EAAC-S2S: East Asian Atmospheric Circulation S2S Forecasting with a Deep Learning Model Considering Multi-Spheres Coupling
-
Graphical Abstract
-
Abstract
Subseasonal-to-seasonal (S2S) forecasting for East Asian atmospheric circulation poses significant challenges for conventional numerical weather prediction (NWP) models. Recently, deep learning (DL) models have demonstrated significant potential in further enhancing S2S forecasts beyond the capabilities of NWP models. However, most current DL-based S2S forecasting models largely overlook the role of global predictors from multiple spheres, such as ocean, land, and atmosphere domains, that are crucial for effective S2S forecasting. In this study, we introduce EAAC-S2S, a tailored DL model for S2S forecasting of East Asian atmospheric circulation. EAAC-S2S employs cross-attention mechanism to couple atmospheric circulations over East Asian with representative multi-spheres (i.e., atmosphere, land, and ocean) variables, providing pentad-averaged circulation forecasts up to 12 pentads ahead throughout all seasons. Experimental results demonstrate, on S2S timescale, EAAC-S2S consistently outperforms the European Centre for Medium-Range Weather Forecasts (ECMWF) Ensemble Prediction System by decreasing Root-Mean-Square Error (RMSE) of 3.8% and increasing Anomaly Correlation Coefficient (ACC) of 8.6% averaged on all 17 predictands. Our system also has good skill for examples of heat waves and the South China Sea Subtropical High Intensity Index (SCSSHII). Moreover, quantitative interpretability analysis including multi-spheres attribution and attention visualization are conducted for the first time in a DL S2S model, where the traced predictability aligns well with prior meteorological knowledge. We hope that our work will accelerate research in data-driven S2S forecasting.
-
-