1. Introduction
Atmospheric mercury (Hg) comprises three operationally defined forms. Gaseous elemental Hg (GEM) is the dominant form, whereas gaseous oxidized Hg (GOM) and particulate-bound Hg (PBM) are the oxidized forms typically present at ambient concentrations in the pg m−3 range. GEM can undergo dry deposition via uptake (Obrist et al. 2017), while GOM and PBM can undergo dry and wet deposition. Once deposited, oxidized Hg can form methylmercury and cause toxic effects on wildlife and humans (Evers 2018; Wright et al. 2018). The oxidation of GEM is one of the major pathways leading to the production of oxidized Hg in the atmosphere; however, the knowledge on many aspects of the transformation process, including the reaction mechanisms and kinetics, oxidant concentrations, stability of the oxidation products and the role of multiphase surfaces, remain incomplete.
There are a number of oxidants involved in the gas-phase oxidation of GEM, such as Br, BrO, Cl, ClO, HO2, I, NO2, O3, and OH, which are still a matter of debate and subject to ongoing evaluations (Hynes et al. 2009; Subir et al. 2011; Ariya et al. 2015). Earlier experimental work focused on investigating the role of O3 and OH in oxidizing GEM (Hall et al. 1995; Sommar et al. 2001; Pal and Ariya 2004a,b); however, these mechanisms were unable to account for the rapid oxidation of GEM and subsequent high concentrations of GOM observed in the polar springtime (Lu et al. 2001). The HgO and HgOH oxidation products were also considered too unstable (Goodsite et al. 2004; Calvert and Lindberg 2005; Hynes et al. 2009). BrO and Br were proposed as the oxidants driving the atmospheric Hg depletion events as well as ozone depletion in the Arctic due to their fast reaction kinetics and active bromine cycling over the snow and sea ice surfaces (Goodsite et al. 2004; Dastoor et al. 2008, 2015; Mao et al. 2010; Stephens et al. 2012; Toyota et al. 2014; Wang et al. 2019). Subsequent modeling studies suggested that the role of Br and BrO as oxidants of GEM extend beyond the polar region, including in the free troposphere and tropical upper troposphere (Holmes et al. 2006; Coburn et al. 2016; Saiz-Lopez and Fernandez 2016; Shah et al. 2016; Bieser et al. 2017), midlatitude marine boundary layer (MBL) (Holmes et al. 2009; Obrist et al. 2011; Ye et al. 2016), and on a global scale (Holmes et al. 2010; Schmidt et al. 2016; Shah and Jaeglé 2017; Travnikov et al. 2017). In addition to Br, theoretical studies suggested that the intermediate product HgBr of the GEM + Br reaction could further react with BrO, ClO, HO2, I, NO2, or OH and yield stable oxidation products (Dibble et al. 2012; Jiao and Dibble 2017), which is supported by modeling studies (Wang et al. 2014; Ye et al. 2016, 2018; Horowitz et al. 2017). While progress has been made on understanding the fate of HgBr, there are large uncertainties on oxidant concentrations in the atmosphere and the rate coefficients (Holmes et al. 2006; Hynes et al. 2009; Dibble et al. 2012; Wang et al. 2014; Shah et al. 2016). The photoreduction of GOM and its implications on Hg transformation and the lifetime of GEM also have not been fully explored (Saiz-Lopez et al. 2018).
In this study, a chemistry box model developed by Ye et al. (2016) was used to simulate the Hg oxidation mechanisms occurring in the atmosphere, and the model results were evaluated using 8 years of GOM measurements at Kejimkujik, Nova Scotia, Canada. This Atlantic Canada site is unaffected by local anthropogenic Hg emissions implying that local sources contribute minimally to variations in atmospheric Hg. It is impacted by continental and MBL air masses, which allows for an investigation into the Hg oxidation processes occurring in different air masses. We identified the dominant GEM oxidation pathways in simulations where all GEM oxidation reactions occurred simultaneously and in simulations excluding oxidation by O3 and OH. The model results were used to explain the observed seasonal and diurnal concentrations in GOM. Findings from this study can help improve future model predictions of GOM and provide further insight into the speciation of GOM, which are important to the understanding of the fate and toxicity of Hg in ecosystems and calibration of GOM measurements.
2. Methods
a. Datasets for model input and evaluation
GEM, GOM, and PBM (aerodynamic diameter <2.5 μm) concentrations measured from 2009 to 2016 at the Kejimkujik National Park, Nova Scotia, Canada, were used in the model simulations and evaluations. Kejimkujik (44.43°N, 65.2°W; 158 m MSL) is a remote forested site located in eastern Canada (Fig. S1 in the online supplemental material) and is ~65 km from the Atlantic Ocean. The Hg measurements were collected using the Tekran mercury speciation system (Models 1130/1135/2537A), which is programmed to measure GEM every 5 min and GOM and PBM every 3 h. The data were quality controlled and archived by the National Atmospheric Deposition Program (NADP)’s Atmospheric Mercury Network (NADP 2017). Previous studies reported measurement uncertainties of a factor of 1.3–5 for GOM due to variability in the sampling efficiency for different GOM species and biased measurements at high ambient ozone (i.e., >50 ppb) and high humidity (Lyman et al. 2010; Ambrose et al. 2013; Gustin et al. 2013, 2015; Huang and Gustin 2015; Weiss-Penzias et al. 2015). At Kejimkujik, ozone mixing ratios typically ranged from 20 to 40 ppb. In addition, a Model 1102 air dryer was used in conjunction with the Model 1130 instrument as per standard AMNet operations to remove water vapor from ambient air. Hence, the uncertainties in GOM measurements are not a major concern at Kejimkujik. We obtained collocated hourly O3 measurements [National Air Pollution Surveillance (NAPS) station ID: 30501] and NO2 measurements from the nearest regionally representative monitoring site (NAPS station ID: 30701) and collocated hourly meteorological data from Environment and Climate Change Canada (ECCC; ECCC 2017a,b). Collocated hourly global horizontal irradiance (GHI) data produced from the Physical Solar Model were obtained from the National Solar Radiation Database (DOE 2017).
b. Back trajectory modeling and airmass classification
Back trajectories were produced from the HYSPLIT 4 model (Stein et al. 2015; Rolph et al. 2017) using the EDAS 40-km archived meteorological data; 72-h back trajectories were generated every 3 h from Kejimkujik at a starting height corresponding to one-half of the modeled mixing height. To determine the dominant daytime airmass pattern for each Hg sampling day, back trajectories arriving at the site between 0600 and 1800 LST were categorized into one of the following airmass origins: continental-north, continental-south, coastal, or open-ocean regions (Fig. S1). We defined a dominant airmass pattern if ≥70% of the trajectory endpoints for the daytime trajectories lies within one of the four origins. We analyzed back trajectories exclusively from continental regions or marine (i.e., open ocean) regions. The proportion of air masses arriving at Kejimkujik from continental and marine regions were ~86% and ~14%, respectively. The different airmass types were used for determining initial conditions in model simulations (Table S1), as further discussed in the next section. Back trajectories associated with mixed air masses were not analyzed due to the complex interactions between land and marine air, which is not entirely understood (Ye et al. 2016).
