An Improved Point Clouds Model for Displacement Assessment of Slope Surface by Combining TLS and UAV Photogrammetry

Round 1

Reviewer 1 Report

The paper addresses the very important topic of slope deformation detection in civil engineering. The innovative part of the paper is the proposal of an approach to integrating airborne photography with terrestrial laser scanning for displacement assessment. In this regard, the paper is interesting and has potential to contribute to the research community. However, the reviewer also found several issues need to be clarified before publication in the journal. A detailed list of comments is as follows:

• Line 34 to Line 36, the sentence is grammatically incorrect, with two sentences fused into one without the use of conjunction.
• Line 38, what does “these” refer to?
• The authors are suggested to provide a more comprehensive review of the latest literature. There have been many related research efforts in detecting slope damages from aerial photographs, registering aerial image with 3D models for structure condition assessment, and developing algorithm that integrates PCA and ICP to register partial photogrammetric point cloud to a larger cloud for indoor localization purpose.
• Line 65 to 76, The authors seem to confuse airborne laser scanning with UAV structure from motion. Although they both generate point cloud, the technical rationale is totally different. What the authors review here is the latter, not ALS.
• Line 108 to 110, the paragraph is grammatically incorrect. In addition, why UAV can generate right amount of point clouds? Please justify.
• Line 215 to 217, are the digital cameras used here the one mounted on UAV?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

It is a paper with a topical subject, clearly presented, theoretically supported and argued by the practical results presented.
Congratulations to the authors.

Author Response

Thank you for your congratulation

Reviewer 3 Report

line 44: What does the phrase "high risk operation" mean? Do you have issues with accuracy or safety at work?
line 147: A theoretical background on PMVS is required.
line 149: unified to where? Global coordinate system or local coordinate system?
line 257: ICP may not work properly in areas with few feature points, such as a long slope to the left and right. It may work well in this specific area, but it may not work well in others. Please provide your comments on this issue.
Why did you opt for ICP in 2.3.2? Is there any reason not to use reference points?
Is the number 0.707 from sqrt(2)/2 on line 319?
Examine the error and characteristics of each point in Figure 16 to see if there is a pattern.

Please read the following papers as well, and refer to them if you find any helpful information.
Brideau, M. A., Sturzenegger, M., Stead, D., Jaboyedoff, M., Lawrence, M., Roberts, N. J., ... & Clague, J. J. (2012). Stability analysis of the 2007 Chehalis lake landslide based on long-range terrestrial photogrammetry and airborne LiDAR data. Landslides, 9(1), 75-91.

Collins, B. D., & Sitar, N. (2004, May). Application of high resolution 3D laser scanning to slope stability studies. In 39th Annual Symposium on Engineering Geology and Geotechnical Engineering, Butte, Mont.

Kemeny, J., & Turner, K. (2008). Ground-based lidar: rock slope mapping and assessment (No. FHWA-CFL/TD-08-006). United States. Federal Highway Administration. Central Federal Lands Highway Division.

Khosravipour, A., Skidmore, A. K., Wang, T., Isenburg, M., & Khoshelham, K. (2015). Effect of slope on treetop detection using a LiDAR Canopy Height Model. ISPRS journal of photogrammetry and remote sensing, 104, 44-52.

Kim, M. K., Kim, S., Sohn, H. G., Kim, N., & Park, J. S. (2017). A new recursive filtering method of terrestrial laser scanning data to preserve ground surface information in steep-slope areas. ISPRS International Journal of Geo-Information, 6(11), 359.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

The authors use Terrestrial laser scanning (TLS) and UAV photogrammetry to compute displacement of slope surface for fixed terrain points. However, the applied methodology and the measurement routines are not clear enough. The authors give a brief overview of the bibliography, but they do not define the structures for processing and how they are represented and processed. Although we have a general diagram of processing flow, the following subsections do not precise the methods and their implementations. Some of the methods are only mentioned (eg. PCA, ICP registration, color distinguishing), some are enormously extended (eg. coordinate conversion), some of them are missed (eg. selection of monitoring points, data acquisition, the usage of registration procedures, photogrammetry reconstruction, scaling the photogrammetry). There is no description of experimental measurements including the parameters of measurement units and quantities of obtained data. The interpretation of the comparison between different methods is not convincing, especially, regarding the practical cases. The topic itself is interesting but I suggest rewriting and restructuring the whole text, and also performing additional experiments.

