 and 201, the two main days used for analysis.Model evaluation also showed good agreement of BrO concentrations (Figs. 2, 3 and Supplement Fig. S1), particularly during the occurrence of mid-day AMDEs.We focused further detailed analyses on comparing results for JDs 188 and 201 as they are characterized by inherently different BrO x levels and associated [Br]/[BrO] ratios, with maximum BrO levels of ∼90 pmol mol −1 (JD 188) -as frequently observed over the Dead Sea -and levels of ∼30 pmol mol −1 (JD 201), as typically observed in the polar boundary layer.

Reaction rates of GEM with Br and BrO
We performed sensitivity analyses to calibrate the model with respect to reaction rates of Br and BrO with GEM (Figs. 2-4 and Supplement Fig. S1).Note that for JD = 188, GEM depletions occurred too early and were too intense for most combinations of rate constants (Fig. 2b), except in one case (L Br L BrO) when GEM depletion could not be simulated to its full extent.We chose to constrain GEM concentrations between 07:30 and 13:30 LT (i.e., prior to the onset of the AMDE; Fig. 2c); the reason for the elevated measured GEM levels is unknown but is likely due to entrainment of GEM from other areas or from other sources that are not accounted in our chemical box model.Figure 4 shows scatter plots of measured GEM concentrations (x-axis) with correspondingly simulated concentrations (y-axis) during mid-day AMDEs for all combinations of reaction rates for the three days of simulations, and Table 3 shows corresponding summary statistics.GEM depletions were best reproduced when employing "BASE" ("LBr HBrO") and L Br M BrO simulations.Values of k Hg+BrO smaller than used in these two cases yielded fair agreement only for the day with high BrO levels and a high [Br]/[BrO] ratio (i.e., JD 188), and strongly underestimated GEM depletion for the days with lower BrO levels (i.e., JD 197 and 201).On these days, using k Hg+BrO values smaller than M BrO greatly underestimates GEM depletions.This suggests that k Hg+BrO should be at least as high as in L Br M BrO (i.e., 5 × 10 −14 cm 3 molecule −1 s −1 ) and could be as high as in "BASE" simulation (i.e., 1.5 × 10 −13 cm 3 molecule −1 s −1 , see below).
We attribute overestimation of GEM depletion on JD 188 to under-estimation of simulated O 3 during that AMDE (between ∼13:30 and 14:45 LT; Fig. 1) which would overpredict [Br]/[BrO] ratios.Based on our analysis (see Fig. 5), [Br]/[BrO] ratios on that day could have been overestimated by up to a factor of ∼2, which would overestimate ratios -suggest that the rate constants used by "BASE" fit observations quite well.We suggest that a value of 2.7 × 10 −13 cm 3 molecule −1 s −1 for k Hg+Br and a lower and upper limit of 5 × 10 −14 cm 3 molecule −1 s −1 and 1.5 × 10 −13 cm 3 molecule −1 s −1 for k GEM+BrO seem reasonable rate coefficients to reproduce observed AMDEs at the Dead Sea.It should be emphasized that improved knowledge of kintetics, particularly temperature-dependence of rate constants (Ariya et al., 2008;Xie et al., 2008) are needed to confirm the above suggested rate constants.
Previous studies (Goodsite et al., 2004;Ariya et al., 2002;Calvert and Lindberg, 2004;Peleg et al., 2007) indicated that GEM oxidation by Br predominates over oxidation by BrO, since theoretical kinetic calculations suggest that the reaction of BrO with Hg is endoergic and, therefore, unlikely to occur in the atmosphere (Tossell et al., 2003;Holmes et al., 2006;Shepler et al., 2007).Experimental studies, however, found conflicting data on the energetics of the species existing in the vapor over heated HgO (s) (Tossell et al., 2006), suggesting that oxidation of GEM by BrO may be significant.Stephens et al. (2012) found that while Br had the greatest overall contributions to GEM depletion in the Arctic, BrO on some occasions nearly matched the importance of Br.Overall, a wide range of reaction rates have been published during the last decade, suggesting that GEM oxidation by Br ranges between a high value   of 1 × 10 −13 − 3 × 10 −12 cm 3 molecule −1 s −1 (see Ariya et al., 2008 and references within) and lower oxidation rates by BrO of 10 −13 -10 −15 cm 3 molecule −1 s −1 (Raofie and Ariya, 2003;Spicer et al., 2002).Recently, Xie et al. (2008) showed that model simulations could best reproduce observed correlations between GEM and O 3 for rate constants of GEM oxidation by Br and BrO on the magnitude of 3 × 10 −13 and ≤ 1.0 × 10 −15 cm 3 molecule −1 s −1 , respectively, which are different than values proposed in our study (see above).One reason for the discrepancy between theirs and our rate constants may be that they used correlations between GEM and O 3 to assess GEM and O 3 depletion rates by BrO x .Correlations between O 3 and GEM, however, are subject to absolute O 3 , GEM, and BrO x levels (e.g., Fig. 5).In our study, GEM depletion rates were not based on correlations to O 3 but on ratios of [Br]/[BrO] to assess relative influences of BrO and Br on GEM depletion rates.Figure 5a shows relationships between measured GEM, BrO, and O 3 .We graphed GEM depletion per unit change in BrO (i.e., GEM/ BrO) as a function of respective O 3 concentrations, and only daytime values and data when BrO > 0 were used.Clearly, GEM/ BrO increases with higher O 3 levels.Ratios of [Br]/[BrO], however, are expected to anti-correlate with O 3 (e.g., Wayne et al., 1995;Tas et al., 2008); therefore we expect that oxidation of GEM by BrO will increase at the expense of oxidation by Br for high O 3 levels (Xie et al., 2008).Figure 5a implies that GEM oxidation by BrO predominates over oxidation by Br, in agreement with model results in Fig. 6.We rule out that these patterns are caused by correspondingly different GEM concentrations (blue bars in Fig. 5a) as these did not well correlate with GEM/ BrO.days 188, 201, and 197.In these graphs, GEM/ BrO is graphed as a function of [Br]/[BrO] (which is based on, and hence analogous to, O 3 levels in panel a).Our results indicate that modeled GEM/ BrO versus [Br]/[BrO] is similar to observed patterns only for the "BASE" simulation (Fig. 5b).Even for this case, however, GEM/ BrO ratios tend to peak at somewhat lower O 3 , which may suggest that the ratio of rate constants for BrO and Br (e.g., k GEM+BrO /k GEM+Br ) still may be underestimated for the "BASE" simulation.For simulations "H Br L BrO" and "L Br L BrO", GEM/ BrO was greatly overestimated for O 3 levels below 10 nmol mol −1 , which is expected if k GEM+BrO /k GEM+Br is underestimated.This result further suggests that the rate constant for L BrO is too low to accurately represent observed results.Possibly, a decrease in efficiency of GEM depletion by BrO at O 3 levels above ∼30-40 nmol mol −1 as observed in the "BASE" simulation may be due to the importance of O 3 for GEM oxidation at high O 3 levels.

