the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 15 Apr 2024
Exploring the driving factors of compound flood severity in coastal cities: a comprehensive analytical approach
Abstract. Coastal cities frequently face various types of flooding triggered by heavy rainfall and storm surges, such as fluvial flooding and pluvial flooding. Currently, Currently, there is a lack of comprehensive methods to analyse the sources of severe compound flooding. This study, using the Shahe River Basin in Guangzhou, China as an example, establishes and validates a coupled 1D and 2D hydrodynamic model. Based on historical data, it constructs joint probability distributions of rainfall and tidal levels with different return periods and durations. Using the results from the coupled model under various design scenarios, it proposes an impact index to quantify the contributions of rainfall and tides to flooding. Furthermore, it quantifies the interactions between fluvial flooding and pluvial flooding. Flood-prone areas are delineated, and the causes of flooding are analyzed. The results show that when the return periods of rainfall and tide level are both 10 years, the Kendall return period for the combined event of rainfall and tide level is 36.35 years, greater than the “Or” return period (5.40 years) and less than the “And” return period (66.88 years). The impact degree index of rainfall on flooding varies between 0.5 and 1, with the minimum at 24-hour duration, indicating that the study area is primarily affected by rainfall and the influence of tide level is most significant at 24-hour duration. The pluvial flooding caused by the influence of river water level on the drainage outlet accounts for 19.08 % of the total volume at most. This shows that fluvial flooding affects the seriousness of pluvial flooding by influencing the water levels of outlets. The flood-prone area is divided into different regions based on the main natural factors (rainfall and tidal level) and social factors (pipeline network, drainage outlets, and riverbank defenses) to help decision-makers identify the causes of flooding in each drainage unit and better formulate targeted disaster reduction strategies to improve flood control capabilities.
- Preprint
(1806 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on hess-2024-100', Anonymous Referee #1, 06 May 2024
In this study, the authors conducted a comprehensive investigation of the driving factors behind compound flooding in coastal cities using a combination of hydrodynamic modeling and mathematical statistics. Their research yielded insights into the impact of rainfall and tidal levels on compound flooding, as well as the contributions of different types of floods to compound flooding. The conclusions of this study are novel and have positive implications for risk management. The chosen topic also aligns with the scope of the journal "Hydrology and Earth System Sciences." , I have several suggestions and recommendations that would help enhance the manuscript.
Specific comments
- It is necessary to indicate whether there are tide prevention facilities in the study area and how this factor is considered in the model.
- In the "Model construction and validation" section, a one-dimensional and two-dimensional hydrodynamic coupling model was constructed, but the analysis of flood severity primarily relied on flood volume, with limited analysis on indicators such as flood area and flood depth. This aspect needs to be supplemented.
- The paper only mentioned the length of the original data but did not elaborate on the sample data situation for constructing the Copula function.
- The paper is lengthy, and to be concise, compressing the research methods section would be helpful. For example, common formulas in Correlation and Copula.
- In the section "Spatial interaction of drainage units," the analysis of the interaction forces between different drainage units is not highly relevant to the main theme of this paper.
- In section 4.4 "Causes and prevention measures of floods in drainage units," flood prevention measures should not be discussed in the research results. It is suggested to elaborate on them in the discussion section.
- The conclusion needs further refinement.
Technical comments
L11: The word "Currently" is repeated.
L31: Change "in this year" to a specific year.
Figure 1: Mark all drainage outlets in the figure.
Table 1: Explain why the RMSE of the edge distribution function corresponding to the optimal tidal level for 3h is not the best.
L430-431: What is the specific relationship? It needs to be clarified.
L457-459: Drainage unit 14 appears in both cases simultaneously, please verify.
L460-461: Same as above.
L508-509: This is an observation, not a conclusion.
Figure 14: The discussion of the causes of sudden changes is crucial.
L520-522: This should be in the Methods section.
Citation: https://doi.org/10.5194/hess-2024-100-RC1 - AC1: 'Reply on RC1', Ting Zhang, 04 Jun 2024
-
AC2: 'Reply on RC1', Ting Zhang, 04 Jun 2024
Dear Reviewer,
Thank you again for your comments. I am writing to apologize for the inadvertent inclusion of Chinese text in our last response to your review comments. Due to a browser error, our intended English text was partially replaced by Chinese characters, which might have caused confusion.
