Fine Identification of Lake Water Bodies and Near-Water Land Using Multi-Source Remote Sensing Fusion: A Case Study of Weishan Lake, China

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors, 

What is the main question addressed in this study? 

In their paper, "Fine Identification of Lake Water Bodies and Near-Water Land Using Multi-Source Remote Sensing Fusion: A Case Study of Weishan Lake, China," the authors utilised the ability to analyse data from a multi-source remote sensing system to monitor the hydrology of the area. Their primary goal was to confirm the applicability and adaptability of multi-source remote sensing in water monitoring. Specifically, they compared the SAR, RF and CART methods and assessed the efficiency of extraction in coastal buffer zones and reservoirs. 

Do you consider this topic original or relevant to the field? Does it fill a specific gap in the field? Please also explain why this is/is not the case. 

In their paper, the authors confirmed the reliability and scalability of the synergistic approach to water identification. They proposed a research framework and the potential for application across multiple sources. It also demonstrated significant adaptability and usefulness in areas with complex hydrography. 

What does it contribute to the field compared to other published materials? 

In the context of this problem and other studies widely described in the literature, the solution to the problem can be considered significant. Despite its complexity, the use of a multi-source remote sensing system allows for identification results at a satisfactory level. 

What specific methodological improvements should the authors consider? 

The research contribution of this work is valuable. The analysis and graphical presentation were carefully prepared. It was also demonstrated that this comprehensive model maintains spatial consistency in the test areas at a satisfactory level. 

Of course, additional factors and environmental variables could be included to enhance the predictive stability of the results, resulting in additional results. This point has already been emphasised by the authors in the body of the work. 

- Are the conclusions consistent with the evidence and arguments presented and do they answer the main question? Please also explain why this is the case. 

Yes, the conclusions are consistent with the presented evidence. Based on the results and analyses obtained, they confirm clear differences between the algorithms in terms of accuracy, stability, and misclassification mechanisms. The combination of remote sensing data from multiple sources significantly increased the robustness and precision of water body extraction. The summary also confirms the applicability and adaptability of multi-source remote sensing for monitoring water and coastal areas. 

- Are the sources appropriate? 

The source materials are definitely appropriate. 

- Any additional comments regarding the tables and figures. 

 In Fig. 2, the text will be difficult to read when enlarged due to the resolution. 

 Fig. 17: Units used for evapotranspiration 

Author Response

Dear Editor and Reviewer,

We sincerely thank the reviewer for the careful evaluation and positive comments on our manuscript entitled “Fine Identification of Lake Water Bodies and Near-Water Land Using Multi-Source Remote Sensing Fusion: A Case Study of Weishan Lake, China”. We appreciate the reviewer’s recognition of the relevance, methodological rigor, and applicability of our work. Below, we provide detailed responses to each comment. All revisions mentioned have been incorporated into the revised manuscript and are marked using the Track Changes function.

Comments1: What is the main question addressed in this study?

Response1: Thank you for this summary and interpretation of our work. We fully agree with the reviewer’s understanding. As correctly stated, the central research question of this study is to evaluate whether a multi-source remote sensing fusion framework can improve the accuracy, robustness, and adaptability of water and near-water land identification under complex hydrological and agricultural conditions. The manuscript has been carefully revised to further clarify this core research question in the Introduction (Section 1, last paragraph), where the objectives and scope of the study are explicitly stated.

 

Comments 2: Do you consider this topic original or relevant to the field? Does it fill a specific gap in the field?

Response 2: We appreciate the reviewer’s positive assessment. As noted, while multi-source remote sensing and machine learning have been widely applied in water extraction studies, this work addresses a specific gap by (i) explicitly introducing and operationalizing a “near-water land” class to characterize transitional lake–land zones, and (ii) integrating frequency-based temporal analysis with factor interpretation (RF and SHAP) to improve both classification robustness and physical interpretability. These aspects have been further emphasized in the revised Introduction and Discussion sections to more clearly articulate the study’s novelty and contribution.

 

Comments 3: What does it contribute to the field compared to other published materials?

Response 3: We thank the reviewer for recognizing the significance of our contribution. In response, we have strengthened the discussion of how the proposed framework differs from existing approaches, particularly in terms of its transferability across heterogeneous lake systems and its ability to quantitatively interpret environmental drivers of classification uncertainty. These clarifications can be found in the revised Discussion (Section 4.1 and 4.2).

 

Comments 4: What specific methodological improvements should the authors consider?

