Stratospheric ozone loss over the Eurasian continent induced by the polar vortex shift

The Montreal Protocol has succeeded in limiting major ozone-depleting substance emissions, and consequently stratospheric ozone concentrations are expected to recover this century. However, there is a large uncertainty in the rate of regional ozone recovery in the Northern Hemisphere. Here we identify a Eurasia-North America dipole mode in the total column ozone over the Northern Hemisphere, showing negative and positive total column ozone anomaly centres over Eurasia and North America, respectively. The positive trend of this mode explains an enhanced total column ozone decline over the Eurasian continent in the past three decades, which is closely related to the polar vortex shift towards Eurasia. Multiple chemistry-climate-model simulations indicate that the positive Eurasia-North America dipole trend in late winter is likely to continue in the near future. Our findings suggest that the anticipated ozone recovery in late winter will be sensitive not only to the ozone-depleting substance decline but also to the polar vortex changes, and could be substantially delayed in some regions of the Northern Hemisphere extratropics.

The manuscript is nicely written, presentation of results is appealing and the underlying science and methods are sound. This work contains novel results which highlight the important connection between climate change and stratospheric ozone and the conclusions reveal its regional manifestation, also in the crucial context of climate change effects to future ozone hole recovery. The topic should be of interest to fields other than atmospheric chemistry and climate change, for example human health or biodiversity (for the possible implications on surface UV). I recommend publication once the following minor points are dealt with (referring to the manuscript with the corresponding lines).
Reviewer #3 (Remarks to the Author): It is not at all surprising or novel that ozone concentrations co-vary with the location of the polar vortex (since the polar vortex acts as a containment vessel and facilitates ozone depletion).
In my view, the authors would have done better to combine the key results of this paper with those of Zhang et al. 2016, creating one coherent and consistent story. This currently feels like an attempt to extract two separate manuscripts from the same story.
The introduction lacks some basic introduction to the dynamical differences between the Arctic and Antarctic polar vortices. Some of the factors alluded to arise simply from the fact that wave driving is larger in the Northern Hemisphere, making the polar vortex more perturbed than it is in the SH. The Artic vortex is more disturbed on both intraseasonal and interannual timescales, whereas the Antarctic vortex is climatologically stronger, allowing for substantial ozone depletion. (e.g, Waugh and Polvani 2010;Solomon et al. 2014;Sheshadri et al. 2015 and probably many others, this is now almost textbook material).
In looking at the polar vortex shift, did you make allowances for the fact that in about 6 years out of 10, the polar vortex underwent major warming events? Particularly when there is a "split" event, the vortex splits into two daughter vortices, the bigger one of which is over Eurasia and the smaller one of which is over North America (Matthewman et al. 2008). Perhaps what you are seeing as a trend towards a shift in the location of the vortex is simply averaging over these events? There were 12 subsequent years with SSWs starting in the early 2000's.
L23 Most such studies have focused on the SH, where ozone depletion has been substantial. Ozone anomalies or trends have been quite small in the NH in comparison. L47 EOF analysis of what? Daily ozone concentrations as a function of longitude and latitude? Was the variable of interest weighted in any way? L62 There is some longitudinal structure L117 you mean vortices L126 it is not active "transport" in the way that this sentence implies. The "balanced" state is that that vortex itself is located in a different position, and therefore the ozone concentrations appear shifted in location. "Transport" implies transport by the circulation.

