top of page

Book of


Innovation to Scenario Discovery

[34] War in Ukraine: Resilience requirements under extreme uncertainty 

Author: Leena Ilmola-Sheppard, International Institute for Applied Systems Analysis


"Europe is a stage for war again. There is no way to predict how the war in Ukraine will unfold, how long it will take and will the war escalate. All this will have an impact on the global geopolitical order, the globalization of the economy, the green transition of energy, the speed of climate change and social systems. In this challenging environment for decision making we should be able to define resilience requirements. We are conducting a special study that will define the resilience requirements for national level decision-making. The basic sources of resilience are defined by extreme scenarios’ methods and the robust portfolio modeling. The dynamic requirements of resilience are defined with qualitative systems modeling. The study is conducted by a panel of researchers that covers a wide set of cognitive frames; the cross-disciplinary researchers operate in the Americas, Europe, Africa and Asia. The group has defined five key uncertainties and composed extreme, low probability and high impact scenarios for each of the drivers of uncertainty. The time scope is 15 years and the hypotheses is that the war in Ukraine has ended, but there is no requirement of peace. Resilience analysis is based on the portfolio modeling of a set of scenario specific success strategies, whose feasibility is assessed across all of the extreme scenarios. The final portfolio is able to distinguish the core actions (useful in all of the futures) and environment contingent actions that are dependent on the state of operating environment. The actions portfolio as such does not guide decision-makers, but it reveals examples of the type of actions that are potentially source of resilience, thus the conclusions are driven by the shared features of awareness, adaptation and agility. The study will also compare the feasibility and explanation power of the method in the case of war-generated extreme uncertainty with our previous resilience studies in the context of Covid Pandemics and Energy Transition. "

[73] Likelihood-informed scenario discovery: an approach to explore robustness under probabilistic knowledge and conditional dependencies of deeply uncertain factors   

Authors: Mehmet Umit Taner; Bramka Arga Jafino; Kees van Ginkel, Deltares


"Scenario discovery (SD) is an analytical approach for characterizing and communicating deep uncertainties from large simulation experiments resulting in vulnerabilities or performance failures. SD is often called ‘scenario-neutral’ because scenarios - such as predictions or forecasts - are not prescribed in advance but do result from the analysis. However, implicit assumptions in the design of underlying simulation experiments, such as the bandwidth and independence of the uncertain factors, can influence the outcomes and, in turn, policy recommendations. To circumvent this issue, we propose an alternative SD method that takes into account the relative likelihoods of possible states defined by the underlying factors. The relative likelihoods can be estimated using multiple sources of probabilistic information and propagated using a Bayesian Network. We argue that this method can reduce the influence of a priori assumptions in the analysis and provide transparency. We demonstrate the method for a stylized sea level rise-induced flood risk and house price dynamics model for Rotterdam. We calculate the relative likelihoods based on probabilistic sea-level projections and stakeholders’ belief in RCP likelihoods and measure implementation times. These likelihoods then feed a new SD algorithm that identifies scenarios with likely, undesirable outcomes. The stylized study shows three potential benefits of the new SD method. First, the influence of the implicit experimental designs is reduced and rather replaced by a more transparent and participatory process by including stakeholders’ beliefs and climate science information. Second, the method allows interdependence of the uncertain factors in the analysis. Third, it unravels different types of policy-relevant insights compared to the standard SD, as the identified scenarios here are also those that the stakeholders believe are and reflect on being more likely to happen."

[21] Causal Metamodeling for efficient scenario discovery 

Author: Francisco Pereira, 

Technical University of Denmark


"Simulation is a natural computational tool for exploration of possible futures, given present decisions, exogenous parameters and their uncertainty. However, already with a small number of variables to explore, one very quickly one runs into tractability issues if we want to analyse a representative number of different scenarios. This problem becomes even more relevant when a single run requires minutes or hours, as happens with many complex simulation tools [1]. A response from the Machine Learning field, has been to train metamodels (or surrogate, response surface models) to allow for exploration orders of magnitude faster [2], but these suffer from a fundamental challenge: they replicate the input data distribution very well, but fail out-of-distribution [3]. In other words, if one wants to study scenarios very different than those n the original simulation runs, they are not trustworthy. This happens because the large majority of these models do not reflect the internal cause-effect relationships embedded in the simulation tools. In this presentation, I show and discuss my recent work on causal metamodeling, which combines three recent developments in machine learning: causal inference, active learning and algorithmic alignment. Causal metamodeling allows for out-of-distribution generalisation together with faster inference, and is therefore a very promising direction towards efficient multi-variate scenario discovery considering deep uncertainty. 1. Balmer, M., Meister, K., Rieser, M., Nagel, K., & Axhausen, K. W. (2008). Agent-based simulation of travel demand: Structure and computational performance of MATSim-T. Arbeitsberichte Verkehrs-und Raumplanung, 504. 2. Kasim, M. F., Watson-Parris, D., Deaconu, L., Oliver, S., Hatfield, P., Froula, D. H., ... & Vinko, S. M. (2020). Building high accuracy emulators for scientific simulations with deep neural architecture search. arXiv preprint arXiv:2001.08055. 3. Wang, R., Yi, M., Chen, Z., & Zhu, S. (2022). Out-of-distribution Generalization with Causal Invariant Transformations. arXiv preprint arXiv:2203.11528."

[68] Acknowledging deep uncertainty in urban nexus modelling

Author: Dr. Rebecka Ericsdotter Engtström,

KTH Royal Institute of Technology


"The water-energy-food (WEF) nexus has grown into an impactful field of sustainability research over the past decade, developing quantitative modelling of the integrated resource systems that our societies fundamentally depend on. While a majority of ‘nexus’ research to date has focused on national and transnational system interactions, shifting the focus to the local scale holds potential to reframe the WEF-nexus “as a driver for policy change that structures alternatives” (Artioli et al. 2017). However, city-scale data is typically less comprehensive or systematically made available compared to national ditto (Horta & Keirstead, 2017). Furthermore, time-series data is often lacking to comprehensively study change over time. But, even if historical data was available to a larger extent, it would provide limited information to future focused assessments of rapidly changing cities that depend on WEF supplies from global supply chains, that in turn are vulnerable to multiple global sustainability challenges. This deep uncertainty embedded in any modelling of the future WEF nexus of cities has received limited attention in urban nexus research. This project reviews urban nexus literature from the past 5 years to map how uncertainty – and in particular deep uncertainty – is acknowledged in this body of work. A thematic analysis shows how both study scopes and robustness of results become limited in the face of data challenges. Taking inspiration from DMDU approaches, the project recommends refinements of methods and practices, to make urban nexus studies provide the decision-support that it holds the promise to do. "


[42] Exploring normative uncertainty in adaptive mitigation policies: a methodological approach to incorporate morally informed rival framings in Integrated Assessment Models 

Authors: Palok Biswas; Jazmin Zatarain Salazar; Jan H. Kwakkel, Delft University of Technology


The Integrated Assessment Models (IAMs) used to design climate change mitigation policies are beset with scientific and normative uncertainties. The normative component of IAMs is inherently complex and fraught with deep uncertainty. Adaptive mitigation strategies are more robust in the face of deep uncertainty, and previous studies demonstrated the effectiveness of adaptive strategies in managing scientific uncertainty using temperature data as feedback. Besides temperature, future population growth plays a crucial role in long-term economic growth, thereby determining future emissions and the number of people vulnerable to climate change. Thus, future demographics are a significant source of normative uncertainty. However, normative uncertainty in IAMs is largely overlooked, and the impact of demographic data on normative uncertainty is currently unexplored. This study builds on the Regional Integrated Climate-Economy (RICE) IAM by reformulating the optimization problem into a closed-loop control problem using the Evolutionary Multi-Objective Direct Policy Search (EMODPS) framework using temperature and population as feedback. Next, we explore four rival framings of the optimization problem drawing on four different ethical principles: utilitarian, prioritarian, sufficientarian, and egalitarian. We demonstrate how these rival framings significantly affect mitigation pathways -- each with different normative consequences -- compared to the human-Earth system uncertainties. We argue that it is vital to acknowledge normative uncertainty in IAMs besides scientific uncertainty. Therefore, we introduce a methodology to incorporate morally informed rival framings in IAMs to explore normative uncertainties.

[44] Making normative assumptions explicit: Managing uncertainty in model-based climate change planning with rival ethical framings

Authors: Damla Akoluk; Jazmin Zatarain Salazar ; Alexander Verbraeck, Faculty of Technology, Policy and Management, TUDelft


"Justice in climate change planning has gained more attention in the last decade due to existing inequalities between various sectors, regions, and generations. Conventional model-based decision support tools do not consider the dynamic uncertainty of values in climate change futures. Particularly, Integrated Assessment Models (IAMs), which are central to inform climate policy, do not dynamically adapt over time to cope with changing temperature, emissions, or welfare inequality, and make premature aggregations which cause unnecessary loss of information, and make unrealistic assumptions about the expected damages across actors. Many of these aggregations are a source for contestation over climate justice. This study sheds light on how our aggregation choices within IAMs (i.e. across actors, time and space), and how our normative assumptions impact the conclusions that we reach. Questioning the philosophical underpinnings of climate change policies entails expanding our analysis to a suite of distributive justice principles. To this effect, this study explores different distributive justice principles in integrated assessment models using a descriptive approach. It results in a classification of the five most common ethical principles: Utilitarianism, Rawlsianism, Egalitarianism, Prioritarianism, and Sufficientarianism. We operationalize these justice principles by expanding the Regional Integrated Climate-Economy (RICE) model into a Many Objective Robust Decision Making (MORDM) framework to observe the trade-offs between emission rates, overshooting years, and distributive justice assumptions under the dynamic and uncertain features of policy design. In addition to these arrangements, we use the Evolutionary Multi-Objective Direct Policy Search (EMODPS) framework to present the feedback from both global temperature and welfare changes that have the power to fundamentally alter the course of mitigation policies. Initial results suggest that, dispersion around the mitigation changes drastically within the rival ethical frameworks. These results also provide additional insights into the trade-offs between multiple distributive justice metrics, which rely on dynamic adaptive policy strategies. We show that changing the justice perspective, i.e. the underlying assumptions for the ethical principles embedded into the model, can dramatically change the plausible climate policies and significantly affect the distributional outcomes across actors."

[98] A new approach to parameterizing the effectiveness, costs, and benefits of multi-sectoral and multi-national decarbonization strategies

Authors: Nidhi Kalra; Edmundo Molina-Perez; James Syme; Fernando Estevez, RAND Corporation; Tec de Monterrey


Coordinated sector-specific decarbonization strategies – in energy, transportation, industry, waste, agriculture, livestock, and forestry sectors -- are needed to turn broad national decarbonization goals into a carbon neutral future, while also meeting Sustainable Development Goals. However, the heterogeneous and deeply uncertain estimates found in the literature of the applicability effectiveness, costs, and benefits of specific strategies pose a serious challenge to identifying and modeling specific strategies in a DMDU analysis. This talk describes a process to parametrize sectoral transformations in the context of Latin America and the Caribbean (LAC) to achieve both net zero emissions and net positive socioeconomic development benefits. The approach support exploratory modeling and allows researchers to identify more robust combinations of transformations across sectors and countries.

[33] Exploring Policy Mixes for Low-carbon and Just Energy Transitions Systems: An Australian case

Authors: Angela M. Rojas-Arevaloa, Fjalar de Haana, Seona Candyb, Greg Folientea, Lu Ayea, The University of Melbourne


The electricity system, which powers critical services that underpin our modern economy and society, needs to achieve its function sustainably. But there is more to energy sustainability transitions than reducing emissions. A costly energy Transition could be detrimental to the most vulnerable and risky for a population on the brink of hardship. Therefore, Energy Justice should be considered to avoid compromising end-users energy security and affordability. Going beyond the low-carbon narrative and avoiding unintended consequences requires understanding the complexity of socio-technical systems for energy provision and the many variables and uncertainties of a Transition. Computer models are essential to inquire into complex systems and aid decision-making. This research simulates the electricity system in Victoria, Australia---one of the most polluting systems in the country, with an agent-based model and explores the most appropriate policy mixes to achieve three main types of Transitions: i) Low-carbon Transition, which is a technologically driven transition, guided by the Victorian government targets to limit global warming below 1.5°C and 2°C; ii) Just Transition, which is a socially driven transition, enabling new markets and institutions to prioritise energy security and affordability; or iii) Sustainability Transition, which is a combination of the Low-carbon and Just Transitions. The significant role of computer simulations and exploratory modelling methodologies in decision-making under uncertainty is key to getting the Transition's type and timing right. Findings show a variety of incremental to radical policy mixes achieving Sustainability Transitions, indicating that Low-carbon and Just Transitions, if well managed, are achievable and compatible.

[3] post-MORDM: mapping policies to synthesize optimization and robustness results for decision-maker compromise

Author: Nathan Bonham; Joseph Kasprzyk; Edith Zagona, University of Colorado Boulder Civil, Environmental and Architectural Engineering department; Center for Advanced Decision Support for Water and Environmental Systems (all authors)


This paper introduces post-MORDM, a decision-support framework that augments Many Objective Robust Decision Making (MORDM). MORDM often creates an intractable number of environmental management policies, characterized by decision variable, objective, and robustness values. This large number of policies inhibits decision support, causing disagreements among decision-makers. Post-MORDM addresses these challenges via the Self-Organizing Map (SOM), synthesizing MORDM data as layers organized in a map-like coordinate system. Post-MORDM is a structured platform that encourages negotiation and compromise. It uses the SOM to cluster policies, discover salient characteristics, and assess cause-effect relationships between decision-maker choices (i.e. decision variable values) and performance (objective and robustness values). Post-MORDM is demonstrated with a case study of two illustrative decision-makers for reservoir operation policy in the Colorado River Basin, USA. We use post-MORDM to communicate tradeoffs between storage and delivery objectives, relate tradeoffs to shortage policies, and identify mutually feasible policies.

[32] How can open-source modelling support robust global climate governance?

Author: Agnese Beltramo, Will Usher: division of Energy Systems, KTH Royal Institute of Technology (Stockholm, Sweden); Taco Niet: Delta-E-plus Research Group, Simon Fraser University (Vancouver, Canada).


"Models are relevant for supporting decision-making, particularly in the field of climate governance. The Intergovernmental Panel on Climate Change (IPCC) is relying on Integrated Assessment Modelling (IAM) to advance the knowledge and understanding of future possible climate change scenarios and inform the Conference of the Parties. However, these tools are characterised by deep uncertainties, which are often difficult to address due to the high computational requirements of the tools. The Global Least-cost User-friendly CLEWs Open-Source Exploratory model (GLUCOSE) provides a complementary approach to IAM. Thanks to its highly aggregated structure, its light computational requirement, and its open-source nature, GLUCOSE can be used for stochastic modelling, to explore deep modelling uncertainties and support policy makers in their investigation and better understanding of the long-term dynamics behind integrated resource systems and their contribution to climate change. This study investigates the quality and validity of GLUCOSE by exploring its solution space and testing its robustness against uncertainties affecting possible policy pathways towards net-zero emissions, thus exploring a wide range of sectoral policy combinations that can help reducing global greenhouse gas (GHG) emissions, while also limiting the depletion of natural resources. This study showcases to decision-makers the wide range of policy options that GLUCOSE can help exploring, and provides an open and transparent methodology for identifying robust, cross-sectoral solutions to address climate change via quantitative modelling. Finally, this study demonstrates that multi-sectoral policy approaches can produce resource-effective solutions to meet defined demands in the model and at the same time reach low emissions for the entire system. Therefore, it identifies solution spaces that are realistic and feasible, while ensuring sustainability for the interlinked resource system."

[25] Seizing the means of distribution: Using Rawls’ framework of justice to evaluate Pareto Optimal policies for transboundary water resource allocation under deep uncertainty

Author: Sahiti Sarva; Jazmin Zatarain Salazar; Jan Kwakkel; Neelke Doorn; Delft University of Technology


Transboundary water resource allocation often causes water shortages and floods in upstream and downstream areas due to the uncertainty in rainfall, runoff volumes, and water availability. Techniques such as multi-objective evolutionary algorithms account for uncertainty in input parameters and aid policymakers in generating optimal allocations to every actor that benefits from the transboundary water body. However, these algorithms generate hundreds of Pareto optimal solutions, thus leaving analysts searching for additional criteria to evaluate them. We propose Rawls' conceptualization of justice: a combination of fairness and stability, as a framework to evaluate Pareto optimal policies in multi-objective resource allocation problems. We used the Susquehanna River basin case study to generate Pareto Optimal policies. Subsequently, we evaluated their fairness using utilitarianism and egalitarianism. We evaluated their stability using fallback bargaining. Given context-specific thresholds, we found a desirable Pareto space where policies could be considered just according to both the fairness and stability principle. We also found that fallback bargaining guarantees the selection of a stable policy if all participating actors agree to a social contract. Using Rawls’ framework as the basis for additional criteria resulted in an approach for evaluating the justice of Pareto Optimal policies. This approach can also be employed within other resource allocation multi-objective models that seek to identify just policies under deep uncertainty.

[16] Adaptive planning capacity building through a community of practice

Author: Abel Immaraj; Sam Skinner; Shane Tyrrell; Alexandra Hare; Andrew Warren

Aurecon; Deltares 


"Nothing is fair or certain in the heat of battle. The Second World War was a period of global uncertainty in military supply chains which accelerated the development of Operations Research to assist strategic and logistical decision-making in deep uncertainty with potential catastrophic consequences. As computing capacity grew, it enabled the application of a range of methods incorporating statistics, stochastics, optimisation, real-time control, iterative algorithms to complex processes such as financial investments, rocket guidance systems and even human behaviour. Adaptive approaches are helping water utilities in Australia and New Zealand deal with volatility, uncertainty, complexity and ambiguity (VUCA) to support decision-making on strategy, planning, operation and investment. VUCA induces decision paralysis (perplexity for decision-makers) or path dependency (business-as-usual solutions) or option lock-in (pick one solution). Adaptive approaches enable alternative pathways that keep future options open, linking current decisions to future choices, and increase flexibility to respond to future information, trends and shocks. An adaptive approach to planning is a measured and considered process that does not replace but augments good decision-making. Extensive change, such as that facing Aotearoa New Zealand as we aim to improve the safety, quality, resilience and accessibility and performance of three waters, creates opportunities and challenges, debate on risks and trade-offs, which includes the obsolescence of information, systems, products and services, which affect long lived linear infrastructure planning. Traditional planning approaches are ill-equipped to account for spatial, temporal and optional complexity, leading to suboptimal solutions. The challenge for the water sector across the world is to address deep uncertainty and complexity arising from mega trends and forces of change such as droughts and floods, ever increasing competition for water and associated costs of infrastructure, contaminants of concern and their impact on the environment. This paper draws on our experience in Australia where a Community of Practice was formed with membership of Urban Utilities, Seqwater, Melbourne Water, Aurecon, Deltares, Institute for Sustainable Futures, and in collaboration with the Water Services Association of Australia (WSAA). Through the year-long ‘learning project’ with presentations, workshops and discussion sessions, a body of knowledge (BoK) was co-created to assist planners understand the approach and methodology to implement adaptive planning. The BoK includes a guide, a series of Q&A and utility case studies describing how the approach was adopted. A collaborative approach was initiated to build on the collective strengths of utilities, researchers and consultants in creating the BoK, which is being converted to a living BoK for members to continue to share and update their experiences. This paper provides an overview of the key learnings and explores how we can draw learnings from the opportunities that such communities can create and look to explore ways/communities of a similar nature may be established to gain similar collaborative and outcomes based contributions to ensure we continue to Restoring and preserving the balance between water, the environment and people. "

[50] Examining preferences for sea-level rise adaptation and mobility in Miami-Dade County, FL, USA.

