the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 May 2026
Towards improved integration of hydrological uncertainty and hydraulic model sensitivity in flood hazard mapping
Abstract. The study investigates the impact of hydrological uncertainties and the related sensitivity of the hydrodynamic modelling results on flood hazard mapping. Uncertainties in flood frequency analysis (FFA), including anticipated impact of climate change and sensitivity to channel and floodplain roughness, were examined. The study area is the lower Vipava river, a transboundary catchment shared between Slovenia and Italy with variable floodplain topography. Sensitivity analysis revealed that uncertainty in FFA and river channel roughness significantly influenced the inundation spatial extent and inundation depth, the impact of floodplain roughness appears to be limited. The analysis shows that natural successional changes in river channel roughness substantially impact the results of a 10-year RP event, increasing flood extent by 45%. The increase in inundated areas is less pronounced for 100- and 500-year RP floods, with increases of 15% and 11%, respectively. The assessed probability of inundation based on scenario ensembles provided an informative identification of areas most susceptible to potential changes in flood hazard. Our findings highlight the need to address critical scenario ensembles that incorporate FFA uncertainty and hydraulic roughness sensitivity, leading to more informative flood hazard mapping for steering future land use planning.
- Preprint
(9681 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-2602', Michael Nones, 25 May 2026
-
AC3: 'Reply on RC1', Simon Rusjan, 17 Jul 2026
Reply to Reviewer 1 (RC1) comments
We would like to thank Dr. Nones for his insightful comments and suggestions, which we found very helpful for improving our manuscript and presenting the results more concisely. Please find below our detailed replies, where we explain how we will address the general and specific comments. All changes will be tracked in the revised version of the manuscript (with track changes), which we will resubmit after receiving comments from all reviewers. We will also indicate where (i.e. by giving line numbers) each point of the reviewers' comments has been incorporated.
General comments:
The work presents an interesting approach on flood mapping, which includes the analysis of hydrological uncertainty and model sensitivity while developing flood hazard maps. The case study is a reach of the Vipava River, a transboundary river flowing from Slovenia to Italy.
The goal of the study is proper and timely, but, in my opinion, the manuscript should be significantly better developed before publication, starting from a more explicit stressing of the novelty, and how this specific work goes beyond the current state of the art. Indeed, the need for considering uncertainty in flood mapping is not a new topic, and there are examples in the literature, which should be addressed in the Introduction to point out knowledge gaps.
General comment #1 reply: Thank you for this comment. We will extend the “Introduction” section and add examples from the literature to better emphasize the novelty of the proposed approach and explain how our work goes beyond the current state of the art. We agree with the reviewer that considering uncertainty in flood mapping is not a new topic; however, in our view, and as noted by the reviewer, there is still a substantial knowledge gap between most theoretical approaches proposed for incorporating uncertainty and sensitivity into flood hazard mapping and the more practical implementation of these approaches.
The methodological part could be further expanded, also following my detailed comments below, to better guide readers through all the steps needed to produce the maps you show, and also what is needed to evaluate them from a quantitative point of view.
General comment #2 reply: We will incorporate the reviewer’s suggestions to improve the methodological part of the paper and better present the steps involved in the proposed methodological approach and the resulting maps.
I strongly advise separating the Results from the Discussion, expanding the latter section to better highlight the novelty of the study and what other scholars and stakeholders can learn from your application. Likely in the Introduction, a more in-depth review of the state of the art would support your statements in a more thorough manner, moving from reporting results to actually discussing them from a critical point of view.
General comment #3 reply: Following the reviewer’s comments, we will extend the Introduction section by providing a more in-depth review of the state of the art on different approaches for incorporating uncertainty and sensitivity analysis into flood hazard mapping. We will make a clearer distinction between the “Results” and “Discussion” sections to highlight the novelty of the study and provide a more concise overview of our results in relation to previous studies.
Detailed comments:
Introduction
- please avoid excessive use of AI-generated text. The first line of the Introduction could be found in many other similar studies, always with the same phrasing.
Detailed comment #1 reply: Following the reviewer comment we will rewrite the first sentence to improve the introduction into the topic. We would like to point out that we used AI tools only to support us in grammar and syntax review.
- there are no such things as “natural disasters”. As pointed out by UNDRR (https://www.undrr.org/our-impact/campaigns/no-natural-disasters), we have to change the way we communicate, and acknowledge that nature is not causing disasters, while is the presence of humans plays the major role.
Detailed comment #2 reply: We would like to thank the reviewer for this comment, we completely agree with reviewer’s notion. We will replace the term “natural disaster” with “natural hazard” throughout the manuscript.
- line 27: how can mapping reduce hazard? The link is not fully clear, as, usually, mapping is a good way to reduce risks. Please rephrase this paragraph.
Detailed comment #3 reply: Thank you for noting this. We will rephrase the sentence.
- lines 40-43: please double-check the language. I guess it should read “This…”. But also in this case, the language could be improved
Detailed comment #4 reply: Thank you for noting this. We will rephrase the sentences to improve the language.
- line 54: what critical scenarios and selected how? It would be good to have more details already in the Introduction, also to better catch the transferability of results/approach
Detailed comment #5 reply: Thank you for the comment. We will extend the description of the “scenario ensembles” to ensure a clearer link between the “Introduction” and “Methods” sections.
- line 59: additional information on how this roughness is calculated would help in understanding the novelty of the study. For example, have you considered time-changing roughness during the simulation or just values depending on the land use? Even if addressed later in the study, adding some comments in the Introduction would help readers better follow you.
Detailed comment #6 reply: Thank you for the comment. We will rephrase this paragraph to include the points raised by the reviewer, specifically potential changes in river channel roughness and land use-related roughness conditions. We have not considered time-changing roughness during the simulation but just values depending on the land use.Methods
- please change the title to a more general “Materials and Methods”, as you are also presenting the input data
Detailed comment #7 reply: Thank you for the comment. As suggested by the reviewer, we will change the title to “Materials and Methods”.
- line 77-78: where can readers see the flood inundation extent? Can you provide some additional references?
Detailed comment #8 reply: We will improve Figure 1 by adding the flood inundation extent. The inundation extent polygon is based on the observed flood extent during few past flood events. The shown flood extent will provide a general representation of a flood event in the range of a 100-year return period.
- Figure 1: what is the source of the buildings polygons? I suggest acknowledging data sources more properly.
Detailed comment #9 reply: Thank you for pointing this out. By including simplified building polygons in Figure 1, we aimed to represent the locations of larger built-up clusters. The simplified built-up cluster polygons was based on the detailed land-use classification provided by the Slovenian Ministry of Agriculture, Forestry and Food (MKGP). Details on the data source are provided in Table 2.
We have decided to remove the building polygons from new Figure 1, since we encountered the problem with readability of figure when adding inundation polygons to Figure 1. Also, the build-up areas are shown also in Figure 3 (lower map) where a clear acknowledgement of land use data will be provided.
- please add more references to past flooding events, as the second part of Sec. 2.1 remains a bit too vague in the present version
Detailed comment #10 reply: We will add references to second part of section 2.1 to improve the presentation of past flood events in the study area.
- would it be good to also have reference years in Table 1, to better understand what the baseline of the model is
Detailed comment #11 reply: Thank you for this comment. We will add reference years of the national datasets considered in the hydrodynamical model development.
