NN4CAST: una aplicación de red neural de extremo a extremo para predicciones climáticas estacionales

Abstract

Seasonal climate prediction is critical for decision-making across sectors, but current dynamical models have limitations. Traditional statistical and recent data-driven approaches offer promise but face challenges. NN4CAST is a specialized Python library, which applies deep learning techniques to address these challenges by enabling nonlinear statistical analysis of teleconnections in climate data. This tool represents a significant advancement in seasonal forecasting, with potential applications in practical decision-making. Examples demonstrate its effectiveness in predicting key variables impacting the Iberian Peninsula.