Tyche: One Step Flow for Efficient Probabilistic Weather Forecasting

TL;DR

Tyche is a novel one-step flow model that enables efficient probabilistic weather forecasting by directly mapping Gaussian noise to future states, reducing inference cost while maintaining high accuracy.

Contribution

It introduces a destination-aware average-velocity flow with a JVP-regularized objective and transformer parameterization for scalable, accurate, and reliable weather forecasts.

Findings

Tyche matches or exceeds state-of-the-art ensemble methods in skill and calibration.

Uses a single NFE to achieve high-quality probabilistic forecasts.

Outperforms operational ECMWF IFS ensemble in experiments.

Abstract

Probabilistic weather forecasting requires not only accurate trajectories, but calibrated distributions over plausible atmospheric futures. Recent data-driven systems have achieved remarkable deterministic skill, and diffusion-based ensemble forecasters have substantially improved sample realism and uncertainty quantification. However, their inference cost scales with forecast horizon, ensemble size, and the number of denoising steps required for each transition, making large operational ensembles expensive. To address this, we present Tyche, a one-step conditional flow model for efficient probabilistic weather forecasting. Tyche models the conditional forecast distribution with a destination-aware average-velocity flow that maps Gaussian noise directly to future weather states in a single function evaluation (1-NFE). To make this one-step transport learnable in high-dimensional geophysical fields, we derive a JVP-regularized rectification objective that enforces temporal self-consistency across source and destination flow timesteps without explicitly forming Jacobians. The transport field is parameterized by an isotropic Swin-style transformer that preserves fine-scale spatial structure while remaining scalable on global grids. To improve ensemble reliability under autoregressive forecasting, we further introduce a rollout-based finetuning stage with curriculum CRPS calibration supervision. Experiments on ERA5 at 1.5$^\circ$ and 6-hour resolution show that our Tyche, using merely a single NFE, matches or exceeds the forecast skill and calibration of state-of-the-art multi-step generative baselines and the operational ECMWF IFS ensemble.