Long-term evolution of stratospheric water vapor from the Hunga eruption

Dubé, Kimberlee; Randel, William; Bourassa, Adam; Tegtmeier, Susann; Wang, Xinyue; Zhang, Jun; Hlady, Eilidh; Brehon, Meghan; Khaykin, Sergey; Degenstein, Douglas

doi:10.5194/egusphere-2026-4022

Preprints

https://doi.org/10.5194/egusphere-2026-4022

Preprints

06 Aug 2026

| 06 Aug 2026

Long-term evolution of stratospheric water vapor from the Hunga eruption

Kimberlee Dubé, William Randel, Adam Bourassa, Susann Tegtmeier, Xinyue Wang, Jun Zhang, Eilidh Hlady, Meghan Brehon, Sergey Khaykin, and Douglas Degenstein

Abstract. The Hunga underwater volcanic eruption in January 2022 injected 150 Tg of water vapour (H₂O) into the stratosphere, increasing the total stratospheric H₂O mass by 10 %. This study investigates the transport of the Hunga H₂O within, and out of, the stratosphere from 2022–2025, using H₂O observations from the Microwave Limb Sounder (MLS) and the Atmospheric Chemistry Experiment – Fourier Transform Spectrometer (ACE-FTS), along with model simulations from the Whole Atmosphere Community Climate Model (WACCM) and the FLEXible PARTicle dispersion model (FLEXPART). The Hunga H₂O is isolated by using the tropical cold point temperature to account for H₂O that entered the stratosphere through the tropical tropopause, rather than via the eruption. The resulting residuals show the detailed evolution of the Hunga H₂O over time while it moves to higher latitudes and lower altitudes. Including the lower stratosphere, approximately two-thirds of the Hunga H₂O remains in the stratosphere in late 2025. There is good agreement between the observed and modelled H₂O above 20 km. However, in the lower stratosphere observations show substantially more H₂O compared to the model during 2024 and 2025, suggesting that the modelled transport across the lower stratosphere is too fast. Observations show excess H₂O in the tropics in 2023 and 2024, suggesting evidence of recirculation from SH mid-latitudes back to the tropics. This mixing signature is found in WACCM simulations and confirmed in FLEXPART transport calculations. Including the lower stratosphere in the calculation increases the decay time scale of the Hunga stratospheric water vapor by 1–2 years.

Received: 06 Jul 2026 – Discussion started: 06 Aug 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Kimberlee Dubé, William Randel, Adam Bourassa, Susann Tegtmeier, Xinyue Wang, Jun Zhang, Eilidh Hlady, Meghan Brehon, Sergey Khaykin, and Douglas Degenstein

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-4022', Hugh C. Pumphrey, 27 Aug 2026
General comments
This paper considers the fate of the water vapour deposited in the stratosphere by the 2022 Hunga eruption. This topic has been considered in a number of earlier papers; the paper reviewed here makes further progress by more detailed modelling studies and by using a more sophisticated technique to separate the Hunga water from the background.
The main scientific suggestion that I would make would be to re-work the plots using the newly-released version 6 of the MLS data. Version 6 is very similar to version 5, but the artefact at ∼33km altitude has been eliminated, so the process to remove the artefact recommended by Millán et al. (2024) is not required. While the authors would probably regard version 6 as more satisfactory I understand that it may be too much work to re-do all of the analysis, so the paper could be published using version 5 if the authors consider that this is the most practical way forward.
The paper appears generally well written and illustrated. I consider that it should be accepted subject to some minor corrections.
Specific comments
Line 77: "Version 5 of the MLS retrieval is used here.'' As noted in the general comments I would recommend re-working the paper using version 6 of the data.

Line 157: I found figure A1 to be rather illuminating; it seems a shame to relegate it to an appendix.

Lines 163--64: I think that the first paper to report the water vapour injected by the ANY event was Schwartz et al. (2020) ( https://doi.org/10.1029/2020GL090831 ) At any rate it was published ahead of the papers currently cited here.