Author: Nadia A. Seeteram; Mahadev Bhat; Brett F. Sanders; Jochen E. Schubert; and Katharine J. Mach: Earth and Environment and Institute of the Environment, Florida International University; Civil and Environmental Engineering, UC Irvine; Urban Planning and Public Policy, UC Irvine; Environmental Science and Policy, Rosenstiel School of Marine and Atmospheric Science, University of Miami; Center for Ecosystem Science and Policy, University of Miami, Coral Gables


"Nothing is fair or certain in the heat of battle. The Second World War was a period of global uncertainty in military supply chains which accelerated the development of Operations Research to assist strategic and logistical decision-making in deep uncertainty with potential catastrophic consequences. As computing capacity grew, it enabled the application of a range of methods incorporating statistics, stochastics, optimisation, real-time control, iterative algorithms to complex processes such as financial investments, rocket guidance systems and even human behaviour. Adaptive approaches are helping water utilities in Australia and New Zealand deal with volatility, uncertainty, complexity and ambiguity (VUCA) to support decision-making on strategy, planning, operation and investment. VUCA induces decision paralysis (perplexity for decision-makers) or path dependency (business-as-usual solutions) or option lock-in (pick one solution). Adaptive approaches enable alternative pathways that keep future options open, linking current decisions to future choices, and increase flexibility to respond to future information, trends and shocks. An adaptive approach to planning is a measured and considered process that does not replace but augments good decision-making. Extensive change, such as that facing Aotearoa New Zealand as we aim to improve the safety, quality, resilience and accessibility and performance of three waters, creates opportunities and challenges, debate on risks and trade-offs, which includes the obsolescence of information, systems, products and services, which affect long lived linear infrastructure planning. Traditional planning approaches are ill-equipped to account for spatial, temporal and optional complexity, leading to suboptimal solutions. The challenge for the water sector across the world is to address deep uncertainty and complexity arising from mega trends and forces of change such as droughts and floods, ever increasing competition for water and associated costs of infrastructure, contaminants of concern and their impact on the environment. This paper draws on our experience in Australia where a Community of Practice was formed with membership of Urban Utilities, Seqwater, Melbourne Water, Aurecon, Deltares, Institute for Sustainable Futures, and in collaboration with the Water Services Association of Australia (WSAA). Through the year-long ‘learning project’ with presentations, workshops and discussion sessions, a body of knowledge (BoK) was co-created to assist planners understand the approach and methodology to implement adaptive planning. The BoK includes a guide, a series of Q&A and utility case studies describing how the approach was adopted. A collaborative approach was initiated to build on the collective strengths of utilities, researchers and consultants in creating the BoK, which is being converted to a living BoK for members to continue to share and update their experiences. This paper provides an overview of the key learnings and explores how we can draw learnings from the opportunities that such communities can create and look to explore ways/communities of a similar nature may be established to gain similar collaborative and outcomes based contributions to ensure we continue to Restoring and preserving the balance between water, the environment and people. "

[89] A cognitive modeling approach for understanding computational intelligence-human interactions in deeply uncertain decision-making environments

Author: Isaac Molina; Edmundo Molina-Perez, Tecnologico de Monterrey


Decision-making in complex and uncertain environments is a high-level individual or group process that depends on various cognitive, psychological, and social mechanisms, such as perception, attention, memory, abstract thinking, and debate. Deeply uncertain environments require series of decisions to be made, with each decision depending on rapidly changing information, complex computational intelligence tools (CITs), detailed data analysis tasks and multiple agents’ perspectives. This talk discusses results of a series of experiments that combine behavioral experimentation and neuroscientific methods to develop a cognitive model that quantitatively describes the impact that CITs have on decisions being made in uncertain environments. We find that, by themselves, experimental conditions, participants characteristics or CIT’s features, do not sufficiently influence cognitive or neurological processes to change decision outcomes. However different combinations of these parameters lead to a diverse set of results such as improving decision outcomes in experiments, but also to overreliance, mistrust, and model rejection. We discuss how this integrative approach can be used to develop formal cognitive models of decision making under deep uncertainty.

[90] Using Bayesian Networks for developing exploratory models in data scare contexts

Author: Edmundo Molina-Perez; Adolfo Javier De Unanue-Tiscareño, Tecnologico de Monterrey


Decision-making in complex and uncertain environments is a high-level individual or group process that depends on various cognitive, psychological, and social mechanisms, such as perception, attention, memory, abstract thinking, and debate. Deeply uncertain environments require series of decisions to be made, with each decision depending on rapidly changing information, complex computational intelligence tools (CITs), detailed data analysis tasks and multiple agents’ perspectives. This talk discusses results of a series of experiments that combine behavioral experimentation and neuroscientific methods to develop a cognitive model that quantitatively describes the impact that CITs have on decisions being made in uncertain environments. We find that, by themselves, experimental conditions, participants characteristics or CIT’s features, do not sufficiently influence cognitive or neurological processes to change decision outcomes. However different combinations of these parameters lead to a diverse set of results such as improving decision outcomes in experiments, but also to overreliance, mistrust, and model rejection. We discuss how this integrative approach can be used to develop formal cognitive models of decision making under deep uncertainty.

[14] Requirements for combining DMDU and stakeholder participation

Author: Karoline Führer; Jan Kwakkel; Els van Daalen; Etiënne Rouwette, Delft University of Technology; Delft University of Technology; Delft University of Technology; Radboud University Nijmegen


"Socio-technical transitions are complex societal challenges, involving many stakeholders and enormous uncertainty in the developments of many parts of that system. Transitions pose the challenge of a changing decision environment and evolving perceptions among actors during policy processes. The two research fields of decision-making under deep uncertainty (DMDU) and modelling with stakeholders aim to offer model-based support for multi-stakeholder decision making. Their shared aim and differing angles offer a basis for benefiting each other. In order to create transformative agency, a range of relevant actors needs to be mobilized. Decisions can be implemented with less conflict and more success when multiple stakeholders drive them. A high level of participation is required to give stakeholders a sense of ownership or even a mandate to act. Modelling with stakeholders allows for integrating scientific information and local knowledge and enables participants to gain deeper understanding of the system. However, methods for modelling with stakeholders fall short when it comes to integrating deep uncertainty. Decision-making under deep uncertainty provides a variety of analytical approaches for analyzing the consequences of uncertainty. However, there is a lack of involvement of stakeholders in the analysis. Most steps of the process are executed by analysts, then the final results and interpretation are communicated to the decision maker. The analysis is usually catered to one decision maker, while in reality there is a variety of relevant actors. They all have their own view on the system as well as different decision drivers and decision impact. This study presents a list of requirements for using models in supporting multi-stakeholder deliberation under uncertainty, based on a literature review of participatory modelling methods and DMDU approaches. The first part considers the impact of uncertainty and requirements for addressing it. Typologies such as types of uncertainties relevant for strategic decision-making, or frameworks for assessing and communicating uncertainty in model-based analysis studies provide us with the necessary concepts. In the second part, we focus on the integration of stakeholder participation. It is important that the participatory modelling process is transparent, follows a just procedure, and yields a high quality model output. The requirements for addressing uncertainty as well as the ones for stakeholder participation are operationalized into an evaluative framework. This can be used to evaluate the efficacy of new methods to bridge the gap between deep uncertainty and participation. "

[80] Data to Decision; Uncertainty in Environmental Data Science

Author: Katherine Wright, Lancaster University, UK


"Environmental data science is an emerging research area, combining techniques from environmental science, computer science, and statistics, to make sense of the complex changes occurring in our natural environment. Alongside the ambiguities created from the collaboration of researchers with different disciplinary backgrounds, other types of uncertainty occur during scientific research – some inherent due to natural variability (aleatory), some due to limited knowledge (epistemic). Understanding the many sources of uncertainty, along the data to decision pipeline, will aid provision of robust scientific evidence to underpin decision-making. Grounded in data collected from interviews and focus groups, this work will discuss the uncertainties experienced by experts from environmental science, computer science, and statistics. A new typology of uncertainty for environmental data science will be presented. "

[99] Climate Change Impacts on Wildlife Habitat Suitability: Developing a Resilient Management Strategy

Author: Hilda Zamora-Maldonado; Sophie-Avila Foucat, Tecnológico de Monterrey; Instituto de Investigaciones Económicas, UNAM


For wildlife stakeholders, conserving species in the face of climate change can be a complex endeavor. As conditions change on the ground, the landscapes that species use now may be different from the ones are available or they choose in the future. Exploratory analysis for habitat suitability in the future is essential to inform management and conservation strategies of wildlife species today. This study implemented the ecological niche modeling algorithm Maximum Entropy (Maxent) to examine present climatic niche and potential distribution of bighorn sheep (Ovis canadensis) populations in Mexico and climate change effects on its potential future habitat suitability and distribution. Maxent was built with 1,158 georeferenced occurrence records and uncertainty of climate change was addressed by exploring possible scenarios under four Representative Concentration Pathways (RCPs) (2.6, 4.5, 6.0 and 8.5) and four different General Circulation Models (GCM) in a time span of 20 years (2041-2060). The results showed that current protection strategies on public and private lands are inadequate for the conservation of target specie given projected climate impacts. Therefore, exploratory analysis may be most useful in supporting resilient management strategies by directing attention toward locations that may be suitable for wildlife management under future conditions. This analysis can be extended by considering multiple species, exploring strategies that employ an array of environmental policy options, and potential for learning that ease the reframing and refocusing of goals with which the wildlife management community and stakeholders may need to engage in the face of climate change.

[54] Robust Climate Adaptation Policies for Subsistence Agriculture in Nepal

Author: Nicolas Choquette-Levy; Frank Errickson; Michael Oppenheimer; Klaus Keller, School of Public and International Affairs, Princeton University; School of Engineering, Dartmouth University


"The evolving nature of climate risks represents new challenges for the world’s 500 million subsistence farmers.1,2 Farmers make livelihood decisions in deeply uncertain environments regarding weather conditions, economic markets, and the social-political systems in which they operate. In turn, uncertainty in how farmers make such decisions challenges the ability of policymakers at multiple governance scales to achieve key objectives such as maximizing economic growth, minimizing inequality, ensuring food security, and limiting rates of outmigration. Farming households and policymakers must therefore make climate adaptation decisions under exogenous sources of deep uncertainty, e.g. unknown probability distributions of climatic and economic states of the world, and endogenous deep uncertainty resulting from misalignment or misunderstanding of objectives among key stakeholders in a system. We create an integrated robust decision-making – agent-based model (RDM-ABM) framework to analyze how both types of deep uncertainty affect the robustness of potential climate adaptation policy interventions in the agricultural sector of Nepal. Nepal is a particularly important case study for this topic, owing to its high exposure to multiple climate risks,3–5 high dependence on subsistence agriculture and remittances from outmigration,6,7 and influx of international development financing.8 Previous work has typically applied robust decision making (RDM) and dynamic adaptive policy pathways (DAPP) frameworks to investigate “hard” infrastructure decisions under climate uncertainty.9–13 More recent work evaluates the robustness of farming strategies to deep climate and economic uncertainty,14,15 and assesses the prospects for reconciling resource tradeoffs between agriculture and other economic sectors under future climate scenarios.16,17 We contribute to this growing literature by analyzing (i) potential tradeoffs that emerge from differences between smallholder farmer objectives and those of policymakers at multiple scales regarding climate adaptation in Nepal's agricultural sector, (ii) the importance of different exogenous and endogenous sources of uncertainty in determining the likelihood of meeting these various objectives, and (iii) how different potential policy interventions - including investing in irrigation infrastructure, providing cash transfers, subsidizing weather-based crop insurance, and supporting migration remittances - compare in their robustness to these sources of uncertainty. Our work expands on a previous analysis of smallholder farmer livelihood decisions under climate risk,18 which used an ABM calibrated with survey data on Nepali farmers’ livelihood decisions and risk preferences. This model allows us to explore how farmers who are heterogeneous in their financial resources, risk preferences, and access to information make livelihood decisions under various climatic, economic, and policy states of the world. We also incorporate endogenous feedbacks between policymakers at multiple levels of government, international development agencies, and farming households. We apply this framework to investigate how real-world complexities of the policymaking process – including potentially competing objectives, misalignment between scales of policymaking, and lags in policy implementation – affects the robustness of proposed climate adaptation policies for subsistence agriculture in Nepal. References 1. Howden et al. (2007) 2. Lowder et al. (2016) 3. Karki & Gurung (2012) 4. MoALMC (2018) 5. Government of Nepal (2019) 6. Nepal Central Bureau of Statistics (2014) 7. Ghimire et al. (2020) 8. World Bank Group (2018) 9. Groves & Lempert (2007) 10. Keller et al. (2008) 11. Hallegatte & Lempert (2012) 12. Haasnoot et al. (2013) 13. Kwakkel et al. (2015) 14. González et al. (2020) 15. Jafino et al. (2021) 16. Knox et al. (2018) 17. Yoon et al. (2021) 18. Choquette-Levy et al. (2021)"

[72] How is Europe preparing for uncertain sea level rise? 

Author: Sadie McEvoy; Marjolijn Haasnoot, Deltares; Utrecht University


"Sea level rise is a growing threat to most coastal countries, but there is no systematic overview of how different countries plan for it and handle uncertainty in sea level rise. For example, some countries have national policies and sea level rise scenarios, while others leave coastal management to regional or even local planning. To develop an overview of how countries in Europe are preparing for sea level rise, an expert survey was conducted, covering topics such as time horizons and sea levels used in planning. Additionally, the study focused on how uncertainty in sea level rise is handled in the country and whether countries are considering high-end scenarios. The survey results indicate roughly half of countries in Europe use at least one approach for handling uncertainty in sea level rise planning guidance, including using multiple scenarios, planning cycles and adaptive plans. Respondents in many of these countries indicated that more than one approach is used. However, in almost half of countries respondents reported that there is no handling of uncertainty in SLR planning, and in almost every country. Results also suggest that uncertainty language and methods are not widely understood in the field of sea level rise planning and that approaches for uncertainty handling appear not to be formalized in many countries’ planning documents. Survey results on whether countries are preparing for or considering high-end sea level rise scenarios, indicate than less than half of countries are either using or exploring these scenarios in planning. Only a small minority of countries are considering sea levels above 1m or time horizons beyond 2100. Similar to planning under deep uncertainty, respondents from some countries appeared to conflate high-end sea level rise with RCP8.5, further suggesting that uncertainty may not be well understood in the field. This presentation will summarize the main findings of the research and share possible implications of disparate approaches to sea level rise planning and uncertainty handling."

[1] Adapting scenario planning to create an expectation for surprising extremes 

Author: James Derbyshire; Ian Belton; Mandeep Dhami; Dilek Onkal, Middlesex University; Northumbria University


Scenario Planning (SP) is designed to aid anticipation of surprises in the form of highly uncertain and extreme changes to an organisation’s external environment. Common forms of SP employ the concept of plausibility because probability is considered too constraining on the range of futures being considered. However, like probability, plausibility is founded on knowledge derived from past occurrences or presently available information. It may therefore also constrain the full potential extremity of the future. This is problematic because the extremity of a potential outcome can extend way beyond what was conceivable earlier in time based on information available at that point. We test a methodological adaptation to the common Intuitive Logics (IL) approach to SP. The proposed method is designed to allow extreme potential futures to come to mind by focusing participants’ attention on surprising extremes.

[65] Achieving a climate-resilient electricity grid: beyond optimality, searching for robustness & embracing uncertainty 

Author: Liyang Wang; Andrew D. Jone; David Anthoff, Energy & Resources Group, UC Berkeley; Lawrence Berkeley National Laboratory 


The changing climate conditions and extreme weather we are experiencing today expose our energy infrastructure's fragility. Deep uncertainties around climate change projections and future socioeconomic conditions increase the complexities of energy infrastructure planning. Electricity grid planners are confronted with competing objectives such as achieving cost-effectiveness, integrating low-carbon energy sources, maintaining reliability, and ensuring equitable access. The competing objectives layered with the high level of uncertainty pose a myriad of difficulties in making robust long-term planning decisions. In this exploratory study, we operationalize one main concept in the DMDU literature, robustness, to a simplified capacity expansion planning model, SWITCH 2.0 and illustrate challenges in climate-aware decision making. We simulate a simplified decision-making environment with multiple decision-makers with different priorities using a capacity expansion planning model. Using two different analysis approaches, regret-based and satisficing, we examine the dynamic nature of robustness, illuminate potential perplexities in implementing robustness to a climate-aware decision-making process, and demonstrate the applicability of the DMDU toolkit for practitioners. Finally, we offer recommendations for paving a pathway to better incorporate climate change in electricity grid planning.

[66] Media Narratives and Decision Making under Deep Uncertainty 

Authors: Robert Lempert; Andrew C. Revkin; Edmundo Molina-Perez, RAND Corporation; Columbia University; Tecnologico de Monterrey


Contemporary journalism, including data-based journalism, relies substantially on narratives. This includes the blend of story-telling writing style with detail, precise and well-research information on individual characters, human emotions, and real social problems. Narrative journalism is seen as a potent tool to increase society’s understanding of complex phenomena and as a form to spark public conversations on important societal challenges. However, if narrative journalism creates compelling stories for the public based on misunderstood facts and/or unidentified complexities, it can potentially shape perceptions based on narrative fallacies. This session will explore how the DMDU school of thought can help journalists tell stories that are compelling and memorable, but also accurate when dealing with complexity and deep uncertainty. Three topics will be discussed and debated in a panel comprised of professional journalists and DMDU practitioners (and researchers): 1) how journalist can blend DMDU methods and principles in narratives, 2) how DMDU work, and findings can be made more accessible to journalists, and 3) the research agenda required to address the first two challenges.

