Exploring present and future ENSO dynamics within the recharge oscillator framework

The dynamics of the El Niño–Southern Oscillation (ENSO) can be succinctly represented by the Recharge Oscillator (RO) framework. In this presentation, we systematically explore how ENSO properties depend on RO parameters and assess the framework’s ability to diagnose both present-day variability and externally forced changes.

We first show that parameter regimes producing self-sustained oscillations yield unrealistically low Niño3.4 kurtosis compared to observations, indicating that ENSO is more consistently represented as a stochastically forced, linearly stable system. Rather than relying on standard fitting approaches, we conduct a broad exploration of RO parameter space to identify configurations that simultaneously reproduce observed ENSO amplitude, seasonality, skewness, and lagged autocorrelation. The simplest and most realistic configuration is a strongly damped oscillator, with a decay timescale shorter than the dominant ENSO period, forced by multiplicative white noise and modulated by weak deterministic nonlinearities.

These simulations generate interdecadal ENSO fluctuations comparable in magnitude to those observed, raising questions about the interpretability of slowly evolving RO parameters inferred from single realizations. Using idealized twin experiments, we show that fitting the RO to individual time series produces spurious interdecadal parameter shifts that appear to “explain” ENSO variability through changes in the Bjerknes–Wyrtki–Jin index, but do not reflect a forced response.

We then impose 20–40% linear trends in selected RO parameters over 200 years and test their recoverability using ensemble fits. About 50 ensemble members are sufficient to robustly detect linear parameter changes and most ENSO property trends over 40-year windows, while detecting changes in skewness and nonlinear parameters associated with extreme ENSO events requires 100 members or more. These idealized experiments demonstrate that ensemble simulations are essential for diagnosing externally forced changes in ENSO dynamics and provide a proof of concept for applying the RO framework to large-ensemble climate model experiments.