Some of the multi-panel figures (especially figures~4 and 6) contain a lot of small detail. The authors should ensure that they end up printed large enough. Journal layout staff sometimes seem to prefer huge white margins to sufficiently large detail on two-column figures; the authors should resist this tendency.

Technical corrections
Line 84--85 "(Millán et al., 2024)'' should be "Millán et al. (2024)'' i.e. the citation should be a \citet{} and not a \citep{}. (If the authors follow my suggestion to use version 6 this sentence will be removed anyway.)

Line 96 and elsewhere: Units should appear in the same font as the running text. The fact that they do not is a long-standing bug in the LaTeX package supplied by the journal; the authors should not waste their time trying to fix it as it will be fixed anyway when the final version of the paper is typeset.

Figure 1 The legend items for "ACE'' in (a) and "Residual ACE'' in (b) do not match the plot in that the plot has only dots and the legend items have a line as well.

Line 174: km is a unit and should appear in a non-italic font, as is already the case for ms^-1 and gmol^-1. (As noted

above, the journal's LaTeX package will use the wrong non-italic font; the authors should not worry about that.)

Lines 216, 322, 377: Replace "likely'' with "probably''. Likely should not be used as an adverb in formal writing; it is an adjective synonymous with "probable''. (Yes, this is a personal beef with all of North America which I have no hope of winning.)

Line 229: "(shown in Figure 7)''. Journals usually expect the first mentions of figures to be in order, in that the first reference to figure n should come after the first mention of figure n-1 and before the first mention of figure n+1. There are a variety of ways to address this. One possibility is to put the current figure~3 and the current figure~7(b) into an extended figure~3; this would make the paper clearer and easier to follow.
Citation: https://doi.org/10.5194/egusphere-2026-4022-RC1
CC1: 'Comment on egusphere-2026-4022', Graham Mann, 05 Sep 2026

I am making here a community comment on this study, mainly in relation to what I consider to be a substantial change this study is reporting, in relation to the picture that is emerging from this new analysis of the latest satellite measurements (MLS with ACE-FTS).
I make 3 main comments in this comment, and do so entirely as an individual, whilst noting my recent international leadership role as one of the four co-ordinating scientists for the World Climate Research Program’s Hunga atmospheric impacts report (APARC, 2025).
The 3 comments each highlight a separate point, but the 1^st sets the over-arching theme towards the study’s finding of an increased longevity for the ~10% anomaly in global stratospheric water vapour that resulted from the January 2022 Hunga eruption.
The aim here is towards asking the authors to comment on these over-arching issues, with each comment also stemming from specific findings/Figures from the submitted manuscript.
1) Re: the finding from last sentence of the Abstract (line 14): this new satellite measurements analysis “increases the decay time scale of the Hunga stratospheric water vapour by 1-2 years”

The authors state categorically the cause of this change (compared to other recent studies, e.g. Schoeberl et al, 2024; Zhou et al., 2024; 2026) is their methods include the lower stratosphere in the calculations, aligned to the statement on lines 10-11 “In the lower stratosphere observations show substantially more H2O compared to the model during 2024 and 2025”.
That model-obs difference is apparent from comparing Figures 7a (where 60S-60N mass burdens are calculated upwards from 2km above the tropopause) with Figure 7b (same 60S-60N mass burdens, but calculated only above 20km).
The Figures 7c and 7d are clear that both MLS and ACE-FTS are seeing the emergence of an extra 25-30 Tg of water vapour in the Southern Hemisphere lower stratosphere (30-80S), not represented in the model simulations.
The above I agree fully with, but the main comment I make here, is to query the interpretation the authors have given to (fairly definitively) identify this difference as evidence of recirculation from SH mid-latitudes back to the tropics.
Firstly, the Figure 7c includes latitudes into the polar regions, through to 80oS, and it is quite possible that a good portion of that emerging water vapour could be simply the arrival of Hunga excess water vapour in the deep branch Brewer Dobson circulation. The timing to emerge in 2024 and 2025 would be consistent with timescales associated with deep branch BDC southward meridional transport from the tropics in the upper stratosphere, and subsequent descent in the wintertime polar vortex.