[22] Key challenges for defining and designing robust infrastructure investment partnerships in deeply uncertain many-actor water supply networks 

Authors: Andrew L. Hamilton; Rohini S. Gupta; Patrick M. Reed; Harrison B. Zeff; Gregory W. Characklis, Cornell University; University of North Carolina at Chapel Hill


The last decade has seen significant advancement in computational frameworks for multi-objective decision-making under deep uncertainty. Techniques such as Multi-Objective Robust Decision-Making (MORDM) allow stakeholders to navigate tradeoffs across conflicting goals, risk tolerances, and problem framings. This can aid decision-making in both single-stakeholder contexts (e.g., a water utility balancing reliability vs. cost vs. equity) and multi-stakeholder contexts (e.g., flooding vs. hydropower vs. habitat in reservoir operations). However, both contexts implicitly assume that the decision-maker (or set of decision-makers) is well-defined and fixed a priori. In reality, planning processes in complex, interconnected socio-environmental systems are often a looser, bottom-up phenomenon in which different organizations band together to form voluntary partnerships that are perceived to be in their mutual interest (e.g., investing in jointly operated infrastructure). Thus, we are interested not only in finding robust infrastructure projects, but also robust partnerships to support investment in them. From a planning perspective, this complicates our ability to compare the robustness of alternative investment partnerships, since the set of partners involved (i.e., “robust for whom”) varies across alternative partnership structures. This presentation will explore these issues in the context of water supply infrastructure investments in the Central Valley of California. MORDM is used to discover infrastructure partnerships that provide robust benefits to regional water suppliers under an array of deeply uncertain climatic, regulatory, and economic factors. Scenario discovery is then used to understand the most consequential uncertainties and how they vary across the water supply and financial performance metrics and alternative partnership structures. This research will highlight how cooperative partnerships and robustness shape the financial resilience and performance vulnerabilities for major infrastructure investments in many-actor systems.

[87] Considering Uncertainty in Human Systems Modeling 

Authors: Jim Yoon; Patricia Romero-Lankao; Y. C. Ethan Yang; Christian Klassert; Alfred Wan; Brent Daniel; Nathan Urban; Kendra Kaiser; Klaus Keller; Vivek Srikrishnan; Brinda Yarlagadda; Nathalie Voisin, Patrick M. Reed, Richard Moss, PNNL; NREL; Lehigh University; UFZ Helmholtz; Brookhaven National Laboratory; Boise State University, Dartmouth University; Cornell University; University of Maryland;


The role of individual and collective human agency is increasingly recognized as a prominent and arguably paramount determinant in shaping the behavior, trajectory, and vulnerability of multisector systems. This human influence operates at both short-term (hourly to daily) to long-term (annually to decennially) timescales, pushing systems towards either desirable or undesirable outcomes. However, the representation of human actors in computational models is prone to severe ontological, structural, and parametric uncertainties: Who are the modeled actors in MSD systems? What are their actions? How are these actors and actions operationalized and parameterized in a computational model? We present a new typology for classifying how human actors are represented in the broad suite of coupled human-natural system (CHNS) models that are applied in MultiSector Dynamics (MSD) research and organizing modeling efforts to tackle questions regarding uncertainty in human systems. The typology is demonstrated on CHANCE-C, an agent-based model of coastal urban development. Utilizing the typology and an initial prototype of the model, we conduct stylized experiments interrogating the relative influence of structural versus parametric uncertainty in the representation of residential location choice on urban development trajectories in flood-prone housing markets.

[107] Transforming Complex Social Systems: Insights from Systems Analysis and Negotiation 

Author: Tim McDonald, RAND Corporation


Effective response to public problems, and achieving societal goals, typically require efficient problem solving and adapting on the part of public leaders – especially under conditions of deep uncertainty. Sometimes however effective response requires reshaping social systems to transform them over time. The challenge for leaders in democratic societies is to meet to near-term needs and expectations while also addressing the complex, systemic origins of challenges and creating new opportunities – such as promoting health, personalized learning, or rapidly pursuing renewable energy – that are only available through system transformation. This paper examines what the literatures from two fields – systems analysis and negotiation – can contribute to understanding how to transform complex systems, and how they can improve public decision making under conditions of deep uncertainty.

[7] Assessing the Role of Travel Networks in Pandemic Risk from Emerging Infectious Diseases 

Authors: Pedro Adami Oliboni; Siméon Campos; Ross Tieman, Bocconi University and Forethought Foundation; École normale supérieure de Lyon


"COVID-19 has demonstrated the need to be prepared for future pandemics in order to minimize deaths, illness and economic impacts. Future pandemics could arise from several sources, of which, Emerging infectious diseases (EID) from wildlife are considered significant for future pandemics. Fortunately recent advances in infectious disease, information technology and biotechnology provide the necessary building blocks to mitigate some pandemic risks if deployed intelligently. The global nature of infectious diseases and increasing complexity of emergence and transmission dynamics due to travel networks poses a significant challenge to effective deployment of mitigation measures. Thus improving understanding of global EID risk landscapes represents an avenue to improve mitigation efforts. Currently EID risk assessments are limited to country level and do not describe connectedness to exogenous EID risk through air travel networks. A better understanding of the EID risk landscape that results when travel networks are considered is required to improve risk mitigation efforts. This paper combines existing EID risk assessment frameworks and global travel data in the creation of three risk models: a heuristic model, a Global Epidemic and Mobility Model (GLEAM) instantiated through, and a Susceptible Infectious Removed (SIR) model. The three models probe different aspects of global EID risk, allowing important characteristics to be identified. The results of these models indicate several policy recommendations concerning prioritization of pandemic risk mitigation such as focusing preventative mitigation efforts in high EID genesis countries, specifically across sub-Saharan Africa, and concentrating screening and mitigation efforts in key transport hubs."

[8] Adaptive tools for decisions on compounding climate change impacts on water infrastructure  

Authors: Andrew Allison 1; Scott Stephens 1; Judy Lawrence 2; Shailesh Singh 1; Paula Blackett 1; Yvonne Matthews 1; and Jan Kwakkel 3, 1 National Institute of Water and Atmospheric Research, Hamilton, New Zealand ; 2 Victoria University of Wellington, Wellington, New Zealand ; 3 TU Delft, Den Hague, Netherlands.


Rising sea levels, due to climate change, are putting increasing stress on coastal infrastructure like drinking water, wastewater and stormwater infrastructure services. This will force adaptive action, which requires us to robustly evaluate the most promising climate-change adaptation actions under conditions of uncertainty, such as which sequencing of infrastructure upgrades would be most robust to future uncertainty, and which sequencing would likely fail. This is especially difficult when multiple hazards interact with adaptation actions through time, compounding impact and risk. Our research addresses this capacity gap by developing assessment tools for transitioning and transforming infrastructure where compound hazards exist (flooding hazards from river, sea and groundwater which propagate as shocks and as slowly emerging and persistent impacts). We present two case studies of New Zealand wastewater treatment plants located on low-lying coastal plains and discharging treated effluent into a tidal receiving environment. We outline the specific challenges faced, the adaptation solutions proposed during workshops between researchers and practitioners, and we show how Robust Decision Making tools can be used within a Dynamic Adaptive Pathways Planning process to develop adaptation plans that are robust under a range of conditions.

[27] Equitable, Robust, Adaptive and Stable Deeply Uncertain Pathways: a new framework for exploring cooperative stability and multi-actor vulnerabilities in regional water supply management and infrastructure investment pathways  

Authors: David Gold; David Gorelick; Gregory Characklis; Patrick Reed, Department of Civil and Environmental Engineering, Cornell University; Center on Financial Risk in Environmental Systems, Gillings School of Global Public Health, UNC Institute for the Environment, University of North Carolina at Chapel Hil, Department of Environmental Sciences and Engineering, Gillings School of Global Public Health


Rising sea levels, due to climate change, are putting increasing stress on coastal infrastructure like drinking water, wastewater and stormwater infrastructure services. This will force adaptive action, which requires us to robustly evaluate the most promising climate-change adaptation actions under conditions of uncertainty, such as which sequencing of infrastructure upgrades would be most robust to future uncertainty, and which sequencing would likely fail. This is especially difficult when multiple hazards interact with adaptation actions through time, compounding impact and risk. Our research addresses this capacity gap by developing assessment tools for transitioning and transforming infrastructure where compound hazards exist (flooding hazards from river, sea and groundwater which propagate as shocks and as slowly emerging and persistent impacts). We present two case studies of New Zealand wastewater treatment plants located on low-lying coastal plains and discharging treated effluent into a tidal receiving environment. We outline the specific challenges faced, the adaptation solutions proposed during workshops between researchers and practitioners, and we show how Robust Decision Making tools can be used within a Dynamic Adaptive Pathways Planning process to develop adaptation plans that are robust under a range of conditions.

[95] Uncertainty and Governance in Complex Transboundary Water Systems: A Comparative Analysis of Adaptiveness and Robustness in the Great Lakes and Rio Grande/Bravo Regions  

Authors: Debora VanNijnatten; Yena Bassone-Quashie; Carolyn Johns, Wilfrid Laurier University; Toronto Metropolitan University; Toronto Metropolitan University


"Uncertainty – or, rather, many types of uncertainty – are deeply ingrained in water management and, in transboundary water basins, the management task is even more complex (VanNijnatten and Johns 2020). Climate change is further deepening the uncertainties faced by those who must manage shared water resources, with increasing demand for water, yet degraded water quality due to altered ecosystem conditions. This means that water governance and management systems must shift from the dominant “predict-then act” approach (Stanton and Roelich 2021) to one which recognizes the nonstationary of water resources and ecosystems (Cosgrove and Loucks 2015) and thus the need for more adaptive and responsive water governance mechanisms. This paper will provide a conceptual and methodological framework for analyzing uncertainties in water basins arising from climate change, and provide insights into the kinds of processes, practices and tools water governance systems will need in order to enhance their robustness and adaptiveness in the face of uncertainties, using a ‘adaptive governance barometer’ we have constructed. Both the uncertainty framework and adaptive governance barometer will then be applied to two complex transboundary water systems in North America, one which is a system of water abundance, the Canada-U.S. Great Lakes Basin, and one which is a system of scarcity, the U.S.-Mexico Rio Grande-Bravo Basin. This two-pronged analysis, as applied in these very different basins, aims to illuminate the many facets of deep uncertainty in complex transboundary water governance systems and also indicate whether these two governance systems possess the processes, practices and tools for responding to these uncertainties. More generally, this work is intended to provide analytical tools for analyzing uncertainty and system readiness that can be applied in other basins."

[41] Intra and Intergenerational Equity in Water Resources Planning: Prospects and challenges for robust equity in Mexico City  

Authors: Sarah St. George Freeman; Jacob Tracy; Casey Brown; Veronica Martínez; Homero Paltán, Patrick Ray; Diego Rodríguez, University of Massachusetts Amherst; University of Cincinnati; Independent Consultant; World Bank Group


"Urban systems play a critical role in way that we manage natural resources, including water. From the Romans and Persians to the modern day, civilizations have relied on large imports of water from rural areas to feed their urban growth. This paradigm of large-scale supply-side projects is an oft-unspoken prioritization of one section of the population over the other. In the epoch of the Anthropocene these management schemes also represent a prioritization of the current generation over future generations if they are not carefully planned to account for long term sustainability under uncertain future conditions. Such elements of intra and intergeneration equity are rarely accounted for explicitly in water resources planning. This presentation focuses on efforts to quantify tradeoffs in intra and intergenerational equity in water supply planning decisions under climate uncertainty in Mexico City and the surrounding region. Currently, about on third of the population of Mexico City—over 7 million people—do not have access to daily water supply and sanitation services. The majority of these shortfalls in services are concentrated in poorer neighborhoods of the city. While there is a need to increase water supply across the city, the largest portion of the city’s current water supply (over 40%) is sourced from a rapidly and irreversibly depleting aquifer. The next largest water supply—the Cutzamala System—provides about 30% of the city’s water from a complex surface water system that is increasingly under pressure from deeply uncertain shifts in agricultural demands and climatic factors. In this context, the question of how to improve the equity of water services for the current population as well as for future generations is wicked, uncertain, and not getting easier. Given the acute need to address present challenges, how could prioritizing current needs jeopardize future generations? Is ‘robust equity’ achievable and at what cost? In this presentation we explore the nature of equity tradeoffs in water allocation and planning decisions for the city and its uncertain future."

[79] Exploring the adaptation of military organisational structures  

Authors: Y.L. Lont ; T. Comes; J.H. Kwakkel, Delft University of Technology


"The success of military operations depends on the timely availability of military units, equipment, and essential supplies anywhere in the world. The timeliness creates interdependencies in resources, which requires the coordination of operational decisions (e.g., timing, quantity, or quality). The mechanism used to effectively and efficiently coordinate these decisions is context-specific. Here, a coordination mechanism is defined by all the elements that structure an organisation's individual behaviour; the responsibilities, roles, information structure, decision authorities, and incentive structure. However, military organisations struggle to realign their coordination mechanism with the changing context. They face deeply uncertain environmental changes, such as changes in funding due to shifting political priorities and coalitions, changing public expectations regarding military capabilities, and changes in war theatre, such as a shift from desert to tundra. The rapidly changing strategic landscape creates additional dynamics at the operational level. The realignment of the coordinating structure with the continuously changing context is further complicated because military organisations tend to be massive institutions organised in a vertical hierarchy with limited horizontal coordination. Any coordination of decisions in such organisations is very procedural and inherently involves various information asymmetries. What is the effect of information asymmetry on the emergence of a coordinating structure that can effectively and efficiently deliver the required military units, equipment, and essential supplies given the deeply uncertain contexts? As a first step towards answering this question, we theoretically investigate the emergence of a vertical hierarchy in the presence of information asymmetry and assess how deeply uncertain changing contexts affect the effectiveness and efficiency of coordinating decisions. Specifically, we extend Epstein’s model (Epstein, 2006) of growing adaptive organisations with information asymmetry. In this model, individual agents endogenously generate internal organisational structures (e.g., local hierarchies and trading regimes) and share information to make decisions. Next, we experiment with dynamically changing contexts and assess how this affects the organisation's performance and when and how this leads to organisational change."

[51] Identifying and understanding equity challenges in adaptive water supply investment and management pathways in The Federal District of Brazil  

Authors: Bruna M. Araujo; David F. Gold; Lillian Lau; Conceição de Maria A. Alves; Patrick M. Reed, Universidade de Brasília; Cornell University;


Urban water supply systems are increasingly subject to many uncertainties from climate, economic and social perspectives. Adaptive and flexible planning and decision processes are important tools for facing these issues and identifying more robust policies. However, in economically unequal societies, these policies may not properly reflect the different levels of vulnerability expressed by communities, resulting in actions and measures that aggravate the conditions of the most vulnerable communities. These issues are particularly acute for urban water supply systems in many large cities of developing countries. In this study, we focus on the Federal District of Brazil (FDB), a metropolitan area that is home for more than 3 million people. The FDB Water Supply System is comprised of two major service areas (Descoberto and Santa Maria) with highly unequal average income, water consumption patterns and adaptive capacity to manage drought crisis events. Despite being initially planned and developed in early 60s, the FDB region has been characterized by unregulated urban occupation, increasing population growth rates and socio and economic disparities that often result in unequal access to water and sanitation services. These challenges were highlighted during a recent drought driven water crisis from 2016 to 2018. Short-term drought management actions (water transfers and conservation) were implemented at high costs for consumers, government and the water supply utility. This work aims to improve regional water management by developing adaptive urban water supply investment and management portfolio pathways using the Deeply Uncertain Pathways (DU Pathways) framework. Our modeling formulation searches for alternative policies to increase the performance of the water utility given changing drought extremes, demand pressures and other deeply uncertain conditions in a multi-objective context. Results illustrate strong inequality between the service areas, showing that policies need to be also evaluated in terms of distributive costs among service areas when planning short-term actions and long-term infrastructure sequencing. We show a strong interdependence and potential for asymmetry in how short-term actions (e.g., water transfers or restrictions) benefit the two water service areas in drought crises and influence the relative intensity of their required long-term infrastructure investments. These findings provide insights on how water policies and inequalities in socio-economic conditions may agravate disparities between service areas in urban water supply systems. We further reveal tradeoffs between the objectives that guide candidate policies performance, and identify key uncertain conditions that can lead to system failures. Results of this study are broadly applicable to urban water supply planning problems globally in regions with high inequality.

[35] Exploring urban development under uncertain conditions. A model-based approach to investigate plausible futures in a flood-prone Municipality in Rome.   

Authors: Simona Mannucci; Jan H. Kwakkel, Sapienza University of Rome; TU Delft


This study presents an interdisciplinary approach for including uncertainties in the planning process, using a case study of a flood-prone area, the X Municipality of Rome. The aim is to explore plausible futures considering the uncertainties that could influence the urban development in the area. For this reason, we used a land-use change model implemented in Metronamica, a generic integrated spatial decision support system to simulate land-use dynamics. The exploration is up to 2050 and accounts for future uncertainties for land-use demands and the influences of new infrastructures discussed by institutions to be implemented in the area. Five thousand computational “what if” experiments are performed using exploratory modelling; the resulting land-use maps for 2050 are clustered according to the similarity of the land-use patterns. Next, the clusters are analysed to assess which clusters are relevant. Lastly, Scenario Discovery is performed for each decision-relevant cluster to understand the combination of uncertain input parameters under which the land-use pattern emerges.

[19] Understanding decision making in post-disaster recovery: The role of values, rules, and knowledge in recovery of winegrowers after the 2021 floods in the Ahr valley   

Authors: Jonathan Hassel; Dr. Saskia E. Werners, United Nations University Institute for Environment and Human Security (UNU-EHS)


"Against the backdrop of arising uncertainty due to climate change, it is important to not only make progress in decision making concerning climate change mitigation and adaptation, but to also engage with the topic of post-disaster recovery. The guiding principle of recovery as defined in the Sendai Framework for Disaster Risk Reduction 2015–2030 is “Building Back Better”, and involves systematically incorporating disaster risk reduction measures into the recovery process. Under this framing, research to date has focused primarily on how to implement Building Back Better as a top-down process. Offering an alternative approach to understanding recovery, we investigate how values, rules, and knowledge shape the recovery pathways of winegrowers decide upon after the 2021 floods in the Ahr valley. We derive from conducted interviews, that most winegrowers confronted with uncertainty decide to continue winegrowing at their former location. The choice of this recovery pathway is driven by held values like “achievement” and “self-direction” and by the perception that winegrowing is generally a profitable business opportunity. Along the wine production chain, the most fundamental change is observable in the marketing and sales section. As technical rules inherent in existing sales infrastructure like restaurants and hotels are broken, while the digital infrastructure improved significantly in response to COVID-19, a long-lasting shift towards online markets and wholesale is discussed among winegrowers. Knowledge of the threat of a flood reoccurrence or other climate change-related impacts does not fundamentally influence first recovery decisions. In the analysis, we not only discuss how affected actors envision their recovery but follow up on the interaction of values, rules, and knowledge, revealing dilemmas that prevent the implementation of Building Back Better in recovery and stories of hope, which enable it. On this basis, we argue that decision making in recovery is embedded in ongoing political struggles, which began before the disaster. Furthermore, establishing a plan to adapt to climate change before a disaster strikes can serve as a guiding vision for recovery. Measuring the recovery success in the time passed until the disaster site is physically reconstructed falls short of considering aspects of recovery like building a shared memory of the disaster as a basis for risk awareness. Lastly, the cooccurrence of COVID-19 and the flood showed that winegrowers were better equipped to respond to the flood due to recovery and adaptation measures previously set up in response to COVID-19. Building on this observation, we suggest moving forward with the idea of “systemic recovery”, propagating the success of interventions from one hazard to another."

