Review for “Quantifying the Sources of Increasing Stratospheric Water Vapor Concentration in the 21st Century” by Sheese et al.

This study utilizes ACE-FTS satellite data to assess the contribution of several factors, including tropical tropopause temperature, methane, and the Brewer-Dobson circulation, to the trends in stratospheric water vapor over the 21st century. Given that many satellite missions are nearing the end of their lifetimes, the analysis using ACE-FTS data makes this work timely and valuable. However, the topic of stratospheric water vapor trends has been explored previously, and some of the analyses in this paper could be refined. I recommend a major revision before the paper can be accepted.

Major Comments

1. The Title:
   
   Since the study covers the period 2004–2021, referring to the “21st century” in the title may be misleading. I suggest rephrasing the title to reflect the time span analyzed more accurately.
2. Lines 3 & 49: 'None of them parse the trend'
   
   This statement is not entirely accurate. Yu et al. (2022) quantified the contributions of TTL temperature and methane oxidation to stratospheric water vapor trends. Please revise this part accordingly to acknowledge previous relevant studies.
3. Contribution of Brewer-Dobson Circulation and Tropical Tropopause Temperature:
   
   This section needs further clarity, as the results are somewhat confusing.
   • I recommend adding time series of tropical tropopause temperature and Brewer-Dobson circulation, similar to what is presented in Figure 1, to provide additional context.
   • Studies like Fu et al. (2019) and Polvani et al. (2018) report a slowdown of the Brewer-Dobson circulation during the first two decades of the 21st century. However, in your study, the Brewer-Dobson circulation is shown to increase stratospheric water vapor in certain regions (e.g., Southern Hemisphere high latitudes, Northern Hemisphere mid-latitudes) and decrease it elsewhere. Could you provide further explanation or discussion to reconcile this difference?
   • Separating the contributions of the Brewer-Dobson circulation and tropopause temperature is challenging, as a weaker Brewer-Dobson circulation can lead to lower tropopause temperatures. I suggest conducting a more careful analysis to distinguish the respective influences of these two factors.
4. Sampling and Reliability of Water Vapor Trends:
   
   The paper mentions that the ACE-FTS instrument has low sampling in the tropics, and that CMAM is used to estimate CH₄ trends. I recommend addressing the following points:
   • In Section 2, please provide more detailed descriptions of the H₂O sampling over different latitude bands.
   • It would also be helpful to include an analysis of how reliable the tropical water vapor trends are, given the known sampling limitations.
5. Line 114: Tropopause Height and Seasonality
   
   Since the height of the tropopause varies with season, have you accounted for seasonal variability? Additionally, have you tested the results using higher altitudes to ensure robustness?

Specific Comments

1. Line 17:
   
   There is a typo in "greenhouse gases."
2. Line 82:
   
   The SWOOSH database (Davis et al., 2016) also utilizes ACE-FTS H₂O data.
3. Line 219:
   
   The statement, “with an increase of less than 1% per decade near the tropical tropopause region,” raises a question. Methane does not undergo oxidation in this region, so how can its contribution be non-zero?
4. Line 253 (Suggestion):
   
   As ACE-FTS provides HDO data, you might consider using it as a proxy for deep convection, following Hanisco et al. (2007). While this analysis may be beyond the scope of the current paper and the results may be uncertain, it could be worthwhile to explore how HDO could provide insights into the role of deep convection in the observed trends.

References

• Davis, S. et al. (2016). The Stratospheric Water and Ozone Satellite Homogenized (SWOOSH) database: A long-term database for climate studies. Earth System Science Data, 8(2), 461–490. https://doi.org/10.5194/ESSD-8-461-2016
• Fu, Q. et al. (2019). Observed changes in Brewer-Dobson circulation for 1980–2018. Environmental Research Letters, 14(11), 114026. https://doi.org/10.1088/1748-9326/ab4de7
• Hanisco, T. F. et al. (2007). Observations of deep convective influence on stratospheric water vapor and its isotopic composition. Geophysical Research Letters, 34, L04814. https://doi.org/10.1029/2006GL027899
• Polvani, L. M. et al. (2018). Significant weakening of Brewer-Dobson circulation trends over the 21st century as a consequence of the Montreal Protocol. Geophysical Research Letters, 45(1), 401–409. https://doi.org/10.1002/2017GL075345
• Yu, W. et al. (2022). Variability of water vapor in the tropical middle atmosphere observed from satellites and interpreted using SD-WACCM simulations. Journal of Geophysical Research: Atmospheres, 127, e2022JD036714. https://doi.org/10.1029/2022JD036714

