Terrain Shadow Interference Reduction for Water Surface Extraction in the Hindu Kush Himalaya Using a Transformer-Based Network

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper provides a detailed description of the research objective, methodology, and experimental results overall. However, there are a few areas that could be improved or expanded upon:

1.More detailed explanation of the dataset composition is needed. Providing specific characteristics of the dataset, such as the ratio of terrain shadow samples, distribution of water/non-water areas, etc., would increase the reproducibility of the experiments.

2.A deeper discussion is required on why the addition of terrain data led to a decrease in water extraction performance. Instead of simply attributing it to differences in physical characteristics, providing more concrete evidence would be beneficial.

3.A quantitative comparison and analysis with other existing water extraction methodologies is lacking. To objectively demonstrate the performance of the proposed method, quantitative comparisons in terms of accuracy or other metrics would be necessary.

4. An evaluation of the applicability of the proposed method to various terrain environments beyond the HKH region would be valuable to assess its generalization capability.

5. An assessment of computational efficiency in real-world operational environments could also be added to demonstrate the practical utility of the methodology.

Comments on the Quality of English Language

Moderate editing of English language required

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The paper is well organized, the methods used are adequately described, the objective of the research is clearly stated and the conclusions of the paper are supported by the results, their analysis showing that the proposed tool has better performance (in terms of OA, IoU and Kappa) than others taken into consideration.

One remark regarding the references, it seems that no. 35 and 36 are not cited in the paper.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

"Terrain Shadow Interference Reduction for Water Surface Extraction in the Hindu Kush Himalaya Using a Transformer-based Network" entitled paper aims to develop a robust deep learning network that will accurately detect water surfaces in the Hindu Kush Himalaya (HKH) region. This is essential for effectively monitoring the status and changes of water resources in this region. The work focuses on significantly reducing the misclassification of land shadows as water surfaces. Images from satellite-based sensors were used to analyze water surface properties. The Vision Transformer, a cutting-edge deep learning network, was employed for this purpose. To enhance the accuracy of water surface detection, terrain shadows have been incorporated into the training dataset and adjusted in different ways. The findings obtained from the proposed method have been validated by comparing it with global surface water (GSW) data, and it has been shown that it exhibits superior performance.

An insight into how the proposed method can perform in different geographical regions should be given.

More detailed information is needed about the sources of training and testing datasets and how they were prepared.

In addition to terrain shadows, you can offer detailed insights into other sources of misclassification and the effective methods used to address them.

By the way, you should check the spelling and language of your paper. A native speaker must revise it, or you may prefer proofreading.

Comments on the Quality of English Language

A native speaker must check the paper for grammar and spelling.

Author Response

Please see the attachment.

Author Response File: Author Response.docx