Reverse-Time Migration Imaging of Ground-Penetrating Radar in NDT of Reinforced Concrete Structures

Round 1

Reviewer 1 Report

Non-destructive techniques for identifying the diameter and position of reinforcing bars within concrete structures are of particular interest in structural engineering. 
The paper reports the results of an experimental campaign conducted to verify the reliability and accuracy of GPR in conjunction with the RTM algorithm to derive the required data.
In my opinion, the experimentation was conducted on a too limited number of samples, therefore not sufficient to test the reliability of the method.

The accuracy of the results also raises some doubts.
1. the real measure of the concrete cover in the cut column is not reported and in any case there seems to be an overestimation;
2. for the diameter and position of the bars an error of 2 mm is estimated, which for 22 mm bars is equal to over 9%.

With errors of this magnitude it becomes more practical to use other instruments (e.g. pacometer) that are faster and less expensive.

The determination of the position and diameter of the bars was carried out on a column, in an area where there are no transverse bars that could generate significant disturbances and difficulties.
I think it is essential that results are presented on a larger number of samples and with different reinforcement distributions, possibly even on different layers. 

Some minor errors in the text:
page 2, line 85 (1)-(3) instead of (2)-(4)
page 4, figure 1 - Back instead of Bcak

I believe that in order to be published the experimentation must be completed with more tests to be performed on different structural elements and with different reinforcement distributions. 

Author Response

Comment 1: The accuracy of the results also raises some doubts.

1. the real measure of the concrete cover in the cut column is not reported and in any case there seems to be an overestimation;

Response: Thanks for your comments. Sorry to tell you a lack of the photo of the real measure of the concrete cover in the cut column. The real-life thickness of the concrete cover in the cut column is approximately 10cm. In the imaging result of the cut column, the reflections are suppressed well. In the imaging results of four sides of the other column, there exists an overestimation.

Comment 2: 2. for the diameter and position of the bars an error of 2 mm is estimated, which for 22 mm bars is equal to over 9%. 

With errors of this magnitude it becomes more practical to use other instruments (e.g. pacometer) that are faster and less expensive. 

Response: Thanks for your comments. In the imaging result of the cut column, the estimated relative positions are consistent with those of the real-life ones; the estimation error of diameters is unsatisfying. The hyperbolic reflections are retrieved well. So we can localize the rebars. However, in the synthetic data test, the diameters, as well as the positions, are correct. In the real case test.

Comment 3: In my opinion, the experimentation was conducted on a too limited number of samples, therefore not sufficient to test the reliability of the method. 

The determination of the position and diameter of the bars was carried out on a column, in an area where there are no transverse bars that could generate significant disturbances and difficulties.

I think it is essential that results are presented on a larger number of samples and with different reinforcement distributions, possibly even on different layers. 

I believe that in order to be published the experimentation must be completed with more tests to be performed on different structural elements and with different reinforcement distributions.

Response: Thanks for your comments. The testing site is a new camp where the specifications of all columns are same. So we select one column to process the dataset instead of all columns. We think it is representative.

Comment 4: Some minor errors in the text: page 2, line 85 (1)-(3) instead of (2)-(4); page 4, figure 1 - Back instead of Bcak .

Response: Thanks for your comments. Sorry to make such a mistake. All the mistakes have been corrected.

Reviewer 2 Report

The NDT application of reverse-time migration of Ground-penetrating radar

by

Ruiqing Shen, Yonghui Zhao, Shufan Hu, Bo Li and Wenda Bi

The paper deals with the evaluation and inspection of steel bars in reinforced concrete structures. In particular, the paper reports experimental data obtained by Ground-penetrating radar (GPR) on a test-site area inside a camp on Changxing Island, Shanghai, China. The work is good and I think that it deserve merits to be published in the present form. I would only suggest: to change the text removing the words ‘authors in Ref.’, which are repeated many times in the Introduction section; to delete the following sentence: ‘In section 2, the principle of the RTM method and its workflow are introduced. Then, we present the synthetic data test and the real data test in section 3. The results are discussed in section 3 and section 4. Finally, we draw some conclusions in section 5.‘

Author Response

Thanks for your comments. we have removed the words ' authors in Ref '., and deleted the sentence in section 2 according to your suggestions.