- what is the source of Figure 3, and how was the clustering made? In my opinion, lines 120-130 should be expanded, providing more details that ensure the reproducibility of the study
Detailed comment #12 reply: We will add a reference to the national land-use database (MKGP, 2026), from which the land-use classes shown in Figure 3 were obtained. We will provide additional information on the clustering of the land-use classes. The main aim of preparing Figure 3 (lower map) with clustered land-use classes was to more clearly represent the fragmented patterns of built-up areas and the general extent of agricultural land-use classes along the studied Vipava River section. We will rewrite the section in which the land-use characteristics of the study area are presented.
- given that you used HecRas (Sec. 2.2.2), this model should be described in the Introduction, pointing out its pros and cons with respect to the specific case study. HecRas has some limitations that should be acknowledged, and the fact that it was already used in the study area does not suffice to confirm that it is a good choice
Detailed comment #13 reply: We will follow the reviewer’s suggestion and introduce the HEC-RAS hydrodynamic model already in the Introduction. We agree that HEC-RAS has some limitations; however, it is widely used software that has been relatively successfully applied in many studies. We would also like to point out that other hydrodynamic models have been used in the past to study flood conditions along the studied Vipava River section (e.g. FLO-2D, MIKE FLOOD). Based on different applications, we find it extremely difficult to clearly identify one hydrodynamic model as superior to others. However, we will extend the description of the HEC-RAS hydrodynamic model settings to more clearly present the potential limitations of the performed hydrodynamic simulations.
- line 144: here you said that roughness was calibrated against water levels. Would it be possible to see such a calibration, and eventually a validation against flooding events? I understand you provided some references, but readers should also be able to follow you without looking at past studies. You can add calibration/validation details as supplementary material.
Detailed comment #14 reply: Thank you for this comment. Following the reviewer’s suggestion, we will provide the observed combined flood extent during the more recent flood events in 2010 and 2012 as supplementary material. The observed flood extent was used for calibration of the hydrodynamic model simulation. Considerable differences in the flood peak return periods were reported for specific water stations along Vipava River. The peak discharge return periods of the observed flood events along the studied Vipava River reach were assessed to locally range between 50 and 100 years. For few other flood events with lower return periods, only water levels at specific locations are available. These events were used to calibrate and validate the channel roughness characteristics of the Vipava River.
- line 158: with the current development of IT infrastructures, the computational effort is not a limitation anymore. I suggest deleting this sentence. In addition, there are other free and/or commercial models that perform better than HecRas in terms of computational speed and efficiency (and this comment connects to the need of explaining why HecRas was used).
Detailed comment #15 reply: We agree with the reviewer that recent developments in IT infrastructure have considerably reduced computational limitations when performing numerous hydrodynamic simulations. However, we believe that performing numerous simulations with changing hydrodynamic settings remains a considerable challenge in many cases. We will rewrite this section to better address the points raised by the reviewer. As stated in the reply to Detailed comment #12, in our view, it is extremely difficult to clearly identify one hydrodynamic model as superior to others based on different criteria, such as computational speed, efficiency, numerical stability etc.
- from the description reported in Section 2.2.2, I can understand that simulations were done with clear water. Could you please confirm that no morphological changes were considered, and that this reflects the conditions of the study reach?
Detailed comment #16 reply: The reviewer is correct; the simulations were performed without considering sediment transport processes, and no potential hydromorphological changes were considered. We consider this assumption suitable for the studied Vipava River section, as only relatively limited and spatially constrained hydromorphological changes have been observed during the last few flood events.
- line 201: why a threshold of 0.01 m water depth? What’s the rationale behind it?
Detailed comment #17 reply: Thank you for this comment. Since several floodplain sections along the studied Vipava River section are wide and flat, we selected a relatively low threshold (0.01 m) to obtain a more detailed overview of the potential changes in inundation extent driven by the selected scenario ensembles. We are aware that selecting such a low threshold might raise questions about hydrodynamic model sensitivity. However, the general idea behind selecting the scenario ensembles was to better cover different uncertainty and sensitivity aspects that are reflected in changes in inundation extent.
Results and Discussion
- line 210: please add a reference to the first sentence.
Detailed comment #18 reply: Thank you for noting this, we will add relevant references.
- Figure 4: how where the flood hazard classes defined? I might have missed it.
Detailed comment #19 reply: Flood hazard classes were defined following the criteria presented in Table 4.
- line 230: this raises a question about the actual need of using statistical methods. I agree with your point of view, but I suggest deepening the discussion on this aspect, eventually looking at other studies.
Detailed comment #20 reply: Thank you for pointing this out. We will extend this section to better address the problem of assessing the design discharges in view of the observed flood peaks.
- Sections 3.2 and 3.3 need more comments, as roughness plays a major role in water levels rather than in spatial extent, as you also noticed. Do water levels play a role in the study area (e.g., evacuation routes, buildings)
Detailed comment #21 reply: Thank you for this comment. We will extend the comments in Sections 3.2 and 3.3 to better address the influence of roughness on water levels and the spatial extent of inundation. Following the reviewer’s suggestions, we will also extend the comments in Section 3.4 related to the impact of water levels on inundation characteristics in built-up areas.
- line 295: Louise Slater (she, so should be “her results”) looked at the contribution of a changing morphology on flood risk, while your model considers clear water, if I understood correctly. Could you please provide more comments on that? Moreover, there is ample recent literature on the influence of sediments and active bed on flood risk, and I suggest deepening the literature review if you wish to address this topic.
Detailed comment #22 reply: Thank you for noting our mistake; we apologize for that. As noted by the reviewer, we did not consider the impact of changed channel hydromorphology, but rather the impact of changed river channel hydraulic roughness, which was also one of the drivers of changed hydraulic capacity identified by Slater (2016). Following the reviewer’s suggestion, we will extend the discussion by adding references to support our discussion of the impact of successional changes in river channel roughness on inundation extent and inundation depth.
- would it be possible to have a clearer Figure 10, eventually using the same colour scale to help compare results?
Detailed comment #23 reply: We will make a new version of Figure 10 in order to make the presentation clearer. We will unify the colors scale of both graphs and increased text fonts.
- line 348-361: this remains a rather vague discussion. I suggest adding some clear, actionable steps to take advantage of flood mapping, also considering the potential use of the affected areas. Appropriate spatial planning should go beyond abandoning areas, and I would like to see more comments on this, also considering the current state of the art.
Detailed comment #24 reply: Thank you for these suggestions. We will enhance the discussion by following the reviewer’s suggestions to more clearly present the potential advantages of the flood maps shown in Figures 11 and 12. We will also add references to support the discussion of state-of-the-art approaches to spatial planning and the potential uses of the information provided by the weighted exceedance probability of inundation.
Conclusions
- line 382: I would use “changes” (or a similar word) instead of “deterioration”, as some changes in roughness could eventually reduce local flood hazard.
Detailed comment #25 reply: Thank you for this attentive comment. We will implement the suggested corrections (“deterioration” to “changes”).
- line 388: again, I do not see computational time as a limitation nowadays. It’s possible to address it using adequate models and IT infrastructures.
