the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 31 Jul 2026
Understanding drivers of inter-model uncertainty in the dynamical response to stratospheric heating
Abstract. Stratospheric aerosol injection has emerged as a candidate climate intervention strategy to partially offset global surface warming. Previous work has demonstrated that lower stratospheric heating drives modifications to the stratospheric thermal profile, circulation, and water vapor, with downstream consequences for surface climate; however, the impact of stratospheric heating has not previously been characterized across a multi-model ensemble. This work presents first results from the Stratospheric Heating Model Intercomparison Project (SHeatMIP), in which an idealized 0.3 K/day tropical lower-stratospheric heating tendency is imposed across five climate models (CESM, GFDL, GISS, MIROC, UKESM). All models show robust increases in lower-stratospheric temperature, water vapor, and polar night jet strength in both hemispheres, with corresponding surface shifts in the subtropical and eddy-driven jets and a polar cap pressure response resembling a positive North Atlantic Oscillation phase. Despite qualitative agreement, inter-model spread is substantial, with differences of 1 K in the cold point temperature adjustment and 0.38 K in the global mean surface temperature response. The surface temperature spread strongly co-varies with the stratospheric water vapor response (R2 = 0.82), implicating water vapor as a key source of surface warming uncertainty. The forced polar vortex response shows strong co-variability with the climatological polar night jet strength (R2=0.98), and projects onto the surface as a polar cap pressure anomaly. The results show that inter-model differences in the response to stratospheric heating may be traceable to the climatological mean state, offering a pathway toward observationally-constrained evaluation of model suitability for SAI research.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Atmospheric Chemistry and Physics.
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.- Preprint
(53174 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 17 Sep 2026)
- RC1: 'Comment on egusphere-2026-4200', Anonymous Referee #1, 01 Sep 2026 reply
-
RC2: 'Comment on egusphere-2026-4200', Anonymous Referee #2, 08 Sep 2026
reply
Review of 'Understanding drivers of inter-model uncertainty in the dynamical response to stratospheric heating' by Golja et al.
I enjoyed reading this very well-written paper on climate models’ response to stratospheric heating. Stratospheric heating is one of the key side effects of stratospheric sulfate injection, and isolating the consequences of stratospheric heating in an idealised setup is extremely useful. More such experiments should be done to disentangle climate model uncertainty as it relates to SAI.
I only have a few minor comments for the authors’ consideration.
L.42 : update to published version https://acp.copernicus.org/articles/26/7463/2026/
L.115 : Are we to interpret the heating as what you would get if SAI was used to produce a -4 W/m2 total (SW+LW) forcing? I am a little confused here.
Section 3.1 : I would find it useful to situate these stratospheric temperature changes within what one would expect from typical SAI scenarios e.g. G6-1.5K-SAI (which injects aerosols at 30N/S and leads to less stratospheric warming than equatorial injection (G6sulfur)).
Fig 7 : The near-total absence of stippling indicates that nowhere has a value significantly different to zero, correct? If so, that would be worth emphasising.
L.315 : How should I interpret the t values in fig 8? Does the statement on GFDL in L. 315 follow from that?
I am quite surprised by the results on precipitation. I always assumed the larger reduction in precipitation under equatorial injection was at least partly due to stratospheric heating.
L.354 : typo on number
L.395-6 : Is the ‘robust’ statement here consistent with the discussion on fig 8 (L.315-7)?
L.449 : typo parenthesis
Citation: https://doi.org/10.5194/egusphere-2026-4200-RC2
Data sets
Replication Data for : Understanding drivers of inter-model uncertainty in the dynamical response to stratospheric heating (Version 1) Colleen Golja https://doi.org/10.5281/zenodo.21345483
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 226 | 89 | 21 | 336 | 28 | 26 |
- HTML: 226
- PDF: 89
- XML: 21
- Total: 336
- BibTeX: 28
- EndNote: 26
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
As part of ongoing efforts to understand the climate impacts of stratospheric aerosol injection, this study investigates the responses to prescribed heating in the tropical stratosphere across five climate models. The authors examine both the mean response and the associated inter-model uncertainty. Two notable relationships are identified: the global-mean surface temperature response is related to the stratospheric water vapor response, and the forced Northern Hemisphere stratospheric polar vortex response is related to the climatological strength of the polar vortex.
