the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 03 Feb 2026
Characterizing variability and vertical structure of water vapor in the extratropical lower stratosphere
Abstract. Stratospheric water vapor strongly affects Earth’s radiative balance, especially in the extratropical lower stratosphere, yet large uncertainties remain regarding its variability, long-term trends and relationships to tropopause behavior. Here we characterize the seasonality, vertical structure, and variability of water vapor near the extratropical tropopause and the extratropical transition layer (ExTL). We analyze 17 years (2005–2021) of satellite observations from the Aura Microwave Limb Sounder (MLS) and the Atmospheric Chemistry Experiment Fourier transform spectrometer (ACE-FTS), with simulations from the Whole Atmosphere Community Climate Model (WACCM) using specified dynamics calculations. Analyses are performed in geometric and tropopause-relative vertical coordinates to assess the influence of coordinate on variability. Tropopause-relative coordinates substantially reduce variance in ExTL water vapor, but artificially enhance variability in the tropical upper troposphere. Tropopause-relative coordinates reveal a clear distinction between variability in the ExTL and the stratospheric overworld approximately 2.5 km above the tropopause. Above this level, there are hemispherically coherent anomalies linked to transport of air originating in the tropics. In contrast, ExTL variability lacks hemispheric coherence and likely reflects smaller-scale stratosphere-troposphere exchange. Interannual changes in ExTL water vapor show a statistically significant negative trend in both hemispheres in MLS data, but that behavior is not reproduced in ACE-FTS measurements or WACCM simulation. Despite this, the datasets agree well on overworld variability and regression responses to climate oscillations. The uncertainty in ExTL trends highlights the need for continued evaluation of long-term satellite records and improved model representation in this climatically sensitive layer.
- Preprint
(15036 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-412', Jian Guan, 09 Mar 2026
-
RC2: 'Comment on egusphere-2026-412', Anonymous Referee #2, 11 Mar 2026
Review of “Characterizing variability and vertical structure of water vapor in the extratropical lower stratosphere” by Emily N. Tinney and William J. Randel
The manuscript examines the variability of water vapor in the extratropical lower stratosphere using geometric and tropopause-relative coordinates. For this purpose, the authors use MLS, ACE-FTS, and WACCM-SD data. The hemispheric coherence is an interesting result. I suggest major revisions, particularly the inclusion of ACE-FTS and WACCM results in all analyses and figures. However, once the authors address these points, the manuscript should be ready for publication.
General concerns:
Why is the study limited to 2005-2021? It seems the model was run up to 2022. Is this due to the possible influence of Hunga in the results? If it is, this needs to mention in the text. If there is another reason, please include it in the text.
Why are you using PTGT? This is a new definition, please provide some background and add some details about it. What is the advantage of this tropopause over WMO or cold point tropopause? Perhaps the cold point tropopause will be a better tropopause for water vapor studies?
When using tropopause-relative coordinates to analyze water vapor variability, it is important to note that the natural variability of water vapor cannot be reduced by any coordinate transformation. Coordinate systems can only reduce the artificial variability introduced during data binning. This artificial variability arises from averaging data originating from different dynamical regimes. Although this distinction is subtle, it is important to reflect it clearly in the text; therefore, please revise the manuscript accordingly.
Please specify how are you computing the trend errors. The anomalies shown in Figure 1 will suggest big errors in the trends, but figure 13 shows statistical significance everywhere (at least for MLS). Note there is autocorrelation in the time series so you will need to take that into account.
Specific comments are:
L2 Add comma after long-term trends
L36 Tao et al 2023 and Yu et al 2022 are not the correct references for this, the importance of methane oxidation has been know for a while please add more appropriate references, such as, 10.1029/JZ055i003p00301, among others.
L40 I don’t think “alternatively” is the right word, the complexity of the transport in the LMS is not an alternative to the simplicity of the overworld, I think “in contrast” will work better
L48 add e.g., before Charlesworth et al 2023
Section 2.1. Are you using quality screening as recommended in Livesey et al 2022 to screen out bad data in the MLS record? If you are please mentioned it here.
Please summarize the water vapor drift here (which may influence your results). See 10.5194/acp-21-15409-2021
Also summarize the MLS water vapor validation efforts
L69-70: using the version 5.2 retrieval -> using version 5.2
L70: You didn’t provided the spectral coverage of the MLS instrument. If you think this is important for ACE-FTS please add it as well for MLS. Also summarize the ACE-FTS water vapor validation efforts
L81 What are the other problems? The impact of sampling biases could be mentioned here 10.1002/jgrd.50874, 10.5194/acp-18-4187-2018
Section 2.3 What is the temporal resolution of WACCM? Is there any study validation the WACCM water vapor? Is it representative of the atmospheric state despite the wet biases?
