Round 1

Reviewer 1 Report

GENERAL COMMENTS

This study delves into an intriguing examination of hydrodynamic behavior, utilizing a newly introduced Machine Learning (ML) model termed the Expanded Framework of Group Method of Data Handling (EFGMDH). While the study's subject matter holds promise, it is imperative to acknowledge several critical issues that warrant careful attention to ensure the study's clarity, robustness, and ultimate relevance within the field. Addressing these concerns will be pivotal in maximizing the potential impact and contribution of the study.

The manuscript exhibits a sound scientific foundation, yet some sentences require refinement for improved clarity and syntax.

The literature review could be strengthened by offering a more comprehensive survey of pertinent studies that bolster the motivations and contextual background of the research. For instance, the successful application of machine learning techniques in developing time-varying cross-section rating curves. Please see suggested reference at the end.

The use of a specific numerical model (HEC-RAS) for generating the comprehensive dataset of river flow discharge is mentioned. However, further clarification regarding the choice of this numerical model (for example, why not LISFLOOD-FP or FLO-2d) and its appropriateness for the study's objectives is recommended.

Regarding the Conclusions section, while the site-specific conclusions drawn from the study's findings are valuable, it is essential to distill these into general lessons that underscore the broader applicability of the EFGMDH method or the successful representation of hydraulic behavior. This will provide readers with insights into the wider implications of the study's outcomes.



SPECIFIC COMMENTS

**Location:** Indeed, a comprehensive dataset of river flow discharge has been generated using a numerical model, including a wide range of potential future floods. This significantly improves the ML training process to generalize the accuracy of results.
**Comment:** Both sentences should be consistently in the simple past tense to maintain coherence.

**Location 1:** The importance of river depth concerning floods can be understood through the following points: (i) Flood risk assessment [13,14], (ii) Hydraulic capacity, (iii) Floodplain mapping [15], Infrastructure design [16], (v) Emergency response planning.
**Location 2:** Here are some reasons why velocity is essential in understanding floods: (1) Flow dynamics [17], (2) Flood hazard mapping [18], (3) Sediment Transport [19], (4) Hydraulic engineering design [20], and (5) Flood modeling and forecasting [21].
**Comment:** While you've listed relevant categories, it's important to elaborate on each point. This will provide a more comprehensive literature review to support the motivations and background of your study.

**Location:** The "flow velocity" during different floods
**Comment:** The use of double quotes is unnecessary here.

**Location:** A comprehensive dataset of river flow discharge will be generated using the numerical model,
**Comment:** Clarify which numerical model will be employed for generating the dataset.

**Location:** Table 1
**Comment:** Since the standard deviation is already provided, the inclusion of the variance (SV column) seems redundant.

**Location:** Equations (15) - (17)
**Comment:** Consider reordering the equations, placing Equation (17) after (15) for better logical flow.
**Comment:** The definitions of Z and E should include subscript "i".

**Location:** The descriptive performance of the R2, NSE, and NRMSE are provided in Table 2.
**Comment:** A more clear description is "The model efficiency characterization based on R2, NSE, and NRMSE intervals is provided in Table 2."

**Location:** Table 2
**Comment:** The criteria upon which the performance intervals of this table are based should be explained.

**Location:** Figure 3
**Comment:** Please clarify if M1 to M8 correspond to models 1 to 8.
**Comment:** Avoid using brackets in the definition of M1 to M8 as it may be confused with a matrix representation.


REFERENCES

Rozos, E.; Leandro, J.; Koutsoyiannis, D. Development of Rating Curves: Machine Learning vs. Statistical Methods. Hydrology 2022, 9. https://doi.org/10.3390/hydrology9100166.

Xu G, Fan H, Oliver DM, Dai Y, Li H, Shi Y, Long H, Xiong K, Zhao Z. Decoding river pollution trends and their landscape determinants in an ecologically fragile karst basin using a machine learning model. Environmental Research. 2022 Nov 1;214:113843

Moderate editing of English language is required.

Author Response

Reviewer #1:

This study delves into an intriguing examination of hydrodynamic behavior, utilizing a newly introduced Machine Learning (ML) model termed the Expanded Framework of Group Method of Data Handling (EFGMDH). While the study's subject matter holds promise, it is imperative to acknowledge several critical issues that warrant careful attention to ensure the study's clarity, robustness, and ultimate relevance within the field. Addressing these concerns will be pivotal in maximizing the potential impact and contribution of the study.

Response: Thanks. We sincerely thank the reviewer for their valuable insights and constructive feedback on our manuscript. We greatly appreciate their time and effort in evaluating our work. We are committed to addressing all the recommended revisions and concerns raised in their review. These suggestions will undoubtedly enhance our study's clarity, robustness, and overall relevance within the field. Your guidance will be instrumental in maximizing the impact and value of our research contribution.

