Unveiling the Human-Climate Nexus through the CLIMAgenAI framework: a Multi-Source Causal Framework for Climate Risk Analysis and AI-Supported Disaster Governance
This paper introduces CLIMAgenAI, a multi-source causal framework that integrates judicial, archival, and multimedia evidence into a unified representation of climate-related disasters to support AI-driven risk analysis and governance, as demonstrated through the reconstruction of the 1998 Sarno debris-flow event.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Disasters rarely happen in a vacuum. When a flood sweeps through a valley or a landslide buries a village, the physical force of nature is only part of the story. The true scale of the damage often depends on a complex web of human decisions, forgotten warnings, and the way communities are built. For decades, scientists and engineers have tried to measure this risk, but their tools have often been too narrow, looking at the weather in one department and the buildings in another, rarely connecting the two. This separation leaves a gap in our understanding: we know the rain fell, but we often miss the chain of choices that turned a storm into a tragedy. To fix this, researchers are developing new ways to piece together the full history of a disaster, treating it not just as a natural event, but as a human one.
A team of researchers at the University of Rome Tor Vergata has introduced a new method called CLIMAgenAI to bridge this gap. Their goal is to create a single, unified picture of how climate-related disasters unfold by weaving together three very different types of evidence that are usually kept apart: court rulings, old government archives, and photographs or news footage. Instead of relying on a single report, this framework treats a disaster as a puzzle where the legal documents explain who was responsible, the archives reveal what was planned or ignored, and the images show exactly what the ground looked like when the water rose. By combining these sources, the researchers can trace a clear line from the initial weather event all the way to the final damage, showing exactly where human choices amplified the danger.
To test this approach, the team turned to a devastating event that occurred in 1998 in Sarno, Italy. On that day, heavy rains triggered a massive flow of mud and debris that rushed down steep hillsides and into the town below, causing significant loss of life and destruction. The researchers gathered every available document related to the event. They pulled court decisions that detailed the legal findings of what went wrong, dug through historical records to find reports on infrastructure and planning, and collected photographs and news clips that captured the scene. They then fed all this information into their new system, which acts like a structured organizer. It sorts every piece of information into three categories: natural factors like the rain and the soil, contextual factors like the location of houses and the state of roads, and behavioral factors like the decisions made by officials or the actions of residents.
The result is a detailed map of cause and effect that reveals how different threads of the disaster were connected. In the case of Sarno, the framework showed that the disaster was not just a result of intense rainfall. It highlighted how the natural danger was made worse by specific human choices, such as building homes in vulnerable areas and failing to maintain drainage channels. It also pointed to behavioral failures, such as delays in issuing warnings or a lack of coordination among emergency services. By laying these elements out side by side, the system makes it possible to see exactly how a natural hazard transformed into a human catastrophe. It shows that the vulnerability of the town was not a fixed condition but the outcome of a series of decisions and oversights that could have been identified and addressed beforehand.
This method offers a powerful tool for understanding risk because it treats exposure and vulnerability as things that are built over time, rather than just static numbers. The researchers found that by integrating these diverse sources, they could reconstruct the event with a level of detail that single-source reports never achieve. The court rulings provided a clear account of institutional failures, the archives offered a timeline of planning decisions, and the visual materials confirmed the physical reality of the damage. When combined, they created a robust record that tells the full story of the disaster, from the first drop of rain to the final rescue effort.
The ultimate aim of this work is to use this structured knowledge to help people make better decisions in the future. The researchers propose that this detailed, evidence-based map of a disaster can be used by artificial intelligence systems to generate clear, understandable explanations for the public and for policymakers. Instead of presenting complex data that only experts can read, these systems could translate the findings into plain language, helping communities understand exactly why a disaster happened and what specific steps could prevent it next time. This approach does not replace human judgment; rather, it provides a solid foundation of facts that ensures any advice or warning is grounded in the actual history of the event.
While the study focused on a single event in Italy, the method itself is designed to be applied to other disasters around the world. The researchers acknowledge that their work is a proof of concept, demonstrating that it is possible to merge legal, historical, and visual records into a single, coherent analysis. They note that future work will need to test this framework on different types of hazards and in different legal systems to see how well it holds up. However, the core idea remains clear: by looking at the full record of a disaster, including the human choices that shaped it, we can move beyond simple warnings and start building a deeper, more accurate understanding of climate risk. This shift allows society to see disasters not as unavoidable acts of nature, but as complex events where human agency plays a decisive role, offering a clearer path toward safer communities.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.