Landscape Character Classification with a Deep Neural Network: A Case Study of the Jianghan Plain
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe article is both relevant and timely as a research proposal. It explores the use of deep neural networks for landscape classification, a method that complements traditional Landscape Character Assessment (LCA) approaches with quantitative deep learning techniques such as the ResNet-34 model. For this purpose, the authors used Jianghan Plain in China as the study area, leveraging remote sensing images and the LANMAP methodology to create an intelligent landscape recognition system.
This study contributes to the field of environmental studies by integrating natural and social data into an intelligent model—an advancement that could be widely applied to other regions with complex ecosystems. The article identifies limitations related to image resolution and accuracy at the mesoscale level, suggesting directions for future research, such as using higher-resolution images and adapting the model for specific habitat analyses.
The introduction and objectives are clear and showcase progress in the research field, emphasizing the potential of artificial intelligence tools to enhance environmental management and landscape planning with high precision. The methodology is well-structured, detailing the segmentation of remote sensing data into analytical units at different scales (macro and meso), enriching the analysis procedures by allowing model adaptation to various levels of detail. The adaptation of ResNet-34 for multispectral images demonstrates methodological innovation and technical adaptability. Additionally, a category optimization module was employed to consolidate similar classes, which improved model accuracy and generalizability.
The results are presented clearly, showing that the proposed model achieved accuracy rates of 85% and 89% after category optimization, surpassing traditional methods. The discussion clearly highlights how categorization at different scales reveals relationships between natural and social landscape characteristics, making a significant contribution to LCA. However, a more in-depth comparative analysis with other deep learning methods used in landscape classification would be beneficial.
Author Response
Thank you very much for taking the time to review this manuscript.
Please see the attachment for point-to-point responses.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe article is relevant in the technical-scientific context. The methodology and results obtained were positive. In my opinion, the article needs some adjustments to be published in Land/MDPI. Below are some recommendations.
1. Some bibliographic citations need to be included in item 1.2 RESEARCH OBJECTIVES.
2. The METHODOLOGY needs to detail and specify the remote sensing data used in the study in terms of spatial, spectral and temporal characteristics. A detailed table on the remote sensing data should be included.
2. The DISCUSSION presented requires the inclusion of bibliographic citations of results from the literature to corroborate those obtained in the study.
4. The article presented only thirty scientific references. I see that the article requires more bibliographic citations and scientific references.
Author Response
Thank you very much for taking the time to review this manuscript.
Please see the attachment for point-to-point responses.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for Authors
1. General comments: this research has already been completed, so details must be written in past tense. See highlighted verbs in the abstract for your attention.
2. References are required for the details of the study area and methods.
3. The structure of the manuscript should be organized. For example, Section 2 Materials and Method can be divided into two sections 2.1 Study area and Data, and 2.2 Method, and then sub sections for 2.2.
4. Results section should include classification results as maps/tables/figures etc.
5. Discussion – discuss positive and negative points of your results, compare them with existing research.
6. Conclusion – incomplete.
7. Abstract
It is not necessary divide section in the abstract such as context, objectives, methods etc. Please remove them and make one paragraph. This paragraph should include few sentences to introduce the project, its aim and objectives, data and method used, results and discussion and a conclusion. Basically, it should be the summary of the project.
8. Keywords – please do not use abbreviations
9. Introduction (page 1) – introduce abbreviations before using them and use past tense to talk about the project in the first paragraph
10. Introduction – there are ten other comments in the pdf.
11. Please see manuscript for other comments.
Comments for author File: Comments.pdf
Author Response
Thank you very much for taking the time to review this manuscript.
Please see the attachment for point-to-point responses.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsDear authors,
The manuscript has significantly improved, and I am glad about that. I would further suggest rewriting the conclusion section to represent the summary of the study. It should contain:
1. a few sentences about the introduction, including research motivation
2. brief description of data and the method
3. Results and discussion of significant findings
4. What can you conclude from the results?
Hope you will amend this as well,
Thanks.
Author Response
Dear reviewer,
Thanks for the suggesions in the 2nd round review, we revised the conclusion section according to the described structure (line 553 to 565), and we believe it now better summarizes the motivation, method, results and conclusions.
Thanks again for all the constructive comments for this work.
