Very good work. Please find my comments in the attached pdf file.

REVIEW OF GODIN-BEEKMANN ET AL., 'Updated trends of the stratospheric ozone vertical distribution in the 60°S-60°N latitude range based on the LOTUS regression model'.

General comments

This manuscript addresses what must be considered one of the most important questions in atmospheric chemistry and physics of all, namely is stratospheric ozone recovering as anticipated?

The authors provide an update of the work of the SPARC "LOTUS" project, and systematically present vertically resolved post-2000 trends in ozone from multiple carefully constructed long-term satellite records, calculated with a defined regression model.

The work is extremely relevant and suitable for publication in ACP.

The authors deserve credit for their careful and systematic approach and also for their transparency and openness. The results are presented clearly and readers can assess for themselves the level of agreement between different datasets and the significance of the different trends (and in fact there are some places where, to me at least, the results seem disappointing, for example in Figure 2 the spread between different datasets is wide and the altitude region of unambiguous increase is very small, and non-existent in the tropics. In Figure 4 the results from Lauder look terrible frankly. In Figure 7 the R-squared is disappointingly small, but still a good improvement on the previous report).

The chosen approach can be considered quite minimal. The use of a purely linear trend term rather than physical proxies such as GHG concentrations and chlorine and bromine concentrations in different regions of the stratosphere of course restricts the interpretation of the results. It is also an interesting decision that the regression model does not include a term for any large scale dynamical variability beyond QBO and ENSO. This choice also limits the interpretation of the results.  

The manuscript itself does seem somewhat rushed in its preparation to me. The core subject (satellite trends) I felt was of a much higher standard than both of the other main topics, namely the comparison with ground-based instruments, or the comparison with models, which are both treated quite cursorily.

There is an overall tendency for the text to simply describe the figures without adding much commentary to help the reader draw any conclusions. Section 5 was an exception and was very welcome, but only considered the core satellite trends.

An example giving the impression of a rushed writing process is that the details of the seasonal variation of proxies were never properly described – section 3 refers the reader to section 5 and section 5 to section 3.

Another serious omission in this reviewer's opinion is the lack of a comparison with contemporary analysis of total ozone trends (Weber, M., Arosio, C., Coldewey-Egbers, M., Fioletov, V., Frith, S. M., Wild, J. D., Tourpali, K., Burrows, J. P., and Loyola, D.: Global total ozone recovery trends derived from five merged ozone datasets, Atmos. Chem. Phys. Discuss. [preprint], https://doi.org/10.5194/acp-2021-1058, in review, 2022.)

I wonder if the pre-1997 trends could be included as a supplement? In the LOTUS report this was very useful for providing context to the post-2000 trends.  

I strongly dislike the use of the term 'supersite' throughout the manuscript, which is both unjustified and ill-advised. Table 1 states of the five European sites, two have only one instrument providing vertically resolved ozone measurements, two have two instruments and only one site has three. What is the basis then for calling them 'supersites'? The clear implication is also that other NDACC sites are in some way inferior (or at least, not as "super") and less worthy of support.  Further, figure 4 shows better agreement between the five different but geographically close European sites than between the single sites of Mauna Loa and Lauder, suggesting combining stations is the better way to go anyway for good results.  

Please check that the terminology of upper/middle/lower/lowermost stratosphere is being used consistently throughout the manuscript to refer to the same height regions each time.

Specific comments

Lines 10 'Finish' should be 'Finnish'

Lines 49-51 Saying "close agreement, especially over the European Alps" is an exaggeration, the agreement is hardly "close" even over Europe while it's worse at Mauna Loa and much worse at Lauder.

Line 54 Newman et al. 2007 had ODS peaking in the mid 1990s not the 'turn of the century'.

Line 56 WMO 2010 reported an increase in the NH upper stratosphere, albeit very cautiously.

Line 62 You should add WMO 2018 for discussion of Antarctic ozone trends.

Line 63 "Numerous" to me implies more than just four.

Line 72 "performed in the frame of the REF-B2 CCMI" is jargon.

Lines 74-76 The reader might struggle to see how 'enhanced mixing' can cause decreasing trends in both the tropics and extra-tropics, which is what you have said.

Lines 63-76 This section discusses recent work on trends in the lowermost stratosphere, but the reader might wonder why you don't also give any background for the other altitude regions of the stratosphere.

Lines 63-76 For this paragraph, personally I would prefer less detail (in particular relating to Ball et al. 2020) and more discussion of how the different results do or don't fit together. In other words, I like the the first half of the paragraph better than the second half.

Lines 91-93 I think this sentence would be clearer if re-arranged to ' … the northern hemisphere mid-latitudes of 2-3% per decade in the 5-1 hPa range and in the tropics of 1-1.5% per decade in the 3-1 hPa range'

Line 110 The implication of your wording is that the trend is maximum in winter because of photochemical control – the reader might have trouble following the reasoning.

Line 119 As I wrote earlier, I dislike the use of 'supersite' .

Lines 125 – 167 This section is fine and we don't want any more detail, but nonetheless the big questions seem skipped – (1) are these datasets stable over decades? (2) How successfully is it possible to merge records from different instruments?

Lines 170-208 Again, do we expect these instruments to be stable over decades?

Line 204 Why would you use a different trend model? That seems very odd. Couldn't the data have been supplied to you and run through your model for consistency?

Lines 219 Aerosol is a proxy term in your regression model though so won't this introduce an inconsistency? (Is the aerosol term significant in your regression model?)

Lines 232-232 It is surprising to me that you don't deseasonalize all datasets so you can treat them consistently. This approach seems to introduce a small but unnecessary inconsistency between the results for the different datasets.

Line 240 This link didn't work for me – it gave security warnings. Looking at the website there only seems to be a mixture of daily and monthly values for different time periods available – how did you treat these?

Lines 241-243 It would be interesting to know how well the piecewise linear approach actually could fit the data, ie did the residuals show any systematic departure from the three straight lines?

Lines 274-276  You don't comment on why they have become significant now? Is it four more years of data, improvements in the regression model, improvements in the data or something else?

Lines 286 – "eg" should be "ie"

Lines 341 "general good" – I wouldn't say "good" – my expectation is the trends from the different instruments would look more similar than this – if this is not the case than more explanation is needed to explain why not.

Line 346 This is an example of the text simply describing the figure without adding much to help the reader draw any conclusion.

Lines 351 I don't see how an upgrade in 2015-2017 is going to effect this??

Lines 355 Well can you use it then? (Or at least, should you put it on the same plot as the others?)

Lines 358 It is very hard for the reader to come away with a clear impression after this discussion. What does it all really mean?

Lines 368 Except stratospheric aerosol, right?

Lines 369 I worry you are "comparing apples and oranges" here. This uncertainty seems completely different to the LOTUS uncertainty. It would be comprised mainly of unforced variability in the models wouldn't it?

Lines 360-374 This section is also very cursory and it is left to the readers to form their own impressions.

Lines 375 I like this section a lot but it is limited in scope.

Lines 380-383 This is really good but I think it would be helpful to the reader to make clearer which uncertainties are contributing to these two terms (eg, disagreements between different satellites, instrument drift, lack of ability of the regression model to capture the variability etc).

Lines 434-440 The text seems to be verging into an advertisement for SPRAC activities.

Line 615 (Figure 2) Could there be some extra tick marks on the left-hand y-axis (pressure) please?