Inﬂuence of the Change in Total Ozone Column (TOC) on the Occurrence of Tropospheric Ozone Depletion Events (ODEs) in the Antarctic

. The occurrence of the tropospheric ozone depletion events (ODEs) in the Antarctic can be inﬂuenced by the change in Total Ozone Column (TOC). In this study, we combined the observational data obtained from ground observation stations with two numerical models (TUV and KINAL), to ﬁgure out the relationship between the TOC change and the occurrence of ODEs in the Antarctic. A sensitivity analysis was also performed on the change in ozone and major bromine species (BrO, HOBr and HBr) to ﬁnd out key photolysis reactions determining the impact on the occurrence of tropospheric ODEs brought 5 by the change in TOC. From the analysis of the observational data and the numerical results, we suggested that the occurrence frequency of ODEs in the Antarctic seems negatively correlated with the variation of TOC. Moreover, major ODE accelerating reactions (i.e. photolysis of ozone, H 2 O 2 and HCHO) and decelerating reactions (i.e. photolysis of BrO and HOBr), which heavily control the start of ODEs, were also identiﬁed. It was found that when TOC varies, the major ODE accelerating reactions speed up signiﬁcantly, while major ODE decelerating reactions are only slightly affected, thus leading to the negative 10 dependence of the ODE occurrence on the change in TOC.


Introduction
Ozone is a short-lived trace gas in the atmosphere, with about 90% located in the stratosphere and 10% in the troposphere (Seinfeld and Pandis, 2006;Akimoto, 2016). In the stratosphere, ozone plays a role in absorbing the ultraviolet (UV) radiation from the sun, thus protecting the lives on the earth. In contrast, ozone in the troposphere is a pollutant. It causes eye irrita-15 tions and disorders of the lung function of human beings at a high concentration (Lippmann, 1991) Moreover, ozone in the troposphere also acts as a greenhouse gas, contributing to the global warming (Seinfeld and Pandis, 2006). It was suggested by Toumi et al. (1996) that the tropospheric ozone mostly originates from three sources: downward entrainment from the strato-over the time period . They suggested that a low temperature (<-20 • C) is probably a necessary condition for the occurrence of ODEs, because heterogeneous reactions that activate bromide from substrates and the formation of frost flowers 50 are favored under this cold condition. However, in a later analysis of ozone data obtained from a transpolar drift, Bottenheim et al. (2009) found the temperature well above -20 • C during the most persistent ozone depletion period over the Arctic Ocean.
It was also reported by Koo et al. (2012) that there is no evidence in observations for the threshold value of temperature for the occurrence of ODEs. Instead, they suggested the variability of temperature potentially an important factor for the depletion of ozone.
(2) Passing of pressure systems. By analyzing the values of ozone mixing ratio and meteorological parameters from 55 balloon sondes during the 1994 Polar Sunrise Experiment (PSE94), Hopper et al. (1998) suggested that the occurrence of ODEs in the Arctic is strongly correlated with high-pressure systems. This dependence of the ozone decline on pressure systems was also confirmed by Jacobi et al. (2010) who proposed that mesoscale synoptic systems are able to transport air masses with low ozone mixing ratio to the observational site, leading to the detection of ODEs at Arctic coastal stations. It was also suggested by Boylan et al. (2014) that the transport caused by synoptic patterns acts as the major factor for the occurrence of ODEs at Barrow,60 Alaska, rather than the change