Identifying key convection-sensitive oceanic regions to weaken the ENSO spring predictability barrier.

The El Niño-Southern Oscillation (ENSO) exhibits its weakest predictability during boreal spring, a phenomenon known as the Spring Predictability Barrier (SPB). The SPB arises from weak air-sea coupling that limits the growth and persistence of the ENSO signal. Improving springtime prediction therefore requires identifying oceanic regions most relevant for convection variability. Here, we introduce a Sea Surface Temperature Range Index (SRI), which quantifies the spatial extent of Sea surface temperatures favorable for convection. Using SRI, we show that regions exceeding 26 °C in the east-central Pacific and 28.5 °C in the eastern Atlantic during spring are critical for initiating persistent intense convection. The expansion of these convection-sensitive areas strengthens the Bjerknes feedback by modulating the Walker circulation, providing an effective predictor of ENSO evolution. We further develop a Long Short-Term Memory deep learning model incorporating SRI, which achieves higher predictive skill than the average of dynamical and statistical models, especially for multiyear La Niña events. These results underscore the central role of convection-sensitive oceanic regions in alleviating the SPB.