L109 the Gelaro et al citation seems to be out of place. Was it supposed to be “For ACE and MLS, a reference tropopause is computed from three-hourly MERRA-2 reanalysis data (Gelaro et al., 2017) and interpolated in space and time to each profile location” Gelaro et al does not interpolate to measurements locations.
L131-134: Please quantify the reduction of the H2O variance? You could simply show the standard deviation or the normalized standard deviation. (either in the text or in the manuscript). Note that this “reduction on variability” has been noted before 10.1029/2008JD009984 Please cite accordingly
L139 degree symbol not 30o
L138-145: It seems to me that in tropopause coordinates, the variability associated with the jets is binned below the tropopause (the two maxima around 30S and 30N). Also, the variability in polar regions seems to be clearly related to the tropopopause height, which will explain the large variability in tropopause relative coordinates in the polar regions.
I will not cause those results artifacts, it is simply a consequence of averaging the data differently and it gives you a sense of where the variability is coming from.
That said, the variability around the equator is surprising, where is it coming from. Could this be a consequence of the PTGT definition, did you explore the thermal or cold point tropopause, is that something you could consider?
L142 What is “low frequency changes”? Do you mean day to day changes or month to month, please clarify.
Figure 2 Please consider adding the ACE-FTS and the WACCM-SD results to see how robust the MLS results are?
Figure 3: presumably panels a,b,and c are in pressure and not in pseudo pressure.
What is the meaning of the curly brackets in the color bar? They are not mentioned in the text. Consider removing them.
L152 the equation should be (WACCM-MLS)/MLS*100
Figure 4 panel a not pseudo. The results are no fractional, they are in percent. The equation should be (WACCM-MLS)/MLS*100
Figure 5 remove curly brackets from the colorbar. Please show the WACCM and ACE-FTS results as well to show how robust the MLS results are
Figure 6 ordinate -> coordinate
Remove the curly brackets
Add ACE-FTS
L217 ExTL , (delete extra space)
Figure 7 Please show the ACE-FTS and the WACCM results as well.
Figure 9 Please show the ACE-FTS and the WACCM results as well.
Figure 10 Please show the ACE-FTS and the WACCM results as well.
Figure 11 Why is panel (a) the fourth layer above the tropopause rather than the third as in panel b and in figure 8 panel a?
Figure 12, Could you add the same analysis for ozone? Since that is discussed in figure 10
L269-270 & Figure 13: I think based on Figure 13 this sentence is not justified. ACE-FTS display non statistical significant trends in the region, even suggesting negative trends in the tropics (in the tropical regions where it can actually measure there are some hints of blue) and WACCM-SD magnitudes are so positively bias that it is hard to trust them.
That said, I think the authors could paraphrase what they have in the conclusions here, that is: The MLS decreases in the tropical upper troposphere seen in Fig. 13 also disagree with other infrared and microwave satellite observations, which show consistent specific humidity increases in this region over the recent decades (e.g., John et al., 2025; Allan et al., 2022). Likewise, results for ERA5 reanalysis show moistening of the tropical upper troposphere since 2000 (e.g., Li et al., 2024; Allan et al., 2022).
The authors could add 10.1029/2021GL097609 to the ERA5 references.
L316-318: perhaps something like: As a result, we view the negative trends with caution, as they may represent a potential retrieval artifact in the MLS data that should be further investigated.
Figure 15 Why are you only showing MLS, please show ACE and WACCM as in Figure 13 and 14
L292 extra space between ( and transition
L300 incomplete sentence
Citation: https://doi.org/10.5194/egusphere-2026-412-RC2 -
RC3: 'Comment on egusphere-2026-412', Laura Braschoß, 16 Mar 2026
Review of “Characterizing variability and vertical structure of water vapor in the extratropical lower stratosphere” by Emily N. Tinney and William J. Randel
The authors effectively highlight the importance, current complications and numerous open questions in characterizing water vapour in the extratropical lower stratosphere, providing a strong motivation for their investigation of vertical water vapour profiles, intercomparing model (WACCM) and satellite data (MLS & ACE-FTS). Notably, their results demonstrate that interannual ExTL water vapour variability is distinct from overworld variability and shows patterns that disagree between the datasets (specifically the MLS trend).
The paper is well-written and thoughtfully structured. The presented results take a meaningful step towards understanding LMS water vapour structure and variability and also provide multiple incentives for further investigations. Therefore, I think the paper is clearly of interest to ACP readers.
I have only a few comments/suggestions:- Line 142: "We find that much of the enhanced H2O variance in this region is due to low frequency changes in tropopause height tied to ENSO and QBO variations (Randel et al., 2000)." Please add that these findings are shown/discussed further in section 3.4.