Overall, I find the study interesting and potentially important. In particular, the manuscript highlights how forced climate responses may depend on model climatology, with potentially useful implications for understanding and constraining inter-model uncertainty. Nevertheless, I have several major concerns that I hope the authors will consider as they revise the manuscript.
General comments
My primary concern is the lack of statistical significance for a substantial portion of the results, particularly those shown in Figures 7–10. Likewise, the responses shown in Figures A5, A6, A8, and A9 also likely have quite limited statistical significance. These figures constitute a substantial part of the discussion of the tropospheric circulation response to changes in the polar vortex and the winter precipitation response.
I think this issue needs to be addressed explicitly. If the responses are not statistically distinguishable from internal variability, detailed physical interpretation of their spatial structures becomes difficult to justify. At a minimum, the manuscript should clearly distinguish robust responses from suggestive but statistically uncertain features. The discussion of the latter should be substantially qualified and, where appropriate, shortened. One possibility is to examine the zonally averaged SLP and precipitation, as zonal means may be statistically more robust than analyses at individual grid points.
Relatedly, I find that the lengthy discussion of statistically uncertain responses tends to obscure what I see as the more distinctive points. Both the title and the overall framing suggest that inter-model uncertainty, and particularly its relationship to model climatology, is a central focus of the study. In the current manuscript, however, this key message emerges relatively late and is preceded by extensive discussion of responses that are not always statistically robust. As a result, the main narrative becomes difficult to follow.
I think additional diagnostics and discussion are needed to establish physical interpretations for the inter-model relationships highlighted by the manuscript. For the stratospheric water vapor response versus the surface temperature response, could the authors compare the simulated relationship with theoretical expectations or with previous estimates from the literature? Second, the UKESM result requires further explanation: UKESM shows an increase in stratospheric water vapor but a cooling response in global-mean surface temperature. Given that increased stratospheric water vapor would generally be expected to contribute to surface warming through its radiative effect, what processes account for the opposite net GMST response in this model? As noted in the manuscript, UKESM also experiences a drying in the UTLS below 200 hPa in the tropics, which differs from other models. Any thoughts?
For the stratospheric zonal wind relationship between climatology and response, additional dynamical discussion could be beneficial, including the wind in the neck region, as suggested by earlier studies. Some earlier studies also underscore the importance of the tropospheric jet latitude for the surface response to stratospheric perturbations. It might be worthwhile to examine the possible role of the tropospheric mean state in the results presented.
I also think that the QBO deserves examination. The imposed temperature tendency is located in the tropical stratosphere and substantially overlaps the QBO region. Any resulting QBO response could, in turn, have implications for both the Northern Hemisphere polar vortex and the surface climate responses. This issue may be particularly relevant to the manuscript's emphasis on inter-model uncertainty, as the representation of the QBO varies widely among models. I understand that only three of the models exhibit QBO-like variability. Nevertheless, I think it would still be valuable to examine how the QBO responds in those models.
Individual comments
L54: volcanic literature?
L70: The cited literature doesn’t seem to directly support a polar vortex modulation of tropical cyclone activity.
L101: Some additional information about the nudging is needed, as a water vapor change is still observed in MIROC.
L117: winter?
L121: Please provide a description of the control simulations. Now it can only be inferred that they seem to refer to a preindustrial run of unknown length.
L136: A brief description of the methodology would be appreciated.
L150, 157: 10 hPa or 70 hPa? What region do you use in this analysis?
L174: Please elaborate on the argument. How should we see the consistency between temperature response and polar vortex changes?
L212: How to understand the drying?
Figure 3: The color scheme in this plot can be quite misleading. Please use a scheme that is symmetric around 0. Why do the contour intervals appear to be inconsistent across panels?
L240: I thought the consistency between wind and temperature was expected.
Figure 5: To help understand the robustness of the BDC change, I think it could add estimates of internal variability in the control run.
L303: I hardly see the statistical significance in MIROC.
L306: Why not discuss the NAO first? Inferring the NAO response from temperature anomalies seems less reliable.
L324: The statistical significance markers seem limited and sparsely distributed in the plots.
L411: Isn’t UKESM warming smaller than GFDL?
L423: Figure 12b -> Figure 12c
L449: 2.3).
Figure 12a: What is the asterisk in the plot? I assume the pink dot represents MIROC?
L470: Delete the duplicate “shifts in the”.