Response 4: We appreciate this constructive comment. As suggested, the potential inclusion of additional environmental variables has already been acknowledged as a future research direction. In the revised manuscript, we have refined the “Limitations and Future Prospects” section (Section 4.5) to more explicitly discuss feature selection strategies, overfitting control, and computational trade-offs, thereby providing a clearer methodological outlook without altering the core experimental design.

 

Comments 5: Are the conclusions consistent with the evidence and arguments presented?

Response 5: We thank the reviewers for confirming that our conclusions are consistent with the evidence provided. Regarding this, we have slightly revised the wording of the "Conclusions" section to better align the summary findings with the quantitative results discussed in the "Results" and "Discussion" sections.

 

Comments 6: Are the sources appropriate?

Response 6: We appreciate the reviewer’s confirmation that the cited sources are appropriate.

 

Comments 7: Some additional comments regarding the tables and figures.

Response 7: Thank you for these helpful suggestions.

(1) Figure 2 (now Figure 3): We agree that the original resolution could affect readability when enlarged. Accordingly, the workflow figure has been completely redrawn with higher resolution and clearer layout. The updated figure is now presented as Figure 3 in the revised manuscript (around Line 354), with improved font size, spacing, and graphical clarity.

(2) Figure 17 (now Figure 18): Units used for evapotranspiration: Thank you for pointing this out. The evapotranspiration variable used in this study is derived from the FAO WaPOR Level 1 Actual Evapotranspiration and Interception (AETI_D) dataset, which is expressed in mm/day. In addition, surface temperature and precipitation variables are obtained from the ERA5-Land reanalysis dataset, following their standard physical units. We have clarified the evapotranspiration unit in the figure caption and the main text. Moreover, we have explicitly noted that Figure 18 presents the SHAP-based feature contribution and interaction patterns, illustrating the relative importance and shared influence of environmental variables rather than their absolute magnitudes.

Once again, we sincerely thank the reviewer for the constructive comments and positive evaluation. We believe that the revisions have further improved the clarity, rigor, and presentation quality of the manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors, I've read your manuscript entitled “Fine Identification of Lake Water Bodies and Near-Water Land Using Multi-Source Remote Sensing Fusion: A Case Study of Weishan Lake, China” to Sustainability. The topic is relevant and the study addresses an important issue for sustainable water resource management. However, after a detailed review, I believe the manuscript requires major revision before it can be considered for publication. Below are my comments and specific recommendations:   Originality and Contribution

I think that your manuscript is well-written and methodologically sound, but the novelty is not clearly articulated. Multi-source fusion and machine learning for water body extraction are widely studied.

Please you could:

-explicitly state what is new compared to previous works (e.g., unique integration strategy, introduction of “near-water land” class, or methodological improvements).

-Consider adding a short paragraph in the Introduction and Discussion highlighting the innovation.

 

Methodology

I think that the workflow is comprehensive, but some details are missing:

-Training and validation: Clarify how samples were distributed across seasons and land cover types. Was the dataset balanced?

-Statistical significance: Accuracy metrics (OA, Kappa) are reported, but no statistical test is provided to compare methods.

You could: add details on sample representativeness and include a statistical comparison (e.g., t-test or ANOVA) to support claims of superiority.

Results and Interpretation

I consider figures 3–5 and 8–11 difficult to be interpreted; captions lack sufficient explanation.
You could:

- Improve figure resolution and provide more descriptive captions (e.g., what the reader should observe, key differences between methods).

-Include a summary table comparing NDWI, SAR-Otsu, RF, and CART (accuracy, pros/cons) for quick reference.

Discussion

In my opinion the discussion is too long and partially repetitive.
You could:

- Condense the discussion and focus on:

i) Why the proposed framework performs better.

ii)Practical implications for sustainable irrigation management.

iii)Limitations and concrete future directions (e.g., feature selection to reduce overfitting, computational cost analysis).

References

Ensure references are formatted according to journal guidelines and check for recent relevant studies beyond those already cited.

Summarizing I think that your study is promising, but improvements in clarity, originality, and methodological rigor are necessary to meet publication standards.

Comments on the Quality of English Language

 The manuscript would benefit from editing for conciseness and readability.

Some sentences are verbose and could be simplified for clarity.
I ask to the authors for revising for conciseness and readability, especially in the Introduction and Discussion.

