Swift: An Autoregressive Consistency Model for Efficient Weather Forecasting

2025-09-29

AI Summary

Diffusion models for subseasonal-to-seasonal (S2S) probabilistic weather forecasting suffer from slow inference and reliance on multi-step iterative sampling, hindering operational deployment and long-horizon prediction. To address this, we propose a single-step autoregressive consistency model that—uniquely—optimizes the continuous ranked probability score (CRPS) as the objective function, enabling direct end-to-end fine-tuning of the probability flow ODE without ensemble averaging or parameter perturbation. Our approach eliminates iterative sampling while preserving physical consistency. Evaluated at 6-hour intervals, it delivers seamless 75-day probabilistic forecasts with inference speed 39× faster than state-of-the-art diffusion models, matching the skill of the ECMWF Integrated Forecasting System (IFS) ensemble. This work establishes a new paradigm for generative meteorological modeling: efficient, stable, and production-ready.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Probabilistic Circuits and Graphical Models

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Diffusion models offer a physically grounded framework for probabilistic weather forecasting, but their typical reliance on slow, iterative solvers during inference makes them impractical for subseasonal-to-seasonal (S2S) applications where long lead-times and domain-driven calibration are essential. To address this, we introduce Swift, a single-step consistency model that, for the first time, enables autoregressive finetuning of a probability flow model with a continuous ranked probability score (CRPS) objective. This eliminates the need for multi-model ensembling or parameter perturbations. Results show that Swift produces skillful 6-hourly forecasts that remain stable for up to 75 days, running $39 imes$ faster than state-of-the-art diffusion baselines while achieving forecast skill competitive with the numerical-based, operational IFS ENS. This marks a step toward efficient and reliable ensemble forecasting from medium-range to seasonal-scales.

Problem

• Efficient probabilistic weather forecasting using autoregressive consistency models
• Eliminating slow iterative inference in subseasonal-to-seasonal forecasting
• Achieving stable long-term forecasts faster than diffusion baselines

Innovation

• Single-step consistency model for weather forecasting
• Autoregressive finetuning with CRPS objective
• Eliminates need for multi-model ensembling