Diagnosing the representation and causes of the ENSO persistence barrier in CMIP5 simulations

Abstract

In this study, the persistence barrier (PB) of the El Nino–Southern Oscillation (ENSO) phenomenon is investigated using reanalysis data and historical simulations of the Coupled Model Intercomparison Project Phase 5 (CMIP5) models. Both the timing and intensity of the ENSO PB can be quantified using the maximum gradient of autocorrelation decline of Nino sea surface temperature (SST) anomaly indices. Most of the CMIP5 models were found to reasonably reproduce the observed timing of the ENSO PB that typically occurs during the boreal late spring to early summer, and underestimated the PB intensity compared to observations. Furthermore, the PB properties of the Eastern Pacific (EP) ENSO indices were much better represented by the models than those of the Central Pacific (CP) ENSO indices, probably because CP ENSO events are more challenging to simulate than their counterparts. Approximately half of the models can satisfyingly reflect the intensity and timing of PB for indices of EP ENSO and their distinctions from those of the CP ENSO, with a larger uncertainty for the modeled PB timing than intensity. Further diagnosis has revealed the relationship between the ENSO PB intensity and the factors associated with the tropical Pacific background state. The PB intensity exhibits a stronger relationship with the seasonality of the SST amplitude in CP, compared to the SST amplitude, and the strength of seasonal synchronization of EP SST anomalies is highly correlated with the PB intensity. These results suggest that the seasonality of tropical SST variability may fundamentally contribute to the ENSO PB.