Reviewer #4 (Remarks to the Author):
This study can be considered as the "part 2" of a recent paper published by the same group of authors in Nature Clim. Change (Zhang et al. 2016). Therefore, its background mainly relies on this previous "part 1" investigation. In short, sea ice depletion over the North Pole coupled with an increased snow precipitation over the north Asia worked together in shifting the polar vortex over the Eurasian continent and away from North America during the last decades. With this in mind, the new study focuses on the ozone changes induced by this shift. By analyzing ozone reanalysis and simulations carried out with SLIMCAT model the authors attribute these changes to transport of ozone-poor air from Arctic and to local chemical processes caused by the transport of active chlorine species towards lower latitudes.
Overall, the study is well planned, results are convincing and it provides a useful contribution to evaluate the ozone temporal evolution/recovery at northern high latitudes. Nevertheless, in the present form this work can hardly be published in a high ranking journal while I find it more suitable for JGR or ACP. Personally, I would find more convenient having included some of the present findings within the part 1, but I understand the need of splitting the work, so I don't blame the authors. However, having already achieved the main results in the part 1, the additional findings presented in this new paper look somewhat ordinary and far from being very intriguing. The authors state that "The question remains whether this shift of the polar vortex induced by climate change can change trace gas distributions, especially the ozone distribution in the Northern Hemisphere during late winter". In the light of the part 1 of the study, it seems rather plausible that ozone would follow the vortex shift so one can expect what the authors described here. Obviously, the study has some merit e.g. the assessment of the ozone transport with respect to the local chemical processes, but I think that at the present stage this is not enough to warrant publication on Nature communication and some further work should be carried out. Also, the conclusions of this study don't look novel (e.g. "The results suggest that the future ozone recovery in late-winter would be sensitive not only to ODSs change but also to the polar vortex change") and I believe that a broader paper instead of a short communication could better address the issues here mentioned.
I actually don't have specific suggestions on how making this work more appealing, nevertheless I still see that the analysis is basically limited to SLIMCAT and MSR2 datasets. Due to the limited temporal extension of the time series and the expected large internal climate variability in the northern high latitudes, can an additional analysis based on climate models (either from the CCMI or CMIP5) be useful in this context? Moreover, since things seem to be somewhat clear for the "historical" period, I would find even more remarkable the investigation of the future ozone response under different RCP scenarios. Then, is MSR2 the best observational reference for this study? You should justify this choice. Recently, many combined O3 datasets spanning more than three decades are become available. Although they mostly rely on solar occultation instruments and therefore are characterized by a sparse and infrequent sampling at high latitudes, the new SBUV v8.6 datasets could still be useful. Moreover, since both the data assimilation of MSR2 as well as the SLIMCAT model are driven by ERA-interim, it could be interesting including an additional independent assimilated ozone dataset. A good candidate can be the Bodeker's NIWA ozone dataset. Finally, as ERA-interim assimilates also ozone fields, perhaps, despite its limitations, ERA-interim ozone could even be useful for checking the consistency of these results.
We thank the four reviewers very much for their important comments. This study, for the first time, highlights the environmental and chemical influences induced by the polar vortex shift revealed by Zhang et al. (2016). In the original version of this paper, we mainly analyzed the historical influence of the polar vortex shift on the ENAD mode of TCO over the Northern Hemisphere (NH) and proposed that the ENAD could delay the ozone recovery over Eurasia in the future. In the revision, we found the NH clear-sky ultraviolet variation is closely related to ENAD mode. Furthermore, we make more efforts to verify that the ENAD would be significantly enhanced in the near future by using results from Chemistry-Climate Model Initiative (CCMI) experiments. The revised manuscript is substantially improved and we have made the following major changes in response to the reviewers' comments and suggestions:

1) We used several methods to calculate polar vortex strength and shift indices.
The results indicate that our analysis is not sensitive to the method used for determining the vortex edge. 3) Some parts of the introduction have been rephrased in order to highlight the significance of Arctic ozone research and outstanding scientific questions related to this topic.

2) To highlight the importance of ENAD in
Our detailed replies are given below.