The counter-argument might be that we should see a discrepancy earlier, if that were the case, but with the units being shown here being mass burden, any potential model-observation discrepancy in H2O vapour in the upper stratosphere would be less apparent, with the thinner air in the upper stratosphere and lower mesosphere.
When the air descends however, any discrepancy propagating within the downwards transport into the Antarctic polar vortex would then emerge as a difference in mass, with the increasing pressure.
This point relates also to the 2^nd point I make below, that all Figures of the vertical profile (Figures 2, 4, 6 and 9) all show the Hunga-excess water vapour in mass-burden terms.
That choice of unit implicitly weights towards the lower stratosphere, and any model-observation differences in the upper stratosphere would not be apparent when plotted with this unit.Please see the 2^nd point where I discuss this further
This 1^st point however I note also another possible cause of the lower stratosphere model-observation discrepancy, in relation to this timing of emergence aligning with when the El Nino Southern Oscillation progressed into the warm El Nino phase.
Neither the coupled-model or specified dynamics WACCM simulations described in section 2.4 will be representing this transition into El Nino, and then the transport of water vapour within the tropical upwelling would be expected to stronger than represented in the simulations (on average), and this could potentially be part of the difference seen in Figures 7c and 7d.
The timing of emergence is potentially consistent also with when the Hunga aerosol descended to the tropopause, for a tropopause heating of the tropical lowermost stratosphere alleviating the freeze-drying effect, allowing extra water vapour into the lower stratosphere (e.g. as demonstrated for Pinatubo in Zhou et al., 2023). However, the magnitude of the aerosol absorptive heating would be less here, and this seems unlikely to be a significant cause of the observed post-2024 increase.
In summary, this 1^st point is to ask the authors to adjust the certainty of their attributing causative statements, when interpreting the differences seen between the observations, and these particular model simulations.
I’d argue this difference is certainly motivating additional model studies to understand the cause of this emerging discrepancy, but please can the authors make the interpretation statements more open as to other potential causes.
2) Mass burden units shown in vertical profile Figures (2, 4, 6, 9) weighted to lower stratosphere

In this 2^nd point, I am further noting that considering model-observation discrepancies only in mass burden terms, is implicitly weighting differences to those in the lower stratosphere.
I realise the study is focused to understand the progression of the mass of water vapour in the stratosphere, and I agree totally that mass burden is definitely the variable to consider also.
However, when considering vertical gradients in the stratosphere, the 10-fold difference in pressure between the upper stratosphere and the lower stratosphere, and the deep branch of the Brewer Dobson circulation transporting air upwards into much lower pressure air, requires to consider the vertical gradients in mass mixing ratio terms.
I have attached here a screen-shot PNG of Figure 3.2 from the APARC Hunga atmospheric impacts report (Bourassa, Khaykin et al., 2025), which illustrates how the Hunga-excess water vapour was transported upwards into the upper stratosphere during 2023, consistent with the timing of transition into El Nino (see point 1 above), and the absence of aerosol-absorptive heating from the Hunga water vapour layer separating from the Hunga volcanic aerosol.
What’s clear from the Figure, is the transport of the Hunga-excess water vapour into the upper stratosphere and mesosphere, where the air is much thinner.
Atmosphere and climate scientists less familiar with stratospheric composition and circulation, may not realise this transition, and indeed for volcanic impacts, for all other large eruptions seen in the satellite era, impacts have tended to be from volcanic species being transported in the shallow branch of the stratospheric circulation, where there is no significant change in pressure.
This 2^nd comment is to ask the authors to cross-check for any anomaly between the model and observations in mass mixing ratio terms, specifically perhaps an extra Figure in the same format as Figure 4, but with the 4 columns showing 4 selected latitudes, perhaps 50N, 20N, 20S and 50N, to check the northern and southern tropics, and the northern and southern mid-latitudes?
For these view-graphs, the vertical co-ordinate should extend up to 60km, to be able to see the transport up into the upper stratosphere and lowermost mesosphere.

