Asymmetries in the ENSO phase space

Abstract

El Niño Southern Oscillation (ENSO) dynamics are best described by the recharge oscillator model, in which the eastern tropical Pacific sea surface temperatures (T) and subsurface heat content (thermocline depth; h) have an out-of-phase relationship. This defines a 2-dimensional phase space diagram between T and h. In an idealized, stochastically forced damped oscillator, the mean phase space diagram should be a perfectly symmetrical circle with a clockwise propagation over time. However, the observed phase space shows strong asymmetries. In this study we illustrate how the ENSO phase space can be used to discuss the phase-dependency of ENSO dynamics. A normalized spherical coordinate system allows the definition of phase-depending ENSO growth rates and phase transition speeds. Based on these we discuss the implications of the observed asymmetries with regards to the dynamics and predictability of ENSO; with a particular focus on the variations in the growth rate and coupling of ENSO along the oscillation cycle. Using linear and non-linear recharge oscillator models we will show how dynamics and noise are driving ENSO at different phases of the ENSO cycles. The results illustrate that the ENSO cycle with positive phase transitions is present in all phases but has strong variations in its strength. Much of these variations result from presenting the ENSO phase space with estimates of h based on the iso-thermal depth, that is not ideal as it is not out-of-phase with T. Future work should address how h can be estimated better, including aspects such as the vertical temperature gradients and the meridional or zonal range. We further illustrated that a non-linear growth rate of T can explain most of the observed non-linear phase space characteristics.

Improving and Harmonizing El Niño Recharge Indices

“…And Z20 and SLA‐based indices become even closer (Figure S6 in Supporting Information ). Trajectories in the ( T E , h ind ) phase diagram (Dommenget & Al‐Ansari, 2022; Kessler, 2002) also become closer (Figure 3; shown here for longer ORAS5 SLA data set to have a larger density; see Figures S7 and S8 in Supporting Information for satellite SLA and ORAS5 Z20). The transform is geometrically simply a linear transform of coordinates from the ( T E , h ) to the ( T E , h ind ) orthogonal coordinate space.…”

Diagnosing the Linear Periodicity Dynamics of ENSO for Its Two Spatiotemporal Modes

“…Note that the F 1 F 2 factor is on the right‐hand side of Equation . Referring to the recharge oscillator framework, h theoretically has an almost 90° phase difference with T E , although deviations exist in observations (Dommenget & Al‐Ansari, 2023; Meinen & McPhaden, 2000; Timmermann et al., 2018). To simplify Equation , the h ‐related term on the right‐hand side and the T E first‐order derivative term on the left‐hand side can be negligible by operating an inner product defined by the temporal covariance with T E on both sides of Equation .…”

The dynamics of the El Niño–Southern Oscillation diversity in the recharge oscillator framework

“…We integrate each of the three models with random white noise forcings for 10 4 years to estimate the statistics of T n and h n for each model. We further define an idealised ReOsc model with symmetrical parameters following Dommenget and Al-Ansari [2022]:…”