Inﬂuence of the Background Nitrogen Oxides on the Tropospheric Ozone Depletion Events in the Arctic during Springtime

: Ozone depletion events (ODEs) in the springtime of the Arctic have been frequently observed since the early 1980s, and the correlation between the ozone mixing ratio during the ODEs and the nitrogen oxides (NO x ) concentration is still unclear. In the present study, the role of the background level of NO x in ODEs was investigated by using a box model implementing a chemical reaction mechanism containing 49 chemical species and 141 related reactions. A concentration sensitivity analysis was also applied to discover the dependence of the ozone mixing ratio during the ODEs on each constituent of the initial air composition. The simulation results showed that a critical value of the NO x background level exists, with which the ozone depletion rate is independent of the initial concentration of NO x , and the critical value was found to be approximately 55 ppt (ppt = part per trillion, 10 − 12 mol/mol) in the present study. The concentration sensitivity analysis also showed that the existence of NO x has a two-sided impact on the depletion of ozone, depending on the initial amount of NO x . With a low background level of NO x (less than 55 ppt), the increase of the initial NO x can advance the ozone depletion. On the contrary, with a high initial NO x level (more than 55 ppt), NO x would delay the consumption of ozone during the ODEs.


Introduction
Ozone is a trace gas of great significance in the atmosphere, which has an important influence on the global climate and environment [1][2][3]. It is a key factor affecting the dynamics, thermodynamics, radiation and chemical processes in the stratosphere and the troposphere. First of all, by absorbing ultraviolet (UV) radiation from the sun, ozone becomes the major heat source in the stratosphere and protects lives on Earth from harmful UV radiation [4,5]. Second, ozone is a strong oxidant, and its photolysis in the troposphere is the major source of OH radicals in the atmosphere. Thus, the distribution and the change of ozone directly affect the concentration and lifetime of other chemical species and OH radicals. Moreover, the tropospheric ozone is a pollutant, which can cause eye irritation and respiratory diseases [6][7][8]. The major sources of the tropospheric ozone are automobile emissions, power plants using fossil fuel and other human activities [9,10]. The mixing ratio of ozone in the troposphere has doubled since the early 20th century because of the increasing man-made emissions of nitrogen oxides and volatile organic compounds (VOCs) [11].
The depletion of ozone is always a core problem in many fields. There are different types of ozone depletion events in the stratosphere and the troposphere [12][13][14], among which the Antarctic ozone hole, discovered by Farman et al. [15], is the most-commonly ozone depletion phenomenon in the stratosphere. In the present study, we focused on tropospheric ozone depletion events (ODEs) occurring in polar regions. ODEs in the polar troposphere refer to a phenomenon where the amount of ozone in the boundary layer of polar regions decreases rapidly from a background level (∼40 ppb, ppb = part per billion, 10 −9 mol/mol) to less than 1 ppb in a few days. These ODEs occur mostly in the spring of polar regions, from March to May in the Arctic [16] and from August to October in the Antarctic [17], when they are favored by light and low temperature conditions [18].The occurrence frequency of the ODEs also shows great variations between different years, and the inter-annual variability is partly attributable to the influence brought about by the western Pacific teleconnection pattern, especially in the 2000s [19].
Because ozone is a precursor of many atmospheric oxidants, the tropospheric ODEs exert a strong influence on the oxidation capacity of the atmosphere. During ODEs, the oxidation ability originally dominated by ozone is weakened, while the halogen species controls the oxidation ability [13]. This alters the reaction cycles and final products of many atmospheric reactions. For example, during ODEs, the enhanced halogen chemistry can effectively oxidize reactive gaseous mercury (RGM), making it settle from the atmosphere into snow more easily. After that, due to the drift of glaciers and the melt of the snow, the bio-availability of mercury is enhanced [20].

