Why El Niño Arrives in Winter ✛

Original: "A Stochastic-Thermodynamic Constraint on the Seasonal Phase Locking of the El Niño-Southern Oscillation"
· Yuki Yasuda, Tsubasa Kohyama
arXiv:2606.19778 · 2026-06-18 · CC BY 4.0 · 1 min
The ocean is like a bathtub with a leaky plug: water only rises when the leak slows down.

Every winter, something strange happens in the tropical Pacific: the water off the coast of Peru becomes noticeably warmer than usual. This phenomenon is called El Niño, and it affects weather around the globe. The big puzzle is why El Niño's peak almost always falls in December. The ocean has plenty of heat in summer too, but the temperature rise only begins in autumn.

Scientists have compared the ocean to a bathtub where the tap turns on and off, and the plug isn't quite tight. Water comes in, but some drains away. For the level to rise, more must come in than goes out. It turns out that in November, the 'leak' slows down: the ocean seems to stop 'forgetting' its warm state, and temperatures start to climb quickly. In September and February, the leak is at its strongest, so temperature growth is suppressed in those months.

Scientists call this 'leak' irreversibility. It shows how hard it is to reverse a process. In November, irreversibility is minimal, and the ocean can warm up quickly.

Previously, it was thought that only the amount of heat mattered. Now it turns out that another factor is important: how irreversible temperature changes are. This discovery helps explain why climate models often get El Niño's timing wrong. Perhaps they misjudge how the ocean 'forgets' its past in different seasons.

The peak of entropy production in September and February means that in these months ocean temperature transitions are most irreversible: the system 'forgets' the past faster than in other seasons.

\frac{\dot{P}}{2} \ge \frac{1}{(\sigma_T)^2 \langle T^2 \rangle} \left( \frac{1}{2} \frac{d}{dt} \langle T^2 \rangle \right)^2
The partial entropy production rate bounds from above the normalized rate of change of temperature variance.
\dot{P} = \frac{2\dot{Q}}{(\sigma_T)^2} + \frac{dS}{dt} - \dot{I}
Partial entropy production is the sum of an irreversibility indicator, the change in Shannon entropy, and an information flow.
Scientists
Rudolf ClausiusJames Clerk MaxwellLudwig BoltzmannEmmy NoetherAlbert Einstein
Concepts
ENSO El Niño-Southern Oscillation spring predictability barrier TAO/TRITON moorings recharge oscillator model Jin-Timmermann model Multivariate ENSO Index La Niña El Niño index Southern Oscillation ERSST natural variability seasonal forcing sea surface temperature Westerly wind burst North Equatorial Countercurrent seasonal cycle Warm water volume carnot limit relative ONI