The State of the Art in Deep Learning Applications, Challenges, and Future Prospects: A Comprehensive Review of Flood Forecasting and Management

Round 1

Reviewer 1 Report

1. Figure 1 is very generic. The authors can give a more detailed model.

2. The highlights of the paper shall be given

3. Instead of a detailed theory, the authors can give a comparative study report(figure) for different algorithms since it's a review paper(at least for two metrics like accuracy. RMSE, etc).

4. The authors shall collect the data collection part and add to the review (duration, methods, and so on).

5. The selection of the right algorithm based on the inputs can be given in the conclusion or in the discussion section.

6. Rephrasing can be done in a few places to improve readability. 

I suggest rephrasing the sentences in a few places for example"  A DL model that has been trained on flood events in one region, for example, can be fine-tuned to forecast flood events in another region with...

difficult to understand. 

Author Response

1. Figure 1 is very generic. The authors can give a more detailed model.

Reply: The Figure 1 has been changed in accordance with the suggestion of the reviewer to take into account more details and class of machine learning.

2. The highlights of the paper shall be given

Reply: As per the reviewer’s suggestion, the highlights have been now included in the revised manuscript.

3. Instead of a detailed theory, the authors can give a comparative study report(figure) for different algorithms since it's a review paper(at least for two metrics like accuracy. RMSE, etc).

Reply: As per the reviewer’s suggestion, Table 2 is now added to show the detailed comparison of Different DL models including accuracy, RMSE.

4. The authors shall collect the data collection part and add to the review (duration, methods, and so on).

Reply: As per the reviewer’s suggestion, Table 1 is now added to show the detailed comparison of DL models.

5. The selection of the right algorithm based on the inputs can be given in the conclusion or in the discussion section.

Reply: As per the reviewer’s suggestion, the selections of the right algorithm based on the inputs are now added in the revised manuscript in discussion section. Page-25.

6. Rephrasing can be done in a few places to improve readability. 

I suggest rephrasing the sentences in a few places for example"  A DL model that has been trained on flood events in one region, for example, can be fine-tuned to forecast flood events in another region with...

difficult to understand. 

Reply: As per the reviewer’s suggestion, the corrections have been done. 

Reviewer 2 Report

The paper delivers a comprehensive and up-to-date review of deep learning (DL) applications in flood forecasting and management. It efficiently summarizes the current state of the art, critically evaluates existing methods, and provides insights into future research directions. The authors lay a strong foundation for addressing challenges in this field and considering ethical implications, which adds value to the paper. However, particular areas lack depth and require more elaboration.

Strengths:

1. The paper covers many deep learning models, data sources, and evaluation metrics used in flood forecasting and management. This holistic view makes it a valuable resource for researchers in the field.

2. The topic is important given the increasing prevalence of severe flooding due to climate change. The paper provides a timely review of the use of cutting-edge DL techniques to mitigate the effects of this global issue.

3. The authors' insights into future research areas, such as hybrid models, uncertainty estimates, data integration, and interpretability improvement, are well-articulated and forward-looking.

4. The discussion on challenges like data availability and interpretability is valuable. This acknowledges the practical difficulties in deploying DL models for flood forecasting.

Drawbacks:

1. The main drawback is the lack of a review methodology. The search strategy, inclusion, and exclusion criteria should be included.

2. While the paper covers a broad spectrum of models, it could benefit from a more detailed examination of each model. It could explore the strengths and weaknesses of each DL technique in various scenarios of flood forecasting and management.

3. The paper mentions the ethical implications of DL for flood forecasting but does not elaborate on these aspects. Providing examples or case studies to illustrate potential ethical concerns and possible mitigations would have been useful.

4. Although the paper mentions the evaluation criteria used to judge the effectiveness of models, it could have discussed the appropriateness of different metrics in varied scenarios.

5. The paper would have benefited from case studies showing how different DL techniques have been applied in real-world flood forecasting and management scenarios, highlighting successes and improvement areas.