Detailed comment #26 reply: Generally, we agree with the reviewer’s opinion. However, as mentioned earlier (Detailed comment #14 reply), we believe that performing numerous simulations is, in many cases, still a considerable challenge. We decided to rewrite the sentence in the direction indicated by the reviewer.
Data availability
- I was not able to find the Zenodo dataset. As this could be my mistake, could you please help me with this?
Detailed comment #27 reply: We will check the link to the datased uploadad to Zenodo, it worked for us. We will do additional check of the link with the journal office.
Citation: https://doi.org/10.5194/egusphere-2026-2602-AC3
-
AC3: 'Reply on RC1', Simon Rusjan, 17 Jul 2026
-
AC1: 'Comment on egusphere-2026-2602', Simon Rusjan, 24 Jun 2026
Reply to Reviewer comments
We would like to thank Dr. Nones for his insightful comments and suggestions, which we found very helpful for improving our manuscript and presenting the results more concisely. Please find below our detailed replies, where we explain how we will address the general and specific comments. All changes will be tracked in the revised version of the manuscript (with track changes), which we will resubmit after receiving comments from all reviewers. We will also indicate where (i.e. by giving line numbers) each point of the reviewers' comments has been incorporated.
General comments:
The work presents an interesting approach on flood mapping, which includes the analysis of hydrological uncertainty and model sensitivity while developing flood hazard maps. The case study is a reach of the Vipava River, a transboundary river flowing from Slovenia to Italy.
The goal of the study is proper and timely, but, in my opinion, the manuscript should be significantly better developed before publication, starting from a more explicit stressing of the novelty, and how this specific work goes beyond the current state of the art. Indeed, the need for considering uncertainty in flood mapping is not a new topic, and there are examples in the literature, which should be addressed in the Introduction to point out knowledge gaps.
General comment #1 reply: Thank you for this comment. We will extend the “Introduction” section and add examples from the literature to better emphasize the novelty of the proposed approach and explain how our work goes beyond the current state of the art. We agree with the reviewer that considering uncertainty in flood mapping is not a new topic; however, in our view, and as noted by the reviewer, there is still a substantial knowledge gap between most theoretical approaches proposed for incorporating uncertainty and sensitivity into flood hazard mapping and the more practical implementation of these approaches.
The methodological part could be further expanded, also following my detailed comments below, to better guide readers through all the steps needed to produce the maps you show, and also what is needed to evaluate them from a quantitative point of view.
General comment #2 reply: We will incorporate the reviewer’s suggestions to improve the methodological part of the paper and better present the steps involved in the proposed methodological approach and the resulting maps.
I strongly advise separating the Results from the Discussion, expanding the latter section to better highlight the novelty of the study and what other scholars and stakeholders can learn from your application. Likely in the Introduction, a more in-depth review of the state of the art would support your statements in a more thorough manner, moving from reporting results to actually discussing them from a critical point of view.
General comment #3 reply: Following the reviewer’s comments, we will extend the Introduction section by providing a more in-depth review of the state of the art on different approaches for incorporating uncertainty and sensitivity analysis into flood hazard mapping. We will make a clearer distinction between the “Results” and “Discussion” sections to highlight the novelty of the study and provide a more concise overview of our results in relation to previous studies.
Detailed comments:
Introduction
- please avoid excessive use of AI-generated text. The first line of the Introduction could be found in many other similar studies, always with the same phrasing.
Detailed comment #1 reply: Following the reviewer comment we will rewrite the first sentence to improve the introduction into the topic. We would like to point out that we used AI tools only to support us in grammar and syntax review.
- there are no such things as “natural disasters”. As pointed out by UNDRR (https://www.undrr.org/our-impact/campaigns/no-natural-disasters), we have to change the way we communicate, and acknowledge that nature is not causing disasters, while is the presence of humans plays the major role.
Detailed comment #2 reply: We would like to thank the reviewer for this comment, we completely agree with reviewer’s notion. We will replace the term “natural disaster” with “natural hazard” throughout the manuscript.
- line 27: how can mapping reduce hazard? The link is not fully clear, as, usually, mapping is a good way to reduce risks. Please rephrase this paragraph.
Detailed comment #3 reply: Thank you for noting this. We will rephrase the sentence.
- lines 40-43: please double-check the language. I guess it should read “This…”. But also in this case, the language could be improved
Detailed comment #4 reply: Thank you for noting this. We will rephrase the sentences to improve the language.
- line 54: what critical scenarios and selected how? It would be good to have more details already in the Introduction, also to better catch the transferability of results/approach
Detailed comment #5 reply: Thank you for the comment. We will extend the description of the “scenario ensembles” to ensure a clearer link between the “Introduction” and “Methods” sections.
- line 59: additional information on how this roughness is calculated would help in understanding the novelty of the study. For example, have you considered time-changing roughness during the simulation or just values depending on the land use? Even if addressed later in the study, adding some comments in the Introduction would help readers better follow you.
Detailed comment #6 reply: Thank you for the comment. We will rephrase this paragraph to include the points raised by the reviewer, specifically potential changes in river channel roughness and land use-related roughness conditions. We have not considered time-changing roughness during the simulation but just values depending on the land use.
Methods
- please change the title to a more general “Materials and Methods”, as you are also presenting the input data
Detailed comment #7 reply: Thank you for the comment. As suggested by the reviewer, we will change the title to “Materials and Methods”.
- line 77-78: where can readers see the flood inundation extent? Can you provide some additional references?
Detailed comment #8 reply: We will improve Figure 1 by adding the flood inundation extent. The inundation extent polygon is based on the observed flood extent during few past flood events. The shown flood extent will provide a general representation of a flood event in the range of a 100-year return period.
- Figure 1: what is the source of the buildings polygons? I suggest acknowledging data sources more properly.
Detailed comment #9 reply: Thank you for pointing this out. By including simplified building polygons in Figure 1, we aimed to represent the locations of larger built-up clusters. The simplified built-up cluster polygons was based on the detailed land-use classification provided by the Slovenian Ministry of Agriculture, Forestry and Food (MKGP). Details on the data source are provided in Table 2.
We have decided to remove the building polygons from new Figure 1, since we encountered the problem with readability of figure when adding inundation polygons to Figure 1. Also, the build-up areas are shown also in Figure 3 (lower map) where a clear acknowledgement of land use data will be provided.
- please add more references to past flooding events, as the second part of Sec. 2.1 remains a bit too vague in the present version
Detailed comment #10 reply: We will add references to second part of section 2.1 to improve the presentation of past flood events in the study area.
- would it be good to also have reference years in Table 1, to better understand what the baseline of the model is
Detailed comment #11 reply: Thank you for this comment. We will add reference years of the national datasets considered in the hydrodynamical model development.
- what is the source of Figure 3, and how was the clustering made? In my opinion, lines 120-130 should be expanded, providing more details that ensure the reproducibility of the study
Detailed comment #12 reply: We will add a reference to the national land-use database (MKGP, 2026), from which the land-use classes shown in Figure 3 were obtained. We will provide additional information on the clustering of the land-use classes. The main aim of preparing Figure 3 (lower map) with clustered land-use classes was to more clearly represent the fragmented patterns of built-up areas and the general extent of agricultural land-use classes along the studied Vipava River section. We will rewrite the section in which the land-use characteristics of the study area are presented.