The correct expression is: “Please find attached a PDF responding to the comments of Reviewer 1.” The attachment file in the last reply is correct.
Thank you for your patience and support.
Best regards,
Ting Zhang
Citation: https://doi.org/10.5194/hess-2024-100-AC2
-
RC2: 'Comment on hess-2024-100', Anonymous Referee #2, 25 Jul 2024
In this study, the authors thoroughly explored the factors driving compound flooding in an urban area by combining hydrodynamic modeling and multivariate statistics. Their research provided insights into the analysis of the relative contribution of rainfall and tidal levels on compound flooding and the roles of different flood scenarios. The study is fairly novel and findings are beneficial for flood risk management and applications. Further, the scope and findings of the study align with the scope of the journal "Hydrology and Earth System Sciences,” and I recommend its publication but after some revisions. Below, the authors can find my suggestions and recommendations, which I believe would contribute to improving the manuscript and clarifying the methodology.
Major Comments:
- Copulas and marginal distributions are fitted to the “tidal levels.” Is this just the astronomical tide, or do these values include wind-driven surge (in the case of Typhoon/Cyclone conditions) and river discharge components as well? The authors mention “storm surge” as a critical factor for compound flooding several times (L346, L456, L564), but it’s not clear how storm surge is taken into account when modeling “tidal levels” in this study. If the tidal range is higher and the storm surge component is relatively very small (or if only the astronomical tide is considered), fitting an extreme value model might be fundamentally incorrect since the astronomical tide is deterministic. Therefore, a clear explanation of the terminology is needed. Furthermore, are the tidal boundary conditions used at the sluice gate (as outflow boundary conditions?)? The authors mention a tidal gate, but the location is not shown in the figure.
- When selecting the sample for joint probability analysis, the tidal level is chosen as the "highest tidal level of the corresponding day." This method implies that the calculated tidal design levels could occur within a 24-hour period from the rainfall event. In the numerical modeling process, a single event is used as a representative for the tidal levels process, but the timing of the rainfall peak and tidal peaks in generating design scenarios is not explained. Additionally, figure 8(a) shows the peak rainfall and peak tide almost coinciding, which is rare in reality (not the most probable 200-year event). The 3-hour rainfall duration shows the peak rainfall occurring approximately half an hour after the peak tide, whereas the 12-hour rainfall duration shows the peak rainfall occurring approximately 3 hours before the peak tide. How are the corresponding time lags between peak rainfall and peak tide determined for each scenario? Are they coming from a probabilistic basis? This timing significantly impacts the resultant flood volume and might be the main reason for the variations discussed around L430.
- 41 and 42 are used to evaluate the spatial interactions between drainage units (DUs). The equations only rely on the masses and distances of two drainage units. To clarify the conclusions around L445-L450, it is essential to understand what the spatial interaction force represents. For example, the spatial interaction force between DU22 and DU6 (which are far away from each other) is 0.75. Does this relatively higher value suggest a higher level of interconnectedness or influence on each other, than even adjacent drainage units? Please explain. Also, it is not clear how the spatial interaction of the drainage units relates to the main objectives of the study.
- It is necessary to provide additional details on the types of data used in the study. For instance, specify whether rainfall data is gridded or gauge data, and if gauge data is used, indicate the number of gauges involved. What are the locations of the tidal data and rainfall used? If so, do you assume a uniformly distributed rainfall over the entire catchment?
- The paper lacks a proper discussion about the main assumptions and limitations of the process. For instance, using only 16 years of data to estimate a 200-year design event introduces significant uncertainty in modeling the dependence structure and the tail distributions. Since the study proposes a "universal" method, it is crucial for readers to understand these assumptions and uncertainties, especially when the method is to be applied to other study sites.
Minor Comments
- L11 “currently” repeats.
- In the abstract, the authors mention the Kendall return period for the combined event of rainfall and tidal levels, greater than the “Or” return period and less than the “And” return period. The Kendall return period is typically expected to fall between the "Or" and "And" scenarios for combined events. It's not a significant finding of the study to be mentioned in the abstract.
- L32, and L36, could you provide references?
- Figure 2: what represents the purple arrow from correlation analysis to “effects of social factors on flood”?
