|I would like to applaud the authors, this revised version of the article is substantially improved. Below are minor comments with two specific points raised about overlap between this work and Tielidze and Wheate, 2018, and the ability to detect a debris cover change signal. After a consideration of the comments below, I would recommend publication of this article in The Cryosphere. |
P1L31: Missing “per” year superscript “-1”
P1L38: Add this citation along with Nicholson et al., 2018: Østrem, G., 1959. Ice melting under a thin layer of moraine, and the existence of ice cores in moraine ridges. Geografiska Annaler, 41(4), pp.228-230.
P2L1: I think it’s either “energy balance” or “mass balance” not “energy mass balance”.
P2L15: “Ice and snow melt in these mountains are major sources of runoff for populated places” Is this common knowledge or is there a citation for this?
P2L19: I think Scherler et al., 2018 “measured” that debris cover is abundant, not “suggest.”
P2L20: more than 25% [of] glacier area
P2L21: Add citations for the earlier studies that indicate low relative supraglacial debris cover in the Greater Caucasus, or if already present in the three citations at the end of the sentence, move the relevant citations to the middle of the sentence so it is clear which reference is to which statement.
P2L33: For consistency with how you set up this list, give the ratio (130/0).
P2L34: northern/southern slope ratio is not given for the Elbrus massif and not divided by aspect in later analysis (although the eastern slopes are referenced). Please explain why or add this.
P3L7: is 1986 to 2014 explicit or do you mean 1985/86 to 2013/14?
P3L10: It’s not clear to me what work is new here and what is from Tielidze and Wheate, 2018. Looking at this earlier article, effort was taken to include debris cover. If you are using the same work then you do not need to describe the mapping methods in this article, simply cite your earlier work and present your new analysis of debris cover changes. If you have made changes to the earlier inventory, then I think a missing result is the difference between these glacier outlines and those of Tielidze and Wheate, 2018 for the subset regions considered here.
P3L21-22: Manual improvements correcting for snow and shadows, does this step also include mapping debris cover?
P3L24: “Supraglacial debris cover was extracted and saved as a separate layer.” You can remove this sentence, your readers know you did this implicitly.
P4L16-17: “The accuracy is +/-30 m for debris and +/-15 m for clean ice” How did you derive this result? Is this an average value for all seven glaciers?
P6L11: make it a little more clear that these results are no longer per your subregions. Something like “For all regions investigated in the Greater Caucasus mountains, the rate of supraglacial debris cover was different between northern and southern aspects.”
P6L22: missing superscript “-1”
P6L23: (and P6L26): rewrite; something like “[..same period] with a terminus retreat of”
P7L24: “northern slopes are less steep” do you mean the mountain slope above the glaciers? How would this couple with more debris cover? Can you take this one logic step further?
P8L2-3: “..although the total uncertainty is comparable to the obtained relative changes” I have a lot of respect for honest results that fall within suitable error bounds, however, there are clear changes between 1986 and 2014. Why was your method unable to confidently resolve this? In this discussion can you propose a method that might be more successful?
P8L3-4 “Comparison with semi-automated methods shows that debris cover may be considerably underestimated.” First, it is unclear what you are referring to as the semi-automated method, if I follow correctly, this is the bare ice area that was subtracted from manual outlines to derive a debris cover so how could this dataset provide the comparison to make this statement? Beyond descriptive clarity, I’m inclined to agree with the general statement. I had a quick look on Earth Explorer and compared the image from 1986 against your Figure 4 (see attached figure). I added some arrows where Figure 4 misleadingly shows changes, e.g. an entirely new medial moraine network (labeled ‘a’), area that is shown to gain debris cover but if this is the case then the glacier would have had to have grown not shrunk over the investigation period (‘b’), and what looks like two rock avalanche deposits that are missed (‘c’). I think it is a fine argument that these points are within your error bounds and I acknowledge that it is difficult to maintain consistent methods with debris-covered area change measurements because earlier sensors do not have the radiometric resolution that the later sensors do. However, the images do show true changes and it is the challenge of this line of research to be able to extract a meaningful signal.
P8L5: “lake number and area”
P8L7: I do not follow the evidence for an increase of debris cover being explained by resurfacing of englacial debris. Is there a pattern in how the debris cover looks that would lead you to make this assumption?
P8L9: what boundaries in the subsurface are you looking for and why?
P9L1: please make statements without ambiguities like “somewhat”
P9L6: end the sentence after “(2008)”, and I believe you meant to say Rowan et al. (2005) found similar model results.
P10L10: the list of detailed field measurements: “debris thickness, GPR, radiation” give a physical quantity, a geophysical tool and an emission of energy, please rewrite to list measurements.
Figure 2. This figure is very nice. Could you please add the error bounds? Also, the wording “clean ice to supraglacial debris cover [ratio]” should be reversed, the ordering implies numerator and denominator.
Figure 4. I’m not questioning the area you show as clean ice in 1986 that is no longer clean ice in 2014 high on the massif, but area that is former glacier and now bedrock is distinctly different from a discussion of debris cover changes on a glacier’s surface.
Figure S4: A figure showing a permafrost dataset with no real context or discussion seems out of the scope of this paper.