in local meteorological parameters. In contrast, Jones et al. (2006) analyzed the observational data of ozone and meteorological parameters obtained at Halley station in coastal Antarctica, and they figured out that in Antarctica, the occurrence of ODEs is highly associated with low pressure systems, denoting the remarkable differences in the atmospheric system between the Arctic and the Antarctic. (3) Formation of fresh sea ice. Based on sea ice maps obtained from satellite detection, Bottenheim et al. (2009) figured out regions of the Arctic Ocean as the origin of the tropospheric ozone depletion. In these regions, open leads, polynyas and fresh sea ice are frequently formed, which favors the release of bromine and thus the depletion of ozone. These ODE-originating regions proposed by Bottenheim et al. (2009) are also consistent with the "cold spots" discovered in a previous study of Bottenheim and Chan (2006) where the depletion of ozone possibly initiates and develops. The connection between the ODE occurrence and the formation of fresh sea ice was also identified by Jones et al. (2006) by revealing that the air masses causing rapid ozone depletion at Halley station originate in a region where a large 70 amount of fresh sea ice is formed. (4) Other factors. ODEs were also found impacted by the presence of mixed-phase clouds in the boundary layer due to the cloud-top radiative cooling (Hu et al., 2011) and regional climate variability such as the Western Pacific (WP) teleconnection pattern (Koo et al., 2014).
Although there exist many studies discussing the determining factors for the occurrence of the tropospheric ODE, the impact caused by the change in the stratospheric ozone on the occurrence of the tropospheric ODE has not been thoroughly investi-75 gated yet. In previous studies, most of often, ozone change in the stratosphere and the tropospheric ODEs were investigated separately. However, ODEs can be strongly influenced by the change in the stratospheric ozone. For example, the ozonelacking air in the boundary layer can be partly replenished by the ozone-rich air due to the downward entrainment from the lower stratosphere (Kuang et al., 2017). However, because of the strong stability of the polar boundary layer during ODEs, this influence might be minor. Moreover, the variation of the stratospheric ozone would also lead to a change in the solar radiation 80 reaching the ground surface, thus affecting the rates of photolysis reactions in the troposphere. As a result, the lifetimes of many atmospheric constituents in the troposphere and the occurrence frequency of the tropospheric ODEs can be altered. But it is still unclear whether the change in the stratospheric ozone will foster or retard the occurrence of the tropospheric ODE.
Therefore, in the present study, we combined the observational data from ground observation stations with two numerical models, to figure out the impacts on the occurrence of tropospheric ODEs in the Antarctic brought about by the change in the total 85 ozone amount including the stratospheric ozone. A concentration sensitivity analysis was also performed to reveal photolysis reactions heavily responsible for this impact on the tropospheric ODE.
The structure of the manuscript is as follows. In Sect. 2, observational data analyzed in the present study are described. The numerical models and the governing equations are also presented in this section. In Sect. 3, results obtained from the analysis of the observational data and the computations of the numerical models are shown and discussed. At last,in Sect. 4,conclusions 90 achieved in this study are summarized, and studies to be made in the future are also prospected.