- L61, in underdeveloped areas, fluvial flooding may indeed be more prevalent due to the natural terrain and lack of infrastructure to mitigate such events. However, it is not accurate to say that "only" fluvial flooding exists in these areas. Could you explain or provide references?
- Provide references for the “Pilgrim & Cordery rainfall model”, “Maximum Possible Weighting Function” and “Pilgrim & Cordery rainfall model”
- Figure 6 seems incorrect. The theoretical distributions (GEV, Gamma) should be smooth unless you don’t change the distribution parameters.
- What types of information does Figure 7 (b) provide? What are the insights we can gain into joint distribution probabilities? The color scale is also not given.
- Although the peak tidal levels could be negative (depending on the datum used), the gamma distribution is selected to fit the peak tidal levels for some rainfall durations. The gamma distribution is lower bounded by zero thus making it impossible to sample negative realizations from the fitted distribution (if necessary).
- The extreme sample selection is not well explained. Does it consider annual maxima? How to ensure the events are independent? What is the sample size?
- Figure 10: The study area is in the legend (red). But cannot be seen in the figure.
- Figure 12: Do these different colors represent any information? If not use a single color.
- It's not clear how this observation is made “Despite differences in rainfall peak timing, rainfall, and tide peak timings leading to variations in calculated flooding volumes, the impact degree index is not affected by these differences”?
- It's not clear how this conclusion is made “This study also highlights that under the same recurrence interval, rainfall events with larger peak timings are more destructive than those with earlier peak timings”? Have you run many events for the same return period by only changing the peak timing?
- The authors suggest a set of equations (from eq. 30 to eq. 40), but poorly explained. Consider elaborating more about the governing process of equations. Additionally, it's mentioned that “calculate the range of rainfall and tidal level design values for different durations from 2-yr RP to 200-yr RP using the following formula” (L286,), but the formula doesn’t give the range of rainfall, and since it divides the range by 200yr Rainfall value. So, ∆Xt will be a dimensionless parameter that is related to the amount of variability (range) of rainfall. Same for Tides. Same in L304, ∆Vxt and ∆Vyt do not quantify the variation of flooding volume. Accordingly, check Dxt and Dyt.
- The paragraph from L543 to 558 is more about discussing the results. Consider moving it into the discussion section.
- The paper is lengthy. Consider moving some of the extra results to the supplementary materials. For example, Tables 1 and 2. Commonly used equations, such as those for correlation and distributions, can be omitted from the text. Instead, it is sufficient to cite the relevant references for these well-known equations.
- Check the way of citing publications in the manuscript according to the HESS guideline (e.g., L40, L42, L50).
Citation: https://doi.org/10.5194/hess-2024-100-RC2 -
AC3: 'Reply on RC2', Ting Zhang, 09 Aug 2024
We are sincerely grateful to the editor and reviewers for their valuable time and diligent consideration in reviewing our manuscript. We have carefully addressed each comment to improve the quality of our manuscript. Please find attached a PDF responding to the comments of Reviewer 2.
Status: closed
-
RC1: 'Comment on hess-2024-100', Anonymous Referee #1, 06 May 2024
In this study, the authors conducted a comprehensive investigation of the driving factors behind compound flooding in coastal cities using a combination of hydrodynamic modeling and mathematical statistics. Their research yielded insights into the impact of rainfall and tidal levels on compound flooding, as well as the contributions of different types of floods to compound flooding. The conclusions of this study are novel and have positive implications for risk management. The chosen topic also aligns with the scope of the journal "Hydrology and Earth System Sciences." , I have several suggestions and recommendations that would help enhance the manuscript.
Specific comments
- It is necessary to indicate whether there are tide prevention facilities in the study area and how this factor is considered in the model.
- In the "Model construction and validation" section, a one-dimensional and two-dimensional hydrodynamic coupling model was constructed, but the analysis of flood severity primarily relied on flood volume, with limited analysis on indicators such as flood area and flood depth. This aspect needs to be supplemented.
- The paper only mentioned the length of the original data but did not elaborate on the sample data situation for constructing the Copula function.
- The paper is lengthy, and to be concise, compressing the research methods section would be helpful. For example, common formulas in Correlation and Copula.
- In the section "Spatial interaction of drainage units," the analysis of the interaction forces between different drainage units is not highly relevant to the main theme of this paper.
