Robust decision-making under uncertainty – towards adaptive and resilient flood risk management infrastructure

Flood resilience is increasingly discussed in academia and practice as a complement to existing flood risk management approaches (Fekete, Hartmann, & Jüpner 2020). It is seen as a promising concept to deal with increasingly severe consequences of climate change in general, and with increasing flood risk in particular. The debate on flood resilience is linked to the paradigm shift from flood protection to risk management, which started in Europe after the major river flood events in 1993 and 1995 along the river Rhine (Hartmann, 2012), and has developed over the past decades—pushed by further major fluvial and pluvial flood events (Begum, Stive, & Hall, 2007; Hartmann & Juepner, 2014; Klijn, Samuels, & van Os, 2008; Patt & Jüpner, 2020). However, in flood risk management, the academic debate on resilience is only in its infancy (Jüpner et al., 2018; Vis, Klijn, Bruijn, & Buuren, 2003). Many different definitions exist (Disse, Johnson, Leandro, & Hartmann, 2020). Especially when it comes to specific implementation of resilience, the vagueness of the concept—which is sometimes described as one of its strengths (Baggio, Brown, & Hellebrandt, 2015; Brand & Jax, 2007)—can present a problem. How to design a resilient hydraulic infrastructure? This question has not only technical but also concerned with financial and legal aspects. This becomes more complicated when individual protection measures are put in a context of a larger system, such as a city or a region. When and how is a system such as a city or a region resilient, and how does one measure resilience? These questions are important, not only for construction but also for legal (such as liabilities), financial (budgetary), and political reasons. In other words, resilience is not (yet) a ready-to-use concept in flood risk management. Resilience challenges presumptions of traditional flood risk management in many ways. Resilience implies that more and different stakeholders and actors will be involved in flood risk management than before, such as landowners or spatial planners. Hitherto, flood risk management has been and still is mainly the domain of water management (Hartmann & Driessen, 2017), although the prevalence of the traditional civil engineering approach has been questioned before (van den Brink, 2009). This change is not positive for all actors—affected homeowners, but also city planners or mayors now must deal with questions that were previously outside of their realm of competence and responsibility. From a flood risk management perspective, many questions arise, such as: How does resilience add to and change the existing flood risk management system? How can resilience contribute to a more effective and efficient flood risk management approach? What are the specific advantages of integrating resilience in flood risk management? How can resilience be measured and quantified? Which parameters are most relevant? This special collection brings together contributions from different disciplines that address the specific challenges of implementing resilience in flood risk management. The special collection takes the European debate as a starting point. The European Floods Directive from 2007 (2007/EC/60) pushed the debate on flood risk management and resilience in Europe. This Directive established a legal frame on how countries in the European Union have to deal with flood risk for the first time (Hartmann & Jüpner, 2014). At the same time, the shift toward a risk-based approach raised questions such as "what can happen?" "what must not happen?" and "which security level can be realized at which costs?" Ultimately, flood risk management replaces the paradigm of complete protection against floods with the idea of managing the risk (Grünewald, 2005). This makes approaches to resilience and the reduction of damage potential (vulnerabilities)—not only for river floods but also for pluvial rainfall—important. The first three papers in this collection highlight the necessity and consequences of defining the concept of resilience. In their opinion paper, Fekete, Hartmann, and Jüpner (2020) point out that resilience is indeed a trend that neither can nor should be ignored in flood risk management, rather than it should be embraced—with a caveat: the concept is still not exactly defined and cannot be understood as a panacea for all shortcomings and challenges of flood risk management. So, is resilience merely a new buzzword? Dewulf et al. (2019) discuss the power and impact of defining resilience and therewith unravel the political-normative notion of flood resilience as a concept. This illustrates that, with flood resilience, flood risk management turns much more strongly towards social and political science than ever before. Discussing resilience and its definition is thus a crucial step toward successfully implementing it in flood risk management. In her advanced review, Rodina (2019) reveals different notions of resilience in different disciplines when taking on a broader view of the concept (i.e., not solely focusing on flood risks). She concludes that participation and stakeholder involvement are important components of resilience—a theme that is picked up in the next section of this special collection as well. So, though this article had not been part of the original set-up of the special collection, the conclusions are highly relevant for implementing flood resilience in flood risk management. Resilience requires more action by homeowners and citizens. All papers in this section of this special collection leave no doubt about that. Snel, Witte, Hartmann, and Geertman (2020), however, start with a very critical point of departure, namely that many publications on homeowner and citizen involvement do not seem to question if this involvement is necessary. The paper unravels the arguments that are prevalent in the academic debate for more involvement of homeowners in flood resilience, which supposedly comes along with a shift towards flood governance. Four types of main arguments are identified and discussed in the paper. Kuhlicke and the multidisciplinary team of authors writing on the behavioral turn in flood risk management (Kuhlicke et al., 2020) discuss in detail three underlying assumptions of increased involvement of citizens in flood risk management: the motivations of citizens, the effectiveness of measures, and capacities of citizens. In this vein, Rufat et al. (2020) discuss in their opinion paper need for a more nuanced debate on the link between flood risk perception and risk-mitigating behavior. Attems, Thaler, Genovese, and Fuchs (2020) then provide an overview of the measures that homeowners can take to enhance resilience by using property-level flood risk adaptation (PLFRA) measures. The paper confirms that such measures can indeed be an appropriate addition or complement to flood risk management. The third group of papers in this special collection addresses the issue of measuring and communicating resilience. The need for this comes not only from hydraulic engineering but also from the need to communicate about flood risk with homeowners and citizens, as they have to become more involved (as the earlier papers in this special collection clearly show). Pohl (2020) views resilience from a hydrological as well as a hydraulic engineering perspective and outlines what it means to include the concept in flood risk management. In particular, the contribution points out the difficulties of embedding resilience as a design criterion in hydraulic structures. Pohl concludes that—though the fuzziness of resilience is appreciated by some social scientists as a "boundary concept" (Baggio et al., 2015)—hydraulic engineering and hydrology inevitably require some quantifiable measurement of resilience. Fekete (2019) elaborates in his paper how cascading effects and critical infrastructure matter for flood resilience. It thereby shows how flood resilience with this perspective extends existing flood risk management approaches. Cascading effects of critical infrastructure can substantially affect the measurement of flood resilience. Schmitt and Scheid (2020) discuss the need for better risk communication and discuss approaches by focusing on pluvial floods. In particular, they show how a rainstorm severity index may supersede the communicative capacity of statistical rainfall parameters in the communication with nonexperts especially. In conclusion, the contributions in this special collection show how the concept of resilience is advancing flood risk management but also point out some crucial changes and challenges that flood resilience as a concept brings to the practical implementation of flood risk management. It seems inevitable that homeowners especially will play a much more fundamental role in flood resilience, which raises needs for measuring and communication of flood resilience. Along with this shift in flood risk management, the papers illustrate that flood resilience puts flood risk management in a hitherto unknown domain for many water managers, as flood resilience means embracing social and political science. Embedding flood resilience into flood risk management becomes a balancing act—on the one hand, resilience is a political and normative concept, but on the other hand, it needs to be implemented with concrete steps designed via civil engineering and hydrology. This tension is still unsolved and remains a dilemma in implementing flood resilience.