Responses to Referee 1
The authors study the correlation between the variation of total column ozone (TCO) and the deformation of the Arctic polar vortex over the past thirty years. By using empirical orthogonal function analysis (EOF), they find a new mode which explains the 30% variance in TCO and it correlates well with the displacement of the polar vortex towards Eurasia. In light of this, they claim that despite the overall general reduction of ozone-depleting substances after the Montreal Protocol, the expected increase in stratospheric ozone concentration in the coming decades can be spatially inhomogeneous and be significantly influenced by the deformation of the polar vortex. I think that the paper is suitable for publication and well written. I believe, however, that addressing the following comments would improve the quality and the rigorousness of the manuscript.
Response: We thank the reviewer for his/her comments. We have revised the manuscript carefully according to these comments and suggestions. The detailed point-to-point responses are as follows. (a) Is this a common method to address the polar vortex strength? If so, please insert a reference.
(b) Under certain assumptions, the 3D polar vortex is approximately 2D on isentropic surfaces. Hence I suggest this would be a better choice instead of the isobaric surfaces.
Furthermore, this would allow consistency with other analysis the authors did using potential vorticity (PV) and other quantities given on isentropic surfaces.  Examples of material features include the existence, displacement and deformation of a vortex. Secondly, assessing the vortex edge using PV-based methods would identify non-material structures, and hence any conclusion related to transport within the vortex would be unjustified. Even in the ideal case in which PV is conserved, the PV contours which have the largest PV gradient with respect to equivalent latitude (i.e., the Nash method) at two distinct times, do not advect into each others. The authors may look here: https://arxiv.org/abs/1702.05593 for further details on this. Having said that, there is a correlation between the vortex edge and PV. I expect that the results would not change substantially, but may further improve if the polar vortex edge is identified by more rigorous methods.
Response: Thank you very much for your suggestion. We read through the suggested paper (Serra et al., 2017) which was useful. We tried our best to recalculate the vortex edge according to the LCSs method described in the paper.
We downloaded the LCSs code from the author's website but we failed to drive it using our initial data to calculate the polar vortex edge. Instead, we used two other methods without using PV to examine the polar vortex edge according to  5. Lines 122-123. I don't understand the last sentence.

Responses to Referee 2
The manuscript of Zhang et al., presents an analysis of past stratospheric ozone changes for the Northern Hemisphere (NH) using a three-dimensional (3D) chemical transport model (CTM), global meteorological re-analyses and satellite total ozone datasets and statistical methods. It is based on recent previous work by the same authors that established that the late winter NH polar vortex has shifted persistently towards the Eurasian continent over the past three decades, associated with Arctic sea-ice loss induced by global warming. The current work reveals a very important implication of this shift, related to a dipole spatial structure in the inter-annual variability and trend of the NH total ozone column (TOC) in the latitudes northern than 45 degrees. It does so in an unequivocal manner, using eigen-value statistical computations to demonstrate the existence of a Eurasia-North America Dipole (ENAD) pattern in the TCO observed and modelled fields. It is also shown that the TCO decline over Eurasia, linked to the Arctic vortex shift, is brought about by dynamical and chemical ozone loss factors. This is made possible via the SLIMCAT CTM set-up and specialised simulations that can distinguish between "dynamical" and "chemical" ozone as well as for gas-phase only and heterogeneous ozone depletion.
The manuscript is nicely written, presentation of results is appealing and the underlying science and methods are sound. This work contains novel results which highlight the important connection between climate change and stratospheric ozone and the conclusions reveal its regional manifestation, also in the crucial context of climate change effects to future ozone hole recovery. The topic should be of interest to fields other than atmospheric chemistry and climate change, for example human health or biodiversity (for the possible implications on surface UV). I recommend publication once the following minor points are dealt with (referring to the manuscript with the corresponding lines).
Response: We thank the reviewer for the positive evaluation of our study and we appreciate the reviewer's very helpful comments. We have revised the manuscript carefully according to his/her suggestions.
1) line 28: Some more credit to related ground-breaking or more recent previous work on the long-term dynamical and chemical ozone changes is needed, by referring to the papers by Fusco and Salby (1999) and Harris et al. (2008). Response: This is a good suggestion and we have rephrased this sentence in the revised paper (please see Lines 53-55) as follows:

recently reported that the late-winter Arctic vortex has shifted persistently towards the Eurasian continent over the past three decades partly due to the Arctic sea-ice loss"
3) line 117: replace "vortexes" with "vortices" Response: Corrected, thank you. 4) lines 125-127: A brief 1-2 lines sentence is needed here to define the "dynamical and chemical ozone" before you start using it. I had to interrupt the reading and look for the detailed explanation in the Methods (in SLIMCAT model) so a brief definition 12 at the beginning of this paragraph will help.