c. Mercury chemistry modeling
Numerical integration of the chemical reaction mechanisms were performed using the Kinetic Preprocessor (KPP) 2.2.3 software (Damian et al. 2002; Daescu et al. 2003; Sandu et al. 2003; KPP 2018). The chemical reactions and rate expressions used in this study were obtained from the model developed by Ye et al. (2016). The model contains 507 chemical reactions. In the gas phase, the model simulates GEM oxidation by Br, BrO, Cl, Cl2, H2O2, I, O3, and OH and second step oxidation of HgBr or HgCl by Br, BrO, ClO, HO2, NO2, and/or OH (Table 1). Reduction of Hg+2 by HO2, Hg(OH)2, and HgSO3 are simulated in the aqueous phase. Aside from Hg chemistry, the model includes gas-phase bromine (Ye et al. 2018), chlorine, iodine, ozone, nitrogen oxides, volatile organic compounds (VOCs) and sulfur chemistry, and gas–aqueous equilibrium and aqueous phase reactions. The reactions and kinetics were taken from the JPL Chemical Kinetics and Photochemical Data for Use in Atmospheric Studies Report 17 (Sander et al. 2011) and Atkinson et al. (2004, 2008). In the model, VOCs react with various gas-phase species, such as OH, Br, Cl, HO2, NO, NO2, NO3, and O2, and undergo photodissociation. Table S1 shows the initial inputs for GEM, O3, NO2, solar radiation, and temperature at Kejimkujik and other model parameters. Initial concentrations of other chemical species are shown in Table S2. Note that measurements for the chemical species in Table S2 were not available; therefore, the initial concentrations were assumed to be the same as those of the New Hampshire sites modeled in Ye et al. (2016). The three sites in New Hampshire (Appledore Island, Thompson Farm and Pack Monadnock) and the Kejimkujik site are in the Gulf of Maine–northeastern U.S. region. Gas–aqueous equilibriums are included for many chemical species, such as GOM species, OH, HO2, H2O2, NO3, CH3Br, CH3OOH, CH3CHO, and HCHO. It also contains aqueous reactions for oxidized Hg species, SO2, HSO3−, SO32−, H2SO4, HSO4−, H2O2, HO2, HNO3, OH, O3, Br, HOBr, Br−, Br2−, HOBr−, etc. (details in supplemental section S1). The photo-dissociation rates specific to the Kejimkujik site for the photolysis reactions were obtained from the Tropospheric Ultraviolet and Visible Radiation (TUV) model 5.3.1 (NCAR 2018).
Gaseous elemental mercury (GEM or Hg0) oxidation reactions in the gas phase. The reactions listed in the table were used in the chemical scheme with all GEM oxidation reactions. In an alternative chemical scheme, gas-phase oxidation by O3 and OH were excluded in model simulations by setting the rate coefficients for the first two reactions to zero. Gas-phase oxidation by O3, OH, and H2O2 were also excluded in another model simulation by setting the first three reaction rate coefficients to zero. T is temperature (K), and M is the number density of air.
GOM concentrations in the free troposphere were not measured at Kejimkujik; therefore, we used modeled vertical profiles in Bieser et al. (2017). Note that the modeled vertical profiles in Bieser et al. (2017) were generated from specific chemical schemes (O3chem, OHchem and Brchem); thus, they correspond to HgO and HgBr2 gas-phase species. The modeled vertical profiles will also have some uncertainties given that the atmospheric Hg chemistry in the free troposphere is also uncertain.
GEM, O3, NO2, and other species whose concentrations do not vary within a few days were treated as fixed parameters. Br2 and other species whose concentrations change rapidly with time were defined as variable, except in simulations where we treated Br2 as a fixed variable. For Br2, we tested different mixing ratios in the model to determine the optimal mixing ratios that lead to agreement in modeled and observed GOM. We found that the optimal Br2 mixing ratios were too low at this site to be necessarily different for continental versus marine air.
3. Results and discussion
a. Chemical scheme with all GEM oxidation reactions
In this chemical scheme, GEM, O3, NO2, and temperature inputs were different in each simulation varying according to measurements, and the same Br2 were used in all simulations. Based on the model, GOM was predominantly formed by GEM oxidation by O3 and OH (79%), H2O2 (10%), Br with second step oxidation of HgBr by NO2 (7%), BrO (3%), and other oxidants (<0.35%). The percentage in the parentheses indicates the contribution of each reaction to the modeled GOM concentration.
The modeled and observed mean concentrations of GOM were 1.7 ± 1.5 pg m−3 and 1.2 ± 2.0 pg m−3, respectively. The NMB of the modeled GOM concentrations was 50%, which varied by season and airmass type. The NMB was 42% during spring and 78% during summer. The biases were higher during summer than spring not because the model was predicting higher concentrations, but rather it was because of the very low observed concentrations. A potential reason could be the higher dry deposition during summer that has been observed over forest canopies during the growing season (Zhang et al. 2009). The NMB of the modeled GOM concentrations associated with continental and marine airmass origins were 53% and 4%, respectively. The much lower model bias for marine airflows suggests that the reactions, rate constants and initial inputs in the model were more representative of the chemistry in the marine boundary layer than continental boundary layer.
The oxidation of GEM by O3 and OH alone without other oxidation mechanisms was able to reproduce the observed GOM, similar to the results from some modeling studies (Holmes et al. 2010; Weiss-Penzias et al. 2015; Ye et al. 2016; Bieser et al. 2017; Travnikov et al. 2017). GEM oxidation by O3 and OH were also the dominant oxidation mechanisms in both continental and marine air masses. The percent contributions to GOM from these reactions were 80% and 76%, respectively. Differences in the percent contributions between continental and marine air were small for other reactions. The contribution from GEM oxidation by H2O2 was 2.2% higher in continental airflows than marine airflows. The contribution from GEM oxidation by Br with NO2 in the second reaction step was 2.1% higher in marine airflows than continental airflows. The contribution from GEM oxidation by BrO was 1.9% higher in marine airflows than continental airflows.
