Every few years, the Pacific Ocean takes a deep breath. Warm water that usually piles up near the coasts of Indonesia spreads eastward toward South America, and the trade winds that had been pushing it west temporarily ease. This inhale and exhale is called El Niño–Southern Oscillation, or ENSO. It doesn't just change weather in the tropics: its pulse reverberates as droughts in Australia, floods in Peru, and even crop yields in far-off countries. But this pulse has an irregular rhythm, and the planet's cardiogram holds traces that scientists are only beginning to read.
Until recently, the ENSO cardiogram was recorded with a crude instrument. Standard models assumed that deviations from the average rhythm were random noise, like static on a radio: many small blips, with large ones almost impossible. But when scientists compared records spanning 66 years (SOI) and 147 years (BEI), they found that strong pulses occur far more often than such a model predicts. The tails of the distributions follow a power law: the probability of an event with twice the threshold amplitude drops not by hundreds of times, but only by a factor of 23. It's as if, in a calm sea, a wave as tall as a house rolled in every week.
The planet's pulse remembers its past beats. The analysis showed that up to a timescale of about 2.5 years (for SOI) and 2 years (for BEI), fluctuations grow with an exponent of about 1.1–1.4, meaning the ocean's current state depends on what happened several years earlier. Beyond that timescale, memory weakens but does not vanish completely. After removing seasonality, the slope was 0.89 for SOI and 0.84 for BEI, a sign of long-range correlations where the system remembers itself across all scales. Multifractal analysis added another layer: the rhythm isn't just irregular; it consists of many interwoven rhythms, like a complex score where different instruments play at different tempos.
Why does this matter? Models used for El Niño forecasting are built on Gaussian noise and short memory. They systematically underestimate the probability of strong events. The forecast for the famous 2015–2016 El Niño was overestimated precisely because the model didn't account for power-law tails. If models incorporate long memory and power-law distributions, forecasts would become more accurate, and risks to agriculture, health, and marine ecosystems could be assessed more realistically.
So the ocean's breathing ceases to be a mysterious noise. Its cardiogram, recorded over a century and a half, shows that the planet doesn't breathe like a metronome but like a jazz drummer who remembers every beat and sometimes unleashes a wave on the world that we weren't ready for.
The power-law tail exponent μ = 4.52 for SOI means that an event with amplitude twice the threshold occurs roughly 23 times less often than an event at the threshold amplitude. That is much slower than the Gaussian distribution predicts.