INfrastructure and TEchnology for Generative modeling and Readiness in Advanced Twin Environments

Schrift Standard Fett
Login

Forschungsstelle

SBFI

Projektnummer

25.00187

Projekttitel

INfrastructure and TEchnology for Generative modeling and Readiness in Advanced Twin Environments

Texte zu diesem Projekt

	Deutsch	Französisch	Italienisch	Englisch
Abstract	-	-	-

Erfasste Texte

Kategorie	Text
Abstract (Englisch)	MeteoSwiss proposes three complementary research work packages (WPs) in 2026 to bridge Swiss activities into Destination Earth (DestinE) Phase III. Each WP aligns with DestinE Phase II priorities and addresses core capabilities that are not within the scope of the GLORI-A project, which is primarily focused on developing an Alpine digital twin using regional high-resolution numerical weather prediction. The overarching aim is to advance machine learning (ML) and integration of DestinE Phase I and II products into weather and climate services, ensuring Switzerland’s seamless integration into DestinE Phase III in 2027. Specifically, the project will: (i) develop generative ML techniques (diffusion models) for physically consistent convective-scale ensemble weather prediction, (ii) integrate DestinE workflow infrastructure components (FDB, Polytope, Aviso) within MeteoSwiss’ operating systems, and (iii) validate the DestinE Climate Change Adaptation Digital Twin (Climate DT) output for Alpine region extremes against existing regional climate simulations and climate analyses. By leveraging unique Swiss datasets (ICON-CH1/CH2 ensemble predictions and reanalyses) and HPC capabilities, we target scientific excellence, innovation in ML, and direct relevance to DestinE’s vision.

Kategorie

Text

Abstract
(Englisch)

MeteoSwiss proposes three complementary research work packages (WPs) in 2026 to bridge Swiss activities into Destination Earth (DestinE) Phase III. Each WP aligns with DestinE Phase II priorities and addresses core capabilities that are not within the scope of the GLORI-A project, which is primarily focused on developing an Alpine digital twin using regional high-resolution numerical weather prediction. The overarching aim is to advance machine learning (ML) and integration of DestinE Phase I and II products into weather and climate services, ensuring Switzerland’s seamless integration into DestinE Phase III in 2027. Specifically, the project will: (i) develop generative ML techniques (diffusion models) for physically consistent convective-scale ensemble weather prediction, (ii) integrate DestinE workflow infrastructure components (FDB, Polytope, Aviso) within MeteoSwiss’ operating systems, and (iii) validate the DestinE Climate Change Adaptation Digital Twin (Climate DT) output for Alpine region extremes against existing regional climate simulations and climate analyses. By leveraging unique Swiss datasets (ICON-CH1/CH2 ensemble predictions and reanalyses) and HPC capabilities, we target scientific excellence, innovation in ML, and direct relevance to DestinE’s vision.