Search for a command to run...
The resilience of interconnected critical infrastructure systems remains poorly assessed, as are the social and economic impacts of their disruptions. The increase of interdependencies between systems and the emergence of new infrastructure types, such as for renewable energy infrastructure, add further complexity to existing systems. However, a main challenge in addressing the systemic risk of disrupted critical infrastructure is the lack of freely available and consistent data about its impacts. To address this gap, we develop a prototype database on the impacts of infrastructure disruptions in Europe that occurred in the last two decades. We focus on large-scale climate hazards: riverine and coastal floods, severe storms, droughts, heatwaves, and wildfires. For building the database, we leverage an interdisciplinary approach that blends natural language processing with geospatial analysis and network modelling. Our database combines information extracted from scientific publications and newspaper articles by means of an automated framework that orchestrates different language models and their specific tasks. This approach facilitates the extraction of information on infrastructure disruptions within the transport, energy, social, water, and waste treatment sectors, along with their cascading impacts. The reliability of our framework is strengthened by a thorough evaluation of its models and the traceability of the extracted impact data by the user. In addition, we benchmark the ability of our framework to extract such complex information against Google’s LangExtract, a Python library to retrieve structured information from various text sources. The output of our framework is a dataset on the affected critical infrastructure type, its location, and damage type, along with its cascading impacts on other infrastructure components and socioeconomic effects caused by its disruption. In a later step, we enrich our data with information from other state-of-the-art disaster datasets to generate a freely accessible infrastructure impact database comprising climate hazard, exposure, and vulnerability information at a high spatial resolution for Europe. This database will facilitate the analysis and modelling of systemic risks from disrupted critical infrastructure.