Abstract Ocean initialization (OI) plays a crucial role in El Niño–Southern Oscillation (ENSO) predictions. This study investigates both the importance of OI accuracy and the detrimental effect of initialization shock caused by discretization differences between the OI generator and the forecast model, a developmental version of NOAA’s next‐generation operational Seasonal Forecast System (SFS). Thirty years of hindcasts for May and November initializations are analyzed. Three ocean reanalysis products, ORAS5, ORAS5‐Replay (hereafter “Replay”), and GLORe, are tested for initialization. Although GLORe is less accurate than ORAS5 and Replay in representing the ocean state, it is the most consistent with SFS and produces the highest ENSO forecast skill at longer lead times for May initialization. These results suggest that minimizing initialization shock is as important as achieving accurate initial conditions for improving seasonal ENSO forecasts.

Read original article