- given that you used HecRas (Sec. 2.2.2), this model should be described in the Introduction, pointing out its pros and cons with respect to the specific case study. HecRas has some limitations that should be acknowledged, and the fact that it was already used in the study area does not suffice to confirm that it is a good choice
Detailed comment #13 reply: We will follow the reviewer’s suggestion and introduce the HEC-RAS hydrodynamic model already in the Introduction. We agree that HEC-RAS has some limitations; however, it is widely used software that has been relatively successfully applied in many studies. We would also like to point out that other hydrodynamic models have been used in the past to study flood conditions along the studied Vipava River section (e.g. FLO-2D, MIKE FLOOD). Based on different applications, we find it extremely difficult to clearly identify one hydrodynamic model as superior to others. However, we will extend the description of the HEC-RAS hydrodynamic model settings to more clearly present the potential limitations of the performed hydrodynamic simulations.
- line 144: here you said that roughness was calibrated against water levels. Would it be possible to see such a calibration, and eventually a validation against flooding events? I understand you provided some references, but readers should also be able to follow you without looking at past studies. You can add calibration/validation details as supplementary material.
Detailed comment #14 reply: Thank you for this comment. Following the reviewer’s suggestion, we will provide the observed combined flood extent during the more recent flood events in 2010 and 2012 as supplementary material. The observed flood extent was used for calibration of the hydrodynamic model simulation. Considerable differences in the flood peak return periods were reported for specific water stations along Vipava River. The peak discharge return periods of the observed flood events along the studied Vipava River reach were assessed to locally range between 50 and 100 years. For few other flood events with lower return periods, only water levels at specific locations are available. These events were used to calibrate and validate the channel roughness characteristics of the Vipava River.
- line 158: with the current development of IT infrastructures, the computational effort is not a limitation anymore. I suggest deleting this sentence. In addition, there are other free and/or commercial models that perform better than HecRas in terms of computational speed and efficiency (and this comment connects to the need of explaining why HecRas was used).
Detailed comment #15 reply: We agree with the reviewer that recent developments in IT infrastructure have considerably reduced computational limitations when performing numerous hydrodynamic simulations. However, we believe that performing numerous simulations with changing hydrodynamic settings remains a considerable challenge in many cases. We will rewrite this section to better address the points raised by the reviewer. As stated in the reply to Detailed comment #12, in our view, it is extremely difficult to clearly identify one hydrodynamic model as superior to others based on different criteria, such as computational speed, efficiency, numerical stability etc.
- from the description reported in Section 2.2.2, I can understand that simulations were done with clear water. Could you please confirm that no morphological changes were considered, and that this reflects the conditions of the study reach?
Detailed comment #16 reply: The reviewer is correct; the simulations were performed without considering sediment transport processes, and no potential hydromorphological changes were considered. We consider this assumption suitable for the studied Vipava River section, as only relatively limited and spatially constrained hydromorphological changes have been observed during the last few flood events.
- line 201: why a threshold of 0.01 m water depth? What’s the rationale behind it?
Detailed comment #17 reply: Thank you for this comment. Since several floodplain sections along the studied Vipava River section are wide and flat, we selected a relatively low threshold (0.01 m) to obtain a more detailed overview of the potential changes in inundation extent driven by the selected scenario ensembles. We are aware that selecting such a low threshold might raise questions about hydrodynamic model sensitivity. However, the general idea behind selecting the scenario ensembles was to better cover different uncertainty and sensitivity aspects that are reflected in changes in inundation extent.
Results and Discussion
- line 210: please add a reference to the first sentence.
Detailed comment #18 reply: Thank you for noting this, we will add relevant references.
- Figure 4: how where the flood hazard classes defined? I might have missed it.
Detailed comment #19 reply: Flood hazard classes were defined following the criteria presented in Table 4.
- line 230: this raises a question about the actual need of using statistical methods. I agree with your point of view, but I suggest deepening the discussion on this aspect, eventually looking at other studies.
Detailed comment #20 reply: Thank you for pointing this out. We will extend this section to better address the problem of assessing the design discharges in view of the observed flood peaks.
- Sections 3.2 and 3.3 need more comments, as roughness plays a major role in water levels rather than in spatial extent, as you also noticed. Do water levels play a role in the study area (e.g., evacuation routes, buildings)
Detailed comment #21 reply: Thank you for this comment. We will extend the comments in Sections 3.2 and 3.3 to better address the influence of roughness on water levels and the spatial extent of inundation. Following the reviewer’s suggestions, we will also extend the comments in Section 3.4 related to the impact of water levels on inundation characteristics in built-up areas.
- line 295: Louise Slater (she, so should be “her results”) looked at the contribution of a changing morphology on flood risk, while your model considers clear water, if I understood correctly. Could you please provide more comments on that? Moreover, there is ample recent literature on the influence of sediments and active bed on flood risk, and I suggest deepening the literature review if you wish to address this topic.
Detailed comment #22 reply: Thank you for noting our mistake; we apologize for that. As noted by the reviewer, we did not consider the impact of changed channel hydromorphology, but rather the impact of changed river channel hydraulic roughness, which was also one of the drivers of changed hydraulic capacity identified by Slater (2016). Following the reviewer’s suggestion, we will extend the discussion by adding references to support our discussion of the impact of successional changes in river channel roughness on inundation extent and inundation depth.
- would it be possible to have a clearer Figure 10, eventually using the same colour scale to help compare results?
Detailed comment #23 reply: We will make a new version of Figure 10 in order to make the presentation clearer. We will unify the colors scale of both graphs and increased text fonts.
- line 348-361: this remains a rather vague discussion. I suggest adding some clear, actionable steps to take advantage of flood mapping, also considering the potential use of the affected areas. Appropriate spatial planning should go beyond abandoning areas, and I would like to see more comments on this, also considering the current state of the art.
Detailed comment #24 reply: Thank you for these suggestions. We will enhance the discussion by following the reviewer’s suggestions to more clearly present the potential advantages of the flood maps shown in Figures 11 and 12. We will also add references to support the discussion of state-of-the-art approaches to spatial planning and the potential uses of the information provided by the weighted exceedance probability of inundation.
Conclusions
- line 382: I would use “changes” (or a similar word) instead of “deterioration”, as some changes in roughness could eventually reduce local flood hazard.
Detailed comment #25 reply: Thank you for this attentive comment. We will implement the suggested corrections (“deterioration” to “changes”).
- line 388: again, I do not see computational time as a limitation nowadays. It’s possible to address it using adequate models and IT infrastructures.
Detailed comment #26 reply: Generally, we agree with the reviewer’s opinion. However, as mentioned earlier (Detailed comment #14 reply), we believe that performing numerous simulations is, in many cases, still a considerable challenge. We decided to rewrite the sentence in the direction indicated by the reviewer.
Data availability
- I was not able to find the Zenodo dataset. As this could be my mistake, could you please help me with this?
Detailed comment #27 reply: We will check the link to the datased uploadad to Zenodo, it worked for us. We will do additional check of the link with the journal office.
