Dynamics of ENSO Phase–Locking and Its Biases in Climate Models

The contributions of different oceanic feedbacks to the El Niño–Southern Oscillation (ENSO) phase‐locking are examined by deriving ENSO dynamics based on the recharge‐discharge framework. In observations, the significant winter preference of the ENSO peak is determined by a strong seasonal modulation of SST growth rate, which is controlled by the zonal advective and thermodynamic feedbacks. However, the majority of climate models fail to simulate ENSO phase‐locking because the contribution of zonal advective feedback to the seasonal modulation of the SST growth rate is much smaller compared to observations. The weak annual cycle of the SST‐current coupling coefficient and small annual mean of the negative climatological zonal SST gradient are two factors contributing to the weak‐biased seasonality of zonal advective feedback. Further analysis shows that the Niño3.4 SSTA has better phase‐locking performance than Niño3 SSTA in the climate models due to the better simulation of zonal advection feedback in the Niño3.4 region. Plain Language Summary El Niño–Southern Oscillation events typically onset in the boreal spring, grow rapidly during boreal summer and autumn, and reach their peak in wintertime, known as the phase‐locking phenomenon. The understanding of the dynamics of ENSO phase‐locking and the causes of phase‐locking biases in climate models is still inconclusive. In this study, a recharge‐discharge oscillator model is used to investigate the features and mechanisms of ENSO phase‐locking in observations and climate models. The observed characteristics of phase‐locking are determined by the seasonality of ENSO‐related SST growth rate. However, in climate models, the unrealistic ENSO phase‐locking is linked to the bias in the tropical climate basic state and its annual cycle. The contribution of different ocean processes to ENSO phase‐locking is further discussed. Key Points The El Niño–Southern Oscillation phase‐locking is determined by the seasonality of growth rate, which is controlled by the zonal advective and thermodynamic feedbacks The phase‐locking biases in models are attributed to the small contribution of zonal advective feedback to the seasonality of growth rate The errors in simulated zonal advective feedback are caused by the weak annual cycle of the SST‐current coupling and annual mean of climatological zonal SST gradient