The dominant oxidants will change depending on the Br2 mixing ratio in the model. As Br2 increases, GEM oxidation by Br becomes more important than those of O3 and OH. However, the model indicates that it would result in too much oxidation at this site as shown by the large NMB (Fig. 1). At very low Br2 (0.1 ppqv), the NMB was 56% for all cases and 59% for cases associated with continental airflows. Note that this Br2 input corresponds to mean simulated Br and BrO mixing ratios that are 14 and 18 times lower, respectively, than modeled data for this region (Holmes et al. 2010; Parrella et al. 2012; Schmidt et al. 2016). Only cases associated with marine air masses resulted in comparable modeled and observed GOM (NMB = 7%). For marine air masses, GEM oxidation by Br was necessary to compensate for the lower HgO production that resulted from increased gas-particle partitioning of HgO (i.e., higher LWC). This is consistent with the above results showing a slightly lower contribution from oxidation by O3 and OH and higher contribution from oxidation by Br with NO2 for marine air masses compared to continental air masses.
Sensitivity tests on Br2 mixing ratio input and effects on the normalized mean bias (NMB) of modeled GOM (simulation with all GEM oxidation reactions). Increase in Br2 leads to a significant overprediction of observed GOM concentrations at this site.
Citation: Journal of the Atmospheric Sciences 77, 3; 10.1175/JAS-D-19-0120.1
Simulations were also conducted using lower rate coefficients for GEM oxidation by O3 and OH. The simulated average GOM decreased from 6.2 ± 6.7 pg m−3 to 5.2 ± 6.0 pg m−3 and the NMB was reduced from 437% to 352% (Table 2). Note that this simulation assumed a Br2 of 5.6 ppqv, which corresponds to simulated Br that are similar to previous model estimates for this region (Holmes et al. 2010; Parrella et al. 2012; Schmidt et al. 2016). Entrainment of GOM was also simulated in the model. HgO and HgBr2 concentrations in the free troposphere were obtained from modeled values in Bieser et al. (2017), and gas-particle partitioning was assumed to be 50% in the gas and particle phase [CFT, HgO(g) = 60 pg m−3; CFT, HgBr2(g) = 65 pg m−3]. The addition of entrainment resulted in a simulated mean GOM of 6.6 ± 6.1 pg m−3 and an NMB of 469% for the simulation with lower rate coefficients for GEM oxidation by O3 and OH (Table 2). The model biases were still significant despite reducing the reaction rates for O3 and OH. Including entrainment of GOM from the free troposphere increased the model bias. We find that a reduction in the Br2 mixing ratio had the most effect on decreasing the model bias (Table 2).
Statistics of modeled GOM (pg m−3) and normalized mean biases (NMB) for simulations using the chemical scheme with all GEM oxidation reactions (includes oxidation by O3 and OH).
Our results for the chemical scheme with O3 and OH were similar to model simulations conducted at inland and coastal sites in the northeastern United States (Ye et al. 2016). In that study, GEM oxidation by O3 and OH accounted for 62%–88% of the simulated GOM (Ye et al. 2016). However, modeling studies conducted in MBL environments found that Br and second step oxidants (e.g., Br, I, NO2, HO2) were likely the major oxidants (Holmes et al. 2009; Wang et al. 2014; Ye et al. 2016). There are several reasons that may explain the difference in the dominant oxidants. First, the observed GOM concentrations at Kejimkujik were lower than at previous MBL sites (Holmes et al. 2009; Wang et al. 2014; Ye et al. 2016) attributing the observed concentrations at Kejimkujik to oxidation of GEM by O3 and OH alone. At locations where the observed GOM concentrations are much higher (assuming similar concentration levels of O3 and OH), a higher Br2 input and subsequently, a higher contribution of oxidation by Br to GOM production is expected. Second, our input concentrations for O3 and simulated Br and BrO mixing ratios were different from those previous MBL studies. O3 concentrations at Kejimkujik were ~3 times higher than some MBL sites, and Br and BrO mixing ratios in this study were much lower than those in previous model simulations (Holmes et al. 2009; Wang et al. 2014; Ye et al. 2016).
b. Chemical scheme without GEM oxidation by O3 and OH
In recent Hg model simulations, such as in the GEOS-Chem model (Shah et al. 2016; Horowitz et al. 2017; Shah and Jaeglé 2017; Travnikov et al. 2017), gas-phase oxidation of GEM by O3 has been excluded because thermodynamics and laboratory studies suggest the reaction proceeds too slowly in the atmosphere and is likely not in the pure gas phase (Calvert and Lindberg 2005; Hynes et al. 2009; Subir et al. 2011). Likewise, the gas-phase OH reaction has also been excluded in some Hg models because of thermodynamic considerations (Hynes et al. 2009; Horowitz et al. 2017). In recent Community Multiscale Air Quality (CMAQ) modeling of Hg, GEM oxidation by O3 and OH were treated as a solid-phase product (Ye et al. 2018). In this chemical scheme, both GEM oxidation by O3 and OH in the gas phase were turned off in the model. Since measurements of Br2 are scarce, one of the goals of simulating this chemical scheme was to determine an optimal Br2 mixing ratio that would improve the model–observation agreement of GOM. In the model, photolysis of Br2 as well as the recycling of BrO produce Br.
The modeled GOM statistics for the simulations with Br2 mixing ratios ranging from 5.6 to 0.73 ppqv are shown in Table 3. Using a [Br2] = 5.6 ppqv, we obtained simulated mean Br and BrO mixing ratios of 0.013 and 0.1 pptv, respectively, which are comparable to those predicted by atmospheric bromine models for this region (Holmes et al. 2010; Parrella et al. 2012; Schmidt et al. 2016). However, the model significantly overestimated the observed GOM concentrations at this Br2 mixing ratio (NMB = 319%, Table 3). The best model fit to the observations was obtained using a [Br2] = 0.73 ppqv. This corresponds to simulated mean [Br] and [BrO] of 0.003 and 0.02 pptv, respectively, which are substantially lower than modeled bromine estimates and below the detection limit of the Multi-Axis Differential Optical Absorption Spectroscopy (MAX-DOAS) instrument for measuring BrO (0.5 pptv; Coburn et al. 2016). Using the optimal [Br2] = 0.73 ppqv in the chemical scheme without oxidation by O3 and OH, GEM was predominantly oxidized by Br with NO2 in the second reaction step (58%), BrO (23.5%), H2O2 (16%), and Br with HO2 in the second reaction step (3%). The abundance of BrHgNO2 relative to other HgBr compounds was also reported in previous simulations (Coburn et al. 2016; Horowitz et al. 2017).
Statistics of modeled GOM (pg m−3) and normalized mean biases (NMB) for model simulations without gas-phase GEM oxidation by O3 and OH; Vd is the dry deposition velocity of GOM for the Kejimkujik site.
The dry deposition velocity of GOM (Vd,GOM) was increased in the model in an attempt to reduce the model bias of GOM. Increasing Vd,GOM from 2 to 5 cm s−1, the range of Vd,GOM reported over forested sites (Zhang et al. 2009), decreased the peak GOM concentration by ~55% for the simulation using a [Br2] = 5.6 ppqv. However, the model still overestimated the observed concentrations (Table 3), suggesting that the use of modeled Br2 data would result in poor predictions at this site even if Vd,GOM had been underestimated. This example illustrates why reliable bromine measurements are necessary. Currently, atmospheric bromine measurements are very limited and have only been reported for the free troposphere and in polar regions and the Dead Sea, where the mixing ratios are mostly above detection limit. Direct measurements in Alaska showed that the mixing ratios could reach 35 pptv for Br2, 14 pptv for Br, and 25 pptv for BrO during GEM depletion events (Wang et al. 2019).