The paper by Sheese et al., "Quantifying the sources of increasing stratospheric water vapor...", utilizes ACE-FTS data to quantify stratospheric water vapor trends from 2004 to 2021. Given that ACE-FTS is still operational—unlike MLS, which has already experienced significant data reductions—publishing research on stratospheric water vapor trends is essential. This work is critical not only due to the climate relevance of stratospheric water vapor but also because ACE-FTS data may help to fill potential data gaps in the future. An interesting finding of the paper is the statement that "the residual trend could potentially be explained by increases in upper tropospheric molecular hydrogen."

There are two main reasons I recommend a major revision of the paper:

First, although the findings related to stratospheric moistening after 2000, its hemispheric asymmetry associated with changes in the Brewer-Dobson circulation, the tropical cold point temperature, and CH4 oxidation are interesting, they are not entirely novel. These phenomena have already been discussed in previous studies, including Konopka et al., 2022, and Tao et al., 2023.

Konopka et al., 2022, doi.org/10.1029/2021GL097609

Tao et al., 2023, doi.org/10.1038/s43247-023-01094-9

Second, there is a need to explain more critically all the additional assumptions beyond the multiple linear regression model that were used to conclude the potential importance of hydrogen trends (see below).

Minor points:

Title: The phrase "in the 21st century" may be misleading, as only 25 years of data are available. Please consider rephrasing.

Abstract: The sentence "Previous studies have estimated stratospheric H2O trends, but none have simultaneously quantified..." is inaccurate (see the point above on existing studies).

Abstract: The phrase "these unaccounted... sources could potentially" may need rephrasing for clarity.

Methodology: The multi-linear regression model is not well explained. For example, the annual oscillation term appears to mix harmonics and regressors. Additionally, there is no explanation of the time lags, which are introduced later to improve the analysis.

L225-234: This section represents the most critical part of the manuscript, particularly in its attempt to extend the definition of d[CH4]entry/dt by incorporating time lags. This approach differs from that in Fig. 5, where time lags are defined by minimizing the residuals between the data and the multi-linear regression (MLR) model. In this section, however, the same time lags are applied to redefine d[CH4]entry/dt, making their use appear somewhat "a posteriori." It is unclear why the authors did not apply the well-established formalism for analyzing stratospheric trends introduced in Hegglin et al., 2014, and subsequently applied in many studies, such as Poshyvailo-Strube et al., 2022 (doi.org/10.5194/acp-22-9895-2022) and Tao et al., 2023 (see reference above). Additionally, the term "alpha" is defined differently in this paper compared to previous references, which may lead to confusion.

L245-250: The complexity further increases when discussing variable values of alpha. The main objective of multi-linear regression is to minimize the residual by simultaneously varying all relevant parameters, including the lag times and the alpha parameter. A critical discussion justifying the "simplified procedure" used in this paper is necessary.

This paper presents ACE-FTS satellite observations of stratospheric water vapor and investigates trends in these during the last about two decades. Based on these observations it is shown that water vapor mixing ratios have significantly increased during this period by 1-5%/dec, depending on the region. Moreover, a multi-linear regression method is applied to attribute these trends to different causing factors, like changes in tropical tropopause temperatures, stratospheric circulation and atmospheric methane concentrations. All of these are found to contribute to the net trends, with the tropopause temperature contribution most important in the tropical lower stratosphere, methane oxidation most important in the middle to upper stratosphere and circulation changes in the lower to middle stratosphere. Furthermore, the stratospheric circulation change during the considered period causes a hemispheric difference in the water vapor trends. It is finally argued that the unexplained trend residual could potentially be related to an increase in atmospheric hydrogen.

