Machine Learning Approaches to Automatically Detect Glacier Snow Lines on Multi-Spectral Satellite Images
, Jean-Baptiste Thomas 1, Ivar Farup 1 and Jocelyn Chanussot 4
Round 1
Reviewer 1 Report
The full review report is attached as PDF.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Overall, the manuscript is well written and of some scientific interest. However, I think it would have been better if the following issues had been refined.
- More suitable title should be selected for the article.
- Abstract needs to change: the abstract should contain Objectives, Methods/Analysis, Findings, and Novelty /Improvement.
- The necessity and innovation of the article should be presented to the introduction.
- It is suggested to present the structure of the article at the end of the introduction.- More suitable title should be selected for the article.
- The discussion should reinforce the significance of the study and compare the differences between the current studies. In addition, it should also highlight the shortcomings and outlook.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
As an important indicator of glacier mass changes, snowline changes have been of interest to the academic community. This manuscript presents two methods to detect snowline and achieves similar results in a specific study area. Although both methods fail when snowlines are not continuous or at a fairly stable elevation, this study certainly provides a new solution for snowline detection. As such, it is a topic of interest to the researchers in the related areas. However, I personally think that the manuscript needs some minor improvement before acceptance for publication. My detailed comments are as follows:
(1) In the abstract, punctuation is missing at the end of the paragraph.
(2) Partial text in the titles of Figure 1 (a) and (b) is covered by Figure 1 (c).
(3) Lines 250-252: The authors note that for each glacier, the snow bin with the highest number of consecutive snow bins above it will be tagged as the potential snowline. Does this treatment have a significant impact on the study results?
(4) In Figure 6, the left figure shows a 4th order polynomial approximation of the elevation bin given the X coordinates, while the right one shows a 4D-polynomial approximation of the bins given the Y coordinates. Why was the same method not used in both? Or, is it a misrepresentation? I am a bit confused about this.
(5) In the illustration of Figure 6, the 4th order polynomial approximation of the elevation bin given the X coordinates is ineffective and achieves a low R2-score. So, did the authors consider other methods? In addition, is “achieve” a misspelling of “achieves” ? is “approximations” a misspelling of “approximation” ?
(6) Other minor issues, such as the writing format of 4D-polynomial should be consistent in Figures 6 and 7, and the number 2 in R2-score should be superscripted in the title of Figure 6.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
The authors present applications of machine learning techniques to determine the snowline on glaciers using multi-spectral imaging. The paper is write well written with a couple of grammar mistakes, Gaussian not gaussian, for example.
The only comment i have is about transferability, which is to say you mention in the discussion about using more observations, and having to retrain the algorithm, but i think there should be a discussion about how well this approach would work in other locations.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
