Using Deep Learning to Identify Initial Error Sensitivity for Interpretable ENSO Forecasts

Abstract We introduce an interpretable-by-design method, optimized model analog, that integrates deep learning with model-analog forecasting which generates forecasts from similar initial climate states in a repository of model simulations. This hybrid framework employs a convolutional neural network to estimate state-dependent weights to identify initial analog states that lead to shadowing target trajectories. The advantage of our method lies in its inherent interpretability, offering insights into initial-error-sensitive regions through estimated weights and the ability to trace the physically based evolution of the system through analog forecasting. We evaluate our approach using the Community Earth System Model, version 2, Large Ensemble, to forecast El Niño–Southern Oscillation (ENSO) on a seasonal-to-annual time scale. Results demonstrate significant improvements over the unweighted model-analog technique and show comparable skill to a standalone neural network approach. Furthermore, our model exhibits improvements in boreal winter and spring initialization when evaluated against a reanalysis dataset. Our approach reveals state-dependent regional sensitivity linked to various seasonally varying physical processes, including the Pacific meridional modes, equatorial recharge oscillator, and stochastic wind forcing. Additionally, forecasts of El Niño and La Niña are sensitive to different initial states: El Niño forecasts are more sensitive to initial error in tropical Pacific sea surface temperature in boreal winter, while La Niña forecasts are more sensitive to initial error in tropical Pacific zonal wind stress in boreal summer. This approach has broad implications for forecasting diverse climate phenomena, including regional temperature and precipitation, which are challenging for the model-analog approach alone. Significance Statement This study demonstrates that combining deep learning and a simple analog forecasting method can yield skillful and interpretable El Niño–Southern Oscillation forecasts. A convolutional neural network is used to find critical areas for picking analog members. This is important because it is challenging to explain the decision-making processes of recent deep learning approaches. The developed approach can be applied to various climate predictions.