The mean and standard deviation of the simulated GOM concentrations for this chemical scheme ([Br2] = 0.73 ppqv) was 1.1 ± 1.1 pg m−3 and the NMB was −6.0%. The model bias was smallest for cases associated with continental air (NMB = −4.3%). The model underestimated the observed GOM concentrations associated with marine air (NMB = −29%), whereas it overestimated the observed summertime concentrations (NMB = 29%). The results showed that the Br2 mixing ratio was optimal for cases associated with continental air; however, a higher Br2 mixing ratio was necessary for simulating marine air.
GEM oxidation by H2O2 is also under debate due to uncertainties on the gas-phase product of this reaction and the greater likelihood of this reaction occurring in the heterogeneous phases (Subir et al. 2011). In model scenarios that exclude gas-phase oxidation by O3, OH and H2O2, the [Br2] input needed to be increased slightly to 1 ppqv in order for the model to reproduce the observed GOM (Table 3). Optimal [Br2] were 1 and 1.5 ppqv for cases associated with continental and marine air masses, respectively. The Br2 mixing ratio are still lower than modeled Br2 data in this region.
There are several caveats that concern the dominant GEM oxidation mechanisms inferred from model simulations. The mixing ratios of Br and BrO are uncertain because of a lack of measurements in the midlatitudes of the PBL. The reaction rate coefficients also have large variability according to experimental and thermodynamic studies (Subir et al. 2011; Song et al. 2018). The understanding of atmospheric Hg chemistry is also frequently evolving. Recent computational studies proposed a reaction that competes with the HgBr + NO2 addition reaction, which leads to the formation of GEM and BrHgO radical (Jiao and Dibble 2017). The rate constant for this reaction is uncertain (Saiz-Lopez et al. 2018), and another study suggested that the BrHgO radical can form GOM by reacting with other compounds, for example, NO, NO2, CH4, C2H6, and other VOCs yet to be determined (Lam et al. 2019). The roles of aqueous and gas-phase reduction of GOM are also a subject of debate. Although some models indicated that aqueous reduction of GOM played an important role (Bash et al. 2014; Horowitz et al. 2017), a recent study utilizing quantum chemical methods suggested that gas-phase photolysis of GOM to HgBr resulted in improved agreement between modeled and observed GEM (Saiz-Lopez et al. 2018).
c. Trends by season and airmass origins and potential drivers
The model results for either chemical scheme showed good consistency with observed spring/summer trends and differences in concentrations from continental and marine air masses. In terms of seasonal trends, the mean or median observed GOM was higher during spring than summer (Fig. S2). This trend was reproduced in the model results. The larger dispersion in the observed GOM during summer than spring was also seen in the model results. In the observed springtime data, the mean or median GOM concentration was higher from continental areas than the open ocean. This difference was also reproduced by the model. During summer, the trend was reversed. GOM concentrations in the air masses of marine origin were higher than those of continental origin in the observed data, which was largely reproduced in the model results (Fig. S2).
We examined the model parameters driving the seasonal and airmass patterns in simulated GOM concentrations. In the chemical scheme with oxidation by O3 and OH, a large proportion of the decrease in modeled GOM from spring to summer was attributed to HgO since this was the dominant GOM specie. We analyzed parameters closely related to HgO including the reactant concentrations (GEM, O3, and OH) and physical parameters such as temperature, solar irradiance and LWC. Modeled HgO was significantly (p < 0.0001) correlated with measured GEM (r = 0.33) and O3 (r = 0.67) and modeled OH (r = 0.91), solar irradiance (r = 0.73), and LWC (r = −0.54). The correlation coefficients were calculated from the inputs and model output (steady-state concentrations) for each hourly simulation. Thus, the seasonal trends in modeled HgO were also consistent with those of GEM, O3, OH, and LWC. The effect of solar irradiance was primarily on modeled OH (r = 0.93) based on several photodissociation mechanisms in the model. For GEM, O3, and OH, the mean concentrations were higher in the spring than summer, which were similar to modeled HgO. For LWC, the higher value in the summer that we parameterized based on relative humidity (Table S1) resulted in lower modeled HgO in the summer, which agreed with observations. HgO was not correlated with temperature (r = −0.02); hence, the differences in HgO between spring and summer were not attributed to the substantial temperature differences. Differences in the modeled GOM concentrations between continental and marine air masses were also largely attributed to modeled HgO. In the spring, higher HgO and OH and lower LWC were associated with continental air masses than marine masses (Fig. 2), suggesting OH concentrations and LWC were the main parameters driving the differences in HgO in different air masses. The higher relative humidity observed in marine air masses led to a higher parameterized value for LWC, which increased the partitioning of GOM to PBM. Thus, gas-particle partitioning was one of the driving mechanisms for the differences in GOM associated with continental and marine air masses during spring. In the summer, higher GEM and O3 concentrations in marine air explained the higher HgO associated with marine air than continental air (Fig. 2). This analysis provides further insight into the key parameters controlling variations in GOM concentrations. Ensuring high-quality data on these parameters may improve future predictions of GOM.
Trends in mean concentrations of modeled and measured parameters by season and airmass origin (simulation with all GEM oxidation reactions). Modeled parameters include HgO, Hg(OH)2, BrHgNO2, HgBrO (pg m−3), and OH (×10−1 ppt). Measured parameters include GEM (ng m−3) and O3 (×10 ppb). LWC was parameterized based on measured average relative humidity of continental and marine air masses during spring and summer (×10−11 m3 water m−3 air).
Citation: Journal of the Atmospheric Sciences 77, 3; 10.1175/JAS-D-19-0120.1
In the chemical scheme without oxidation by O3 and OH ([Br2] = 0.73 ppqv), BrHgNO2 and HgBrO were the dominant GOM species. The one exception was the similar contributions by Br (42%) and BrO (40%) oxidants for marine air masses during spring (Fig. 3). This is partly driven by the higher temperatures associated with marine air masses relative to continental air masses (14° vs 5°C), which increased the dissociation rate of the HgBr intermediate by a factor of 2.4 and subsequently reduced the production of BrHgNO2. BrHgNO2 was significantly (p < 0.0001) correlated with Br (r = 0.97) and weakly correlated with NO2 (r = 0.05), while HgBrO was correlated with BrO (r = 0.99). As shown in Fig. 3, the trends in BrHgNO2 and Br tracked each other very well. This analysis indicates that Br and BrO are critical in reproducing the observed GOM concentrations for model simulations without GEM oxidation by O3 and OH. While experimental and computational studies have suggested this chemical scheme is highly plausible in the atmosphere with wide adoption in modeling studies, significant uncertainties in the Br and BrO mixing ratios remain. Considering that modeled Br and BrO data are widely used in mercury models owing to a lack of measurements in the midlatitudes of the PBL, there is a greater need to ensure that the modeled atmospheric bromine data are validated.
