Adaptive Bias Correction for Improved Subseasonal Forecasting

• Activity Number: 394 - Spatial and Spatio-Temporal Modeling in Climate and Meteorology
• Type: Contributed
• Date/Time: Wednesday, August 10, 2022 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Soukayna Mouatadid and Paulo Orenstein and Genevieve Flaspohler* and Miruna Oprescu and Judah Cohen and Franklyn Wang and Sean Knight and Maria Geogdzhayeva and Sam Levang and Ernest Fraenkel and Lester Mackey
• Companies: University of Toronto and Instituto de Matemática Pura e Aplicada and Massachusetts Institute of Technology and Microsoft Research New England and Atmospheric and Environmental Research and Harvard University and Massachusetts Institute of Technology and Massachusetts Institute of Technology and Salient Predictions Inc. and Massachusetts Institute of Technology and Microsoft Research New England
• Keywords: subseasonal forecasting; adaptive bias correction; machine learning; numerical weather prediction

Abstract

Subseasonal forecasting—predicting temperature and precipitation 2-6 weeks ahead—is critical for effective water allocation, wildfire management, and drought and flood mitigation. However, accurate forecasts for the subseasonal regime are lacking due to the chaotic nature of weather and the complex dependence on both local weather variables and global climate variables. To address this need, we introduce an adaptive bias correction (ABC) method that combines state-of-the-art dynamical forecasts with observations using machine learning. We evaluate our adaptive bias correction method in the contiguous U.S. over the years 2011-2020 and demonstrate consistent improvement over standard meteorological baselines, state-of-the-art learning models, and the leading subseasonal dynamical models, as measured by root mean squared error and uncentered anomaly correlation skill. To facilitate future subseasonal benchmarking and development, we release our model code through the subseasonal_toolkit Python package and our routinely updated SubseasonalClimateUSA dataset through the subseasonal_data Python package.