Dear Maximillian Van Wyk de Vries and Andrew Wickert,

Thank you for the time and efforts you put in responding to the reviewer comments. I read your replies with great interest and appreciate the detailed answers you gave to the various issues raised by the reviewers (particularly the critical notes by reviewer 2). I was also pleased to read you feel that the suggestions made by reviewer 2 not only helped to improve the manuscript, but also the code itself. From your replies, it is clear that you will change the manuscript in many places: it will be necessary to specifically highlight the changes performed (i.e. include the new/updated text) in the rebuttal letter, which will make it easier for the reviewers to assess the changes performed.

A few comments from my side at this stage:
• I think that you argue well why ‘The Cryosphere’ is a suited journal: you do indeed present an easy-to-use GUI here, and the whole manuscript is very much centered around glaciology topics. In the end, you want to provide a tool that glaciologists can use to easily derive surface velocities, without having to look into the details/code behind this. It would, nevertheless, be good to have some additional graphical information about the structure of your code and the workflow in the main text to give the reader at least an intuitive feeling for what is going on ‘behind the scenes’. The second reviewer suggests you to add such elements (‘workflows’ and ‘dashboard’; Fig. 2 and Fig. 3 from Perks, 2020, GMDD are indeed good examples), but from your answer it was not entirely clear whether you plan to include this.
• You mention that “We were unable to locate a copy of PyCorr software on the internet (although it is mentioned in several papers, as you have highlighted), and are unsure if it is available for use by other teams”: it would good if you could directly contact the people who developed PyCorr to clarify this, and (potentially) update your answer/manuscript based on this.
• From your answers it is clear that you will edit some figures in the updated manuscript. I hereby also appreciate that you took into account some of the initial editorial comments made during the ‘access review’. It would also be good if the reviewers could answer some of the other (minor) issues that were initially raised (e.g. related to Table 1, rainbow colorbars,...etc)
• Removing the ice thickness reconstruction (as suggested by reviewer 1) is indeed fine, as it is more an application of what comes out of your toolbox than a result of the toolbox itself.

At this stage, once you submit the updated manuscript and the corresponding rebuttal letter, it will go through another round of reviews to assess whether the changes you performed satisfy the reviewers.

I look forward to a fruitful continuation of the review process and do of course remain at your disposal if you have any additional questions.

Best regards,
Harry

Dear Maximillian Van Wyk de Vries and Andrew Wickert,

Thanks a lot for your elaborate and detailed rebuttal letter and for updating the manuscript. Given the constructive, yet quite critical remarks by the anonymous reviewer during the first round of review, the updated manuscript will now be sent out for another round of review.

Best regards,
Harry

Dear authors,

Your manuscript has been reviewed a second time by one of the original reviewers. This reviewer remains very critical about your work, and still sees major issues that prevent her/him to recommend your manuscript for publication (in fact, the anonymous reviewer suggests a “rejection” at this point). Given the contrasting views on the manuscript, with one reviewer (Ted Scambos) supporting your submission and the other one (anonymous) being very critical, the review process will be continued.

I first invite you to answer all new queries raised by the anonymous reviewer, after which the manuscript will be sent out for review again. In answering the issues raised by the critical reviewer, it is important to go in detail and provide quantitative information where possible. In particular, pay attention to the comment by the reviewer stating that: “Secondly, the response by the authors is brief and no effort is put in backing-up their argument. The majority of answers include phrases like, "we find little difference", "we found no clear improvement", but results or information is lacking. Thus it is difficult to get convinced, especially since the points I have raised are backed up by references. One can not expect me to run the test for you, in my opinion the burden of proof needs to be at the side of the authors”. When answering this question, I would also like you to expand on some of the answers you previously gave, by providing additional quantitative information (and update the manuscript accordingly) on:
• "we find little difference between 8 bit and 10 bit 70 data even in shadowed/clouded areas. The orientation filter tends to cancel out contrasts within the data. The 10 bit data may prove advantageous where orientation filtering is not suitable and simple low-pass filtering and/or contrast limited histogram equalization (CLAHE) is used
o What is ‘little difference’?