[57] A graduate-level course in adaptive planning under deep uncertainty 

Author: Garry Sotnik, Stanford University


Many fields that universities aim to prepare graduate students for require decision-making under deep uncertainty (DMDU). A primary example is the field of sustainability, where decision-makers face social dilemmas in complex and adaptive systems that are characterized by non-linear dynamics and unpredictable futures. Yet, the theories and methods taught as part of graduate-level curricula often assume that dynamics are linear and uncertainty is probabilistic. This undermines a student's ability to make decisions, such as those aimed at resolving social dilemmas in pursuit of sustainability. We first review existing courses that teach adaptive planning under deep uncertainty. We then describe a new course that builds on game theory, complexity science, and DMDU; and describe lessons learned from its pilot with students in Stanford University's Sustainability Science and Practice Master's Program. We conclude by sharing plans for future course development.

[30] Engaging decisionmakers on the costs and benefits of reaching net-zero emissions in Latin America 

Authors: Adrien Vogt-Schilb; Nidhi Kalra; Edmundo Molina; James Symes; Fernando Esteves; Marco Butazzoni; Claudio Alatorre, Interamerican Development Bank; RAND; Agence Francaise de Developement


Countries are setting targets to meet net-zero emissions by around 2050, calling for profound changes spanning many sectors, including energy, transport, food production, housing, industry, and waste management. Decisionmakers from these sectors are thus crucial to the success of net-zero targets. While sympathetic with climate change goals, however, they often face other priorities, limited resources, and a limited understanding of what they can do – especially in developing countries. We use Robust Decision Making to address this challenge at the Interamerican Development Bank (IDB), the largest provider of official development assistance in Latin America. We organized workshops with IDB sector specialists and asked about development priorities, feared uncertainties, trusted data sources, and ideas on options to reduce emissions. We built and calibrated a modular multi-sector model from this information and simulated pathways to reach net-zero emissions, using a set of transformations that are each expressed in the language of the relevant sector. We show how decarbonization can help advance development benefits in each sector and sum up to a net economic benefit. Finally, we assess what transformations are most important and what uncertainties could derail plans. Our results can help align IDB sector strategies with climate change goals.

[31] The Risk of Delay: Using Robust Decision Making to Improve Implementation of Louisiana's Climate Action Plan 

Authors: Patrick Bodilly Kane; Allison DeJong; Jordan Fischbach, The Water Institute of the Gulf


"In 2022, Louisiana’s Climate Initiatives Task Force approved the state’s first-ever Climate Action Plan: an ambitious set of strategies and actions to put the state on a path to meet Governor John Bel Edwards’ goal of net zero greenhouse gas (GHG) emissions by 2050. While Louisiana is not the first state to develop a climate action plan, Louisiana’s GHG emissions profile is starkly different from other states; over two-thirds of Louisiana’s emissions are from its industrial sector, including energy and chemical production. The success of the Plan depends on a number of deeply uncertain factors outside of the control of state policymakers, including the timing and speed of shifting to renewable electricity production and powering Louisiana’s industrial sector with low- or no-carbon fuels and renewable electricity. Researchers at the Water Institute of the Gulf previously applied a Louisiana-specific version of the Energy Policy Simulator, an open-source simulation tool developed by Energy Innovation Policy & Technology, to inform development of the Climate Action Plan. Building on this work, we conducted a Robust Decision Making analysis to examine the effects of varying economic and policy scenarios on the success of the plan. Specifically, we looked at the effects of a range of statewide economic projections, fuel pricing, policy implementation rates, and technical advancements in areas like clean hydrogen on key measures of plan success including carbon emissions, GDP growth, health benefits from reduced emissions, and job creation. Results suggest that reaching net zero emissions by 2050 represents a highly optimistic scenario, and that overall implementation success is especially sensitive to the timely implementation of the Plan’s key pillars of renewable electricity generation, low- and no-carbon hydrogen production, and industrial electrification. "

[61] Assessing economy-wide decarbonization options in Guatemala under uncertainty

Authors: Jairo Quirós-Tortós; Luis Victor-Gallardo; Jam Angulo-Paniagua; Lourdes Socarrás; Benjamin Leiva; Daniella Suger; Cristian Guzmán; Diego Incer; Edgar Miranda; Andrea Rivas; Rodrigo Leonardo Castellanos; Francisco Molina; Marcela Jaramillo; Andreas Fazekas; María José Leiva; Omar Samayoa, University of Costa Rica; Universidad Galileo; Universidad del Valle de Guatemala; 2050 Pathways Platform; Inter-American Development Bank


Guatemala is a Central American country of over 16 million inhabitants that relies on firewood for more than half of its energy needs. It is also experiencing a rapid rise in private vehicle ownership and relies on biofuels, oil, coal, and hydropower for its electricity generation mix. The risks of drought, inefficient transport systems, and deforestation challenge maintaining society’s needs sustainably and affordably. These risks are parallel to climate change and worldwide economic shifts, which will affect Guatemalans. Energy, agricultural, climate, and economic policymakers should prepare for these conditions while meeting the country’s international commitments to reduce greenhouse gas emissions. This work will present a technological assessment of options in the energy, transport, agricultural, and land-use sectors of the economy that maximize future socio-economic benefits and decarbonize the country. The assessment will be based on bottom-up systems models simulated under a multiplicity of futures to consider deep uncertainties. The exercise will also be an example of capacity building in local Universities, which will be prepared to support government knowledge requests in the future.

[111] Impact of Ethical Premises on Pareto-Optimal Climate Abatement Pathways

Author: Max Reddel


Humanity faces the unprecedented global challenge of climate change. Integrated Assessment Models (IAMs) are central to informing decision-making to avoid catastrophic consequences. However, policy recommendations resulting from IAMs commonly prompt a very heterogeneous distribution of risks and benefits across the globe. During the recent 2021 United Nations Climate Change Conference (COP26), it became clear that equity is a central issue in the climate action debate. Emerging economies consider currently suggested abatement policies unjust in light of the historical CO2 generation of high-income countries and the strongly increasing need for energy in low-income countries. Equity is therefore an eminently pressing topic, yet most IAM studies largely neglect it due to the implicit use of a utilitarian social welfare function that aggregates risks and benefits over space and time, thus losing sight of distributional consequences. Using alternative social welfare functions (rooted in e.g., egalitarianism, prioritarianism, or sufficientarianism) are likely to yield very different climate abatement pathways within an optimization setup. In this study, the RICE model is transformed into a simulation model and embedded in a many-objective simulation-optimization setup in order to find Pareto-optimal climate mitigation pathways. We compare these pathways across four different ethical problem formulations and show how global emissions, economic damages, and the increase in atmospheric temperature are substantially lower in non-utilitarian problem formulations. These differences in resulting pathways can then be used to support the decision-makers with more transparent policy recommendations that are based on explicit normative assumptions which can be chosen by the decision-makers.

[26] Options to Achieve Carbon Neutrality in Chile: An Assessment Under Uncertainty

Authors: Benavides, Carlos; Cifuentes, Luis A.; Díaz, Manuel; Gilabert, Horacio; Gonzales, Luis; González, Diego; Groves, David G.; Jaramillo, Marcela; Marinkovic, Catalina; Menares, Luna;Meza, Francisco; Molina, Edmundo; Montedónico, Marcia; Palma, Rodrigo; Pica, Andrés; Salas, Cristian; Torres, Rigoberto; Vicuña, Sebastián; Valdés, José Miguel; Vogt-Schilb, Adrien, Inter-American Development Bank (IDB); Energy Center of the University of Chile; Centro Cambio Global UC (CCG-UC); Latin American Center for Economic and Social Politics UC (CLAPES-UC) ; Tecnológico de Monterrey; Rand Corporation.


"Chile aims to reach carbon neutrality. Its Nationally Determined Contribution (NDC) commits the country to reach net-zero emissions of greenhouse gases by 2050 and sets targets for emissions to be reduced progressively over time. To comply with the goals of the NDC, line ministries have considered a set of sectoral transformations, such as closing coal-fired power plants, promoting electric mobility, and increasing forest captures which, taken together, could bring emissions down to zero. This study evaluates how these sectoral transformations would fare under a wide range of economic, environmental, and technological uncertainties. It identifies the vulnerabilities of the strategy, that is, under what conditions sectoral transformations are insufficient to achieve net-zero emissions. It then quantifies options for making sectoral plans to deliver the NDC more robust, that is to reduce the likelihood of not achieving carbon neutrality. The study follows the Robust Decision Making (RDM) method with the objective of analyzing multiple possible futures. In the case of long-term climate strategies, the method allows evaluating which conditions the emission reduction strategies have more benefits than costs, and allows identifying which sectoral transformations are more important to reach the goal of zero net emissions. Additional measures discussed include speeding up retirement of coal-fired power plants, promotion of telework and non-motorized transport, reduction of beef consumption, expansion of thermal retrofitting of houses, increased afforestation, sustainable forest management, and expansion of protected areas. These measures are based on ideas proposed by sectoral experts during a participatory process. Finally, a macroeconomic evaluation finds that enhancing the set of measures put forward to comply with the NDC would result in a net gain of 0.8% of gross domestic product (GDP) by 2050, on the top of 4.4% GDP gain that the current NDC plans would bring."

[18] People and policy integration in planning for human-natural systems

Authors: Enayat A. Moallemi; Fateme Zare; Brett A. Bryan, Dekin University; The University of New South Wales; Deakin University


Planning under uncertainty is important for informing environmental decisions under poorly understood future state of human-natural systems and limited data availability in long timescales. However, the multiplicity of societal actors with various knowledge and stakes in the planning process poses challenges, such as diverse priorities, complex relationships, and different management approaches. Within this context, legitimate plans are those that are 'co-developed' through integrating credible people and policy expertise with formalised methodologies to deepen the understanding of different values, approaches, and relationships. Despite the increasing importance of using people and policy expertise in the planning process, there is a limited conceptual clarity and systematic understanding about it. To address this gap, we undertake a critical appraisal of 43 planning studies related to various human-natural systems, such as sustainable management of water resources, conservation policy, disaster management, transportation planning, and energy policy, in the last two decades to serve two aims. First, we characterise different aspects of using people and policy expertise in the planning process along 9 dimensions to clarify why societal actors are involved (i.e., to frame, to analyse, to act), how they participate (i.e., stakeholders, timing, interaction), and how they create impact on the ground (i.e., power, politics, change). Choices made (un)knowingly along these 9 dimensions will have a strong influence on the complexity of the planning process and result in particular challenges. As a second aim, we identify these challenges and suggest strategies to address them. This will provide researchers and practitioners with a guide to better navigates trade-offs in engaging with people and policy-makers in planning for human-natural systems. It also helps stimulate and promote knowledge co-production aiming at an inclusive representation of societal actors and the interaction of knowledge system with action and societal change.

[52] Transgressive DMDU to enhance sustainability transformations 

Authors: Luis A. Bojórquez-Tapia; Hallie Eakin; Patrick M. Reed; Yosune Miquelajauregui; Ileana Grave; Tatiana Merino-Benítez; Edmundo Molina-Pérez, Laboratorio Nacional de Ciencias de la Sostenibilidad, Instituto de Ecología, UNAM, México; School of Sustainability, Arizona State University, Tempe, AZ, USA; School of Civil and Environmental Engineering, Cornell University, Ithaca, NY, USA; Tecnológico de Monterrey, Escuela de Ciencias Sociales y Gobierno, Monterrey, México


Diverse forms of uncertainty play instrumental roles in decision-making for sustainability transformations. Sustainability transformations involves fundamental changes in human and environmental interactions, feedbacks and power structures to enable trajectories for sustainable development. DMDU originated to deal with inherent disagreements and uncertainties in modeling future states in support of decision-making. In sustainability transformations, however, DMDU has been called on to address a broader scope of uncertainty and normative concerns. Here we examine three DMDU implementations to explore the contribution and limitations of DMDU to sustainability transformations in the context of large-scale infrastructure investments. Our analysis points to a DMDU agenda for future work that would, through an embrace of more transgressive approaches, enable sustainability transformations. We argue that transgression entails making tractable and transparent the social and political processes that undergird policymaking, requiring that scientists must be aware of the power dynamics inherent in participatory processes. Transgressive DMDU is as much a political act as a research challenge, and one that may not be welcome in all spheres of decision-making. We conclude that transgressive DMDU can be effectively used as a means of revealing the values that are fundamental to sustainability transformations, while empowering actors to grapple with uncertain future trajectories.

[71] The connection between uncertainty and project delay - using data to understand the dynamics of combining many samples from long tailed distributions. 

Authors: Gerard Cardoso i Negrie; Alan Mosca; Yael Grushka-Cockayne, nPlan; UVA Darden School of Business


"An investigation of historical construction projects highlights that most activities in a project are delivered on time, yet we all know that many projects are delayed. In this presentation, we will explore how certain properties of structure and risk in projects can begin to explain this phenomenon, all backed up with real examples from nPlan’s dataset of 500k schedules. The availability of schedule data in construction has largely been a blocker in doing large scale analysis of project behaviour. Over the last 5 years, nPlan has collected the largest known data set of construction schedules from heavy infrastructure, rail, oil & gas, and more, which has allowed us to investigate and ask questions about what leads to project delay. The purpose of this presentation is to present some of our key findings, which will inspire project managers and risk practitioners alike to alter the way they may think about project planning in future. The presentation will first introduce the landscape of construction projects, as seen through the actual project data we’ve collected. This will help frame the complexity of understanding project delay. Following this, the presentation will introduce some fundamental concepts of risk and forecasting that will shine the light on how delay is almost inevitable when there is a long tail of risk on activities in a project. Furthermore, the way activities are sequenced in a project can directly impact the amount of risk which will be explored through some real life examples in our data."

[77] Climate and Land Management Vulnerability and Robustness in the Northern Drakensberg Strategic Water Source Area 

Authors: Nalin Singh; Homero Paltán; Kevin Wheeler; Diego Juan Rodriguez, Oxford University School of Geography and the Environment; World Bank 


The Northern Drakensberg strategic water source area (SWSA) is a vital water resource for Lesotho and South Africa. This study uses a Cathedral Peak research catchment as a case study to determine how land management impacts the climate change robustness of the SWSA, using a bottom-up climate change assessment. Through a combination of the decision scaling and robust decision making frameworks, climate-performance thresholds and key trade-offs between the natural and degraded/disturbed conditions were identified. The degraded conditions are more robust for annual hydrological yield. However, the degraded conditions are also far more vulnerable to extreme floods and excessive dam siltation, which undermines hydrological yield. The natural conditions better attenuate climate variability and have a better response across a wider range of climate change projections/scenarios. Therefore, they are more robust. Recommendations are to supplement the hydrological yield with demand-side solutions and integrate bi-lateral land management into the Lesotho Highlands Water Project.

[63] Direct policy search for a risk-based levee design framework 

Authors: Jingya Wang; David Johnson, Purdue University


"The Surge and Waves Model for Protection Systems (SWaMPS) model is a processed-based model of surge-based flood risk which simulates the physical processes of flooding (e.g., surge and wave overtopping, climate change, rainfall, etc.,) within a single-polder ring levee system. It is an efficient model for coastal flood risk management that can be used to inform risk-based levee design heights, achieving greater risk reduction at lower cost than a standards-based design (e.g., a “100-year” standard protecting against an event with 1-in-100 chance of occurring in a given year). The model is fast enough to feasibly evaluate thousands of designs over a large ensemble of future conditions, allowing us to use it in an optimization framework with the design heights of individual levee reaches as decision variables. This framework provides guidance for making optimal decisions of upgrading levees at the beginning of the planning horizon. However, there are other plausible ways to manage the system, such as changing the frequency we want to make decisions of lifting the levee system (e.g., every 25 years). We can raise the levee height based on the actual amount of the sea level rise in that 25-year period. Direct Policy Search (DPS) has been proved to be an efficient state-dependent method to incorporate new information and make adaptive decisions over time. We propose to explore the value and use of DPS method to upgrade the levee periodically to counteract the uncertainties such as sea level rise, and land subsidence. This study embraces the cost-effective risk analysis which can be achieved by simultaneously optimizing the two objectives: the construction cost of upgrading the system; and the expected flood economic losses. The uncertainties in climate change (e.g., sea-level rise and storm surge) are dealt with as an ensemble of states of worlds (SOW) to include different plausible future states. We design a series of state variables which describe the state of the system and exogenous forces acting on it. Then they are mapped to signposts that trigger actions in mapping functions. We finally identify the Pareto frontier of non-dominated policies by using the Borg Multi-Objective Evolutionary Algorithm. We evaluate the performance of the DPS algorithm to a static intertemporal optimization by comparing the Pareto frontiers, reliability, and the speed of convergence. We also use time series analysis to examine how the dike height increase over time in different SOWs, investigating how adaptive policies are when generated by DPS in a realistic problem context. The robustness analysis is processed by using the Multi-objective Robust Decision-Making (MORDM) framework to evaluate how robust the policies are to deal with uncertainties in the flood risk management. We quantify the robustness over alternative single SOWs by employing metrics (e.g., maximin, maximax, etc.,). The performance of the levee system management under different assumed SOWs can be compared, leading us to identify which parameter or factor in the SOWs would affect the performance the most. We also explore the sensitivity of the performance to different discount rates. "

[47] Applying Non-Market Values to DMDU Frameworks: Example of Hydroelectric Dams 

Author: Nihar Chhatiawala, Pardee RAND Graduate School (RAND Corporation)


"Non-market valuation methods are used to assess costs and benefits of decisions for which outcomes carry values that are not aptly measured through markets. Such methods are principally used to measure the value of ecosystem services that would be affected by proposed uses of natural resources. To practitioners, the validity of non-market measurements is rooted in economic theory as it concerns utility maximization. Recent academic critiques of non-market valuation argues that its focus on individual utility fails to address the complexities of human and natural systems and as a result may undervalue ecosystem services. Furthermore, the calculation of non-market values is frequently contingent upon assumptions of the future with questionable validity given the deep uncertainties of climate change. This research aims to explore whether (1) decision support informed by DMDU practices can bolster decisionmakers’ ability to navigate multidimensional complexity in conditions that call for non-market valuations and (2) DMDU methods can enhance confidence in the validity of non-market valuations that result under conditions of deep uncertainty. Following a critical literature review (documenting how policymakers apply non-market valuation to balance the utility of hydroelectric dams against the disutility of their environmental impacts), I will assess methodological strengths and weaknesses in the context of varying policy applications and measured outcomes. Then, I will construct a hypothetical case study to demonstrate integration points between non-market values and DMDU methods which, depending on my prior findings, may include the following: •Leveraging calculated utility functions to compare robust and optimal strategies •Proposed contingent valuation methodology to evaluate deeply uncertain outcomes •Inclusion of non-market valuation models in multi-objective analyses The research is intended to contribute to a broader, policy-relevant research agenda on navigating environmental risks of the energy transition."

