Stability Analysis Of Two Air-sea Coupled Modes In The Tropical Pacific And Their Key Roles In Spatiotemporal Diversity Of ENSO

El Ni(?)o–Southern Oscillation(ENSO)events exhibit a diversity of amplitudes,spatial patterns,and life cycles,with the main ENSO periods concentrated in the 3–7-yr [lowfrequency(LF)] and 2–3-yr [quasi-biennial(QB)] bands.In this study,the spatiotemporal diversity of ENSO is quantitatively examined by extracting the two ENSO modes,namely,the LF and QB components of ENSO,from the traditional Ni(?)o-3.4 index and connecting them with the spatial types of ENSO.El Ni(?)o events can be regrouped as the QB-dominated centralPacific ENSO-like(QB-CP),LF-dominated eastern-Pacific ENSO-like(LF-EP),and LFdominated mixing(LF-mixing)types.La Ni(?)a events with vague spatial patterns can also have the same categorization.The QB-CP and LF-EP El Ni(?)o types both have a high-amplitude QB component.Meanwhile,the former is less affected by its powerless LF component,but the latter is controlled by its strong LF component.Ocean dynamics of the two El Ni(?)o types are distinct from each other.The thermocline feedback dominates the growth of the two El Ni(?)o types and contributes to the phase transition of the LF-EP type,while the zonal advective feedback is of increasing importance in the QB-CP El Ni(?)o and mainly contributes to the phase transitions of the two El Ni(?)o types.Additionally,the LF-mixing type with ambiguous spatial features and complex life cycles is distinguished from the other two types.These results indicate that the two ENSO modes coexist in the tropical Pacific air–sea system,and their combination with changing amplitude is the key to explaining the spatiotemporal diversity of ENSO.The Zebiak-Cane model can reproduce the QB-CP and LF-EP El Ni(?)o,which means that the tropical Pacific can generate the spatiotemporal diversity of ENSO without external forcings.The majority of CMIP5 and CMIP6 models can simulate the ENSO-related QB and LF variability but lack the ability to similar the two spatial types of ENSO.Among CMIP models,MRI-ESM2-0 is the only model that can effectively capture the relationship between the ENSO interannual timescales and spatial patterns and can reproduce the QB-CP and LF-EP El Ni(?)o.We propose a new method to empirically diagnose linear growth rates of the EP and CP ENSO types and measure the influence of different dynamic processes on the diversity of ENSO.We spatially project the peak-time symmetric component of the mixed-layer oceanic heat budget terms onto the ENSO-related SSTA patterns to estimate the ENSO grow rate.After validating the method by comparing it with the Bjerknes(BJ)index,we show that the growth rate is positive for the CP ENSO type and weakly negative for the EP ENSO type in the past 40 years.The CMIP5-based multi-model ensemble mean shows generally similar characteristics as the reanalysis products for the EP ENSO type rather than CP ENSO type.