Author Response

Dear Editor and Reviewer,

We sincerely thank the reviewer for the thorough evaluation and constructive comments on our manuscript entitled “Fine Identification of Lake Water Bodies and Near-Water Land Using Multi-Source Remote Sensing Fusion: A Case Study of Weishan Lake, China”. We appreciate the reviewer’s recognition of the relevance of the topic and the potential contribution of this work to sustainable water resource management. In response to the comments, we have carefully revised the manuscript to improve clarity, originality, and methodological transparency. All revisions are marked using the Track Changes function. Our detailed responses are provided below.

Comments 1: Originality and Contribution: The novelty is not clearly articulated. Multi-source fusion and machine learning for water body extraction are widely studied.

Response1: Thank you for this important comment. We agree that clearly articulating novelty is essential. In the revised manuscript, we have explicitly clarified the innovative aspects of this study in both the Introduction (Lines 107–125) and the Discussion section. Specifically, we emphasize:(i) the explicit introduction and operationalization of “near-water land” as an independent transitional class,(ii) a multi-seasonal, frequency-based integration strategy within the GEE framework to enhance temporal robustness, and(iii) the combination of multi-source fusion with interpretable machine learning analysis to reveal misclassification mechanisms and environmental controls.These additions clearly distinguish our work from previous studies focusing solely on binary water–non-water classification.

 

Comments 2: Methodology: Clarify training and validation across seasons and land cover types; include statistical significance tests.

Response 2: We appreciate your methodological suggestions. To this end, we have revised Section 2.2.3 to more clearly articulate the seasonal distribution and representativeness of the training samples, including the rationale for constructing seasonally specific training datasets to account for vegetation and hydrological variability. The principles of sample balancing and selection have also been more explicitly described. (Lines 251–254)

Regarding statistical significance, we have added a bootstrap-based confidence interval analysis to the accuracy metrics to support method comparisons, thereby enhancing the robustness of the reported OA and Kappa results without relying on overly stringent assumptions. These methodological improvements enhance the rigor of the comparative evaluation. (Lines 268–275, with corresponding explanations added to the results section)

 

Comments 3: Results and Interpretation: Figures are difficult to interpret; captions lack explanation.

Response 3: Thank you for this helpful feedback. In the revised manuscript, figures in Sections 3.1–3.3 have been systematically improved through higher resolution rendering, optimized color schemes, and clearer visual layouts. Figure captions have been expanded to guide readers on key patterns and differences among methods. In addition, the main text now includes enhanced quantitative descriptions to better support visual interpretation. To facilitate comparison across methods, we have also strengthened the textual synthesis of NDWI, SAR-based, RF, and CART results. These revisions collectively improve the readability and interpretability of the Results section.

 

Comments 4: Discussion: The discussion is too long and partially repetitive.

Response 4: We agree with the reviewer’s assessment and have accordingly condensed and restructured the Discussion section. Without altering the overall section framework, we reduced redundancy and sharpened the focus on three key aspects:(i) why the proposed multi-source fusion framework achieves improved performance, (ii) its practical implications for sustainable irrigation and lake water management, and(iii) clearly defined limitations and concrete future research directions, including feature selection and computational considerations. These changes improve conciseness while preserving the depth and continuity of the discussion.

Once again, we sincerely thank the reviewer for the constructive comments and positive evaluation. We believe that the revisions have further improved the clarity, rigor, and presentation quality of the manuscript.

Reviewer 3 Report

Comments and Suggestions for Authors

The comments of “Fine Identification of Lake Water Bodies and Near-Water Land Using Multi-Source Remote Sensing Fusion: A Case Study of Weishan Lake, China”

The content of the article is complete, the innovation is average, there are many errors in text and symbols, and the images are not standardized enough.

Here are some revision suggestions for the paper:

The text in the image is too small and the font is inconsistent with the main text. It is recommended that each image be modified to a uniform format, with added legends and annotations.

L29-30 The main text does not extensively discuss "irrigation water resources" and "sustainable water management". It is suggested to delete them.

L42 An additional point(.) has been added.

L162 Table 1 lists data of different resolutions, how are they fused? You should discuss them in detail.

L225 The three classes should be introduced in detail using typical ground pictures.

L232 How much training is required for different seasons?

L235 It is recommended to list the model parameters in a table for clearer representation.

L258 “.[22]” should be replaced with “[22].”

L273 These two extraction schemes should be highlighted in the flowchart (Figure 2).

L321 The diagram and text are unclear; it is recommended to enlarge and rearrange the flowchart. How are interference factors such as clouds, vegetation, and complex terrain screened? How are environmental factors reflected in the flowchart?

L334-335 What are the reasons for the missing months of July and December? The image is not clear, and the boundary classification result cannot be seen.

