Status: this preprint is open for discussion and under review for Natural Hazards and Earth System Sciences (NHESS).
A Fluvial Flood Risk Model for Quantifying the Benefit of Mitigation Measures under Uncertainty
Mara Ruf,Amelie Hoffmann,and Daniel Straub
Abstract. We present a dynamic probabilistic flood risk model that addresses key challenges in the implementation of integrated flood risk management. These include the need for holistic, large-scale risk assessments that adopt a system-based perspective, and a decision-making framework based on benefit-cost analysis. The proposed model allows for the explicit simulation and dynamic coupling of the flood process components, including downstream flood wave propagation and possible dike failures, in a computationally efficient and data-sparse manner. It enables the consideration of aleatory and epistemic uncertainties in a 2-level Monte Carlo framework. By separating these uncertainties, the model supports robust risk assessments and facilitates the uncertainty-aware evaluation of the benefit of mitigation measures. The model is applied to the Bavarian Danube, demonstrating its ability to estimate the flood risk reduction potential from mitigation measures.
Received: 02 Oct 2025 – Discussion started: 10 Nov 2025
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
Review of manuscript „A Fluvial Flood Risk Model for Quantifying the Benefit of Mitigation Measures under Uncertainty” by Mara Ruf et al., submitted to NHESS
The manuscript presents an interesting extension to the existing flood risk estimation and uncertainty quantification methods, with the aim of quantifying flood risk in economic terms for large river reaches. It applies a concept of simple surrogate models to the required flood model chain (rainfall-runoff simulation, river routing, dike breach/overtopping, inundation and damage estimation). It builds on the probabilistic flood risk assessment concepts developed about 20 years ago, including an uncertainty estimation using the concept of aleatory and epistemic uncertainty. The main advancement of the work is the inclusion of dike failures in a probabilistic manner and to include this in the risk and uncertainty estimation by a2-staged Monte Carlo analysis. Moreover, the work also presents the framework for a cost-benefit analysis of large-scale flood protection measures, like flood retention basins. This is possible because the model concept enables the downstream effect of flood protection measures in a river reach.
The proposed concept is valid and the work takes large scale flood estimation a step further, mainly by including probabilistic dike failures in the risk assessment. The manuscript is properly structured and the work is well presented. Overall, I recommend the publication in NHESS, but some issues should be addressed beforehand:
Emphasize in the introduction that this is the concept presentation, and not the final risk assessment for the river reach.
The inundation modelling is based on non-hydraulic methods, only volume-depth and volume-damage relationships are derived based on GIS analysis. In the light of now-existing fast GPU hydraulic models and AI flood modelling, this appears to be somewhat outdated and needs to be justified. The non-hydraulic nature of the inundation modelling is thus the major limiting factor in this study, because flood dynamics are not captured floods. Highly dynamic flood, where flow velocities and/or inundation duration has an impact on flood damage, cannot be properly captured with the presented approach. Elaborate on this limitation and the resulting limit of applicability of the approach, which is from my point of view large river floods.
The epistemic uncertainty analysis is quite encompassing, but it surely does not capture all possible epistemic uncertainty sources. This is hardly ever possible in practice. Please mention this aspect in the discussion. Also, all the epistemic uncertainty sources should be shown with their assumed distributions and impacts on the risk estimate. Since you don’t show the final risk assessment and cost-benefit analysis for the river reach, but rather demonstrate the concept, this should be included. The readers can also learn from this about the uncertainty range and the impact of the different uncertainties on the risk assessment. Showing only the Sobol Index is not sufficient from my point of view.
The work limits the risk assessment for events with T > 100 a. Please explain why this limitation was taken. This is important, because also less extreme / more frequent floods can cause damage (see e.g. DOI: 10.5194/nhess-9-1033-2009)
I also miss a short discussion of the proposed model concept with existing fully coupled model chains (e.g. Sairam et al. 2021, where only the probabilistic dike breach modelling is missing, but the models are all the actual hydrological/hydraulic models, i.e. no surrogates), or a notion to the increasing number of AI approaches. Your point is here the fully probabilistic approach with uncertainty and the very quick model chain execution. This should be emphasized, but also set in relation to the limitations introduced by the simplified models you developed.
Next to these more general points I made some comments in the attached annotated manuscript.
We developed a flood risk model that estimates the benefit of mitigation measures under uncertainty. The model is computationally efficient and is embedded in a framework that separates the influence of natural from model-related uncertainties. It allows the comparison of (combined) measures and thus provides a robust basis for decision-making. We applied the flood risk model to the Bavarian Danube, Germany, to assess the effectiveness of a flood detention basin.
We developed a flood risk model that estimates the benefit of mitigation measures under...
Review of manuscript „A Fluvial Flood Risk Model for Quantifying the Benefit of Mitigation Measures under Uncertainty” by Mara Ruf et al., submitted to NHESS
The manuscript presents an interesting extension to the existing flood risk estimation and uncertainty quantification methods, with the aim of quantifying flood risk in economic terms for large river reaches. It applies a concept of simple surrogate models to the required flood model chain (rainfall-runoff simulation, river routing, dike breach/overtopping, inundation and damage estimation). It builds on the probabilistic flood risk assessment concepts developed about 20 years ago, including an uncertainty estimation using the concept of aleatory and epistemic uncertainty. The main advancement of the work is the inclusion of dike failures in a probabilistic manner and to include this in the risk and uncertainty estimation by a2-staged Monte Carlo analysis. Moreover, the work also presents the framework for a cost-benefit analysis of large-scale flood protection measures, like flood retention basins. This is possible because the model concept enables the downstream effect of flood protection measures in a river reach.
The proposed concept is valid and the work takes large scale flood estimation a step further, mainly by including probabilistic dike failures in the risk assessment. The manuscript is properly structured and the work is well presented. Overall, I recommend the publication in NHESS, but some issues should be addressed beforehand:
Next to these more general points I made some comments in the attached annotated manuscript.