A regime view of ENSO flavours through clustering in CMIP6 models

El Niño Southern Oscillation (ENSO) flavours in the tropical Pacific are studied from a regime perspective. Five recurring spatial patterns or regimes characterising the diversity of ENSO are established using a clustering approach applied to the HadISST sea surface temperature anomalies (SSTA). Two warm (eastern and central El Niño), two cold (basin wide and central La Niña) and a neutral reference regimes are found. Simulated SSTA by the models from the latest Coupled Model Intercomparison Project (CMIP6) are then matched to these reference regimes. This allows for a consistent assessment of the skill of the models in reproducing the reference regimes over the historical period and the change in these regimes under the high-warming Shared Socio-economic Pathway (SSP5.8.5) scenario. Results over the historical period show that models simulate well the reference regimes with some discrepancies. Models simulate overly strong and broad ENSO patterns and have issues in capturing the correct regime seasonality, persistence and transition between regimes. Some models also have difficulty simulating the frequency of regimes, the eastern El Niño regime in particular. In the future, eastern El Niño and central La Niña regimes are expected to be more frequent accompanied with a less frequent neutral regime. The central Pacific El Niño and La Niña regimes are projected to increase in amplitude and variability. Compared to previous studies, our approach gives a common characterisation across models and observations of the diversity of the warm and cold phases of ENSO at the same time established from observations.