[39] Keep The Benefits Flowing: Finding Robust Release Policies for The Akosombo Dam in a Changing World 

Author: Ted Buskop; Jazmin Zatarain Salazar; Afua Owusu; Jill Slinger, TU Delft and UNESCO IHE


The Akosombo dam in Ghana's Lower Volta River Basin provides essential economic benefits through hydropower generation, flood protection, and irrigation opportunities. However, after its construction finished in 1965, the livelihood of the riverine communities drastically changed. The seasonal peaks and low flows, which provided environmental services for many downstream households, are now replaced with a stable river flow throughout the year to favour hydropower production. Reoperating the dam is crucial to sustain healthy ecosystems and meet the basin's current water demands. Further, climate change and increased water demands for irrigation are expected to strain the water resources in the Lower Volta. Reoperation is also complicated by the absence of water treaties with neighbouring countries. This leaves the possibility of receiving lower flows from upstream riparian states due to their increasing water needs. In this study, we aim to understand the vulnerabilities of the Akosombo system while finding robust policies that can cope with challenging climate and demographic conditions. Towards this goal, we first take a set of Pareto optimal policies found through Evolutionary Multi-Objective Direct Policy Search. Then we evaluate the performance of these candidate release policies under a broader set of climate, social, and political uncertainties in which energy storage and water treaties are also considered. Further, a Patient-Rule-Induction Method is applied to find the boundaries between system success and failure, and feature scoring is used to discover which variables contribute most to system performance. Lastly, we use multiple robustness metrics to analyse the effects of decision maker's risk aversion. This study highlights the challenges in balancing irrigation, environmental, and hydropower goals while also identifying system vulnerabilities and opportunities for robust operations to meet the Lower Volta River water demands under future challenging conditions.

[13] Factors influencing the adoption of adaptive policy approaches by water utilities 

Author: Emily Ryan, TU Delft


Over recent decades there have been advances in the research behind adaptive policy approaches. More recently, an approach known as Dynamic Adaptive Policy Pathways (DAPP) has emerged, which has drawn the attention of planners in the water industry, particularly in water utilities. Despite interest from both the water industry and the research community to see these novel approaches applied, there have been limited applications, and no published guidance to support the operationalization of these approaches in water utilities specifically. This project seeks to bridge this gap by answering the question: “What are the most important factors influencing the adoption of adaptive policy approaches by water utilities?” This project uses expert interviews and research by design to understand current barriers and opportunities in the adoption of such methods in water utilities, and to design a taxonomy to support operationalization of adaptive policy in the industry. I will conduct this research through content analysis of interviews of relevant stakeholder in the utility sector and research community in Australia and the Netherlands. The expected outcome of the research is a taxonomy of barriers and opportunities to the uptake of DAPP in water utilities, along with other artefacts that can support further operationalization research.

[56] Oil as a conflict fuel 

Authors: Willem L. Auping; Jan H. Kwakkel, TU Delft


"There has been an extensive debate on the effects of conflict resources on how severe conflicts play out. Resources often mentioned in these debates are the so-called 3T’s (Tin, Tungsten, and Tantalum), and Gold, but also Diamonds (e.g., the Kimberly process), especially in the African context. The 2022 war in Ukraine made undeniably clear, however, that the influence of sales of energy resources, especially oil and gas, on financing a conflict can be enormous. The problem is, however, that in the case of oil – and many other alleged conflict resources – simply stopping to buy the conflict fuel will only have a limited effect, as rerouting will remain possible. As the oil price shows hog cycle-like behaviour with alternating periods of relatively high and low prices – and 2022 being in one the upswings – it would be interesting to see to what extend and at what costs it would be possible to either turn back of speed up time to extend the previous period of low prices, of accelerate the occurrence of the next. For this purpose, we used two energy-price simulation models to simulate how energy prices would react to a cut in the demand caused by a combination of a direct hit on the economy (i.e., self-induced economic downturn) or rapid decoupling (i.e., reducing the energy intensity of the economy). The models were both modelled in System Dynamics, both simulate on a global scale and for the coming decades, but have completely different levels of aggregation. We found that a crucial factor in these dynamics is the influence of either accumulation of resources or the production capacity have on energy prices. In the first case (accumulation/demand induced price reduction), the demand cut accelerates the occurrence of a new period with low prices. In the second case (production capacity/demand induced price reduction) the opposite seems to occur, as the prices behave similar as they did before that time, where the significance of the effect is of course dependant on the amount of the demand reduction. In all cases, however, the effects on energy prices were temporary, where in some cases the energy prices after the demand cut seemed to become even higher then without the demand cut. The only policies that have a lasting effect on the energy prices are continuous pressure on the demand of (fossil) energy resources due to significant climate mitigation policies. While climate mitigation policies are hardly a silver bullet in this case, as their effect plays more on the longer term, it does show that short term demand reduction now in combination with an ambitious set of climate mitigation policies can reduce the effect of oil as a conflict fuel. "

[9] Housing Price Prediction in the Face of Deep Uncertainty

Author: Swaptik Chowdhury, RAND Corporation


"The broad aim of this analysis is to expand the scope, awareness, and professional practice of DMDU methods. Much of the future decision-making will increasingly occur in the context of deeply uncertain phenomena. We use the prosaic question of housing price forecasts to create a broadly understood example to disseminate the efficacy of DMDU methods as a tool for promoting robust and responsive decision-making. In doing so, this study proposes novel approaches to several aspects of DMDU. Volatility in forecasted housing prices, especially during the pandemic, has shown the limits of the current prediction paradigm. Housing prices depend on many physical (location, district, number of floors and rooms, and age), behavioral (presence of a home owner’s Association, preference for a certain style, and fear of missing out), and economic features (GDP, inflation rate, interest rate, and housing supply). However, as the pandemic has shown, many factors can be ideally characterized as deeply uncertain. When such deeply uncertain features are used in traditional optimization-based prediction methods, it usually leads to erroneous decision making and adoption of short-term outlook which further constrains effective policymaking. The implosion of Zillow Offer’s iBuy program which relied on accurate prediction for rapid flipping of houses is emblematic of the vulnerabilities and limitations of current methods used for predicting housing prices and decision-making. The goal of this study is to leverage the Uncertainties (X), Levers (L), Relationship ( R), and Metrics (M) framework from DMDU and use multi-class scenario analysis (which is an extension of binary-class scenario discovery) to identify an ensemble of prices and isolate key features which may affect those prices. A literature review on existing housing price prediction models will help in identifying the X’s and a convolutional neural network (CNN)-based machine learning model will be employed. Typically, for scenario discovery, clustering of output space, and identification of predictive subspaces are executed sequentially. But this study will establish an ML method called multivariate regression tree (MRT) as a novel use case to simultaneously cluster the output prices from the CNN model and identify highly predictive feature subspaces within those clusters. Analysts, policymakers, and other stakeholders can then use the discovered scenarios to analyze tradeoffs and evaluate different pricing strategies. "

[92] Parallelization of WEAP-MODFLOW models using open-source development environments 

Authors: Edmundo Molina-Perez; Hermilo Cortés-González, Tecnologico de Monterrey


This talk presents a computational approach that can be used to parallelize WEAP-MODFLOW models using Vagrant, an open-source development environment. The approach allows users to parallelize WEAP-MODFLOW models on local machines, dedicated serves, or cloud environments. The approach is exemplified in the study adaptive and robust implementation for the Water Resources Management Plans in the Ligua and Petorca basins, Valparaíso Region, Chile, which aims to propose implementation strategies for the national water plan of Chile with a robust and adaptive approach, through the use of RDM.

[6] COVID-era Decisionmaking in American Institutions of Higher Education

Author: Malia K. Du Mont, Bard College


Institutions of higher education typically operate in a highly predictable way: the academic calendar is set years in advance, and - except for some technological differences at the margins - life at residential American colleges and universities is much the same for today's students and professors as it was 100 years ago. These historic institutions, which continue to produce the vast majority of BA degree holders in the US and remain a major draw for students from around the world, were generally not designed or prepared to adapt rapidly in the face of crisis. Taking Bard College and its network partners as an example, this talk will explore how residential institutions of higher education adapted quickly to the requirements of the pandemic and managed the complex needs of a diverse international population when little was known about the virus and government guidance did not address many of the circumstances being faced at American colleges and universities. Bard was one of the first institutions in the country to return to fully in-person residential learning, in August 2020.

[82] Dealing with the unknown in planning processes: some methodological explorations

Authors: Maria Rosaria Stufano Melone; Domenico Camarda, Politecnico di Bari


"This contribution is about reflections on the awareness of how the lack of knowledge and the unknown are crucial elements to consider during any territorial and environmental planning process. The unknown can affect the effectiveness of choices and start a chain of unpredictable consequences. Plans often deal with policy and planning decisions that interest collectivity, human and non-human beings, the space we live, our territories and our time (or portions of time). Such plans (either for households or a city or a region) have to cope with unexpected events, uncertainties, and unwanted consequences. After exploring some theoretical aspects of knowledge and non-knowledge, we propose applied ontologies as a conceptual approach to deal with the lack of knowledge and the unknown in planning. "

[93] Advancing the design of national decarbonization strategies for an uncertain future 

Authors: Edmundo Molina-Perez; David G. Groves; Adrien Vogt-Schilb, Tecnologico de Monterrey; World Bank; Inter-American Development Bank 


Many parties to the Paris Agreement are designing long-term strategies to coordinate government action towards reaching carbon-neutrality by mid-century. Their effective design and implementation require navigating the economic, social, and technological uncertainty that surrounds any plan to 2050, and garnering support from many stakeholders that do not naturally prioritize climate action. Here, we describe an approach to develop and evaluate whole-economy pathways that achieve decarbonization goals under many possible futures while bringing socioeconomic benefits. The approach gathers stakeholders’ input from greenhouse-gas-emitting sectors to list transformations that they understand and that can help meet their own development goals while reaching net-zero emissions. It then uses simulation models and data to evaluate multi-sector strategies under thousands of plausible futures and different metrics for success, identify key vulnerabilities, and design more robust strategies. We integrate and systematically compare findings of the application of this approach in Chile, Costa Rica, and Peru, and drawing lessons that are applicable globally.

[84] Multi-method design considerations in Sustainable Energy Transitions under Deep Uncertainty 

Authors: Matías Paredes-Vergara;Rodrigo Palma-Behnke;Jannik Haas, University of Chile (Energy Center);University of Chile (Energy Center, Department of Electrical Engineering);University of Canterbury (Department of Civil and Natural Resources Engineering)


Sustainable Energy Transitions (SET) develop under conditions of deep uncertainty in several dimensions such as techno-economic, political, and sociotechnical. To support model-based decision making under deep uncertainty in complex systems, such as SET, a body of literature known as Decision Making under Deep Uncertainty (DMDU) proposes several methodologies, each with its strengths and limitations for SET application. Although a multi-method perspective has been recognised in the DMDU literature as a means for leveraging the strengths of DMDU methodologies to cope with the deep uncertainty condition, the literature lacks a discussion of this perspective for the specific characteristics of SET. This work contributes to the DMDU literature by providing considerations to perform adequate mixing designs among DMDU methodologies to cope with model-based SET under deep uncertainty. The main findings include SET conditions under which a sequential, parallel or integrated mixing design is better suited to identify possible alternatives instead of using a single-method approach. Under these circumstances, candidate mixing designs among DMDU methods are explored and practical recommendations are also provided to implement these multi-method designs within possible SET contexts. An illustrative case is provided based on the evolution of the Chilean carbon neutrality strategy in the context of the Paris Agreement.

[59] Advancing energy decarbonization with fiscal instruments under uncertainty: the case of Costa Rica

Authors: Luis Victor-Gallardo; Mónica Rodríguez-Zúñiga; Jairo Quirós-Tortós; Marcela Jaramillo; Adrien Vogt-Schilb, University of Costa Rica; 2050 Pathways Platform; Inter-American Development Bank


Decarbonizing energy can increase firms' productivity and reduce household expenses through transport and energy efficiency gains. However, governments face losing fuel and vehicle tax revenue. Moreover, they can provide tax exemptions or subsidies to incentivize technology adoption. Here we analyze the trade-off of maintaining tax revenues while enabling decarbonization benefits. We first estimate the financial impact of energy decarbonization on the Government (or fiscal impact), firms, and households by income level. Then, we evaluate energy, property, import, and distance-based tax options to mitigate revenue loss or incentivize decarbonization while accounting for distributional impacts on firms and households. We study the case of Costa Rica in the 2023-50 term, a country that pledged to reach net-zero emissions by 2050 in its National Decarbonization Plan, and 20% of its tax revenue is from energy and transport. We find the Plan would cause a fiscal impact of -0.41% of GDP (yearly 2023-50 average), which occurs mainly in the long term and is lower than the financial benefits on households and firms: +1.49% of GDP (yearly 2023-50 average). We show that combining tax options to eliminate the fiscal impact and distribute the benefits among households and firms would not overaffect one single actor or tax instrument. We also find the combinations of uncertainties and climate policies that would trigger two risks: i) the need for strict fiscal adjustments to redistribute financial impacts and ii) insufficient decarbonization benefits to eliminate the fiscal impact.

[11] The Role of Exploratory Modelling Approaches and their integration into UK Net-Zero Policy Design - A Greenhouse Gas Removal technologies and the 6th Carbon Budget Case Study

Authors: Quirina Rodriguez Mendez; Mark Workman; Geoff Darch, Sustainable Energy Futures Masters, Energy Futures Lab, Imperial College London, UK; Foresight Transitions Ltd, Grimstead Road, Farley, Near Salisbury, Wiltshire, SP5 1AT and Energy Futures Lab, Imperial College London, South Kensington, London SW7 2AZ, UK; Anglian Water, Thorpe Wood House, Thorpe Wood, Peterborough PE3 6WT 


"The limited remaining global carbon budget to maintain the 1.5°C target, coupled with delayed mitigation action, has elevated Greenhouse Gas Removal (GGR) technologies to the forefront of climate policy design. Removing carbon from the atmosphere at the GtCO2 scales envisaged could, however, be hindered by the plethora of techno-economic, sustainability, and social, political and ethical uncertainties inherent to GGR value chains. This is compounded by the lack of large-scale GGR value chains in operation today. The ability for orthodox consolidative decision support approaches to integrate such insufficiently characterised, disruptive technologies and inability to accommodate for uncertainties - such as the recent sixfold increase in wholesale gas prices - has called into question their central role in Net-Zero policy design. The need to deal with deep uncertainty, system complexity and emergence has resulted in a growing body of literature supporting alternative approaches [1, 2]. Exploratory modelling emerges as a potential approach capable of bridging the gap between the demanded urgency of GGR deployment and inherent deep uncertainty [3]. By exploring a wide range of conceivable futures, exploratory approaches specifically Robust Decision Making (RDM), explicitly embrace deep uncertainties while seeking to reduce vulnerability to future plausible states. The RDM process assesses the performance of different policy levers attempting to achieve a set of defined goals under varying uncertain conditions, with performance measured using metrics identified through stakeholder engagement. The consequences of risks to GGR policies to overall objectives such as national Net-Zero targets are tested without privileging any set of assumptions, thereby exposing potential fragilities in current strategies. This participatory and iterative approach allows for the identification of co-benefits and drawbacks associated to GGR deployment in a more active trade-off process, whereby stakeholders can clearly discern the performance of different portfolios using multiple criteria. However, though RDM approaches have been well documented there is little insight as to how such exploratory approaches might be integrated into Net-Zero policy design processes. This contribution seeks to assess the role of GGR technologies in the UK’s 6th Carbon Budget and how these technologies might be integrated into Net-Zero policy design as a case study. It will be undertaken through stakeholder consultation and engagement with the modelling and policy community. It will seek to inform policymakers and GGR specialists how RDM approaches might compliment present decision support tools and net-zero policy design; as well as to develop an understanding as to how exploratory approaches might be adopted by the climate policy community at a national and international level. References: [1] Pye, S., Broad, O., Bataille, C., Brockway, P., Daly, H., Freeman, R., Gambhir, A., Geden, O., Rogan, F., Sanghvi, S. et al. (2021), ‘Modelling net-zero emissions energy systems requires a change in approach’, Climate Policy 21(2), 222–231. [2] Workman, M., Dooley, K., Lomax, G., Maltby, J. & Darch, G. (2020), ‘Decision making in contexts of deep uncertainty - an alternative approach for long-term climate policy’, Environ- mental Science & Policy 103, 77–84. [3] Workman, M., Darch, G., Dooley, K., Lomax, G., Maltby, J. & Pollitt, H. (2021), ‘Climate policy decision making in contexts of deep uncertainty - from optimisation to robustness’, Environmental Science & Policy 120, 127–137."

[24] System Models in Decision Making for Low Carbon Transitions Under Deep Uncertainty

Authors: Sheridan Few; Muriel Bonjean Stanton; Katy Roelich, University of Leeds


"The global low carbon transition requires the transformation of a wide range of social and physical systems. Bringing about these transformations necessitates pro-active decision making, with a key role for the public sector in driving change. However, the complexity of systems involved leads to a high degree of uncertainty in their response to interventions, whilst the critical role these systems play in the functioning of society means that stakes are high. This can serve as a barrier to ambitious policymaking. Decision making under deep uncertainty (DMDU) provides a useful framework to catalyse pro-active decision making in this context. DMDU techniques are designed to support decision making where actors don’t know or can’t agree upon (i) appropriate models of relationships in considered systems, (ii) probability distributions of key variables, and/or (iii) desirable outcomes. The balance of emphasis upon each of the three categories in practical applications of DMDU has been little examined, and the appropriate balance will vary between contexts. Notably, DMDU methods have primarily been applied to making systems more resilient to external forces, and have only recently begun to be applied in the context of transforming systems in order to reduce their external impacts. Many complex systems may be conceived of and modelled in diverse manners, with substantial implications for model outputs. These differences may become more important in decision making for system transformation. Further, deeper leverage points for system transformation relate to changing perspectives and system relationships, whilst parameter changes typically access only shallow ones. As such, consideration of model diversity and transition between system states is particularly important in decision making for sustainability transformations. This work presents a structured literature review of approaches to system models in 40 practical applications of DMDU methods to support infrastructural decision making. It examines the extent to which previous DMDU studies (i) explicitly include alternative sets of system relationships, (ii) include uncertainties which encompass different conceptions of the system, and (iii) include actions which can access deeper leverage points for system transformation. Risks associated with insufficient treatment of model uncertainty and leverage points in decision making under uncertainty are outlined. Insights based upon this review are used to develop best practice recommendations for the use of models in decision making for decarbonisation under deep uncertainty."

[67] An interactive framework for multiobjective robust decision making under deep uncertainty

Authors: Babooshka Shavazipour; Jan Kwakkel; Kaisa Miettinen, University of Jyvaskyla, Faculty of Information Technology; Faculty of Technology, Policy and Management, Delft University of Technology


The Robust Decision Making (RDM) framework is successfully applied in various real-world applications. It also extended for the simultaneous consideration of multiple objectives and scenarios (known as MORDM). However, MORDM is a posteriori method; i.e., the analyst first generates alternative solutions and performs all the robustness analyses, then displays the results to the decision-makers to choose the most preferred one among many solutions. Indeed, the decision-makers need to track all the trade-offs between objectives to find a balanced solution that is also robust in various scenarios, introducing a substantial cognitive load. In problems with many objectives, tracking all the trade-offs and uncertainty effects is too laborious and makes decision-making tricky. This issue has been counted as one of the reasons why MORDM is not widely applied beyond academia. To overcome this issue, in this study, we propose a novel interactive framework involving the decision-makers in the search for the most preferred robust solutions utilizing interactive multiobjective optimization methods. In this way, the decision-makers can learn about the problem limitations, the feasibility of their preferences, how uncertainty may affect the outcomes of a decision, and interactively study the trade-offs between objectives in various scenarios. This involvement and learning give them additional insight into the problem and allows them to directly control and lead the search during the solution generation process and decision-making, boosting their confidence and assurance in implementing the identified robust solutions in practice.