---------------------------------------------------------------------------------------------------------------------

Comment #1

The manuscript exhibits a sound scientific foundation, yet some sentences require refinement for improved clarity and syntax.

Response: Thank you. The revision that was requested has been done accordingly. (Please see manuscript)

---------------------------------------------------------------------------------------------------------------------

Comment #2

The literature review could be strengthened by offering a more comprehensive survey of pertinent studies that bolster the motivations and contextual background of the research. For instance, the successful application of machine learning techniques in developing time-varying cross-section rating curves. Please see suggested reference at the end.

Response: Your input regarding this suggestion is greatly appreciated. As requested, the revision has been made. (Please see references & P. 3)

---------------------------------------------------------------------------------------------------------------------

Comment #3

The use of a specific numerical model (HEC-RAS) for generating the comprehensive dataset of river flow discharge is mentioned. However, further clarification regarding the choice of this numerical model (for example, why not LISFLOOD-FP or FLO-2d) and its appropriateness for the study's objectives is recommended.

Response: Thank you, the requested clarification has been accomplished. (Please see P. 3)

---------------------------------------------------------------------------------------------------------------------

Comment #4

Regarding the Conclusions section, while the site-specific conclusions drawn from the study's findings are valuable, it is essential to distill these into general lessons that underscore the broader applicability of the EFGMDH method or the successful representation of hydraulic behavior. This will provide readers with insights into the wider implications of the study's outcomes.

Response: Thank you for your comment. The requested revision has been done. (Please see Conclusions)

---------------------------------------------------------------------------------------------------------------------

Comment #5

SPECIFIC COMMENTS

**Location:** Indeed, a comprehensive dataset of river flow discharge has been generated using a numerical model, including a wide range of potential future floods. This significantly improves the ML training process to generalize the accuracy of results.

**Comment:** Both sentences should be consistently in the simple past tense to maintain coherence.

Response: Thank you. It was done. (Please see Abstract)

---------------------------------------------------------------------------------------------------------------------

Comment #6

**Location 1:** The importance of river depth concerning floods can be understood through the following points: (i) Flood risk assessment [13,14], (ii) Hydraulic capacity, (iii) Floodplain mapping [15], Infrastructure design [16], (v) Emergency response planning.

**Location 2:** Here are some reasons why velocity is essential in understanding floods: (1) Flow dynamics [17], (2) Flood hazard mapping [18], (3) Sediment Transport [19], (4) Hydraulic engineering design [20], and (5) Flood modeling and forecasting [21].

**Comment:** While you've listed relevant categories, it's important to elaborate on each point. This will provide a more comprehensive literature review to support the motivations and background of your study.

Response: Thanks for your valuable comments. The requested revisions have been made accordingly. (Please see P. 2-3)

---------------------------------------------------------------------------------------------------------------------

Comment #7

**Location:** The "flow velocity" during different floods

**Comment:** The use of double quotes is unnecessary here.

Response: Thanks. It was modified. (Please see P. 3)

---------------------------------------------------------------------------------------------------------------------

Comment #7

**Location:** A comprehensive dataset of river flow discharge will be generated using the numerical model,

**Comment:** Clarify which numerical model will be employed for generating the dataset.

Response: Thanks. The requested revision was done accordingly. (Please see P. 4)

---------------------------------------------------------------------------------------------------------------------

Comment #9

**Location:** Table 1

**Comment:** Since the standard deviation is already provided, the inclusion of the variance (SV column) seems redundant.

Response: Thank you. Based on your comment, the variance has been removed from Table 1. (Please see Table 1)

---------------------------------------------------------------------------------------------------------------------

Comment #10

**Location:** Equations (15) - (17)

**Comment:** Consider reordering the equations, placing Equation (17) after (15) for better logical flow.

**Comment:** The definitions of Z and E should include subscript "i".

Response: Thank you. Eq. (17) moved after Eq. (15), and subscripts were added to both Z and E. (Please see Eqs. (16) & Eq. (17))

---------------------------------------------------------------------------------------------------------------------

Comment #11

**Location:** The descriptive performance of the R2, NSE, and NRMSE are provided in Table 2.

**Comment:** A more clear description is "The model efficiency characterization based on R2, NSE, and NRMSE intervals is provided in Table 2."

Response: Thank you for your recommendation. It was modified based on your suggestion. (Please see P. 10)

---------------------------------------------------------------------------------------------------------------------

Comment #12

**Location:** Table 2

**Comment:** The criteria upon which the performance intervals of this table are based should be explained.

Response: Thanks. The reference for these intervals has been included in the manuscript. (Please see P. 10)

---------------------------------------------------------------------------------------------------------------------

Comment #13

**Location:** Figure 3

**Comment:** Please clarify if M1 to M8 correspond to models 1 to 8.

**Comment:** Avoid using brackets in the definition of M1 to M8 as it may be confused with a matrix representation.

Response: Thank you. Accordingly, revisions were made. (Please see Figure 3)

---------------------------------------------------------------------------------------------------------------------

REFERENCES

Rozos, E.; Leandro, J.; Koutsoyiannis, D. Development of Rating Curves: Machine Learning vs. Statistical Methods. Hydrology 2022, 9. https://doi.org/10.3390/hydrology9100166.

