Five Roadblocks Keeping Disaster Science From Tackling Cascading Climate and Health Risks

When a magnitude 7.2 earthquake struck Haiti’s southern peninsula in August 2021, it destroyed roughly 137,000 homes, killed at least 2,248 people, and triggered landslides that severed remote communities from help. Days later, Tropical Storm Grace swept in, its torrential rains paralysing rescue operations just as the Covid-19 pandemic was already straining the country’s healthcare system. By 2022, the collapse of water and sanitation infrastructure had ignited a cholera outbreak that claimed more than 200 lives. The catastrophe was not a single disaster but a chain of interacting crises, and it exemplifies precisely the kind of compound, cascading risk that the world’s disaster science enterprise remains poorly equipped to understand, model, and manage.

That is the central argument of a major new perspective paper published in the journal Natural Hazards and Earth System Sciences. An interdisciplinary team led by Philip J. Ward of the Institute for Environmental Studies at Vrije Universiteit Amsterdam, together with Marleen C. de Ruiter, Kees Boersma, Jasper Verschuur, Nicole van Maanen and more than a dozen colleagues spanning climate risk, public health, behavioural science, crisis management and disaster governance, sets out five core challenges that stand between today’s fragmented risk research and what the United Nations has called Comprehensive Disaster and Climate Risk Management, or CRM. The paper, published as an invited highlight contribution, arrives in the wake of the Sendai Framework’s Mid-Term Review, which concluded bluntly that science is falling short in understanding complex multi-risks.

The authors’ starting point is a conceptual shift that has gathered momentum over the past decade. Risk, as the UN Office for Disaster Risk Reduction defines it, is a function of hazard, exposure and vulnerability. Traditional assessments treated these ingredients as static, studying each hazard in isolation. Researchers now recognise that hazards interact, triggering, amplifying or compounding one another, and that exposure and vulnerability themselves evolve over time, sometimes for decades. Slow-onset crises such as droughts and pandemics erode financial resources, education, health and social networks, while rapid-onset events like floods and earthquakes displace people into new exposures. A person displaced by flooding may encounter vector-borne disease; a community enduring an outbreak may face heightened exposure because its drinking water infrastructure has been damaged.

The links between climatic hazards and infectious disease are particularly striking. Flooding and high temperatures can expand breeding grounds for mosquitoes, amplifying malaria and dengue transmission, while floodwaters spread the bacteria responsible for cholera. A systematic review cited in the paper found that 58 percent of all known human infectious diseases, including Zika, Chikungunya and Ebola, have been aggravated by climatic hazards through more than 1,000 unique transmission pathways. Yet disaster risk models almost never include biological hazards, and health systems rarely plan for geophysical and meteorological shocks, leaving what the authors describe as double or even triple health burdens when physical injury, disease and mental health impacts coincide.

The first of the five challenges is observational: we lack comprehensive, high-quality data on how societies are impacted by, prepare for, and respond to multi-risks. Health scientists curate disease data, hydrologists curate flood and drought data, and seismologists curate earthquake data, each in disciplinary silos with incompatible standards. Spatial and temporal scales rarely match, so urban heat islands, flash floods and informal-settlement vulnerabilities fall below the resolution of global datasets. Vulnerability is captured in episodic snapshots such as censuses that quickly go stale, while impact databases like EM-DAT record only aggregated, largely tangible losses, overlooking mental health, social cohesion and long-term path dependencies. In many low- and middle-income countries, the sustainability of health information systems itself is threatened by budget cuts to external funding.

The second challenge follows directly: our understanding of why people, communities and institutions prepare, or fail to prepare, for complex multi-risks remains incomplete. Classic theories such as Expected Utility Theory, Prospect Theory, the Theory of Planned Behaviour and Protection Motivation Theory were largely built for single hazards and one-time decisions. In a multi-hazard world, households facing frequent floods and rare earthquakes may misallocate scarce resources, or conversely, experience with one risk may raise awareness of another. Systemic risks transmitted through food prices, supply chains and disease outbreaks compound the problem, because people perceive remotely materialising risks differently from local ones. Most empirical studies capture a single snapshot, so feedback loops between protective behaviour and declining risk perception remain theoretically hypothesised but empirically unverified.