I find this a very interesting study on a highly timely subject, i.e. stratospheric water vapor trends and their causing factors, which without doubt falls within the scope of the journal. It is particularly important to explore different observational datasets for their capabilities to address this subject, as done here. In that sense, this paper is an important contribution to the field and will be of great relevance for the community. The paper is well written and the presentation is concise and clear. I have one more general but minor and a few specific comments below which I think are easy to address and will hopefully help to further improve the paper. After having addressed these comments I will strongly recommend publication.

Minor comment (Description and discussion of the methodology, Sect. 3):

I have one general and a few more specific comments here. My general comment concerns the choice of regressors in the MLR method: To what degree are the chosen regressors independent? In particular there are interrelationships between tropical tropopause temperatures and stratospheric circulation (here N2O is chosen as proxy for these), as well as between these two and annual cycle, QBO, ENSO, etc. I know that such a regression approach is applied frequently and I don't argue for changing the methodology. But I think it would be worth discussing these aspects (and the ones below) in more detail, either in the method section or later in a discussion section.

Another issue which could be discussed in more detail is the effect of ENSO. Aren't ENSO effects on stratospheric water vapor likely lagged (compared to the ENSO index used in the MLR method), in particular when considering upper stratospheric levels and high latitude regions. Hence including ENSO in the regression could likely be improved by taking into account lag time, as is done already for the impact of tropopause temperatures. Again, I'm not suggesting new analysis but only a more detailed discussion.

Also concerning the calculation of the methane oxidation effect there is an approximation applied but not clearly explained. The exact calculation of the CH4-entry values in Eq. 2 would require the concolution with the stratospheric age spectrum. As this is not available, here a lag time is used as approximation. A recent paper discusses the effect of approximating the propagation of methane entry mixing ratios in a different context (Poshyvailo-Strube et al., 2022, https://doi.org/10.5194/acp-22-9895-2022). I'd suggest to discuss also this approximation in a bit more detail. 

A more technical comment concerns the notation in Eq. 1: I find the indices i, j here somewhat confusing. In particular, shouldn't there also be a sum over j? Later in L94, aren't the cos and sin functions the r_i's in Eq. 1, so shouldn't there be a similar index i in these and the beta-coefficients? In Eq. 4 the j-index is just 2 for AO and SAO, is this correct? Please explain the notation more clearly.

Specific comments:

L10: I'd precise the formulation here: "increases in the Northern Hemisphere below about 30km and decreases above and in the Southern Hemisphere throughout the stratosphere"

L30: Actually, the first study emphasizing the general point of stratospheric dehydration at the tropical tropopause would be Brewer (1949), the main point of the paper cited here is the impact of regional temperature anomalies. I leave the decision what to cite here to the authors...

L79: isotopologues

L114: How can the water vapor time series over 2004-2021 be regressed against a temperature time series which ends in 2018? Please clarify.

L171: To better distinguish the circulation effect from the effect of tropopause temperature changes I'd suggest to write here "structural Brewer-Dobson circulation changes". Perhaps this would be a better formulation also at other places in the manuscript.

L184: I was a bit confused about this sentence, as "residual trends" are shown only in the right panel of Fig. 2. I'd suggest to write either "right panel of Fig. 2" or just "Fig. 2".

L192: Also here I don't find the formulation very clear. My suggestion: "The remaining trends are all increasing (Fig. 3, middle panel), but the hemispheric asymmetry in the lower stratosphere flips sign and remaining trends are more strongly positive in the SH (about 3%/dec) compared to the NH (1-3%/dec). The difference ... circulation, showing that the net hemispheric asymmetry in water vapor trends can be attributed to changes in stratospheric circulation. In particular, the circulation influence ... " These findings are consistent with results of another recent paper by Tao et al. (2023, https://doi.org/10.1038/s43247-023-01094-9) based on different data and I'd suggest to mention that here.

L231: I think the correct reference here is "Fig. 6b".

L273: To me the H2O increases at high latitudes and altitudes between 25-35km in the SH due to circulation changes in Fig. 3c seem very small and almost negligible. Thus, I'd not highlight them here in the conclusions.

L287: Maybe worth to incude a sentence after "...instruments." similar to: "There is urgent need for new satellite missions to continue the observational record for more reliable trend estimation, as are model studies to determine ..."