Improving subseasonal‐to‐seasonal prediction of the ENSO–East Asia teleconnection: A dynamical–statistical model approach

The El Niño–Southern Oscillation (ENSO) strongly influences the surface climate over East Asia during the boreal winter with a pronounced subseasonal variability, with the ENSO–East Asia teleconnection becoming strong in December, weak in January and strong again in February. This subseasonal variability, determined by atmospheric circulation anomalies over the Kuroshio Extension region, is largely explained by a linear combination of teleconnections driven by ENSO‐920‐related convection anomalies over the three tropical regions, i.e., the Indian Ocean, the Maritime Continent, and the central Pacific. However, the operational subseasonal‐to‐seasonal (S2S) forecasting system, GloSea6, does not accurately predict the subseasonal variability of the ENSO–East Asia teleconnection even at one‐month lead time. The forecast errors are mainly due to the failure of the model to predict teleconnection patterns induced by ENSO‐related tropical convection anomalies, not by convection anomalies themselves. These errors are reduced by adopting the hybrid dynamical–statistical model, which combines the dynamical prediction of tropical convection anomalies in the three key regions with the observed teleconnection patterns induced by these anomalies. The hybrid model significantly improves the prediction skill of the ENSO–East Asia teleconnection in December and February, outperforming the GloSea6 S2S forecasts of East Asian surface air temperature at one‐ and two‐month lead times. This result highlights the potential use of the hybrid model in operational S2S forecasts.