Third, the capacity to simulate how current and future risk could be reduced is hampered by these observational and behavioural gaps. A review of roughly 70 large-scale risk models found that only two included dynamic vulnerability. Agent-based models and system-dynamics approaches have advanced for single hazards such as floods, droughts and outbreak response, but little work extends them to multi-risk settings. Validation is another bottleneck: disasters are inherently local processes, yet validation data arrive as coarse aggregates, leaving low-probability, high-impact events, which climate change may intensify, dangerously under-constrained. The authors call for simulations that couple hazards with dynamic exposure, vulnerability and human decision-making across societal groups, enabled by high-performance computing and richer empirical foundations.

Fourth, governance structures often reinforce the very silos that plague data and models. Mandates overlap or leave gaps when risks interact, political and budgetary cycles favour short-term action, and operational institutions are organised in isolated pillars that slow decision-making. Public participation, where it exists, is frequently semi-formal and structurally weak. The authors advocate poly-centric and net-centric governance approaches: adaptive, networked arrangements that distribute authority vertically and horizontally, connect state and non-state actors, and support collective sensemaking, with digital infrastructures for data interoperability and innovations such as citizen climate assemblies as potential test beds for deeper engagement.

The fifth challenge concerns knowledge exchange at the science, policy and practice interface. The multi-risk problem is, in the authors’ words, super wicked: complex, evolving and without definitive solutions. During the Covid-19 pandemic in the Netherlands, biomedical advice dominated early scientific guidance while social and economic perspectives were added later in parallel structures, illustrating how advisory architectures can entrench compartmentalisation. Research from Western, Educated, Industrialized, Rich and Democratic contexts travels more readily into policy than locally grounded and tacit knowledge, reinforcing epistemic injustices and blind spots about marginalised groups. Knowledge translation is often treated as linear dissemination rather than dialogue and co-production, and even co-located disciplines fail to integrate without deliberate frameworks and boundary objects such as participatory maps, outbreak scenarios and serious games.

Looking forward, the team sketches an ambitious agenda built on five pillars: a Disaster Risk Observatory integrating satellite imagery, census data, surveys and disaster forensics into a living system that tracks risk dynamics in near real time; a new generation of risk-behaviour theories incorporating learning, feedbacks and systemic interdependence; integrative modelling frameworks, notably agent-based, that are optimised locally but scalable globally; poly-centric governance for sharing actionable multi-risk data and models; and systematic study of the science-policy-practice interface itself, supported by knowledge brokers who span disciplinary and administrative boundaries. Crucially, the authors stress that the five challenges are mutually reinforcing, each compounding the others into a persistent actionability gap. Closing it, they argue, will require funding mechanisms and research efforts that reach well beyond the traditional one-to-four-year project cycle, engaging meaningfully with the Sendai Framework, the WHO Health Emergency and Disaster Risk Management Framework, and the global climate agenda.

Subject of Research: Comprehensive disaster and climate risk management and the scientific challenges of multi-risk assessment

Article Title: Invited Perspectives: Science for Comprehensive Disaster and Climate Risk Management

Article References: Ward, P. J., de Ruiter, M. C., Boersma, K., Cesuroglu, T., Clark, N., Comes, T., Dahal, A., Fransen, S., Verschuur, J., Anthonj, C., Balakrishnan, S., de Bruijn, J., Hendriks, E., Mena Fluhmann, R., Perlaviciute, G., Steg, L., van den Homberg, M. J. C., van Westen, C., Atun, F., … van Maanen, N. (2026). Invited Perspectives: Science for Comprehensive Disaster and Climate Risk Management. Natural Hazards and Earth System Sciences, 26(9), 4589-4609. https://doi.org/10.5194/nhess-26-4589-2026

Image Credits: AI Generated

DOI: 10.5194/nhess-26-4589-2026

Keywords: disaster risk management, climate risk, multi-hazard, cascading disasters, infectious disease, vulnerability dynamics, agent-based modelling, disaster governance, Sendai Framework, risk perception, knowledge exchange, Haiti earthquake