Seasonal prediction and potential predictability of the northern tropical Atlantic SST anomalies in current coupled climate models

The northern tropical Atlantic (NTA) pattern, a key mode of interannual sea surface temperature (SST) variability in the tropical Atlantic, significantly influences climate variability within the region and globally. But gaps remain in understanding its seasonal predictability in current dynamical coupled climate models. This study examines forecasts and hindcasts from 18 operational seasonal forecast systems, revealing that there is pronounced seasonality in the prediction skill of the NTA pattern. Effective prediction extends to 4-5-month leads when initialized in boreal spring and summer, while it is limited to 2-3-month leads for boreal autumn and winter. Further assessment demonstrates that both the prediction skill and potential predictability tend to stabilize within an ensemble size of 5-9 for the individual models. The NTA pattern generally has higher potential predictability than actual forecast skill, with an exception that a few models exhibit a weak signal-to-noise paradox across varying initialization months and lead months. Moreover, the enhanced potential predictability coincides with a higher signal-to-noise ratio during boreal spring and early summer, suggesting a dominant role of external forcing in driving the NTA SST variability during the peak season. Consistently, investigation into the potential factors influencing prediction skill suggests joint influences of El Ni & ntilde;o-Southern Oscillation (ENSO) and the North Atlantic Oscillation (NAO) on the NTA pattern. In particular, among the current dynamical models, the NAO has a greater influence on the prediction uncertainty of the NTA pattern compared to ENSO. This work reveals the performance of current coupled climate models in predicting NTA SST anomalies and demonstrates the potential scope for improving the prediction skill.