PROSWIN: Probabilistic Solar Wind Speed Forecasting Using Deep Distributional Regression From Solar Images
8 affiliations
- 1Space Physics and Space Weather GFZ Helmholtz Centre for Geosciences Potsdam Germany
- 2Department of Electrical Engineering and Computer Science Technical University of Berlin Berlin Germany
- 3Institute of Physics and Astronomy University of Potsdam Potsdam Germany
- 4Department of Earth, Planetary, and Space Sciences University of California Los Angeles Los Angeles CA USA
- 5German Aerospace Center Munich Germany
- 6Columbia Astrophysics Laboratory Columbia University New York NY USA
- 7Scientific Computing Center Karlsruhe Institute of Technology Karlsruhe Germany
- 8Einstein Center Digital Future Berlin Germany
Abstract
Accurately predicting fast solar wind conditions is challenging, as uncertainties are large and unquantified by traditional single‐value prediction models. In particular, the risks of high‐speed solar wind streams (HSSs), which can cause damage to technological infrastructure, cannot be reliably assessed without probabilistic forecasts. We present PROSWIN, a probabilistic machine learning model that forecasts the hourly solar wind speed (SWS) at Earth with a 4‐day lead time. The approach combines solar images and magnetograms using a deep neural network coupled to a distributional regression algorithm. Because standard error metrics underweight the relevance of HSS peaks, we further introduce the prediction score (PS) , a model‐selection metric that jointly rewards timeline and HSS peak accuracy. On 14 years of data, our forecast achieves very well‐calibrated uncertainties (<1% average absolute coverage deviation). Using the continuous ranked probability score (CRPS), a metric that assesses distributional accuracy, we obtain a timeline CRPS of 41.0 km/s, an HSS peak CRPS of 45.3 km/s, and a PS of 42.3 km/s. We find that the 171 Å channel is an important complement to the typically used 193 and 211 Å channels and that the PS for model selection improves the applicability of the model. Compared to selected models from the literature, ours is the only one that is accurate for both timeline and HSS peak values, rather than trading one off against the other. These results support the advantages of probabilistic over single‐value solar wind models. The introduced methods are also transferable to other forecasting problems.
Versions
Related
- Graph-based probabilistic earthquake clustering and forecasting in Sumatra
- Advancing Near-Term Water Quality Forecasting with Explainable Machine Learning and Probabilistic Uncertainty Quantification
- SDECast: Probabilistic Weather Forecasting in Continuous Time with Neural SDEs
- Njord: A Probabilistic Graph Neural Network for Ensemble Ocean Forecasting
- Predictability-Guided Multiscale Probabilistic Forecasting of Wind Direction under Extreme Shear
BibTeX
@article{collin2026proswin,
title = {{PROSWIN: Probabilistic Solar Wind Speed Forecasting Using Deep Distributional Regression From Solar Images}},
author = {Daniel Collin and Yuri Shprits and Luca Chiarabini and Stefan J. Hofmeister and Nadja Klein and Guillermo Gallego},
journal = {AGU Advances},
year = {2026},
month = oct,
doi = {10.1029/2026av002524},
url = {https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2026AV002524},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - PROSWIN: Probabilistic Solar Wind Speed Forecasting Using Deep Distributional Regression From Solar Images
AU - Daniel Collin
AU - Yuri Shprits
AU - Luca Chiarabini
AU - Stefan J. Hofmeister
AU - Nadja Klein
AU - Guillermo Gallego
PY - 2026
DA - 2026/10/30
JO - AGU Advances
DO - 10.1029/2026av002524
UR - https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2026AV002524
AB - Accurately predicting fast solar wind conditions is challenging, as uncertainties are large and unquantified by traditional single‐value prediction models. In particular, the risks of high‐speed solar wind streams (HSSs), which can cause damage to technological infrastructure, cannot be reliably assessed without probabilistic forecasts. We present PROSWIN, a probabilistic machine learning model that forecasts the hourly solar wind speed (SWS) at Earth with a 4‐day lead time. The approach combines solar images and magnetograms using a deep neural network coupled to a distributional regression algorithm. Because standard error metrics underweight the relevance of HSS peaks, we further introduce the prediction score (PS) , a model‐selection metric that jointly rewards timeline and HSS peak accuracy. On 14 years of data, our forecast achieves very well‐calibrated uncertainties (<1% average absolute coverage deviation). Using the continuous ranked probability score (CRPS), a metric that assesses distributional accuracy, we obtain a timeline CRPS of 41.0 km/s, an HSS peak CRPS of 45.3 km/s, and a PS of 42.3 km/s. We find that the 171 Å channel is an important complement to the typically used 193 and 211 Å channels and that the PS for model selection improves the applicability of the model. Compared to selected models from the literature, ours is the only one that is accurate for both timeline and HSS peak values, rather than trading one off against the other. These results support the advantages of probabilistic over single‐value solar wind models. The introduced methods are also transferable to other forecasting problems.
SN - 2576-604X
SN - 2576-604X
ER -