Probabilistic & ensemble forecasting
Diffusion Models for High-Resolution Solar Forecasts
In one sentence. Score-based diffusion models super-resolve coarse NWP to satellite-resolution day-ahead solar irradiance, giving probabilistic forecasts with many samples.
Abstract
Forecasting future weather and climate is inherently difficult. Machine learning offers new approaches to increase the accuracy and computational efficiency of forecasts, but current methods are unable to accurately model uncertainty in high-dimensional predictions. Score-based diffusion models offer a new approach to modeling probability distributions over many dependent variables, and in this work, we demonstrate how they provide probabilistic forecasts of weather and climate variables at unprecedented resolution, speed, and accuracy. We apply the technique to day-ahead solar irradiance forecasts by generating many samples from a diffusion model trained to super-resolve coarse-resolution numerical weather predictions to high-resolution weather satellite observations.
Labels
- Topic
- Probabilistic & ensemble forecasting, Downscaling & super-resolution, Applications & impacts
- Type
- New model or method
- Method
- Diffusion / flow matching / score-based
- Time scale
- 6 h–3 days
- Domain
- Regional
- Data
- NWP analysis or forecasts (IFS, GFS, ...), Satellite
- Status
- Research
- Identifiers
- arXiv:2302.00170 OpenAlex:W4320513550
- Added
- 2026-10-22
Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.