Discovery of ODEs
In the spring of 1978, Oltmans [21] observed a phenomenon at Barrow, Alaska that the amount of ozone in the boundary layer rapidly decreased from the background level (∼40 ppb) to less than 1 ppb in a few days. The low concentration of ozone also lasted for several days until the local meteorological conditions significantly changed. Then in 1986, Bottenheim et al. [16] observed an analogous phenomenon at Alert, Canada, confirming the findings of Oltmans. They also found a strong negative correlation between the ozone mixing ratio and the bromine concentration during ODEs, which highlights the importance of bromine-related species for the destruction of ozone.
The correlation between the bromine species and the ozone destruction remained unclear until 1994. Hönninger and Platt [22] gave an explanation for the increase of bromine concentration in ODEs. They proposed that bromine atoms react with ozone, forming BrO as and the self-reactions of BrO are able to form bromine atoms or molecular bromine through Under light conditions, bromine molecules photolyze and form bromine atoms again, continuously consuming ozone in the troposphere via In the series of (R1)-(R3), bromine acts as a catalyst. Apart from these reactions, BrO can also be oxidized by free radicals in the atmosphere and form gas-phase hypobromous acid (HOBr) through BrO + HO 2 → HOBr + O 2 . (R4) A following heterogeneous process with the participation of HOBr can activate bromide from various polar substrates such as the suspended aerosols and ice/snow-covered surfaces: leading to an explosive growth of the total bromine amount in the atmosphere and a rapid reduction of ozone in the boundary layer. This process is thus called "bromine explosion" mechanism [23][24][25]. As mentioned above, the tropospheric ODEs are favored under low temperature conditions [26]. However, the dependence of the ODEs on the temperature has not been fully clarified [27][28][29]. Lehrer et al. [18] and Koo et al. [29] proposed that a very strong temperature inversion is essential to the occurrence of ODEs because a stable boundary layer tends to be formed, leading to an accumulation of bromine and thus the destruction of ozone. Moreover, it was also suggested that wind shear [30,31], passage of lows [28], and a downward diffusion of warm air caused by the existence of clouds [32,33] may contribute to the termination of the ODEs. In general, the atmospheric conditions in which ODEs could occur in polar regions are still not clear.