Flood and coastal erosion risk management has always faced the challenge of decision making in the face of multiple uncertainties relating to the climate, the economy and society. Traditionally, this has been addressed by adopting a precautionary approach that seeks to protect against a reasonable worst case. However, a managed adaptive approach can offer advantages. The benefits include improved resilience to negative changes, enabling opportunities from positive changes and greater cost effectiveness. The absence of clear methods and tools to value adaptive approaches has been recognised as an obstacle to wider adoption. This paper proposes a staged approach to building in adaptive capacity, which systematically analyses uncertainties, identifies opportunities to incorporate adaptability and appraises benefits through the analysis of decision trees. A case study is presented. The methodology is set in the context of appraisal processes used in England by the Environment Agency, based on cost–benefit analysis guidance issued by HM Treasury. The approach is transferable to other situations worldwide, where decision making is based on quantified assessment of costs and benefits. The work should help decision makers fully appraise the benefits of building in adaptive capacity and make the economic and technical case for adaptive flood risk management in an uncertain environment.

9 - Societal Choices in Flood Risk Management, from Individual Responsibility to National Policy

Flood risk management in the context of Integrated Water Resource Management (IWRM) is becoming widely accepted as an approach to improving resilience in light of increasing flood risks due to climate change and other factors. This paper contributes to a better understanding of the governance arrangements needed for effectively implement integrated approaches to managing flood risk. We compare how IWRM and flood risk management have been operationalized within “top-down” and “bottom-up” governance arrangements in the European Union and the United States. We focus in particular on two case study regions, the Catalan coastal region in Spain and the San Francisco Bay Area in California, which have strong similarities in economy, climate, and environmental values, but different institutional settings. Our findings contribute empirical evidence of the need for a balance between “top-down” and “bottom-up” approaches. While the San Francisco Bay Area’s strongly collaborative and participatory approach has generated new connections among flood managers and other stakeholders, the lack of a central entity with the capacity and mandate for on-going coordination and region-wide risk assessments appears to constrain its ability to support integrated and adaptive management. The European Union’s top-down approach and the presence of a central authority at the river basin scale have led to a consolidated regional plan in Catalonia encompassing all phases of flood risk management, but the degree of engagement and opportunities for knowledge-sharing among participants may be more limited.

