the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Sep 2026
Temperature sensitivity of glacier ablation models in high-elevation regions
Abstract. Glacier mass balances in high mountain regions across the world are projected to decline as a result of climate warming. However, the modelled response of glaciers to increasing temperatures depends strongly on the temperature sensitivity of the mass-balance model used. Most modelling studies employ empirical modelling approaches for simulating glacier mass balances, but the effectiveness of empirical approaches has been questioned for high-elevation regions due to a partial decoupling between temperature variations and ablation. This study compares the temperature sensitivities of two empirical and two energy-balance models when applied at high elevations (~6500 m) on the Khumbu Glacier, Mount Everest. The empirical models have a temperature sensitivity more than double that of the energy-balance models when the same warming experiments were applied. This bias indicates that future projections of glacier mass balances in high-elevation regions produced using empirical mass-balance modelling will likely be more negative than if energy-balance approaches were applied, and consequently that existing global mass-balance projections may overestimate future glacier recession in similar high-elevation regions.
- Preprint
(4621 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 30 Nov 2026)
- RC1: 'Comment on egusphere-2026-5187', Anonymous Referee #1, 30 Sep 2026 reply
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 199 | 100 | 31 | 330 | 29 | 24 |
- HTML: 199
- PDF: 100
- XML: 31
- Total: 330
- BibTeX: 29
- EndNote: 24
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This study investigates the sensitivity of four different glacier ablation models to rising air temperatures in the Central Himalaya. The authors apply four models of different complexity, ranging from an empirical temperature-index model to the fully-fledged energy balance model COSIPY. The temperature sensitivity of the models is tested by exercising three synthetic warming scenarios, increasing the air temperature by +1°C, +2°C and +3°C, respectively. The authors find clearly increased temperature-sensitivity of empirical glacier ablation models compared to energy balance models at their study site.
While the study is highly synthetic (synthetic warming scenarios, no model calibration and validation with in-situ glacier observations, point-based simulations in the accumulation area), I still think the results hold relevant conclusions. The discrepancy in temperature-sensitivity between empirical glacier ablation models and energy balance models is of high relevance for modelling studies and future projections of glaciers in high mountain regions. The motivation and relevance of this study are well outlined in the introduction. The manuscript is overall well structured, and the figures are well presented. Overall, I see the manuscript in between major and minor revisions, depending on how the authors handle my two major comments.
Major comments
The lack of in-situ observations, especially of ablation, is a big limitation to the results of this study. In my opinion, the SMB models should be correctly calibrated with observations, and validated afterwards. Therefore, if there is any other suitable study site with station data and glaciological measurements available in this region, I would highly recommend to include it into this study.
In general, I think your study would benefit a lot from including more modeling sites. I see in your 2026 paper in Journal of Glaciology, Table 2, that there are more stations with all required meteorological input data available (e.g., Pyramid-G, South Col-N) and located at a range of elevations. Why not include these sites as well? I think this would make your results way more robust, and you could also address the importance of sublimation vs. melt across elevations.
The entire model calibration and determination of parameter ranges needs revision. First, the entire process of how the other models were calibrated to mirror the COSIPY range is not explained. Did you have predefined ranges for the model parameters that you tried out, or did you start at a “best guess” and then just randomly extended the parameter values in both directions until you had a similar range of ablation? If so, how did you define the step size? Please explain this in more detail in the methods section.
Secondly, when it comes to parameter sensitivity and ranges, you have to always consider that many parameters are dependent on each other, i.e. parameter equifinality. For example, for the ETI increasing the solar radiation factor could be compensated by increasing the albedo at the same time. Or a low ψmin could most likely be compensated by a higher temperature factor c or a lower albedo for GO. Therefore, you could probably vary the parameters in a range way beyond what’s physically reasonable and still somehow hit the COSIPY range. This might also be an alternative explanation of your parameter values that are differing from literature values (like the strongly negative ψmin ).
Regarding parameter sensitivity, I have experienced that the sensitivity of a model parameter is not independent of the applied range of values. A parameter that is varied in a larger range is often noted as more sensitive compared to a parameter that is varied in a smaller range, just because the larger range causes results to differ more strongly.
For all these reasons, you should explain in detail how you defined the parameter ranges for the models, and how you calibrated them to fit to the COSIPY range.
Specific comments
Introduction general points:
I miss more details on the differences between empirical and SEB models. I suggest describing in more detail that SEB models are physically based and consider more processes, but require high quality climate data input, which is why they are usually not used for global simulations and future projections. (paragraph 3)
Then make clear, that empirical approaches can only calculate melt, and no sublimation at all. They were not developed for regions where sublimation is important.