Recommendations:

1. Including case studies would provide a more practical understanding of the usage and performance of DL models in flood forecasting and management.

2. The paper should provide a deeper exploration of ethical concerns and provide suggestions for handling such concerns.

3. A more detailed comparison of DL models in terms of their performance, computational requirements, and suitability for different flooding scenarios could provide a more nuanced understanding of state of the art.

The paper contributes to the literature on deep learning applications in flood forecasting and management. With more depth in areas such as model comparison, ethical discussions, and real-world applications, it could become an even more comprehensive guide for researchers and practitioners in this field. However, the review methodology description is compulsory.

Minor editing of English language required

Author Response

The paper delivers a comprehensive and up-to-date review of deep learning (DL) applications in flood forecasting and management. It efficiently summarizes the current state of the art, critically evaluates existing methods, and provides insights into future research directions. The authors lay a strong foundation for addressing challenges in this field and considering ethical implications, which adds value to the paper. However, particular areas lack depth and require more elaboration.

Strengths:

1. The paper covers many deep learning models, data sources, and evaluation metrics used in flood forecasting and management. This holistic view makes it a valuable resource for researchers in the field.
2. The topic is important given the increasing prevalence of severe flooding due to climate change. The paper provides a timely review of the use of cutting-edge DL techniques to mitigate the effects of this global issue.
3. The authors' insights into future research areas, such as hybrid models, uncertainty estimates, data integration, and interpretability improvement, are well-articulated and forward-looking.
4. The discussion on challenges like data availability and interpretability is valuable. This acknowledges the practical difficulties in deploying DL models for flood forecasting.

Reply: We appreciate the reviewer positive feedback and the recognition of the strengths of our work. We also appreciate your recommendations for improvement, which we have addressed in the replies below.

Drawbacks:

1. The main drawback is the lack of a review methodology. The search strategy, inclusion, and exclusion criteria should be included.
2. While the paper covers a broad spectrum of models, it could benefit from a more detailed examination of each model. It could explore the strengths and weaknesses of each DL technique in various scenarios of flood forecasting and management.
3. The paper mentions the ethical implications of DL for flood forecasting but does not elaborate on these aspects. Providing examples or case studies to illustrate potential ethical concerns and possible mitigations would have been useful.
4. Although the paper mentions the evaluation criteria used to judge the effectiveness of models, it could have discussed the appropriateness of different metrics in varied scenarios.
5. The paper would have benefited from case studies showing how different DL techniques have been applied in real-world flood forecasting and management scenarios, highlighting successes and improvement areas.

Reply: As per the reviewer’s suggestion, the recommendations for the drawbacks have been corrected in the revised manuscript. Page-5, Page-23, Table-1

Recommendations:

1. Including case studies would provide a more practical understanding of the usage and performance of DL models in flood forecasting and management.

Reply: As per the reviewers suggestion, case studies have been now included in the review manuscript. Page-5.

2. The paper should provide a deeper exploration of ethical concerns and provide suggestions for handling such concerns.

Reply: As per the reviewer’s suggestion, deeper explorations of ethical concerns are included in the revised manuscript. Page- 23.

3. A more detailed comparison of DL models in terms of their performance, computational requirements, and suitability for different flooding scenarios could provide a more nuanced understanding of state of the art.

Reply: As per the reviewer’s suggestion, Table 1 is now added to show the detailed comparison of DL models in terms of their performance, computational requirements, and suitability for different flooding scenarios.

The paper contributes to the literature on deep learning applications in flood forecasting and management. With more depth in areas such as model comparison, ethical discussions, and real-world applications, it could become an even more comprehensive guide for researchers and practitioners in this field. However, the review methodology description is compulsory.

Reply: As per the reviewer’s suggestion, all the corrections have been done in revised manuscript.

Minor editing of English language required

Reply: As per the reviewer’s suggestion, the manuscript has been cross checked for English language correction.