Description of the Observational Data and the Numerical Method
In the present study, we first analyzed the observational data of ozone obtained from ground observation stations, and then tried to figure out the relationship between the variation of the total ozone column (TOC) and the occurrence frequency of tropospheric ODEs. Then we took the Halley station as an example, and applied two numerical models, TUV model and 95 KINAL box model, to capture the temporal change in ozone and major bromine species during ODEs under the conditions of the Halley station. Sensitivity tests were also performed in these models to figure out the influence caused by the change in the total ozone column (TOC) on the time variation of these air constituents (i.e. ozone and bromine species). Then the most influential photolysis reactions were identified through a concentration sensitivity analysis on the change in these air constituents.

Observational Data of Ozone
Two different types of ozone data were used in the present study. One is the total ozone column (TOC) obtained from groundbased observations. This kind of data mainly indicates the total ozone amount in a vertical column extending from the ground surface to the top of the atmosphere. Because approximately 90% of ozone in the atmosphere resides in the stratosphere, the change in TOC can heavily reflect the variation of the stratospheric ozone. The other type of the ozone data is the near-105 surface ozone mixing ratio recorded at ground observation stations, and it can partly represent the ozone concentration in the atmospheric boundary layer. The details of these ozone data are given below.

Total Ozone Column (TOC)
The TOC data used in this study were obtained from World Ozone and Ultraviolet Radiation Data Center of Canada (WOUDC, https://woudc.org/home.php) for all the registered stations in the Antarctic. These TOC data were observed using a Dobson 110 instrument (Balis et al., 2007), and cover a time span from the year 2000 to 2016. The time resolution of these data is 1 day.
After filtering out stations that possess only out-dated or incomplete data, we picked up Halley station (75.52 • S, 26.73 • W), which has the most complete TOC data, as an example to indicate the typical temporal change in TOC in the Antarctic. In order to guarantee the representativeness of the data from this station, we compared the time series of TOC recorded at the Halley station is analogous to those obtained at other observation stations. Furthermore, the correlation coefficients between the TOCs obtained at the Halley station and other stations mostly possess a value above 0.7 (shown in Tab. S1 of the supplements), indicating that the TOC data obtained at the Halley station is significantly correlated with those obtained at other stations, and thus can represent a typical TOC change in the Antarctic. The low correlation coefficient between the Halley station and the Faraday-Vernadsky station might be caused by the special geographical location of the Faraday-Vernadsky station, which is 120 out of the scope of the present study.

Surface Ozone
After choosing the Halley station as an example representing the typical TOC change in the Antarctic, we adopted the surface ozone data at the Halley station from World Data Center for Greenhouse Gases (WDCGG, https://gaw.kishou.go.jp), and the time span of the surface ozone data is between the year 2007 and 2013. The time resolution of the surface data is 1 hour.

125
Because the tropospheric ODE, which we focused on in the present study, mostly occurs in the springtime of every year, we thus only analyzed the surface ozone data during the springtime of the Halley station (from Sept. 1 to Nov. 30).
We then calculated the occurrence frequency of the tropospheric ODE at the Halley station for each month based on the daily data of the surface ozone. The occurrence frequency of the ODE, f , is estimated as follows, 130 in which N ODEs is the number of time points that ODE occurs, and N total is the total number of time points in this month.
However, the definition of the ODE occurrence is still under debate. In some previous studies (Tarasick and Bottenheim, 2002;Koo et al., 2012), the occurrence of ODEs was recognized by the surface ozone mixing ratio. When the surface ozone drops to lower than 20 ppb, it is called partial ODEs. Moreover, when the surface ozone declines to a level lower than 10 ppb, it is called major or severe ODEs. In contrast, the ODEs can also be judged by the change in the ozone mixing ratio. This method was 135 suggested by Bian et al. (2018) to indicate extreme events in polar regions. In the present study, we used the method suggested by Bian et al. (2018) and picked up the time point representing the occurrence of ODEs from the observational data when it fulfills the following criterion,

Numerical Methods
Two numerical models, a radiation model (TUV, Tropospheric Ultraviolet and Visible) Flocke, 1997, 1999) and a chemical box model (KINAL, KInetic aNALysis of reaction mechanics) (Turányi, 1990), were used in this study. The TUV model was used to estimate the photolytic rates of atmospheric constituents in the troposphere of the Antarctic. The KINAL box model was used to capture the concentration change in these constituents such as ozone in the boundary layer over time. KINAL was also used to compute the sensitivity of these constituents to each photolysis reaction in the chemical mechanism.

TUV model
The TUV model (Tropospheric Ultraviolet and Visible Radiation Model), provided by NCAR (National Center for Atmospheric Research), is able to calculate the tropospheric photo-dissociation coefficient Flocke, 1997, 1999), based on input parameters such as the total ozone column (TOC). The vertical ozone profile assumed in the model is taken from the US standard atmosphere. A total of 112 photolysis reactions are implemented in the TUV model.
The photolytic rate constant j p (unit: s −1 ) for each photolysis reaction is calculated by the TUV model as follows, In Eq. (3), σ(λ) represents the absorption cross section at the wavelength λ. Φ(λ) denotes the photolytic quantum yield.

KINAL model
After obtaining the photolytic dissociation rates of many atmospheric constituents using the TUV model, we then applied the chemical box model, KINAL (KInetic aNALysis of reaction mechanics) (Turányi, 1990) to capture the temporal evolution of chemical species such as ozone and many bromine species. Moreover, sensitivities of these chemical species to each photolysis reaction in the chemical mechanism were also computed using KINAL.

