Bridging the "Predictability Desert": A Probabilistic Bias Correction Framework for AI and Dynamical Subseasonal Forecasts

Hannah Guan, Soukayna Mouatadid, Paulo Orenstein, Judah Cohen, Zekun Ni, Haiyu Dong, Genevieve Elaine Flaspohler, Alex Xijie Lu, Jonathan A. Weyn, Lester Mackey

Abstract: Decision-makers rely on weather forecasts to allocate water resources, manage wildfires, plant crops, 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-based artificial intelligence (AI) models. However, model skill drops precipitously at subseasonal timescales (weeks 3 -- 6) due to compounding errors and persistent biases. To counter this degradation, we introduce \emph{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 model from the European Centre for Medium-Range Weather Forecasts (ECMWF), PBC boosts subseasonal forecasting skill by 70–80\% for precipitation and over 200\% for temperature and sea-level pressure. 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 of six government agencies, 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.

Submission Number: 68