Reviewer 3 Report

The authors have presented a State-of-the-Art in Deep Learning Applications, Challenges, and Future Prospects: A Comprehensive Review of Flood Forecasting and Management. The paper tried to review documents in the areas of deep learning in flood forecasting and management. There is a lot of work to be done before the paper can be accepted for publication in this reputable journal. Hence, the authors have to look into the following concerns:

1.      The motivation and contribution of this paper should be stated more clearly in the abstract for a better understanding from the beginning of the study. Authors are advised to be precise in the abstract, and structure your abstract as follows- 1) Background 2) Aim/Objective 3) Methodology 4) Results 5) Conclusion. Write 2-4 lines for each and merge everything in one paragraph (200-300 Words) without any subheading.

2.      The Introduction section should be improved by focusing more on recent works in this area, presenting the background of problems, and works done in recent times. The introduction is too long to move some of the parts to the literature review. Also, the motivation and contribution should be stated more clearly.

3.      The related work section is very small, an updated and complete literature review should be conducted and should appear in section 2- Related Work. Some latest papers which studied similar effects problems can be discussed to help the readers.

4.      The contribution of the paper should be added at the end of the introduction section for better visibility.

5.      The figures presented are not too clear, the authors should please work on their figures for better quality. For example, figure 1 to 5 is not readable.

6.      The resource and performance analysis of deep learning for flood forecasting and management in this paper is deficient and needs to be further strengthened. The analysis of relevant papers in the area reviewed is not complete.

7.      The authors did not show the methods used to select the paper used for the review, I would suggest the authors use the PRISMA methods to show the various papers and journals used in this review

8.      As a survey, the first and second paragraphs have not impressed me. The background for the deep learning enabled for flood forecasting and management should be more detailed and convincing. Some standards, policies, and implements should be discussed. 

9.      It is strange to me that after going through this paper, it is not a survey paper. The main contribution is introducing deep learning in flood forecasting and management. But the main motivation is not highlighted well. Also, the technicality is not well introduced

10. A taxonomy missing of the works in the areas reviewed is not presented

11. The authors are recommended to provide more technical illustrations, and it would be better if they chose performance metrics to compare among the deep learning enabled for flood forecasting and management techniques they mentioned in the literature review. For example, the table within this manuscript includes figures depicting complexity comparisons among the proposed techniques. This would give the paper a more technical look and presentation.

I did not appreciate the style of presentation of this paper, authors need to incorporate the points mentioned above for a better and possible publication in the journal. I, therefore, recommend a major revision.

Extensive editing of the English language required

Author Response

The authors have presented a State-of-the-Art in Deep Learning Applications, Challenges, and Future Prospects: A Comprehensive Review of Flood Forecasting and Management. The paper tried to review documents in the areas of deep learning in flood forecasting and management. There is a lot of work to be done before the paper can be accepted for publication in this reputable journal. Hence, the authors have to look into the following concerns:

1. The motivation and contribution of this paper should be stated more clearly in the abstract for a better understanding from the beginning of the study. Authors are advised to be precise in the abstract, and structure your abstract as follows- 1) Background 2) Aim/Objective 3) Methodology 4) Results 5) Conclusion. Write 2-4 lines for each and merge everything in one paragraph (200-300 Words) without any subheading.

Reply: As per the reviewers suggestion the abstract has been modified in the revised manuscript. Page-1.

2. The Introduction section should be improved by focusing more on recent works in this area, presenting the background of problems, and works done in recent times. The introduction is too long to move some of the parts to the literature review. Also, the motivation and contribution should be stated more clearly.

Reply: The introductory section has been improved, per the reviewer's recommendation, by highlighting new research in flood forecasting and management and revised the background of the problems being addressed.  A separate subsection of literature review has been now added. For context and relevance, recent developments have been emphasized. In a separate section, the paper's contribution has been described more precisely. Page: 2 to 6.

3. The related work section is very small, an updated and complete literature review should be conducted and should appear in section 2- Related Work. Some latest papers which studied similar effects problems can be discussed to help the readers.

Reply: As per the reviewer’s suggestion, a separate subsection of (1.2 literature review) has been now added along with latest papers. Page: 3 to 5.

