A New Approach for Adaptive GPR Diffraction Focusing

Round 1

Reviewer 1 Report

The authors presented works that combed the autofocus strategy and multi-path summation, with which the migrated GPR data showed impressive improvement. In my view, this is of great significance in actual applications. 
However, the presentation was not good enough for publication. The manuscript reads like an oral report but not a technical paper. The figures are not clear enough. Many sections, including the abstract, the Synthetic Example, and the Discussion, need to be improved because these parts were lengthy and organized without a clear clue. 
My comment is a major revision.

Author Response

We are thankful for your constructive comments, which helped us in improving our manuscript. We have done an effort to improve the language and arrangement of our manuscript based on your and others reviewers’ comments. We have replaced all figures by higher resolution versions, and most of the figures were enlarged to make all numbers and letters clear. The revised version was marked up using the “Track Changes” function, such that any changes can be easily viewed by the editors and reviewers.

Reviewer 2 Report

Dear authors,

I enjoyed reading your manuscript about a very interesting topic. The writing is easy to follow and the results are astonishing.

Nevertheless, I have some remarks that need more explanation. I added most of them as comments in the pdf, but I would also like to list the most important ones here:

• I feel that the mathematical description is a bit lengthy, but I don't feel able to say exactly at what point. Maybe you could check again if you need all the formulas.
• The numbers in the figures need to be larger and sometimes the markers are hard to see.
• My most important remark is this: I am wondering about the comparison in the synthetic example, where the picture with using the right velocity model for migration gave worse results compared to your approach. Why is this? Is it because of the spectral whitening? Or is there maybe a mistake in the figures? And another thing maybe also related to this is the question about lateral/vertical velocity changes. I understood your approach that you use the same mixing of constant velocity models over your complete profile, i.e. at all depths and places the mixed velocity is the same. But this mixed velocity might be too high for e.g. lower parts of the figure (if there is a decrease in velocity with traveltime). Therefore, I would expect that the migration with using the right velocity model (with varying velocity!) should produce the best result over the complete profile. It would be necessary to add a discussion of these effects and also to check the correctness of the figures.

Comments for author File: Comments.pdf

Author Response

First, we would like to thank you for your good words and we are happy to hear you enjoyed reading our manuscript. We are very thankful for your insightful comments, which helped us in improving our manuscript. Here is a point-by-point response to your comments:

• All comments in the provided pdf file were addressed in the revised version of our manuscript.
• Regarding the mathematical description, we checked the mathematical description and we removed one of the formulas. We feel it would be difficult to remove more equations in our manuscript, as this might make it difficult of the readers to understand and implement our approach, especially since we provide no code.
• All figures have been replaced by higher resolution versions, and most of the figures were enlarged to make all numbers and letters clear.
• Regarding your last remark: a) we have modified out text in section “5. Discussion”, where we commented on Figure 6 to address your valuable remark. b) in relevance to the multipath migration approach efficiency in focusing the diffracted energy, as mentioned in the “Introduction” section (lines 71-73): “The resulting migrated images are then stacked to enhance the lateral continuity of reflections by superimposing the migrated diffraction hyperbola’s apexes and collapsing their tails”, hopefully this will answer your comment. We have checked the figures to make sure there were no confusion in them as well.

Reviewer 3 Report

It is an interesting investigation, which may be recommended for publication.

However, some revision is necessary before the manuscript will be accepted.

The quality of many figures is bad and sometimes cannot be analyzed. Besides this, the authors do not use colors for the GPR image analysis (why?). Figures 2, 3, 4, 8, 10, 12, and 13 (and it is possible some others) should be replaced by high-resolution versions.

In the Introduction, it is necessary to add some small review about other approaches applied to the GPR focusing.  

For instance:

Alperovich, L., Eppelbaum, L., Zheludev, V., Dumoulin, J., Soldovieri, F., Proto, M., Bavusi, M. and Loperte, A., 2013. A new combined wavelet methodology applied to GPR and ERT data in the Montagnole experiment (French Alps). Journal of Geophysics and Engineering, 10, No. 2, 025017, 1-17.

Ming-Chih Lin, Yu-Ming Kang, Kun-Fa Lee, and Hui-Chi Hsu, 2009. A Study on the Technologies for Detecting Underground Water Level and Processing Image. International Journal of Applied Science and Engineering, 7, 1, 61-68.

 

 

 

 

Author Response

Thank you for your comments and suggestions, that helped us improve our manuscript. Here is a point-by-point response to your comments:

• Your comment on the figures was very valid as there was probably some problem when transferring the figures to the word file. We have replaced all figures by higher resolution versions, and most of the figures were enlarged to make all numbers and letters clear.
• Thank you for raising this issue. We have added a small review about other approaches applied in GPR focusing in the introduction section lines 48-54.

Round 2

Reviewer 1 Report

I have no more comments.

Author Response

Thank you again for your constructive comments. We hope that our manuscript is in a much better shape now.

Reviewer 2 Report

Dear authors,

thank you very much for revising the manuscript carefully and adressing my comments.

Most of them are answered now, but unfortunately I still do not find any explanation/discussion regarding the spatial velocity variation and how your approach deals with it. Maybe it is hidden somewhere and I didn't find it, but even then I would suggest to discuss it again clearly in the discussion section. Until now, I cannot understand how (or even if) your approach handles spatial velocity variations. Even if it does not handle them explicitly, you could comment on this in the discussion. Nevertheless, the results are convincing!

Another small comment: Some of the figures seem to be misplaced or even on top of the text, but this will probably be fixed during layouting.

Author Response

Thank you for raising this issue, and reminding us to emphasize one of main advantages of the multi-path summation approach. The key word for the efficiency of the multi-path summation is stacking. By stacking the different migrated sections using different velocities, the apices of the hyperbolas will be enhanced against the moving tails of the over or under migrated parts of the constant velocity migrated sections. Using different velocities for migration, the apices will always be clear and in the same location in migrated sections even if an optimum velocity is not used. The tails of the diffractions hyperbolas on the other hand, will change place in every different velocity migrated sections. Therefore, after the stacking of the different migration sections, using the same range of constant velocity migration, this approach will focus most of diffracted energy despite spatial velocity variation. However, multipath summation images may suffer incomplete cancellation of the hyperbola tails due to not equally focused diffractions. This is because the output of this methodology corresponds to an optimum combination of the existing migration velocities within the data, which produces an average focusing of all diffractions.  This is clear in Figures 8 and 11 where some of the diffractions are not efficiently focused. Therefore, the main advantage of this approach is that it manages to focus most of the diffracted energy without requiring the knowledge or selection of a velocity model. We have added a paragraph in the discussion section lines 823-837 related to clarify this issue.

You are correct as many of the figures are misplaced when opening the world file. This is probably because of the different versions of the office word used. We have attached  a pdf file as well to overcome this matter, which I am sure will be fixed during the proofing of the text.

Author Response File: Author Response.pdf