Dear Enze Zhang and co-autors,

Your submitted manuscript "Automatically delineating the calving front of Jakobshavn Isbræ from multi-temporal TerraSAR-X images: a deep learning approach" received two positive reviews where both reviewers mentioned the importance of the application and study of emerging machine learning techniques towards automatic calving front detection.

Both referees had several comments which have been addressed in the revised version of the manuscript. The revised manuscript was sent again to reviewer R2, who suggested (similar to reviewer R1) to accept the paper subject to minor revisions explained below.

Given this recommendation, I suggest to incorporate these comments in your revised version.
Please also acknowledge the referees in the acknowledgments.

Best regards,
Stef Lhermitte, 22 May 2019

Minor Comments:
Page 8 Line 17: The authors explained in the response letter to reviewers that linear stretching is applied to the pixel values to increase the contrast. I think this needs to be explained in the main text as well, as this could be confused with a linear distortion to the dimensions of the image itself, which is also a common augmentation technique when training neural networks (to reduce sensitivity to specific geometric configurations).

Page 15 Lines 11-13: Convolutional neural networks have the intrinsic property of being applicable to consecutive small batches of an image instead of the whole image (due to the nature of convolution). Therefore, it is not true to say the architecture of previous studies need to be changed in order to be applicable to small batches rather than the whole image. The only difference is in post-processing.

Technical/Language Comments:
Page 8 Line 27:change “takes longer time” to “takes more time”
Page 9 Line 2: change “increase with patience of five epochs” to “increase for five consecutive epochs” (is it consecutive?)

Dear authors,

I want to thank you for the implementation of the minor changes and I think the paper can be published in its current form. Congratulations!

Best regards,
Stef