- In section 4.4 "Causes and prevention measures of floods in drainage units," flood prevention measures should not be discussed in the research results. It is suggested to elaborate on them in the discussion section.
- The conclusion needs further refinement.
Technical comments
L11: The word "Currently" is repeated.
L31: Change "in this year" to a specific year.
Figure 1: Mark all drainage outlets in the figure.
Table 1: Explain why the RMSE of the edge distribution function corresponding to the optimal tidal level for 3h is not the best.
L430-431: What is the specific relationship? It needs to be clarified.
L457-459: Drainage unit 14 appears in both cases simultaneously, please verify.
L460-461: Same as above.
L508-509: This is an observation, not a conclusion.
Figure 14: The discussion of the causes of sudden changes is crucial.
L520-522: This should be in the Methods section.
Citation: https://doi.org/10.5194/hess-2024-100-RC1 - AC1: 'Reply on RC1', Ting Zhang, 04 Jun 2024
-
AC2: 'Reply on RC1', Ting Zhang, 04 Jun 2024
Dear Reviewer,
Thank you again for your comments. I am writing to apologize for the inadvertent inclusion of Chinese text in our last response to your review comments. Due to a browser error, our intended English text was partially replaced by Chinese characters, which might have caused confusion.
The correct expression is: “Please find attached a PDF responding to the comments of Reviewer 1.” The attachment file in the last reply is correct.
Thank you for your patience and support.
Best regards,
Ting Zhang
Citation: https://doi.org/10.5194/hess-2024-100-AC2
-
RC2: 'Comment on hess-2024-100', Anonymous Referee #2, 25 Jul 2024
In this study, the authors thoroughly explored the factors driving compound flooding in an urban area by combining hydrodynamic modeling and multivariate statistics. Their research provided insights into the analysis of the relative contribution of rainfall and tidal levels on compound flooding and the roles of different flood scenarios. The study is fairly novel and findings are beneficial for flood risk management and applications. Further, the scope and findings of the study align with the scope of the journal "Hydrology and Earth System Sciences,” and I recommend its publication but after some revisions. Below, the authors can find my suggestions and recommendations, which I believe would contribute to improving the manuscript and clarifying the methodology.
Major Comments:
- Copulas and marginal distributions are fitted to the “tidal levels.” Is this just the astronomical tide, or do these values include wind-driven surge (in the case of Typhoon/Cyclone conditions) and river discharge components as well? The authors mention “storm surge” as a critical factor for compound flooding several times (L346, L456, L564), but it’s not clear how storm surge is taken into account when modeling “tidal levels” in this study. If the tidal range is higher and the storm surge component is relatively very small (or if only the astronomical tide is considered), fitting an extreme value model might be fundamentally incorrect since the astronomical tide is deterministic. Therefore, a clear explanation of the terminology is needed. Furthermore, are the tidal boundary conditions used at the sluice gate (as outflow boundary conditions?)? The authors mention a tidal gate, but the location is not shown in the figure.
- When selecting the sample for joint probability analysis, the tidal level is chosen as the "highest tidal level of the corresponding day." This method implies that the calculated tidal design levels could occur within a 24-hour period from the rainfall event. In the numerical modeling process, a single event is used as a representative for the tidal levels process, but the timing of the rainfall peak and tidal peaks in generating design scenarios is not explained. Additionally, figure 8(a) shows the peak rainfall and peak tide almost coinciding, which is rare in reality (not the most probable 200-year event). The 3-hour rainfall duration shows the peak rainfall occurring approximately half an hour after the peak tide, whereas the 12-hour rainfall duration shows the peak rainfall occurring approximately 3 hours before the peak tide. How are the corresponding time lags between peak rainfall and peak tide determined for each scenario? Are they coming from a probabilistic basis? This timing significantly impacts the resultant flood volume and might be the main reason for the variations discussed around L430.
- 41 and 42 are used to evaluate the spatial interactions between drainage units (DUs). The equations only rely on the masses and distances of two drainage units. To clarify the conclusions around L445-L450, it is essential to understand what the spatial interaction force represents. For example, the spatial interaction force between DU22 and DU6 (which are far away from each other) is 0.75. Does this relatively higher value suggest a higher level of interconnectedness or influence on each other, than even adjacent drainage units? Please explain. Also, it is not clear how the spatial interaction of the drainage units relates to the main objectives of the study.