Role of Nitrogen Oxides in ODEs
Investigations have been carried out into the relevant chemical reactions of gaseous halogens in the Arctic region for decades. However, the role of nitrogen oxides in the halogen cycle during ODEs is not relatively well-known.
The sources of nitrogen oxides in the Arctic are different from those in other regions, because the Arctic is sparsely populated and has few human activities. The typical NO x background level in the Arctic boundary layer ranges from 10 ppt to 100 ppt, mainly due to photochemical products of snow [34], long-distance transport of organic nitrates from cities nearby [35], and emissions from international shipping with the thawing of the Northwest passage during warm seasons [36]. Thus, human-induced climate change will affect the amount of snow and ice cover in the Arctic, leading to a change in the intensity of nitrogen oxide emissions. In addition, oil exploitation and transportation activities in surrounding cities will increase the NO x transported to the Arctic, affecting the NO x level through atmospheric circulation.
Nitrogen oxides play an important role in the recycling of active free radicals (such as reactive halogens) in the atmosphere and indirectly affect ozone depletion. It controls the production of ozone through the titration reaction cycle: Nitrogen oxides can also react with halogens and are essential for the conversion between bromine compounds, and the ultimate sink for NO x is nitric acid (HNO 3 ): The abbreviation "mp" in (R11) denotes a multiphase reaction. Since the early 1990s, box model calculations have shown that NO x is quickly lost during ODEs [37,38]. However, the correlation between the concentrate change of nitrogen oxides and ozone is relatively weak [39]. Analysis of TOPSE observation data in 2003 [40] shows that, in the early stage of ozone destruction, NO x concentration decreases rapidly. However, when O 3 dropped below 10 ppb, NO x concentration recovered. Evans et al. [40] simulated this situation using a 0-D photochemical model, and attributed the initial NO x consumption to the hydrolysis of BrONO 2 . They suggested that with the continuous decrease of O 3 , the ratio of Br/BrO and NO/NO 2 increases. As a result, the BrONO 2 formation through (R10) weakens, leading to a rate decrease of (R11) and (R12), which are sinks of NO x . Then the release of NO x from the snow leads to a recovery of NO x in the ambient air. Thus, the intensity of this NO x recovery depends on the strength of the surface source, and thus may vary geographically and seasonally. The depletion of ozone also affects the life cycle of NO x and HO x radicals due to the enhancement of the bromine species. In 2010, Bloss et al. [41] used a box model to replicate the diurnal variation of NO and NO 2 concentrations. They found that the halogen species will increase the NO 2 /NO ratio, and substantially reduce the NO x lifetime through heterogeneous loss of the halogen nitrates.
The role of nitrogen oxides in the tropospheric ODEs is also unclear. Some researchers proposed that the presence of NO x in the polar troposphere above a certain mixing ratio promotes the formation of ozone [42,43]. However, Cao et al. [44] studied the relationship between halogen release and ozone depletion by using a box model. They found that when the initial concentration of NO x in the Arctic troposphere is less than 15 ppt, the presence of NO x has a positive correlation with ozone depletion, which is influenced by the absorption coefficient of BrONO 2 hydrolysis. It means that an enhancement of the nitrogen oxides in a low-NO x situation would cause a speed-up of the ODEs [44]. In contrast to that, Custard et al. [45] made simulations with two different NO x concentration ranges, representing cleaning conditions (50-100 ppt NO x ) and pollution conditions (700-1600 ppt NO x ), respectively. Their modeling results suggest that high concentrations of NO x may delay ozone depletion. They explained that, on one hand, NO x participation would terminate the halogen radical chain reaction, which would prohibit the ozone destruction. On the other hand, the process can also lead to the formation of bromine explosion products and thus the advancement of ODEs. The relative importance of these two mechanisms depends on the sensitivities of the related reactions to NO x and the rate of each reaction.
Therefore, in this study, the role of nitrogen oxides in ODEs during the springtime of the Arctic is further investigated by using a box model. We applied a concentration sensitivity analysis to investigate how the nitrogen oxides influence the evolution of ozone during ODEs and explore the controversial role of nitrogen oxides in the halogen cycle.

Mathematical Model and Methods
In the present study, a box model developed by Cao et al. [44] was adopted. In this model, a halogen-related chemical mechanism was implemented into a software KINAL (KInetic aNALysis of reaction mechanics [46]). Moreover, a strictly defined air composition was used as the initial condition of the model. Then a following concentration sensitivity analysis was performed on the chemical mechanism.
In the chemical reaction mechanism (141 reactions and 49 species) used in this study, the initial air composition used in the model represents a typical Arctic atmosphere (see Table 1) [44], including a small amount of bromine and chlorine substances to trigger the bromine explosion mechanism. Compounds released from the underlying surface such as NO x , HONO and HCHO are also included ( Table 2) [47][48][49][50]. The temporal variation of the species concentration is conveyed as follows: where c i denotes the concentration of the i-th species with the initial value of c 0 . k j in Equation (1) represents the rate constant of the j-th reaction, and t denotes time. E i is the emission rate of the i-th species from the underlying surface. In Equation (1), the term in the left side denotes the concentration change with time, and the right items represent the chemical production/consumption and the emission, respectively. Thus, the temporal evolution of species is determined by the net effect of chemical reactions and the local emissions from the surface. By solving the stiff ordinary differential Equation (1) using KINAL [46], the temporal evolution of ozone, principal bromine containing species and nitrogen related substances in the ODEs can be obtained. The reaction rate constants of gas-phase reactions are calculated based on the Arrhenius equation, assuming a constant temperature of 258K [18], which represents a typical Arctic temperature. A listing of all the reactions and the rate constants is given in the Appendix A. Table 1. Initial atmospheric composition used in the present model study, adopted from Cao et al. [44] (ppm = part per million, 10 −6 mol/mol). Apart from the gas-phase reactions discussed above, during the ODEs, there are several multiphase reactions occurring on the surface of the substrates including the suspended aerosols and the ice-/snow-covered surfaces. The estimation of the heterogeneous reaction rates is associated with the properties of the suspended aerosols in the atmosphere and the local meteorological conditions such as the stability of the boundary layer and the wind speed [51]. For the heterogeneous reactions occurring in the aerosols, the reaction rate constant can be expressed as [52]:

Species
where r is the typical aerosol radius in polar conditions and is estimated as 0.3 µm in this model. D g in Equation (2) denotes the diffusivity of gas-phase molecules and here it values 0.2 cm 2 ·s −1 , V therm represents the average speed of molecular thermal motion. γ means the uptake coefficient of HOBr or BrONO 2 for the heterogeneous reactions occurring on the surface of aerosols. For BrONO 2 , γ in Equation (2) takes 0.06 [53], while for the uptake of HOBr in the acidic condition, the value of γ is estimated according to the approach of Hanson et al. [54]. The item α eff in Equation (2) denotes the surface area density of the suspended aerosols. V aerosol represents the aerosol volume density, and under the condition of Arctic region in springtime it takes 10 −5 cm 3 ·s −3 as a typical value. Thus, the surface area density for aerosols α eff is calculated as 10 −4 m −1 [55]. The parameterization scheme for the heterogeneous reactions occurring on ice or snow refers to previous model studies [18,44,51], in which the wind speed is set to 8 m·s −1 [56] and the roughness of the reaction surface is assumed as 10 −5 m [57]. The uptake coefficient γ for the heterogeneous reactions of HOBr and BrONO 2 on the ice or snow surface is assumed as 0.06 [53]. Besides, the height of the boundary layer and the surface layer are assumed as 200 m and 20 m, respectively for the typical height observed in springtime of Arctic [56].
In order to investigate the role of nitrogen oxides in the ODEs, different initial values of NO x (from 3 ppt to 450 ppt) were tested in the present study. When the value of NO x concentration was modified, the model outputs the corresponding temporal change of ozone and principal bromine species for comparison.
Afterward, we performed a concentration sensitivity analysis on the present reaction mechanism to identify the relative importance of each component in the initial air composition during ODEs by showing the concentration sensitivitỹ The sensitivity coefficientS ij is a dimensionless parameter representing the percentage change of the i-th species concentration due to 1 percent variation of the j-th species. C i (t) is the concentration of the i-th chemical species at the time t, and C 0 j denotes the initial concentration of the j-th species. The relative concentration sensitivity is computed in KINAL by using a Decomposed Direct Method [58], which has been proven robust and highly efficient [59].

Results and Discussion
In this section, we first show the temporal evolution of the principal chemical species such as ozone, bromine species and nitrogen oxides in the 200 m boundary layer during the ODEs. Then, the evolution of ozone and principal bromine species with different initial NO x concentrations is presented. At last, the relative concentration sensitivities of ozone to the initial air composition under different initial NO x conditions are given and discussed. Figure 1 shows the temporal evolution of ozone and principal bromine containing species. It can be seen that before day 3, ozone is hardly consumed and maintains the background level of about 40 ppb, and the average depletion rate during these 3 days is about 0.01 ppbv·h −1 . After day 3, ozone begins to deplete rapidly and decreases below 0.05 ppb within two days. At the same time, both the mixing ratios of BrO and HOBr increase exponentially. They construct the major bromine amount in this period, reaching a peak value of 50 ppt and 30 ppt, respectively. These values are in consistence with the findings of Liao et al. [60], who reported an average value of BrO within a range of 20-60 ppt and HOBr approximately 20-30 ppt during ODEs. When the ozone is consumed completely, BrO and HOBr decrease sharply to near-zero values, while atomic Br begins to accumulate in the atmosphere from day 5 and then reaches a maximum value of about 80 ppt. In the last period, bromine is nearly entirely converted to HBr by the absorption of aldehydes such as HCHO and CH 3 CHO in the atmosphere, which has been confirmed by the measurements of Langendörfer et al. [61].