3) The final point I would like to make, in this community comment, is to note a close analogue situation to the current post-Hunga period, from the vertical profiles of radiocarbon emitted into the stratosphere from thermonuclear testing in the 1950s and early 1960s.
Specifically I refer the authors (and other readers of this journal) to the 1989 JGR study by Harold Johnston (Johnston, 1989), analysing the progression stratospheric circulation of excess concentrations of long-lived radionuclides in the quiescent periods of the initial testing moratorium, in 1958-1960, and the years after the subsequent partial test ban in October 1963.
The very extensive programs of in-situ balloon-borne and aircraft stratospheric monitoring of these radionuclides in the 1950s and 1960s (e.g. Friend, 1966, Telegadas et al. 1969) in these two periods, provide useful observational constraints for the circulation of stratospheric aerosol, and in the case of radio-carbon, tracing specifically air circulation relevant to this study’s analysis of the Hunga excess water vapour progression in the last 4 years.
Radio-carbon (specifically carbon-14) was monitored extensively, and whereas long-lived fission products such as radio-strontium and radio-tungsten were transported within aerosol particles, the excess carbon-14 followed as essentially a passive tracer of stratospheric air circulation.
Specifically, I refer the authors to the discussion on pages 18,488 and 18,489 of the Johnston (1989) study, and the summary sentence in the Abstract viz-a-viz the statement explaining how the initial skewed-Gaussian shape of vertical profile layers of excess carbon-14 in the stratosphere (e.g. shown in Figure 5 and Figure 11c, for 31 N balloon soundings) progressed in the 2-3 years after the partial test ban to a “step-staired” profile, with essentially constant mass mixing ratio above 20km.
It is quite striking the similarity between the Oct 1963, 1964, 1965 and 1966 excess radiocarbon profiles at 31N from Figures 5 and 11c of Johnston (1989) with the Hunga-excess water vapour mass mixing ratio profiles in Figure 3.2 from the Hunga report. Specifically, considering the vertical profiles for Oct 2022, 2023, 2024 and 2025 as equivalent timings seem to illustrate very similar progression, for the Hunga excess water vapour.
Since all previous volcanic cases so high into the stratosphere were considering aerosol transport, which (from sedimentation) can never follow the deep branch stratospheric circulation, this analogue situation from radio-carbon monitoring in the years after the partial test ban could be an illustrative and well-observed analogue situation.
Specifically, I ask whether the authors would be willing to plot-up for these months (Oct 2022, Oct 2023, Oct 2024 and Oct 2025) the 31N profiles of the water vapour anomaly from MLS and ACE-FTS, and in the model simulations, which could potentially make an interesting equivalent comparison test case for interactive stratospheric CCMs.
4) A final note I make, within this extra 4^th comment, is to ask the authors whether as a community we should expect a progression into a longer timescale residence time, with this vertical-shift of the Hunga excess water vapour into the upper-half of the stratosphere and lower mesosphere.
This now appears to be what the MLS and ACE-FTS measurements are showing, that initial expectations of the 5-10 year timescale (e.g. from Millan et al., 2022) have lengthened initially to a 2031 expected return date (from Zhou et al., 2026), and suggested here in this new study to be a further 1-2 years later, i.e. 2032 or 2033.
The timescales for removal from the stratosphere must tend to be longer for air masses at this altitude, and then should we implicitly have expected any initial decay timescale, seen through 2022 to 2024, progress later to a slower decay-rate more representative of upper stratospheric air masses, or so. Please can the authors make some comment about this point.
References
APARC (2025) : The Hunga volcanic eruption atmospheric impacts report. Yunqian Zhu, Graham Mann, Paul A. Newman, and William Randel (Eds.), APARC Report No. 11, WCRP-10/2025, https://doi.org/10.34734/FZJ-2025-05237 .

