This paper presents a new dataset of stratospheric age-of-air obtained from ACE-FTS satellite measurements of SF6. The dataset spans the period from 2004 to 2021, making it the longest satellite record of the age of air. The raw SF6 measurements and methodology are described in detail, and a comparison with other satellite measurements is provided.

In my opinion, the approach is sound, the paper is well-written, and the figures are clear and well-designed. I recommend publication pending a few minor revisions described below.

My two concerns are as follows:

1. There is limited discussion of the uncertainty in the final age product. The SF6 measurements are presented with error bars representing the sampling uncertainty from ACE, but these are not extended to the AoA product. Additionally, a discussion of estimated uncertainty arising from the SF6 retrieval process and its impact on AoA is missing.

2. A potential weakness in the analysis, particularly in extracting different signals of variability from the time series in Figures 13, 14, 16, and 17, may stem from the limited sampling by ACE of the time-evolving SF6 field. This field is subject to synoptic variability associated with atmospheric motions (e.g., isopleths deformed by Rossby waves). For MIPAS, the redundancy of daily measurements mitigates this effect, but this is not the case for ACE. It seems straightforward to address this issue by using equivalent latitude (and potential temperature) (see, e.g., Allen and Nakamura, 2003; Hegglin et al., 2006).

    

Other minor comments/typos:

• Line 1: "increased" → "increasing"

• Line 32: "faster transit times" → "shorter transit times"

• Lines 32-34: The sentence "Since neither models nor observations allow the examination of infinitesimally small air parcels, it is necessary to address the fact that any region under consideration is made up of air of different ages." is confusing and may be deleted altogether (see comments below regarding infinitesimally small air parcels).

• Line 39: Consider changing "calculated" to "estimated."

• Line 105: "high-resolution" → "high spectral resolution"

• Lines 120-123: It might be useful to provide more information regarding the sampling by ACE.

• Line 150: Have the authors tried different distances in the longitudinal and latitudinal directions? Using equivalent latitude may be an option here.

• Line 166: "one standard error" → "one standard deviation"

• Line 173: Change "Fig. 1" to "Figure 1," as per ACP convention (it is the subject of the sentence).

• Figure 1: It might be worth showing the average for the entire MIPAS dataset to get a sense of the uncertainty caused by ACE sampling.

• Line 261: Consider removing "...were actually infinitesimally small and therefore...". In practice, it is only at the molecular/free path scale that mixing/diffusion, and thus the mixing ratio, does not make sense.

• Line 267: Consider using another symbol for the ratio of moments instead of $w$.

• Line 267: Add "years" to "constant value of 0.7".

• Line 282: Shouldn't the left-hand side of Equation 2 read [SF6]modeled? Further detail on the numerical evaluation of the integral should be provided, such as the discretization step along transit time and the actual upper boundary of the integral (which I imagine is finite).

• Line 292: As far as I understand, this is Newton's method. You might mention this explicitly.

• Line 375: Which tracers would not be affected by this difference in vertical gradients?

• Lines 377-378: I recommend either making the first half of the sentence more quantitative by mentioning the accuracy of the different measurements or removing it.

• Figure 11: Would it be possible to generate a similar wing plot for the SF6 measurements during the MIPAS time period? As it stands, it is not entirely clear where this difference in age stems from.

• Line 540: For the impact of the sink correction on the trend, Loeffel et al. (2022) may be a better reference.

• Line 575: Is this impact of decadal variability in MIPAS also found in ACE data?

Allen, D. R., and N. Nakamura, 2003: Tracer Equivalent Latitude: A Diagnostic Tool for Isentropic Transport Studies. J. Atmos. Sci., 60, 287–304, https://doi.org/10.1175/1520-0469(2003)060<0287:TELADT>2.0.CO;2

Hegglin, M. I., Brunner, D., Peter, T., Hoor, P., Fischer, H., Staehelin, J., Krebsbach, M., Schiller, C., Parchatka, U., and Weers, U.: Measurements of NO, NOy, N2O, and O3 during SPURT: implications for transport and chemistry in the lowermost stratosphere, Atmos. Chem. Phys., 6, 1331–1350, https://doi.org/10.5194/acp-6-1331-2006, 2006

Loeffel, S., Eichinger, R., Garny, H., Reddmann, T., Fritsch, F., Versick, S., Stiller, G., and Haenel, F.: The impact of sulfur hexafluoride (SF6) sinks on age of air climatologies and trends, Atmos. Chem. Phys., 22, 1175–1193, https://doi.org/10.5194/acp-22-1175-2022, 2022

This manuscript describes a new stratospheric age of air (AoA) dataset derived from ACE-FTS measurements of SF6.  Spanning 2004-present, this is the longest continuous AoA dataset available, and it represents a timely and valuable contribution to the ongoing question of whether AoA is changing with time.  Overall, I find the manuscript to be well-written and the presentation of the derivation of AoA from SF6 and comparison to satellite and in situ measurements to be robust and informative, though lacking key details regarding sampling density.  However, I find that there are significant issues with the analysis and discussion of AoA trends that warrant major revisions prior to publication in ACP.