Trends in mean concentrations of modeled and measured parameters by season and airmass origin (simulation without GEM oxidation by O3 and OH, [Br2] = 0.73 ppqv). BrHgNO2 is the dominant GOM species in this chemical scheme and its trends closely followed that of atomic Br.
Citation: Journal of the Atmospheric Sciences 77, 3; 10.1175/JAS-D-19-0120.1
d. Diurnal variations
Diurnal variations in the observed GOM and modeled GOM derived from the two chemical schemes are shown in Fig. 4. In the spring, summer and continental airflow cases, the timing of the daily peak concentration differed between the observations and model results for both chemical schemes. The largest discrepancies were found over 0800–1200 LST, when the growth of the planetary boundary layer led to strong downward mixing of air in the residual layer contributing to surface GOM concentrations. However, vertical mixing was not simulated in the model. In comparison, afternoon (1500–2100 LST) concentrations were more likely contributed by in situ chemistry and thus, the model–observation agreement was much better. This point is reinforced by the diurnal variation associated with marine air (Fig. 4), where the model–observation agreement was better because in the case of marine air the vertical mixing would be less vigorous than in continental air due to a weaker diurnal variation in the atmospheric boundary layer over the ocean (Holtslag et al. 2013).
Mean diurnal variations of observed and modeled GOM concentrations. The all reactions chemical scheme includes oxidation by O3 and OH. For the no O3 and OH chemical scheme, the diurnal variations correspond to the simulations using an optimal Br2 mixing ratio of 0.73 ppqv.
Citation: Journal of the Atmospheric Sciences 77, 3; 10.1175/JAS-D-19-0120.1
Figure 5 shows the diurnal variations of the dominant oxidants (Figs. 5a,b) and GOM species (Figs. 5c,d) from the two chemical schemes. Diurnal patterns were similar among GHI, modeled OH and modeled GOM in the chemical scheme with all GEM oxidation reactions (Fig. 5a). Values increased from 0400 to 1000 LST, leveled off between 1000 and 1300 LST, and then decreased afterward. These diurnal trends differed from that of O3, which exhibited nearly constant mixing ratios throughout the afternoon. Correlation analysis results also indicated a stronger relationship for modeled GOM and OH (r = 0.87) than modeled GOM and O3 (r = 0.66) during daytime. Thus, more of the variability in the modeled GOM can be explained by that of OH than O3. H2O2 mixing ratios were typically higher during nighttime and decreased to a minimum at noon. In the chemical scheme without OH and O3 (Fig. 5b), the diurnal variations of the mixing ratios ranged from 0 to 6 ppqv for Br and from 0 to 45 ppqv for BrO. The diurnal variation in modeled GOM was similar to that of Br, whereas a broader peak was observed in the modeled BrO.
Mean diurnal variations of the (a),(b) dominant oxidants and (c),(d) GOM species from the two chemical schemes. (a),(c) With all GEM oxidation reactions including oxidation by O3 and OH and (b),(d) without oxidation by O3 and OH. Modeled GOM (pg m−3), O3 (×10 ppb), modeled OH (×107 radicals cm−3) and H2O2 (×1010 molecules cm−3), GHI (W m−2), modeled BrO and Br (ppqv), and NO2 (×102 ppt). All GOM species in (c) and (d) are in pg m−3. Modeled Br correspond to the simulations using an optimal Br2 mixing ratio of 0.73 ppqv.
Citation: Journal of the Atmospheric Sciences 77, 3; 10.1175/JAS-D-19-0120.1
Diurnal variations of the concentrations for individual GOM species differed (Figs. 5c,d). The percentage contributions by the various oxidants to modeled GOM concentrations were also strongly dependent on the hour of day. In the chemical scheme with OH and O3, these oxidants dominated GOM production throughout the day and night (Fig. 5c). During daytime (0900–1500 LST), GEM oxidation by Br with second-step oxidation of HgBr by NO2 was the second most important mechanism, while the contribution by GEM oxidation by H2O2 was insignificant. In the chemical scheme without OH and O3 (Fig. 5d), Br and BrO were the dominant oxidants from the early morning to evening, whereas H2O2 provided some oxidation during nighttime.
4. Conclusions
GOM concentrations were predicted using a box model that simulates GEM oxidation by various oxidants in the gaseous (e.g., O3, OH, H2O2, Br, BrO, NO2, HO2) and aqueous phases. Two chemical schemes were simulated in this study: one scheme considering all GEM oxidation reactions and an alternative scheme that excludes gas-phase oxidation by O3 and OH. In the former scheme, the dominant GEM oxidation mechanisms inferred from the model are reactions with O3 and OH. GEM oxidation by Br with second-step oxidation of HgBr by NO2 and GEM oxidation by BrO were the dominant oxidation pathways in the chemical scheme without oxidation by O3 and OH. To reproduce the observed GOM at this site using the latter chemical scheme, the mixing ratios of Br and BrO needed to be lower than the modeled atmospheric bromine for this region. In the scheme with oxidation by O3 and OH, the variability in the GOM concentrations between seasons and between continental and marine air masses is primarily attributed to the variability in GEM, O3, OH, and aerosol liquid water content. In the scheme without oxidation by O3 and OH, Br and BrO are the key parameters accounting for the seasonal trends and differences in GOM concentrations between continental and marine air masses. Thus, validation of modeled Br and BrO data is critical to improving model predictions of GOM. There are also additional limitations beyond the scope of this modeling study. Computational and experimental studies on mercury chemistry are needed to inform modeling studies on viable GEM oxidation reactions and their reaction kinetics. Improved accuracy in the atmospheric oxidized mercury measurements will also constrain model parameters and build confidence in the model predictions.
Acknowledgments
This work is funded by Environment and Climate Change Canada’s (ECCC) Air Pollution Program. We acknowledge the National Atmospheric Deposition Program, National Air Pollution Surveillance Program, Meteorological Service of Canada, and National Renewable Energy Laboratory for the datasets used in this publication; and NOAA Air Resources Laboratory for the provision of the HYSPLIT transport and dispersion model. We thank Philip Cheung, Balbir Pabla, Young-Min Cho, and Deyong Wen from ECCC for technical support.
Data availability statement: The data and model can be accessed and reused by contacting the corresponding author.
REFERENCES
Ambrose, J. L., S. N. Lyman, J. Huang, M. S. Gustin, and D. A. Jaffe, 2013: Fast time resolution oxidized mercury measurements during the Reno Atmospheric Mercury Intercomparison Experiment (RAMIX). Environ. Sci. Technol., 47, 7285–7294, https://doi.org/10.1021/es303916v.
