Identifying Extreme Regimes in Climate-Scale Digital Twins: a Roadmap

Abstract

Digital twins relevant for the climate sector are complex, coupled, multi-scale models, which often entail real-world data assimilated onto physical equations. Classifying the predictions of these models, and in particular identifying any rare or atypical outcome, is a formidable challenge which needs to be overcome if digital twins are to achieve their full predictive potential. Here I outline the technical details connected to this challenge and discuss a possible direction forward.