170
The governing equation describing the temporal evolution of chemical species in the KINAL model is as follows (Turányi, 1990): with the initial condition c| t=0 = c 0 , where c represents a vector of chemical species concentrations. k in Eq. (4) is a vector of rate constants of chemical reactions and t denotes time. E indicates the near-surface source emissions, and dc dt is the temporal 175 evolution of chemical species such as ozone. In the present study, a chemical mechanism including the bromine and chlorine chemistry was adopted from previous box model studies (Cao et al., 2014(Cao et al., , 2016aZhou et al., 2020), and the reaction rate constants were updated with the latest chemical kinetic data (Atkinson et al., 2006). There are in total 49 species and 141 chemical reactions included in the latest version of the chemical mechanism, which are listed in Tab. S2 of the supplementary material. Among these reactions, there are 23 photolysis reactions of which the rates are associated with the change in TOC, Table 1. Listing of photolysis reactions in the chemical mechanism of the KINAL model, of which the rates vary with TOC. The reaction numbers correspond to those listed in Tab. S2 of the supplements.

Reaction Reaction
Number and these photolysis reactions are listed in Tab. 1 along with the reaction numbers in the mechanism. Among these photolysis reactions, a part of them can enhance the occurrence of ODEs, while the others will retard it.
In the KINAL model, it is also assumed that bromide stored in the ice/snow-covered ground surface is inexhaustible. As a result, the rates of heterogeneous reactions such as HOBr (gas) + H + (liquid) + Br − (liquid) → Br 2 (gas) + H 2 O (liquid) , which are responsible for the bromine explosion, only depend on the availability of HOBr in the atmosphere in model calculations.

185
Moreover, the wind speed was assumed as 8 m s −1 (Beare et al., 2006), which is a typical wind speed in polar regions. In addition, the roughness of the ice/snow surface is set to 10 −5 m (Stull, 1988), and the height of the polar boundary layer is Table 2. Initial atmospheric composition in the boundary layer of the Antarctic (ppm = parts per million, ppb = parts per billion, ppt = parts per trillion) (Piot, 2007), and the prescribed intensities of emission fluxes from the ground surface (units: molec. cm −2 s −1 ) (Hutterli et al., 2004;Riedel et al., 2005;Jones et al., 2011), assumed in the model.

Species
Mixing assumed as 200 m because the typical thickness of the boundary layer in polar regions is about 100-500 m (Simpson et al., 2007;Anderson and Neff, 2008). The initial atmospheric composition used in the model is listed in Tab. 2, which represents a typical air composition in the Antarctic (Piot, 2007). Constant emission fluxes from the ground surface were also prescribed in 190 the model according to observations (Hutterli et al., 2004;Riedel et al., 2005;Jones et al., 2011), and the prescribed intensities of the emission fluxes are also presented in Tab. 2.

Concentration Sensitivity Analysis
After obtaining the temporal evolution of ozone and major bromine species, relative concentration sensitivities of these species to different photolysis reactions in the chemical mechanism were computed to reveal the dependence of these species on each 195 photolysis reaction of the mechanism. The relative concentration sensitivity S ij is calculated by: which shows the importance of the j-th reaction for the concentration change in the i-th chemical species. In Eq. (5), i is the index of chemical species, and j is the index of chemical reactions in the mechanism. c i is the concentration of the i-th species, and k j is the rate constant of the j-th reaction. S ij , an element of the relative sensitivity matrix, indicates the change in the i-th 200 species concentration resulted from a small perturbation in the j-th reaction rate. The evaluation of the concentration sensitivity is helpful for discovering the importance of specific reactions in the chemical mechanism for the concentration change in the focused species.
In the following section, the computational results are presented and discussed.
In this section, we first show the relationship between the TOC change and the occurrence frequency of the tropospheric ODE

