A Data-Driven Multi-Step Flood Inundation Forecast System

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper entitled “ A data-driven multi-step inundation forecast system” by Schmid and Leandro deals with a data-driven multi-step inundation forecast system for fluvial flood events with a forecast system  based on a convolutional neural network (CNN).

The paper is well written and structured, and it presents a main topic for pluvial inundation forecasting in real-time using data-driven modelling approach, instead a physically-based approaches which are affected by different sources of uncertainty, and they require longer computation time.

Figures and Tables are generally good, the introduction is robust with discussion and conclusions which that reasonable describe the main findings.

However, I hereafter highlight some minor comments.

 

Title: I should introduce something related to water level map, in order to better highlight the topic of this research, otherwise, it is difficult to get the application.

P2, L57: what do you mean with syntactic return period?

P8, L271: What is the ReLu function? Please, introduce it or add some references.

P8, L272: I suggest “commonly” instead of “comely”.

P8, L274: “be described”

P8, L287: What is the “falltend” layer?

P8, L288: “be interpreted”.

P8, L289: Please, revise this sentence.

P9, L299-300: Please, revise this sentence.

P9, L319: I suggest; “…but it make the training on…”

Figure 1: What are NN in the DEM legend?

P11, L398: Physically-based…please pay attention to how it is written in the whole text

P12, L419-420: Can you describe the difference between the validation and the test phase in you study?

P12, L420: discaged?

P12, L421: Please, revise this sentence.

P12, L428: Is there any reason for this value of discharge? Please, motivate it.

Table 3: It would be interesting to see even the scores for each lead time, beside the average value of them. This would also justify what it is written in section 4.1. I understand that better scores are for hour lead time, and this is very important in real-time applications, but not always. Sometimes, you need 3-6 hours or more. Hence, if better highlight the performance at different lead-times would be much better.

P17, L566: In calculating the CSI skill score, you selected a threshold equal to 0.05 m. Please, can you give why it is a significant threshold for you? Explain a little better this sentence.

Comments on the Quality of English Language

Just some comments above highlighted.

Author Response

Please see attachment

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

This paper develops a data-driven multi-step inundation forecast system using a convolutional neural network  with recursive connections to predict future flood events. The proposed model is validated against physically-based models, demonstrating its potential for accurate and efficient real-time flood forecasting.

The paper is generally good but still has some room for improvement.

(1)The introduction could be improved by including a more detailed review of recent advances in data-driven flood forecasting models. 

(2)A more detailed explanation of the model structure and choice of hyperparameters could have been provided.For example, in Figure 1, providing more detail about each step in the workflow would make it easier for readers to follow.

(3)More detail on how they were optimised or selected would have been helpful. For example, how were emission thresholds determined and what is the rationale?

(4)The methodology section is too brief in its description and lacks clarity in the details of the key steps and model architecture, especially the specific operational steps of the data preprocessing and model training processes.

(5)The discussion and analysis of the results in the experimental section are not deep enough, lacking detailed comparison and discussion of different experimental results. Whether the model still has good applicability over longer time series or even different study areas.

(6)It would also be helpful to consider including a sensitivity analysis to show how changes in the input data affect the performance of the model. This will help to provide insights into the robustness of the model.

(7)The discussion could also be extended to illustrate potential limitations in more detail. For example, discuss the impact of data quality and availability on model performance.

Comments on the Quality of English Language

As a whole the quality of English language in this paper is good. The sentences are clear and coherent, effectively conveying the technical content. However, there are a few areas that could be improved to enhance readability and ensure precision:

(1) Avoid mixing past and present tenses in the introduction, as this can cause confusion.

(2) Pay attention to subject-verb agreement, especially in complex sentences. 

(3) Avoid using redundant phrases. For instance, "in order to" can be simplified to "to," making the text more concise without changing its meaning.

(4) Proper use of commas, especially in lists and complex sentences, can improve readability. 

(5) Be mindful of typographical errors, as they can detract from the professionalism of the paper. A thorough proofread is recommended to catch any typos.

Overall, the paper is well-written and the technical content is effectively communicated. Addressing the above points will further enhance the clarity and professionalism of the manuscript.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

Paper "A data-driven multi-step inundation forecast system" presents an interdisciplinary approach to solving problems in hydrology.

 

Several suggestions will be emphasized:

 

-        Exclude keywords that are the same words from the title.

-        In the Introduction, it is necessary to add short comments for certain references that are only listed in sequence.

-        Also, it is necessary to add a larger overview of references related to other methodologies for solving similar problems as well as references related to the applied methodology, which contributed to some other fields of science.

-        The methodology is presented quite vaguely. There are a large number of references that are only listed, without discussion. Figures and diagrams are technically not so neat; also, the titles of the figures are too long and as such are not acceptable - put clear titles, and transfer the comments and discussion to the text of the paper.

-        In general, as for the algorithm, it is suggested to make it unique and not from multiple parts.

-        Authors are requested to check the equations; I believe that the research area should be formed as a separate chapter, and not in the methodology chapter, because it is not a scientific method. likewise, add the meaning of the parameters in some.

-        In the description of the research area, add details related to geological units, as well as a description of the aquifers & aquitards that occur in that area.

-        The Results chapter should be titled Results and Discussion. On the other hand, it is necessary to comment in more detail on the obtained results, especially the diagrams.

-        The conclusion is not adequate. Here, it is necessary to specifically state what is novelity in research, as well as a specific scientific contribution, and also a contribution to the academic community and application in practice.

 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The author have made an effort to follow the suggestions and improve the quality of the scientific article. Please read the first report carefully. A few more suggestions will be added.

 

-        Some figure titles are too long. A clear title should be given here, and detailed explanations should be transferred to the text of the paper. In the previous report I mentioned this.

-        Chapter 3 is too short, consisting of only a few sentences. In the previous report I mentioned that it would be good to describe geological or hydrogeological units.

-        I believe that the results and the discussion should be combined in one chapter. In this case, some facts are not so clear, and some are repeated.

-        In addition to the conclusion, which is like a summary, add a concrete scientific contribution of the conducted research.

Author Response

Please see the attachment

Author Response File: Author Response.docx