El Niño reshapes the planet's climate every two to seven years, but what triggers it remains a puzzle. Classic low-order models, such as the recharge oscillator, treat the influence of neighboring regions as external noise. This work sits alongside studies where gravity weaves a quantum lace or two pendulums catch the invisible, but here the subject is different: not fundamental fields, but the climate system. The authors replace noise with deterministic coupling between two regions and watch how their interaction births or dampens oscillations.
They took a dimensionless delay differential equation for the temperature anomaly T in each of two regions: dT/dt = T - $T^3$ - α T(t-δ) + γ $T_{othe}$r. The parameter α sets the strength of self-feedback with delay δ, and γ is the coupling between regions. They solved the system numerically for different α, δ, and γ, building time series and phase portraits. To assess basins of attraction, they swept a grid of initial conditions and noted which attractor the solution reached. Stability was checked by adding Gaussian noise of varying strength.
The system shows a rich set of states, from complete cessation of oscillations (amplitude death and oscillation death) to synchronous and irregular oscillations. With weak coupling and large delay, ENSO-like oscillations emerge. The big surprise: if one region alone decays to equilibrium and the other oscillates, the coupling makes both oscillate. Basins of attraction showed that strong coupling widens the regions where fixed points and oscillations coexist. Noise revealed that of four fixed points, only two with large values are noise-resistant. For the real Niño 3 and Niño 4 regions, the model reproduces the 0.5 °C threshold but suggests replacing it with an interval: El Niño progresses if the anomaly in the first region is from 0.5 to 1.6 °C, and in the second from -0.06 to 2 °C.
Unlike neighboring works where regional coupling is modeled as noise or ignored, here two deterministic oscillators interact directly. This allows us to see how oscillations transfer from one region to another and how synchronization arises. The work stands apart: the closest studies concern quantum gravity or neural network training, so direct comparison is difficult. But within its niche, the model offers a new tool: basin-of-attraction maps can be used to estimate El Niño probability from initial anomalies.
The model can be extended to more regions and include the seasonal cycle to get closer to real ENSO dynamics. It would be interesting to test how noise and parameter inhomogeneity affect switching between states. Perhaps statistical predictions could be derived from the geometry of the basins.
This work could influence El Niño monitoring and forecasting in meteorological agencies by proposing an interval criterion instead of a point threshold. It also provides an example of how coupled oscillators with delay generate complex dynamics, which is useful in neurobiology and laser physics.
Test the model on longer SST observation records and compare basin predictions with real El Niño and La Niña episodes. Investigate how basins change with the addition of seasonal forcing and spatial inhomogeneity.
For climate dynamics, an open question remains: why ENSO is irregular and how to predict its amplitude. This work links irregularity to the coexistence of attractors and sensitivity to initial conditions, bringing us closer to understanding internal climate variability.
Kelvin and Rossby waves, which carry heat across the Pacific, travel at different speeds: the fast Kelvin wave crosses 13,350 km in 110 days, while the slow Rossby wave takes 329 days. This difference creates the delay needed for oscillations.
In Kim Stanley Robinson's novel 'Forty Signs of Rain', climate models predict abrupt changes in ocean currents. Here a similar idea appears: small changes in initial conditions can switch the system between states, as in scenarios of climate surprises.