Relationship between the TOC Change and the Tropospheric ODE
The time series of the monthly-averaged TOC at the Halley station and the occurrence frequency of ODEs are presented in Fig. 2. In general, it seems that the occurrence of ODEs is negatively correlated with the temporal change in TOC, with a 215 correlation coefficient of -0.27. It is seen that during the springtime of each year at the Halley station, mostly the ODEs occur more frequently in September and October than in November. Meanwhile, TOC was also found increasing from September to November in most of the years investigated (see Fig. 2). However, this negative correlation does not guarantee that the occurrence of ODEs is negatively dependent on the change in TOC, because that apart from the change in TOC, the variation of solar zenith angle (SZA) between different months would also heavily influence the solar radiation reaching the boundary 220 layer, consequently leading to an altering of the ODE occurrence. Thus, in order to clarify it, we ranked the values of TOC and the ODE occurrence frequencies belonging to different months for the years 2007-2013 (see Tab. 3). From the comparison between the ranks of TOC and the ODE occurrence frequencies, it is still difficult to determine the association between the occurrence of ODEs and the variation of TOC. In October, it seems that higher TOC leads to a lower occurrence frequency of ODEs (see Tab. 3). However, this tendency is different in November, in which higher TOC seems to cause a higher occurrence 225 frequency of ODEs. But it should be noted that the occurrence frequency of ODEs in November is remarkably lower than those in September and October. Thus, from the point view of the observational data, despite the small sample size of the ranks of the monthly averaged ozone data and the ODE occurrence frequency, we suggested that the occurrence of ODEs seems negatively dependent on the change in TOC, although the negative correlation is not very evident. To validate this suggestion, more observational data with extended time periods are needed, which unfortunately we are currently lack of.   The temporal change in ozone and bromine species under the conditions of the year 2013 with an average TOC of 220.03 DU is shown in Fig. 3. From the temporal behavior of these chemical species, we can better understand the interconversion of It can be seen in Fig. 3 that the whole process of the ODE can be divided into four time stages, according to the types of 240 bromine species in the troposphere. The first stage is the beginning of ODEs, in which the concentration of bromine is low, and ozone is only slightly depleted. This time stage would last for about 6 days in the present simulation (see Fig. 3). Subsequently, in the second time stage, Br 2 released from substrates such as the ice/snow-covered surface due to the bromine explosion mechanism is continuously photolyzed, forming Br atoms. The Br atoms then react with ozone and form BrO: A part of BrO then gets oxidized and converted to HOBr: Thus, in the second time stage, the major bromine species are BrO and HOBr. In contrast, the concentration of Br is almost zero due to the presence of ozone. Under this high-HOBr environment, a large amount of bromide is activated from substrates, so that ozone is rapidly consumed by the massive amount of bromine in the atmosphere. Thus, this second time stage is also 250 the key time period that the majority of the tropospheric ozone is depleted, and this time stage lasts for about 1 day, in which the majority of ozone is depleted with a rate of 0.5-1.0 ppb hr −1 . to Br, thus not existing in the atmosphere. Meanwhile, the formed Br is eliminated by aldehydes and HO x free radicals in the atmosphere, and is converted into HBr. As a result, at this last stage, a complete depletion of ozone in the troposphere is achieved, and the major bromine species is HBr, which is in accordance with previous observations (Langendörfer et al., 1999).

Impact of the Change in TOC on the Occurrence of ODEs
We then compared the model results for different months of the springtime of 2013. The TOC values of these three months, September, October and November, are 175.0, 179.9 and 305.2, respectively. However, it should be noted that aside from the change in TOC, the value of SZA also varies between different months, which may significantly affect the radiation fluxes reaching the boundary layer and thus the rates of photolysis reactions.

265
The temporal change in ozone during ODEs under the conditions of different months belonging to the year 2013 is shown in Fig. 4. It is seen that compared with the situation in September, the decline of ozone in November is delayed. Moreover, the depletion rate in November was also found lower than that in September. It suggests that under the weather conditions of November 2013, the occurrence of ODEs is more difficult to achieve, compared with that in September 2013. This result is also consistent with the change in the occurrence frequency of ODEs shown above. However, as mentioned above, although TOC 270 increases from September to November in 2013, it does not guarantee that the occurrence of ODEs is negatively dependent on the variation of TOC, because SZA also decreases, which may heavily enhance the radiation fluxes reaching the ground surface.
Therefore, we continued to test different TOC that corresponds to the October month for three different years ( The mechanism we proposed is that when TOC decreases, a larger amount of solar radiation would reach the troposphere, leading to an acceleration of photo-chemical reactions in the troposphere. As a result, the formation of major bromine species such as BrO and HOBr as well as the bromine activation become faster, and the occurrence of ODEs is also accelerated.