[48] Rhodium-SWMM: An Open-Source Tool for Green Infrastructure Placement Under Deep Uncertainty 

Authors: Nastaran Tebyanian, Jordan Fischbach, Robert Lempert, Hong Wu, Lisa Iulo, Klaus Keller, The Pennsylvania State University, The Water Institute of the Gulf, RAND Corporation, Dartmouth College


"Green Infrastructure (GI) measures are increasingly used to adapt to climate change in urban areas, but it remains a challenge to evaluate their effectiveness and successfully allocate investments in such systems. Climate change and rapid urbanization increase both the need and the challenge of planning for GI. Green infrastructure can provide multiple benefits such as improving water quality, reducing peak flows, and providing cooling. Climate change often leads to increased frequency, intensity, and duration of heavy rainfall, as well as rapid urbanization, which can increase the flood frequency and volume by modifying the hydrological systems. Thus, designing and planning GI are subject to deep uncertainties and require navigating trade-offs between multiple objectives. Traditional Green Infrastructure modeling approaches are typically silent on the effects of deep uncertainties. Robust Decision Making (RDM) and Many-objective Robust Decision Making (MORDM) can address these modeling challenges. While a limited number of studies have applied RDM or MORDM to green infrastructure planning, there are two main gaps: 1) lack of incorporation of site-scale GI levers and 2) the absence of a generalizable flexible open-source tool for applying MORDM to green infrastructure planning. The Rhodium-SWMM software presented in this study addresses these gaps. Rhodium-SWMM is an open-source python library for green infrastructure planning under deep uncertainty. The Rhodium-SWMM library connects EPA Stormwater Management Model (SWMM) to Rhodium, a Python library for MORDM. Rhodium-SWMM provides a generalizable, flexible open-source interface for taking any swmm input file and setting up a multi-objective optimization problem with the ability to define a wide range of parameters in the SWMM input file as uncertainties or levers. The software tool can help to investigate a wide variety of research questions in multi-scale green infrastructure placement under deep uncertainty. "

[100] SISEPUEDE: A rapidly deployable and scalable modeling framework for multisector dynamic exploratory emissions

Authors: James Syme; Edmundo Molina-Perez; Nidhi Kalra; Fernando Esteves, Pardee RAND Graduate School; Tec de Monterrey; RAND Corporation; Pardee RAND Graduate School


We introduce SYSEPUEDE, a rapidly deployable and scalable modeling framework for multisector dynamic exploratory emissions and cost/benefit modeling. SYSEPUEDE is comprised of two key parts: subsector-level models and a data management pipeline. First, SYSEPUEDE contains four emissions sector models--AFOLU (including agriculture and livestock, forestry, and land use), Circular Economy (solid waste and wastewater), IPPU (industrial processes and product use), and Energy (including electricity, industrial energy, transportation, and stationary emissions and carbon capture and sequestration)--and a Socioeconomic model. Each of the emissions models can be run as independent sectors or as part of an integrated emissions model. Second, SYSEPUEDE includes a data pipeline management system to facilitate and support the exploration of subsector-level strategies across a large number of scenarios, capturing exogenous uncertainties and the uncertainties in transformations. SYSEPUEDE is implemented in Python and Julia (NemoMod), though the models can be used from the command line and other programs like R. This talk will detail the models, the modeling framework, and demonstrate how it can easily be modified to explore the costs and benefits of decarbonization in different regions and countries at a subsector level.

[60] Using well-being metrics to characterize flood impacts in Argentina and identify robust responses 

Authors: Sara Turner; Brian Walsh; Julie Rozenberg; Lourdes Rodriguez Chamussy; Evelyn Vezza, The World Bank Group


"While asset losses are commonly relied on metrics used to assess disaster impact, they fail to adequately account for the indirect effects of disasters on income and consumption, and they inadequately account for disproportionate impact of disasters on the poor, who, while they have fewer assets to lose, experience greater income and consumption effects. Existing tools also limit options to compare the benefits and costs of options that do not influence asset losses, which constrains policy making and prevents an integrated approach to policy planning and coordination. New tools take steps to remedy this gap and allow asset loss estimates to be extended to consider losses to well-being, a metric which better represents the disproportionate impact of disasters on the poor. Researchers at the World Bank developed a method to translate asset loss estimates into utility losses, enabling the use of well-being losses as a measure of disaster impact. These tools enable cost benefit comparison across a wider range of potential policy options, including those that are not typically considered by flood risk planners such as social protection payments, which do not reduce asset losses, but can compensate for the well-being impacts of disasters and enable households to recover faster following flood events. This study extends the approach by using a previously developed provincial level model of flood risk, characterizing asset and well-being losses and resilience at a provincial level, and then analyzing the costs and benefits of interventions to reduce both asset and well-being losses in a unified, deep-uncertainty based, framework. Results find that both infrastructure investment and social support can reduce losses due to flooding, and have synergies, being most effective under different conditions, and targeting different subpopulations. "

[29] Evaluating Future Stormwater Flood Risk in New Orleans 

Authors: Jordan R. Fischbach; Patrick Kane; Daniel Gilles; Brett McMann; Nathan Young, The Water Institute of the Gulf


"The City of New Orleans, like many cities in the U.S. and globally, faces significant challenges in managing flooding from heavy rainfall under present and future conditions. Present-day challenges include an inadequately sized drainage and pumping system and aging infrastructure with substantial deferred maintenance. Future climate change is likely to exacerbate these challenges in the decades to come. Several major efforts are planned or underway in New Orleans to improve water management. For example, the city is currently in the process of designing and constructing over $200 million in green infrastructure projects throughout the city. To date, however, there has been no systematic investigation of rainfall flood risk in New Orleans: specifically, how uncertainty related to future climate change, especially increases in rainfall intensity and sea level rise, or operational uncertainties or failures might lead to changes in the frequency or extent of rainfall flood risk for city residents. To address this need, we use a detailed model of the New Orleans drainage system in a Robust Decision Making analysis focused first on vulnerability in the present-day system. This analysis examines the potential for flooding across a range of storm types and frequencies and under hundreds of scenarios representing uncertainty in future rainfall, sea level rise, drainage obstruction, and potential pump failures. We use scenario discovery to identify conditions that lead to significant increases in the frequency and duration of flooding and spatial analysis methods to highlight areas of the city that are or could become more vulnerable to flooding. We use analysis results to develop new visualizations and other compelling communication products, designed to help decision makers and residents understand these risks 1) as a step towards a new consensus vision on how to best “live with water” and 2) to improve flood mitigation project planning and design amidst increasingly growing stressors."

[17] Mitigating the Deeply Uncertain Risks to the Bulk Electric System from Geomagnetically Induced Currents: A Robust Decision Making Approach

Author: Kurt Klein, RAND Corporation; Pardee RAND Graduate School


"The uncertainties looming over the potential effects that geomagnetically induced currents (GIC) could have on United States’ energy systems are alarming. Predictions range from the apocalyptic—a yearlong blackout of the nation’s power grid resulting in 90% of Americans dying from hunger and societal collapse —to the manageable, and that blackouts are not even certain to occur. This uncertainty among scholars and pundits means that the extent of this threat is highly politicized and not conducive to sound policy deliberation. Deep uncertainty exists about the incidence of the GIC threat as well. GIC could be propagated by a geomagnetic disturbance arising from fluctuations on the sun such as coronal mass ejections. But they could also arise from a nuclear explosion in the high atmosphere to create a magnetohydrodynamic electromagnetic pulse (MHDEMP). Likelihoods are difficult to calculate. The alarming prospect of an adversary attacking the United States with a MHDEMP is dictated by geopolitics, the defense planning of the US and its adversaries, and many other variables. Similarly, we are not able to reliably estimate the frequency of large geomagnetic storms of sufficient strength to arrive to earth and produce GIC capable of damaging the power grid. Only a handful of such storms have occurred during the stage of human history when expansive continental sized networks of power grids existed for GIC to affect. One attempt to estimate the 10-year recurrence probability for a severe geomagnetic disturbance state that it is “somewhere between vanishingly unlikely and surprisingly likely.” From an extensive literature review, this research describes what are the United States’ power grid’s vulnerabilities to GIC. It also elaborates an application of DMDU concepts and methods to outline possible solutions for mitigating catastrophic risk to the bulk electric system from GIC. With these vulnerabilities and solutions in mind, this research proposes a robust decision making (RDM) approach to discover and evaluate adaptive solutions to navigate both the deep uncertainty and polarized positions around this policy problem. The author will present both progress towards this effort and the setbacks."

[55] Understanding recovery, lessons from recovery pathways after extreme flood events

Authors: Saskia E. Werners; Jonathan Hassel; René Kemp, United Nations University - Institute for Environment and Human Security (UNU-EHS); United Nations University - Maastricht Economic and Social Research Institute on Innovation and Technology (UNU-MERIT) 


"Extreme events are projected to intensify. Impacts on lives, livelihoods and wellbeing extend through time. Recovery from disasters is an inherently prolonged and uneven process. Yet, across the world, long-term recovery is typically a lower-priority aspect of disaster management policy and practice. Improvements in early warning systems, emergency preparedness plans and response protocols are seldom matched by coherent, multi-sectoral recovery plans and protocols. The aim of this paper is to better understand recovery. As communities recover, the will may be to recover the initial state, without learning from the disaster. We ask how we can learn from the recovery efforts, in particular for transformative recovery or what is called ‘Building-Back-Better’ in the United Nations’ Sendai Framework for Disaster Risk Reduction. Can the post-extreme event recovery period be utilized for building back better? How can the recovery process incorporate learning and decision making that leads up to more climate-resilient futures? What does the recovery period tell us about the navigation of ideas, interests and institutional work under uncertainty in the wake of an extreme event? We address these questions by combining conceptual and empirical insights through literature review and an assessment of recovery pathways from recent extreme flood events. In particular, we analyse the summer 2021 floods in the Rhine-Meuse region in Belgium, Germany and the Netherlands, which shocked communities and decision makers with its unexpected devastation. We juxtapose this evidence with experiences from other flood events around the world, including the 2022 floods in Durban South Africa, the 2021 floods in Lagos Nigeria, the 2019 Punjab region floods in Pakistan and the recurring floods in the Hoar region Bangladesh. Although we start from cases of extreme flood events, we explicitly recognize the systemic nature of risk and the uncertainties surrounding cascading impacts and complex recovery decisions. As such we also invite for exchange with other cases at the DMDU annual meeting. Conceptually we build on decision making under deep uncertainty approaches, including adaptation pathways, systemic risk management, transformative change, and a broad set of recovery frameworks. Working with partners across the main flood-affected countries in Europe, as well as other flood-prone areas of the world, this paper and presentation aims to share knowledge, conceptualize recovery, drive action, and ultimately shift the focus from passive coping to pro-active adaptation, innovation and transformative recovery. "

[85] What is the right level of DMDU guidance vs. prescription, and how does that change depending on agency, project, location scale, and affected resources? How should an agency support the application of DMDU methods where they are appropriate? How do we determine guidance approaches? What have other agencies done? How could we coordinate these guidance efforts?

Authors: Dagmar Llewellyn; Rebecca Smith; Deena Larsen, Bureau of Reclamation


The Bureau of Reclamation, a Federal agency that manages water in the arid Western United States, supports using Decision-making Under Deep Uncertainty (DMDU) to frame and analyze complex water issues. We are developing a guidebook to explain the basic DMDU concepts and how they relate specifically to our modeling, decision-making, and implementation contexts. Many U.S. Federal agencies are incorporating resilience planning and DMDU into their operations (see However, DMDU methods are complicated, there is often a lack of data and computational resources, and applying DMDU analyses can result in ambiguous outcomes with little guidance on how expert judgement and stakeholder sentiment should factor in.

[96] DMDU and the City: Impact and Implementation in Culver City, CA

Author: Thomas Aujero Small, CEO Culver City Forward, former Mayor of Culver City 


"Culver City, California, a small city of 40,000 at heart of the Los Angeles metropolis has experienced tremendous economic growth over the last decade. Over the past five years, Apple, Amazon, HBO and Tiktok chose to locate their media headquarters in the city, which is also home to Sony Pictures and has a long history as a film studio town. Thomas Small was elected to City Council in 2016, and then as Mayor in 2018. Previously a consultant in urban design and planning, Small had a mandate to help solve the problems engendered by this rapid growth, including environmental issues, mobility and traffic congestion, a severe housing shortage, intense commercial development, and rapid demographic/community change. Working with Robert Lempert and Steven Popper of RAND, Mayor Small initiated studies and projects involving DMDU to address these challenges shortly after his election. Culver City hosted the DMDU Society Annual Meeting in 2018. Focused largely on the crucial issues of mobility, housing, infrastructure development, and community outreach, these engagements began with a project designed to accelerate the implementation of a new plan for mobility in Culver City’s Transit Oriented District. This project was published as “Implementing a New Mobility Vision for Rancho Higuera in a Deeply Uncertain, Fast-Changing World: Results from RAND Corporation Participatory Local Planning Workshops by Robert J. Lempert, Tim McDonald, Steven W. Popper, Diogo Prosdocimi, Thomas A. Small. Both the process and the results of this project have had a significant influence on development in the city and on the ongoing Culver City General Plan update, also initiated by Small as Mayor in 2018. The Covid pandemic had a tremendous effect on these processes, slowing the arrival of Apple and Amazon personnel, but also accelerating the possibility of changes in mobility and exacerbating the housing shortage and the homeless crisis. Leaving public office in 2020, Small started the non-profit Culver City Forward to continue the work of engaging these issues in a public/private partnership. The next official DMDU engagement with RAND was called Culver City Bounce Forward, a research and policy project intended to determine what policies would help the city and community not just recover from the crisis, but actually bounce forward, taking advantage of the plasticity of the catastrophic situation to make changes that would be more difficult or impossible in normal times. Culver City Bounce Forward is in the process of being published as a stand-alone website that we hope will be become an important resource for the community and the public-at-large that is facing many of the same issues around the world. This presentation would analyze and trace the history of these DMDU engagements with Culver City, with particular focus on the real-world implementation of initiatives and policies through Adaptive Planning and Adaptive Pathways and our encounters with challenges including the Covid crisis, political opposition, and reactions from local constituents and the broader community. What have we learned and how do we envision the future of Culver City now? "

[12] Finding a diverse set of representative criminal supply chain models to identify robust interventions

Authors: Isabelle M. van Schilt; Jan H. Kwakkel; Alexander Verbraeck, Delft University of Technology


"Governments try to intervene robustly in criminal supply chains. However, governments must come up with possible interventions in the face of deep uncertainty, and use models to analyse the data and study the effect of potential interventions. Criminals try to mask their actions so there is limited available information and data to inform the modelling process. Theories on supply chains might offer an alternative source of information which could supplement the limited available data. However, even by supplementing data with theories, the analysts still face the problem that there is a large set of theoretically distinct models which are all coherent with the available data and information. So, how to generate a representative and diverse set of possible models which can in turn be used for (many objective) robust decision making in pursuit of robust interventions in criminal supply chains? To answer this question, we explore the potential of quality diversity algorithms. Quality diversity algorithms are a type of evolutionary algorithm that aims to find a diverse group of optimal solutions. Quality refers to the high quality (i.e., optimality) of these solutions in the outcome space. Diversity refers to the diverse values of the input parameters for the optimal solutions in the input space. Quality diversity algorithms can be used for our problem by minimizing the difference between the behaviour of the simulation model and the available data, while ensuring diversity across the various models. In this study, we use a stylized discrete event simulation model of a counterfeit Personal Protective Equipment supply chain as case study. We evaluate the quality-of-fit and the diversity of the quality diversity algorithm for generating a representative and diverse set of possible criminal supply chain models."

[43] Scale transformation and uncertainty propagation in loosely-coupled multi-models

Authors: Jin Rui Yap; Yilin Huang; Igor Nikolic; Jan Kwakkel, TU Delft


"In modeling for supporting decision-making on complex socio-technical challenges under deep uncertainty, one finds the need to couple system models of differing disciplines, formalisms, and scales. This is sometimes also referred to as multi-modeling and entails configuring multiple sub-models such that they exchange information with one another. Configuring existing models into multi-models to address new problems saves time and effort. It further enables the bridging of expertise across multiple domains, accounting for multiple perspectives and scales, and allows for the encapsulation of diverse, complex systems. However, multi-modeling introduces complex challenges with regard to scaling approaches and uncertainty analysis. Firstly, models from different domains are often developed at scales relevant to the original problem domain. In coupling these models, one often finds that they are misaligned in scale. Thus, scale bridging approaches are required to facilitate the meaningful transfer of outputs across multi-scale sub-models. The specific aspect of scale discussed is resolution (i.e., level of spatial, temporal, or object detail), whereby downscaling and upscaling refer to the disaggregation and aggregation of information. Across various disciplines of study, scaling functions are used to transform data across different levels of detail within multi-scale simulations. While relatively simple mathematical functions may suffice as scaling functions, such an approach is limited as it can result in significant loss of information. Alternatively, system-specific methods that account for domain characteristics are more realistic and may yield improved accuracy and consistency but are more costly to implement. Concurrently, for multi-models to meaningfully support decision-making under deep uncertainty, there must be a sufficient understanding of the effect of uncertain sub-model inputs and inherent system variability on multi-model outcomes. Uncertainty analysis aims to understand this effect, which is important for understanding system dynamics and informing meaningful interpretation of model outputs. Analyzing the propagation of uncertainties in a multi-model is a complex endeavor as the interactions of uncertain parameters within and across sub-models may be non-linear and interdependent. The selection of scaling approaches is thus non-trivial as it can influence the dynamics of uncertainty propagation through a multi-model. This study uses an interdisciplinary multi-model composed of multi-scale sub-models to address questions in the energy systems domain. The objective is to evaluate the effects of different scaling approaches on the dynamics of uncertainty propagation in a multi-model. This research focuses on loosely-coupled multi-models, explicated by automated two-way data transfer between independently developed models. Specifically, we implement multiple system-agnostic and system-specific scaling functions on the multi-model and compare their impacts on multi-model uncertainties. The results inform the selection of scaling approaches to better reduce and manage uncertainties in using multi-scale multi-models to support decision-making under deep uncertainty."

