Duration of El Niño and La Niña Events: Dynamics and Multiyear Predictability

Abstract:

El Niño and La Niña events are characterized by anomalous warming and cooling in the tropical Pacific and can impact global weather and climate through atmospheric teleconnections. El Niño and La Niña events typically develop in boreal spring-summer and peak in boreal winter. After the peak, about two-thirds of El Niño and half of La Niña events terminate in the following year, but other events persist for another 12 months or longer, prolonging and exacerbating their climate impacts. The predictability of these long-lasting events has remained largely unknown, and current operational forecasts of El Niño and La Niña events are limited to 12 months. This presentation will provide an overview of our recent research on understanding the mechanisms and predictability of the duration of El Niño and La Niña events. Analyses of observational data and the Community Earth System Model version 1 (CESM1) show that the duration of El Niño events is strongly affected by the timing of their onset, while the duration of La Niña events is largely influenced by the amplitude of the preceding warm event. These leading factors control the event duration through oceanic and atmospheric feedbacks in the tropical Pacific as well as the Indian and Atlantic Oceans. The potential predictability of event duration based on these factors is tested using idealized prediction experiments conducted with the CESM1. We further explore the real-world predictability of event duration by conducting multiyear ensemble CESM1 forecasts initialized with observed oceanic conditions during 1954-2015. The CESM1 can skillfully predict the duration of observed El Niño and La Niña events with lead times ranging from 6 to 25 months. The dynamical processes contributing to the predictability of event duration in these forecasts are consistent with observational analyses and idealized forecasts. The high predictability of event duration indicates the potential for extending the current operational forecasts for up to two years.