Ariya, P. A., A. Khalizov, and A. Gidas, 2002: Reactions of gaseous mercury with atomic and molecular halogens: Kinetics, product studies, and atmospheric implications. J. Phys. Chem., 106A, 7310–7320, https://doi.org/10.1021/JP020719O.
Ariya, P. A., and Coauthors, 2015: Mercury physicochemical and biogeochemical transformation in the atmosphere and at atmospheric interfaces: A review and future directions. Chem. Rev., 115, 3760–3802, https://doi.org/10.1021/cr500667e.
Atkinson, R., and Coauthors, 2004: Evaluated kinetic and photochemical data for atmospheric chemistry: Volume I—Gas phase reactions of Ox, HOx, NOx and SOx species. Atmos. Chem. Phys., 4, 1461–1738, https://doi.org/10.5194/acp-4-1461-2004.
Atkinson, R., and Coauthors, 2008: Evaluated kinetic and photochemical data for atmospheric chemistry: Volume IV—Gas phase reactions of organic halogen species. Atmos. Chem. Phys., 8, 4141–4496, https://doi.org/10.5194/acp-8-4141-2008.
Balabanov, N. B., B. C. Shepler, and K. A. Peterson, 2005: Accurate global potential energy surface and reaction dynamics for the ground state of HgBr2. J. Phys. Chem., 109A, 8765–8773, https://doi.org/10.1021/JP053415L.
Bash, J., A. Carlton, W. Hutzell, and O. R. Bullock Jr., 2014: Regional air quality model application of the aqueous-phase photo reduction of atmospheric oxidized mercury by dicarboxylic acids. Atmosphere, 5, 1–15, https://doi.org/10.3390/atmos5010001.
Bieser, J., and Coauthors, 2017: Multi-model study of mercury dispersion in the atmosphere: Vertical and interhemispheric distribution of mercury species. Atmos. Chem. Phys., 17, 6925–6955, https://doi.org/10.5194/acp-17-6925-2017.
Brooks, S., and Coauthors, 2014: Airborne vertical profiling of mercury speciation near Tullahoma, TN, USA. Atmosphere, 5, 557–574, https://doi.org/10.3390/atmos5030557.
Calvert, J. G., and S. E. Lindberg, 2005: Mechanisms of mercury removal by O3 and OH in the atmosphere. Atmos. Environ., 39, 3355–3367, https://doi.org/10.1016/j.atmosenv.2005.01.055.
Coburn, S., and Coauthors, 2016: Mercury oxidation from bromine chemistry in the free troposphere over the southeastern US. Atmos. Chem. Phys., 16, 3743–3760, https://doi.org/10.5194/acp-16-3743-2016.
Daescu, D., A. Sandu, and G. R. Carmichael, 2003: Direct and adjoint sensitivity analysis of chemical kinetic systems with KPP: II—Numerical validation and applications. Atmos. Environ., 37, 5097–5114, https://doi.org/10.1016/j.atmosenv.2003.08.020.
Damian, V., A. Sandu, M. Damian, F. Potra, and G. R. Carmichael, 2002: The Kinetic Preprocessor KPP—A software environment for solving chemical kinetics. Comput. Chem. Eng., 26, 1567–1579, https://doi.org/10.1016/S0098-1354(02)00128-X.
Dastoor, A. P., D. Davignon, N. Theys, M. Van Roozendael, A. Steffen, and P. A. Ariya, 2008: Modeling dynamic exchange of gaseous elemental mercury at polar sunrise. Environ. Sci. Technol., 42, 5183–5188, https://doi.org/10.1021/es800291w.
Dastoor, A. P., A. Ryzhkov, D. Durnford, I. Lehnherr, A. Steffen, and H. Morrison, 2015: Atmospheric mercury in the Canadian Arctic. Part II: Insight from modeling. Sci. Total Environ., 509–510, 16–27, https://doi.org/10.1016/j.scitotenv.2014.10.112.
Dibble, T. S., M. J. Zelie, and H. Mao, 2012: Thermodynamics of reactions of ClHg and BrHg radicals with atmospherically abundant free radicals. Atmos. Chem. Phys., 12, 10 271–10 279, https://doi.org/10.5194/acp-12-10271-2012.
DOE, 2017: National Solar Radiation Database (NSRDB). National Renewable Energy Laboratory, accessed 10 May 2018, https://nsrdb.nrel.gov.
Donohoue, D. L., D. Bauer, and A. J. Hynes, 2005: Temperature and pressure dependent rate coefficients for the reaction of Hg with Cl and the reaction of Cl with Cl: A pulsed laser photolysis-pulsed laser induced fluorescence study. J. Phys. Chem., 109A, 7732–7741, https://doi.org/10.1021/JP051354L.
ECCC, 2017a: NAPS data products. Accessed 10 December 2017, http://maps-cartes.ec.gc.ca/rnspa-naps/data.aspx?lang=en.
ECCC, 2017b: Historical climate data. Accessed 10 December 2017, http://climate.weather.gc.ca/historical_data/search_historic_data_e.html.
Evers, D., 2018: The effects of methylmercury on wildlife: A comprehensive review and approach for interpretation. The Encyclopedia of the Anthropocene, D. A. DellaSala and M. I. Goldstein, Eds., Elsevier, 181–194.
Faloona, I., and Coauthors, 2005: Observations of entrainment in eastern Pacific marine stratocumulus using three conserved scalars. J. Atmos. Sci., 62, 3268–3285, https://doi.org/10.1175/JAS3541.1.
Goodsite, M. E., J. M. C. Plane, and H. Skov, 2004: A theoretical study of the oxidation of Hg0 to HgBr2 in the troposphere. Environ. Sci. Technol., 38, 1772–1776, https://doi.org/10.1021/es034680s.
Goodsite, M. E., J. M. C. Plane, and H. Skov, 2012: Correction to A theoretical study of the oxidation of Hg0 to HgBr2 in the troposphere. Environ. Sci. Technol., 46, 5262–5262, https://doi.org/10.1021/es301201c.
Gustin, M. S., and Coauthors, 2013: Do we understand what the mercury speciation instruments are actually measuring? Results of RAMIX. Environ. Sci. Technol., 47, 7295–7306, https://doi.org/10.1021/es3039104.
Gustin, M. S., H. M. Amos, J. Huang, M. B. Miller, and K. Heidecorn, 2015: Measuring and modeling mercury in the atmosphere: A critical review. Atmos. Chem. Phys., 15, 5697–5713, https://doi.org/10.5194/acp-15-5697-2015.