Relative contributions of oxidants to GEM depletions
Figure 6 shows relative contributions of various oxidants to GEM depletion rates and associated GEM lifetimes for Julian days 188 and 201 using the "BASE" simulation.This figure demonstrates that BrO x contributed more than 90 % and 85 %, respectively, to GEM depletion on a 24-h basis.
Most of the GEM oxidation (>80 %) was caused by BrO, especially under lower BrO x levels associated with higher O 3 levels and hence low [Br]/[BrO] ratios.The associated diurnal average GEM lifetime against oxidation by BrO for Julian day 188 is 41.2 min, equivalent to a lifetime of 6.86 min for the daily peak BrO level of 90 pmol mol −1 .Under lower BrO x concentrations, the relative contributions of Br and Br 2 are comparable due to relatively lower [Br]/[BrO] ratios.The integrated sum contribution of NO 3 , O 3 , and OH was between ∼4 and 8 % on a 24-h basis for these simulation days.
The similarity in rate constants of the reactions of Br and BrO with GEM, using the rate constants of the "BASE" simulation, suggests that the role of BrO in GEM oxidation may dominate over Br as long as [Br]/[BrO] remains below ∼0.2-0.5, which is expected for O 3 > 4-5 nmol mol −1 (Fig. 5).This is usually the case in the troposphere (e.g.Fig. 5; Wayne et al., 1995).It is important to note that contributions of Br and BrO to GEM oxidation, as well as suggested rate constants, are based on best fits of observations and model results during mid-day AMDEs under Dead Sea conditions; relative contributions of Br to GEM oxidation are expected to be higher in other areas, where lower temperatures are associated with lower back dissociation of HgBr.