• "On some glaciers we have found that near infrared Band 8 of Sentinel 2 (842 nm) may produce a better feature contrast than other bands on some glaciers. We have also found that shortwave infrared bands 10 and 11 (1375 and 1610 nm) can be suitable in some cases (high contrast between ice and supraglacial debris), however suffer from a lower spatial resolution"
o ‘may produce better feature contrast’: by how much? And can you specific about the resolution where the problems originate for the shortwave IR?

• "We found no clear improvement in matches relative to frequency domain cross correlation even in areas with significant ice shear (glacier margins with around 1000 m/yr velocity gradient across one kilometre). We also found that the multi-pass frequency domain matching would usually outperform single passes in both frequency and spatial domain. However, we found that changes to the pre and post-processing would often have the greatest improvement on final velocity maps, so have emphasized this aspect in GIV. This is indeed less of an issue within fluid dynamics where the sampling rate, tracers and illumination 110 may be controlled."
o You discuss the (lack of) ‘improvement’, several times. Can you be more specific / quantify the (lack of) improvement

• "We test an implementation of ‘neighbourfilter.m’ using nlfilter, however it results in little overall speed-up and some unexpected artefacts. "
o Can you quantify the speed up and describe the unexpected artefacts?

The comment about the suitability of the journal (‘The Cryosphere’) has been addressed sufficiently in my opinion, and will not be a point of debate in the final decision (as I already indicated before). I agree that the study could have been considered for a more technical journal, like Geoscientific Model Development, but I am still convinced that The Cryosphere is a suitable journal (given that the study fulfils the scientific standards of this journal, which needs to be further assessed through peer review).

Thank you for your efforts,

Best regards,
Harry

Dear authors,

Your manuscript was now reviewed by a new/external reviewer. This reviewer has several (some rather technical) comments, which I now invite you to address in detail (point-by-point rebuttal + manuscript update in accordance). This includes a comment related to the suitability of TC as a journal, which I also ask you to answer for the record. As previously indicated, as an editor, I understand a part of the critique by the reviewers, but I nevertheless also think that TC is a suitable option and therefore support your choice.

Many thanks for your efforts,

Kind regards,
Harry

Dear Maximillian Van Wyk de Vries and Andrew Wickert,

Many thanks for sending in a new version of your manuscript and accompanying rebuttal letter. Your manuscript is now in a good shape and almost ready acceptance. I have formulated a list of mostly minor and technical comments that I would like you to address when resubmitting your manuscript, before proceeding to a final acceptance. Pay particular attention to the remarks related to the figures, which will need some reworking in some cases:

- l. 4: “...and often a safety risk”: sounds a bit odd to me. Maybe change to: “...and often associated with safety risks”
- l. 6: “...hence, velocity over time.”: just to make this entirely clear, suggest referring to “...hence, surface velocity over time”
- l.8-10: please be consistent when referring to glaciers and ice caps. You should refer to all ice bodies as glaciers (in line with IPCC) or instead consistently refer to certain ice bodies as ice caps (in line with for instance EGU, which mentions ‘ice caps’ in session titles and has a ‘science officer for ice caps’): you now refer to Vavilov ice cap as a “very large glacier”. Probably easiest to not refer to ice caps (in line with IPCC): i.e. “...as well as a tropical glacier (Volcán Chimborazo)”. Please make sure that this is consistent throughout the manuscript.
- introduction: is really very broad and elaborately touches upon topics that are not treated at all in this manuscript. You can have a general introduction, but it should become more compact: l.15-37 should ideally be compacted to max 10 lines, where the only part really relevant here seems to be around l. 28-32 + l. 35-37 (add references for these statements)
- l. 31-32: ‘Wal et al., 2008’ should be ‘van de Wal et al., 2008’. Could also refer to some more recent works that focus on remotely sensed velocities on glaciers here, as this is the main topic of your paper (vs. ice-sheets): e.g. Altena and Kääb (2017, Frontiers in Earth Science), Altena et al. (2019, The Cryosphere) and Dehecq et al. (2019, Nature Geoscience).
- l. 76: “...and lighting conditions depend strongly...”: maybe also explicitly mention the shadow(ing)?
- Table 1: very useful table for the reader! A few remarks. For EMT: any indication about the environment/language in which the ‘worklfow’ is? Should ‘matpiv’ not be ‘MatPIV’? I found it a bit strange to have the explanation in the caption go from 3 to 1: maybe change order from 1 to 3?
- Figure 1: width of the individual boxes seems a bit random: why having them smaller in steps 1-3-4, then wider for step 5, and then smaller again: suggest having the same width (except step 5 maybe, as you want to emphasize this I guess? Maybe here also explain in caption why this is shown in bold) or having different widths to have content on single line (e.g. for step 8 and step 9)
- section 2.2: was easy to follow in general, also as a non-specialist. I was wondering if the readability would not further improve by subdividing this section in several subsections, as quite different things are treated here: e.g. l.118-131: frequency-domain matching, l. 132-144: single- and multiple pass approaches, l.145-148: move this to part where FFT is introduced, l.149-156: parallel computing, l. 157-161: non-consecutive imaging, l. 162-178: accuracy assessment and final velocity map improvement, l.179-200: temporal resampling through iteration, l.201: georeferencing
- l. 139: ‘matpiv’ (two occurrences) should be ‘MatPIV’
- Figure 2: nice visualization! Could you indicate which area/glacier we are looking at here? Add this information in caption, and potentially complement with coordinates (lon-lat) in image.
- l. 149-156: very nice to read about this option to parallelize the code. Small question here, which may be related to my misunderstanding: how are you accounting for displacement that occurs between the different image pairs (which are treated on different cores)? Maybe add a short 1-2 sentence description of this, or explain how this is not a problem (I guess others may have the same question?)
- l. 177: “if the dataset is smooth enough to allow it”: how is this determined? Criterion for this? If so, could you give a short indication about how this is done?
- Figure 3: please do replace the red or green color by another: is problematic to have both colors for line data for color-blind people (deuteranopia) (I noticed that you took this into account for other figures as e.g. mentioned in l.203-204: nice!). In panel b: how is the speed-up defined? As a fraction I guess? If so, probably more intuitive to formulate as % speed up.
- l.190-192: found this sentence difficult to follow as it is lengthy + use of numbers for weight, while ‘one’ is used to refer to velocity as well. Consider splitting sentence in two and reformulating second part for clarity.
- l.195-196: “Outlier detection and maximum velocity filters are implemented”: could you provide some (even compact) information about how this is done / based on which criteria?
- l.197-198: ” ... are not be adapted...” should be “...are not adapted...”
- Figure 5a: change the color scheme to one that is suited to represent sequential data! The color scheme that you use here is suited for diverging data. See e.g. https://colorbrewer2.org/ This is of large importance, as you refer to this as: ‘generation of publication-quality images of the velocity and flow direction maps’. I agree that the figure looks nice, but the color scheme you use does not align with your purpose.
- l. 219: write out the two occurrences of ‘3’: ‘three’
- l. 221: ‘based off of...’ : should this be ‘based on...’?