Temporal Evolution of Nitrogen Oxides
We also show the evolution of nitrogen containing species during ODEs in Figure 2. It is seen that the concentrations of NO and NO 2 drop sharply in the first two days, while HNO 3 , HONO and PAN (peroxyacetyl nitrate) increase steadily in the simulated ten days due to the conversions: This enhancement of PAN during the ODEs was also observed by Bottenheim et al. [16] in measurements conducted at Alert, Canada, for nitrogen containing compounds. It is also interesting to note that the concentration of HNO 4 peaks at the beginning of the reaction, then drops to a near-zero value in two days and remains a low value until the end of the ozone depletion simulation. This is because that at the beginning of the simulation, HNO 4 is generated by and then quickly decomposed via reactions:

Influence of the NO x Initial Concentration
To investigate the influence might be caused by the change of the initial concentration of NO x on the ODEs, we varied the initial amount of NO x from 3 ppt to 450 ppt. Figure 3 shows the evolution of ozone with different initial NO x concentrations. It is shown that when the initial concentration of NO x increases from 3 ppt to around 55 ppt, the onset of ozone depletion is advanced. However, when the initial NO x reaches approximately 55 ppt, the existence of the nitrogen oxides exerts a minor influence on the occurrence of ODEs. After that, when the initial NO x keeps increasing, the onset of the ozone depletion is retarded. It is seen the ozone depletion rate in the 150 ppt NO x simulation is nearly identical to that in the 15 ppt NO x case, and the 300 ppt case is similar to the 3 ppt case. It indicates that when the initial NO x concentration exceeds a critical value (i.e., about 55 ppt in this study), a higher NO x initial value tends to retard the ODEs. It is also interesting to note that, under the situations of the initial 150 ppt, 300 ppt and 450 ppt NO x , during the beginning period of ODEs (i.e., the ozone depletion rate is less than 0.1 ppb h −1 [44]), the amount of ozone even increases. This is caused by the high background level of NO x , leading to a rapid ozone formation before the consumption caused by bromine. Figure 4 shows the evolution of Br and BrO under the situations of the initial NO x of 15 ppt, 55 ppt and 150 ppt, respectively, representing the NO x concentration with low, middle and high initial values. By doing that, we can observe the influence of the initial concentration of NO x on the bromine-containing compounds during the ozone depletion. It is seen in Figure 4 that the increase of the initial concentration of NO x from 15 ppt to 55 ppt promotes the occurrence of the bromine explosion and thus enhances the rate of ozone depletion. Conversely, when NO x concentration increases from 55 ppt to 150 ppt, the production of bromine species is delayed, thus reducing the ozone depletion rate. However, the change in the NO x initial concentration has no significant impact on the peak value of bromine species.
Our findings can be compared with those of Tas et al. [62]. In their study, based on extensive field measurements [63] performed in the Dead Sea area and a box model simulation, Tas et al. [62] investigated the connection between the level of NO 2 and the activity of reactive bromine species during the destruction of ozone. They found that the anthropogenic emission of nitrogen oxides may cause an enhancement of the bromine activity under a low nitrogen condition. However, if the level of NO 2 exceeds a certain threshold, a further increase in the NO 2 concentration leads to a decrease of the bromine concentration. Because the destruction of ozone in the Dead Sea area investigated by Tas et al. [62] is also caused by the bromine related chemistry, which is similar to ODEs, considering the negative dependence of the ozone mixing ratio on the bromine species, the findings of our study are qualitatively in agreement with those of Tas et al. [62].

