AI-driven technique boosts accuracy of two-to-six-week weather forecasts
An international research team unveiled an AI post-processing method that improves weather forecasts, especially the notoriously hard 2- to 6-week period.
The scientists detailed a machine-learning system in Nature that refines existing forecasts across all timescales. By intelligently correcting model output, it outperforms traditional approaches for both short-term and extended forecasts. The biggest gain is seen in the 2- to 6-week window, previously called the predictability desert, where climatological averages were the best guess.
Why it matters
More reliable 2- to 6-week forecasts help agriculture, disaster preparedness and energy planning.
In this story
AIweather forecastingpredictability desertmedium-range forecastsmachine learningNature studymeteorologyclimatological averages
