A Transformer‐Based Deep Learning Model for Successful Predictions of the 2021 Second‐Year La Niña Condition

A purely data‐driven and transformer‐based model with a novel self‐attention mechanism (3D‐Geoformer) is used to make predictions by adopting a rolling predictive manner similar to that in dynamical coupled models. The 3D‐Geoformer yields a successful prediction of the 2021 second‐year cooling conditions that followed the 2020 La Niña event, including covarying anomalies of surface wind stress and three‐dimensional (3D) upper‐ocean temperature, the reoccurrence of negative subsurface temperature anomalies in the eastern equatorial Pacific and a corresponding turning point of sea surface temperature (SST) evolution in mid‐2021. The reasons for the successful prediction with interpretability are explored comprehensively by performing sensitivity experiments with modulating effects on SST due to wind and subsurface thermal forcings being separately considered in the input predictors for prediction. A comparison is also conducted with physics‐based modeling, illustrating the suitability and effectiveness of 3D‐Geoformer as a new platform for El Niño and Southern Oscillation studies. Plain Language Summary The tropical Pacific experienced the prolonged cooling conditions during 2020–2022 (often called a triple La Niña), which exerted great impacts on the weather and climate globally. However, physics‐derived coupled models still have difficulty in accurately making long‐lead real‐time predictions for sea surface temperature (SST) evolution in the tropical Pacific. With the rapid development of deep learning‐based modeling, purely data‐driven models provide an innovative way for SST predictions. Here, a transformer‐based deep learning model is used to evaluate its performance in predicting the evolution of SST in the tropical Pacific during 2020–2022 and explore process representations that are important for SST evolution during 2021, including subsurface thermal effect and surface wind forcing on SST, the crucial factors determining the second‐year prolonged La Niña conditions and turning point of SST evolution. A comparison is made between the completely differently constructed physics‐derived dynamical coupled model and the pure‐data driven deep learning model, showing they both can be used for predictions of SST evolution in the 2021 second‐year cooling conditions. This indicates that it is necessary to adequately represent the thermocline feedback in predictive models, either in dynamical coupled models or purely data‐driven models, so that El Niño and Sout