Citation: https://doi.org/10.5194/egusphere-2026-2602-AC1 -
RC2: 'Comment on egusphere-2026-2602', Antonio Annis, 26 Jun 2026
I have read the manuscript together with the RC1 report and the authors' reply. My concerns below are mainly about the stated novelty and about the specification of some methodological steps that are central to the conclusions but cannot, as written, be reproduced or fully interpreted.
Major comments
The RC1 has already asked for a clearer statement of novelty; I would make the request more specific. The comparison of hydrological versus hydraulic uncertainty in inundation mapping is well established (e.g. Merz & Thieken, 2005; Apel et al., 2008; Mosquera-Machado & Ahmad, 2007; Dimitriadis et al., 2016; Annis et al., 2020). As written, the manuscript reads as a careful site-specific application rather than a methodological advance. Please state explicitly in the Introduction what is new relative to this literature — e.g. the "boundary scenario ensemble" construction, the weighted exceedance-probability aggregation, or the joint treatment of FFA confidence intervals and climate scenarios.
Table 1 reports the flood-frequency analysis at "daily" resolution, and the text (l. 109) refers to "annual discharge peaks". For a catchment with the torrential tributary behaviour and karst response you describe, and at this reach scale, the time of concentration is plausibly of the order of hours, so annual maxima of daily-mean discharge can substantially underestimate the instantaneous flood peak. Please clarify whether the FFA target variable is the instantaneous annual maximum or the daily-mean annual maximum and, if only daily data were available, whether a peak/daily-mean adjustment was applied.
Three candidate distributions are mentioned (Pearson III, Log-Pearson III, GEV; l. 109), but the manuscript does not state which distribution was adopted for the design values and the confidence intervals, nor does it report any goodness-of-fit assessment. Since the S1 rests entirely on the 10% and 90% confidence intervals, which distribution was selected and on what basis? Have you done the goodness-of-fit evidence, and how the confidence intervals were estimated (parameter/sampling uncertainty, bootstrap, etc.)? At present this is delegated to a thesis written in Slovenian (Piry, 2020), which is not sufficient for reproducibility. Please include the essential FFA results in the manuscript or as supplementary material — at minimum the selected distribution, fitted parameters, design quantiles with confidence intervals, and a fit diagnostic.
Section 2.2.3 mentions the Morris screening method, but the manuscript does not report the information needed to interpret or reproduce it: the number of trajectories, the number of levels, the parameter ranges sampled, and the resulting total number of model runs. This matters because Table 3 describes only eight boundary scenarios constructed in a one-factor-at-a-time manner, whereas a Morris analysis requires randomized trajectories of runs each across the parameter space. It is therefore unclear whether the Morris indices in Fig. 10 were computed from the eight listed runs (in which case they are not a proper Morris sample) or from an additional, undocumented set of simulations. Please clarify it.
The "weighted exceedance probability of inundation" is one of the manusript's central outputs, but it is not defined precisely enough to be reproduced or interpreted. As described, it combines binary inundation indicators from simulations that span different return periods and different epistemic choices (FFA confidence intervals, floodplain roughness, channel roughness), each weighted by an "annual exceedance probability". Please provide the governing equation: which simulations enter the aggregation, the weight assigned to each, and the normalization. Moreover, the return period is an aleatory (frequency) quantity, whereas the roughness scenarios and the FFA confidence-interval bounds represent epistemic uncertainty. I 'd ask the authors either to separate the weighting across return periods from the epistemic spread across scenario members)or to justify the chosen aggregation and state clearly what the resulting map represents.
The RCP-driven peak-increase scenarios are currently treated somewhat separately from the main sensitivity framework: they appear in the narrative comparison with the FFA confidence intervals (Sect. 3.1) and in Fig. 12, but not in the comparative sensitivity analysis of Fig. 10. Since the climate signal is a scaling of the peak-discharge factor, it could be incorporated directly into the same comparative framework — for example as an additional level/axis in Fig. 10, or in a single figure that ranks FFA confidence-interval width, channel-roughness change, floodplain roughness, and the RCP-driven increase on the same footing. This would let the reader judge how much the climate-change contribution weighs relative to the other sources, and it would directly support the paper's claim that the FFA uncertainty already "covers" the RCP 2.6/4.5 ranges.
Specific comments
- Figure 1. The elevation colour scale used in the left (overview) panel does not match the colour bar/legend shown for the right panel. Please harmonize the two.
- The floodplain-roughness effect on inundation extent (≈4–6%) is of the same order as the reported validation uncertainty (mean 6%, s.d. 3%; l. 154–156), which you partly acknowledge at l. 265. Please state this explicitly where the S2 results are first presented.
- Scenario ensemble S3 compares the 1970s design channel roughness (n = 0.035) with the current calibrated This is a comparison between two temporal states of the channel (a change scenario) rather than a parameter "sensitivity" in the usual sense. Consider adjusting the terminology so that the distinction between epistemic sensitivity (S2) and physical/temporal change (S3) is clear to the reader.
References
Apel, H., Merz, B., & Thieken, A. H. (2008). Quantification of uncertainties in flood risk assessments. International Journal of River Basin Management, 6(2), 149-162.
Merz, B., & Thieken, A. H. (2005). Separating natural and epistemic uncertainty in flood frequency analysis. Journal of Hydrology, 309(1-4), 114-132.
Mosquera-Machado, S., & Ahmad, S. (2007). Flood hazard assessment of Atrato River in Colombia. Water resources management, 21(3), 591-609.
Dimitriadis, P., Tegos, A., Oikonomou, A., Pagana, V., Koukouvinos, A., Mamassis, N., ... & Efstratiadis, A. (2016). Comparative evaluation of 1D and quasi-2D hydraulic models based on benchmark and real-world applications for uncertainty assessment in flood mapping. Journal of Hydrology, 534, 478-492.
Citation: https://doi.org/10.5194/egusphere-2026-2602-RC2 -
AC2: 'Reply on RC2', Simon Rusjan, 17 Jul 2026
Reply to reviewer 2 (RC2) general remarks:
We would like to thank Dr. Annis for his insightful comments and suggestions, which we found very helpful for improving our manuscript and presenting the results more concisely. Please find below our detailed replies, where we explain how we will address the general and specific comments. All changes will be tracked in the revised version of the manuscript (with track changes), which we will resubmit after receiving comments from all reviewers and the Editor’s decision. We will also indicate where (i.e., by giving line numbers) each point of the reviewers' comments has been incorporated.
Major comments
The RC1 has already asked for a clearer statement of novelty; I would make the request more specific. The comparison of hydrological versus hydraulic uncertainty in inundation mapping is well established (e.g. Merz & Thieken, 2005; Apel et al., 2008; Mosquera-Machado & Ahmad, 2007; Dimitriadis et al., 2016; Annis et al., 2020). As written, the manuscript reads as a careful site-specific application rather than a methodological advance. Please state explicitly in the Introduction what is new relative to this literature — e.g. the "boundary scenario ensemble" construction, the weighted exceedance-probability aggregation, or the joint treatment of FFA confidence intervals and climate scenarios.
Major comment #1 reply:
We thank the reviewer for this important comment. We agree that uncertainty in flood-inundation mapping has already been addressed in previous published literature, and we will therefore present more explicitly how the present contribution differs from the earlier studies. In the revised Introduction, we will clarify that the novelty of the study does not lie in considering impacts of hydrological or hydraulic uncertainty separately, but in combining them within the suggested boundary scenario ensembles that could be more effectively incorporated into operational flood hazard mapping.
