Abstract
Understanding and extending the predictability of El Niño‒Southern Oscillation (ENSO) has been an important research topic because of ENSO’s large influence on global weather and climate. Here, we develop an empirical model of tropical atmosphere-ocean interactions that has high ENSO prediction skill, comparable to the skills of well performing dynamical models. The model is used to investigate the effects of the main atmosphere-ocean interaction processes—thermocline and zonal wind feedbacks and zonal wind forcing—on its ENSO predictability. We find that all these processes significantly affect ENSO predictability and extend the predictability limit by up to four months, with the largest effect coming from the thermocline feedback followed by the total zonal wind forcing. The other processes with progressively smaller effects are the external zonal wind forcing and zonal wind feedback. The two most influential processes, however, affect ENSO predictability in the VAR model differently. The thermocline feedback improves the forecast skill by predominantly maintaining the correct phase, whereas the total zonal wind forcing improves the skill by maintaining the correct amplitude of the forecast ENSO events. This result suggests that the dynamical seasonal prediction models must have good representations of the major ENSO processes to make skilful ENSO predictions.
Highlights
The SST, Taux and Z20 data used for model fitting and verification were obtained from the HadISST32, ERA-40/ERA-Interim[33,34] and the Australian Bureau of Meteorology’s POAMA Ensemble Ocean Data Assimilation System (PEODAS)[35], respectively
The anomalies were divided by their respective domain standard deviations before being subjected to separate empirical orthogonal function (EOF) analysis over the tropical Pacific domain (130E-80W, 20S-20N)
The corresponding principal component (PC) time series were obtained by projecting the SST, Taux and Z20 anomalies for the entire period (1960–2017) onto the respective EOF sets
Summary
These results, obtained from an observation-based empirical model, may be useful for understanding the causes of variations in ENSO forecast skill in dynamical seasonal prediction models[2]. The memory associated with ENSO evolutions is determined by the mean state and various forcings and feedbacks simulated by the models. The ENSO forecast skill of an APEC Climate Center seasonal forecast model was found to be adversely affected by the mean state bias caused by thermocline feedback errors at long lead times[31].
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