Hall, B., P. Schager, and J. Weesmaa, 1995: The homogeneous gas phase reaction of mercury with oxygen, and the corresponding heterogeneous reactions in the presence of activated carbon and fly ash. Chemosphere, 30, 611–627, https://doi.org/10.1016/0045-6535(94)00428-W.
Hogan, R. J., A. L. M. Grant, A. J. Illingworth, G. N. Pearson, and E. J. O’Connor, 2009: Vertical velocity variance and skewness in clear and cloud-topped boundary layers as revealed by Doppler lidar. Quart. J. Roy. Meteor. Soc., 135, 635–643, https://doi.org/10.1002/qj.413.
Holmes, C. D., D. J. Jacob, and X. Yang, 2006: Global lifetime of elemental mercury against oxidation by atomic bromine in the free troposphere. Geophys. Res. Lett., 33, L20808, https://doi.org/10.1029/2006GL027176.
Holmes, C. D., D. J. Jacob, R. P. Mason, and D. A. Jaffe, 2009: Sources and deposition of reactive gaseous mercury in the marine atmosphere. Atmos. Environ., 43, 2278–2285, https://doi.org/10.1016/j.atmosenv.2009.01.051.
Holmes, C. D., D. J. Jacob, E. S. Corbitt, J. Mao, X. Yang, R. Talbot, and F. Slemr, 2010: Global atmospheric model for mercury including oxidation by bromine atoms. Atmos. Chem. Phys., 10, 12 037–12 057, https://doi.org/10.5194/acp-10-12037-2010.
Holtslag, A. A. M., and Coauthors, 2013: Stable atmospheric boundary layers and diurnal cycles: Challenges for weather and climate models. Bull. Amer. Meteor. Soc., 94, 1691–1706, https://doi.org/10.1175/BAMS-D-11-00187.1.
Horowitz, H. M., and Coauthors, 2017: A new mechanism for atmospheric mercury redox chemistry: Implications for the global mercury budget. Atmos. Chem. Phys., 17, 6353–6371, https://doi.org/10.5194/acp-17-6353-2017.
Huang, J., and M. S. Gustin, 2015: Uncertainties of gaseous oxidized mercury measurements using KCl-coated denuders, cation-exchange membranes, and nylon membranes: Humidity influences. Environ. Sci. Technol., 49, 6102–6108, https://doi.org/10.1021/acs.est.5b00098.
Hynes, A. J., D. L. Donohoue, M. E. Goodsite, and I. M. Hedgecock, 2009: Our current understanding of major chemical and physical processes affecting mercury dynamics in the atmosphere and at the air-water/terrestrial interfaces. Mercury Fate and Transport in the Global Atmosphere, N. Pirrone and R. Mason, Eds., Springer Science and Business Media, 427–457.
Jiao, Y., and T. S. Dibble, 2017: First kinetic study of the atmospherically important reactions BrHg· + NO2 and BrHg· + HOO. Phys. Chem. Chem. Phys., 19, 1826–1838, https://doi.org/10.1039/C6CP06276H.
KPP, 2018: KPP source code release 2.2.3. Virginia Polytechnic Institute and State University, accessed 15 June 2018, http://people.cs.vt.edu/asandu/Software/Kpp/.
Lam, K. T., C. J. Wilhelmsen, A. C. Schwid, Y. Jiao, and T. S. Dibble, 2019: Computational study on the photolysis of BrHgONO and the reactions of BrHgO• with CH4, C2H6, NO, and NO2: Implications for formation of Hg(II) compounds in the atmosphere. J. Phys. Chem., 123A, 1637–1647, https://doi.org/10.1021/acs.jpca.8b11216.
Lu, J. Y., and Coauthors, 2001: Magnification of atmospheric mercury deposition to polar regions in springtime: The link to tropospheric ozone depletion chemistry. Geophys. Res. Lett., 28, 3219–3222, https://doi.org/10.1029/2000GL012603.
Lyman, S. N., D. A. Jaffe, and M. S. Gustin, 2010: Release of mercury halides from KCl denuders in the presence of ozone. Atmos. Chem. Phys., 10, 8197–8204, https://doi.org/10.5194/acp-10-8197-2010.
Mao, H., R. W. Talbot, B. C. Sive, S. Y. Kim, D. R. Blake, and A. J. Weinheimer, 2010: Arctic mercury depletion and its quantitative link with halogens. J. Atmos. Chem., 65, 145–170, https://doi.org/10.1007/s10874-011-9186-1.
NADP, 2017: Atmospheric Mercury Network. Accessed 4 April 2017, http://nadp.slh.wisc.edu/data/AMNet/.
NCAR, 2018: Tropospheric Ultraviolet and Visible Radiation (TUV) model 5.3.1. Accessed 20 June 2018, https://www2.acom.ucar.edu/modeling/tropospheric-ultraviolet-and-visible-tuv-radiation-model.
Obrist, D., E. Tas, M. Peleg, V. Matveev, X. Faïn, D. Asaf, and M. Luria, 2011: Bromine-induced oxidation of mercury in the mid-latitude atmosphere. Nat. Geosci., 4, 22–26, https://doi.org/10.1038/ngeo1018.
Obrist, D., and Coauthors, 2017: Tundra uptake of atmospheric elemental mercury drives Arctic mercury pollution. Nature, 547, 201–204, https://doi.org/10.1038/nature22997.
Pal, B., and P. A. Ariya, 2004a: Studies of ozone initiated reactions of gaseous mercury: Kinetics, product studies, and atmospheric implications. Phys. Chem. Chem. Phys., 6, 572–579, https://doi.org/10.1039/b311150d.
Pal, B., and P. A. Ariya, 2004b: Studies of ozone initiated reactions of gaseous mercury: Kinetics, product studies, and atmospheric implications. Phys. Chem. Chem. Phys., 6, 572–579, https://doi.org/10.1039/b311150d.
Parrella, J. P., and Coauthors, 2012: Tropospheric bromine chemistry: Implications for present and pre-industrial ozone and mercury. Atmos. Chem. Phys., 12, 6723–6740, https://doi.org/10.5194/acp-12-6723-2012.
Rolph, G., A. Stein, and B. Stunder, 2017: Real-time Environmental Applications and Display System: READY. Environ. Modell. Software, 95, 210–228, https://doi.org/10.1016/j.envsoft.2017.06.025.
Saiz-Lopez, A., and R. P. Fernandez, 2016: On the formation of tropical rings of atomic halogens: Causes and implications. Geophys. Res. Lett., 43, 2928–2935, https://doi.org/10.1002/2015GL067608.
Saiz-Lopez, A., and Coauthors, 2018: Photoreduction of gaseous oxidized mercury changes global atmospheric mercury speciation, transport and deposition. Nat. Commun., 9, 4796, https://doi.org/10.1038/s41467-018-07075-3.
Sander, S. P., and Coauthors, 2011: Chemical kinetics and photochemical data for use in atmospheric studies. JPL Publ. 10-6, Evaluation 17, 523 pp., http://jpldataeval.jpl.nasa.gov.
