Towards Long-Range ENSO Prediction with an Explainable Deep Learning Model

TL;DR

This work tackles long-range ENSO prediction by introducing CTEFNet, a CNN–transformer hybrid that fuses nine ocean–atmosphere predictors to extend forecast lead times to $20$ months while mitigating the SPB. The model delivers superior predictive skill over dynamical ensembles and contemporary DL approaches, aided by training on CMIP6 SSP370 data and validated with reanalyses. A novel gradient-based sensitivity analysis provides physically meaningful precursors and reveals inter-basin teleconnections, aligning with established recharge and Bjerknes feedback mechanisms. The combination of high predictive performance and interpretable sensitivity analyses demonstrates the potential of multivariate, data-driven climate models to advance long-range ENSO forecasting and climate decision support.

Abstract

El Niño-Southern Oscillation (ENSO) is a prominent mode of interannual climate variability with far-reaching global impacts. Its evolution is governed by intricate air-sea interactions, posing significant challenges for long-term prediction. In this study, we introduce CTEFNet, a multivariate deep learning model that synergizes convolutional neural networks and transformers to enhance ENSO forecasting. By integrating multiple oceanic and atmospheric predictors, CTEFNet extends the effective forecast lead time to 20 months while mitigating the impact of the spring predictability barrier, outperforming both dynamical models and state-of-the-art deep learning approaches. Furthermore, CTEFNet offers physically meaningful and statistically significant insights through gradient-based sensitivity analysis, revealing the key precursor signals that govern ENSO dynamics, which align with well-established theories and reveal new insights about inter-basin interactions among the Pacific, Atlantic, and Indian Oceans. The CTEFNet's superior predictive skill and interpretable sensitivity assessments underscore its potential for advancing climate prediction. Our findings highlight the importance of multivariate coupling in ENSO evolution and demonstrate the promise of deep learning in capturing complex climate dynamics with enhanced interpretability.

Figures (4)

• Figure 1: ENSO correlation skill in CTEFNet and other models. The all-season correlation skill of the three-month-moving-averaged Niño 3.4 index as a function of the forecast lead month in CTEFNet (solid orange), CNN(solid deep blue), Geoformer (solid purple), ResCNN (solid green), ResoNet (solid brown), STPNet (solid pink) and the dynamical forecast systems included in the NMME project (dash with other colors). The validation period is between 1980 and 2021.
• Figure 2: The seasonality and lead-time of CTEFNet's performance.(A) Contour plot of correlation skills for CTEFNet across calendar months during the test period (1980–2021) at different lead times. The horizontal axis denotes the forecast lead month, while the vertical axis represents the calendar month. (B) Same as A, but showing the correlation skill difference between CTEFNet and CNN. (C) Same as (A), but showing the correlation skill difference between CTEFNet and Geoformer.
• Figure 3: The precursors and underlying mechanisms of ENSO forecasting revealed by CTEFNet. (A) Sensitivity analysis periods, with red indicating predicted target periods and gray representing 12-month input periods. (B to E) Averaged sensitivities across multiple El Niño events, retaining only grid point values that are statistically significant at the 95% confidence level, illustrating the contributions of various predictors across different months. Colors denote the sensitivities of SST, HC, SLP, and MLD, while vectors represent those of UO, VO, TAUU, and TAUV. The green box marks the Niño 3.4 region.
• Figure 4: Architecture of CTEFNet for ENSO predictions. CTEFNet consists of an input layer, a CNN-based feature extractor, a Transformer spatiotemporal analysis module, two fully connected layers, and an output layer. The input predictors include SST, HC, MLD, SSS, SLP, UO, VO, TAUU, and TAUV anomaly fields, all spanning 12 consecutive months in the region defined by ($60^{\circ}S-60^{\circ}N$, $0^\circ-360^\circ$E). The Niño 3.4 index for the subsequent 24 months serves as the predictands for supervised training.