A Global Map for Selecting Stationary and Nonstationary Methods to Estimate Extreme Floods
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsSee the file attached
Comments for author File: 
Comments.pdf
Try to avoid using non-scientific words. I saw some of these in the discussion and conclusions.
Author Response
Dear Reviewer :
Thank you for your time and efforts in reviewing our manuscript. We are pleased to receive the positive feedback from the referees on our paper. We have considered all of the comments and made the appropriate modifications and revisions as recommended. If you require any further assistance, please do not hesitate to contact us. Please find attached point-to-point responses regarding the reviewers’ comments (marked in blue) and made corresponding changes in the main manuscript (in red).
Hope the revised manuscript are to your satisfaction.
Best.
Comments and Responses:
The authors use a global hydrological model with different scenarios to develop a world map of stationary and nonstationary regions. The work has on overall good quality, the methods and the need for such a product are well described, the results are well presented, and the conclusions are mainly supported. I found some minor issues that I described in the specific comments section. I also left some general comments that, if attended, will improve the manuscript’s quality. Finally, I would like the authors to address my general comment number 2 (model validation). With proper validation, the authors’ results would gain significance and (sorry for the redundancy) validity for the community.
Reply: We thank you for the opportunity to revise our manuscript before publication, and we hope that our updated version meets the standards of the journal. We have carefully reviewed your comments and made the necessary revisions to improve the clarity of our manuscript. We hope that our map can serve as an important resource for hydrological modelers and researchers in the field, and we appreciate your time and effort in reviewing our work. Please find attached a point-by-point response to your feedback below.
General comments
- The authors run a daily model to estimate maximum annual discharge values at each grid cell. This approximation may be valid for relative large watersheds (>10000 km2), however, it may underestimate peak flows at smaller scales. How do the authors address this limitation?
Reply: This is a very good point that scaling issues are still big challenges in the hydrological fields. Considering the resolution of the available GCM outputs, we select the 0.25 degrees as the spatial resolution of the modeling in this study. We recognize that the peak flow sometimes may be underestimated at smaller scales, but large-domain flood-related maps are now being used for various decision-making purposes. Hirabayashi et al. (2022) provided recommendations for the practical application of large-domain hazard maps in corporate practice. In this field, no previous study generated such a reference map which serves as a powerful guideline for the use of stationary versus nonstationary approaches for estimating extreme floods. In this sense, our selected resolution could be appropriate for the large-domain research filed, and our generated map can be used for users to consider which method should be used for flood modeling even at smaller scales. When more discharge data available, it is possible to generate the reference map at smaller scale at local regions.
- Figure 1 presents the mean annual maximum discharge, one of the main results, However, I do not see any mention of a validation of this product comparing simulated peak flows with observations. Without validation, the results and analysis (which seem good) lack a base that supports them, I strongly suggest providing a validation that at least compares magnitudes. Besides, a statistical validation of the peak flows is highly relevant for this kind of analysis.
Reply: An important reference can be used to support the validation of our study. Kimura et al. (2023, HESS) constructed a flood-hazard map using the same model and GCM runoff products (and the same spatial resolution).
In their figures 5 and 6, you can find the monthly mean discharge climatology, exceedance probability curve and Gumbel distributions for the annual maximum discharge based on CaMa-Flood simulation results using each runoff type as input values. The comparison sites were Global Runoff Data Centre (GRDC) observation sites, specifically the Ubon station (104.8617∘ E, 15.2217∘ N) in the Mekong River basin and the Itapeua station (63.0278∘ W, 4.0578∘ S) in the Amazon River basin. Monthly average discharge data were expected to be similar to the reanalysis data. The annual maximum river discharges are also checked and to as similar to the reanalysis data.
The reference was added in our revised manuscript, and hope readers can better understand our product.
On the other hand, we may need to re-clarify our new contribution to the scientific community. For the food-hazard studies, many flood-hazards maps are generated at local or large-domain scales. However, there has not been such a map which serves as a powerful guideline for the use of stationary versus nonstationary approaches for estimating extreme floods. Given the limited availability of the in situ discharge, we employed the GCM-output runoff products. Nine GCMs of CMIP6 are used to account for the uncertainties. The validation of the associated flood inundation maps are validated in Kimura et al. (2023), while we go further in this filed to generate a new map in our study.
Kimura, Y., Hirabayashi, Y., Kita, Y., Zhou, X., and Yamazaki, D.: Methodology for constructing a flood-hazard map for a future climate, Hydrol. Earth Syst. Sci., 27, 1627–1644, https://doi.org/10.5194/hess-27-1627-2023, 2023.
- The authors need to link their work more to previous work on the topic at different scales, Compare their results with others, contrast their scale with work done at smaller scales. And discuss what has a more significant influence, the hydrological model or the forcings. That will enrich the manuscript.
Reply: Thank for your good suggestion. Since there are no previous study focusing on generating such a reference map, the comparisons may be a bit difficult. Please see the response 4 which included the discussion of the limitations of our research.
