SDECast: Neural SDEs Bring Continuous-Time Probabilistic Weather Forecasts Out to 5 Days
canaesseth · x · 2026-10-05
A new arXiv paper, SDECast, by Marchenko, Andrae, Lindsten and Naesseth moves past the pseudo-time framing of diffusion models and flow matching to do probabilistic weather forecasting in continuous physical time.
- Method: extends SDE Matching to learn stochastic atmospheric dynamics directly in physical space with Neural SDEs, avoiding error accumulation from autoregressive rollouts and requiring no repeated SDE simulation during training.
- Results: on a simulated geophysical flow it recovers meaningful drift dynamics and faithfully reproduces continuous-time behavior; it scales to global forecasting at hourly resolution with skillful probabilistic forecasts up to 5 days ahead.
- Accepted to the AI for Stochastic Dynamics and Sim2Science workshops at NeurIPS 2026.
More from Research
- Dream4ACT Unifies Video-Action Modeling Across Robot Embodiments, Hitting 89% on RoboTwin 2.0 — Xiangyu Zhu · 2026-10-05
- SMI Brings Understanding-Driven Spatial Memory Management to Long-Video World Models — Ying Yang · 2026-10-05
- LVMT Sets New SOTA in Long-Term Video Segmentation While Running 10X Faster — tue-mps · 2026-10-05
- Schmidhuber: Google's 2017 Transformer Builds on His 1991 Linear Attention Work — SchmidhuberAI · 2026-10-05
- Near-identical image scores, huge gaps: AI denoising must serve science, not looks — bravo_abad · 2026-10-05
- Dev shares how he trained a PII redaction model for European languages — auto_grad_ · 2026-10-05