|This paper forces a 1d coupled glacier-plume model with future climate from the CMIP5 climate models to project the future behavior of 12 of Greenland’s tidewater glaciers. The topic of sea level rise and ice-ocean interactions in Greenland is of great contemporary interest, and the use of a plume model is certainly an improvement (subject to concerns about tuning and true representation of coupling between melting and calving) on previous similar studies such as Nick et al. 2013.|
Overall I think it is easy to criticize papers doing future projections because there are a great many factors to consider and decisions to be made, and many compromises are necessary. I have tried to write my review with this in mind, and I commend the authors for bringing many datasets and models together and for addressing the many challenges inherent in projecting future sea level from Greenland. Contrary to the previous reviewers I did not find the paper to be clearly written; perhaps it is simply the nature of describing a complex model with many inputs, but I found several passages to be rather hard to follow (notably the ocean forcing description), which will also come across in my review.
I do have some major concerns, some of which might be misunderstandings arising from a lack of clarity, but these should be addressed or the writing improved. For example, I am concerned there may be some important errors with the surface mass balance. On the use of a flowline model, I agree with the previous reviewers that this is a serious limitation and the field is beginning to move beyond flowline models. On balance however, I do not think this should preclude publication provided this is clearly acknowledged, the paper focuses more on the main qualitative points of interest and the major comments below can be addressed.
For me, the two strongest points this paper makes are (i) that when upscaling results from a handful of glaciers to the whole ice sheet we should be careful about which and how many glaciers we sample, and (ii) quantifying the importance of dynamics-SMB coupling in future projections, though I found the discussion and figures on this point to be highly confusing. In general I felt the paper could benefit from focusing on/emphasizing/clarifying these points.
I have split my comments in major and minor comments. I have not noted every spelling and grammatical error as these are rather numerous and should be easy to spot.
1. Representation of melting and the coupling of melting to calving
A plume model is used to represent submarine melting of the calving front. When a floating tongue is present, this is naturally applied as a thinning of the tongue. When the glacier has no floating portion the submarine melting is applied as a thinning of the last grounded cell, which is less natural (it is essentially treating the submarine melting as a surface mass balance). Calving is represented via the commonly used crevasse depth criterion. My major comment is: how do we know that this treatment of combined melting/calving actually captures the effect of submarine melting on calving? How does it relate to emerging understanding of the coupling between melting and calving through undercutting (e.g. Benn et al. 2017, J. Glac.)? For calving fronts without a floating portion, how well does the application of submarine melting as a ‘vertical’ surface mass balance term represent the ‘truer’ process of melting incising horizontally into the calving front? At the moment it feels as though the authors have taken their approach because it is the one which works for their model rather than on the basis of representation of processes. Reaching the conclusions of this paper without discussing at all whether the model actually captures ice-ocean processes seems remiss. I see that this is very briefly mentioned in the discussion and conclusions sections but this is too little and too late at the moment.
2. Tuning of model
In order to initialize the model to obtain glaciers close to their present day state, the authors first vary 4 key parameters with a fixed (non-plume) melt rate and find a combination of these parameters which puts the glacier close to its present day state. They then turn on the plume model and tune 2 key parameters to maintain the glacier close to present day. The resulting set of parameters is used for the future simulations.
Overall, I have two main issues. First, do you know that the values of the first 4 parameters are the only values which would work? It is not at all obvious that there should be 1 unique set of parameters which work. This is important because a different set of parameters might lead to a different evolution in the future. Second, I am uncomfortable with how much the parameters vary from glacier to glacier (by two orders of magnitude in some cases); one would hope that if the model has a good representation of the physics then the parameters should take reasonably close values between glaciers. Once more this leads me to question how robust the future projections are. For example, it might be that due to uncertainties in other aspects of the model (e.g. bed topography), you have to tune up the melt rate a lot to match the glacier in the present day (high value of beta). But then presumably you are hard-wiring that glacier to be highly sensitive to ocean warming and prejudicing its future evolution. I am not convinced that this is physical rather than just an artifact of the initialization and missing model physics or poor input data.
3. Clarity of description – notably ocean forcing
I found certain aspects of the description of the model and its inputs rather confusing - in general I think the paper would be much improved if the description of the model and inputs could be clarified and simplified.
The most confusing part for me at the moment is the fjord temperature and salinity profiles. If I understand correctly, you use either the CTD profiles or the reanalysis data for the spin-up, and then you add the CMIP5 trends on top of the spin-up period to do the future projections. If this is the case it needs to be made clearer. It also wasn’t clear to me when the CTD data was used and when the reanalysis data was used. There is a long discussion of how the CTD data and reanalysis data differ, but ultimately this discussion comes to nothing because you use both anyway. This is one example of how this paper could be a lot more readable – perhaps move the detailed discussion comparing CTD and reanalysis to the supplement, allowing you to focus on describing what actually goes into the model. It would be great if you could provide an equivalent to equation 17 for the fjord temperature – this would really help the reader understand what is being done.
4. Use of CMIP models
The authors use 1 CMIP model for the surface mass balance (or in fact a regional climate model forced by the CMIP model) and 3 CMIP models for the ocean forcing. This disparity has been commented on by the other reviews and I am not convinced by the authors’ response. As the authors themselves state in their response to previous reviews, “climate change scenarios (both in terms of GHGs concentration and model output) are the major source of uncertainties.” This makes it sound like you might have reached different conclusions if you had used different CMIP models – can you be sure that your conclusions are independent of the CMIP models or that the CMIP models you have used are in some way representative of others?
5. Upscaling of SLR
I think the discussion on scaling up of sea level rise from a handful of glaciers to the whole ice sheet is very interesting and important. I wonder if this could be emphasized more in the paper by bringing supplementary figure 7 into the main paper and expanding the discussion? For a direct comparison to Nick et al. 2013, could you do the linear regression with the same 4 glaciers as in their paper?
6. Surface mass balance
I am a little surprised about how small the surface mass balance contribution is without dynamics (Fig. 11, brown). According to Fettweis et al 2013, MAR forced by MIROC5 results in SLR of 9.2 cm by 2100 due to surface mass balance alone. I appreciate this is for the whole ice sheet, but your 12 glaciers probably cover ~5% of the ice sheet area and therefore I would expect a rough SMB-only SLR of 0.05*92 mm = 4.5 mm from your glaciers which is much larger than your brown shading. Why is this? Possibly I am getting confused about your separation in Fig. 11 – what is the difference between the orange and brown shading? Could you clarify this in the text as well? For example, you say “that the SMB-forcing alone derived from MAR (without the glacier’s response) has an almost negligible effect on SLR (Fig. 11 b, brown curve)” – how can this be the case when MAR projects 9.2 cm of SLR for the whole ice sheet when forced by MIROC5 (Fettweis et al. 2013)?
Page 15, lines 16-18, Fig. 15 and Fig S1: Similarly, I find it hard to believe that the SMB contribution to sea level is negative for some glaciers under an RCP 8.5 scenario (e.g. Rink – Fig. 15). Looking at SMB anomalies by 2100 in MAR forced by MIROC5 (Fettweis et al. 2013, Fig. 5, bottom left panel), it certainly doesn’t look like any glacier would have an increasingly positive surface mass balance and it doesn’t look like there is any reason for Rink to be very different than Store (which is nearby), as is implied by Fig. 15. Can you check these numbers?
A second comment: you say in the introduction that you neglect the effect of ice sheet boundary retreat on subglacial discharge. Do you also neglect the effect of ice sheet boundary retreat on SLR from surface mass balance? In other words, are you still summing up the surface mass balance contribution to sea level from areas where the ice sheet has retreated (e.g. the 30 km over which Jakobshavn is projected to retreat). If so, presumably you might be substantially overestimating the contribution to sea level from SMB?
In general I was quite confused by how you are splitting up the different components of sea level - could you make this very clear (e.g. particularly the difference between the brown and orange shadings in Fig. 11)?
Page 1 line 19: you indicate that 70% of SLR is associated with a response to increased submarine melting. Could you clarify here and throughout the paper exactly what is meant by this? Is it that increased calving and submarine melting alone are accounting for 70% of SLR, or is it that increased calving and submarine melting together with decreased SMB due to dynamic thinning are accounting for 70%? This is an important distinction – the latter possibility would include a dynamics-SMB coupling whereas the former is pure dynamics.
Page 3 line 28: you have assessed the uncertainty for calving and melting parameters (at least for a single calving law) but in relation to climate scenarios you have only considered RCP8.5 and a single CMIP model on the SMB side – so I think you have not really quantified uncertainty related to climate change scenarios and maybe this statement should be removed.
Page 5 line 10: ‘initial boundary condition’ – it would be clearer if you changed this to the ‘boundary condition at the ice divide’ or ‘boundary condition at the top of the glacier catchment’.
Section 2.2: it would be good to acknowledge briefly the extent to which this plume model approach captures what is known about submarine melting. E.g. it does capture vertical variability in melt rate within a plume (Jenkins 2011), but it can’t capture variability across the calving front due to presence/absence of plumes (Fried et al. 2015), and it can’t capture melting outside of plumes due to melt-driven convection (Magorrian & Wells, 2016) or fjord-wide circulation (Slater et al. 2018).
Section 2.3: it might improve the readability of the paper if some of these technical details which are not central to the main messages of the paper (e.g. the definition of the total submarine melting) could be moved to supporting information – just a suggestion so feel free to ignore.
Page 16 line 15: pvalue=0 is a bit meaningless – better to say p<0.01 or similar.
Page 18 line 19: I think it would be more natural to state the proportion of SLR which is attributable to dynamics (i.e. the dynamical response of Greenland’s outlet glaciers can account for 5/13.8 = 36% of total sea level contribution from the Greenland ice sheet).
Table 2: some of the column headings have a ‘Delta’ symbol in them which are not mentioned in the figure caption – are they meant to be there?
Figure S2 – you say here you used Bedmachinev3 topography but in the main article you state you used Bedmachinev2 (page 8 line 18). Which was it?
Supplement – please improve figure captions throughout – at the moment they are sloppy. For example Fig S4 – annual subglacial discharge from what? Fig S5 has two missing references.
References which are not in the paper
Fried, M. J., Catania, G. A., Bartholomaus, T. C., Duncan, D., Davis, M., Stearns, L. A., et al. (2015). Distributed subglacial discharge drives significant submarine melt at a Greenland tidewater glacier. Geophysical Research Letters, 42, 9328–9336. https://doi.org/10.1002/2015GL065806
Magorrian, S. J., & Wells, A. J. (2016). Turbulent plumes from a glacier terminus melting in a stratified ocean. Journal of Geophysical Research: Oceans, 121, 4670–4696. https://doi.org/10.1002/2015JC011160
Slater, D. A., Straneo, F., Das, S. B., Richards, C. G., Wagner, T. J. W., & Nienow, P. W. (2018). Localized plumes drive front-wide ocean melting of a Greenlandic tidewater glacier. Geophysical Research Letters, 45. https://doi.org/10.1029/2018GL080763