- The authors limit their discussion to the benefits of their work, which I appreciate. However, I do not see any discussion or comment regarding the limitations of their research and possible future work avenues to tackle those.
Reply: We added some sentences to discuss this issue in the revised manuscript.
4.4. Limitation and Future Research
While our research marks a significant step in improving flood risk assessment by introducing a reference map for methodological selection, several limitations warrant consideration. The use of Generalized Additive Models for Location, Scale, and Shape (GAMLSS) presents certain constraints, and future investigations could explore alternative statistical methods or hybrids for enhanced model selection. Additionally, the reference map's resolution may not capture localized variations in hydrological behavior, suggesting a need for higher-resolution mapping. The choice of a 0.25-degree spatial resolution in our reference map was influenced by the availability of Global Climate Model (GCM) outputs. While this resolution serves the purpose of large-domain flood-related decision-making, it's important to acknowledge that it may result in some underestimation of peak flow at smaller scales. Future research endeavors could explore the feasibility of employing higher resolutions, such as 0.01 degrees, for more localized studies, which could provide a more fine-grained representation of hydrological conditions. Looking ahead, the reference map offers a versatile platform for various lines of inquiry. Researchers can leverage this map to estimate extreme flood characteristics, encompassing flood magnitudes and inundation extents. Validation, a critical step, in-volves comparing model-derived estimates with historical flood events, field observations, or remote sensing data to ensure the reliability of our methodology. Furthermore, future research can delve into exploring the disparities between nonstationary and stationary methods for estimating extreme floods. This investigation can shed light on the contrasting performance of these methodologies and the implications for flood risk assessment. Additionally, there is ample scope for conducting flood analyses under climate change scenarios, utilizing the reference map as a foundation for understanding how evolving climatic conditions may impact future flood events. In summary, while the 0.25-degree spatial resolution was chosen due to GCM data constraints and the requirements of large-scale flood-related decision-making, the potential exists for future research to employ higher resolutions for localized investigations. The reference map serves as a valuable resource for estimating and validating extreme flood events, exploring methodological disparities, and considering the influence of climate change on future flood scenarios.
Specific comments
Line 31, are these percentages relative to the 2 billion number?
Reply: we have rephrased it: In the period between 1998 and 2017 alone, floods affected more than two billion people worldwide with approximately 11% fatalities (142,088 deaths) and a 23% economic loss (USD $656 billion) [2].
Line 195: Please include some references with similar results,
Reply: One reference is included to support this.
Kimura, Y., Hirabayashi, Y., Kita, Y., Zhou, X., and Yamazaki, D.: Methodology for constructing a flood-hazard map for a future climate, Hydrol. Earth Syst. Sci., 27, 1627–1644, https://doi.org/10.5194/hess-27-1627-2023, 2023.
Line 199: Does the hydrological model have temporal variable land use?
Reply: in this study, we do not use the hydrological model. The land use changes effects the runoff generation, thus influence the discharge estimation.
Figure 3: Expand the caption to explain the colors (it will be better if you include a color bar)
Reply: sorry for the confusion, we removed the colors.
Line 251 and 262: The word “geological” confuse me as it makes me thinly of geology, rock etc. You should explain it better before using or changing the word.
Reply: we have revised the “spatial variation”.
Figures 4 and 5, what is white? Why do white areas not coincide between the frames?
Reply: The white region means that some data are not available/invalid. And these differences are result from the difference of different GCM outputs.
Line 329 and 330: hard to understand,
Reply: We have rephrased:
Of particular interest is the fact that non-stationary methods are spatially dispersed across the globe and of limited size.
Figure 8: The legend between the nonstationary and stationary methods seems spatially mixed as the titles appear close to the line with the opposite color. I suggest to organize both names in a legend box and display them only for the first time.
Reply: sorry for the confusion, we double checked it. It is correct, the color of the font is the same with that of the methods.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper discussed the stationary and no stationary methods to estimate the extreme floods, and finally, they generated the 100-year flood magnitude as a reference using GCM runoff data sets. The authors present some interesting results. However, some points need to be clarified further. Using GCMS and existing frameworks could solve flood reference mapping. Has it been tested or validated in any region or country? The 3 (Lognomal, Gamma, and Weibull) distribution functions are commonly used in many GCM-based flood models, and I did not see new in this approach. The GAMs main limitation is overfit; how has this been handled? The reference map shows the Nepal & Himalayan regions for the extreme flood-prone areas. Surprisingly, it did not show the extreme floods in India, such as Bihar, Odisha, and recent Kerala floods. Has the reference map been verified with any current flood events? If not, how could this be used as a reference map? What are the parameters being considered to generate the reference map? The authors should explain the limitations of their approach in the manuscript.
Author Response
Dear Reviewer :
Thank you for your time and efforts in reviewing our manuscript. We are pleased to receive the positive feedback from the referees on our paper. We have considered all of the comments and made the appropriate modifications and revisions as recommended. If you require any further assistance, please do not hesitate to contact us. Please find attached point-to-point responses regarding the reviewers’ comments (marked in blue) and made corresponding changes in the main manuscript (in red).