Reviewer 3 Report

The manuscript investigates the use of reverse-time migration in Ground Penetrating Radar (GPR) investigations of rebars in concrete. The subject is appropriate for publication in Remote Sensing. The Authors have provided a comprehensive theoretical background, supported by mathematics and by experimental work on simulated and real data.

However, there are some improvements that could be made, so as to make the study clearer and easier to understand. In view of this, I recommend this manuscript be accepted with minor revisions.

The structure of the paper could be improved by reorganising the sections. In more detail, the synthetic and real case studies are presented in the ‘Result’ section, which is rather confusing. The Authors should consider introducing the datasets in the previous section. Therefore, I suggest that the ‘Materials and Methods’ section should include the following subsections:

• Theoretical Background, which should include the current ‘Reverse-Time Migration’ section
• Synthetic Dataset
• Real Case Dataset

Similarly, the ‘Results’ section could have subsections with matching titles (i.e. Synthetic Dataset and Real Case Dataset), where the outcomes of the study are displayed for each dataset.

Author Response

Thanks for your comments. The structure of the paper has been reorganized. The synthetic dataset and the real case dataset are moved to section 2. The section 2 includes 3 subsection : 2.1 Reverse-time migration; 2.2 Synthetic dataset; 2.3 Real case dataset.

Reviewer 4 Report

GENERAL COMMENTS

This paper deals with the application of GPR reverse time migration for the detection of rebars in reinforced concrete structures. The idea is interesting and focuses on a demanding area of application in civil engineering. However, the paper requires major amendments to be considered as suitable for publication in Remote Sensing. First, the application area (i.e., reinforced concrete structures) could be mentioned in the title, which is currently quite general.

Presentation aspects also need improvement. To elaborate, quality of images is rather poor and should be improved. A few of them could be also reorganised as some results seem to be repetitive. Good portions of text could be also summarised with Tables, as it is rather difficult to find the most relevant aspects emerged from the analysis of the GPR images if text is only provided.

Organisation of the paper contents should be also improved. In more detail, aims and objectives could be clearly disclosed and presented in a dedicated section. Also, current contents in section “Discussion” seem redundant and require to be summarised.

In view of the above, my recommendation for this paper is “major revisions”.

Best regards.

SPECIFIC COMMENTS

• Introduction is quite short, considering that a good number of studies have been dedicated to the GPR inspection of reinforced concrete structures and materials. Authors may want to consider expanding this section more comprehensively. Also, a few suggestions for integration are as follows:

