Recent history and future demise of Jostedalsbreen, the largest ice cap in mainland Europe

Author(s): Henning Åkesson, Kamilla Hauknes Sjursen, Thomas Vikhamar Schuler, Thorben Dunse, Liss Marie Andreassen, Mette Kusk Gillespie, Benjamin Aubrey Robson, Thomas Schellenberger, and Jacob Clement Yde

MS No.: egusphere-2025-467

General comments

This manuscript presents a model study that uses the Ice-sheet and Sea-level System Model (ISSM) and temperature index model forced with seNorge_2018 gridded daily mean temperature and daily total precipitation and 4 Climate models with two scenarios to simulate recent history (from 1960) and future demise of Jostedalsbreen, until 2100 and then onwards with constant climate until 2300 to assess the committed mass loss for several scenarios. Clear description of model components and observations used for the model initialisation and a thorough analysis of the results is made. This paper presents first simulations of Jostedalsbreen as a whole and has important message about its projected evolution in the future and about the difference in mass loss of different model approaches, i.e. the difference in mass loss between global/regional assessments that use flow line models to model individual outlets of complex ice caps separately, and models of whole ice masses as presented here (lines 578-589). The paper is extensive with much information, 12 Figures and 2 tables and additional 5 figures and a table in appendix.  There are a few issues regarding consistent terminology and sequence of figures listed in the specific comments that if addressed would increase the clarity of the paper.

Specific comments:

The use of the terms committed mass loss, ‘in the pipeline’ (line 103) and response time of glaciers is confusing and could be clarified by using one consistent way of describing committed mass loss due to long response time of glaciers.  This would be (at least, perhaps other places as well) on page 3, lines 88-90, 101-104, page 12  line 281, page 14 line 356 and 362, line 491, 495, page 25 line 507, line 598, page 35 line 740.  The explanation on lines 100-107 is a good place to define the terminology and use it consistently throughout the paper.

The figures are not discussed in the same order as appearing, causing readers to have to jump back and forth and reduces the readability.  this could be improved by adding results of Future8.5-ECC on Figure 2 and rearranging the sequence of figures to fit better the flow of the text.

The uses of the terminology  “X and Y, due to A and B, respectively”  is not always correct, for example lines 480-481, line 601, line 722, 710-711, 722, and perhaps other places, the correct way is in line 609

The colloquial term “from scratch” sounds strange in this context, it is in several places, including line 573, the figure caption of figure A3 and in text, suggest to use “from no ice conditions” instead.

The use of ice caps – glacier unit is not consistent, in lines 13, 17 on page 1, page 3 lines 57-60, page 34 lines 688-690

Technical comments:

Line 131 – suggest to replace “surface” catchments with “ice” – or “water” catchments

Line 190 here can be clarified, is the velocity along surface slope that is the 2D velocity produced?

Line 215  Oldedalen is not shown on Figure 1, could be added?

Line 217-218  this assumption is not discussed again, could add some discussion of the consequences of this assumption on page 33, line 688?

Line 229, here could the term cost function be added to clarify:  The misfit between the modelled u and observed ice velocities, the cost function J(u, alpha)

Line 258  is the RMSE difference between the modelled and observed velocities?

Line 261  what does “more or less accurate” mean? Can it be quantified?

Line 264 is the RMSE difference between the observed and modelled thickness? (rewrite for clarity)

Line 269  suggest to replace “dynamics” with “parameters”

Line 270  how many?  Suggest to replace “positions” with “length” or add terminus positions

Table 2 could references for the climate models be added to table?

Line287  How is this found?  Can be rewritten for clarification

Line 301-302  here text can be clarified, how low RMSE?

Line 333  what are the monthly linear trend models based on, text can be edited for clarification

Line 381 – what does “well within uncertainty” mean?

Line 385 suggest to replace “variations” with “periods” (and rewrite as in figure caption)

Line 387  replace Fig. 2c) with 2 a)

Line 391 repalce fige 9b with 2 b)

Line 392  suggest to replace “apparent” with “modelled”

Line 395 should the be “not” in front of glacierised?

Line 398  strange wording of “should be considered satisfactory” suggest to rewrite for clarification

Line 401 and 406 inconsistency of 30-80 m and 20-50 m

Line 409 suggest to add “modelled” in front of frontal position and “longer” instead of down valley

Line 465  suggest to add “modelled” for clarification

Line 507  “on their way to steady state” is not clear, see comment above on response time, how long time does it take to reach steady state?