[64] Incorporating learning into direct policy search to identify more adaptive strategies for flood risk management

Authors: Jingya Wang, David Johnson, Purdue University


"Direct policy search (DPS) is an approach for identifying optimal policies for managing a system by expressing decisions as a function of variables characterizing the system’s current state and exogenous forces acting on it. While the method originated in the literature on reinforcement learning for robotics, it can be applied to a much broader class of applications. As an example, we can use DPS to develop adaptive rules for upgrading dike systems for flood risk management in response to the rate of sea level rise and observed storm surge events. Policies are optimized based on average performance over a large ensemble of future states of the world (SOW). This presents an opportunity to improve the method by incorporating learning over time: we add the predicted future values of state variables to the policy definition. This is done in a way that effectively reweights the SOWs according to our beliefs about their relative likelihoods. This new algorithm incorporates learning into the standard DPS method. The result is a hybrid approach where today’s deep uncertainties, treated using ensemble analysis, are characterized probabilistically as the system evolves and we update our beliefs about what type of SOW is being experienced over time. System performance is improved via adaptive meta-policies that define how management of the system should switch between policies identified by a standard DPS approach. The benefits of the new method are demonstrated in a case study of the problem of storm surge under sea-level rise which involves planning optimal dike heightening for flood risk management. We simultaneously minimize two objectives: the discounted costs of (i) dike upgrades and (ii) expected residual economic damage. Incorporating learning about the likelihood of different SOWs makes management of dike heightening more state-dependent and adaptive. Consequently, the identified Pareto frontier of efficient policies shows improvement in the two objectives over a large ensemble of plausible SOWs. We evaluate the degree of improvement achieved by different ways of implementing the algorithm, such as the frequency of updating beliefs or the ways in which different SOWs are characterized for learning. In other words, we explore questions like whether policies should be reevaluated every 10 years or every 25 years, or whether predicted sea-level rise should be classified using three levels (e.g., low, medium, high) or with a greater degree of fidelity. The direct policy search method has been shown to be an efficient way to identify high-performing strategies for automatically adapting a system in response to changing conditions. This new method makes management strategies even more adaptive by incorporating new information gained over time; this better reflects the real world where we learn more each year about factors like sea level rise. In another part of the dissertation, I plan to apply the new method to a realistic process-based model of coastal flood risk in order to evaluate its potential to improve policy design for flood risk management. "

[70] Application of Steps 4 and 5 of the CRIDA approach in the Limarí  Basin, Chile.

Authors: Yena Bassone-Quashie; Sadie McEvoy, Toronto Metropolitan University (formerly Ryerson University); Deltares


"In applications of pathways approaches, such as Dynamic Adaptive Policy Pathways, the analysis so far often ends with the development of the pathways. However, implementation of adaptive planning, like pathways, requires different approaches to economic assessments and institutionalization. This session will discuss the application of the Climate Risk Informed Decision Analysis (CRIDA) approach in the Limarí Basin in Chile which faces significant issues related to water scarcity. CRIDA is a five-step approach for decision-making that provides water managers and policy makers with a framework for assessing the impact of climate change and climate uncertainty on their water resources and infrastructure, and supports them in developing effective adaptation strategies (UNESCO & ICIWaRM, 2018). Despite its increasing use, most applications of CRIDA undertake the first three steps of the approach, which consist of establishing the decision context, completing a quantitative climate vulnerability analysis and identifying alternative strategies and adaptation pathways. The Limarí project is one of the few projects that have undertaken CRIDA Steps 4 and 5, comparing the pathways and institutionalizing the plan, respectively. This work builds on analyses completed for the first three steps by Verbist et al., 2020 and Luger, 2020 and demonstrates the implementation of Steps 4 and 5. The results consist of an economic evaluation (primarily an Incremental Cost Analysis) of different adaptation options in the basin, and an institutional and socio-cultural analysis (ISA) (based on the work of van der Brugge & Roosjen., 2015) to assess the extent to which institutional and socio-cultural conditions support or hamper adaptation measures. The session will discuss how the final two steps of CRIDA were applied in the Limarí Basin. It will discuss the key findings from the analysis and the lessons learned from the application of these steps. It will also highlight opportunities for further development of the approach which were identified as a result of undertaking the analysis."

[23] WHEN TO PLAN? Understanding temporality for urban planning for an uncertain future

Authors: Supriya Krishnan; Nazli Aydin; Tina Comes, Resilience Lab, TU Delft


"While the need to adopt an uncertainty-oriented approach to urban planning is acknowledged, there is a substantial knowledge gap in systematically integrating it into plans and planning processes. Current planning finds it challenging to account for the flexibility and tradeoffs required for long-lasting investments. Urban Planning problems, especially under climate change, are complex. This complexity, combined with time pressure to react to urgent development needs, forestalls full exploration of the uncertain space using traditional DMDU methods. Managing uncertainty requires (a) Methods to generate exhaustive scenarios and (b) apriori assumptions based on each urban context - so the uncertainty sampling is more focused. A pivotal roadblock to using this combination effectively is the lack of understanding of temporal aspects of planning. An urban system comprises sub-systems with different temporal rhythms of change ranging from 3-5 years for open spaces to 100+ years for infrastructure. These rhythms must be synced within the actual planning timelines of cities. Climate change adds another layer of temporality that disrupts these rhythms in the form of stresses and shocks that have varying return periods and probabilities. Urban systems with heterogeneous adaptive capacities must respond to disruptions, not lock-in risks. However, time is both a relative and subjective concept in planning. The planning timeline is influenced by financial, political, and socio-economic processes, which change for every geographical context. Hence, there is no widely accepted theory of shifting lifecycles for long-term planning. As a result, cities prefer to work with the 'best guess' future. In this talk, we will present critical scholarship that contributes to the understanding of temporality in urban planning: (A) We examine existing theories on uncertainty and planning methods that deconstruct lifecycles and temporality in the urban environment. We dive into how these theories discuss windows of opportunity and long-term tradeoffs in planning; (B) We analyze two urban case studies in the Global North and Global South using the lens of temporality and their respective growth dynamics. We applied 40 semi-structured interviews with senior planning practitioners regarding their experience and perception of long-term thinking, uncertainty approaches, and challenges; (C) We synthesize the two steps to present critical recommendations for a theory for urban planning under uncertainty. "

[40] Not a silver bullet: Searching for lockdown alternatives in an artificial province

Authors: Mikhail Sirenko; Alexander Verbraeck; Tina Comes; Srijith Balakrishnan; Hidde Menno Bijlard, Delft University of Technology; Singapore-ETH Centre


"Lockdowns have shown their effectiveness in reducing the number of COVID-19 infections around the globe. They vary in type, and there is no standard practice in when to introduce a lockdown, how long it will last, how strict it is, and most importantly, at what scale. For example, in March 2022, the Netherlands had ""stay at home"" advice for individuals in case of symptoms, while Beijing was experiencing probably the most strict version of a lockdown: citizens are not allowed to leave their homes under any circumstances. While a strict lockdown seems to be a robust policy for decreasing infections, previous research has shown numerous adverse effects. What remains unclear is whether a set of more targeted measures and under which circumstances can replace a strict and straightforward lockdown. To find what can replace a strict lockdown, we propose a high-resolution agent-based model (ABM) of COVID-19 spread in South Holland, the Netherlands. With around 3,500,000 individual agents of different profiles (e.g. students, pensioners) performing activities (e.g. studying, shopping) at more than 1,500,000 locations (e.g. schools, supermarkets), the model captures the main aspects of urban life. We use a modified SEIR-like model that accounts for asymptomatic transmission and vaccination. We simulate the ABM under a wide range of epidemiological parameters and run scenario discovery methods to determine which policies perform similarly or better than the lockdown."

[110] Resilience and Uncertainty  

Author: Jose Ramirez Marquez


The term “system resilience’ by itself incites debate about its inherent concept and its computation. To clarify this concept, recent studies have proposed metrics to understand and to quantify the ability of systems to recover from an external disruptive event. Unfortunately, most of this research is focused on deterministic metrics. This presentation will introduce recent advances in the analysis of network resilience- via description of corresponding stochastic time dependent formulae for addressing the problem of quantifying network resilience. The presentation will unveil and discuss two optimization problems of stochastic order inherent to the resilience process when considering recovery activities. The presentation will highlight different concerns regarding uncertainty for implementing a systems resilience framework in practice.

[62] A novel decision support framework for resilient energy system planning when faced with extreme weather events

Authors: Nathan Lee; Nicholas Laws, Colorado School of Mines;National Renewable Energy Laboratory


The frequency and severity of extreme weather events have been rising due to climate change. As a result, weather-related outages and their associated economic consequences have also been increasing in frequency and magnitude. This trend is expected to continue. To mitigate these losses, opportunities to improve energy system resilience must be explored, however, the uncertainty of extreme weather adds complexity to this research. Current energy system modeling tools lack sufficient capabilities to fully consider uncertain severe weather events for investment planning. In this work, a novel decision support framework for energy system planning is introduced, which leverages robust decision-making under deep uncertainty (RDMDU) to improve energy system resilience against future extreme weather scenarios. The framework is applied to an example energy system to demonstrate resilience improvements via the RDMDU process.

Methodological Innovations

Behavioral approaches



Complexity Sciences

Water & Natural Resources I

Adaptive Planning: Methods & Applications


Climate Change

New Perspectives on DMDU

Water & Natural Resources II

Methodological Explorations


Methodological Innovations

Extreme Events & Risk Management

Problem solving session

Discussion Forum

Experts' Panel

Novel Applications

Resilience Science & Methods

Climate Change and IAMs

[97] Fear and Curiosity: How Implicit Narratives Limit and Enlarge the Future 

Authors: J. Henry Graumlich; Ian Prichard, Calleguas Municipal Water District; Camrosa Water District 


"In the context of climate change, it is a commonplace among water resource management practitioners that historical hydrology is no longer a reliable guide to future decision making. As a result, DMDU approaches to scenario planning and adaptive pathways are increasingly employed to assist policy deliberations. The extent to which implicit mental models and their narrative expression have been shaped by engineering and policy responses to that historical hydrology is less recognized and rarely discussed. What ensues are classic wicked problem discussions where uncertainty over cause and effect devolves into values discussions, resulting, predictably, in acrimony and impasse. We cling to the conceptual foundations of our narratives precisely because they are invisible and, as such, carry the emotional weight of reality; our detractors are not simply different, but irrational. When, on the other hand, we make visible the synthesis of our own narratives, we automatically become curious about the prospect of an alternative future. The authors are interested in exploring with others how explicit consideration of implicit narratives may provide effective pathways to futures that would otherwise be constrained by conflict or fear. The authors draw upon over 45 years of combined practice in southern California public sector water resource management across municipal and special district organizations, retail water and wastewater development, and regional wholesale strategic planning and policy. As practitioners and colleagues working together to address issues across multiple scales, the authors can provide case studies of implicit narrative transformation in water quality regulation, water resource development, and water policy strategy. The issue lends itself to the open format of a discussion forum or problem-solving session. The authors are curious to work with others to compare experiences and extend the practical implications of this approach. "



[74] DMDU in the IPCC's 6th Assessment Report   

Authors: Sadie McEvoy; Rob Lempert; Marjolijn Haasnoot; Judy Lawrence, Deltares; RAND, Deltares/Utrecht University, Victoria University Wellington 


"The UN Intergovernmental Panel on Climate Change (IPCC) has increasingly embraced decision making under deep uncertainty in its assessment and special reports. Founding members of the Society for Decision Making under Deep Uncertainty (DMDU) held myriad positions in the IPCC’s recent 6th Assessment Report for Working Group II, focused on Impact, Risk and Adaptation to climate change. This session proposes to share these authors’ experiences of the IPCC process and how deep uncertainty is included in the most recent reports. The format for the proposed session is a series of presentations by IPCC author members, followed by Q&A. The proposed topics and IPCC author presenters are: DMDU in the IPCC – an overview, Rob Lempert: Since AR5, DMDU and its influence are now recognized in the framing of AR6-WGII (impact, risk and adaptation), in the presentation of sea level rise projections in AR6-WGI (the physical science basis), and in SROCC. Adaptive pathways in the AR6 WGII, Marjolijn Haasnoot: Pathways are used throughout the WGII report, in sectoral and regional chapters. Sea level rise and deep uncertainty in AR6 WGII, Judy Lawrence: Sea level rise is a topic where deep uncertainty methods and thinking are especially present in the WGII report. Climate Resilient Development (Pathways), Ben Preston/Sadie McEvoy: Climate Resilient Development Pathways (CRDPs) are an important focus of the latest WGII report. Conceptually, CRDPs aim to integrate adaptation and mitigation into sustainable development for a more just and resilient future for people and planet. Regional and sectoral chapters assessed CRDPs for various chapters as contribution to the dedicated CRDP chapter."

Machine Learning & Computational Intelligence

[28] Policy tree optimization for near-real-time decision making under deep uncertainty

Authors: Irene S. van Droffelaar; Jan H. Kwakkel; Alexander Verbraeck, Delft University of Technology


"The fugitive interception problem asks, ""what is the optimal set of routes for a fleet of police units to traverse in order to maximize the probability of intercepting a fleeing fugitive?"" Models can be used to support the decision, as long as the optimal solution is calculated in near-real-time. The unpredictability of the fugitive, and not knowing if, where, and when a sensor detects the fugitive, introduce deep uncertainty to the problem. Moreover, there is clear path dependency: sending police units in a certain direction constrains their possible rerouting in the future. There is a trade-off between the flexibility to react to new information and the timeliness of decisions. Traditional stochastic optimization methods used for solving the fugitive interception do not account for either the deep uncertainty or the path dependency. However, most techniques developed for adaptive decision-making under deep uncertainty are developed for long-term planning problems, with ample time for analysis and intermediate input from decision makers. Therefore, we explore the potential of policy tree optimization for the identification of the robust set of routes, given deep uncertainty and time pressure, for a fugitive interception problem. Policy tree optimization applies heuristic methods to find optimal actions, based on both short-term and long-term information. The method yields a binary tree that delineates under what conditions (i.e., sensor input) what actions should be taken. This study evaluates the finite-time performance of policy tree optimization on a stylized fugitive interception problem. We demonstrate an approach to find optimal actions when faced with deep uncertainty and time pressure using policy tree optimization."

[91] Mapping Climate Policy Interdependences across Regions under Uncertainty

Authors: Hermilo Cortés-González; Edmundo Molina-Perez, Tecnologico de Monterrey


This study uses DMDU methods to evaluate and analyze processes of technological change between advanced and emerging countries, and its connection with economic growth and climate change through the EDIAM modeling framework (Molina, 2016). In this study, the EDIAM framework is expanded to include updated CMIP6 data ensembles and disaggregated to three different global regions: America, Europe-Africa and Asia. This new version of EDIAM is used in a computational experiment that estimates the optimal mitigation route for a wide range of model parametrization capturing uncertainty about climate and technological change. The resulting database is then mined using high-dimensional stacking and machine learning method to produce a portfolio of adaptive strategies that balance cost across regions and global risk of not meeting the 1.5C target.

[88] Training a multi-sectoral exploratory model (SISEPUEDE) using cross-validation methods: a process for automated estimation of baseline historical and policy conditions in an RDM study 

Authors: Edmundo Molina-Perez; Hermilo Cortés-González; James Syme; Nidhi Kalra; Fernando Estevez; Víctor Manuel Espinoza Juárez; María Esperanza Plaza-Ferreira; Fernando Gómez-Zaldívar; Hilda C. Zamora-Maldonado; Adolfo Javier De Unanue-Tiscareño


Developing exploratory models requires to balance model detail and level of aggregation with the ability to use these models in large-scale computational experiments and mine experimental results to inform concrete policy questions. However, since the introduction of exploratory modeling, thirty years ago, data availability, computer power and the empirical demands of policy analysis have increased markedly. This has created new opportunities for developing more desegregated and data-intensive exploratory models that can enrich RDM analysis. In this talk we introduce a data pipeline and cross-validation calibration process that is used to estimate historical baseline parameters and baseline policy conditions for the SISEPUEDE multisector dynamic exploratory emissions modeling framework. The approach is used to estimate hundreds of sectoral parameters, across 26 different countries, while controlling for overfitting, and by integrating over 30 different data repositories. We discuss possibilities for deploying this pipeline for updating the analysis periodically, expanding it to include more countries and applying to other modeling contexts.

[78] Reinforcement Learning for Decision-Making under Deep Uncertainty

Authors: Zhihao Pei; Fjalar de Haan; Nir Lipovetzky; Enayat A. Moallemi; Angela M. Rojas-Arevalo, The University of Melbourne


"Decision-making process subject to multiple aspects of uncertainty is known as Decision-Making under Deep Uncertainty (DMDU). To support this process, different exploration methods have been applied to evaluate the performance of candidate policies across multiple scenarios, with the objective of identifying a robust plan that performs satisfactorily over a wide range of possible futures. Different exploration methods show complementarity, hence the joint use of them can improve the overall performance of DMDU. The main aim of our research is to introduce reinforcement learning as a complementary exploration method for DMDU. This is motivated by the suitability of reinforcement learning to handle uncertainty, given its inherent plan adaptability. The plan produced by reinforcement learning acts as a function that maps actual system states to actions in real-time, in contrast to stateless solutions provided by the existing exploration methods. State-based plans can adapt to multiple trajectories representing uncertain futures, hence providing better efficiency and robustness. To achieve this aim, we first review existing studies on multi-objective reinforcement learning and robust reinforcement learning. On this basis, we propose a reinforcement learning algorithm applicable in DMDU. We then compare this algorithm with two baseline evolutionary algorithms, which have already been applied to DMDU, in domain problems from the reinforcement learning and DMDU communities. Our experiments empirically show the viability of reinforcement learning as a complementary exploration method to evolutionary algorithms for DMDU problems. The reinforcement learning algorithm not only produces plans robust to deep uncertainty, but also exhibits different advantages compared to the baseline evolutionary algorithms. The former generally provides higher efficiency and robustness to parameter uncertainty, and can handle random initial states. The latter performs better in the presence of objective uncertainty, and has lower convergence variance. Overall, this research also reveals the different characteristics of domain problems and algorithms in the reinforcement learning and DMDU fields, which contributes to setting an agenda for future robust planning research in both fields. "

Novel Technologies and Models

[86] An Implementation of SLEUTH as an Open Platform for Doing Scenario Planning to Predict Urban Growth

Authors: Gonzalo G Peraza Mues; Roberto Ponce Lopez, Institute for the Future of Education, Tecnologico de Monterrey; School of Government and Public Transformation, Tecnologico de Monterrey


"This project proposes a ready to use implementation of a cellular automata for scenario planning applied to urban growth. The most observed geographical pattern of growth characterizing fast growing cities is sprawling. The urban land consumption per capita significantly increased, on average, in cities over the 1990-2000 and 2000 and 2014 periods, according to the Atlas of Urban Expansion. Urban sprawl has dire consequences on miles per vehicle traveled, CO2 and the provision of public services. The evidence has shown that reversing this trend requires decisive policy actions from local governments. The problem is that local governments, especially in developing countries, do not possess the human capital, technical proficiency and financial resources to develop a sophisticated and easy to test policy interventions under a framework of scenario planning. Our project addresses such limitations by reimplementing the SLEUTH model/ simulator developed by Keith C. Clarke, a tool for predicting urban growth that is well established in the literature. We have reimplemented the SLEUTH model/simulator to increase the access of local governments to scenario planning for urban growth through an easy and ready to use digital tool that works with open access data. SLEUTH is a cellular automaton that simulates 3 growth processes: spontaneous growth, edge growth, and road influence growth. Each growth phase is controlled by a set of parameters calibrated using past growth data. Three are the main limitations of current implementations of SLEUTH to be widely used by local governments and regions. First, the code of most implementations is written in C++, and does not follow modern and best practices for coding. Second, the model needs to be fed with satellite imagery and raster files on historic urbanized land, elevations, and roads access. Third, the process of calibrating the parameters to historical data is long, tedious, and not automated. These three elements are a barrier of entry for local governments without a developed technical capacity to do scenario planning for urban growth. To address these limitations in SLEUTH, we built an interactive digital platform that performs remote data integration from Google Earth Engine to feed the satellite imagery and rasters into the SLEUTH model in an automated way, allowing a real time exploration of SLEUTH generated scenarios. We reimplemented SLEUTH program in Python to better integration with modern data formats and frameworks and facilitate experimenting and new feature development for SLEUTH based growth models. We tested new calibration methods based on machine learning to improve SLEUTH's calibration speed. The platform simulates growth for all major cities in Latin America in an automated way, relying on open access platforms and data."

