The three referees are in agreement that the paper has improved significantly in revision. Referee 3 recommends accept as is. Referee 1 identifies further revisions/corrections that are required before publication -- I think that it will be straightforward to address these. Referee 2 identifies a number of points that he/she feels were not addressed adequately in revision and notes that the responses to referees did not in many cases make it completely clear what changes had been made.

Please can you address the comments of Referees 2 and 3, for Referee 2 in particular making it clear in your responses how you are addressing the Referee's comments and what further changes you have made. Please can you also clarify what changes you made to the Discussion section in your previous revision -- since according to the revised version submitted in which new text was highlighted all the text in that section appears to be 'new'. (I imagine that it is not -- but clarification would be helpful.)

If you make revision to address the comments of Referees 2 and 3, and if your responses to Referee 2 in particular are clear and you provide some explanation of changes to the Discussion then I expect to accept the paper without further delay.

Your further changes and responses seem generally adequate to me and I think that the paper makes some interesting points. I understand that you do not wish to address all of Referee 3's comments by making changes to the paper -- and, in the end, you as authors, have to take responsibility for what is in the paper. On the other hand I do feel that some of Referee 3's comments remain valid and that they can be addressed by simple further clarifications.

Please provided a revised version of the paper which makes the following minor changes. I regard these as straightforward, but if you really do not want to make one or two of them then provide clear arguments why this is the case. After these changes have been made I will accept the paper.

[The next comment is quite long -- but the summary is that I am requesting a clearer specification of the model and a clearer argument as to why it is fit for purpose.]

l98-117: Referee 3 requested evidence that the SILAM model was fit for purpose. You have cited the Sofiev et al (2015) paper. That demonstrates that the SILAM model performs well on a number of basic tests, but it does not demonstrate that, when driven by e.g. ERA-I velocity fields, it provides a good simulation of stratospheric tracers. Please take a little more time to explain the logic here.

Sofiev et al (2015) show that the SILAM numerical model performs well in a range of problems in simulating tracer advection by a specified wind field.

Others, e.g. Diallo et al (2012) have shown that using other numerical approaches that ERA-I winds give a good (to the extent that can be assessed) simulation of stratospheric tracers. Therefore on this basis you consider that the SILAM model driven by ERA-I winds will also give a good simulation of stratospheric tracers.

But this logic ignores the point that the vertical winds are calculated in SILAM using a certain method -- can you be sure that this method works for ERA-I wind fields? Have you for example, done any tests of prediction of age-of-air or some other artificial tracer vs the prediction of Lagrangian methods such as those used by Diallo et al?

Furthermore the procedure for calculating vertical transport in given in 3.5 is quite complicated and I am a bit confused whether this is the procedure described in Sofiev et al (2015), or whether it is a new procedure -- please clarify.

You should also make it clear that the model is 3-dimensional and give details of the resolution. You have said that the velocities are retrieved on a 500x250 lat-lon grid -- I think actually you mean lon-lat grid -- but you haven't said explicitly that this is the resolution of the numerical model. Again re vertical levels you have given details of the levels on which the velocities are provided, but have not confirmed explicitly that these are the levels in the SILAM model. (The velocities could have been interpolated to some other grid.) Actually I now see that some of this information is in Section 3.5 -- and the information on the resolution of the velocity data is provided again here -- that is a bit confusing. At the very least the term 'setup' used in Section 2.1 needs to be more explicit and mention grid resolution etc.

This then returns to Referee 3's overall point -- can you provide some concrete evidence that the SILAM model gives similar results to other approaches for calculating stratospheric tracers from ERA-I winds or similar. If you cannot then you should say explicitly that whilst the SILAM model has been carefully constructed it is not yet know how it performs against, say, Lagrangian calculations using diabatic vertical velocities and any results are therefore subject to this uncertainty.


l449: Give the difference from Kovacs et al in the same units as are used in the plots (and elsewhere), i.e. pmol/mol rather than pptv. Also include a following sentence 'Note that whilst we regard this newer version of MIPAS SF6 data as an improvement, it has not yet been reported in a publication, and on that basis is subject to uncertainty.'

Figure 6: include information on meaning of dashed lines in the caption.

l471: 'non-monotonous' > 'non-monotonic'

l500: 'overstating' -- not the right word -- I suggest 'overestimating' is better:

l540: as above 'over-stated/overstating' should be 'over-estimated/over-estimating'

l755: 'and the SF6 observations are potentially a good means to evaluate it'

l758: 'downdraft' > 'descent'

l769: 'braking' > 'breaking'

(Some of the above are straightforward typographical or language errors -- the copy editors will correct some of them, but it is in your interests to remove as many as possible before submitting the final version of the paper.)

Thank you for implementing further changes as recommended. I am pleased to accept the paper for publication in ACP.

I've accepted subject to technical corrections because you seem to have submitted a supplement which is actually the new version of the paper with changes explicitly marked. We should make sure that the unintended supplement does not appear. Please can you negotiate with the Editorial Office about how to sort this out -- it might be that simply submitting the same version again without any supplement is the best way to sort this out.