What the study found
The study found that post-processed wind-speed ensemble forecasts generally performed better than raw ensemble forecasts. It also found that spatial resolution mattered more than ensemble size, and that adding high-resolution members to low-resolution forecasts could improve skill, especially when more high-resolution members were included.
Why the authors say this matters
The authors suggest this matters because forecast skill depends on both resolution and ensemble composition, and because post-processing can reduce differences among forecast configurations. The study indicates that, in some cases, adding members does not necessarily improve skill, so the balance between resolution and ensemble size is important.
What the researchers tested
The researchers compared raw and post-processed medium-range and extended-range wind-speed ensemble forecasts from the European Centre for Medium-Range Weather Forecasts at 9 km and 36 km horizontal resolutions. They used an ensemble model output statistic approach for calibration with three spatial training data selection techniques, and they examined a 150-member dual-resolution combination as well as mixtures made by adding 1, 2, 4, 8, 16, or 32 high-resolution members to a 50-member low-resolution forecast.
What worked and what didn't
In general, all post-processed forecasts outperformed the raw ensemble predictions in probabilistic calibration and point forecast accuracy. Post-processing also reduced differences among the various forecast configurations. The study reports that augmenting a sufficiently large high-resolution ensemble with low-resolution predictions did not necessarily improve forecast skill, while incorporating high-resolution members into low-resolution ensemble forecasts showed clear benefit, with the largest gains in configurations with the most high-resolution members.
What to keep in mind
The abstract does not describe limitations in detail beyond the specific forecast systems and configurations studied. The findings are limited to the wind-speed ensemble forecasts, resolutions, and post-processing methods described in the study.
Key points
- Post-processing improved probabilistic calibration and point forecast accuracy for the wind-speed ensembles.
- Forecast differences among configurations became smaller after calibration.
- Spatial resolution was reported to be more important than ensemble size.
- Adding low-resolution members to high-resolution forecasts did not necessarily improve skill.
- Adding high-resolution members to low-resolution forecasts produced clear gains, especially with more high-resolution members.
Disclosure
- Research title:
- Post-processed wind-speed ensembles outperformed raw forecasts
- Authors:
- Sándor Baran, M Lakatos
- Institutions:
- University of Debrecen, University of Debrecen
- Publication date:
- 2026-04-20
- DOI:
- 10.1002/qj.70201
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with scholarly research without reading academic papers — one filtered digest, every Friday.