[81] Development of an Online Global Platform for Rapid Bottom-up Climate Risk Assessment

Authors: Ümit Taner; Homero Paltan; Diego Juan Rodriguez; Peter Gijsbers; Helene Boisgontier; Laurene Bouaziz; Mark Hegnauer; Nathan L. Engle; Laura Bonzanigo, Deltares; The World Bank


"Water planners are interested to know how much additional risk they face due to climate change. Practitioners often estimate risk “top-down” by using bias-corrected climate projections as their starting point. However, this may lead to a poor estimation of natural variability and the true range of uncertainties from climate change. Bottom-up risk assessment methods address these limitations by principally focusing on stress testing of water systems to a wide range of plausible futures to identify undesired outcomes or vulnerabilities. While bottom-up methods are shown to be better equipped for deep climate uncertainties, their widespread use in water planning is still limited and mainly restricted to academia. This can be attributed to multiple factors, including the need for somewhat more sophisticated modeling tools such as weather generators, analytical procedures with steep learning curves, and the considerable effort needed to process, visualize and communicate the results with decision-makers. To address these challenges, we present a free, open-source web platform for rapid bottom-up climate risk assessment. The platform enables the assessment of hydroclimatological risk in any geographic region or river basin through cloud-based computing, freely available hydrological modeling tools, and global gridded datasets. Users will be able to assess natural variability and climate change signals in a selected project location, design and execute a climate stress test for mapping drought and/or flood-related vulnerabilities and finally make a judgment on the plausibility of vulnerabilities identified using climate model projections. The graphical user interface consists of interactive elements to enable the selection of climate stress test parameters, hydrological performance metrics, and related acceptable performance thresholds. Results are displayed through appropriate tabular formats and interactive visualizations that are downloadable on demand. The platform is intended to provide an initial diagnostic of climate vulnerabilities of water projects as part of structured bottom-up and resilience frameworks."

[94] Near real-time forecasting to support dynamic decision making during project execution to enhance outcomes

Authors: Alan Mosca; Yael Grushka-Cockayne, nPlan; UVA Darden School of Business


"Large projects (construction, engineering, infrastructure) are planned over long periods of time, and in a mostly static fashion. Critical Path Methodologies and large Gantt charts are still the main tools used in industry to plan projects that may take several years to complete. The outcomes are typically well known: projects are often delayed according to the original plan and this results in budgets being overrun significantly. We propose a new methodology of project execution centered around forecasting delay and conditional planning, designing from the beginning points of “flexibility” around the original Gantts, and even creating novel options for deferred decision making. The forecast itself will be able to highlight the points of weakness in the original plans, enabling the project managers to execute the deferred decisions and implement the pre-planned options, as well as implementing new ones in a flexible way. This session will cover the main points of how to use forecasts for better decision making rather than suggest methods for forecasting more efficiently."

[10] Stochastic simulation with informed rotations of Gaussian quadratures

Authors: Davit Stepanyan; Georg Zimmermann; Harald Grethe, bJohann Heinrich von Thünen-Institut, Federal Research Institute for Rural Areas, Forestry and Fisheries, Institute of Farm Economics, Braunschweig, Germany; University of Hohenheim, Institute of Applied Mathematics and Statistics (110) and CSL, Stuttgart, Germany; Humboldt-Universität zu Berlin, International Agricultural Trade and Development Group, Berlin, Germany


"Given the fast growth of available computational capacities and the increasing complexity of simulation models addressing agro environmental issues, uncertainty analysis using stochastic techniques has become a standard modeling practice. However, conventional uncertainty/sensitivity analysis methods are either computationally demanding (Monte Carlo-based methods) or produce results with varying quality (Gaussian quadratures). In this article, we present a computationally inexpensive and reliable uncertainty analysis method for simulation models called informed rotations of Gaussian quadratures (IRGQ). We also provide an R script that generates IRGQ points based on the required input data. The results demonstrate that this method is able to produce approximations that are close to the estimated benchmarks at low computational costs. The method is tested in three different simulation models using different input data in order to demonstrate the independence of the proposed method on specific model types and data structures. This is a methodological paper for practitioners rather than theorists. This is a published work."

Multi-actor Systems and Exploratory Analysis

[37] Process-Rich Scenario Generation Enhances Exploratory Decision Support for Deeply Uncertain Hydroclimatic Extremes

Authors: Rohini S. Gupta; Scott Steinschneider; Patrick M. Reed, Cornell University: Department of Civil and Environmental Engineering; Cornell University: Department of Biological and Environmental Engineering


"Robust planning and management of institutionally complex food-energy-water (FEW) systems requires a thorough exploration of system behavior under a broad set of hydroclimatic scenarios that capture plausible pluvial and drought extremes. It is important for these scenarios to be (1) physically informed and (2) representative of natural variability and possible change inherent in the region’s weather system. Furthermore, scenario generation methods support exploratory modeling efforts to compare and contextualize results with CMIP5/6 projections. We contribute and demonstrate a process-rich stochastic scenario generation framework to aid bottom-up water vulnerability assessments for California’s (CA) FEW system. This system is characterized by complex interactions among state-wide transfer projects, federally managed reservoirs, and local irrigation districts, water utilities, and groundwater banks. Ensembles of temperature and precipitation for 12 key river basins in CA’s Central Valley are generated using a weather-regime based stochastic weather generator, linking local weather to associated patterns of large-scale atmospheric flow (i.e., weather regimes). This weather generator allows for the systematic exploration of water system response to specific mechanisms of change in the climate system, including thermodynamic intensification of precipitation and dynamic shifts in atmospheric circulation. Conditioning the generator solely on the short historic record can greatly limit the characterization of extremes and natural variability. Thus we incorporate an extended representation of natural variability by reconstructing weather regimes based on 600 years of tree-ring based moisture proxies to create weather ensembles that are capture megadrought and pluvial periods. We then use these weather scenarios to force hydrologic models for each basin, generating internally consistent, daily, ensembles of streamflow across the Central Valley. Preliminary results show that a large portion of variance in extreme hydroclimatic events can be attributed to natural variability rather than thermodynamic climate changes representative of anthropogenic warming. However, ensembles conditioned on both paleo-informed natural variability and thermodynamic climate changes result in hydrologic extremes that far exceed those observed historically or generated under natural variability or thermodynamic changes in isolation."

[108] DMDU In a Multi-Actor Participatory Setting

Authors: Vincent Marchau and Warren Walker


This paper deals with the issue of improving the applicability of DMDU approaches in the context of multi-actor participatory decisionmaking. DMDU studies have traditionally focused mainly on the problems and approaches to solving them than on the fact that there are usually multiple stakeholders with their own goals, means, stakes, knowledge, and values that need to be taken into account. Under these conditions, predictions are not possible and policy choices may be highly contested. So, decisionmaking has to shift from a ‘predict and act’ approach to a ‘prepare and adapt’ approach. This ‘prepare and adapt’ approach involves 3 phases, which we have labelled framing (the problem), exploring (uncertainties), and choosing (actions). In this paper, we show how this 3-phase approach can be applied in a multi-actor setting to the recent problem of ‘adaptive pandemic management’ as applied to a region in the Netherlands (South Gelderland).

[46] DMDU approach in Participatory Integrated Assessment: crafting regulations to wicked problems using qualitative modeling

Authors: Tatiana Merino-Benítez;Ileana Grave;Luis A. Bojórquez-Tapia, Laboratorio Nacional de Ciencias de la Sostenibilidad, Instituto de Ecología, Universidad Nacional Autónoma de México, México


Participation has been a buzzword in planning for sustainable development. It is a means through which governments, civil society and stakeholders can work together to devise and implement practical solutions, including crafting regulations, to sustainability problems. Participatory Integrated Assessment (PIA) has been implemented as a transdisciplinary process for organizing and combining the constituent elements of a sustainability problem into a coordinated harmonious scheme to improve decision-making. In most countries, PIA has been applied in decision-making contexts that mainly involve qualitative knowledge and information (e.g., in Mexican Ecological Ordinance (EO), a policy-making tool for multi-sectoral, environmental planning). Given that PIA involves imagining futures and aims to address wicked problems through trans-sector participation, it involves huge amount of knowledge and decision analytics that has methodologically led toward Decision Making under Deep Uncertainty (DMDU). However, DMDU has not yet been applied in real-world contexts that involve data only available in qualitative terms. In this paper we propose a procedure to engage with the highly conflictive and uncertain environmental decision-making arena, based upon an implementation of system dynamics mediated modeling and DMDU techniques in an environmental risk assessment. We present our proposed methodology through the forecast of adverse socio-environmental effects of the EO of Yucatán, México. Our contribution serves to (1) expand the scope of DMDU approaches to incorporate qualitative data in a highly conflictive and politicized multi-sectoral planning context; and (2) acknowledge the uncertainty involved in PIA for environmental planning.

Infrastructure Planning

[38] Planning for the unexpected: Supporting flexibility in infrastructure decision making 

Authors: Bonjean Stanton, Muriel; Roelich, Katy; Few, Sheridan,   Sustainability Research Institute, School of Earth and Environment, University of Leeds, Leeds LS29JT, United Kingdom


Infrastructure are socio-technical systems composed not only of physical assets but also of the institutions that manage, govern, finance, and regulate them. Yet, recent shocks (e.g. the Covid pandemic) and rapid changing conditions (e.g. more frequent extreme flooding events) highlighted the limited capacity of existing infrastructure to provide services in a future riddled with deep uncertainties. Navigating these deep uncertainties and ensuring the continuity of services in the near- and long-term future require new ways of planning. Decision Making Under Deep Uncertainty (DMDU) approaches have real potential to support robust and adaptive decisions for complex systems. Yet, to date, little is known about how DMDU can be operationalised in infrastructure decision-making. This work presents some insights about how a UK transport organisation can make use of DMDU methods and principles at different stages of the decision-making process to ensure infrastructure are more resilient to future changes.

[75] At the Crossroads: implications of not developing mega hydro in the Madeira River Basin on Brazil and Bolivia’s energy development

Author: Stuart, Hayley; University of Oxford School of Geography and the Environment


"Since the early 2000s, Brazilian and Bolivian national development plans have been oriented toward the goal of energetically connecting the two nations. This interconnectedness is largely contingent on harnessing the power of the Madeira River through the construction of four mega-dams in Bolivia and exporting that energy to Brazil. These projects are the Chepete-Bala, the Río Grande Complex, Cachuela Esperanza, and the binational Guayaramerín mega hydroelectric projects. However, in 2020 the Universidad Católica Boliviana published a proposed roadmap for Bolivia’s energy transition by 2050 that does not include these four mega-projects and instead advocates for Bolivia to develop internal energetic security. Guided by a Decision Making Under Uncertainty framework with a special on epistemic risk, I comparatively analyze each energy plan and explore the tradeoffs of whether or not to electrically integrate Bolivia and Brazil via the Madeira River projects. Epistemic risk arises “from incomplete theory, incomplete understanding of a system, modeling limitations, and/or limited data.” (Gorris et al, 2014). I draw from three major documents: Bolivia’s Energy Development Plan 2025 (Bol 2025); the recently proposed Roadmap for Energy Transition in Bolivia 2050 (HdR 2050); and Brazil’s National Energy Plan 2050 (PNE 2050). I argue that despite various tradeoffs, the HdR proposal ultimately offers a pathway forward for Bolivia and Brazil that contains significantly less epistemic risk than the option to develop the Madeira River. Finally, I identify information that is still needed to reduce epistemic risk and make sound transboundary decisions moving forward."

[15] Uncertainty in action: Implementation uncertainty and how it shapes dynamic and adaptive cooperative regional water supply infrastructure pathways.

Authors: Lillian Bei Jia Lau; Patrick M. Reed; David F. Gold, Cornell University


"Urban water utilities are increasingly exploring the use of cooperative regional infrastructure investment and water supply portfolio pathways as a key strategy for confronting the deep uncertainties driven by a growing population and climate change pressures. Recent work emphasizes the importance of regional cooperation in crafting dynamic and adaptive infrastructure pathways that must satisfy reliability, financial and equity objectives while remaining robust to deeply uncertain hydroclimatic and socioeconomic scenarios. Theoretically, regional cooperation should afford utilities better resource efficiency and supply reliability. To date, however, little work has explored how utilities may be exposed to performance and robustness degradations induced by uncertainties in member utilities’ actual implementation of the cooperative infrastructure investment and water supply portfolio management actions. In this study, we characterize the effects of implementation uncertainty in cooperative regional water supply pathways with a focus on three central questions. First, can implementation uncertainty impair the performance of operational objectives and system robustness? Second, how do these risks cascade into the planning of investments in long-term supply infrastructure investment pathways? Third, how do alternative compromise strategies across candidate regionally robust portfolios influence their inherent vulnerabilities to implementation uncertainty across actors and time? Our framework, demonstrated on the challenging Sedento Valley DMDU benchmarking test case, examines and characterizes the path-dependency of implementation uncertainties in short-term operational drought mitigation instruments and long-term, risk-triggered infrastructure investment pathways. Our results yield safe operating spaces beyond which significant vulnerabilities may emerge, causing unexpected shifts in individual and regional robustness and attained performance objectives. These shifts alter the investment pathways of new supply infrastructure and increase the potential for regional conflicts due to asymmetries between member utilities’ vulnerabilities and robustness. Overall, this methodology is broadly applicable to water supply systems seeking to better understand how to better implement cooperatively stable and robust regional water supply infrastructure and portfolio management pathways."

[36] Targeting Adaptive Infrastructure Management using Climate Variability and Information

Author: Sarah Fletcher, Stanford University


To enable low-cost water reliability in a changing climate, adaptive management approaches use near-term indicators (e.g., trends in precipitation change) to trigger actions like new infrastructure development or updated operating policies. This adaptive approach assumes that near-term indicators provide useful information about future conditions. While this approach is effective in hydroclimates where trend dominates precipitation uncertainty, learning about long-term trends is confounded in regions with high variability on annual to decadal scales. In these regions, an apparent trend in precipitation may ultimately be a persistent anomaly, defined as a period of several years that is unusually dry or wet yet eventually returns to normal conditions. This reduces the effectiveness of near-term indicators in adaptive management. Here, I present progress towards an adaptive planning approach that uses patterns of climate variability to quantify opportunities to learn about precipitation uncertainty. This information is then used to design differentiated adaptive management approaches based on the modes of climate variability. I apply this to a case study in the Mwache River in Kenya. Results show that learning about precipitation trends and using information about climate oscillation patterns can increase the effectiveness of both adaptive storage planning and adaptive reservoir operations.


[69] Understanding Complex Adaptive Human-Earth Systems through Multisector Dynamics and DMDU

Authors: Antonia Hadjimichael; Patrick Reed, Penn State University; Cornell University


The field of MultiSector Dynamics (MSD) explores the dynamics and co-evolutionary pathways of human and Earth systems with a focus on critical goods, services, and amenities delivered to people through interdependent sectors. Central to the MSD vision is the ambition of advancing the next generation of complex adaptive human-Earth systems science to better address interconnected risks, increase resilience, and improve sustainability. These challenges require dynamic and adaptive action pathways that balance diverse societal objectives and account for complex feedbacks and deep uncertainties. This session aims to serve as a venue to strengthen the connections between the MSD and DMDU communities, by discussing common themes and challenges, as well as the potential for more integration across the two fields of research.


[83] Quo vadis adaptation pathways? – Recent advancements, experience and challenges in adaptation pathway research

Authors: Sadie McEvoy; Marjolijn Haasnoot; Julius Schlumberger; Sarah Wright; Jon Lamontagne; David Gold; Susanne Hanger-Kopp, Deltares, NLD; Deltares, NLD; Deltares, Vrije Universiteit Amsterdam NLD; Deltares, University of Utrecht, NL; Tufts University, USA; Cornell University, USA; IIASA, AUT


"Pathways thinking and specific approaches, like Dynamic Adaptive Policy Pathways (DAPP), are tackling decision making under deep uncertainty and ensuring objectives are maintained into the future. Pathways thinking and approaches sequence decisions and measures over time to build flexibility and robustness into decision-making and planning processes involving complex and deeply uncertain aspects and developments of the climate and society. Pathways allow the exploration of scenarios of plausible characteristics and dynamics of current and future systems and has been applied using a wide range of qualitative and quantitative methods. While pathways thinking and approaches are well established, methods, concepts and tools used in the application of pathway thinking are advancing on multiple fronts. These advances fall into two categories. The first category focuses on advancing existing approaches and tools to capture the decision-making context as realistically as possible by, for example accounting for (a) (dynamics of) multiple stakeholders with varying, sometimes contested objectives, (b) a multitude of (interconnected) drivers for adaptation, and (c) the integration of adaptation, mitigation and sustainable development ambitions into Climate Resilient Development Pathways. The second category of advances focuses on novel tools and methods to create adaptive plans, for example to (a) enable modelling of power, equity, and cooperative stability in multi-sector systems, (b) to understand and operationalize path-dependencies. We propose an interactive session of leading adaptation pathways researchers and practitioners to share their latest advances and experiences in pathways research and applications. The session is envisaged to comprise three parts. First, experiences and reflections from ten years of developments in pathways research will be reflected upon. Second, the contributors will share their work and experiences on key areas of advances in pathways. Finally, the session will facilitate an exchange on experiences, current challenges and topics, using a world café or similar format. We intend to organize the session fully in person and have organized the following topics and presenters: 1. Experiences and observations from a decade of pathways developments (Presenter: Marjolijn Haasnoot) 2. Tailoring a framework for adaptation pathways in a multi-hazard risk, multi-sector context (Presenter: Julius Schlumberger) 3. Operationalizing Climate Resilient Development Pathways for European cities (Presenter: Sarah Wright) 4. Reflection on past efforts and ongoing research on multi-sector dynamics (Presenter: Jon Lamontagne) 5. Equitable, Robust, Adaptive and Stable Deeply Uncertain Pathways: a new framework for exploring cooperative stability and multi-actor vulnerabilities in regional water supply management and infrastructure investment pathways (Presenter: David Gold) 6. Path dependencies in flood risk management: operationalization and empirical evidence (Presenter: Susanne Hanger-Kopp)"

bottom of page