Work

Unveiling the Human-Climate Nexus through the CLIMAgenAI framework: a Multi-Source Causal Framework for Climate Risk Analysis and AI-Supported Disaster Governance

Year: 2026
Type: preprint
Abstract: Climate risk emerges from the interaction of hazards, exposure and vulnerability within complex socio‑economic and governance systems, yet existing approaches to risk assessment often remain fragmented across disciplines and rely on sector‑specific assessment and mitigation methods that rarely capture complex multi-sector event trajectories. This paper introduces CLIMAgenAI, a multi-source causal framework that extends the classical cause–effect matrix to integrate judicial rulings, archival records and multimedia materials into a unified process-based representation of climate-related disasters. Building on Fault Tree Analysis (FTA), Event Tree Analysis (ETA) and Human and Organizational Factors Engineering (HOFE), CLIMAgenAI systematically extracts and codes causal relations from narrative and visual documents, reconstructing multi‑level pathways linking natural, contextual and institutional factors to disaster outcomes. The framework is demonstrated through the 1998 Sarno (Italy) debris-flow disaster and produces structured outputs—including extended cause–effect matrices, causal pathways and comparative datasets—that provide a knowledge base for AI-supported climate risk communication, stakeholder engagement and forensic disaster governance. Unlike previous approaches that analyse individual sources separately, CLIMAgenAI reconstructs climate-related disasters through a unified causal representation integrating judicial, archival and multimedia evidence. The objective is not to develop a new GenAI approach itself, but to demonstrate that integrating heterogeneous documentary sources into a standardized causal framework creates a robust, replicable and scalable knowledge base for future AI-supported analysis and decision support. .
Language: English
Open Access status: green
OpenAlex ID: W7204178300
Cites: 36
Cited by:
Related to: 10