Uncertainty reduction in ENSO periodicity projection based on the Wyrtki index

Using the Wyrtki index, which comprehensively characterizes the dynamics of ENSO, we evaluated the fitting performance of 24 CMIP6 models for the ENSO dynamic processes during the period from 1980 to 2014. We identified the high-skill (HS) models with better simulation capabilities and the low-skill (LS) models with poorer simulation abilities. Compared to observational and reanalysis data, the HS models better simulate the average state of the tropical Pacific and the associated dynamic processes of ENSO from 1980 to 2014. In contrast, the LS models show a colder equatorial cold tongue, a steeper thermocline slope, and stronger trade winds in the central and western Pacific. The zonal advection feedback in the LS models is weaker, while the thermocline feedback is stronger, which may contribute to the deviations observed in the LS models when simulating the historical ENSO periodicity. The HS models indicate a trend of shorter conventional ENSO periodicity for the period from 2066 to 2100, which contrasts sharply with the conclusion drawn from the 24 CMIP6 models that show no significant change in ENSO periodicity duration. Additionally, the period of CP ENSO is projected to become shorter under SSP585 scenario for the HS models and all models. In contrast, the CP ENSO period change is insignificant for the LS models.