Every few years, the Pacific Ocean off the coast of Peru warms up, and then fields dry out in Australia while deserts in Chile get drenched. This phenomenon is called El Niño. Usually it arrives with nearly the same intensity, but sometimes the ocean heats up as if someone had given the pendulum a mighty shove: the heat lasts longer, and the consequences are devastating. Why this happens has long remained a puzzle.

Scientists have built a simplified model in which the ocean and atmosphere exchange heat. The model has one slow variable, the depth of the warm water layer in the western Pacific, and fast variables, the surface temperatures in the east and west. When the parameter controlling heat transport along the equator crosses a critical value, the system's behavior changes abruptly. What mathematicians call an interior crisis occurs: the chaotic attractor suddenly expands, and trajectories begin to visit regions that were previously inaccessible.

A key role is played by a channel in phase space, a narrow passage that trajectories must go through. If the channel is wide, events occur one after another with similar strength, like ordinary El Niños. But if the channel narrows, trajectories get stuck and then break out with a giant temperature surge. That is why rare extreme events show large variability in amplitude and return time, and their height distribution is multimodal with a bounded tail.

It was previously known that El Niño manifests as chaotic mixed oscillations, but the origin of rare giant events remained unclear. Now it has been shown that they arise through a specific mechanism: a period-doubling cascade merges with a period-adding cascade, and at that moment the attractor expands. This explains why strong El Niños are fundamentally less predictable than ordinary ones. For climatology, this means that even a simple model can reproduce the complex statistics of extreme events, and similar mechanisms should be sought in more realistic models and in observational data. Today, this approach helps improve our understanding of how the ocean and atmosphere generate rare catastrophic events, although accurate prediction is still far off.