Temperature effects
Temperature can significantly influence the efficiency of GEM oxidation by BrO x in the gaseous and aqueous phases.Temperature-dependencies for many mercury reactions are unknown, but the thermal back-dissociation of HgBr is expected to significantly slow down GEM oxidation in the gas phase (Goodsite et al., 2004).Using the rate coefficient given by Goodsite et al. (2004) for the back dissociation rate of HgBr, the thermal back-dissociation of HgBr is ∼2.6 orders of magnitude higher under the average Dead Sea summer temperature measured during the campaign (∼310 K) compared to typical temperatures in the polar regions (240 K).Hence, if Br plays a major role in GEM oxidation one should expect significantly slower oxidation of GEM by BrO x at the Dead Sea compared with the polar regions.
Figure 7 presents the GEM diurnal profile obtained for "BASE", "WINTER", and "POLAR" simulations in order to investigate temperature effects on GEM oxidation.Only direct effects of temperature on the Hg-involved chemical reactions were investigated, while indirect effects of temperature -for instance, on the release of BrO x into the gas phase (Sander et al., 2006) -were ignored here.Figure 7b demonstrates that under relatively moderate BrO levels (∼40-50 pmol mol −1 ; JD 201), oxidation of GEM under typical polar temperatures was more efficient than for the Dead Sea summer by 69.9 % on average, and by 10.1 % between typical Dead Sea winter and summer temperatures.Figure 7a shows that temperature effects were lower under high BrO x levels (JD 188), where differences in GEM depletions averaged 19.7 % between Dead Sea summer and polar temperatures and 10.9 % between typical Dead Sea winter and summer temperatures.We attribute these differences to the fact that GEM depletion becomes highly limited by low GEM concentrations in the presence of high BrO x concentrations.Obrist et al. (2011) showed that RGM formation by BrO x at the Dead Sea occurred even at low BrO concentrations between ∼4 and 6 pmol mol −1 (and possibly below because lower BrO could not be measured due to the DOAS detection limits), suggesting that BrO x has important implications for GEM depletions above oceans in the mid-latitudes and tropics.The present analysis indicates that GEM depletions in such areas may not be subject to strong temperature dependencies; at low BrO x levels (e.g., BrO <∼ 6.5 pmol mol −1 , Saiz Lopez et al., 2004), [Br]/[BrO] ratios are expected to be significantly lower than ∼0.2-0.5 (e.g., Fig. 5; Wayne et al., 1995), and thus BrO may be a dominant GEM oxidant if our proposed rate constants are confirmed for these areas.

Summary and conclusions
This study demonstrates that best agreement between model simulations and observed AMDEs at the Dead Sea was obtained by using k Hg+Br and k Hg+BrO of 2.7 × 10 −13 and 1.5 × 10 −13 cm 3 molecule −1 s −1 , respectively ("BASE" kinetics).A value k Hg+BrO of 5.0 × 10 −14 cm 3 molecule −1 s −1 is suggested as a minimum value.The "BASE" kinetics suggests a high efficiency and central role of BrO x with daytime relative contribution of more than ∼90 % for AMDEs at the Dead Sea, with BrO being the dominant oxidant with relative contributions above ∼80 %.Using the "BASE" kinetics, BrO was found to be the dominant GEM oxidant for [Br]/[BrO] ratios below ∼0.2-0.5, which is usually the case in the troposphere.It should be emphasized that these rate constants are based on best agreements of observations and model results during AMDEs under Dead Sea conditions, and kinetic studies are necessary in order to confirm these suggested rate constants.High contributions of BrO to GEM oxidation suggests a low temperature dependency of GEM oxidation by BrO x , indicating that BrO x -induced GEM depletion can be important outside of the polar region even under relatively low BrO x levels.