4. The contribution of the paper should be added at the end of the introduction section for better visibility.

Reply: As per the reviewer’s suggestion, a separate sub-section of the contribution of the paper has been incorporated at the end of the introduction section for better visibility. Page 5 and 6.

5. The figures presented are not too clear, the authors should please work on their figures for better quality. For example, figure 1 to 5 is not readable.

Reply: As suggested by the reviewer, better quality figures, have been replaced in the revised manuscript. Figure 1 to 5.

6. The resource and performance analysis of deep learning for flood forecasting and management in this paper is deficient and needs to be further strengthened.The analysis of relevant papers in the area reviewed is not complete.

Reply: As per the reviewer’s suggestion, Table 2 is now added to show the detailed comparison of Different DL models including accuracy, RMSE to further strengthened the paper.

7. The authors did not show the methods used to select the paper used for the review, I would suggest the authors use the PRISMAmethods to show the various papers and journals used in this review

Reply: As per the suggestion, PRISMA method has been included in the revised manuscript. Figure 6.

8. As a survey, the first and second paragraphs have not impressed me. The background for the deep learning enabled for flood forecasting and management should be more detailed and convincing. Some standards, policies, and implements should be discussed.

Reply: As per the suggestion, the first and second paragraphs have been improved. Page-1 and 2. The background of the deep learning has been improved. Page-3.

9. It is strange to me that after going through this paper, it is not a survey paper. The main contribution is introducing deep learning in flood forecasting and management. But the main motivation is not highlighted well. Also, the technicality is not well introduced

Reply: As per the suggestion, the abstract, the introduction, literature review, motivation, contribution of the paper has been improved. Page-1 to 6.

10. A taxonomy missingof the works in the areas reviewed is not presented

Reply: As per the suggestion, PRISMA method has been included in the revised manuscript. Figure 6.

11. The authors are recommended to provide more technical illustrations, and it would be better if they chose performance metrics to compare among the deep learning enabled for flood forecasting and management techniques they mentioned in the literature review. For example, the table within this manuscript includes figures depicting complexity comparisons among the proposed techniques. This would give the paper a more technical look and presentation.

Reply: As per the reviewer’s suggestion, Table 2 is now added to show the detailed comparison of different DL models including accuracy, RMSE, advantage and disadvantage to further strengthened the paper.

I did not appreciate the style of presentation of this paper; authors need to incorporate the points mentioned above for a better and possible publication in the journal. I, therefore, recommend a major revision.

Extensive editing of the English language required

Reply: As per the suggestion the manuscript has been modified in the revised manuscript and English editing has been done.

Round 2

Reviewer 1 Report

The authors have addressed my comments. Suggest them to address the following. 

In table 2 description. the term RMSC is confusing with RMSE

Fine

Author Response

The authors have addressed my comments. Suggest them to address the following. 

In table 2 description. the term RMSC is confusing with RMSE

Reply: Thank you for your valuable feedback.

As suggested by the reviewer the correction has been done. Page 15 and 17.

Reviewer 2 Report

Thanks for the authors for considering reviewer's comments and recommendations. In my opinion, now the paper can be accepted.

Author Response

Thanks for the authors for considering reviewer's comments and recommendations. In my opinion, now the paper can be accepted.

Reply: The authors are thankful to the reviewer for accepting the manuscript.

Reviewer 3 Report

The authors significantly improved the paper, and almost all my concerns were addressed.

But authors still need to discuss their findings in the discussion section this will help the readers that want to do research in this area.

Authors should read all through again to correct grammatical and spelling mistakes.

Moderate editing of the English language required

Author Response

The authors significantly improved the paper, and almost all my concerns were addressed.

But authors still need to discuss their findings in the discussion section this will help the readers that want to do research in this area.

Reply: As per the suggestion, the findings are added in discussion. Page-24.

Authors should read all through again to correct grammatical and spelling mistakes.

Reply: Thank you for your suggestion. We have carefully review the paper to address any grammatical and spelling mistakes.