L336-337 There is Chinese in the picture. Where is this area? Is it related to Figure 3? Is this the result extracted from Sentinel-1's data? Comparison charts before and after extraction should be added.

L359-360 What does 1-12 represent in the figure? Does 20-90 represent a percentage?

L369 Is it Figure 6?

L382-383 Should Figure 8 be in Section 3.3?

L390-391 What do the numbers in Figure 9 represent?

L417 At what time were the three lakes?

L441-442 Why do we have data for July and December here?

Figures 15-16-17 are not cited in the main text.

L509 An additional point(.) has been added.

L567 [33.44] should be replaced with [33,44]

L581 (2025) should be deleted.

L648 Please revise all references according to the reference format of the journal

L649 “References……and listed” should be deleted.

Author Response

Dear Editor and Reviewer,

We sincerely thank the reviewer for the careful evaluation and positive comments on our manuscript entitled “Fine Identification of Lake Water Bodies and Near-Water Land Using Multi-Source Remote Sensing Fusion: A Case Study of Weishan Lake, China”. We appreciate the reviewer’s recognition of the relevance, methodological rigor, and applicability of our work. Below, we provide detailed responses to each comment. All revisions mentioned have been incorporated into the revised manuscript and are marked using the Track Changes function.

Comments1: The text in the image is too small and the font is inconsistent.

Response1: Thank you for your suggestions. All charts have been redrawn or reformatted, and legends and annotations have been optimized. The resolution and layout of the charts have been optimized to ensure legibility in both digital and print formats.

 

Comments 2: L29–30: The manuscript does not extensively discuss “irrigation water resources” and “sustainable water management.”

Response 2: We agree with this comment. To avoid overstatement, references to “irrigation water resources” and “sustainable water management” in Lines 29–30 have been removed. Related implications are now discussed more explicitly and appropriately in the Discussion section.

 

Comments 3: L42: An additional point (.) has been added.

Response 3: Thank you for pointing this out. The extra punctuation has been corrected.

 

Comments 4: L162: Table 1 lists data of different resolutions; how are they fused?

Response 4: Thank you for this important comment. The data fusion strategy has already been described in Lines 182–187 of the revised manuscript and has been further refined for clarity. We now explicitly explain how multi-resolution datasets were temporally aggregated and spatially resampled to ensure consistency with classification inputs (see optimized sentence in the revised text).

 

Comments 5: L225: The three classes should be introduced in detail using typical ground pictures.

Response 5: We agree with this suggestion. Detailed descriptions of the three land cover classes (water bodies, near-water land, and non-water areas), supported by representative reference examples, have been added in Lines 251–266 of the revised manuscript (especially water bodies and near-water areas).

 

Comments 6: L232: How much training is required for different seasons?

Response 6: Thank you for this comment. The seasonal distribution of training samples has now been clarified in Lines 251–266, including the rationale for constructing season-specific training datasets to account for vegetation and hydrological variability.

 

Comments 7: L235: It is recommended to list the model parameters in a table.

Response 7: We appreciate this suggestion. A concise table summarizing the key parameters of the RF and CART models has been added to improve clarity and transparency (see revised Methods section).

 

Comments 8: L258: “.[22]” should be replaced with “[22].”

Response 8: Thank you for noting this. The citation format has been corrected.

 

Comments 9 & 10: L273 and L321: Extraction schemes should be highlighted in the flowchart; the diagram and text are unclear.

Response 9 & 10: Thank you for your feedback. The methodology flowchart has been redesigned and expanded, clearly highlighting the two main extraction schemes (optical-based and SAR/machine learning-based). Congestion factors such as clouds, vegetation, and complex terrain are now represented through dedicated preprocessing and feature building steps in the flowchart, and environmental variables are explicitly included as inputs for subsequent analyses.

 

Comments 11 & 12: L235: It is recommended to list the model parameters in a table.

Response 11 & 12: Thank you for your valuable feedback. Explanatory text has been added before the relevant charts (now on lines 359-364) to clarify the time span. The resolution and layout of the charts have been improved, related plates have been rearranged to enhance consistency between charts, all annotations have been standardized, and a new set of charts with corresponding quantitative interpretations has been added.

 

Comments 13: L359–360: What do 1–12 and 20–90 represent?

Response 13: Thank you for pointing this out. The figure axes and legends have been revised to clearly indicate that 1–12 represent months and that the numerical range corresponds to percentage-based metrics.

 

Comments 14 & 15: L369 and L382–383.