Response: We have added some SLIMCAT introductions in the revised paper
(please see L143-147) as follows:

"In the SLIMCAT simulation, the dynamical ozone concentrations can be calculated from a 'passive ozone tracer' advected by the model dynamical processes, while the chemical ozone depletion is represented by the difference between the modeled ozone concentrations with full chemistry and dynamical ozone concentrations (see Methods section)."
It is not at all surprising or novel that ozone concentrations co-vary with the location of the polar vortex (since the polar vortex acts as a containment vessel and facilitates ozone depletion).      this is now almost textbook material).

Response: Thanks for your good suggestion. We have rewritten Introduction
section and added information required by the reviewer in the revised paper, particularly, the first paragraph (L21-L37) is rewritten as follows:

Particularly, Antarctic ozone has been reported to experience the third-stage of ozone recovery 21 . In contrast, the polar vortex in Arctic region is more dynamically disturbed by planetary waves than in the Antarctic vortex 22-24 and complex interactions between chemical and dynamical processes at high northern latitudes make it more challenging to identify Arctic stratospheric ozone trends 25-28 . Recently, some studies have argued that dynamical processes such as the Arctic polar vortex changes will play a more significant role in affecting Arctic ozone in the future;
consequently, the timing of Arctic stratospheric ozone recovery is not so robust 28-32 ." All the above references have been added into the revised paper and please see the new reference list.
In looking at the polar vortex shift, did you make allowances for the fact that in about 6 years out of 10, the polar vortex underwent major warming events? Particularly when there is a "split" event, the vortex splits into two daughter vortices, the bigger one of which is over Eurasia and the smaller one of which is over North America (Matthewman et al. 2008). Perhaps what you are seeing as a trend towards a shift in the location of the vortex is simply averaging over these events? There were 12 subsequent years with SSWs starting in the early 2000's.

Response: Zhang et al. (2016) pointed out that the shift of the polar vortex towards Eurasia is closely related to the strengthened upward wave associated
with Arctic sea-ice loss in past decades. However, more upward propagating waves may not necessarily cause a sudden stratospheric warming (SSW) event.
On most occasions, it may cause a weaker polar vortex, which is more likely to be displaced, rather a SSW event. From the climatological point of view, Figure R9 shows the annual occurrence of major SSWs during December, January and

L62 There is some longitudinal structure
Response: Thank you. We have emphasized this point as follows:

Response: Corrected.
L126 it is not active "transport" in the way that this sentence implies. The "balanced" state is that that vortex itself is located in a different position, and therefore the ozone concentrations appear shifted in location. "Transport" implies transport by the circulation.

Response: We have rephrased this expression in the revised paper as follows:
" Figure 3d-f clearly show negative dynamical ozone anomalies over the Eurasian continent, i.e., the low ozone center is also shifted along with the polar vortex shift."