Specifically, we will more explicitly emphasize the following aspects: (1) the selection of boundary scenario ensembles that jointly cover FFA uncertainty, floodplain-roughness sensitivity, and channel-roughness change; (2) the aggregation of binary inundation outputs into weighted exceedance-probability maps, providing a spatially explicit measure of inundation occurrence across FFA uncertainty and hydrodynamic simulation sensitivity; and (3) the comparison of FFA confidence intervals with RCP-based peak-discharge increases to evaluate the extent to which the considered hydrological uncertainty could encompass the climate change projections. We will also revise the Results and Discussion sections to ensure that these elements are presented as a reproducible and operational oriented workflow that also supports the methodological novelty.
Table 1 reports the flood-frequency analysis at "daily" resolution, and the text (l. 109) refers to "annual discharge peaks". For a catchment with the torrential tributary behaviour and karst response you describe, and at this reach scale, the time of concentration is plausibly of the order of hours, so annual maxima of daily-mean discharge can substantially underestimate the instantaneous flood peak. Please clarify whether the FFA target variable is the instantaneous annual maximum or the daily-mean annual maximum and, if only daily data were available, whether a peak/daily-mean adjustment was applied.
Major comment #2 reply:
Thank you for this comment and for highlighting the importance of clearly specifying the temporal resolution of the discharge data used in the flood-frequency analysis (FFA). We will clarify that the FFA was performed using instantaneous annual maximum discharges from the national hydrometeorological monitoring system, not daily mean annual maximum discharges. The discharge records are based on sub-daily measurements, generally available at 15-minute recording intervals and subsequently stored or aggregated at 30-minute or hourly intervals depending on the station and period. Therefore, daily mean design discharges were not used in the FFA. We will revise the manuscript to state explicitly that the target variable was the instantaneous annual maximum discharge series and to provide the relevant information on the measurement and aggregation interval of the discharge data used for the FFA.
Three candidate distributions are mentioned (Pearson III, Log-Pearson III, GEV; l. 109), but the manuscript does not state which distribution was adopted for the design values and the confidence intervals, nor does it report any goodness-of-fit assessment. Since the S1 rests entirely on the 10% and 90% confidence intervals, which distribution was selected and on what basis? Have you done the goodness-of-fit evidence, and how the confidence intervals were estimated (parameter/sampling uncertainty, bootstrap, etc.)? At present this is delegated to a thesis written in Slovenian (Piry, 2020), which is not sufficient for reproducibility. Please include the essential FFA results in the manuscript or as supplementary material — at minimum the selected distribution, fitted parameters, design quantiles with confidence intervals, and a fit diagnostic.
Major comment #3 reply:
Thank you for this comment. We will include the description of method used for parameters estimation, design quantiles with confidence intervals, and the goodness-of-fit diagnostic in the revised manuscript. The distribution parameters were estimated using the method of L-moments (Hosking and Wallis, 2005; Bezak et al., 2014). The Akaike Information Criterion (AIC) was used for goodness-of-fit assessment and selection of the best-fitted distribution (Šraj and Bezak, 2025). Additionally, the Kolmogorov–Smirnov test was used to check whether the selected distribution function adequately describes the measured data on flood peaks (Šraj et al., 2025; Šraj and Bezak, 2025). We examined whether the selected distribution could not be rejected at the selected significance level (0.05) based on the Kolmogorov–Smirnov test. Confidence intervals of the design flood quantiles were estimated using a parametric bootstrap procedure (Meylan et al., 2012), with 10,000 synthetic samples generated from the fitted distribution, and further used to derive confidence intervals for the selected return-period quantiles.
We will add references (listed below) to support the presentation of FFA analysis and uncertainty interval assessment procedures.
Bezak, N., Brilly, M., and Šraj, M.: Comparison between the peaks-over-threshold method and the annual maximum method for flood frequency analysis, Hydrological Sciences Journal, 59, 959–977, https://doi.org/10.1080/02626667.2013.831174, 2014.
Hosking, J. R. M. and Wallis, J. R.: Some statistics useful in regional frequency analysis, Water Resources Research, 29, 271–281, https://doi.org/10.1029/92WR01980, 1993.
Meylan, P., Favre, A.-C., and Musy, A.: Predictive Hydrology: A Frequency Analysis Approach, CRC Press, Boca Raton, USA, 180 pp., https://doi.org/10.1201/b11832, 2012.
Šraj, M., Bezak, N. Modifying Design Standards: The 2023 Extreme Flood’s Impact on Design Discharges in Slovenia. Water, 17, 2198. https://doi.org/10.3390/w17152198, 2025.
Šraj, M., Bezak, N., Begelj, Ž., and Lebar, K.: Expert basis related to the preparation of flood-frequency analysis (FFA): final report, University of Ljubljana, Faculty of Civil and Geodetic Engineering, Chair of General Hydrotechnics, Ljubljana, Slovenia, 59 pp.,2025. (in Slovenian)
Section 2.2.3 mentions the Morris screening method, but the manuscript does not report the information needed to interpret or reproduce it: the number of trajectories, the number of levels, the parameter ranges sampled, and the resulting total number of model runs. This matters because Table 3 describes only eight boundary scenarios constructed in a one-factor-at-a-time manner, whereas a Morris analysis requires randomized trajectories of runs each across the parameter space. It is therefore unclear whether the Morris indices in Fig. 10 were computed from the eight listed runs (in which case they are not a proper Morris sample) or from an additional, undocumented set of simulations. Please clarify it.
Major comment #4 reply:
Thank you for this important comment. We will include additional information to present clearly how the scenario ensembles were considered in the Morris screening analysis. The Morris indices shown in Fig. 10 were not computed from only the eight scenarios listed in Table 3. Because hydrodynamic simulations following a full randomized Morris design would have been computationally demanding, we used a surrogate-assisted Morris screening approach based on the available ensemble of 24 hydraulic simulations representing the considered scenario ensembles: 8 simulations for each flood peak return period (Q10, Q100, and Q500). For each available hydrodynamic simulation, flood-response metrics, including inundation area and maximum depth, were extracted from the raster-based simulation outputs.
We applied the Morris analysis to a piecewise linear surrogate model constructed by interpolating the hydrodynamic simulation results. The method was applied to the three input factors: peak discharge Qp, floodplain roughness and channel roughness. The parameter ranges were defined based on the scenario ensemble, with Qp ranging from 318 to 577 m³/s, floodplain roughness from 0.023 to 0.120 and channel roughness from 0.030 to 0.060. A surrogate-based Morris method with 40 trajectories, 8 grid levels, and a grid jump of 2 was used, corresponding to 160 requested surrogate model evaluations for the three parameters. After trajectory filtering, 39 valid trajectories were retained, corresponding to 156 effective Morris evaluations. These response metrics were then related to peak discharge Qp, channel roughness and floodplain roughness using linear surrogate models. The sensitivity analysis based on the surrogate model is consistent with recent flood inundation uncertainty studies that use response-surface or emulator approaches to reduce the computational burden of hydrodynamic uncertainty propagation (Herman et al., 2013; Aitken et al., 2024; Fraehr et al., 2024; Siripatana et al., 2025), as well as with applications of Morris screening to 2D flood-inundation modelling, where roughness and hydraulic boundary conditions are key uncertain parameters (Ullah et al., 2024). However, we will state more explicitly that the μ* and σ indices should be interpreted as exploratory, surrogate-based sensitivity estimates rather than as direct elementary effects of the hydraulic model. In particular, near-zero σ values mainly reflect the approximately additive linear structure of the fitted surrogate models and should not be interpreted as definitive evidence that the full hydrodynamic system lacks nonlinearities or parameter interactions. We will revise Sect. 2.2.3, Fig. 10, and the related Results and Discussion text to present the implemented Morris screening approach more clearly.
