Abstract. Seasonal climate predictions are essential for climate services, being the changes in tropical sea surface temperature (SST) the most influential drivers. SST anomalies can affect the climate in remote regions through various atmospheric teleconnection mechanisms, and the persistence/evolution of those SST anomalies can give seasonal predictability to atmospheric signals. Dynamical models often struggle with biases and low signal-to-noise ratios, making statistical methods a valuable alternative. Deep learning models are currently providing accurate predictions, mainly in short range weather forecast. Nevertheless, the blackbox nature of this methodology makes necessary the identification of its explainability. In this context, we present NN4CAST (Neural Network foreCAST), a versatile Python deep learning tool designed to assess seasonal predictability. Starting from the original files, NN4CAST performs all the methodological steps, enabling researchers to rapidly explore the predictability of a target variable and identify its main drivers. This flexible framework allows for the quick testing of predictive skill from different sources of predictability, making it a valuable asset for climate services. As SST is the primary source of seasonal predictability, we illustrate the application of NN4CAST to tropical and extratropical teleconnections forced by the Pacific SSTs. We show that NN4CAST can provide skillful seasonal forecasts in regions where the atmospheric response to SST anomalies is predominantly linear, such as the tropics, as well as in remote regions where the signal is highly non-linear, like Europe. Two key examples are the prediction of SST anomalies in the tropical Atlantic region during boreal spring and precipitation anomalies over the European continent during boreal autumn. The former exemplifies a predominantly linear tropical linear ENSO-Tropical North Atlantic, whereas the latter involves a highly non-linear and non-stationary ENSO-Euro-Atlantic teleconnection. Our results demonstrates NN4CAST’s potential to determine and quantify the influence of specific drivers on a target variable, offering a useful tool for improving climate predictability assessments. NN4CAST enables the attribution of predictions to specific input features, helping to identify the relative importance of different sources of predictability over time and space. In summary, NN4CAST offers a powerful framework to better characterize and understand the complex, non-linear, and non-stationary remote climate interactions.
Fraile et al. (Fri,) studied this question.