Khaykin, S., A. Bourassa et al. (2025): “Atmospheric transport and evolution of Hunga water vapour and aerosols”. In APARC, 2025: The Hunga Eruption Atmospheric Impacts Report [Yunqian Zhu, Graham Mann, Paul A. Newman, William Randel (Eds.)]. APARC Report No. 11, WCRP Report No. 10/2025, https://doi.org/10.34734/FZJ 2025-05240 .
Friend, J. P. (1966): “Properties of the stratospheric aerosol”, Tellus, vol. 18 (2-3), pp. 465-473, https://doi.org/10.1111/j.2153-3490.1966.tb00259.x
Johnston, H. et al. (1989): “Evaluation of excess carbon-14 and strontium-90 data for suitability to test two-dimensional stratospheric models”, J. Geophys. Res., vol. 94 (D15), pp. 18,485-18,493, https://doi.org/10.1029/JB074i006p01339 .
Schoeberl, M. et al. (2025): “Stratospheric injection lifetimes”, J. Geophys. Res. Atmos., 130, e2025JD043928. https://doi.org/10.1029/2025JD043928 .
Telegadas, K. and R. List (1969): “Are particulate radioactive tracers indicative of stratospheric motions”, J. Geophys. Res., vol. 74, no. 6, 1339-1350, https://doi.org/10.1029/JB074i006p01339 .
Zhou, X. et al. (2023), “The influence of internal climate variability on stratospheric water vapor increases after large-magnitude explosive tropical volcanic eruptions”, Geophys. Res. Lett., vol. 50, e2023GL103076, https://doi.org/10.1029/2023GL103076 .
Zhou, X. et al. (2024), “Antarctic vortex dehydration in 2023 as a substantial removal pathway for Hunga Tonga‐Hunga Ha'apai water vapor”, Geophys. Res. Lett., vol. 51, e2023GL107630. https://doi.org/10.1029/2023GL107630 .
Zhou, X. et al. (2026), “Residence time of Hunga stratospheric water vapour perturbation quantified at 9 years”, Comms. Earth Env., vol. 7, 198, https://doi.org/10.1038/s43247-026-03216-5 .

Citation: https://doi.org/10.5194/egusphere-2026-4022-CC1
RC2: 'Comment on egusphere-2026-4022', Anonymous Referee #2, 18 Sep 2026

This paper investigates the changes in the distribution of stratospheric water vapor, in the years following the large direct injection of steam by the Hunga volcano in Jan 2022, from contemporary satellite observations and comparisons with model simulations. The Hunga excess H2O lingers on in the stratosphere, transported around by the BDC, until various removal mechanisms take control, such as photolysis following ascent into the mesosphere, the seasonal dehydration caused by sedimentation of ice PSCs (overwhelmingly in the SH) and the stratosphere/troposphere exchange of air masses; the latter two must await for the advection of the excess H2O into their domains of influence. The 2022 Antarctic vortex was already in place before the H2O excess was transported far enough poleward so the dehydration removal process did not come into effect until the following year. Similarly, strat/trop exchange is not effective at removing the H2O excess until it has descended to low enough altitudes.
In the study, interannual changes in the water vapor are modeled using CPT temperature vs H2O lag regressions trained on the 2005-2021 tropical time series and extended to the post-Hunga time frame using the observed RO CPT temperature time series to track the vertical and horizontal propagation of the resulting correlated H2O anomalies.
The conclusion from the paper is that the model simulations agree reasonably well with the observations on the redistribution of water vapor above 20 km, but do not represent an accurate depiction of the removal of stratospheric water vapor from the lowermost regions of the stratosphere. Evidence of excess tropical H2O during 2023/2024 seen in the observations is interpreted as a recirculation of water from the southern hemisphere mid-latitudes back to the tropics rather than increased transport upwards through the tropopause.
This interpretation seems reasonable, but reporting of the results of earlier studies of the Hunga H2O excess could be more appropriately worded e.g. L224-L233 and elsewhere seems unusually harsh since the LMS region was omitted or could not even have been considered relevant in some of the works cited given their earlier publication dates.
I recommend that the paper be accepted subject to some minor corrections.
Typos, Suggestions and Other Comments:
/xxx/ means remove xxx