Hope the revised manuscript are to your satisfaction.
Best.
Comments and Responses:
The paper discussed the stationary and no stationary methods to estimate the extreme floods, and finally, they generated the 100-year flood magnitude as a reference using GCM runoff data sets. The authors present some interesting results. However, some points need to be clarified further. Using GCMS and existing frameworks could solve flood reference mapping. Has it been tested or validated in any region or country? The 3 (Lognomal, Gamma, and Weibull) distribution functions are commonly used in many GCM-based flood models, and I did not see new in this approach. The GAMs main limitation is overfit; how has this been handled? The reference map shows the Nepal & Himalayan regions for the extreme flood-prone areas. Surprisingly, it did not show the extreme floods in India, such as Bihar, Odisha, and recent Kerala floods. Has the reference map been verified with any current flood events? If not, how could this be used as a reference map? What are the parameters being considered to generate the reference map? The authors should explain the limitations of their approach in the manuscript.
Reply: We would like to express our gratitude for your thoughtful and constructive review of our paper. We appreciate the opportunity to address your concerns and provide clarifications regarding the points you raised.
Validation of the Approach, Reference Map and Extreme Flood Events: You inquired about the validation of our approach in any specific region or country. We acknowledge the importance of validating flood estimation methods, and it's an essential step in flood risk assessment. However, our study aimed to provide a reference map for suggesting the use of non-stationary or stationary methods to estimate floods, rather than conducting the validation itself. We intend to emphasize this in our manuscript to prevent any misconceptions about the scope of our work. The application of the referenced map for estimating flood magnitudes using stationary and nonstationary methods is our next step, as you rightly pointed out, we need to validate the estimated flood magnitudes in the next step that goes beyond the scope of our current study.
You pointed out that our reference map did not display extreme floods in certain regions, such as Bihar, Odisha, and Kerala in India. Based on our findings, we found that when estimating extreme floods for these regions, non-stationary method may be a good choice. Please note Figure 7 only shows the 100-year flood magnitudes estimated by stationary method. Sorry for this confusion.
Distribution Functions: You mentioned that the use of Lognormal, Gamma, and Weibull distribution functions is common in GCM-based flood models. Indeed, these distributions are widely used for modeling flood data. In our paper, we did not introduce new distribution functions but rather employed these established methods as part of the approach to estimate extreme floods. We will clarify this point in the manuscript to ensure it is correctly interpreted. The issue of overfitting in GAMs is a valid concern. To address this, we followed best practices by employing Bayesian Information Criterion method to determine the optimal parameter of our models.
Parameters for Generating the Reference Map: The parameters considered for generating the reference map were the distribution parameters of different distribution functions (GAMLSS models).
Limitations of the Approach: You correctly highlighted the importance of addressing the limitations of our approach. In response, we will incorporate a section in the manuscript dedicated to discussing the limitations of our study.
4.4. Limitation and Future Research
While our research marks a significant step in improving flood risk assessment by introducing a reference map for methodological selection, several limitations warrant consideration. The use of Generalized Additive Models for Location, Scale, and Shape (GAMLSS) presents certain constraints, and future investigations could explore alternative statistical methods or hybrids for enhanced model selection. Additionally, the reference map's resolution may not capture localized variations in hydrological behavior, suggesting a need for higher-resolution mapping. The choice of a 0.25-degree spatial resolution in our reference map was influenced by the availability of Global Climate Model (GCM) outputs. While this resolution serves the purpose of large-domain flood-related decision-making, it's important to acknowledge that it may result in some underestimation of peak flow at smaller scales. Future research endeavors could explore the feasibility of employing higher resolutions, such as 0.01 degrees, for more localized studies, which could provide a more fine-grained representation of hydrological conditions. Looking ahead, the reference map offers a versatile platform for various lines of inquiry. Researchers can leverage this map to estimate extreme flood characteristics, encompassing flood magnitudes and inundation extents. Validation, a critical step, in-volves comparing model-derived estimates with historical flood events, field observations, or remote sensing data to ensure the reliability of our methodology. Furthermore, future research can delve into exploring the disparities between nonstationary and stationary methods for estimating extreme floods. This investigation can shed light on the contrasting performance of these methodologies and the implications for flood risk assessment. Additionally, there is ample scope for conducting flood analyses under climate change scenarios, utilizing the reference map as a foundation for understanding how evolving climatic conditions may impact future flood events. In summary, while the 0.25-degree spatial resolution was chosen due to GCM data constraints and the requirements of large-scale flood-related decision-making, the potential exists for future research to employ higher resolutions for localized investigations. The reference map serves as a valuable resource for estimating and validating extreme flood events, exploring methodological disparities, and considering the influence of climate change on future flood scenarios.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have addressed all comments satisfactorily, and I recommend the publication after the editorial process.
