| 27 Jul 2026
Towards more realistic calving simulations: the effects of subglacial hydrology on stochastic calving at Sermeq Kujalleq (Store Glacier), west Greenland
Abstract. Accurate modelling of long-term calving behaviour at tidewater glaciers remains a major challenge in glaciology, with existing methods being either overly computationally intensive, or too simplistic to capture observed calving behaviour. Building on previous work that showed the potential of a stochastic positional approach to modelling calving, we investigate how far a stochastic calving model can reproduce observed terminus dynamics and calving behaviour at Store Glacier (Sermeq Kujalleq), West Greenland, with and without coupling to a subglacial hydrology model. We show that the stochastic model is capable of accurately reproducing both qualitative and quantitative dimensions of calving at Store Glacier, while the inclusion of the subglacial hydrology increases calving volumes and produces the observed seasonal calving cycle. We conclude that simple hydrological parameterisations that leave the basal water pressure fixed in time cannot produce the full spectrum of observed calving behaviours of tidewater outlet glaciers.
Competing interests: At least one of the (co-)authors is a member of the editorial board of The Cryosphere.
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.
Cooke, Wheel, and others demonstrate new results obtained with recent developments in the Elmer/Ice model applied to Store Glacier, Greenland. The key components are the hydrology model, and the stochastic calving model. These are linked indirectly through the ice flow model, and directly through the appearance of water pressure Pw in the tensile stress indictor* at the base of the ice. The paper contrasts results that do and do not implement this link and finds that, for example, evolution of the hydrology system can drive larger calving events via this link, and that it introduces seasonal features e.g a link between summer runoff and greater calving.
*The tensile stress indicator is essentially the crevasse depth formula from earlier work e.g Benn et al 2007, Nick et al 2010) , but the authors are keen avoid that interpretation.
General Comments
I found the model experiments and results to be credible and interesting, and overall agree with the authors view that (a) the stochastic calving model is a fruitful avenue for large scale models and (b) the direct coupling between hydrology and this class of calving model should not be neglected, especially because it is a modest development for any model that has the ’hard’ parts (ice stress balance, hydrology evolution, a practical calving model) working.
I do think the conclusions drawn are too strong given the results that were obtained (although I think they would likely be correct in the light of more results). First, it seems incorrect to say the coupled cases reproduce a seasonal cycle, because the simulations cover only three months. Rather, they show some aspects of the seasonal cycle. Second – and this is harder to address – the small number of simulations and absence of statistical analysis leave open the possibility that the (IMHO modest, with interesting features) differences between the simulations are due to pseudo-random number generator sequence differences (unless the same PRNG sequences were used in pairs?). More in specific comments
I also think that several of the figures could be improved.
The typesetting is odd – there are no blank lines between paragraphs, together with ragged justification this makes the start and end of arguments hard to spot.
At the same time some paragraphs are in fact long (e.g in 4.3, discussion in general). e.g 5.2 appears to be a single paragraph more than 40 lines long
Specific Comments
Abstract:
“… with and without coupling to a subglacial hydrology model.”. I was initially misled by this clause, because the various components are fact coupled in all cases – e.g hydrology->sliding->stress->calving. The distinction is between simulations where the hydrology model does and does not inform Pw in the calving component.
"and produces the observed seasonal calving cycle". Perhaps part of the cycle.
Introduction
41: “Fundamentally, the fracture mechanics that control calving are too computationally expensive...” – no – they are in practice expensive now, but that is not a fundamental issue. At the same time, it is clear that the model of this paper is also computationally expensive (if less so) – which is why the authors run only few short simulations.
Methods
100 – a brief summary of GlaDs physics and behaviour would be helpful here
105 – ‘ feeding back to the ice velocity via a Coulomb sliding law’ – include the formula, which IIRC is in fact not a plain Coulomb rule (Tb = fN), but a law that tends to Coulomb (Tb = fN) for N -> 0 and Weertman (|u| = C Tb^3) for 1/N -> 0. I know it is written elsewhere but is a key component of the hydrology/ice flow coupling.
130 – OK, but please briefly summarize the reasoning for shorter relaxation times from Benn 2026 and Wheel 2024b.
197 : “Finally, values of the parameter k were chosen for comparison with uncoupled model studies of Jakobshavn Isbrae (k = 24; Wheel et al., in prep.)”. What happens if Wheel at al is delayed/rejected? Need a short explanation here – what is the objective function?
Results
Fig 2 (b) – How about making *one* panel, with month on the x-axis. The observations in a light pen overlying one another, to make clear the amplitude and phase of cycles. Then the model runs in bold pen, with their t= 0 corresponding to their notional start month. As it is *all* of the models are advancing, so it would be good to see if their advance lines up with observations.
Fig 3 – the first of several figures with N at y = 0 (bottom), opposite to say fig 2a (N at top). Why not have all figures in the same (approximate) orientation. (so N at top of fig 3 , fig 5 panels, etc).
Around line 220: SK24U vs SK20C – the discussion here is about rather minor differences between these simulations. This somewhere that the modification I suggest to fig 2 might help – it is not at all clear to me that SK24C is more in line with observations. The same is true of the Winter simulations. – the WK2OU advance amplitude is in line with some years.
236 – comparison of fig 3 and 5. I think the U panels resemble the observations more than the C panels, e.g both C simulations produce larger bergs at the N margin. The bubble plots are not easy to compare – it might be easier compare box plots for (say) 1.5 km wide band.
267, Repeat of SK20C description – this might be avoided if you swapped fig 4-> (raw results) and 2 -3(summary of terminus position) and 3 (observed bergs) -> 4
275: if you know SK20U is wrong a-priori, why include it? Seems to be a point about the stochastic calving law rather then coupling with the hydrology.,
290- Descriptive statistics only – why not a test (e.g a signed rank test of SK24U vs SK20C and same for winter). It is possible that the difference is just a PRNG effect (I don’t believe that, but it would be good to see formal analysis)
Fig 7 – these are good for comparing the models but panel axes are small. I think the columns could be safely merged – i.,e 3 panels in total the main points will be just as clear.
Discussion
350: ” The main outcome of this paper is that including an explicit representation of subglacial hydrology in our model is crucial in improving the realism of our simulations of Store”. I would say it is a modest improvement.
Section 5.2 – repeats many of the point made in the result description, and some speculation (if we had done x, then y would be the outcome)
Discussion section in general – is long and reiterates points from the results. This obscures the synthesis points that it does make (e.g 450-470, 475-485). I suggest rewriting with a target length (say, half the current length). It is for this reason I chose 'major revisions'.