https://doi.org/10.1007/s10712-019-09565-5

https://doi.org/10.3390/rs12172778

https://doi.org/10.1016/S0950-0618(03)00093-X

• Please check correct citation style across the manuscript (e.g., In Ref. [19], Hossain Zadhoush et al. should be referred to the main author only, without the name; Iraklis Giannakis et al. [18] should be Giannakis et al. [18] etc.)
• Figure 1: please check the spelling of the “Back propagation” step in the flowchart
• Lines 170-171: “Moreover, diameters of the whole circles are equal.”: please explain what it is implied with this comment. Are the authors implying that information on the rebar diameter can be inferred in these specific numerical examples? If this is the case, then a theoretical explanation is needed.
• Lines 171-172. “These results indicate that the RTM algorithm has a good ability to possess the noisy data.”: please check the correctness of the term “possess”.
• Line 179: “GPR measurements were performed”.
• Figure 8 is missing. Please provide it according to the contents in the text.
• Lines 193-194: “[..] and the distance between two adjacent traces is 0.0025m.”: please cross-check the correctness of this statement (i.e., meaning that the horizontal step is 2.5mm).
• Line 197: “We can also see a weak hyperbola at the rightmost position near 0.9 m.”: please indicate this feature in the figure as well, if relevant.
• Lines 195 – 208: the amount and organisation of details provided could be improved. I suggest the authors to include a Table listing the main features for the identified targets. The table could be then commented on the most important features. This will help to improve the clarity of the description.
• Lines 219-221: “The dug column in Fig. 10 (a) demonstrates that the position and the size of rebars in the imaging result is well retrieved.”: I am struggling to see the matching claimed in this statement. Perhaps, using markers/indicators in the real-life picture could help to highlight the evidence mentioned by the authors.
• Lines 221-222: “Notably, as shown in Fig. 10 (b), the right hyperbola is generated by a steel pipe.”: please indicate in the picture where is the steel pipe located. Also, this target should be duly referenced with respect to the whole inspected area – to prove a matching between the outcome in the processed radargram and the ground-truth scenario.
• Lines 231-234: “The front and back profiles contain five hyperbolas. Similarly, the left and right profiles include six hyperbolas. Notice that this column was tested in counter-clockwise order. Moreover, the front and left sides were detected from left to right instead of that in the back and right sides.”: please clarify this statement. First, why the authors specify that counter-clockwise inspections have been carried out? Does this imply that data are to be interpreted differently from the dataset of the previously-discussed column? If this layout affects the interpretation of results (i.e., it makes data analysis as a difficult task) it is suggested to invert the dataset and make it coherent with the other acquisitions. Secondly, it is not clear how – if the column has been scanned counter-clockwise – scans can be collected from left to right in certain prospects (i.e., front and left sides) and from right to left along the other two sides (i.e., back and right sides). Please clarify.
• Lines-241-284: please see the comment made for the text comprised between lines 195-208. Although many details are provided, these are purely descriptive and use of a Table with main features is strongly recommended. Supporting comments could be then provided on the main features to highlight.
• Lines 288-341: results from the application of the RTM algorithm should be reorganised. Similarly to the descriptive features in previous paragraphs, tables must be used to summarised main results in terms of target ID, their position, longitudinal/transversal axis length etc.
• Lines 333-334: “The imaging results in Fig. 15 and Fig. 16 show more accurate estimations of diameters of rebars.”: where and how the authors have quantified the diameters of the rebars? This needs a more comprehensive discussion.
• Line 334: “The maximum error of diameters is several millimeters [..]”: please be more quantitative in this statement.
• 17: please consider redrawing the sketch of the rebars position and scanning directions. Lines seems to be crooked. Also, what the internal black lines represent?
• Section “Discussion”: this session contains a lot of repetitions from section 3.2. With the main features in section 3.2 being summarised in Tables (i.e., see my previous comments), the authors could summarise contents – avoiding repetitions – and include what is really necessary for discussion of main results.
• Conclusions should be re-elaborated to point out if the set objectives have been achieved.

Author Response

Comment 1: However, the paper requires major amendments to be considered as suitable for publication in Remote Sensing. First, the application area (i.e., reinforced concrete structures) could be mentioned in the title, which is currently quite general.  

Response: Thanks for your comments. The titile has been changed into “ Reverse-time migration imaging of Ground-penetrating radar in NDT of reinforced concrete structures”.

Comment 2: Presentation aspects also need improvement. To elaborate, quality of images is rather poor and should be improved. A few of them could be also reorganised as some results seem to be repetitive. Good portions of text could be also summarised with Tables, as it is rather difficult to find the most relevant aspects emerged from the analysis of the GPR images if text is only provided.  

Response: Thanks for your comments. So sorry that the quality of images is poor. Some of them have been revised, they are 300 dpi and more. But there is a little problem that Word seems to compress the quality of images automatically. For a instance, the photo of the cut column is originally clear and all scales are clear, however, it has to become vague. In the synthetic data test, the images look like being repetitive because the noisy data has little influence on the imaging results. In addition, we changed some portions of text into table according to your suggestion.

Comment 3: Organisation of the paper contents should be also improved. In more detail, aims and objectives could be clearly disclosed and presented in a dedicated section. Also, current contents in section “Discussion” seem redundant and require to be summarised.

Response: Thanks for your comments. The structure of the paper has been revised. The contents in section “Discussion” have been simplified and summarized well with tables.