FigureFig. 2 .
Fig. 2. Simulated and measured diurnal profiles of GEM and BrO.(a) Measured and simulated diurnal profiles of BrO, and simulated diurnal profile of [Br]/[BrO] for Julian day 188.(b) GEM diurnal profiles for different combinations of k GEM+BrO and k GEM+Br values.(c) Same as (b)with simulated GEM mixing ratios being constrained between 07:30 and 13:30 LT.Note that constrains of fluxes were not added during AMDEs which was the main emphasis of our analyses.

Figure 3 .Fig. 3 .
Figure 3. Simulated and measured diurnal profiles of GEM and BrO.(a) Measured and simulated diurnal profiles of BrO and simulated diurnal profile of [Br]/[BrO] for Julian day 201.(b) Different combinations of values for kGEM+BrO and kGEM+Br were used to evaluate the simulated GEM diurnal profiles. 32

Fig. 5 .
Fig. 5. Relative importance of Br and BrO to GEM-oxidation based on underlying LP-DOAS O 3 mixing ratios.(a) Measured GEM depletion per unit BrO (i.e., GEM/ BrO) is presented as a function of O 3 .GEM/ BrO was calculated for all campaign days and averaged for respective O 3 values.(b-d) Simulated GEM/ BrO as a function of [Br]/[BrO] for simulations L Br H BrO (i.e., "BASE"; b), L Br L BrO (c), and H Br L BrO (d).Measured O 3 are expected to correspond to [Br]/[BrO] ratios in simulations due to known correlations between the two.Simulated GEM/ BrO values were calculated only for periods of BrO > 0 based on Julian days 188, 197, and 201.

Figure
Figure5b-d shows simulated GEM/ BrO for Julian days188, 201, and 197.In these graphs, GEM/ BrO is graphed as a function of [Br]/[BrO] (which is based on, and hence analogous to, O 3 levels in panel a).Our results indicate that modeled GEM/ BrO versus [Br]/[BrO] is similar to observed patterns only for the "BASE" simulation (Fig.5b).Even for this case, however, GEM/ BrO ratios tend to peak at somewhat lower O 3 , which may suggest that the ratio of rate constants for BrO and Br (e.g., k GEM+BrO /k GEM+Br ) still may be underestimated for the "BASE" simulation.For simulations "H Br L BrO" and "L Br L BrO", GEM/ BrO was greatly overestimated for O 3 levels below 10 nmol mol −1 , which is expected if k GEM+BrO /k GEM+Br is underestimated.This result further suggests that the rate constant for L BrO is too low to accurately represent observed results.Possibly, a decrease in efficiency of GEM depletion by BrO at O 3 levels above ∼30-40 nmol mol −1 as observed in the "BASE" simulation may be due to the importance of O 3 for GEM oxidation at high O 3 levels.

Figure 6 .
Figure 6.Relative contributions of various oxidants to GEM depletion.Relative contributions (in %) of different species to GEM oxidation and associated GEM lifetimes for Julian days 201 (a) and 188(b), based on "BASE" simulation.Maximum mixing ratios of 20 pmol*mol -1 were used for NO 3 .

Figure 7 .
Figure 7. Influence of BrOX and temperature on GEM depletion.The influence of BrOX and temperature on GEM depletion rate.The GEM diurnal profiles are shown for the "BASE", "WINTER", "POLAR", "NOBROX," and "ONLYBR" simulations for Julian days 188 (a) and 201(b).

Table 3 .
Correlation and average error during AMDEs as an indicator for simulated GEM agreement with measured GEM during AMDEs.
Relative contributions of various oxidants to GEM depletion.Relative contributions (in %) of different species to GEM oxidation and associated GEM lifetimes for Julian days 201 (a) and 188 (b), based on "BASE" simulation.Maximum mixing ratios of 20 pmol mol −1 were used for NO 3 .