Additional references:
Aitken, G., Beevers, L., and Christie, M. A.: Advanced uncertainty quantification for flood inundation modelling, Water, 16, 1309, https://doi.org/10.3390/w16091309, 2024.
Fraehr, N., Wang, Q. J., Wu, W., and Nathan, R.: Assessment of surrogate models for flood inundation: The physics-guided LSG model vs. state-of-the-art machine learning models, Water Research, 2024, https://doi.org/10.1016/j.watres.2024.121202, 2024.
Herman, J. D., Kollat, J. B., Reed, P. M., and Wagener, T.: Technical Note: Method of Morris effectively reduces the computational demands of global sensitivity analysis for distributed watershed models, Hydrology and Earth System Sciences, 17, 2893–2903, https://doi.org/10.5194/hess-17-2893-2013, 2013.
Siripatana, A., Wilson, A. L., and Beevers, L.: Uncertainty quantification for multi-input fluvial flood inundation using GPR- and PCE-based surrogates, Water Resources Research, 61, e2024WR039668, https://doi.org/10.1029/2024WR039668, 2025. [agupubs.on....wiley.com]
Ullah, A., Haider, S., and Farooq, R.: Sensitivity analysis of a 2D flood inundation model. A case study of Tous Dam, Environmental Earth Sciences, 83, 213, https://doi.org/10.1007/s12665-024-11500-w, 2024.
The "weighted exceedance probability of inundation" is one of the manuscript's central outputs, but it is not defined precisely enough to be reproduced or interpreted. As described, it combines binary inundation indicators from simulations that span different return periods and different epistemic choices (FFA confidence intervals, floodplain roughness, channel roughness), each weighted by an "annual exceedance probability". Please provide the governing equation: which simulations enter the aggregation, the weight assigned to each, and the normalization. Moreover, the return period is an aleatory (frequency) quantity, whereas the roughness scenarios and the FFA confidence-interval bounds represent epistemic uncertainty. I 'd ask the authors either to separate the weighting across return periods from the epistemic spread across scenario members) or to justify the chosen aggregation and state clearly what the resulting map represents.
Major comment #5 reply:
Thank you for this important comment, which will help us improve the description of the weighted exceedance probability of inundation (WEP). We will describe the derivation of the WEP more explicitly, particularly because the simulation ensembles combine aleatory uncertainty with epistemic scenario choices related to FFA confidence limits and Manning roughness values. For this reason, we did not assign probabilistic weights to the epistemic scenario members; instead, we averaged them within each RP and applied annual exceedance probability weighting only across considered return periods. WEP represents a probability-weighted inundation occurrence metric derived from multiple return-period classes and hydraulic parameter realizations. In the revised manuscript, we will provide the governing equation and clearly present which simulations enter the aggregation, how they are weighted, how the normalized map was prepared, and how it should be interpreted. We will improve the consistency of the WEP interpretation throughout the manuscript. More detailed description of the WEP calculation is provided below.
Within each return-period class, all epistemic scenario members were treated as alternative realizations with equal weight. The RP contribution was then calculated as the mean inundation occurrence across the epistemic scenario members for that RP class, multiplied by the corresponding annual exceedance probability. The total WEP was obtained by summing these weighted contributions across all considered RP classes. For the scenario ensemble used in our study, the maximum attainable WEP would be 0.112 for cells inundated in all scenario ensembles across all considered RP classes. The WEP is in our case spatially weighted inundation occurrence metrics derived from multiple hydraulic simulations representing different flood frequencies and parameter realizations. This metric should be interpreted as a relative exceedance score, because the considered scenarios are not treated as mutually exclusive events. The resulting WEP therefore represents a probability-weighted measure of inundation occurrence under the selected scenario ensembles. For visualization and relative spatial comparison, the WEP raster was additionally normalized to a 0–1 scale by dividing each cell value by the maximum attainable WEP in the ensemble. A normalized value of 0 indicates no inundation in any simulation, whereas a value of 1 indicates inundation in all scenarios across all considered return period classes. We will revise the Methods and Results sections accordingly and add non-normalized WEP values to Figs. 11 and 12 to retain the exceedance probability interpretation, while the normalized WEP scale will be retained for visualization purposes.
The RCP-driven peak-increase scenarios are currently treated somewhat separately from the main sensitivity framework: they appear in the narrative comparison with the FFA confidence intervals (Sect. 3.1) and in Fig. 12, but not in the comparative sensitivity analysis of Fig. 10. Since the climate signal is a scaling of the peak-discharge factor, it could be incorporated directly into the same comparative framework — for example as an additional level/axis in Fig. 10, or in a single figure that ranks FFA confidence-interval width, channel-roughness change, floodplain roughness, and the RCP-driven increase on the same footing. This would let the reader judge how much the climate-change contribution weighs relative to the other sources, and it would directly support the paper's claim that the FFA uncertainty already "covers" the RCP 2.6/4.5 ranges.
Major comment #6 reply:
We thank the reviewer for this comment. We agree that the RCP-driven peak discharge increases can be compared with the FFA confidence intervals because both act as perturbations of the peak discharge boundary condition. However, we will retain the distinction between hydrological uncertainty, hydrodynamic sensitivity, and RCP-based climate change perturbations. The FFA confidence intervals are derived directly from the observed discharge record and quantify sampling and distributional uncertainty in the design flood estimates. By contrast, the RCP-driven peak discharge increases are externally modelled climate change perturbations that include additional assumptions related to climate projections, downscaling, hydrological transformation, future catchment conditions etc. Accordingly, in the manuscript we consider the RCP scenarios more suitable as a contextual benchmark against which the magnitude of the FFA uncertainty can be compared, rather than as an additional factor treated equivalently to FFA uncertainty and roughness sensitivity. To address the reviewer’s concern, we will strengthen the comparative presentation by explicitly presenting the RCP 2.6, RCP 4.5, and RCP 8.5 peak-discharge increases alongside the simulation resluts and by discussing their relative magnitudes in the Results and Discussion sections. This will allow readers to assess the magnitude of the potential climate change while avoiding the implication that the simplified RCP perturbations and the FFA uncertainty were generated within the same probabilistic framework. We will also emphasize that the proposed workflow remains flexible, as additional climate change scenarios can be incorporated in future applications and their influence can be evaluated explicitly against the hydrological uncertainty represented by the updated FFA confidence intervals.
Specific comments
- Figure 1. The elevation colour scale used in the left (overview) panel does not match the colour bar/legend shown for the right panel. Please harmonize the two.
Reply to specific comment No. 1: Thank you for this notion, we will harmonize the color legend for both panels.
