AI summaryⓘ
The authors present a new machine learning method called probabilistic bias correction (PBC) that helps improve weather forecasts 2 to 6 weeks ahead, a time frame where predictions are usually less accurate. PBC works by learning from past forecast errors to reduce systematic mistakes in both physics-based and AI weather models. Tested on leading models from the European Centre for Medium-Range Weather Forecasts, PBC notably doubled the skill of AI forecasts and improved dynamical model predictions for most weather variables. Their approach also won first place in a 2025 global forecasting competition, showing better performance than many international models. These improvements can help decision-makers prepare better for weather extremes and manage resources like water and energy more effectively.
probabilistic bias correctionmachine learningsubseasonal weather forecastingdynamical weather modelsAI forecasting systemsforecast skillEuropean Centre for Medium-Range Weather Forecastssystematic errorweather extremesforecast calibration
Authors
Hannah Guan, Soukayna Mouatadid, Paulo Orenstein, Judah Cohen, Haiyu Dong, Zekun Ni, Jeremy Berman, Genevieve Flaspohler, Alex Lu, Jakob Schloer, Joshua Talib, Jonathan A. Weyn, Lester Mackey
Abstract
Decision-makers rely on weather forecasts to plant crops, manage wildfires, allocate water and energy, and prepare for weather extremes. Today, such forecasts enjoy unprecedented accuracy out to two weeks thanks to steady advances in physics-based dynamical models and data-driven artificial intelligence (AI) models. However, model skill drops precipitously at subseasonal timescales (2 - 6 weeks ahead), due to compounding errors and persistent biases. To counter this degradation, we introduce probabilistic bias correction (PBC), a machine learning framework that substantially reduces systematic error by learning to correct historical probabilistic forecasts. When applied to the leading dynamical and AI models from the European Centre for Medium-Range Weather Forecasts (ECMWF), PBC doubles the subseasonal skill of the AI Forecasting System and improves the skill of the operationally-debiased dynamical model for 91% of pressure, 92% of temperature, and 98% of precipitation targets. We designed PBC for operational deployment, and, in ECMWF's 2025 real-time forecasting competition, its global forecasts placed first for all weather variables and lead times, outperforming the dynamical models from six operational forecasting centers, an international dynamical multi-model ensemble, ECMWF's AI Forecasting System, and the forecasting systems of 34 teams worldwide. These probabilistic skill gains translate into more accurate prediction of extreme events and have the potential to improve agricultural planning, energy management, and disaster preparedness in vulnerable communities.