Tree Stem Detection and Crown Delineation in a Structurally Diverse Deciduous Forest Combining Leaf-On and Leaf-Off UAV-SfM Data

Round 1

Reviewer 1 Report

The manuscript is comprehensibly and carefully written. I appreciate the extensive introduction including the description of utilized methods and the table of recent studies related to the topic of the manuscript.

The description of the methodology is mostly clear as well as the presentation of the results. There are some formulations in the methodology that deserve better explanation:

1.       Line 398: various parameters: can you be more specific in the list of various parameters?

2.       Line 399: (difference from) the minimum and maximum: The meaning of brackets is not clear, Should it be 1) the minimum, 2) the maximum, and 3) the difference between them? Please reformulate.

3.       Line 400: (difference from) the 3-dimensional distance: Again, the meaning is not clear; a difference from a distance is nonsense.

4.       Line 403: described by their center coordinate: please explain how the center coordinate is assessed.

5.       Line 437: very small objects were merged: please specify the criteria of defining very small objects

The main message of the manuscript that using tree stem positions as starting points help the accuracy of crown delineation is not really innovative, it has been used in many 3D data processing. However, applying this on the combination of leaf-on and leaf-off data might be an interesting approach for assessing data that are difficult to get with standard UAF SfM approaches.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript presents interesting research on the possibilities of UAVs equipped with RGB cameras, and algorithms for accurate detection and delineation of individual trees and their crowns in dense forests. A good overview of recent studies on tree detection and crown delineation in forest ecosystems using UAV optical data and tree detection algorithms is also presented. The manuscript, in its current form, needs some revision. The presentation of the obtained results could be improved.

Note to the manuscript:

Figure 3. There is a bar in the figure with a mark of 55 without a unit. What is this- meters or something else? Similar is for Figure 7.

Tables 4 and 5. There are color codes in the Precision, Recall, and F1-score columns. There is no information about what the different colors mean.

Figure 6. It would be good to put a scale for the colors corresponding to the height of the trees.

 

Figure 10. If the lines from the corresponding figures (A and B; C and D) are superimposed, an idea of the accuracy of crown delineation will be more clearly obtained.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

For a long time now, there have been works that try to detect individual trees in a stand based on remote sensing data, either manually, semi-automatically or automatically, see examples as

- Matějka K. (2009): Assessment of tree layer biomass and structure using aerial photos in lake catchments of the Šumava Mts. - Journal of Forest Science, 55(2): 63-74.

- Machala M (2016): Design of Application for Assessing the Height of Trees in Forest Stands Based on Images from an Unmanned Aerial Vehicle. Doctoral Thesis (web available)

Hájek F. (2007): Automated classification of tree species composition from remote sensing data. PhD Thesis (web available)

So, the assessed work is very interesting and inspiring. The methodology is appropriately chosen, the data are adequate and the results illustrate it. I have only several small comments.

The text is not easy to read because it uses many abbreviations. Therefore, it would be advisable to include a list of abbreviations (perhaps as one table).

I do not see info about average tree age and size.

Line238-240: It is not clear, what is mean under 40% decrease of precipitation sum (in which period: week, month, year or some one other?)

Why don't Figures 6 and 7 show the same view of space? Thus, the results cannot be clearly compared.

Line 478-479: Modify references according to the journal's proposals.

Fig. 10: Include calculated numerical parameters for both individual examples.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Reviewer’s Report on the manuscript entitled:

Tree Stem Detection and Crown Delineation in a Structurally Diverse Deciduous Forest Combining Leaf-On and Leaf-Off UAV-SfM Data

 

The authors developed a method for precise tree stem detection and crown delineation in dense deciduous forests in the Hainich National Park, Germany via UAV equipped with RGB cameras. I found the method and results interesting, but the presentation requires improvement. Generally, the manuscript is lengthy and carries some unnecessary information that can be found in literature and simply be cited. Lines 109,121,131,143 just have heading in bold face that is not recommended in Introduction. Introduction is usually recommended to be about 1000 words with 5-8 paragraphs. The main contributions of the manuscript can be highlighted at the end of Introduction using preferably bullet points or Arabic numbers.  

 

Table 1 is nice but very lengthy. It could be shortened while conveying the main message.

 

Lines 61-65. The following recent article about the use of UAV in extraction of tree crown can also be included here: https://doi.org/10.3390/rs14051239

Line 154. Also, please add the ANN model as described in the article above.  

 

The following article also describes an algorithm for individual tree crowns segmentation using UAV oblique photos that can be discussed:

https://doi.org/10.1016/j.jag.2022.102893

 

It is in my view, more useful to have the geographic latitudes and longitudes in degrees, minutes for the study region.

 

Section 2.2 What are the specifications of the digital camera, such as Camera Model, Resolution, Optical zoom, Focal Length, Pixel Size? Perhaps adding a Table?  

 

Equations (1)-(3) need a reference with application to UAV, for example:

https://doi.org/10.3390/rs14061523

 

Table 5. Third column is just Prop. Method! So, this column can be removed. I also suggest comparing your model with at least another model on the same dataset.   

 

Adding an acronym table at the end of the manuscript would be useful.

 

Thank you!

Regards,

Please check for grammar/typos/punctuation issues.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 5 Report

I spent over six years in Jena and visited Hainich National Park. It is quite magical for me to be reviewing a manuscript from the Jena Tower. Therefore, I would like to offer all my knowledge to help you improve your work, in the hope that it can contribute something new to our society. Your manuscript is good enough for publication now. All my comments are suggestions, and you can choose to adopt them or not.

From a higher perspective, your work lacks innovation. It integrates two conventional studies in processing SfM point clouds from forests. However, I see a very good intention in your work. You discussed the potential of using UAVs as an alternative to human forest field surveys. For this purpose, you conducted two flights.

For the assessment of a forest, it would be better to know the height and DBH of individual trees. Your dataset is good enough to do this work, so why didn't you do it? If this work could be included, your work would become more useful for helping the majority of people who have to collect this data in the forest field.

Secondly, I would like to suggest that you use the leaf-on data to make a prediction of the stem position, and even the DBH. Your leaf-off data would be good training material for AI models. I did a little research on this, and it seems like a few studies have done similar work. The CHM model for a crown contains much more information than canopy cover or height. Is it possible for us to gain something from this? I hope to hear something new from you.

 

Say hallo to the people on Jena Tower(which floor?).

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 6 Report

Thank you.  Your research approach and results were clearly presented and easy to follow.  I think the main benefit of this work is with respect to leaf on and leaf off applications and the use UAC-SfM approach. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 4 Report

Dear authors,

Thank you for addressing my comments satisfactorily.

Regards,

Please carefully proofread the manuscript and correct any typos/style/punctuation/grammar issues.