Delineating and Reconstructing 3D Forest Fuel Components and Volumes with Terrestrial Laser Scanning

Round 1

Reviewer 1 Report

General comments

This well written manuscript presents a methodology to assign points in TLS point clouds to various classes corresponding to ground and canopy fuel elements to facilitate the computation of fuel volumes. The complex methods are well described and represent an advancement in the field.

I think you need to include a definition of ladder and ladder fuels as you use it in the study in the introduction. You define ladder fuels in lines 212-215 but it is a little confusing to talk about other studies that look at ladder fuels in the introduction and then disclose that your definition does not include lower canopy branches in your methods section. In some forest types, the lower, dead branches make up the bulk of the ladder fuels and play a significant role in fire behavior.

I am curious at what point a leaning dead tree becomes a downed log. In beetle-killed lodgepole stands, it is common to have downed trees suspended a few meters above the ground until the large branches rot sufficiently to allow the stem to drop to the ground. In your classification results, it looks like downed logs are generally horizontal and ladder trees are generally vertical. Is this just a side effect of the way filtering is applied or is there logic that forces dead trees into one class or the other based on their orientation and proximity to the ground?

Comments referenced to line numbers

Lines 189-195: A better description of the co-registration and georeferencing processes is needed. How were tie points selected and how many tie points were used? How was the registration error assessed? For georeferencing with ALS data, were control points collected to provide world coordinates or were the point clouds manually aligned?

Lines 198-205: this paragraph is duplicated in lines 188-195

Line 209: I think “seeding” should be “seedling”

Lines 228-229: Caption for figure 4 is the same as caption for figure 3.

Lines 264: Was transparency adjusted for panels (b) & (c) or (c) & (d) or (b), (c), & (d)?

Lines 285-286: this sentence is duplicated in lines 283-284 (but has “predetermined” added). You may have intended this duplication but I don’t think it is needed since line 284 mentions “these steps” which included the comparison to a threshold value and merging if the error was below the threshold.

Line 393: Figure 6 caption: There is no scale bar in the figure.

Line 476: Just increasing the number of training samples might not be as beneficial as ensuring that the full range of conditions is represented in the training samples. You mention your process for locating training samples in lines 248-253 and emphasize that the randomization process was intentional to avoid overfitting. I wonder if overfitting is really a concern with your process.

Figure 11: What is a “duff branch”? This term is not used in the text of the paper.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

Very complete article that provides a comprehensive and insightful analysis of forest fuel determination. The article is well-structured and written, making it an enjoyable and informative read. I only have some minor issues related to the sensor employed and the methodology:

* The introduction is focused on TLS, as is to be expected from the objective of the article, but this does not prevent excluding other methods available for a quick 3D reconstruction for fuel mapping, such as portable mobile mapping systems. Please, complement the introduction with a brief comment on these other alternative methods to avoid any bias.

* The authors employed the intensity attribute for branch classification, but such values cannot be used as is, not only for the reasons given in line 347 but also because of issues such as laser signal attenuation and angle of incidence (unless the authors have performed a vicarious calibration of the TLS used). I consider necessary to point out this aspect in the manuscript’s methodology and as a limitation of the branch classification.

* In relation to the TLS co-registration, it was carried out manually (line 190). While it is true that for a large number of plots, for efficiency reasons, this is the method that is usually chosen, the correct procedure is to use targets (spheres). Therefore, please, make reference to this issue in the text, and that if portable mobile mapping systems were used, it would not even be necessary to have targets. Moreover, the TLS georeferencing to ALS point cloud also increases the error budget, so it has to be estimated and reflected in the text.

* Also, why was chosen the spatial resolution of 5 mm (line 193)? Please, state the reason for choosing this value and not another (like 10 mm for example)

* Lines 198 to 205 are repeated and thus should be eliminated.

Author Response

Please see the attachment

Author Response File: Author Response.docx