Dear Authors,
After two reviews, authors response, manuscript revisions and a further re-review of the revised version, I decided that in principle the paper should be published but there are some remaining issues that should be addressed before acceptance. I try to summarize my reasoning for the decision below and based on the re-reviewer make suggestions for further revisions.

The reviewers picked up a range of smaller rather technical or language issues which have been addressed well in the revised version.

Also, the undertaken research is in the revised version well explained and clear and the quality of the figures ok.

One of the main issues, highlighted by both reviewers and that remains to some degree in the revised version, is the significance and related interpretation/conclusions from the results of the initial submission.

Although the detailed effects and related explanations of grounding line treatment in the Elmer-full-stokes model may technically be interesting, it is rather narrow in its scope (very specific experiments, one type of model) and more critical, it is an issue that seems to be strongly related to insufficient spatial resolution, and the effect from the groundling line treatment seems no bigger than the error from insufficient spatial resolution.
I agree with the reviewers that this of course hampers the significance and value of the paper to some degree, or as one reviewer puts it ‘…I (the reviewer) don’t see the point of comparing different basal drag parameterizations for numerical models that are still mesh dependent…’.

The authors argue and justify the significance of this paper in relation to the importance for the MISMIP experiments and that the used model serves for that as a reference as a full stokes model.
I agree that the MISMIP intercomparison experiments take an important role evaluation and development of grounding line motion models and thus I agree with the re-review that the grounding line parametrization issues of the full stokes reference case should be communicated, even though or maybe exactly because the MISMIP reference case is not at sufficient resolution.
Whether this justifies a full TC-publication or rather a condensed short communication (as suggested by the re- review) is in my view not that crucial. It is important, however, the issue is communicated to the marine ice sheet modelling community and which in my view as an editor justifies a publication in TC (in whatever format).

I agree with the reviewers that the focus and weight of the paper should not overstate this issue of grounding line parametrization too much, and better integrate the findings of the experiments into the context of grid size dependency.
The revised version addressed this issue to a large degree (removed ‘substantial’ in titel and clarifies that issue is related to specific MISMIP resolution,…).
But I agree somewhat with the re-review that the relevance of the paper is perhaps still too much on this basal drag parameterization and not well enough integrated into the context of the grid size dependence (which dominates the errors). The re-review therefore suggests to reduce some parts on the parametrization sensitivity and focus it more towards the context of grid size dependency and pack it as a short communication.

Although, regarding significance a ‘short communication’ is perhaps more appropriate, I have no strong opinion on the type of publication and could potentially also envisage a full TC publication. But I would recommend (consistent with re-recview) to better stream line and the integrate the grid size dependence into the context of the basal drag parametrization issue. This would likely lead to a shorter more concise publication that could be changed to ‘short communication’ if the authors which so.

Specifically such streamlining could be done by:

• Taking sections such as on the diagnostic experiment (section 3and in particular 3.2 (P75D-experiment) out and move them perhaps to the supplements (as they do not tell us much about errors, and are a rather technical and lengthy discussion of differences in initial condition for the experiments in section 4. What is important that the initial positions for experiments in section 4 are different).
• Give the context of grid resolution (or insufficient resolution) in the aim/motivation (introduction), interpretation and conclusion a bit more consideration (clarify this context from beginning). This should include a clear discussion of errors (no bigger than from grid resolution) and convergence with decreasing grid size (independent of method).
• Maybe take Figure 4 out as it is almost redundant when also showing Fig 5 (which seems much easier to be read). Some of the Figures 7 to 10 could perhaps also be summarized or in parts be moved to the supplements.


A further point raised by the reviewers that may weaken the significance of this paper is that only the rather un-natural case of a step change from high to no friction is considered and in the real world the transition in friction from grounded to floating ice is likely much smoother and thus the investigated parametrization issue potentially not that important.
I agree in this point with the reviewers, but I think to focus on the MISMIP step-change case is consistent with the authors focus on the issues with the MISMIP reference case. Further, I agree with the authors and the re-reviewer that including the ‘smooth sliding’ transition case is beyond the scope of major revisions of this paper.

Editors decision
By Andreas Vieli, 31 Dec 2015

After the re-review, the authors followed the recommendations of the re-reviewer (Cornford) and of the editor and addressed all major issues satisfactorily. Specifically:

-The authors refocused and rephrased the paper now around gridsize dependence (convergence) of MISMIP experiments for the reference model Elmer/Ice.
-For this, they added also a new experiment (also recommended by reviewer Cornford) to investigate the impact of a smooth friction transition at the groundingline which substantially adds to the significance of the paper (which clearly addresses an earlier a point of criticism by the first two reviews). In particular, the obtained result of this additional experiment are not that obvious. The authors also already discussed their new findings with the reviewer Cornford by email (communication asked to be made public in review discussion).
-the numerical implementation of discontinuous friction and the related results are still included but are now well integrated into the mesh-size dependence investigations and the details moved to the supplementary material. The inclusion of this numerical implementation of discontinuous friction is well justified given the role of the Elmer/Ice as a reference model in the MISMIP intercomparison.
-They also shortened the manuscript and resubmitted it as a ‘short-communication’ which is now well justified (novel aspects of numerical model issues relevant for MISMIP’ and of short enough format).

Thus the current manuscript is close to be acceptable. Besides a few very minor points, there remains two more slightly more substantial issues that need addressing before final acceptance which should not be difficult and which are both related to the MISMIP3d experiments (most likely be a result of rearranging of parts of the previous version):

1) Steady state differences (diagnostic), lines 182 to 193: Somehow its is not so clear anymore that you refer to the steady state runs/positions here, this should be clarified near beginning of these two paragraphs. Specifically I would suggest the following:
• Line 178: it is no longer clear what experiment you mean here for the friction implementation comparison, and do not mean resolution instead of discretization? I assume you mean the steady state case (prognostic), so maybe modify to: ‘First, the steady state geometries for the three friction implementation are compared with similar grid resolutions…’

• Line 180: further I do not really understand what you mean by …you varied the element size of the mesh along main flow direction…, from above I thought the grid-resolution is the same/fixed. Clarify.

• Line 184: again you refer to the steady states here and ‘conduct’ seems awkward, so I would rather say ‘…the three methods result in three different steady state positions….’

2) for the prognostic experiments (probably more substantial), the paragraph from line 194-205 initially heavily refers to the Figure S4 in the supplement, which is not shown however in the main text. This is not ideal, as details of this figures are described that one is not able to follow at all without the figure. And I think these details are also not necessary here, the essence of the paper is on gridsize dependence now (and the method comparison is useful but not the major part). I assume the authors could not add another figure due to the short communication format (space) and thus moved it to the supplement.
I suggest to the authors two options of how they could address this problem:

-Option 1 (which is my much preferred one): Reduce this paragraph to a summary of the main message from the figure: e.g. …In the P75S experiment (for the coarse transverse resolution Ny=20) we found that the results in transient responses (relative to the initial positions) differ substantially between the three methods (see supplement Fig. S4), also for the reversal of the perturbation. You then continue with your paragraph on line 206: …Such large differences in transient….
Maybe you could add a little description/discussion paragraph in the supplement if you are keen to explain these details more.

-Option 2 (but not that ideal): you move Fig. S4 to the main text as Fig. 3. But then I would expect some substantial shortening of the text on the diagnostic and transient part Lines 178-205.

Further I would clarify a bit what is shown in Fig. 3 by changing the text on line 213 to; ‘The results for lateral resolutions with Ny=20, Ny=40 and Ny=80 grid points are shown in the Supplementary Figures S4, S5 and S6 and the differences from such lateral resolution visualized relative to the highest resolution Ny=80 in Figure 3.’…


Minor points

Line 41-43: i would add before the two questions (on line 41): ‚This raises the following questions. Is the ....’

Line 65: is there not a ‘problem’ missing here: ‘….contact problem between…’ (you can not solve a contact slone, but a contact problem.

Line 97: should it not rather say ‘formulation’ instead of ‘formalism’

Line 124: ‘result’ (plural) and nor ‘results’

Line 124: I would say ‘different flow approximations’ instead of ‘different models’

Line 127: please check with the Cornford whether he is happy with this formulation of the personal communication. I will try and add the email conversation to him to the review discussion part on TCD.

Line 133-138: this is a very long sentence and thus hard to follow, break it up.

Line 168: remove the ‘then’

Line 189: ‘…it should also be noted that….’

Line 225: ‘purpose’ does not seem the right word here: maybe ‘reasons’ is better

Line 234: ‘…different friction EVOLUTION….’ is awkward (evolution usually refers to time …), maybe change to ‘….different friction transition schemes…’

Line 244: ‘In doing so,…’ (rather than ‘so doing’).

Line 250: ‘…using mesh resolutions that are…’


Comments to Supplementary material:

p.4 line 2: verb should be in plural: ‘…models do not …’
p.4 line 11-12: you refer to a discussion on the transient case (at least this is what I expected) but there is no such discussion in the supplement, there is only fig. S4. So add such a discussion or detele this reference (but leave reference to Fig. S4).

Make sure the reference list and figures are kept apart (first figures then reference list).

Editor decision

The authors have fully addressed the minor points raised by the editor after the revisions and I am pleased to announce that the paper can now be accepted after one tiny technical revision:

p. 5 line 125: I am pretty sure it should say ‘…probably result FROM the two different…’ (and not ‘…probably result IN the…’.


In order to include the email conversation of the author with the reviewer Cornford I would suggest to include the author response document after the re-review (review of the revised version) which I will send to the authors in a seperate email. If the authors agree, I will ask TCD to upload this file to the interactive discussion of this paper.

Andreas Vieli