Additional remarks:
- Include the dictionary of abbreviations.
- l.65 - show the samples of terrain data showing the similarities and differences.
- Section 2 - the description is too general, describe the data structures, their specific features, and acquisition procedures; change the title of the section - it is not relevant.
- Section 2.1 - list the steps of UAV reconstruction, describe them and their parameters.
- l.139-141 - the text suggests that it is your work but the results are cited from outer publication.
- Fig. 2 - describe the process of UAV generation in text.
- Fig. 3 - support the figures with quantitive values, show the scale and interpret markers.
- Sect. 2.1.2 - this description looks like from another fairy tale, adapt it to your research and show correspondences to your routines.
- Sect. 2.2.1 - Be concrete and precise, rely on parameters and abilities of your equipment, change the whole part.
- l.216 - point methods that are similar.
- Sect. 2.3 - give a concrete definition of the registration procedure. Show the registration routines and parameters. Describe the implementation or software tool.
-
- l.242 - what is x,y,z, explain how cov is used
- l.249 - define the overlap rate.
- Fig. 8 - no description in the text, describe subimages.
- Fig. 9 - The ICP description is too general, describe the specification of your application. There are usually a lot of improvements in this algorithm as in general form it is not stable - do you have such experience, can you describe or cite your approach.
- Fig. 10 - Describe subimages
- l.272-274 - point the paper or technical report including the details, you can also enclose sources for analysis.
- Table 1 - Can you formulate the ranges for color selection
- Section 2.5 - Describe the way of obtaining the corresponding monitoring points.
- Fig. 14 - Explain the difference between datum points and monitoring points, how they are distinguished. The figure is too general: show and describe details.
- Section 3 - Give more description of the equipment for experiments and their arrangement.
- Fig. 15 - how is it possible that values that are proportionally growing in one image have contradictive tendencies in the 2nd and 3rd drawing?
- l.399-400 - define these errors, how they were calculated.
- Fig. 16 - how to interpret the scores in this fig???
- l.470 - the proposed method of registration is not clearly defined.
- l.473 - the term 'improved to a certain part' is not precise as for results

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 5 Report

In this study, the method to improve terrestrial laser scanning point clouds by UAV photogrammetric point clouds on highway high slope surface displacement was used.

Table 2 shows that using TLS+UAV increases Error-mean and RMSE, compared to the case where only TLS is used. This shows that using UAV can increase the system error. Can you explain this error increase? 

You mention that the data acquired by TLS often omits some parts of slope surface because the instrument perspective. In this case, you can simply change the location of TLS, so that all slope surface can be covered. In other words, integrating the data of multiple TLSs (which are located at multiple places) can be used to cover the entire space. Why did you use a drone image, which can increase the system error?

In conclusion, you mention that ". If the altitude exceeds 60m, the measurement error will increase sharply." Can you explain the reason for this error increase in detail? How can you control the drone altitude so that it does not exceed 60 m?

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

You've been through a lot. Lastly, please check grammar and typo.

Author Response

Thanks to your suggestion, we have modified this paper according to your seggestion.

Reviewer 4 Report

- There is no dictionary of abbreviations, eg. I cannot see the explanation for GPS RTK (l.40, 42 ...) and GPDS RTK (l.23).
- In Review 1 comment 2 I meant the visual similarities and differences of the considered measurement types.
- l. 388: I still lack the description and discussion of the software implementation; maybe pointing to the resources would be a good idea.
- There are minor language mistakes in the text; spell check is required.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 5 Report

I am satisfied with revision. Thank you.

Author Response

Thanks to your suggestion, we have modified this paper according to your seggestion.