Comment 4: Introduction is quite short, considering that a good number of studies have been dedicated to the GPR inspection of reinforced concrete structures and materials. Authors may want to consider expanding this section more comprehensively. Also, a few suggestions for integration are as follows:

https://doi.org/10.1007/s10712-019-09565-5 

https://doi.org/10.3390/rs12172778 

https://doi.org/10.1016/S0950-0618(03)00093-X 

Response: Thanks for your comments. The introduction has been revised. These papers are so useful. Again, thanks for your comments.

Comment 5: Please check correct citation style across the manuscript (e.g., In Ref. [19], Hossain Zadhoush et al. should be referred to the main author only, without the name; Iraklis Giannakis et al. [18] should be Giannakis et al. [18] etc.)

Response: Thanks for your comments. The citation format has been revised.

Comment 6: Figure 1: please check the spelling of the “Back propagation” step in the flowchart

Response: Thanks for your comments. The little mistake has been corrected duly.

Comment 7: Lines 170-171: “Moreover, diameters of the whole circles are equal.”: please explain what it is implied with this comment. Are the authors implying that information on the rebar diameter can be inferred in these specific numerical examples? If this is the case, then a theoretical explanation is needed.

Response: Thanks for your comments. Sorry to mislead you. The author has wanted to express that all the circles are same. In the revised edition, the circles in the migration images of the synthetic datasets have been moved.

Comment 8: Lines 171-172. “These results indicate that the RTM algorithm has a good ability to possess the noisy data.”: please check the correctness of the term “possess”

Response: Thanks for your comments. The term “possess” has been corrected into “against”.

Comment 9: Line 179: “GPR measurements were performed”.

Figure 8 is missing. Please provide it according to the contents in the text.

Response: Thanks for your comments. The figure 8 was originally the B-scan profile of the cut side. Maybe it is due to the Word edition of 2016. Sometimes there exists some problems during the conversion of the edition. Whatever is fine, the figure has been provided.

Comment 10: Lines 193-194: “[..] and the distance between two adjacent traces is 0.0025m.”: please cross-check the correctness of this statement (i.e., meaning that the horizontal step is 2.5mm).

Response: Thanks for your comments. The statement has been corrected duly. Again, thanks for your comments.

Comment 11: Line 197: “We can also see a weak hyperbola at the rightmost position near 0.9 m.”: please indicate this feature in the figure as well, if relevant.

Response: Thanks for your comments. According to your suggestion, the figure has been revised and the hyperbola has been marked by a black-dotted box.

Comment 12: Lines 195 – 208: the amount and organisation of details provided could be improved. I suggest the authors to include a Table listing the main features for the identified targets. The table could be then commented on the most important features. This will help to improve the clarity of the description.

Response: Thanks for your comments. The table has been used according to your suggestion. The text has been re-organized.

Comment 13: Lines 219-221: “The dug column in Fig. 10 (a) demonstrates that the position and the size of rebars in the imaging result is well retrieved.”: I am struggling to see the matching claimed in this statement. Perhaps, using markers/indicators in the real-life picture could help to highlight the evidence mentioned by the authors.

Response: Thanks for your comments. The indicators have been used in the real-life picture. Nevertheless, sorry to the incorrectness of the statement. The ‘size’ should be moved.

Comment 14: Lines 221-222: “Notably, as shown in Fig. 10 (b), the right hyperbola is generated by a steel pipe.”: please indicate in the picture where is the steel pipe located. Also, this target should be duly referenced with respect to the whole inspected area – to prove a matching between the outcome in the processed radargram and the ground-truth scenario.

Response: Thanks for your comments. The steel pipe has been indicated in the real-life picture. However, the real-life picture of the cut column is vague due to Word’s function of compressing automatically (mentioned at the previous response of comment 2 ). The relative positions are consistent with the real-life ones.

Comment 15: Lines 231-234: “The front and back profiles contain five hyperbolas. Similarly, the left and right profiles include six hyperbolas. Notice that this column was tested in counter-clockwise order. Moreover, the front and left sides were detected from left to right instead of that in the back and right sides.”: please clarify this statement. First, why the authors specify that counter-clockwise inspections have been carried out? Does this imply that data are to be interpreted differently from the dataset of the previously-discussed column? If this layout affects the interpretation of results (i.e., it makes data analysis as a difficult task) it is suggested to invert the dataset and make it coherent with the other acquisitions. Secondly, it is not clear how – if the column has been scanned counter-clockwise – scans can be collected from left to right in certain prospects (i.e., front and left sides) and from right to left along the other two sides (i.e., back and right sides). Please clarify.

