Dear authors,

in line with the reviewers, I find this version of the paper much improved over the previous version. However, based on the evaluation of reviewer 2 and on my own reading of this manuscript, I have come to the conclusion that another major revision is necessary before this paper may become suitable for publication.

Such major revision does not necessarily require a major amount of additional work, but possibly a major change in the notion of this paper regarding the superiority of ORB over SIFT and SURF. I agree with reviewer 2 that the current evaluation is not convincing. In particular, apparently using the same images that were used for tuning ORB to then evaluate the performance of ORB is not convincing, given that the other two algorithm were not specifically tuned. Because of this, in my opinion this paper does currently not "prove" that ORB is the best option for feature tracking of sea ice on Sentinel-1 SAR imagery (l.371 f.), but primarily shows that ORB can be tuned to give great results.

I see two rather straight-forward possibilities to address these concerns:

1. The data basis on which the relative assessments of ORB vs. SIFT/SURF is based could be increased by applying all algorithms to additional images, without retuning. If ORB still remains clearly superior, this gives much more credibility to the notion that this really is a better algorithm

or

2. The entire notion of this being a fair evaluation could be removed from this paper, which then becomes much more a description of ORB and an indication that ORB can be useful. This would primarily require a toning down of all references regarding quality of the algorithms, with more reference to future work.

Given the positive evaluation of the two other reviewers, and given that I also find this paper a generally interesting read, I very much hope that you will be able to address these concerns.

Please get in touch if any questions should come up.

Thank you very much for submitting your work to The Cryosphere,

best,

Dirk Notz

Dear Stefan,

thanks for the revised version. I find that the evaluation has gained substantially through the additional comparisons you carried out. I am hence happy to accept the paper for publication in The Cryosphere.

I would recommend that you add, for the final version, a sentence or two in which you explain that the 43 images have not been used for the tuning of the algorithm and hence allow for an independent evaluation.

Thank you very much for submitting your paper to The Cryosphere, and for the constructive work on the paper in response to the reviewers' comments.

Best,

Dirk