From out-competing grandmasters in chess to informing high-stakes healthcare decisions, emerging methods from artificial intelligence are increasingly capable of making complex and strategic decisions in diverse, high-dimensional and uncertain situations. But can these methods help us devise robust strategies for managing environmental systems under great uncertainty? Here we explore how reinforcement learning (RL), a subfield of artificial intelligence, approaches decision problems through a lens similar to adaptive environmental management: learning through experience to gradually improve decisions with updated knowledge. We review where RL holds promise for improving evidence-informed adaptive management decisions even when classical optimization methods are intractable and discuss technical and social issues that arise when applying RL to adaptive management problems in the environmental domain. Our synthesis suggests that environmental management and computer science can learn from one another about the practices, promises and perils of experience-based decision-making. This article is part of the theme issue 'Detecting and attributing the causes of biodiversity change: needs, gaps and solutions'.

Flood risk is an increasingly critical concern in urban development as the frequency and severity of floods escalate due to climate change and intensified economic activities. Despite extensive studies on flood risk management (FRM), a significant gap persists in practical frameworks that systematically integrate FRM principles across the entire lifecycle of development projects. To address this gap, this study conducted a scoping review of 27 peer-reviewed articles published between 2014 and 2024, retrieved from Scopus, ScienceDirect, and Google Scholar databases. Using a thematic analysis approach, four major themes were identified: (1) the risk–hazard model, (2) risk assessment theory, (3) project lifecycle theory, and (4) risk management theory. These themes were synthesized to develop a unified conceptual framework that embeds FRM throughout project planning, design, implementation, and operation. The proposed framework emphasizes early risk identification, continuous stakeholder engagement, adaptive management, and interdisciplinary collaboration, enabling proactive integration of FRM into development processes. This novel approach aligns flood resilience with broader urban sustainability and planning objectives, offering a practical tool for policymakers, project managers, and urban planners. Future research should focus on empirical validation and contextual adaptation of the framework across diverse socio-economic and geographical settings to enhance its global applicability.

Adaptive water resources management recognizes the centrality of learning to its effective implementation.However, the literature on learning in Adaptive Management (AM) is disjointed and lacks rigorous treatment.The purpose of this paper is to clarify how learning associated with AM is distinctive and to explain the importance of successfully integrating it into policy design.Because adaptive learning holds a unique position within the adaptive management and assessment process, it should serve as the basis for a systematic redesign of water policy processes.Adaptive environmental assessment and management processes, which highlight the role of adaptive learning, are built upon in this paper.For the purpose of this paper, adaptive learning is defined as "an ongoing process of inquiry that incorporates new knowledge with the aim to continually improve management policies".In order for adaptive learning to live with and profit from nature's dynamism, it is argued that it must be problem-centered, decision relevant, reflective, and inherently transformative.When this kind of learning becomes a fully integrated component of AM, water resource institutions have a greater likelihood of maturing into actual "learning systems".The specific objectives of this paper are to: 1) redefine what adaptive learning is, 2) explain how this particular type of learning is distinctive and 3) outline the role of adaptive learning in policy formation.Regarding policy design, the paramount issue is determining how to respond to the speed, scale, and complexity of the planetary challenges.This paper draws on a number of related areas of research to present propositions for increasing the adaptive capacity of water organizations.These research topics include: demonstrating how adaptive learning should be integrated into policy design; establishing new standards of practice that are based on adaptive learning rather than achieving 'optimal' end-game solutions; using the breakthrough views in order to transform learning into effective action; framing management problems in non-reductionist ways so that nature may serve as an ally and pilot; using shadow networks to inspire innovation, encourage institutional learning, and improve governance rules; and creating a new social reality that is more future-responsive to problems and more hospitable to new ways of thinking about water management.

The occurrence of flood disasters has become a vicious problem in many urban regions. Operationalization of flood resilience has become a significant concern with the increase of urban development and climate change. The objective of this paper is at first to analyze how flood resilience has been framed over the past period of time. Second, this paper has focused to identify the practical difficulties related to the available flood risk assessment and management frameworks for operationalization of flood resilience. The results show that the flood risk and resilience has been framed differently throughout the time, in diverse perspectives but still remains as contested concepts. Evolutionary resilience has been reframed the flood risk management system as an integrated and adaptive system with the recognition of the uncertainties due to the dynamics of socio-spatial interdependencies of flood risk. Therefore, the integration and the adaptive capacity of spatial planning has become a crucial aspect for operationalizing flood resilience which is governed through the knowledge base of the dominant discourses of stakeholders and the way of power and the resource exercised in the context.

