Skip to main content
Aller à la page d’accueil de la Commission européenne (s’ouvre dans une nouvelle fenêtre)
français fr
CORDIS - Résultats de la recherche de l’UE
CORDIS

Reaching AI Projections Trustworthy for Unseen Rainfall Extremes

Objectif

RAPTURE addresses a critical and timely challenge: how can we simulate extreme rainfall on the kilometre-scale under conditions we have never seen, in a trustworthy way? Overcoming this challenge is vital for understanding the dynamics and impacts of localised extreme rainfall (LXR) in a future, warmer world. Two approaches now offer practical paths forward: regional models that can resolve convective rainfall and computationally-efficient machine-learning rainfall emulators.

Models we can trust for studying future LXR must realistically reproduce current LXR diversity across space and time, and its relationship to multi-scale dynamical processes. Yet this diversity is difficult to verify and how far current approaches meet this realism criteria is still poorly understood. Trustworthy emulators must also be able to generalise to weather conditions they have not been trained on, but when emulators can and cannot do this is still unknown. These measures of trustworthiness require domain-specific expertise to evaluate and (for emulators) computer science expertise to improve: a disconnect that has not yet been bridged.

RAPTURE will overcome these barriers to progress. We use an innovative weather systems approach to achieve unprecedented insight into European LXR in observations, regional simulations and emulators, then use that insight to improve models and create accessible diagnostics. We will ask and answer:
1) How realistic is LXR in physical models and emulators?
2) When can emulators generalise, when can’t they, and why?
3) Can we make emulators more trustworthy, and make this goal accessible and assessable for machine-learning researchers?

These answers will unlock the potential of cheap, trustworthy LXR emulation: local rainfall that can be simulated for centuries, opening new cross-timescale research frontiers. Our approach will extend directly to other regions and climate hazards, and our findings will have cascading impacts in core-AI and applied science

Champ scientifique (EuroSciVoc)

Ce projet n'a pas encore été classé par EuroSciVoc.
Proposez les domaines scientifiques qui vous semblent les plus pertinents et aidez-nous à améliorer notre service de classification.

Mots‑clés

Programme(s)

Thème(s)

Régime de financement

HORIZON-ERC - HORIZON ERC Grants

Voir tous les projets financés dans le cadre de ce programme de financement

Appel à propositions

(s’ouvre dans une nouvelle fenêtre) ERC-2026-STG

Voir tous les projets financés au titre de cet appel

Institution d’accueil

UNIVERSITETET I BERGEN
Contribution nette de l'UE
€ 1 500 000,00
Coût total
€ 1 500 000,00

Bénéficiaires (1)