Impact-Detection Algorithm That Uses Point Clouds as Topographic Inputs for 3D Rockfall Simulations

Round 1

Reviewer 1 Report

This study presented an impact-detection algorithm for 3D rockfall simulations. The innovation of the presented method is based on the use of a point cloud as topographic input. These data allowed to increase the detail of the terrain topography on which the simulations were carried out. The point cloud was also used to measure the terrain surface roughness. Accurate measurement of surface roughness increases the realism of the simulation.

The article thoroughly analyzes the influence of different surface roughness, slope inclination, rock size and tree stems interaction in rockfall simulations process. The results of the analyzes were compared with the results of the existing simulation models.

The research topic is very interesting, and the advance in this field could have significant impact on rockfall simulations. However, there are some clarifications needed:

1. In Figure 1. was shown secondary (computed) TIN DTM from a gridded DTM. Please explain, is there a possible to use primary (measured) DTM with higher resolution in existing simulation models?

2. In Figure 2. the 3D point cloud points are presented as evenly distributed but laser scanning data are not evenly distributed because of different scanning density, field conditions, surface properties of the scanned object. This point must be clarify in the paper.

3. There is no information about point density and accuracy of points in 3D point cloud. Please complete these data.

Author Response

Dear reviewer,

Thank you for spending the time for going through our manuscript and giving us constructive remarks. We assessed them following the same numbering:

• It is possible if the rock is not considered as a simple point (lumped mass). Otherwise, the perceived terrain orientation will not be differentiated from the local terrain orientation, which can introduce excessive deviations. We added one more reference showing this problem (lines 33-39 and figure 5) and discuss more about it later in introduction (line 99 to 108 in the attached PDF with revision marks).
• We nuanced the legend of the figure stipulating that they “can” be evenly distributed. Their characteristics are better explained in section 1.1.
• The section 1.1 has been moved in introduction. It explains characteristics of point clouds, with their typical density depending on the acquisition methods. It also covers how to calculate surface orientation depending on the local noise in the case of inaccurate/unprecise point clouds. The missing average point spacings of the first application case have been added. They are now specified for all point clouds used in the two application cases and validations.

The numerous little changes and clarifications on the introduction, methods and results can be reviewed in detail from the attached PDF document (automatic comments “a mis en forme” in the margin about updating styles and/or languages should be ignored). We tracked the edited changes to ease the review process.

Also, the English language of the manuscript has been edited by the American Journal Experts (AJE).

Thanks again for your time.

Sincerely,

François Noël

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper presents a novel impact detection algorithm for 3D rockfall simulations. The algorithm is described in details and its application is compared to simulations conducted using well-known rockfall simulation software.

The paper is generally well written; it presents and interesting approach that could be of significance for the scientific community.

I support its for publication provided that the authors address the following minor revisions:

• page 4 lines 76-79: The significance of the effect of the incident angle on the energy change upon impact has been highlighted by several researchers both in lab and field works: I recommend including in the references the work published by Wyllie 2014, Spadari et al. 2012, Buzzi et al. 2011
• pag 5 line 91: reference for CRSP-3D should be included
• I suggest using “rock” or “impacting rock” or “falling rock” instead of “rock projectile” within all manuscript.
• I could not access reference [26] (pag 6 line 120) and identify how much material and discussion about the new algorithm has already been included (and published) in the previous paper. If there are several overlaps (paragraphs and sections) between the two manuscripts, I recommend not duplicating the content and potentially considering reducing some sections of the current manuscript (recalling the most important aspects of the new algorithm only). This paper is quite long already, and it would remain a substantial amount of work even just focusing on the application and validation of the proposed algorithm and comparison with other software.
• Page 9 line 223: what is an “adequate” parameter?

• Section 2.1 should be moved earlier in the structure of the paper as it includes a short review of remote sensing techniques currently in use to acquire images of rock slopes.
• Sections 3.1.3 pag 16 line 394: suggest rewriting the sentence avoiding using “the rock projectile do not fly away from the terrain”.
• The reference x,y,z, plane is missing in figures 16 and 20.
• Final considerations on the application to case La Verda recall similar observations on the influence of roughness and slope angle presented in Ferrari et al.2016 (Georisk). I suggest considering this published work for reference and comparison.
• Pag 27 lines 674-676: did the authors converted the 3D point cloud with equal 0.12m spacing in both directions? Why did the authors space the 2D profile differently (0.5m)?
• I do not agree with the use of term “tuning” in the second paragraph of pag 27: the approach here refers to “back-analysis”.
• Not many details are provided on the “adjusted dumping parameters” (lines 689 pag 27) and I suggest providing further clarification. Did the authors consider only 1 adjusted parameter for the full slope (see table 3)?
• Final conclusion that CRSP4 does not use the same rebound equations of reference [45] does not really provide a clear identification of the problem. Did the author verify their claim in anyway?
• Pag 31 lines 796: to which elongated conical shape do the authors refer to? It is not clear.
• Figure 25 shows some evident discrepancies in the distribution of trajectories with trees (images  25b and 25d), especially in the south-west zone of runout, however, no much  discussion is provided. 

Author Response

Dear reviewer,

Thank you for spending the time for going through our manuscript and giving us constructive remarks. We assessed them in the same order as follow:

• The relevant suggested references about the significance of the effect of the incident angle on the energy change have been added to the manuscript (lines 78-81 in the attached PDF with revision marks).
• References for CRSP has been added (line 92).
• “rock” is used instead of “rock projectile” within all manuscript.
• The referred extended abstract from the ISL2016 conference differs greatly from the proposed manuscript. Thus, no change has been done concerning this point.
• “adequate parameters” has been changed to “reasonable parameters” and detailed at lines 271-273.
• Section 2.1 has been moved in the introduction (line 136).
• The term “do not fly away” has been replaced by “are not deflected away” (line 399).
• The reference plane (shown with a red-green-blue trihedron) has been added to figures 16 and 20.
• Details concerning the influence of the perceived terrain roughness and terrain geometry were added through the manuscript with the suggested reference (mostly at lines 427-451, 482-488 and 635-642).
• The 0.12 m average point spacing is obtained by subsampling a denser randomly distributed point cloud (the subsampling is done by removing every points closer to a given distance in the 3D space around one point and repeat to the next point until they are all spaced relatively evenly). Details concerning the spacing were added at lines 721-727.
• The paragraph from lines 728-735 was improved by using the terms “back analysis” and “back calculation”.
• Details and clarifications were added concerning the adjusted damping coefficients and artificial roughness (lines 482-488 & 736-746).
• The partial conclusion concerning the rebound model has been nuanced (line 787-796). The recommendations are unchanged, but the conclusions concerning the cause of the mismatching results in between CRSP 4 and the others is left to the reader.
• The elongated conical shape refers to the shape of the stem. This has been clarified by linking to related figures (lines 855-856).
• Details and clarifications were added concerning the differences observed (paragraphs starting at line 893, 915 and 933). Line 918 and followings covered this part, but mistakenly referred to the bottom right portion of the figure instead of bottom left. This has been corrected.

The numerous little changes and clarifications on the introduction, methods and results can be reviewed in detail from the attached PDF document (automatic comments “a mis en forme” in the margin about updating styles and/or languages should be ignored). We tracked the edited changes to ease the review process.

Also, the English language of the manuscript has been edited by the American Journal Experts (AJE).

Thanks again for your time.

Sincerely,

François Noël

Author Response File: Author Response.pdf