Sandu, A., D. Daescu, and G. R. Carmichael, 2003: Direct and adjoint sensitivity analysis of chemical kinetic systems with KPP: I—Theory and software tools. Atmos. Environ., 37, 5083–5096, https://doi.org/10.1016/j.atmosenv.2003.08.019.
Schmidt, J. A., and Coauthors, 2016: Modeling the observed tropospheric BrO background: Importance of multiphase chemistry and implications for ozone, OH, and mercury. J. Geophys. Res. Atmos., 121, 11 819–11 835, https://doi.org/10.1002/2015JD024229.
Shah, V., and L. Jaeglé, 2017: Subtropical subsidence and surface deposition of oxidized mercury produced in the free troposphere. Atmos. Chem. Phys., 17, 8999–9017, https://doi.org/10.5194/acp-17-8999-2017.
Shah, V., and Coauthors, 2016: Origin of oxidized mercury in the summertime free troposphere over the southeastern US. Atmos. Chem. Phys., 16, 1511–1530, https://doi.org/10.5194/acp-16-1511-2016.
Snider, G., F. Raofie, and P. A. Ariya, 2008: Effects of relative humidity and CO(g) on the O3-initiated oxidation reaction of Hg0(g): Kinetic and product studies. Phys. Chem. Chem. Phys., 10, 5616–5623, https://doi.org/10.1039/b801226a.
Sommar, J., K. Gårdfeldt, D. Strömberg, and X. Feng, 2001: A kinetic study of the gas-phase reaction between the hydroxyl radical and atomic mercury. Atmos. Environ., 35, 3049–3054, https://doi.org/10.1016/S1352-2310(01)00108-X.
Song, S., and Coauthors, 2018: Understanding mercury oxidation and air–snow exchange on the East Antarctic Plateau: A modeling study. Atmos. Chem. Phys., 18, 15 825–15 840, https://doi.org/10.5194/acp-18-15825-2018.
Stein, A. F., R. R. Draxler, G. D. Rolph, B. J. B. Stunder, M. D. Cohen, and F. Ngan, 2015: NOAA’s HYSPLIT atmospheric transport and dispersion modeling system. Bull. Amer. Meteor. Soc., 96, 2059–2077, https://doi.org/10.1175/BAMS-D-14-00110.1.
Stephens, C. R., and Coauthors, 2012: The relative importance of chlorine and bromine radicals in the oxidation of atmospheric mercury at Barrow, Alaska. J. Geophys. Res., 117, D00R11, https://doi.org/10.1029/2011JD016649.
Subir, M., P. A. Ariya, and A. P. Dastoor, 2011: A review of uncertainties in atmospheric modeling of mercury chemistry I. Uncertainties in existing kinetic parameters—Fundamental limitations and the importance of heterogeneous chemistry. Atmos. Environ., 45, 5664–5676, https://doi.org/10.1016/j.atmosenv.2011.04.046.
Tan, H., M. Cai, Q. Fan, L. Liu, F. Li, P. W. Chan, X. Deng, and D. Wu, 2017: An analysis of aerosol liquid water content and related impact factors in Pearl River delta. Sci. Total Environ., 579, 1822–1830, https://doi.org/10.1016/j.scitotenv.2016.11.167.
Tokos, J. J. S., B. Hall, J. A. Calhoun, and E. M. Prestbo, 1998: Homogeneous gas-phase reaction of Hg° with H2O2, 03, CH3I, and (CH3)2S: Implications for atmospheric Hg cycling. Atmos. Environ., 32, 823–827, https://doi.org/10.1016/S1352-2310(97)00171-4.
Toyota, K., A. P. Dastoor, and A. Ryzhkov, 2014: Air–snowpack exchange of bromine, ozone and mercury in the springtime Arctic simulated by the 1-D model PHANTAS—Part 2: Mercury and its speciation. Atmos. Chem. Phys., 14, 4135–4167, https://doi.org/10.5194/acp-14-4135-2014.
Travnikov, O., and Coauthors, 2017: Multi-model study of mercury dispersion in the atmosphere: Atmospheric processes and model evaluation. Atmos. Chem. Phys., 17, 5271–5295, https://doi.org/10.5194/acp-17-5271-2017.
Wang, F., A. Saiz-Lopez, A. S. Mahajan, J. C. Gómez Martín, D. Armstrong, M. Lemes, T. Hay, and C. Prados-Roman, 2014: Enhanced production of oxidised mercury over the tropical Pacific Ocean: A key missing oxidation pathway. Atmos. Chem. Phys., 14, 1323–1335, https://doi.org/10.5194/acp-14-1323-2014.
Wang, S., and Coauthors, 2019: Direct detection of atmospheric atomic bromine leading to mercury and ozone depletion. Proc. Natl. Acad. Sci. USA, 116, 14 479–14 484, https://doi.org/10.1073/pnas.1900613116.
Weiss-Penzias, P. S., M. S. Gustin, and S. N. Lyman, 2011: Source of gaseous oxidized mercury and mercury dry deposition at two southeastern U.S. sites. Atmos. Environ., 45, 4569–4579, https://doi.org/10.1016/j.atmosenv.2011.05.069.
Weiss-Penzias, P. S., H. M. Amos, N. E. Selin, M. S. Gustin, D. A. Jaffe, D. Obrist, G.-R. Sheu, and A. Giang, 2015: Use of a global model to understand speciated atmospheric mercury observations at five high-elevation sites. Atmos. Chem. Phys., 15, 1161–1173, https://doi.org/10.5194/acp-15-1161-2015.
Wright, L. P., L. Zhang, I. Cheng, J. Aherne, and G. R. Wentworth, 2018: Impacts and effects indicators of atmospheric deposition of major pollutants to various ecosystems—A review. Aerosol Air Qual. Res., 18, 1953–1992, https://doi.org/10.4209/aaqr.2018.03.0107.
Ye, Z., H. Mao, C.-J. Lin, and S. Y. Kim, 2016: Investigation of processes controlling summertime gaseous elemental mercury oxidation at midlatitudinal marine, coastal, and inland sites. Atmos. Chem. Phys., 16, 8461–8478, https://doi.org/10.5194/acp-16-8461-2016.
Ye, Z., H. Mao, C. T. Driscoll, Y. Wang, Y. Zhang, and L. Jaeglé, 2018: Evaluation of CMAQ coupled with a state-of-the-art mercury chemical mechanism (CMAQ-newHg-Br). J. Adv. Model. Earth Syst., 10, 668–690, https://doi.org/10.1002/2017MS001161.
Zhang, L., L. P. Wright, and P. Blanchard, 2009: A review of current knowledge concerning dry deposition of atmospheric mercury. Atmos. Environ., 43, 5853–5864, https://doi.org/10.1016/j.atmosenv.2009.08.019.