|The authors addressed some of the criticism and overall the manuscript has somewhat improved since the previous version. I am still not convinced that the manuscript represents a significant advance in science however I do appreciate author’s efforts to disentangle statistical links between Arctic stratospheric ozone and surface climate. Below I add some comments; hopefully the authors find them useful for further improvement of the manuscript.|
First, contrary to what the authors assert, I don’t believe the connection between Arctic ozone and surface climate is “settled science”; I simply don’t think the authors approach this problem from the right perspective. In Calvo et al they did demonstrate that, as ozone variability increased from low Ozone Depleting Substances (ODS) period to high ODS period, this was followed by increased stratospheric cooling and stronger changes in surface climate. While their study may have caveats, for example because they only use one model, they could however with reasonable confidence attribute their surface climate response during the second period to increased stratospheric variability caused by increased ODS and low ozone. I don’t think they assumed a direct ozone-surface link because they carefully documented responses of zonal mean temperatures and winds consistent with the dynamical stratosphere-troposphere coupling, although they might have discussed the mechanism in a more clear way.
I also don’t see how the authors of the present manuscript addressed the statement from Ozone assessment, which they cite. It is well-established that extreme low ozone values, observed during some winters in the Arctic, are a result of chemical depletion, rather that stratospheric dynamics alone, see for example Manney et al. 2011. Thus, based on Calvo et al results, surface anomalies that follow low ozone episodes may be attributable to chemical ozone depletion. While the ozone assessment statement does not mention any mechanism, it is certainly consistent with stratosphere-troposphere dynamical coupling similar to what is apparently operating in the Antarctic. Thus I don’t see how the authors “explored the robustness” of the statement in the ozone assessment.
In summary, while I find that the part of the study concerning ASO-ENSO link has improved and is worth reporting, I still don’t find that the part concerning links between ASO and polar surface climate is equally exciting. Below are my suggestions:
1. The authors find strikingly large differences in ozone-climate coupling even between simulations by the same model. In Calvo et al they find similarly large differences in coupling between two periods when ozone was / was not depleted. During ozone depletion period the ozone variability was twice as large, owing to extreme low ozone events. This was linked to larger signal in stratospheric temperatures and thus in surface climate. I suggest that the authors test whether the spread shown in Figures 4 and 6 can similarly be attributed to different interannual variability between these periods. Specifically, do you see larger interannual variability in ozone (measured e.g. by interannual standard deviation) during years when ozone-climate coupling is larger? One could expect stronger coupling when there is larger variability, due to increased signal to noise ratio.
2. Further, can the authors attribute stronger coupling (and possibly larger variability) to different ozone levels and thus possibly to chemistry? The authors showed how the coupling changes between different periods (1970-2010, 2011-2051, 2052-2092) but it is difficult to understand their result without knowing what ozone values were during these runs. Is there, for example, any indication of chemical ozone depletion in those runs? One could test correlation coefficients against minimum ozone values achieved in those runs. I also wonder what could be the cause for seemingly stronger coupling during some future runs (2052-2092). One could imagine that there are still episodes of chemical ozone depletions that, combined with increased Brewer-Dobson circulation in future, would lead to larger variability and stronger coupling. I guess negative answer to these questions would indeed indicate that the variability in coupling is due to internal variability, as the authors assert, but the tests proposed above could still give more insights.
I appreciate large amount of work that authors put into the study but I nevertheless don’t find that the manuscript is complete without this additional (and rather straightforward) analysis.