Using Deep Learning to Model Elevation Differences between Radar and Laser Altimetry

Round 1

Reviewer 1 Report

This article describes the returning height differences from radar and laser altimeter over the surface of Greenland. Despite of high interest on the subject, the article seems to fail to present any significance of their results. 

First of all, the structure is particularly poor making it extremely difficult to comprehend what the authors tried to achieve. Major parts from the Data section must be transferred to the Method section, ie., L122-L124, L155. and the statistical assumption as well as machine learning models must be combined to a single section of Method. Besides, section 2.3 itself is not a simple task. If the authors wanted to make a difference map of radar and lidar without any consideration, just refer to other articles and use their data instead. Moving on, a table of ALL parameters that were used in the trainning is required. It is unclear which parameter and/or datasets were used. 

Lastly, any information on land/ice discrimination? was the analysis done on the glacier surface only? Any difference between the radar and lidar on land surface? were they validated? 

Again, this topic could be very interesting to many others if the article was appropriately designed and proceeded. 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

From what is understood, the study combines elevation data from ICESat-2 LiDAR data and CryoSat-2’s interferometric radar altimeter data using a deep neural network to model elevation differences between them. However, there is no/minimal novelty in the proposed idea of just using a deep NN to the model elevation differences. The authors must more clearly explain the context of the research and the novelty of the proposed method. 

Page 2, Line 92: “These 90 base datasets are augmented by topographical quantities derived from the ArticDEM 91 digital surface model”. Please briefly recollect how the augmentation is achieved using topographical quantities.

Page 4, Line 178 Please replace “The data was” with “The data were”. Please check throughout the manuscript (e.g., Line 188).

Page 5, Line 194 “A multi-variate, ordinary least squares regression model (OLS) was implemented to serve as a benchmark for comparisons to the machine learning models.” Which machine learning models? Also, OLS is not a suitable comparison for deep models.

 

Page 5, Line 196: “A flexible N-layer 195 neural network was then created using Pytorch” What is the architecture used? If the number of inputs, number of hidden layers, and number of nodes are changed, then the model itself changes. Please clarify!

Author Response

Please see the attachment 

Author Response File: Author Response.docx

Reviewer 3 Report

This is a great start to introduce the machine learning techniques to determine the elevation differences between radar and laser altimetry.  The slight problem is the paragraphs are too long in some parts but the figures are quite well presented.

Author Response

Thank you for taking the time to review this manuscript. Minor typos and paragraph rewordings have taken place in the latest revision.

Reviewer 4 Report

Globally, the manuscript is very well written and organized. In my opinion, the topic under study is relevant and deserves to be published.

There are only some minor English/typing “bugs” that should be corrected, before the final publication of the manuscript. I recommend the authors to carefully re-ready the complete manuscript to correct these “bugs”; please also refer to the attached document, where some of them are highlighted.

Comments for author File: Comments.pdf

Author Response

Thank you for the pdf with the list of highlighted focus areas. We have addressed all of the obvious typos. Some highlighted issues did not have obvious corrections (L136, L217, L222, L529 in the original manuscript) so the paragraphs were reworded which hopefully addresses the original concerns.

Round 2

Reviewer 2 Report

The authors have provided acceptable response to the reviewer comments. The paper is in good shape.

Author Response

Thank you for reviewing. It seems this author is happy to approve the current revision.