- The floodplain-roughness effect on inundation extent (≈4–6%) is of the same order as the reported validation uncertainty (mean 6%, s.d. 3%; l. 154–156), which you partly acknowledge at l. 265. Please state this explicitly where the S2 results are first presented.
Reply to specific comment No. 2: Thank you for pointing this out, we will mention the reported calibration and validation uncertainty also where S2 results are presented.
- Scenario ensemble S3 compares the 1970s design channel roughness (n = 0.035) with the current calibrated This is a comparison between two temporal states of the channel (a change scenario) rather than a parameter "sensitivity" in the usual sense. Consider adjusting the terminology so that the distinction between epistemic sensitivity (S2) and physical/temporal change (S3) is clear to the reader.
Reply to specific comment No. 3: Thank you for pointing this out and suggesting “temporal states of the channel roughness” to better describe the distinctions between S2 and S3 scenarios. We will adjust the suggested terminology into a revised version of the manuscript.
The references will be updated the revised version of the manuscript by including the suggested references.
References
Apel, H., Merz, B., & Thieken, A. H. (2008). Quantification of uncertainties in flood risk assessments. International Journal of River Basin Management, 6(2), 149-162.
Merz, B., & Thieken, A. H. (2005). Separating natural and epistemic uncertainty in flood frequency analysis. Journal of Hydrology, 309(1-4), 114-132.
Mosquera-Machado, S., & Ahmad, S. (2007). Flood hazard assessment of Atrato River in Colombia. Water resources management, 21(3), 591-609.
Dimitriadis, P., Tegos, A., Oikonomou, A., Pagana, V., Koukouvinos, A., Mamassis, N., ... & Efstratiadis, A. (2016). Comparative evaluation of 1D and quasi-2D hydraulic models based on benchmark and real-world applications for uncertainty assessment in flood mapping. Journal of Hydrology, 534, 478-492.
Citation: https://doi.org/10.5194/egusphere-2026-2602-AC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 160 | 76 | 22 | 258 | 11 | 9 |
- HTML: 160
- PDF: 76
- XML: 22
- Total: 258
- BibTeX: 11
- EndNote: 9
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments:
The work presents an interesting approach on flood mapping, which includes the analysis of hydrological uncertainty and model sensitivity while developing flood hazard maps. The case study is a reach of the Vipava River, a transboundary river flowing from Slovenia to Italy.
The goal of the study is proper and timely, but, in my opinion, the manuscript should be significantly better developed before publication, starting from a more explicit stressing of the novelty, and how this specific work goes beyond the current state of the art. Indeed, the need for considering uncertainty in flood mapping is not a new topic, and there are examples in the literature, which should be addressed in the Introduction to point out knowledge gaps.
The methodological part could be further expanded, also following my detailed comments below, to better guide readers through all the steps needed to produce the maps you show, and also what is needed to evaluate them from a quantitative point of view.
I strongly advise separating the Results from the Discussion, expanding the latter section to better highlight the novelty of the study and what other scholars and stakeholders can learn from your application. Likely in the Introduction, a more in-depth review of the state of the art would support your statements in a more thorough manner, moving from reporting results to actually discussing them from a critical point of view.
Detailed comments:
Introduction
- please avoid excessive use of AI-generated text. The first line of the Introduction could be found in many other similar studies, always with the same phrasing.
- there are no such things as “natural disasters”. As pointed out by UNDRR (https://www.undrr.org/our-impact/campaigns/no-natural-disasters), we have to change the way we communicate, and acknowledge that nature is not causing disasters, while is the presence of humans plays the major role.
- line 27: how can mapping reduce hazard? The link is not fully clear, as, usually, mapping is a good way to reduce risks. Please rephrase this paragraph.
- lines 40-43: please double-check the language. I guess it should read “This…”. But also in this case, the language could be improved
- line 54: what critical scenarios and selected how? It would be good to have more details already in the Introduction, also to better catch the transferability of results/approach
- line 59: additional information on how this roughness is calculated would help in understanding the novelty of the study. For example, have you considered time-changing roughness during the simulation or just values depending on the land use? Even if addressed later in the study, adding some comments in the Introduction would help readers better follow you.
Methods
- please change the title to a more general “Materials and Methods”, as you are also presenting the input data
- line 77-78: where can readers see the flood inundation extent? Can you provide some additional references?
- Figure 1: what is the source of the buildings polygons? I suggest acknowledging data sources more properly.
- please add more references to past flooding events, as the second part of Sec. 2.1 remains a bit too vague in the present version
- would it be good to also have reference years in Table 1, to better understand what the baseline of the model is
- what is the source of Figure 3, and how was the clustering made? In my opinion, lines 120-130 should be expanded, providing more details that ensure the reproducibility of the study
- given that you used HecRas (Sec. 2.2.2), this model should be described in the Introduction, pointing out its pros and cons with respect to the specific case study. HecRas has some limitations that should be acknowledged, and the fact that it was already used in the study area does not suffice to confirm that it is a good choice
- line 144: here you said that roughness was calibrated against water levels. Would it be possible to see such a calibration, and eventually a validation against flooding events? I understand you provided some references, but readers should also be able to follow you without looking at past studies. You can add calibration/validation details as supplementary material.
- line 158: with the current development of IT infrastructures, the computational effort is not a limitation anymore. I suggest deleting this sentence. In addition, there are other free and/or commercial models that perform better than HecRas in terms of computational speed and efficiency (and this comment connects to the need of explaining why HecRas was used).
- from the description reported in Section 2.2.2, I can understand that simulations were done with clear water. Could you please confirm that no morphological changes were considered, and that this reflects the conditions of the study reach?
- line 201: why a threshold of 0.01 m water depth? What’s the rationale behind it?
Results and Discussion
- line 210: please add a reference to the first sentence.
- Figure 4: how where the flood hazard classes defined? I might have missed it.
- line 230: this raises a question about the actual need of using statistical methods. I agree with your point of view, but I suggest deepening the discussion on this aspect, eventually looking at other studies.
- Sections 3.2 and 3.3 need more comments, as roughness plays a major role in water levels rather than in spatial extent, as you also noticed. Do water levels play a role in the study area (e.g., evacuation routes, buildings)
- line 295: Louise Slater (she, so should be “her results”) looked at the contribution of a changing morphology on flood risk, while your model considers clear water, if I understood correctly. Could you please provide more comments on that? Moreover, there is ample recent literature on the influence of sediments and active bed on flood risk, and I suggest deepening the literature review if you wish to address this topic.
- would it be possible to have a clearer Figure 10, eventually using the same colour scale to help compare results?
- line 348-361: this remains a rather vague discussion. I suggest adding some clear, actionable steps to take advantage of flood mapping, also considering the potential use of the affected areas. Appropriate spatial planning should go beyond abandoning areas, and I would like to see more comments on this, also considering the current state of the art.
Conclusions
- line 382: I would use “changes” (or a similar word) instead of “deterioration”, as some changes in roughness could eventually reduce local flood hazard.
- line 388: again, I do not see computational time as a limitation nowadays. It’s possible to address it using adequate models and IT infrastructures.
Data availability
- I was not able to find the Zenodo dataset. As this could be my mistake, could you please help me with this?