Distribution‐Guided Ensemble Postprocessing for S2S Precipitation Forecasts: A Seamless Pathway Using Deep Generative Models

Wen Shi, Baoxiang Pan, J Huang, Tingfeng Dou, Jie Feng, Huihui Yuan, Anran Wang, Xin Xia, Lei Huang, Yong Luo

Journal of Geophysical Research Machine Learning and Computation · 06 May 2026 · China Meteorological Administration (CMA), Fudan University (FuXi), Institute of Atmospheric Physics, Chinese Academy of Sciences, Tsinghua University (NowcastNet)

Paper

In one sentence. GAN-based post-processing of ECMWF S2S hindcasts over Southern China extends skillful probabilistic precipitation forecasts to week 3, improving CRPS and Brier score.

Abstract

Abstract Atmosphere‐ocean‐land coupled forecasting systems, despite their comprehensiveness, face substantial challenges in the “predictability desert” at subseasonal to seasonal (S2S) timescales, particularly for precipitation—a variable crucial for socioeconomic activities yet of stunning spatiotemporal variance. Post‐processing methods developed for numerical weather prediction and climate projections are not directly applicable to S2S forecasting, as they cannot distinctively address initialization errors, chaos‐induced state uncertainty growth, and model systematic biases. Additionally, regression‐based deep learning corrections introduce smoothing artifacts and ensemble under‐dispersion, limiting their ability to capture key processes and extreme events. We propose an integrated framework using Generative Adversarial Networks (GANs) for ensemble post‐processing. The approach exploits the ability of deep generative models to represent high‐dimensional distributions, combining trajectory constraints from short‐term forecasts with distributional constraints from long‐term climatology. In a case study using ECMWF hindcasts over Southern China, our model calibrates ensemble forecasts while enhancing both ensemble size and spatial resolution. The post‐processed forecasts maintain deterministic skill (anomaly correlation coefficient) while showing improved probabilistic forecast metrics, such as Continuous Ranked Probability Score (CRPS) and Brier score, extending the skillful probabilistic forecast horizon to week 3. The predicted fields demonstrate improved spatial distribution matching and maintain linear covariance across variables. The framework demonstrates strong spread‐error correlations for effective advance error estimation, and helps disentangle forecast uncertainties into propagated dynamical and post‐processing components, each with distinct lead‐time dependencies. This unified framework demonstrates the potential to advance seamless forecast capabilities while addressing the growing demand for high‐resolution, physically consistent S2S products.

Labels

Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.