Line 515-516 , suggest to swap order, first state reality and then response?

line 530 suggest to replace “pathway” with “scenario” for consistency

Line 624, the sentence is strange, how is model performance dependent of observations? Is it dependent on the parameter selection?

Line 630 what is meant by “in theory” here? Are the model output with this high uncertainty?

Line 637 more concise text would be useful here, are they located, or not? Are some? By how much?  Or rewrite for clarity

Line 649, suggest to replace “heavily influenced” by “controlled”

Line 651  suggest to replace “small scale” by “spatially variable” I don’t think wind redistribution or avalances are small scale, they may have small impact on the total, suggest rewriting for clarity

Line 653 suggest to replace “corrected” with “variable”

Line 663 suggest to rewrite, replae “reality” with “observations”  and the next sentence implies that velocity data is not “correct” does it have high uncertainty?

Line 671  what does “representative” mean here?  - low uncertainty?

Line 679-680 suggest to edit, strange wording  “thicknesses diverge”?  simulate observations?

Line 696 and 697 edit (Fig A4ef to Fig A4 e) and f)

Line 713 add “high” after unrealistic?

Line 719 suggest to add “current climate” to sentence and rewrite for clarity

Line 725 suggest to replace “reversed” with “replaced by”

728-730 suggest to edit, this is not clear  “difficult to regrow” what does that mean? “not enough” is not clear

Line  732  take s off appear for plural

Missing references in

Line 31

Line 117 both for temperature-index model and Bayesian approach

Line 164 (QGIS)

Line 168 for1966  DTM for entire ice cap

Line 174  ArcGIS Pro and ANUDEM should be referenced

Line 261 are there references to show the validity of the assumption?

Line 278 what are the references for the reanalysis?

Line 543 reference for uplift?

Line 545, regardless of what? Suggest to rewrite for clarity

Line 560 add transGeo or Area on the SMB

Line 630 add reference to the ice thickness model here?

Line 739 add reference for machine learning

Figure 3  the color scheme for the velocity figure is not helpful to show the variability in magnitude of velocity, could a log scale be used, or longer scale to show the lower velocities better, the map is almost completely green and does not show much.

Figure 7, suggest to add Bedrock to beginning of figure caption – and explain shy the elevation is fluctuating, there appears to be some up and dow in figures c) and d) blue colors between the yellow.

Figure 8  would it be possible to show the sizes of each region by sizes of the circles? So that at glance of the figure readers would have impression of how different sizes region the different circles represent?

Fiugre A2 is not referred to in text and no discussion of 10 or 20 m threshold is in the text, either add that, or delete figure

The manuscript by Åkesson et al. presents model simulations of Jostedalsbreen ice cap in a past, present and future climate. The manuscript is very detailed and well-written, the methods are sound, the structure is clear, and the results are novel, relevant and significant. I only have minor suggestions for improvements and clarification, which are detailed below.

Specific comments:

L7: ”12-74%”. It could be useful to include a median / mid-range estimate here.

L12: “are irreversible”. Please define what is meant with irreversible here, e.g. by referring to hysteresis theory or committed mass loss.

L31-32: “The wide range of geometries … of the world.”. This is rather what you find in many ice caps (rather than glaciers) globally, with slow-flowing interiors and fast-flowing outlet glaciers.  

L45-49: In addition to highlighting the need for 3-D higher-order modelling, it could be emphasized that shallowness assumptions in commonly used ice flow models are not justified for complex ice cap systems.

L50-67: To my taste, the discussion of global glacier modelling is a bit too lengthy and could be condensed to a few sentences.

L84-85: “Some of these models struggle … because each glacier is treated individually”. It could be good to include a reference to which model(s) that applies (e.g. Millan et al. 2022).

L218: “the lake surface is considered the bedrock topography”. Does this create a wall that slows down motion (and induces thickening) near the fronts?

L228: “We constrain the local values of the basal friction parameter”. I suppose the bed topography is kept fixed following Gillespie et al. (2024). Does this imply that all errors in both ice flow model physics, mass balance forcing and bed topography are compensated for by basal friction adjustments? It could be worth adding some discussion on this in the Discussion section.

L237: “For the rest of the ice cap”. What would happen if you apply the above approach (the adjoint-based minimization) also to "the rest of the ice cap"? Would the results be clearly worse than with the elevation-dependent function?