290
However, it is still unclear through which photolysis reactions the change in TOC deeply affects the occurrence of tropospheric ODEs. Thus, we continued to analyze the photo-chemical reactions using the concentration sensitivity analysis, as presented below.

Sensitivities of Ozone and Major Bromine Species to Photolysis Reactions
In the present study, the impact on ODEs caused by the change in TOC in the models is exerted through 23 photolysis reactions, 295 which are listed in Tab. 1. In order to figure out which photolysis reactions are the most important ones during the ODE process, we performed a concentration sensitivity analysis on the change in ozone and major bromine species for the springtime of 2013, so that the dependence of ozone and bromine species on these photolysis reactions can be revealed.
The relative concentration sensitivities of ozone and major bromine species (i.e. BrO, HOBr and HBr) to all the 23 photolysis reactions on Day 6.5, which resides in the second time stage of ODEs when the strongest ozone depletion occurs (see Fig. 3), 300 are shown in Fig. 6. From Fig. 6(a), it can be seen that Reactions (SR7) and (SR11): possess the largest positive sensitivities for the mixing ratio of ozone. It means that the rate increase of these two photolysis 305 reactions leads to an elevation of the ozone value during ODEs and thus a retardation of ODEs. These two reactions are thus named "major ODE decelerating reactions" in the following context. The reason for the delay impact on ODEs brought by Reaction (SR7) is that in this reaction, BrO is photolyzed, forming Br. As a result, the formation of HOBr by the oxidation of BrO is decelerated due to the reduction of the available BrO. Thus, the heterogeneous bromine activation process, i.e. bromine explosion mechanism that HOBr participates in, gets retarded, leading to a slow down of the bromine activation and the ozone 310  depletion. Apart from that, additional ozone is also formed through Reaction (SR7). With respect to Reaction (SR11), HOBr is photo-decomposed through this reaction. Thus, the heterogeneous bromine activation also gets suppressed by the strengthening of this reaction, resulting in a delay of the ozone depletion.
In contrast, Reactions (SR1), (SR57), and (SR58), are largely negatively correlated with the change in ozone (see Fig. 6a). It denotes that when these three reactions speedup, the 320 level of ozone drops, which represents an acceleration of the tropospheric ODE. These three reactions are thus named "major ODE accelerating reactions". For Reaction (SR1), it is not surprising as this reaction is the direct photolysis of the tropospheric ozone. Moreover, Reaction (SR1) also serves as a major formation pathway of OH radicals. OH radicals are important for the occurrence of ODEs and the bromine explosion mechanism, as they are involved in Reaction (SR18): In Reaction (SR18), not only the Br atoms are generated by the conversion from the released Br 2 , but also the formation of HOBr is strengthened, leading to an acceleration of the bromine activation. As a result, an enhancement of Reaction (SR1) would lead to a speedup of the ozone depletion. Regarding Reaction (SR57), the occurrence of this reaction also strengthens the occurrence of ODEs, because this reaction also forms OH radicals, which are critical for the bromine explosion mechanism as mentioned above. With respect to Reaction (SR58), it possesses the most negative ozone sensitivity (see Fig. 6a), which 330 means that this reaction heavily controls the depletion of ozone. It is because this reaction reinforces the formation of HO 2 . As HO 2 is the key oxidant for the formation of HOBr through reaction: the enhancement of Reaction (SR58) thus favors the release of bromine and the depletion of ozone.
From Fig. 6(b)-(d), we can see that the sensitivities of major bromine species (i.e. BrO, HOBr and HBr) to photolysis 335 reactions mostly have an opposite sign, compared with those corresponding to ozone (shown in Fig. 6a). It is because the bromine species in the atmosphere are mostly responsible for the ozone depletion in the troposphere during ODEs. Within these photolysis reactions, Reactions (SR7) and (SR11) have the largest negative sensitivities as they strongly decelerate the bromine explosion mechanism. In contrast, Reactions (SR1), (SR57) and (SR58) exert a positive impact on the change in bromine species. It is because the speedup of these three reactions can reinforce the bromine explosion mechanism as mentioned above, 340 thus leading to a positive dependence of these bromine containing compounds on these three reactions.   (Vaghjiani and Ravishankara, 1989), which overlaps the wavelength range that the TOC change strongly affects. Thus, the photo-decomposition of H 2 O 2 (i.e. Reaction (SR57)) also 365 gets strongly promoted when TOC decreases. (3) The absorption spectrum of HCHO spreads from 260 nm to 360 nm, with many vibrational structures (Rogers, 1990). As a result, the decline of TOC substantially enhances the photolysis of HCHO (i.e. Reaction (SR58)).
On the contrary, for the major ODE decelerating reactions, i.e. photolysis of BrO and HOBr, their rates are only slightly influenced by the decrease of TOC. It is because that BrO has an absorption spectrum in the range of 290-380 nm, peaking at 370 approximately 330 nm (Wilmouth et al., 1999). The strengthening of the solar radiation especially in 200-320 nm due to the decline of TOC thus only exerts a small impact on the photolysis of BrO (i.e. Reaction (SR7)). Regarding HOBr, its photolysis rate relies more on the strength of the UV-A radiation (i.e. in 320-400 nm) reaching the boundary layer, as it has a broad absorption spectrum between 200 and 400 nm (Burkholder et al., 2015). Therefore, the influence caused by the TOC change on the photolysis of HOBr is also weak.