Responses to Referee 4
This study can be considered as the "part 2" of a recent paper published by the same group of authors in Nature Clim. Change (Zhang et al. 2016). Therefore, its background mainly relies on this previous "part 1" investigation. In short, sea ice depletion over the North Pole coupled with an increased snow precipitation over the north Asia worked together in shifting the polar vortex over the Eurasian continent and away from North America during the last decades. With this in mind, the new study focuses on the ozone changes induced by this shift. By analyzing ozone reanalysis and simulations carried out with SLIMCAT model the authors attribute these changes to transport of ozone-poor air from Arctic and to local chemical processes caused by the transport of active chlorine species towards lower latitudes.
Overall, the study is well planned, results are convincing and it provides a useful contribution to evaluate the ozone temporal evolution/recovery at northern high latitudes. Nevertheless, in the present form this work can hardly be published in a high ranking journal while I find it more suitable for JGR or ACP. Personally, I would find more convenient having included some of the present findings within the part 1, but I understand the need of splitting the work, so I don't blame the authors. However, having already achieved the main results in the part 1, the additional findings presented in this new paper look somewhat ordinary and far from being very intriguing. The authors state that "The question remains whether this shift of the polar vortex induced by climate change can change trace gas distributions, especially the ozone distribution in the Northern Hemisphere during late winter". In the light of the part 1 of the study, it seems rather plausible that ozone would follow the vortex shift so one can expect what the authors described here. Obviously, the study has some merit e.g. the assessment of the ozone transport with respect to the local chemical processes, but I think that at the present stage this is not enough to warrant publication on Nature Communication and some further work should be carried out. Also, the conclusions of this study don't look novel (e.g. "The results suggest that the future ozone recovery in late-winter would be sensitive not only to ODSs change but also to the polar vortex change") and I believe that a broader paper instead of a short communication could better address the issues here mentioned.
I actually don't have specific suggestions on how making this work more appealing, nevertheless I still see that the analysis is basically limited to SLIMCAT and MSR2 datasets. Due to the limited temporal extension of the time series and the expected large internal climate variability in the northern high latitudes, can an additional analysis based on climate models (either from the CCMI or CMIP5) be useful in this context? Moreover, since things seem to be somewhat clear for the "historical" period, I would find even more remarkable the investigation of the future ozone response under different RCP scenarios.
Response: We thank the reviewer for his/her comments. In the original version of this study, we mainly analyzed the historical influence of polar vortex shift and gave a perspective that the ENAD could delay the ozone recovery in the future.
We agree with the reviewer that it would be unique and important to include additional analysis based on climate models for this work. That is, we used chemistry-climate models (CCMs) to predict the polar vortex shift and ENAD changes in the future.

Initiative Phase-1 (CCMI-1) in simulating the two leading modes of Northern
Hemisphere TCO and polar vortex shift in historical period for the purpose of picking out suitable CCMs to simulate future ENAD changes. To evaluate the performance of 13 CCMs in simulating historical spatial patterns of EOF1 and ENAD, Taylor diagram ( Figure R11) is plotted, which shows how closely a spatial pattern from models matches that from observation. With regards to EOF1 of the TCO (Fig. R11a)    assimilates also ozone fields, perhaps, despite its limitations, ERA-interim ozone could even be useful for checking the consistency of these results.  In the 2nd revised paper, we added the PIs of all the Chemistry-Climate Model

Initiative (CCMI) models as co-authors, according to the editor's and reviewers'
comments and suggestions. The authors have reviewed the manuscript and confirmed their authorships. Our detailed replies to the 3rd reviewer's comments are given below.

Responses to Referee 3
I continue to think that the results presented in this manuscript are not particularly novel, and that this paper was written more from the desire to publish another manuscript than from any real need to split this work. (Particularly if this is aimed at specialists in the field --specialists in the field would expect that ozone concentrations co-vary with the location of the vortex.) Some comments: 1. You could calculate EOFs of anything and call them modes (which is not even accurate terminology, since EOFs are statistical patterns and not modes of an underlying dynamical system).

Response: Although the EOF analysis is a mathematical tool, it is commonly used
to detect leading spatial-temporal patters in meteorological fields. The ENAD pattern derived from EOF analysis highly resembles the TCO pattern projecting on polar vortex shift index, suggesting that the ENAD mode indeed has a physical meaning. 2. I'm not sure what you mean by evidence that this "mode" could delay future ozone recovery, to me it just seems like you are saying that future ozone recovery will be set by both future emissions and the dynamics of the polar vortex (both of which are pretty obvious statements). I have now read this manuscript twice, and it hasn't convinced me that ENAD has implications for understanding anything really. To repeat, it is not surprising at all that there is a correlation between your ENAD and the shift of the vortex edge in the CCMI models either (one expects ozone concentrations to co-vary with the vortex location --the opposite conclusion would be surprising). Response: Figure R3 shows the SSW duration and polar vortex shift in February.

Response
Note that there is no significant correlation between them. Furthermore, the SSW duration shows no significant trend while the polar vortex shift in February has an increasing trend. These results suggest that SSW may be not the direct cause for the polar vortex shift and its associated ENAD pattern.