- figure 6: in the text you refer to the real names of the glacier (Spanish), while in the figures you use the English names: be consistent. For the location map: would it maybe make sense to have this a bit more focused on Patagonia? I guess most people know where Patagonia is, but have little idea where the two glaciers are in Patagonia (e.g. are they 20 km apart from one another or 500 km?)
- figure 7: same remark as for figure 6 concerning the location map. Where in the European Alps is this glacier vs. where are the European Alps.. On main map: very hard to see where North is: could you make this clearer? I find it quite counterintuitive to not have the north points upwards, as is the case in other studies focusing on this glacier (e.g. see figs. 1b and 2 in Rabatel et al., 2018, Frontiers in Earth Science).
- l. 250: “Validating GIV...”: you cannot really ‘validate’ your model by ‘observing’ dynamics of a glacier. I would rather refer to this as an evaluation, and would in fact suggest to simply remove this part to be consistent with other section heading: i.e. renaming this to ‘3.3 Vavilov ice dynamics’
- l.252: You refer to ‘Arctic land-ice’ here: not clear if this includes the Greenland ice sheet (which is also Arctic land-ice...) or not, as seems to be the case when reading the next sentence (if so: be more specific and refer to the glaciers and ice caps explicitly)
- l. 255-266: many of the elements you mention here are really based on our knowledge of glacier surge: would be good to also refer to more general which this is explained, such as Sevestre and Benn (2015, JGlac; which also explains the phenomenon of glacier surging for glaciers in the Russian Arctic)
- figure 8: same remark as for figure 6 and 7 for the location map. Why did you choose this color bar for the velocities? Very difficult to interpret and seems to be a color scheme that is meant to plot landscapes in fact.. Please change the color scheme to a more classic one that is meant for sequential data! In the caption of the figure, remove ‘present’, ‘displays’ and ‘present’.
- l. 283: ‘...if associated changed in...’ should be ‘...if associated changes in...’
- l. 284: ...’whether a similar peak occurs in 2020‘: not up to date anymore: incorporate 2020 in your explanation / figure above, or change the text here, possibly referring to 2021 instead of 2020.
- l. 286: ‘Method validation’: comparing your results to those from another study is not a validation, but rather an evaluation. Moreover, it may be questioned whether you can evaluate your results by comparing them to another product which also has substantial uncertainties and potential large artefact (as you also mention, in e.g. l. 297-298). A real evaluation would for instance compare the GIV velocities to on-site high-precision velocity measurements based on GPS. Suggest removing the subsections 3.3.1 and 3.3.2 here and having the entire explanation under the section 3.3.
- l.300: ‘Many tropical glaciers and ice caps have limited to no ice-flow data’: well, they have a lot of data based on techniques such as the ones you present here. So be more specific here, e.g.: ‘...have limited to no ice-flow data from direct field measurements’
- l.303: making the bridge from ice velocity to ‘practical decision making’ is quite a big step: how are these connected? Or maybe reword to: ‘...provide information on glacier state, which can contribute...’
- figure 9: remove the ‘shows’ and ‘is’ from the caption.
- l. 305: “...capped with an ice cap”: maybe “...covered with an ice cap”
- l. 317: you mention the runtime for this simulation, but you did not mention it for the other examples. Be consistent. Ideally, summarize all the runtime information in a single table, which could be added to the suppl. mat. Would also be useful, as directly after this (start of Discussion) you mention the computational aspect.
- l. 328-340: summary of other feature tracking algorithms: this information should appear earlier, in the introduction. In the discussion section, you should really focus on... the discussion (l.341-349), instead of giving a long summary of existing toolboxes. In fact, the real discussion is now very short (as is the conclusion): suggest merging the discussion and the conclusion in a single section.
- l. 376: mention that this is a PhD thesis

Thanks a lot for going through this list of comments and updating the manuscript accordingly. I look forward to receiving a new version of the manuscript, which could then be considered for final acceptance.

Best regards,
Harry

Dear Maximilian Van Wyk de Vries and Adrew Wickert,

Thanks a lot for going through the list of final comments so rapidly. I had a detailed look at the rebuttal letter and updated manuscript, and I am now very happy to accept if for publication.

I am convinced that your manuscript and GIV will be valuable to our community, hence my decision to follow your choice to have opted for TC (vs. for instance GMD, as suggested by reviewers). I realize that your manuscript has gone through a relatively long review journey in TCD, but I hope you agree that the input from external reviewers has helped to better focus and present your results.

Thank you once again for choosing for TC, and good luck with all your ongoing and future work,

Best regards,
Harry