- Fig. 5: I think it is currently very difficult to see the dry signals (tape recorder, latitudinal spread etc.) in this colour scheme (everything above 3.5 ppmv looks almost the same to me)...perhaps you could improve this by using different colours or different colour level spacing... For the contour lines: please check that the labels do not overlap (e.g. for the 400K isentrope in JJA).
- Line 223: "The different behavior for [missing word] is of course due to the large reservoir of below the tropopause, together with the source of overworld variability in the tropics, distinct from ozone." Could you please expand on that? (Do you want to imply that the amount of water vapour in the troposphere influences the importance of tropospheric variability contributions to ExTL variability?)
- Line 232: "While the overworld variations are consistent among MLS, ACE and WACCM, the ExTL variations are quite different..." Besides the trends not matching the one found in the MLS data, how are the hemispheric mean ACE and WACCM variations different from each other? (Which features differ?)
Generally, it would be good if a few more ACE-FTS and WACCM plots could be shown (especially for Fig. 7 and Fig. 9) to allow the reader to directly compare them to the MLS results. Given the number of figures, I think it would suffice to add these plots as supplementary material...
I am also currently missing information on the significance of the trends and correlations (what are the confidence intervals? And how were they computed?) - please add this information in some way.
Technical corrections:
- Caption of Figure 1: "at 45 N" -> " at 45°N"
- Line 125: " 45° N" ->" 45°N"
- Line 139: "near 30o N and S and near the equator" -> "near 30°N and S and near the equator"
- Line 155: "as highlighted in (Charlesworth et al., 2023)." -> "as highlighted in Charlesworth et al., 2023."
- Line 217: "and ExTL , as identified" -> "and ExTL, as identified"
- Line 271: missing word in the sentence that starts with "In general,..."
- Line 285: "...and hence have less impact on [missing word] (not shown)"
- Line 297: "In the ExTL and below, variability is tied to tropospheric [missing word]..."
- Line 300: "As a note, LMS ozone interannual variability is very different from [missing word] ,..."Citation: https://doi.org/10.5194/egusphere-2026-412-RC3
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 210 | 148 | 16 | 374 | 14 | 35 |
- HTML: 210
- PDF: 148
- XML: 16
- Total: 374
- BibTeX: 14
- EndNote: 35
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The paper by Tinney and Randel introduces a novel tropopause-relative coordinate to analyze the variability and vertical structure of stratospheric water vapor. This framework effectively removes a large portion of the variability associated with the seasonal cycle of tropopause height, allowing the variability of water vapor itself to be examined more clearly. Using this coordinate system, the authors investigate several aspects of lower stratospheric water vapor, which is the most radiatively important region. The manuscript is well written, clearly organized, and provides several interesting new perspectives on the variability of lower stratospheric water vapor.
My main concern relates to the potential long-term drift in MLS measurements in the tropopause.
Hurst et al. (2016) reported a drift in MLS H₂O since about 2010 using version 4 data. Later, Livesey et al. (2021) identified a drift associated with the 190 GHz sideband fraction and released an updated version of H₂O and N₂O (v5). Although the drift was substantially reduced, MLS H₂O still shows significant drift compared with frost-point hygrometer measurements, and N₂O still exhibits relatively large drift.
The large negative trend in the UTLS observed by MLS appears inconsistent with current observational evidence (as noted in the manuscript), theoretical expectations based on the Clausius-Clapeyron relationship, and model results. In addition, this negative trend does not seem to be explained by either ENSO or QBO as presented in the manuscript. ENSO shows a positive trend, while the negative QBO influence (Figure 15a) occurs at higher levels than the region of strongest negative H₂O trend (roughly one layer above and below the tropopause). Therefore, the possibility that the negative trend in this region is influenced by residual drift in MLS measurements cannot be excluded. I suggest that the authors cite the studies mentioned above and discuss this issue. It would in fact add scientific value if the manuscript explicitly raised this important issue and discussed whether the observed trend could partly reflect instrumental drift based on the current analysis; however, even a brief mention of this possibility would be helpful to provide broader context.
Line 195: The authors state that “There is a distinct separation of patterns in the vertical structure occurring around ∼2.5 km above the tropopause in both hemispheres” for H₂O (Figure 7), whereas O₃ shows a “coherent vertical structure in each hemisphere” (Figure 10). Since both H₂O and O₃ exhibit strong gradients near the tropopause, I would not have expected such markedly different behavior. The potential MLS drift mentioned above might contribute to this difference. I would be interested to see whether the structure in Figure 7 remains the same if the long-term trend is removed, and whether the vertical structure then becomes more coherent.
It is important to note that, regardless of whether such long-term drift exists, it does not diminish the novelty or importance of the manuscript. The focus of this study is primarily on short-term variability (seasonal and interannual), which would not be significantly affected by a slow instrumental drift.
There are also several minor comments:
There are also a few technical points: