Dear Basile de Fleurian and co-authors,

Thank you for the responses to the two reviewers. Their reviews indicate that the paper is strong and will be a valuable contribution, once a number of minor issues are treated. I agree with the points they brought up and with your responses to how you will address these points in your revision. I look forward to reading the revised manuscript.

Cheers,
Kristin Poinar

Dear Basile de Fleurian and co-authors,

Thank you for your revised manuscript and thoughtful response to both reviewers' comments. The manuscript is now improved from its original state and is nearly ready for publication. I have the following list of improvements that are still needed. Though it is rather lengthy, none are very onerous and I have every confidence they can be met.



List of moderate changes needed:

I appreciate the enhanced description of the dual-continuum model (lines 90-102). However, it is problematic because it is word-for-word the same as the description in de Fleurian et al. (2016). This portion of the manuscript needs thorough paraphrasing before it can be published in The Cryosphere.

Figure 1 inset -- It would be clearer to understand if the y-axis were "Runoff" instead of "Temperature". I get that you used a PDD so your base forcing truly is temperature, but the processes that you are studying relate to runoff, and runoff is more frequently discussed in your text (as well as shown as the forcing in all other figure panels) than temperature.

Figure 4 should include altitude values on panels b and c as well, not just panel d.



List of minor changes needed:

Abstract -- remove or rephrase "surprisingly" as it is imprecise / unscientific. Similarly, remove or rephrase "and/or" as it is informal.

Clarify which of the sensitivity runs (low / medium / high runoff) is the reference simulation. It is fairly intuitive that it should be the medium one, but this is never stated.

The elevation of the runoff limit is unclear. I understand that it varies between runs (low, medium, and high runoff sensitivity runs). Presenting this information in either text, a table (perhaps Table 2), or on figures would be helpful, as I found myself searching the manuscript for the runoff limit. From Figure 3 I could infer that it lies between 1215 m and 1465 m for the reference simulation.

SSUmax should be defined specifically. Currently it just says "this shift in behaviour" (line 234), although I think that SSUmax is actually an elevation. What SSUmax stands for should be defined, and what the "max" refers to -- elevation or speed? -- should be stated as well.

line 306-7 -- a few extra words here, "the" and "to"

line 328 -- "more larger" should be corrected

Table 5 -- Instead of "Colder temperatures", it should say more directly what the experiment actually was: "lower intensity of runoff" (or something similar), and same for "Warmer temperatures".

line 361 -- similar to the comment above, replace "colder"

line 379 -- instead of "strong" velocities, this should say "high" or "fast"

line 385 -- rephrase "rather inefficient efficient draining system"

line 403 -- rephrase the sentence starting "However, we think..." in order to make it more clear which you are hypothesizing (that there is a reasonable physical explanation) and which is fact rather than speculation (that your results are robust).

line 454 -- "pauses" should be "poses"

line 462 -- specify that you mean _spatial_ distribution here. Further along in this paragraph, you mention that moulin location is likely unimportant, but would you care to address moulin density (e.g. Banwell et al., 2016)? Essentially, your model has a moulin at every grid node, so it is maximally dense. Finally, the last three sentences in this paragraph (lines 469-473) are a bit convoluted and should be smoothed out to make your point clear: that the highest injection point in real life is not likely to change from year to year. As currently written, the reasoning switches back and forth with "however"s that seem to me to be misleading (i.e. not actually in contradiction with the previous point). Perhaps this made me misunderstand the basic argument. Please check and revise accordingly.

line 475 -- "without performing the actual simulations" -- do you mean simulations with real-life runoff inputs?

line 501 -- change "constant runoff" to something that specifies that the time-integrated runoff volume is unchanged from simulation to simulation. As is, "constant runoff" could be read as unchanging in time.

line 502 -- change "originally" to "initially"

line 512 -- "led" seems like not the right word here and is redundant with "at the beginning of the melt season"

line 514 -- I believe you're actually advocating for use of models with _both_ inefficient and efficient components, yes? As written now, it could be interpreted that a channel-only model is a good approach. Please revise.



With the above suggested changes, I will be very happy to see this work in The Cryosphere. Thanks for work and attention to addressing the reviewers' comments.

Cheers,
Kristin Poinar

Dear Basile de Fleurian and co-authors,

Thank you for the changes to your manuscript. I find that it is nearly ready to publish. However, I am returning the manuscript to you one more time for the opportunity to clarify the discussion paragraph on injection points, which you already edited some in the previous round.

I still find your argument difficult to follow through this long paragraph (lines 433-451 in the clean/undifferenced version). It starts out on the topic of low-elevation velocity driving the domain mean. Soon it switches topics -- in the middle of the paragraph (line ~438) it seems that you are saying that your model results (as compared to what we might expect in reality) may be sensitive to its assumption that runoff sites = moulin sites. However, by line 446, it seems you have changed your mind on that (with support from Sholzen et al. and SHMIP). It would be clearer if you wrapped this up with some statement that you don't in fact think that runoff sites = moulin sites is a problematic assumption (probably citing Sholzen here). Next in this paragraph, you switch to a new uncertainty, the highest injection point. Your conclusion on how sensitive your results are to this stays unclear to me by the end, line 451. This needs to be treated more thoroughly and, again, your overall verdict on the matter should be stated.

My recommendation is to break this long paragraph of three topics into 2-3 paragraphs that each treat the separate topics addressed, and do more smoothing out of each argument, including better organizing the findings from previous work into supporting (or contrasting with, if so) your own conclusions. I think that some attention to this section will greatly improve what readers take away from it and therefore increase the paper's impact even more.

Cheers,
Kristin Poinar

Thank you for the edits. The discussion paragraphs are now stronger and easier to follow. I am pleased to accept this manuscript for publication in TC.