The impact of the extratropics on ENSO diversity and predictability

Several mechanisms originating in the Northern (NH) and Southern Hemisphere (SH) are argued to have the ability to stochastically force ENSO events. In this study, the impact of these extratropical mechanisms on ENSO diversity and predictability are evaluated using linear regression methodologies from information theory and machine learning applied to observational data. Overfitting is often an issue when investigating different extratropical mechanisms, as they are highly correlated in both space and time. The statistical methods in this study are specifically designed to address this issue. Results show that at 1-year lead-times, the extratropics are related to development of Central Pacific (CP), but not Eastern Pacific (EP) ENSO events. In boreal winter, the SH extratropics contribute to the predictability of CP ENSO, much further in advance than previous studies have indicated. The dominant NH predictor of CP ENSO from one winter to the next is identified as a sea surface temperature dipole in the Western North Pacific. Finally, separation of CP ENSO into its extratropical and tropical related components demonstrates that CP ENSO events with strong forcing from the extratropics start one season earlier than events primarily forced from the Tropics and thus have the potential for longer lead predictability, up to 1-year in advance of a CP ENSO event.