aiqbench analysis
This research introduces Xaurora, a stochastic foundation model that transforms deterministic atmospheric priors into efficient generative ensembles for probabilistic weather forecasting. By leveraging Denoising Stochastic Interpolants and SDE rollouts, the method achieves state-of-the-art global metrics with significantly reduced computational overhead compared to training large-scale models from scratch. The work demonstrates a practical pathway to enhance uncertainty quantification and extreme event prediction using existing foundation model architectures.
Xaurora: Generative Weather Forecasting with Denoising Stochastic Interpolants from a Foundation Model Prior
arXiv:2610.06509v1 Announce Type: new Abstract: Deep learning has revolutionised weather forecasting in recent years, especially through atmospheric foundation models, which offer competitive skill for a fraction of the computational costs of classic physics-based models. However, most existing foundation models are deterministic, limiting the generation of large ensembles for accurate uncertainty quantification, extreme weather risk assessment, and long-range weather forecasting. Furthermore, these models incur a large, often prohibitive, computational overhead to train from scratch. To address these shortcomings, we turn a pretrained deterministic prior model, namely the Aurora foundation model, into a generative ensemble-prediction model. To that end, we introduce a novel generative method, Denoising Stochastic Interpolants, combined with a replay buffer for Stochastic Differential Equation (SDE) rollout, enabling probabilistic training of SDE trajectories. Our stochastic foundation model, Xaurora, is finetuned from the small Aurora version, yet it approaches the state-of-the-art on global ensemble metrics and is competitive with the large version of Aurora. Our method is parameter and sample efficient, and generates skilful 15-day forecasts in 13 minutes. Our results demonstrate that deterministic foundation models can be efficiently extended into even stronger stochastic models.
Abstract:Deep learning has revolutionised weather forecasting in recent years, especially through atmospheric foundation models, which offer competitive skill for a fraction of the computational costs of classic physics-based models. However, most existing foundation models are deterministic, limiting the generation of large ensembles for accurate uncertainty quantification, extreme weather risk assessment, and long-range weather forecasting. Furthermore, these models incur a large, often prohibitive, computational overhead to train from scratch. To address these shortcomings, we turn a pretrained deterministic prior model, namely the Aurora foundation model, into a generative ensemble-prediction model. To that end, we introduce a novel generative method, Denoising Stochastic Interpolants, combined with a replay buffer for Stochastic Differential Equation (SDE) rollout, enabling probabilistic training of SDE trajectories. Our stochastic foundation model, Xaurora, is finetuned from the small Aurora version, yet it approaches the state-of-the-art on global ensemble metrics and is competitive with the large version of Aurora. Our method is parameter and sample efficient, and generates skilful 15-day forecasts in 13 minutes. Our results demonstrate that deterministic foundation models can be efficiently extended into even stronger stochastic models.
Machine Learning (cs.LG)
arXiv:2610.06509 [cs.LG]
(or arXiv:2610.06509v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.06509 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 273 words
Related coverage
Proper Scoring Rule-based Diffusion for Probabilistic Weather Forecasting
arXiv · Oct 1, 2026
ABC: Any-Subset Autoregression via Non-Markovian Diffusion Bridges in Continuous Time and Space
arXiv · Sep 1, 2026
Thirty minutes to better weather insights with Earth2Studio
Quantum Zeitgeist · Sep 22, 2026
Turn Your Latest Observations Into Timely Weather Decisions With NVIDIA Earth-2
NVIDIA · Sep 21, 2026