Discovery of a Phenomenological Dynamical Model for Predicting the El Ni\~no-Southern Oscillation

TL;DR

This paper introduces a phenomenological nonlinear dynamical model for ENSO, capturing its physical processes and variability, significantly improving prediction and addressing the Spring Predictability Barrier.

Contribution

A novel nonlinear dynamical model derived from observed ENSO data that incorporates physical processes and enhances predictability.

Findings

Model demonstrates high predictive skill.

Nonlinearity is key to overcoming the Spring Predictability Barrier.

Potential for modeling other geophysical systems.

Abstract

The skill of the statistical as well as physics-based coupled climate models in predicting the El Ni\~no-Southern Oscillation (ENSO) is limited by their inability to represent the observed ENSO nonlinearity. A promising alternative, namely a deterministic nonlinear dynamical model derived from an observed ENSO timeseries, however, has remained elusive. Here we discover such a phenomenological nonlinear dynamical model that embodies known physical processes responsible for the self-sustained quasi-oscillatory character of the ENSO and its observed spectrum of variability. High predictive potential of the model is demonstrated and the intrinsic nonlinearity of the ENSO is shown to be critical for overcoming the Spring Predictability Barrier to a large extent. The unique methodology presented here has the potential for constructing similar models for other geophysical systems.