To make informed flood risk management (FRM) decisions in large protected river systems, flood risk and hazard analyses should include the potential for dike breaching. ‘Load interdependency’ analyses attempt to include the system-wide effects of dike breaching while accounting for the uncertainty of both river loads and dike fragility. The intensive stochastic computation required for these analyses often precludes the use of complex hydraulic models, but simpler models may miss spatial inundation interactions such as flows that ‘cascade’ between compartmentalised regions and overland flows that ‘shortcut’ between river branches. The potential for these interactions in the Netherlands has previously been identified, and so a schematisation of the Dutch floodplain and protection system is here developed for use in a load interdependency analysis. The approach allows for the spatial distribution of hazard to be quantified under various scenarios and return periods. The results demonstrate the importance of including spatial inundation interactions on hazard estimation at three specific locations, and for the system in general. The modelling approach can be used at a local scale to focus flood-risk analysis and management on the relevant causes of inundation, and at a system-wide scale to estimate the overall impact of large-scale measures.

Adaptive management (AM) is one approach to manage migratory waterbirds, but obstacles to the implementation of AM require adaptive capacities in the management system (rules, institutions, action situations). This study aims to examine the adaptive capacity of participatory goose management in Sweden. Considering the biophysical and institutional context, we analyzed how tangible, individual, and governance assets were associated with technical and social learning. Interviews with informants in the national council for geese, swans, and cranes, and local management groups (LMGs) were conducted, and documentation reviewed. Results revealed evidence of a local preparedness in areas with an LMG. Nevertheless, the study highlighted a need to formalize the evolving system, to consider a more systematic implementation of AM (including regulations allowing for adaptive responses), and to ensure stakeholder acceptance for management tools and visions. The study illustrates the need for a broad set of assets to ensure learning in participatory management.

<p>The global water sector is changing and it is in need of more evidence-based responses of emerging global, regional, national and local challenges. Communities are seeking interventions which achieve multiple benefits and outcomes such as: improved quality of water bodies, reduced greenhouse emissions, reliably delivered water for human use but also some that are rather urgent like: flood-risk management. In order to take into account the environmental, technological, economic, institutional and cultural characteristics of river basins, we need to move from current management regimes towards more adaptive regimes with the use of Nature-based solutions (NbS) instead of traditional 'grey' engineering approaches. Quite a vast amount of tools have been developed throughout the years for achieving this transition. This paper identifies the challenges and opportunities that water professionals face when using these tools in the process of planning NbS. An online tailor-made approach, based on a modified nominal group technique (NGT) and Multi-criteria analysus (MCA) was developed and applied. The NGT-based assessment of these tools consists of two rounds during which participants were asked to reflect first individually, and then collectively about the prerequisites and implications of these tools in the process of planning NbS. The participants are water professionals from the European project Co-Adapt. Here we presented one approach where new scientific methods and practical tools are developed for participatory assessment and implementation of adaptive water management.</p>

Background: Climate change is expected to have a range of health impacts, some of which are already apparent. Public health adaptation is imperative, but there has been little discussion of how to increase adaptive capacity and resilience in public health systems.Objectives: We explored possible explanations for the lack of work on adaptive capacity, outline climate–health challenges that may lie outside public health’s coping range, and consider changes in practice that could increase public health’s adaptive capacity.Methods: We conducted a substantive, interdisciplinary literature review focused on climate change adaptation in public health, social learning, and management of socioeconomic systems exhibiting dynamic complexity.Discussion: There are two competing views of how public health should engage climate change adaptation. Perspectives differ on whether climate change will primarily amplify existing hazards, requiring enhancement of existing public health functions, or present categorically distinct threats requiring innovative management strategies. In some contexts, distinctly climate-sensitive health threats may overwhelm public health’s adaptive capacity. Addressing these threats will require increased emphasis on institutional learning, innovative management strategies, and new and improved tools. Adaptive management, an iterative framework that embraces uncertainty, uses modeling, and integrates learning, may be a useful approach. We illustrate its application to extreme heat in an urban setting.Conclusions: Increasing public health capacity will be necessary for certain climate–health threats. Focusing efforts to increase adaptive capacity in specific areas, promoting institutional learning, embracing adaptive management, and developing tools to facilitate these processes are important priorities and can improve the resilience of local public health systems to climate change.

