Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper studies the potential of some machine learning algorithms to predict solid properties based on UAV-based hyperspectral images and LiDAR points. As paper is interesting to read it has major drawbacks which should be addressed for a potential publication as follows:

• Authors mention machine learning algorithms. Main representatives of machine learning algorithms are neural networks/multi layer perceptrons (MLP). What are they not utilized in this paper? On the other hand, some basic algorithms, such as linear regression, would bring more clear interpretation of the results. 

• Reproducibility in this paper is problematic from multiple points of view. Is there any code available of in appendix? Some details are missing and appended code would clarify the details. 

• Data is missing. Can data be published in the supplementary material? I miss exact coordinates of soil sample sites, capture dates, raw data. 

• The authors mention some environmental conditions in the time of capture, such as “calm weather”. Please write in more specific terms: exact dates, temperature, …etc. 

• The authors describe preprocessing steps, such as “The images were processed for geometric and terrain corrections and combined. ” but details are missing: which tools, which algorithms, which parameters, etc. Version of libraries is missing. Which data was used to produce machine learning models? How ML models behave on a training dataset and test dataset? 

• In line 261 the authors mention KS algorithms. but the reference is missing.

• Ablation study is missing. How can various combinations of features improve and worsen the results. 

• Equation 2 has error. What happens, when n = 1?

• The paper is quite wordy in some parts but clear images of the workflow would help more. The paper would benefit if the authors would draw the workflow of the experiment (what are the inputs, outputs, etc…)

• Features, such as elev_AIH_50th should be described in more detail. All features shall be listed in one table. 

• Some figures are low quality and the resolution should be increased. This is evident the most in Figure 4. 

• More results are missing. The authors report only optimal models. What about not the most optimal ones? Additionally, more complete evaluation of the results is missing. Learning curves, convergence curves….

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Good work, but in my point of view some improvements are needed. 

The introduction (chapter 1) is somewhat confusing, and should be rewritten following a logical sequence relating to the work presented.

Lines 84-86 : Why insert this sentence if it is not verified? It would be better to write it differently, insisting on the novelty of this work.

Bibliographical references should be included in the paragraphs describing the results. In particular, some milestones that justify th evaluation of the models (Chapter 3.2, lines 296-301). Also references for the Chapter 3.3 are needed.

The Chapter 4.1 should be moved to the introduction.

Reading the work, it is not clear what contribution Lidar data makes to improving the predictive performance of hyperspectral data. Greater emphasis should be given to this aspect to enhance the study carried out.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The research paper "Mapping soil properties in tropical rainforest regions using integrated UAV-based hyperspectral images and LiDAR points" gives an overview of the hyperspectral and Lidar datasets that were used to test how well the five machine learning techniques for mapping soil properties worked. Here are the suggestions and revisions I can request from the authors.

How can the prediction accuracy or any accuracy assessment be 512.50%, 206.67%, or more than 100%? Also, is it possible to reach at the prediction even 100%?

You should use more sensible expressions (e.g., instead of 2,251.6 mm, you can choose meter, or instead of 371,634.41 m², you can prefer hectare, etc.).

It is hard to read and understand from Figure 1b and 1c to where you get samples.

There is confusion about sorting to transfer to the readers, I suppose. I have read many times and hardly understand what did for what. So, It is necessary to revise to sort the topics of the paper. 

Different contents in Figure 6 should have different legends. Total nitrogen and pH should not be colored the same. Also, the total phosphorus amount in Figure 6 (g/kg) should be presented in a different unit. For example, ppm can be used.

In Figure 6, a and b should be next to each other, not one below the other. To increase clarity, metrics equivalent to each other should be used next to each other.

Comments on the Quality of English Language

should be revised and proofreading

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The paper has been sufficiently improved. The appended supplementary files improve the reproducibility of the results.

Author Response

Comment: The paper has been sufficiently improved. The appended supplementary files improve the reproducibility of the results.

Response: Thank you for your valuable feedback. We are glad to hear that the improvements made to the paper have met your expectations. We also appreciate your positive comments regarding the supplementary files.

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for your response. In my opinion, the work in this version has a more logical structure.

The changes I felt were necessary have been made and, with the revision carried out, the study carried out and the efforts made for this work are highlighted.

Great work

Author Response

Comment: In my opinion, the work in this version has a more logical structure. The changes I felt were necessary have been made and, with the revision carried out, the study carried out and the efforts made for this work are highlighted. Great work.

Response: Thank you for your thoughtful feedback and kind words. We are pleased to hear that you find the revised version to have a more logical structure and that the necessary changes have been successfully incorporated. We appreciate your recognition of the efforts put into this work, and we are grateful for your positive evaluation.