Firstly, I find the discussion of trends based on comparisons to balloon measurements to be a significant over-reach, even with the few caveats provided about the representativeness of the data.  Given the large interannual variability in AoA shown, for example, in the manuscript’s own Figures 13 and 14, it is impossible to draw inferences about trends from 6 profiles taken at different times and locations.  The balloon comparisons are useful for examining the vertical structure of AoA, but should not be used as a basis for inferring changes in AoA with time.  Secondly, the analysis of trends using the full ACE-FTS dataset requires 1) either a different approach than using the solar cycle to represent unknown decadal-scale variability or a much more robust justification for doing so, as it seems to actually introduce some decadal-scale variability that may not be present in the data, and 2) the provision of much more information regarding the statistical robustness of the trends shown.  Given the current focus on trends in AoA as potential markers of changes in the stratospheric circulation, it is crucial that the trend analysis be carefully and judiciously considered

My specific comments are as follows:

Lines 25-27: The statement “this has thus far only been indicated by measurements in the northern lower stratosphere” seems to contradict lines 88-90, which say, based on both the Ray et al. study and MIPAS observations, that “the shallow branch in the southern hemisphere does appear to have accelerated.”

Lines 35-36: “Mixing”, which appears here to refer to large-scale mixing of air from different regions (see a comment below on the use of this term) has a one-sided effect on the spectrum (you cannot get younger ages through mixing) and thus leads to a change in the mean age in addition to widening the spectrum.

Lines 36-38: The manuscript uses the parenthetical “(with no change in the Tropics)” in several places when discussing how mean age trends in the high latitudes are related to changes in the circulation.  The authors need to provide a brief narrative on why this parenthetical is needed (it is because only changes in tropical-extratropical age gradients actually reflect changes in the circulation independent of large-scale mixing changes) and reference the proper papers on this (primarily written by M. Linz).

Line 39: Please add “monotonically” before “increasing”

Line 142: Consistency with what?

Lines 170-171: The decrease in SF6 below 14 km in ACE-FTS is mentioned several times in the manuscript.  The authors should describe the basis on which it is determined to be unrealistic.

Line 204: It would be clearer if the heading were “AoA Methodology” or “Methodology for Deriving AoA”

Line 208: I do not understand what is meant by “separated by year” here – can you please clarify?

Lines 234-239 and 258-259: 2 km below the tropopause probably successfully separates stratospheric and tropospheric air for ACE-FTS, but does it really work for MIPAS with its significantly coarser vertical resolution?

Line 261: Here the term “mixing” is used to describe turbulent-scale interactions between air parcels rather than large-scale stirring that would still transport an infinitesimally small parcel to different regions of the stratosphere.  Since the word mixing is used to describe what is actually large-scale stirring in other places in the manuscript, the distinction should be made clear here.

Line 272: For clarity “than ERA-Interim” should be added after “for ERA-5”

Figure 4 and all latitude-height cross sections: For the purpose of understanding the robustness of the results, companion figures should be provided in the Supplementary materials that provide the number of observations in each bin.

Line 416: The phrase “Aside from this difference” does not make sense here because a difference was not actually described in the previous sentence.  The sentence beginning on Line 416 stands alone without that phrase.

Line 417: Isn’t it self-evident that the difference at 30.5 km is related to differences in the vertical gradient?

Lines 418-428: I realize that the authors are using the word “could” here and that they point out here and in lines 429-430 that the variability is large, but I do not think these profiles should be used to say anything about trends.  In fact, in the conclusions (lines 591-593), they are considerably more bold, stating “Comparisons with age of air derived from balloon measurements of SF6 and CO2 from before the ACE-FTS mission suggested that the shallow branch of the BDC might be accelerating, and possibly that the deep branch might have slowed between 1995 and the mid-2000s” before reminding the reader that this is based on comparisons with single balloon flights. The variability in AoA is so large (see Figures 13 and 14 of this manuscript) that single-point comparisons like these should not be used even speculatively to infer trends.  I recommend that the authors focus on the vertical profiles of age for the balloon-based comparisons.

Lines 471-473: Is it not also possible that this could be due to an issue with the vertical age gradients in ACE-FTS AoA seen in Figure 10? Those results compare two uncorrected SF6 AoA datasets and so should not be influenced by problems in accounting for the mesospheric sink.

Lines 482-484: Again, the number of measurements in each bin (as a function of time) should be provided in the Supplemental materials.

Lines 495-498 and 508-514: I simply do not understand the motivation for using the solar cycle as a proxy for decadal-scale variability of undetermined origin. The solar cycle has a very distinctive and asymmetric signal and this methodology imposes that signal without a clear physical reason for doing so.  Looking at Figure 13 panels E and F, which are called out in lines 525-526 as being most noticeable with respect to the decadal variability, it is clear that the residuals in the first half of the dataset, when the solar cycle term causes the fit to start low and increase with time, are more positive than the residuals during the second half of the time series when the solar cycle term causes the fitted timeseries to decrease. So I see the decadal variability in the fit but not in the data.  Why didn’t the authors remove the other variability and then try to fit a harmonic to the decadal-scale if they wanted to try to account for it.  From my perspective, the clear message about the decadal-scale variability is that the trends we quantify from 17 years of data are highly unlikely to be significant, not that we should be using unphysical methods to try to remove it.

Line 508: “enso” should be “ENSO”

Lines 533-536: The actual slopes and uncertainties should be provided either in the figures themselves or in a Table, and the results should be discussed in terms of whether the changes are significant or not.

Lines 565-566: Again, if these trends are going to be used to infer changes in the circulation then a discussion of the magnitudes and uncertainties is required.