Paragraph 4-5 outline the motivation very clearly.
L19f: “The latter can be summarised by the temperature sensitivity – the change in glacier state (e.g., mass) normalised by change in air temperature.”
This statement is highly simplified, model uncertainty includes many more aspects. First of all, different models consider different processes, leading to different results. Then, depending on the model, other climate variables can have a similarly high impact as changes in temperature (e.g. changes in wind velocity impacting the turbulent heat fluxes, or changes in precipitation).
L22: Are you referring to surface/climatic mass balance and surface/climatic ablation here?
L24f: Changes in surface elevation have an indirect impact on surface mass balance as well.
L45ff: Maybe you could already add a few more details in this paragraph, like that you used ablation models of different complexity, ranging from empirical temperature-index model to physically based energy balance models. Maybe mention the study site here as well.
Fig. 1: This figure is not really necessary in my opinion. You mention in the text that more than half of the glacier surface area lies above 5400m asl, which is enough. Could be moved to the supplement if you add one anyway.
L65ff: Here you say that you use incoming longwave radiation measurements for model calibration. Is this approach described anywhere in more detail?
L75ff: Can you give more details on the precipitation adjustments. It is also not clear if there was any elevation correction applied. Given that the Base Camp Station is located some 1000m lower than Camp II station / Modelling site, it is required to account for this elevation difference.
L125ff: The COSIPY model lacks explanation. It is clear that you cannot explain everything (and it’s also not necessary), but at least follow your strategy from the other models, where you described the most important aspects including the albedo scheme. The parameterization of the turbulent fluxes might be interesting as well, e.g. which stability correction did you use? I would also mention the parameters you are planning to adjust, and why you chose those out of the 30 total parameters you mentioned. Explain what they are controlling in the model (e.g. the impact of snow/firn roughness length of albedo ageing timestep might not be clear to everyone).
For GO you specifically mention that you exclude the refreezing component. Did you do the same for COSIPY?
L136: It would be really good to have in-situ observations to correctly calibrate and validate your models. See major comment 1.
L141: Remove “as we focused on modelling surface mass balance only"
Ice albedo and ice roughness length were not perturbed, as snow and firn layers at the study site are thick enough that underlying ice properties will not affect surface ablation.
L149: So I assume refreezing is excluded for COSIPY as well? What about internal melt within the snowpack?
L157ff: I don’t really understand how you calibrated the other models to mirror the COSIPY range. Please explain this in more detail. See major comment 2.
L193ff: I would like to get a bit more details here on the energy and mass balance results. What is the annual SMB? What is the annual ablation? How big are the contribution of melt and sublimation? How does this partition change between monsoon season and the rest of the year? You answer some of these questions in the discussion, but why not give the most important results in the results section?
Fig. 4: Change m to m w.e. in the figure caption and the y-axis label.
Fig. 5: I would suggest adding the total ablation as well. This would also support your statement from L194 that ablation takes place year-round.
L199f: Could you explain in one or two more sentences how it usually comes to melting with air temperatures below 0°C? I assume this happens on sunny days, where SWin brings enough energy to warm the snowpack to melting point? How high is the surface temperature throughout the year?
L216f.: I have doubts about your parameter calibration. See major comment 2 for details.
L227f: Technically, the increase in the ensemble range of ablation from the empirical models is less than three times the corresponding increase from the energy-balance models (TI 5.45 vs. COSIPY 2.01).
Fig. 7: Replace m with m w.e. in the axis label.
Fig. 8: Replace m with m w.e. in the axis label.
L235ff. / Fig. 9: Why does p2 become more important with warmer temperatures for the ETI?
L240f: Why does the fresh snow albedo become more important with warmer temperatures for the GO?
L264f: Suggest changing to “… indicating that variation in solar radiation will continue to be the dominant control on ablation rates in high altitude regions, …”.
Fig. 10: Replace m with m w.e. in the axis label.
Fig. 11: Replace m with m w.e. in the axis label.
L369: I think including more modelling sites would make your results way more robust. See major comment 1.
L395ff: While I personally agree with your statement, I would be carefull saying “empirical approaches are likely to overestimate future rates of ablation, energy-balance approaches are more appropriate for modelling the future of glaciers in this environment”. Because your results do not proof that empirical models are overestimating future ablation. They only show that empirical models produce a lot more ablation under warmer conditions compared to energy-balance models. Theoretically, the opposite could also be true, i.e. energy-balance models might underestimate future ablation. I think you should phrase these last two sentences much more moderate.