- It is necessary to provide additional details on the types of data used in the study. For instance, specify whether rainfall data is gridded or gauge data, and if gauge data is used, indicate the number of gauges involved. What are the locations of the tidal data and rainfall used? If so, do you assume a uniformly distributed rainfall over the entire catchment?
- The paper lacks a proper discussion about the main assumptions and limitations of the process. For instance, using only 16 years of data to estimate a 200-year design event introduces significant uncertainty in modeling the dependence structure and the tail distributions. Since the study proposes a "universal" method, it is crucial for readers to understand these assumptions and uncertainties, especially when the method is to be applied to other study sites.
Minor Comments
- L11 “currently” repeats.
- In the abstract, the authors mention the Kendall return period for the combined event of rainfall and tidal levels, greater than the “Or” return period and less than the “And” return period. The Kendall return period is typically expected to fall between the "Or" and "And" scenarios for combined events. It's not a significant finding of the study to be mentioned in the abstract.
- L32, and L36, could you provide references?
- Figure 2: what represents the purple arrow from correlation analysis to “effects of social factors on flood”?
- L61, in underdeveloped areas, fluvial flooding may indeed be more prevalent due to the natural terrain and lack of infrastructure to mitigate such events. However, it is not accurate to say that "only" fluvial flooding exists in these areas. Could you explain or provide references?
- Provide references for the “Pilgrim & Cordery rainfall model”, “Maximum Possible Weighting Function” and “Pilgrim & Cordery rainfall model”
- Figure 6 seems incorrect. The theoretical distributions (GEV, Gamma) should be smooth unless you don’t change the distribution parameters.
- What types of information does Figure 7 (b) provide? What are the insights we can gain into joint distribution probabilities? The color scale is also not given.
- Although the peak tidal levels could be negative (depending on the datum used), the gamma distribution is selected to fit the peak tidal levels for some rainfall durations. The gamma distribution is lower bounded by zero thus making it impossible to sample negative realizations from the fitted distribution (if necessary).
- The extreme sample selection is not well explained. Does it consider annual maxima? How to ensure the events are independent? What is the sample size?
- Figure 10: The study area is in the legend (red). But cannot be seen in the figure.
- Figure 12: Do these different colors represent any information? If not use a single color.
- It's not clear how this observation is made “Despite differences in rainfall peak timing, rainfall, and tide peak timings leading to variations in calculated flooding volumes, the impact degree index is not affected by these differences”?
- It's not clear how this conclusion is made “This study also highlights that under the same recurrence interval, rainfall events with larger peak timings are more destructive than those with earlier peak timings”? Have you run many events for the same return period by only changing the peak timing?
- The authors suggest a set of equations (from eq. 30 to eq. 40), but poorly explained. Consider elaborating more about the governing process of equations. Additionally, it's mentioned that “calculate the range of rainfall and tidal level design values for different durations from 2-yr RP to 200-yr RP using the following formula” (L286,), but the formula doesn’t give the range of rainfall, and since it divides the range by 200yr Rainfall value. So, ∆Xt will be a dimensionless parameter that is related to the amount of variability (range) of rainfall. Same for Tides. Same in L304, ∆Vxt and ∆Vyt do not quantify the variation of flooding volume. Accordingly, check Dxt and Dyt.
- The paragraph from L543 to 558 is more about discussing the results. Consider moving it into the discussion section.
- The paper is lengthy. Consider moving some of the extra results to the supplementary materials. For example, Tables 1 and 2. Commonly used equations, such as those for correlation and distributions, can be omitted from the text. Instead, it is sufficient to cite the relevant references for these well-known equations.
- Check the way of citing publications in the manuscript according to the HESS guideline (e.g., L40, L42, L50).
Citation: https://doi.org/10.5194/hess-2024-100-RC2 -
AC3: 'Reply on RC2', Ting Zhang, 09 Aug 2024
We are sincerely grateful to the editor and reviewers for their valuable time and diligent consideration in reviewing our manuscript. We have carefully addressed each comment to improve the quality of our manuscript. Please find attached a PDF responding to the comments of Reviewer 2.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 422 | 98 | 36 | 556 | 26 | 27 |
- HTML: 422
- PDF: 98
- XML: 36
- Total: 556
- BibTeX: 26
- EndNote: 27
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