Section 4.3: It could be useful to briefly mention how the SMB model distinguishes precipitation falling as rain or snow. This is relevant for understanding the future SMB description.

L256: “obtain the best possible representation of ice velocities”.  Aren’t velocities for Tunsbergsdalsbreen and Nigardsbreen forced to match observations by minimizing the cost function (eq. 2)?

L260: “we assume that the ice cap was in steady-state in 1960s.”. Maybe this sentence could be moved to after L270 when the spin-up process is described. Then it is directly clear how the steady-state is achieved.

L264-268: It could be good to merge/move the part about friction inversion with/to section 4.2 for clarity.    

L269-270: “Once the optimal model dynamics is obtained, we perform a 250-year spinup...”. Have you considered/tested alternative methods to spin up the model in 1960? The current method advantageously can use the detailed known bed as a boundary condition, but drawbacks are the steady state assumption in 1960 and lack of agreement between modelled and observed surface height. I wonder whether the authors considered using an inverse approach as in Frank and Van Pelt (2024), which advantageously would have been faster and would have removed the steady-state assumption. On the downside, it would have implied the need not to fix the bed. Please consider adding some discussion on this.

L357-358: “using the mean annual SMB of 2001-2020”. Since the present-day SMB is likely more negative than the 2001-2020 average, it is more a "Commit-2010" scenario that is tested here. I.e. we are already 15 years too late for this scenario ;)

L361: “The Commit experiments”. I am not fully convinced whether the Commit4.5 and 8.5 results are particularly useful. It is rather arbitrary to use the end of the 21st century as the start of a stable future climate. It also adds two more runs to an already large suite of experiments.

L373-377: “The historical simulations ... future projections.”. This is more discussion or introduction than results.

L379: “changed very little since the 1960s”. How much of this may be due to the steady state assumption?

L379-387: A general comment, there are quite many short paragraphs in the manuscript. Please consider merging them with neighbouring paragraphs where possible.

L397-398: “This sensitivity analysis ... satisfactory.”. I wonder why the model forms these large thin ice areas. Is it an artefact of the description of the frontal and lateral boundary conditions and/or flow speeds near the glacier boundaries?

L399: “Overall, the modelled thickness agrees well ...”. This is a good result! Since both velocity (to some extent) and bed topography are constrained with observations, the main errors in ice thickness are errors in the ice flux being too small or large which in turn results from errors in the surface mass balance. Would you agree that most of these thickness discrepancies are due to mass balance uncertainty?

L428: “gains in future precipitation”. Would not much of that fall as rain instead of snow?

L439: “Considering the mid-range climate forcing.” It could have been considered to focus the future analysis only on the mid-range scenarios. The current spread in the climate projections seems unnecessarily large, especially for RCP 8.5. It is also odd that the average of RCP8.5 mass loss is lower than for RCP4.5.

Figure 8: It is hard to see which circle diagram belongs to which glacier on the map. Furthermore, I would prefer discs indicating how much mass is left rather than how much is lost. That would be more intuitive. I.e. a full circle means all ice remains.

Section 6.2.4 (L500-510): The hysteresis effect could be discussed here. Thick (flat) ice caps can maintain a positive or zero SMB because of the high surface elevation. When the average elevations drop too much, the mass loss becomes 'irreversible' unless a very cold period happens over an extended period of time (i.e. even much colder than the 20th century climate). This hysteris effect applies mostly to low-sloping glaciers, ice caps and e.g. the Greenland Ice Sheet.

Section 7.2: Just an idea, there is currently a lot of information and relative change numbers for different glaciers in this section. It could be helpful for the reader to summarize the results in a table showing relative changes for Jostedalsbreen (this study) and other glaciers at different points in the future.

L678-681: “Summary. ...”. This could be removed or moved to the conclusions.

Section 7.4: There is quite a bit of repetition here of the introduction, e.g. there is no need to again refer to global glacier retreat in a warming climate or impacts on society. Please consider shortening the text.  

Technical corrections:

L6: "3-d" --> "three-dimensional (3-D)".

L42: “Meur” --> “Le Meur”.

L84: “Pelt” --> “Van Pelt”.

L174: “ANUDEM”. Please define the acronym.

L191: “GAMMA Remote Sensing software”. Please include a reference.

L206: ”is” --> ”are.

L364-365: “To this end, ... is used as a forcing”. It could be added that this applies to after 2100.

Figure 3 caption: Please remove brackets around 2022.