![[2602.11893] Universal Diffusion-Based Probabilistic Downscaling](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.11893] Universal Diffusion-Based Probabilistic Downscaling
Summary
The paper presents a universal diffusion-based framework for downscaling weather forecasts, enhancing low-resolution predictions into high-resolution probabilistic forecasts without model-specific adjustments.
Why It Matters
This research is significant as it addresses the need for improved accuracy in weather forecasting. By providing a model-agnostic solution, it enhances the reliability of predictions across various weather models, which is crucial for sectors dependent on accurate weather data, such as agriculture and disaster management.
Key Takeaways
- Introduces a universal framework for probabilistic downscaling of weather forecasts.
- Enhances low-resolution forecasts to high-resolution predictions without model-specific tuning.
- Demonstrates improved probabilistic skill in forecasts compared to raw deterministic outputs.
Computer Science > Machine Learning arXiv:2602.11893 (cs) [Submitted on 12 Feb 2026 (v1), last revised 19 Feb 2026 (this version, v2)] Title:Universal Diffusion-Based Probabilistic Downscaling Authors:Roberto Molinaro, Niall Siegenheim, Henry Martin, Mark Frey, Niels Poulsen, Philipp Seitz, Marvin Vincent Gabler View a PDF of the paper titled Universal Diffusion-Based Probabilistic Downscaling, by Roberto Molinaro and 6 other authors View PDF HTML (experimental) Abstract:We introduce a universal diffusion-based downscaling framework that lifts deterministic low-resolution weather forecasts into probabilistic high-resolution predictions without any model-specific fine-tuning. A single conditional diffusion model is trained on paired coarse-resolution inputs (~25 km resolution) and high-resolution regional reanalysis targets (~5 km resolution), and is applied in a fully zero-shot manner to deterministic forecasts from heterogeneous upstream weather models. Focusing on near-surface variables, we evaluate probabilistic forecasts against independent in situ station observations over lead times up to 90 h. Across a diverse set of AI-based and numerical weather prediction (NWP) systems, the ensemble mean of the downscaled forecasts consistently improves upon each model's own raw deterministic forecast, and substantially larger gains are observed in probabilistic skill as measured by CRPS. These results demonstrate that diffusion-based downscaling provides a scalable, model-agnost...
