Interchangeability of Cross-Platform Orthophotographic and LiDAR Data in DeepLabV3+-Based Land Cover Classification Method

Round 1

Reviewer 1 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

The article presents a structurally complete method for river land cover classification (LCC) based on LiDAR and UAV technology combined with a deep learning model (DeepLabV3+), aiming at assessing the ecological changes in rivers, especially for the changes in vegetation and water bodies in downstream rivers.

1. The introduction is generally detailed, but there is some room for optimization in terms of presentation and structure. It is suggested that the transition between paragraphs should be improved so that the various parts are more closely linked. The expression of some sentences is rather complex and lengthy, which may easily lead to difficulties in comprehension. At the same time, some of the terms and concepts lack sufficient background introduction, which may cause trouble to unfamiliar readers. For example, in line 52, “Heretofore, obtaining such data necessitated field surveys that require personnel to enter the site. In this sentence, “Heretofore” may be obscure to the reader, and it is suggested that it be replaced with the more common “Previously”.

2. Section 2.3, Why ALB and GLS data were chosen and how they supported the study. It is recommended to add a detailed discussion of the selection of ALB and GLS data sources, clarifying their strengths, limitations, and applicability scenarios, to help readers understand why they were chosen and how they complement each other.

3. Part 2.5, in the DeepLabV3+ model section mentioned in Section 2.5, lacks a detailed description of the specific experimental validation methods and results, despite mentioning the research objectives of data interchangeability. The effect on cross-platform data interchangeability has not been clearly demonstrated. Adding more details about data interchangeability validation, including experimental design, evaluation metrics (e.g., OA and Macro-F1), etc., helps readers understand the actual performance of the model under cross-platform data.

4. Section 3.4 mentions that Image Fusion performs best across data types, especially on the F1 scores for two land cover types, Bamboo and Tree. Why does Image Fusion perform particularly well on these two land cover types? It is recommended to add a classification result graph to more visually represent the difference in performance across different data.

5. Lines 426-430, suggesting a detailed discussion of the potential improvements and challenges of expanding study sites, adding seasonal data, and experimenting with other models (e.g., U-Net, ResNet, etc.)

6. The language requires additional refinement by a professional, to ensure it aligns with the requirements of this journal.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report (New Reviewer)

Comments and Suggestions for Authors

A. Strengths

- Innovative Topic: The study focuses on a novel area—interchangeability between ALB and GLS platforms for Land Cover Classification (LCC)—and introduces image fusion as a potential enhancement technique.

- Comprehensive Methodology: The paper includes a detailed description of the study site, data collection, preprocessing, and application of the DeepLabV3+ model.

- Clear Results: The use of OA, Macro-F1, and cross-platform interchangeability metrics provides a robust analysis of model performance.

- Discussion of Limitations: The authors address potential limitations, such as limited study sites and seasonal variations, suggesting future directions for extending the work.

 B. Comments

(1) Abstract and Introduction:

   - The abstract is overly dense and could be rewritten for better clarity. Some sentences are unnecessarily long and contain minor grammatical errors (e.g., "these years" instead of "in recent years").

   - The introduction does not sufficiently highlight the significance of the research gap or the practical implications of cross-platform interchangeability.

(2)  Methodology:

- While detailed, the methodology section includes overly technical details that could overwhelm readers unfamiliar with LiDAR systems.

Consider summarizing or providing supplementary material for highly specific aspects as a Methodology flowchart.

-Validation Methodology: Consider providing a separate paragraph explicitly detailing the validation process used in your study.  This section should clearly describe how the validation metrics (e.g., OA, Macro-F1) were calculated, the datasets used for validation, and any specific steps or considerations taken to ensure robustness.

(3) Graphical Abstract:

I recommend including the validation process as part of the graphical abstract Highlight where and how the validation fits into your methodology. Please add this step to the methodolgy flowchart to provide a clearer understanding of how the validation integrates into your workflow.

(4)  Language:

   - There are several grammatical issues and awkward phrasing throughout the manuscript. For example:

     - "And until now these on-site works still have lots of limitation to be overcame" → "These on-site tasks still face significant limitations."

     - "Considering of the mentioned state of affairs..." → "Given the stated conditions..."

   - A thorough proofreading for English grammar and style is recommended.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear authors,

The current manuscript is well written and technically reasonable, which can be accepted after the following revision.

1 some figures need to be improved to be more clear.

2 Limitation & Potential Application should be moved to discussion.

Reviewer 2 Report

Comments and Suggestions for Authors

I do not deny the innovation and workload of this article, but from the writing style of the article, it does not quite conform to the structure of a scientific paper, such as the sections of Introduction, Results, and Discussion.

The authors do not need to separate the Introduction into different sections to write, while the contents in the Results and Discussion should be presented by different sub-titles.