Improving ENSO hindcast skill by adjusting the mean background state in a coupled general circulation model (CAS-ESM2.0)

ENSO (El Niño–Southern Oscillation) is a major mode of interannual climate variability in the global climate system, and improving its prediction is essential for understanding and responding to climate anomalies. This study proposes an iterative air–sea flux adjustment scheme and uses multiyear hindcast experiments with the CAS-ESM2.0 coupled model to systematically assess how model climatological state adjustment influences the seasonal prediction skill of ENSO. The results show that flux adjustment markedly improves the simulation of tropical Pacific sea surface temperature climatology and subsurface thermocline structure, thereby providing a more realistic background state for ENSO prediction. After adjustment, the prediction skills of the Niño3.4, Niño3, and Niño4 indices are all significantly improved, with larger gains at forecast ranges beyond four months, and the overall skill approaches the international multimodel ensemble mean. In addition, the adjusted forecasts produce more realistic El Niño peak intensity and seasonal phase locking, and the spring predictability barrier (SPB) is weakened. Further analysis indicates that the improvement in ENSO prediction skill mainly arises from three processes. First, flux adjustment reduces the initial shock during the early forecast stage. Second, it weakens the excessive positive feedback associated with thermocline bias. Third, it stabilizes thermodynamic damping dominated by latent heat and shortwave feedbacks. Together, these changes improve the magnitude and stability of the Bjerknes stability index and thus enhance the stability and skill of ENSO prediction.