[xxx] means add xxx
L14: suggest /increases/ [prolongs]
L76: No MLS [H2O] observations
L78: /suggested/ [standard]
L80-81: In May 2024 and onwards the MLS H2O measurement runs for only a few days per month to preserve the longevity of the remaining measurement record. The H2O status flag is set to an odd value for profiles when its measurement system is off and it is the primary indicator of bad or unavailable data regardless of the data quality flag i.e. the status flag must be an even value otherwise the data should not be used.
L87-88: What is the procedure used for conversion from geopotential to geometric heights?
L89: The [v5 MLS H2O] artifact correction
L107: monthly CPT [RO] temperature
L113-118: define the acronyms MOZART-TS/TSMLT, MAM4, CMIP6 SSP2-4.5, MERRA2
L133: define ERA5
L137: don't start a sentence with a numerical value... altitude. 300000
L178: observed [RO] CPT temperature
L227: re-arrange the figures to be referenced in numerical order, this page jumps from Fig3 to Fig7 and skips the ones in between
L237: subsequent /F/[f]igures
L248: evolution /is/ [are]
L304: curve/d/
L309: suggest /increases/ [prolongs]
L336: Tracer particles were released in the FLEXPART simulations in December 2023. It does not appear that uncertainties in the back-trajectories were examined. What are the effects of accumulated errors in advection over many months of following these particles? This is in contrast to the treatment of the WACCM simulations, where an ensemble of free-running simulations and a specified dynamics simulation were examined (by the way, why doesn't an SD run deserve the same skepticism and get sampled with probable dynamics errors?).

Citation: https://doi.org/10.5194/egusphere-2026-4022-RC2

Kimberlee Dubé, William Randel, Adam Bourassa, Susann Tegtmeier, Xinyue Wang, Jun Zhang, Eilidh Hlady, Meghan Brehon, Sergey Khaykin, and Douglas Degenstein

Viewed

Total article views: 411 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
246	137	28	411	27	25

HTML: 246
PDF: 137
XML: 28
Total: 411
BibTeX: 27
EndNote: 25

Views and downloads (calculated since 06 Aug 2026)

Month	HTML	PDF	XML	Total
Aug 2026	65	30	5	100
Sep 2026	138	87	19	244
Oct 2026	43	20	4	67

Cumulative views and downloads (calculated since 06 Aug 2026)

Month	HTML	PDF	XML	Total
Aug 2026	65	30	5	100
Sep 2026	138	87	19	244
Oct 2026	43	20	4	67

Viewed (geographical distribution)

Total article views: 387 (including HTML, PDF, and XML) Thereof 387 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 07 Oct 2026

Short summary

The Hunga volcanic eruption in January 2022 increased the total stratospheric water vapor (H₂O) mass by 10 %. The study uses satellite observations and models to track the transport of the Hunga H₂O. We find that there is substantially more H₂O in the observations than the model during 2024 and 2025, suggesting that the modelled transport across the lower stratosphere is too fast. We also find evidence that some of the H₂O that reaches the lower stratosphere is mixed back into the tropics.


Total:	0
HTML:	0
PDF:	0
XML:	0