Response 14 & 15: These issues have been addressed implicitly through figure renumbering and layout optimization in the revised manuscript.

 

Comments 16: L390–391: What do the numbers in Figure 9 represent?

Response 16: Thank you for this comment. Figure 9 has been updated with clearer legends and scale annotations to explicitly define the meaning of numerical values.

 

Comments 17: L417: At what time were the three lakes?

Response 17: We agree that this required clarification. The description has been revised in Lines 475–479 to explicitly state that the lake observations correspond to the same periods used for model training and evaluation.

 

Comments 18: L441–442: Why are there data for July and December here?

Response 18: Thank you for pointing out this issue. This was due to the use of outdated tabular data when creating the chart. The table and corresponding chart have been corrected, and the unexpected test data for July and December have been removed.

 

Comments 19: Figures 15–16–17 are not cited in the main text.

Response 19: Thank you for this observation. All figures are now explicitly cited and discussed in the corresponding sections of the main text.

 

Comments 20–24: Minor text and reference formatting issues (L509, L567, L581, L648, L649).

Response 20–24: All listed typographical errors and reference formatting issues have been carefully corrected. The reference list has been fully revised to comply with the journal’s formatting guidelines.

 

Once again, we sincerely thank the reviewer for the constructive comments and positive evaluation. We believe that the revisions have further improved the clarity, rigor, and presentation quality of the manuscript.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors,

thank you for your thorough revision. I read your manuscript and I consider it now much clearer and methodologically robust. The inclusion of bootstrap confidence intervals, multi-seasonal modeling, and SHAP-based interpretability significantly strengthens the work. The tri-class classification framework and cross-regional validation are valuable contributions to the field.

Before publication, please address the following minor points:

1. Discussion: Reduce redundancy and streamline key messages for better readability.
2. Figures: Improve resolution and clarity of maps and charts, especially those showing seasonal variations.
3. Technical details: Consider moving very detailed parameter settings or intermediate steps to Supplementary Material to keep the main text concise.

Overall, in my opinion your paper is well-prepared and ready for acceptance after these minor adjustments.

 

 

Author Response

Dear Editors and Reviewers:

Thank you very much for your careful review of our manuscript and for your encouraging comments. We have made further targeted revisions based on your other suggestions, as detailed below. All revisions have been marked using the "Track Changes" function in the manuscript.

Comments 1: Discussion: Reduce redundancy and streamline key messages for better readability.

Response 1:
Thank you for this helpful suggestion. We agree that improving conciseness enhances the readability of the Discussion. Accordingly, we further streamlined Section 4 by removing repetitive explanations, merging overlapping interpretations, and focusing more clearly on the key findings, mechanisms, and implications. Specifically, redundant descriptions of seasonal effects and vegetation interference were reduced, and methodological advantages were summarized more succinctly. 

Comment 2: Charts: Improve the resolution and clarity of maps and charts, especially those showing seasonal variations.

Reply 2:

We appreciate your comment and have taken it seriously. The resolution and visual clarity of charts involving multi-seasonal comparisons and multi-source data results have been improved. In particular, maps and charts used to demonstrate the performance of seasonal extraction and temporal variability have been optimized for higher resolution, and legends and axes have been appropriately optimized for improved readability.

Comment 3: Technical Details: It is recommended to move very detailed parameter settings or intermediate steps to the supplementary materials to keep the main text concise.

Reply 3:

Thank you for your constructive suggestions. During the revision process, we added some detailed descriptions of model training and accuracy assessment results based on comments from other reviewers. Highly specific or supplementary information has been summarized in the main text. This adjustment improves the credibility of the results while maintaining the transparency of the method. The corresponding changes are mainly located in Section 2 (Methods).

Reviewer 3 Report

Comments and Suggestions for Authors

There are still some formatting issues, such as L571-L576, L617-L622.

Accept after modification.

Author Response

Dear  Reviewer,

Thank you very much for your follow-up review and for your constructive and supportive comments.

Comments:
There are still some formatting issues, such as L571–L576 and L617–L622.

Response:
Thank you for pointing this out. We have carefully checked and corrected the formatting issues at Lines 571–576 and 617–622, which were mainly related to section title line breaks and layout inconsistencies. These issues were caused by automatic formatting changes after using the “accept all revisions” function. All identified formatting problems have now been fully resolved in the revised manuscript.

We sincerely appreciate your careful attention to formatting details throughout both review rounds. Your thorough comments have been very helpful in improving the overall clarity and presentation quality of the manuscript.