Response: Thanks for your comments. So sorry to mislead you. First, this statement has no additional and complex meaning. The author has wanted to emphasize that the order is consistent with the sequence of the four imaging results. Actually, all columns was tested in counter-clockwise order. Second, different side has different testing direction. As you have known, the column was tested in counter-clockwise order. When testing the front side and the left side, we conducted the inspection from left to right. Similarly, when testing the back side and the left side, we changed the testing direction into inversion.

Comment 16: Lines-241-284: please see the comment made for the text comprised between lines 195-208. Although many details are provided, these are purely descriptive and use of a Table with main features is strongly recommended. Supporting comments could be then provided on the main features to highlight.

Response: Thanks for your comments. As mentioned previously, the tables have been utilized to summarize these features.

Comment 17: Lines 288-341: results from the application of the RTM algorithm should be reorganised. Similarly to the descriptive features in previous paragraphs, tables must be used to summarised main results in terms of target ID, their position, longitudinal/transversal axis length etc.

Response: Thanks for your comments. The contents have been reorganized. The tables are used to summarize main interpreted results.

Comment 18: Lines 333-334: “The imaging results in Fig. 15 and Fig. 16 show more accurate estimations of diameters of rebars.”: where and how the authors have quantified the diameters of the rebars? This needs a more comprehensive discussion.

Response: Thanks for your comments. First, there exists a mistake in the statement. The ‘Fig. 15 and Fig. 16’ should be ‘Fig. 14 and Fig. 15’. Obviously, the hyperbolic reflections collapse better in Fig. 14 and Fig. 15. Second, the diameters of rebars in the imaging results of the four sides are not quantified. So this statement should be a mistake. After the paper was revised, the diameters of the rebars in the synthetic datasets and the column-cut dataset have been quantified. We selected one area surrounding the identified rebar from the imaging result and enlarged it. For calculation of the diameter, we marked the scale. Notably, the real diameter should be obtained by conversion due to the scale not being equal to 0.2m. Disappointingly, the estimation error between the interpreted rebar and the real one reach 7% or more. However, the diffraction wave is suppressed greatly.

Comment 19: Line 334: “The maximum error of diameters is several millimeters [..]”: please be more quantitative in this statement.

Response: Thanks for your comments. The statement is an error, which has been moved in current edition. The relevant quantitative statements is mentioned in previous response (i.e. the response of comment 18).

Comment 20: 17: please consider redrawing the sketch of the rebars position and scanning directions. Lines seems to be crooked. Also, what the internal black lines represent?

Response: Thanks for your comments. The sketch of the interpreted concrete reinforced structure has been redrew. In addition, the internal black lines are moved because it is just a mistake.

Comment 21: Section “Discussion”: this session contains a lot of repetitions from section 3.2. With the main features in section 3.2 being summarised in Tables (i.e., see my previous comments), the authors could summarise contents – avoiding repetitions – and include what is really necessary for discussion of main results.

Response: Thanks for your comments. The contents of Section “Discussion” have been reorganized and summarized.

Comment 22: Conclusions should be re-elaborated to point out if the set objectives have been achieved.

Response: Thanks for your comments. The content of Section “Conclusions” have been revised, in which the set aims are mentioned.

Round 2

Reviewer 1 Report

I  thank the authors for their responses. which I believe satisfied my requirements.

Author Response

Your recognition of our work is much appreciated.

Reviewer 4 Report

Dear Authors,

Thanks for providing the revised version of your papers. Comments previously raised have been responded. Presentation of results has now improved as well as the organisation of contents. Please provide an English spell-check throughout the manuscript. In my opinion, still minor revisions are required to reach the publication level, as reported below.

Best regards.

