Combination Strategies of Variables with Various Spatial Resolutions Derived from GF-2 Images for Mapping Forest Stock Volume

Round 1

Reviewer 1 Report

Reviewer Comments

1.       The study area should have a separate section from the Materials and Methods section. Typically, the study area section would provide a detailed description of the location, including any relevant geographical, ecological, or cultural characteristics.

2.       The paper could benefit from more detailed explanations of the Boruta algorithm, as it's not a widely-known feature selection technique.

3.       The authors could have provided more insight into why they chose the specific vegetation indices they used, as some may be redundant or not as informative as others.

4.       The units for the wood volume measurements are not specified, which could be confusing for readers.

5.       In Figure 3, it would be helpful to have a scale bar to show the size of the down-scaled images.

6.       The authors could have provided more discussion on the limitations of using only optical remote sensing data for mapping forest stock volume, as other factors such as tree species, age, and soil type can affect FSV.

7.       The authors did not mention the specific version of the Boruta algorithm they used for feature selection. As there are multiple versions available, this could potentially affect the reproducibility of the study.

8.       The authors could have provided more information on the tree species and stand ages of the sampled plots. This information could have helped readers understand how well the results could generalize to forests with different characteristics.

9.       The use of only one-year ground measurements for model training and validation could be considered a limitation, as forest conditions can vary between years. It would have been useful if the authors had discussed the potential impact of this limitation on their results.

10.   While the authors did mention the use of topographic correction for optical satellite images, they did not provide details on the specific method used. This could have been useful for readers who are interested in replicating the study.

11.   The study focuses specifically on planted forests, so it would be important to investigate if the results generalize to other types of forests. Different types of forests may have different spectral and textural characteristics, which could affect the optimal feature sets and spatial resolutions for mapping FSV.

12.   The study does not explicitly consider the cost or feasibility of obtaining images at different spatial resolutions. It would be helpful to know if the gains in accuracy from using higher spatial resolution images are worth the additional cost and effort.

13.   The study assumes a linear relationship between the selected features and FSV. It would be interesting to investigate if more complex models (e.g. non-linear models) could improve the accuracy even further.

14.   The study only considers four statistical indicators (MAX, MIN, mean, and CV) to analyze the effect of spatial resolution on spectral information. While these are useful indicators, it would be helpful to explore other statistical measures or feature extraction techniques to better understand the relationship between spatial resolution and spectral information.

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Dear Authors

Abstract: The result might depend on the forest structure. So I think it is better to shortly describe the study area in the abstract. 

Study area: please describe the studied forests with details such as type, mean volume, tree per hectare, etc.

Figure 2: Usually, the relationship between DBH and FSV is shown as DBH in the x-axis.

 

The quality is good. But it is recommended to use better wording in some sentences such as the following one:

"And spectral features (SFs) and texture features (TFs) derived from these optical images with difference spatial resolution are commonly employed to construct various models." 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

15/05/2023

Dear authors,

In the manuscript Combination strategies of variables with various spatial resolutions derived from GF-2 images for mapping forest stock volume you propose combination strategies of variables with various spatial resolutions to improve the accuracy of mapping forest FSV using high spatial resolution of images (GF-2).

General comments

The study is interesting and has the potential for a better insight into the forest’s scene within the use of remote sensing methods. Research and analysis show promising results.

The abstract is too long (the recommendation is 200 words, and you have 384) and contains too detailed a description of the results. In it, you only need to briefly state the motivation and announce the results that you will state and comment on in the Conclusion.

When specifying specific procedures for preparing samples for analysis or performing analysis, you must be more specific. Explain how you are implemented and provide references. Furthermore, for many terms you have given abbreviations (spectral features, texture features, vegetation indices, …), but in the text you mostly mention them by their full name and only occasionally use abbreviations.

I think the main problem with your manuscript is the following. You pointed out the well-known fact that with the improvement of the spatial resolution, the heterogeneity of the canopy also increases. However, have you ever wondered to what extent a 10m x 10m pixel generalizes the forest area? How many clearings can be hidden on that, and larger, surface? I think you approached the problem one-sidedly, relying on statistical parameters that can mask the problem of scene detail.

In the Conclusion you must interpret all the results and highlight what you have proven in your tests. However, do not repeat what you did and what you wrote earlier. In it, you should provide those details about the results from the Abstract.

Specific comments (are in the manuscript)

-          Lines 15-41 – The abstract is too long and contains too detailed a description of the results.

-          Line 58 - You can replace this text with the abbreviation you defined in the previous paragraph.

-          Line 81 - You can replace this text with the abbreviation you defined in the previous text.

-          Lines – 105-106 - What does this mean? Is it band sharpening?

-          Line 135 - GF-2 satellite images are declared as: 0.81 m for PAN and 3.24 m for MS imagery. https://www.eoportal.org/satellite-missions/gaofen-2#mission-capabilities

-          Line 138 - This page can be found (404 Not Found).

-          Lines 148-149 - Please clarify this. How did you scale the images?

-          Line 168 - You should introduce this abbreviation the first time you use this term.

-          Lines 177-178 - This should be explained or referenced.

-          Lines 192-193 - Are you sure about this statement? Better spatial resolutions better show variations in spectral responses, that's true, but is that ultimately true. If so, you need to prove it or provide a reference.

-          Line 232 - Min and mean are wrongly shown in figure a). It should be the other way around.

-          Line 233 - What do you mean by the term 'feature resolution'?

-          Line 272 - Re-introduce the same abbreviations. You have already done this in line 56.

-          Line 371 - Such type of manuscript should be written in the third person, as you have done so far.

-          Lines 450-453 - This is redundant. You stated all this in the earlier text, and this has no place in the Conclusion.

-          Line 453 – Again, such type of manuscript should be written in the third person.

 

 Best regards

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

20/05/2023

 Dear authors,

 

In revised manuscript Combination strategies of variables with various spatial resolutions derived from GF-2 images for mapping forest stock volume you propose combination strategies of variables with various spatial resolutions to improve the accuracy of mapping forest FSV using high spatial resolution of images (GF-2).

 

General comments

You answered all my comments and questions. I am mostly satisfied with them and have no major complaints.

 

However, the Abstract is still too long, 320 words. You need to learn to announce and present your work and results concisely.

 

Best regards

Author Response

Please see the attachment.

Author Response File: Author Response.docx