Concentration Sensitivity Analysis of Ozone and Bromine Containing Compounds during the ODEs
In this section, a concentration sensitivity analysis is applied on the present reaction mechanism for the time period during the ODEs. Figure 5 displays the relative concentration sensitivities of ozone on day 4.5 (noon of the fourth day), which resides in the most intense period of the ozone depletion under the situation of a 15 ppt initial NO x . It is seen that for the ozone mixing ratio, the sign of the sensitivity coefficient corresponding to the initial nitrogen oxide is negative. This means that an increase of the initial concentration of NO x would reduce the ozone mixing ratio during ODEs in the case of 15 ppt. As a result, the depletion is promoted due to the increase of the initial NO x . This is mainly due to the hydrolysis of BrONO 2 via (R11), which increases the production of HOBr by (R10), leading to a further ozone depletion through the heterogeneous bromide activation: For situations with an initial NO x concentration of 55 ppt and 150 ppt, the relative concentration sensitivities of ozone are displayed in Figures 6 and 7, respectively. Compared to the results shown in Figure 5, the sensitivity coefficients of ozone corresponding to the initial nitrogen oxides are close to zero in the case of 55 ppt, while in the case of 150 ppt they are positive. It can be concluded that, for different initial nitrogen oxide concentrations, NO x concentration has different impacts on ozone depletion. With a low background level of NO x , the increase of NO x can promote the ozone depletion. On the contrary, with higher level of NO x , the existence of NO x will prohibit the ozone depletion. This is attributed to the formation of ozone through the reaction cycle between VOCs and NO x with the presence of sunlight: By analogy to CH 4 , reaction of OH with many hydrocarbons (RH) leads to the formation of alkyl peroxy radicals (RO 2 ) through (R19). The formed alkyl peroxy radicals are able to oxidize NO to NO 2 , which is able to compensate ozone through the following photolysis of NO.  Figure 8 compares the time variation of the relative concentration sensitivities of ozone to the initial NO x under different initial NO x amounts. It can be seen that in the case of 55 ppt initial NO x , the sensitivity coefficients of ozone to the initial nitrogen oxides are nearly identical to zero. Thus, a critical value of the NO x background level exists, with which the ozone depletion rate is independent of the initial value of NO x , and the critical value is found approximated 55 ppt in the present study.

Conclusions and Future Work
In this study, the role of NO x in ODEs during springtime in the Arctic was further investigated systematically through a model study by implementing a sensitivity analysis. The timescale of the ozone depletion, as well as the temporal evolution of ozone and bromine containing compounds in the presence of different initial NO x , were also analyzed.
It was found in the present study that when the initial concentration of NO x in the Arctic troposphere is less than 55 ppt, the increase of the NO x background level apparently strengthens the consumption of ozone and the enhancement of halogen. In addition, the onset of the ozone depletion is also advanced. In contrast to that, when the initial NO x concentration is higher than 55 ppt, the ODEs are retarded and the delayed time positively depends on the initial NO x concentration.
The concentration sensitivity analysis also confirms the role of NO x in the ODEs in different scenarios. With a relatively low background level of NO x , the sensitivity of ozone to the initial NO x keeps positive, denoting a positive correlation of the ozone concentration with the initial NO x value during ODEs. In contrast to that, with a high background level of NO x , ozone possesses a negative sensitivity to the initial NO x , representing a delay of ODEs caused by the increase of the initial NO x .
The critical value of the NO x initial concentration separating these two situations was found to be approximately 55 ppt.
The present study still has a few deficiencies to be improved. In terms of the model, the temporal and the spatial variations of the emissions from underlying surface are not taken into account, while the emission rates of the species such as HONO and NO x from the underlying ice-/snow-covered surface should vary during the daytime and nighttime. It would also be beneficial to extend the model to other conditions by modifying the values of the meteorological parameters, such as the wind speed and the temperature, because the variation of these factors alters the chemical reaction rates, which might influence the importance of each reaction in the ODEs as well as the critical turning value of the initial NO x obtained in the present study. In addition, the diurnal variation of the photolysis reactions was not considered in the present study. It is also of great significance to improve the present model by making a comparison with observational data. The research in the future will not only be restricted to Arctic regions, it could also be extended to other related issues such as the regional ozone pollution episodes.

Acknowledgments:
The authors like to thank Liting Liu in School of Atmospheric Physics, NUIST for her fundamental work on this topic. We also wish to thank three anonymous reviewers and the editor for their perspicacious comments that significantly improved our paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A   Table A1. The complete list of chemical reactions used in a present study, a constant temperature T = 258 K is implemented, and the rate of third-body reactions is estimated as   Atkinson et al. [64]