Private precaution is an important component in contemporary flood risk management and climate adaptation. However, quantitative knowledge about vulnerability reduction via private precautionary measures is scarce and their effects are hardly considered in loss modeling and risk assessments. However, this is a prerequisite to enable temporally dynamic flood damage and risk modeling, and thus the evaluation of risk management and adaptation strategies. To quantify the average reduction in vulnerability of residential buildings via private precaution empirical vulnerability data (n = 948) is used. Households with and without precautionary measures undertaken before the flood event are classified into treatment and nontreatment groups and matched. Postmatching regression is used to quantify the treatment effect. Additionally, we test state‐of‐the‐art flood loss models regarding their capability to capture this difference in vulnerability. The estimated average treatment effect of implementing private precaution is between 11 and 15 thousand EUR per household, confirming the significant effectiveness of private precautionary measures in reducing flood vulnerability. From all tested flood loss models, the expert Bayesian network‐based model BN‐FLEMOps and the rule‐based loss model FLEMOps perform best in capturing the difference in vulnerability due to private precaution. Thus, the use of such loss models is suggested for flood risk assessments to effectively support evaluations and decision making for adaptable flood risk management.

Adaptive management (AM) is the process of implementing land management activities in incremental steps and evaluating whether desired outcomes are being achieved at each step. If conditions deviate substantially from predictions, management activities are adjusted to achieve the desired outcomes. Thus, AM is a kind of monitoring, an activity that land management agencies have done poorly for the most part, at least with respect to ground-based monitoring. Will they do better in the future? We doubt it unless costs, personnel, and future commitment are seriously addressed. Because ecosystem responses to management impacts can ripple into the distant future, monitoring programs that address only the near future (e.g., 10-20 years), are probably unreliable for making statements about resource conditions in the distant future. We give examples of this. Feedback loops between ecosystem response and adjustment of management actions are often broken, and therefore AM again fails. Successful ground-based monitoring must address these and other points that agencies commonly ignore. As part of the solution, publics distrustful of agency activities should be included in any monitoring program.

Rapid urbanization and climate change jointly shape future flood exposure in coastal cities, particularly in the Global South, where urban growth rates are high and adaptive capacity is often constrained. While scenario-based frameworks such as the Shared Socioeconomic Pathways (SSPs) and Representative Concentration Pathways (RCPs) provide consistent narratives of socioeconomic and climatic change, their integration into spatially explicit urban growth modelling and flood exposure analysis remains limited.This study applies the SLEUTH urban growth model to project high-resolution (30 m) urban expansion trajectories up to 2050 for nine coastal urban agglomerations across Africa, Asia, and Latin America under SSP1, SSP2, and SSP5. SSP-based urban population projections were translated into scenario-specific SLEUTH parameters, enabling consistent representation of divergent socioeconomic pathways. Model calibration was supported by harmonized historical settlement data from the World Settlement Footprint and enhanced through a kernel density–based zonation and spatial tiling approach to capture heterogeneous urban–rural growth dynamics across large study areas. Projected urban extents were overlaid with RCP-based coastal, fluvial, and pluvial flood hazard maps to assess future flood exposure at the 50th and 83rd percentiles for a 100-year return period.Results reveal pronounced inter-city and inter-scenario variability in both urban expansion and flood exposure. Four cities (Dar es Salaam, Ho Chi Minh City, Khulna, and Surat) are projected to expand by more than 50% by 2050 under SSP1 and SSP5. Flood exposure is driven by the combined effects of climate forcing and urban development patterns: high radiative forcing scenarios amplify hazard extent, while socioeconomic pathways strongly influence where and how cities expand into flood-prone areas [1]. In several cities, newly developed urban areas exhibit disproportionately higher exposure than existing settlements, particularly for fluvial and pluvial flooding. However, exposure patterns are highly city-specific, underscoring the limitations of generalized assumptions.The findings demonstrate that future urban flood exposure cannot be explained by climate change alone, but emerges from the interaction between climate scenarios and socioeconomic pathways shaping urban growth. By combining SSP/RCP frameworks with spatially explicit urban growth modelling across multiple cities, this study provides comparative insights relevant for scenario-based urban planning and flood risk management in rapidly urbanizing coastal regions of the Global South.[1] Bachofer, F., Wang, Z., Huth, J., Eisfelder, C., Reimuth, A., & Kuenzer, C. (2026). Urban growth prediction along Shared Socioeconomic Pathways (SSPs) for future flood exposure risk assessment: a cross-continental analysis of coastal cities. Anthropocene Coasts, 9(1). https://doi.org/10.1007/s44218-025-00109-