Xaurora: Generative Weather Forecasting with Denoising Stochastic Interpolants from a Foundation Model Prior

2026-10-05

AI Summary

This study addresses the challenges of uncertainty quantification and high training costs inherent in deterministic meteorological foundation models by proposing a framework that transforms the pretrained Aurora model into a generative ensemble prediction system. The core methodology introduces Denoising Stochastic Interpolation (DSI), integrated with stochastic differential equation (SDE) rollouts and a replay buffer mechanism, to enable parameter- and sample-efficient probabilistic trajectory training and model fine-tuning. Experimental results demonstrate that this approach achieves state-of-the-art performance on global ensemble forecasting metrics, rivaling the large-scale Aurora model. Notably, it completes 15-day forecasts in merely 13 minutes, substantially reducing computational overhead and establishing an efficient new paradigm for uncertainty quantification in meteorology.

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.

Problem

• weather forecasting
• foundation models
• uncertainty quantification
• ensemble prediction
• computational overhead

Innovation

• Generative Weather Forecasting
• Denoising Stochastic Interpolants
• Foundation Model Prior
• Stochastic Differential Equation
• Ensemble Prediction