375
Hence, it can be concluded that when TOC decreases, the rates of major ODE accelerating reactions (i.e. Reactions (SR1), (SR57) and (SR58)) significantly increase, while the rates of major ODE decelerating reactions (i.e. Reactions (SR7) and (SR11)) are hardly changed. Consequently, the ODEs are accelerated under a low TOC condition, and vice versa, resulting in a negative association between the change of TOC and the occurrence frequency of ODEs as presented above.
In this study, we investigated the connection between the change in the total ozone column (TOC) and the occurrence of the tropospheric ozone depletion events (ODEs) in the Antarctic. Photolysis reactions dominating this connection were also identified using a concentration sensitivity analysis. Major conclusions achieved in the present study are as follows.
Based on the analysis of the observational data belonging to the years 2007-2013, we suggested that the decrease in TOC possibly favors the occurrence of the tropospheric ODEs. Then, the model results with the input data representing the conditions 385 of October belonging to different years indicate that the occurrence of ODEs would be accelerated when TOC decreases.
Moreover, key photolysis reactions that dominate the production and the consumption of ozone during ODEs, i.e. major ODE accelerating reactions and major ODE decelerating reactions, were also figured out. It was found that when TOC changes, the rates of major ODE accelerating reactions are substantially altered, while the rates of major ODE decelerating reactions mostly remain unchanged, leading to the negative association between the TOC change and the occurrence frequency of ODEs.

390
Improvements can be made to the present study. For instance, many other factors that are able to influence the occurrence of ODEs such as the type of the sea ice and the existence of frost flowers should be considered in the future work. Unfortunately, currently we are still lack of these observational data. Aside from that, a study for Arctic conditions should also be conducted, so that the conclusions obtained in the present study can be compared and verified, which is attributed to a future publication.
Code and data availability. The source code of the model and the data of computational results shown in this article can be acquired upon 395 request from the authors.
performed the computations, and wrote the paper with Le Cao together. Simeng Li revised the chemical mechanism and Shuangyan Yang gave valuable suggestions on the improvement of the manuscript. All the authors listed have read and approved the final manuscript.
Competing interests. The authors declare no conflict of interest.