The Dance of Two Pendulums: Why the Monsoon Can Be Predicted 18 Months Ahead ✛ ✎

Original: "Resolving the Paradox of Changing El Niño-Monsoon Relation through Synchronization of Chaotic Oscillators"
arXiv:2603.16346 · 2026-03-17 · arXiv non-exclusive · independent analysis · 3 min
The Indian monsoon and deep tropical ocean waters resonate with an 18-month delay, and this connection does not weaken over time.

Two pendulums hang on the same wall, swinging out of sync, but suddenly they move as one, only with an 18-month delay. One pendulum is the Indian monsoon, the seasonal wind that brings rain to South Asian farmland. The other is the deep water of the tropical oceans, which slowly accumulates heat and remembers it for years.

A century ago, meteorologist Gilbert Walker noticed that pressure in the Pacific Ocean was linked to rainfall in India. Since then, El Niño has been considered the main predictor of the monsoon. But the correlation between rainfall and Pacific sea surface temperature wanders from -0.4 to -0.8 across decades and seemed to weaken after the 1980s. This raised doubts: can the monsoon be predicted on long timescales at all? The problem was not the connection itself but the unfortunate choice of indicator.

Instead of surface temperature, the authors used the depth of the 20°C isotherm (D20), the ocean layer where water warms to twenty degrees. This layer shifts by tens of meters and remembers the ocean's state for years, while the surface forgets everything within a few months.

The scientists built correlation maps between the depth of this layer and monsoon rainfall for 1875–2010, kept only significant regions, and projected anomalies onto them. The result was the Dp index, which captures signals from the Pacific, Indian, and Atlantic Oceans simultaneously. The correlation between Dp and monsoon rainfall reaches 0.8 at an 18-month lead time and grows to 0.9 in recent decades. Meanwhile, the correlation with surface temperature drops to 0.2 at the same horizon.

Next, they reconstructed the dynamics of the index and rainfall in phase space and measured the joint recurrence of states. It turned out that rainfall and the deep ocean layer enter chaotic synchronization at lags of 4–6 and 16–18 months. The phase locking between them at an 18-month lag is 0.75, corresponding to a lag synchronization regime. This means that two chaotic oscillators, the ocean and atmosphere, resonate with a delay, and this resonance provides predictability beyond the usual limit of deterministic chaos.

The sliding 21-year correlation of Dp with rainfall barely changes over time, unlike the Niño 3.4 index. In CMIP6 models under RCP2.6 and RCP4.5 scenarios, the connection remains strong and stable until 2100.

Now it is clear that the El Niño–monsoon link has always been stable; it was just measured with the wrong tool. Already today, forecasts of the Indian monsoon can be made 18 months in advance, which is important for agriculture and water planning in South Asia. Two pendulums on the same wall do not just swing side by side; they hear each other through the ocean's depths.

The depth of the 20°C isotherm in the tropics changes by tens of meters, and these slow waves remember the ocean state for years, while the surface forgets everything within a few months.

PLV = \left| \frac{1}{N} \sum_{t=1}^{N} e^{i\Delta\phi(t)} \right|
A measure of how stably two oscillatory systems maintain their phase difference
Concepts
interannual variability El Niño ENSO El Niño-Southern Oscillation monsoon monsoon onset La Niña spring predictability barrier Indian Ocean capacitor effect Multivariate ENSO Index Southern Oscillation Indian Ocean Dipole relative ONI seasonal cycle natural variability seasonal forcing Jin-Timmermann model TAO/TRITON moorings recharge oscillator model ERSST