D2.4 Data fusion and AI driven innovations for RKBs

Le résumé fourni par la source

D2.4 presents the outcomes of Task T2.4, Ecosystems data fusion and monitoring, focused on the use of artificial intelligence, machine learning, advanced statistics, and data-fusion techniques to support climate-risk analysis, visual analytics, and data interaction. The deliverable documents the scientific and technical development of several advanced models, including the associated data exploration, processing, and exploitation steps. The work covers four main modelling applications: downscaling global and regional climate information to specific locations; local-scale modelling of custom climate-risk indices, illustrated through avalanche-risk assessment; modelling of the socio-economic impacts of climate change and associated susceptibility, illustrated for Norway; and a statistical analysis of the relationship between droughts and extremely hot days across Europe. Climate projections and scenarios were used to train and validate these models, with pattern-recognition and geometric deep-learning techniques supporting the identification of relationships between environmental, climatic, and anthropogenic factors. A central objective of the task was to demonstrate how publicly available datasets can be transformed into useful climate-resilience knowledge through advanced modelling. The resulting models are intended as representative examples of the potential of open data and scientific expertise to support communities, stakeholders, and decision-makers, rather than as an exhaustive set of climate-risk applications. The source code is made available through the IMPETUS GitHub repository and the RKB Resources Repository. At the time of the deliverable, access to some models was temporarily restricted while related scientific publications were being finalized, with the intention of releasing them publicly under open-data licenses once the embargo period ended.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.