• Ranking system of features used in Table 1: please define what is meant by “normal”. This should be done in the text as well as in the Table caption. DO the authors mean “clearly visible”?
• 3/4/6/8/10/13/14/15/16: please indicate units in the colour maps of the radargrams.
• There is still unclarity on how the experimental scenario has been presented. In my view, columns should be given an ID code (for example, Column A and Column B). This will work better than defining them as the “cut column” and the “other column”. Obviously, the cut column is an important feature and it should remain attached to the reference column with the suggested ID code.
• Line 240: “Moreover, the width of the box is equal to 0.2m.”: this information does not add anything to the discussion. I suggest to remove it.
• Figure 9: please provide a short description of the main figure and the detail window explaining what these represent.
• Figure 11: please indicate the ID of the rebar that the green circle is referring to (right-hand side of the figure) with respect to the overall diagram on the left-hand side of the figure.
• 12(b): please better highlight the notation “steel pipe” in the figure. Perhaps sue bold style and larger fonts could improve the readability of this figure.
• 13, 14, 15 and 16 could be reorganised and displayed similarly to Fig. 8. This will help to make the presentation results more homogeneous and will make easier the comprehension of results.
• 17: please include a scale for the x and y axis. This will help to appreciate better the change in thickness of the 10 cm cover.

Author Response

Many thanks for your valuable comments, suggestions, and corrections, which have greatly improved the manuscript.

Comment 1: Ranking system of features used in Table 1: please define what is meant by “normal”. This should be done in the text as well as in the Table caption. DO the authors mean “clearly visible”?

Response: Thanks for your comments. The “normal” shows that the hyperbola is typical and complete. The “clearly visible” is also one of the meanings. The author has defined the meaning of the “normal” in the text as well as in the Table caption.

Comment 2: 3/4/6/8/10/13/14/15/16: please indicate units in the colour maps of the radargrams.

Response: Thanks for your comments. Sorry to ignore the unit. The color maps of the radargrams indicate the colorbar of the electric field intensity E of which the unit is V/m. The paper has been revised duly according to your suggestion.

Comment 3: There is still unclarity on how the experimental scenario has been presented. In my view, columns should be given an ID code (for example, Column A and Column B). This will work better than defining them as the “cut column” and the “other column”. Obviously, the cut column is an important feature and it should remain attached to the reference column with the suggested ID code.

Response: Thanks for your comments. According to your suggestion, the author has corrected the “cut column” and the “other column” into “column A” and “column B”. For emphasizing that column A was cut, the paper uses “cut column A” somewhere.

Comment 4: Line 240: “Moreover, the width of the box is equal to 0.2m.”: this information does not add anything to the discussion. I suggest to remove it.

Response: Thanks for your comments. The sentence has been removed.

Comment 5: Figure 9: please provide a short description of the main figure and the detail window explaining what these represent.

Response: Thanks for your comments. The author has revised the paper according to your suggestion.

Comment 6: Figure 11: please indicate the ID of the rebar that the green circle is referring to (right-hand side of the figure) with respect to the overall diagram on the left-hand side of the figure.

Response: Thanks for your comments. The author marked the rebar 4 inside the black-dotted box. The detail window of figure 11 has been modified.

Comment 7: 12(b): please better highlight the notation “steel pipe” in the figure. Perhaps sue bold style and larger fonts could improve the readability of this figure.

Response: Thanks for your comments. According to your suggestion, the figure has been modified. The notation “steel pipe” is enlarged, and the author marked the steel pipe with a black-dotted box.

Comment 8: 13, 14, 15 and 16 could be reorganised and displayed similarly to Fig. 8. This will help to make the presentation results more homogeneous and will make easier the comprehension of results.

Response: Thanks for your comments. The four figures have been reorganized into Figure. 13 (a) – (d).

Comment 9: 17: please include a scale for the x and y axis. This will help to appreciate better the change in thickness of the 10 cm cover

Response: Thanks for your comments. The fig. 14, i.e. the fig. 17 originally, has been modified. The author drew a scale for the x and y axis, respectively. The scale of the x axis is from 0 to 60cm, similarly, the scale of the y axis is from 0 to 80cm. Moreover, the blue circles are changed into the black circles.