Xu G, Fan H, Oliver DM, Dai Y, Li H, Shi Y, Long H, Xiong K, Zhao Z. Decoding river pollution trends and their landscape determinants in an ecologically fragile karst basin using a machine learning model. Environmental Research. 2022 Nov 1;214:113843

Response: We value your contribution in suggesting valuable references. Both of these references are relevant and have been cited within the text. (Please see References)

---------------------------------------------------------------------------------------------------------------------

 

Reviewer 2 Report

Flood event monitoring is important for disaster control and evaluation under climate change. This study combines a numerical simulation with a machine learning technique to predict flood event regarding floodplain width, flow velocity, and river flow depth. The performance of this model framework has been well demonstrated in the Ottawa River Watershed. This paper has well writing to describe the purpose of this study, and provided detailed information regarding the design of the experiment. Therefore, I suggest it be publishable after a few minor revisions.

 

(1) This study was designed and has been well applied in the Ottawa River Watershed. But it would be better to discuss the approach is extendable and application in other watersheds.

(2) Lines 95-96, 102: the abbreviation of GMDH appears several times. Please make it more concise to describe the method.

(3) Line 161 and Table 1: what are the nLeft, nMiddle, nRight? Does the slope represent the topography, which is derived from DEM?

(4) What is Ln in Eq. (14)?

(5) Line 383: the title of subsection 2.5 should be improved. Methodology has a broad definition.

(6) What are the variables in Figure 3: X, Y, n_L, n_M, n_R? Please give a description in the figure caption.

(7) In conclusion, the information in the first paragraph looks copied from the introduction/abstract. Please improve the discussion to make it concise.  

Ths quality of English writting is good, but the paper requires improvement to make it more concise. 

Author Response

Reviewer #2:

Flood event monitoring is important for disaster control and evaluation under climate change. This study combines a numerical simulation with a machine learning technique to predict flood event regarding floodplain width, flow velocity, and river flow depth. The performance of this model framework has been well demonstrated in the Ottawa River Watershed. This paper has well writing to describe the purpose of this study, and provided detailed information regarding the design of the experiment. Therefore, I suggest it be publishable after a few minor revisions.

Response: We sincerely appreciate your positive review and insightful feedback on our manuscript. We are dedicated to promptly addressing your feedback to ensure that the manuscript aligns with your comments.

---------------------------------------------------------------------------------------------------------------------

Comment #1

This study was designed and has been well applied in the Ottawa River Watershed. But it would be better to discuss the approach is extendable and application in other watersheds.

Response: Thanks. The requested revision has been done. (Please see the last paragraph of the Conclusion)

---------------------------------------------------------------------------------------------------------------------

Comment #2

Lines 95-96, 102: the abbreviation of GMDH appears several times. Please make it more concise to describe the method.

Response: Thank you for your comment. Accordingly, revisions were made. (Please see P. 3)

---------------------------------------------------------------------------------------------------------------------

Comment #3

Line 161 and Table 1: what are the nLeft, nMiddle, nRight? Does the slope represent the topography, which is derived from DEM?

Response: Thank you. The manuscript has been updated to include the requested clarification regarding these parameters. The nLeft, nMiddle, and nRight are Manning's roughness coefficients at the left, middle, and right sides of the channel at each cross-section. In each zone, the slope of the riverbed was derived from the DEM. (Please see P. 5-6)

---------------------------------------------------------------------------------------------------------------------

Comment #4

What is Ln in Eq. (14)?

Response: Thanks. The “Ln” is the natural logarithm. The definition of “Ln” was added in the manuscript (Please see P. 9)

---------------------------------------------------------------------------------------------------------------------

Comment #5

Line 383: the title of subsection 2.5 should be improved. Methodology has a broad definition.

Response: Thank you for your comment. It was changed to “The framework for estimating the hydrodynamic behavior of the river.” (Please see section 2.5)

---------------------------------------------------------------------------------------------------------------------

Comment #6

What are the variables in Figure 3: X, Y, n_L, n_M, n_R? Please give a description in the figure caption.

Response: Thanks. It was done accordingly. (Please see the caption of Figure 3)

---------------------------------------------------------------------------------------------------------------------

Comment #7

In conclusion, the information in the first paragraph looks copied from the introduction/abstract. Please improve the discussion to make it concise.  

Response: Thanks. The first paragraph of the Conclusions has been modified based on your comment. (Please see Conclusion)

---------------------------------------------------------------------------------------------------------------------

Round 2

Reviewer 1 Report

Minor editing of English language required

The authors have addressed successfully all my comments/suggestions.