Forest Change Monitoring Based on Block Instance Sampling and Homomorphic Hypothesis Margin Evaluation
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe manuscript addresses an important topic in forest damage detection using remote sensing techniques. However, there are several areas that require significant improvement:
1. For Figure 1, the characteristics of the area/forest should be explained, and the focus should be on describing the disaster damage it has suffered.
2. Line 203 (as well as in the abstract), why emphasize the difficulty of urban areas? This is not related to forest damage.
3. The text below Figures 8 and 9 is disorganized and difficult to understand.
4. Due to the simplicity of the scenes in the data used for the study, the general applicability of the method is questionable, and the conclusions are not universally applicable. For example, the most suitable window size will vary significantly with resolution, land cover type, land cover scale, and damage extent. Instead of providing the most suitable window, it would be better to describe how to find the most suitable window in an unsupervised manner.
5. The applicability of using out-of-bag (OOB) error instead of validation sample points is questionable. Validation sample points are relatively objective, while OOB error can change with parameter settings, thus lacking universality.
6. For such simple images used in this study, could conventional change detection methods adequately extract changes and assess damage extent? It is recommended to use more images to verify the effectiveness of the proposed method.
7. The cited references are not sufficiently novel.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper proposes a classification method of storm-damaged and undamaged areas. The authors claimed that several new findings are given in their paper. However, I did not touch all the clear findings in this paper. Some of the required improvements are reported below.
(1) The introduction only includes some "classical" techniques from 2022 and earlier. The paper should provide a more thorough survey of recent papers post-2022.
(2) The conribtuions are not clear, please describe the details in Introduction and Methodology.
(3) The proposed efficient trick, sample selection appears similar to existing works. Meanwhile, in section 3.2.2, I do not think it is is reasonable for random selection. Could the authors explain that?
(4) It is also challenging to distinguish and understand the methods from the results, please be clear in what data was used for each step.
(5) More similar comparison algorithms (e.g. deep learning) are needed to show the superiority of the proposed algorithm.
(6) In the results of the paper, the authors should discuss some limitations of the methodology presented.
Comments on the Quality of English LanguageMinor editing of English language required
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
Comments and Suggestions for AuthorsThe paper proposes a novel method for forest change monitoring based on block instances sampling and homomorphic hypothesis margin evaluation. The innovativeness and contributions of the proposed method are sufficient to be published in Remote Sensing. The manuscript is well written and organized. Herein, I only have some minor comments.
1. In the introduction, the authors should describe more precisely the challenges of forest change detection and how the proposed methodology addresses these challenges, presented in a clearer way.
2. In the introduction or in the discussion of the experiment, it is recommended to add some reference to the change detection method. For example,
"Structure consistency-based graph for unsupervised change detection with homogeneous and heterogeneous remote sensing images." IEEE transactions on geoscience and remote sensing 60 (2021): 1-21.
"Nonlocal patch similarity based heterogeneous remote sensing change detection." Pattern Recognition 109 (2021): 107598.
"Similarity and dissimilarity relationships based graphs for multimodal change detection." ISPRS Journal of Photogrammetry and Remote Sensing 208 (2024): 70-88.
3. The first occurrence of the OOB abbreviation in the summary should use the full name.
4. The downward rounding symbols in Eq. 2 should be given as an indication.
5. It is suggested that in giving some detail enlargements of the results, e.g., Figure 8 and Figure 10. It is currently difficult to see the difference between the subfigures.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe quality of the paper has improved compared to the previous round of review. The research objectives, methods, and results are now clearly described. However, the following main issues still exist:
- As mentioned previously, if the number of images is small or the scenes are simple, simple methods could achieve damage area extraction. There is no need to adopt complex approaches in such cases.
- The method lacks innovation. The entire process, including feature extraction, sampling, and classification, employs relatively conventional processing techniques.
Author Response
Please see the attachment
Author Response File: Author Response.docx
Reviewer 2 Report
Comments and Suggestions for AuthorsMost of the problems have been addressed. I recommend accept after revising some text typos.
Comments on the Quality of English LanguageAccept after minor revisions
Author Response
Thank you for your valuable suggestions. We have made approximately 30 corrections, including spelling errors, errors in the use of conjunctions, tense and voice inaccuracies, improper use of articles (a/an/the), addition of em dashes, and missing punctuation. Thank you very much again for taking the time and effort to review our manuscript.