From Shock to Forecast: TOGA, the TAO Array and the Taming of El Niño

Every mature science has its founding embarrassment. For El Niño forecasting, it was the winter of 1982-83, when the strongest event in at least a century developed essentially unobserved — its warming masked by volcanic aerosols, its onset missing the expected precursors — and was recognized only near its peak. The full story of that ambush is told in our article on the secret super El Niño. What followed it was one of the great redemption arcs in environmental science: within fifteen years, El Niño went from unforecastable to the most predictable interannual signal on Earth.

TOGA: a decade for the tropical Pacific

The institutional answer to 1982-83 was TOGA — the Tropical Ocean–Global Atmosphere program — a ten-year international campaign under the World Climate Research Programme, running from 1985 to 1994. Its premise was Bjerknes's coupled vision made operational: if ocean and atmosphere generate El Niño together, then predicting it required observing them together, continuously, across the whole equatorial Pacific.

The program's centerpiece grew mooring by mooring into the TAO array: approximately 70 ATLAS buoys anchored along the equatorial band, measuring surface winds, sea-surface temperature and the ocean's interior down to 500 metres — where the displaced warm water that foretells an El Niño actually travels. Each buoy transmitted its data in real time by satellite. The array was completed in December 1994, transforming the least-observed ocean on Earth into arguably the best-observed. Japan later contributed TRITON moorings in the western Pacific, and in 2000 the system was renamed TAO/TRITON; it continued under the CLIVAR program and remains the backbone of ENSO monitoring today, alongside Argo floats and satellite altimeters.

The first forecast

While the buoys were going in, a parallel revolution was unfolding in the models. At Columbia University's Lamont-Doherty Earth Observatory, Mark Cane and Stephen Zebiak built the first coupled ocean-atmosphere model of ENSO — a deliberately simplified machine that captured the essential Bjerknes feedback and the ocean's slow memory.

In 1986 they published "Experimental forecasts of El Niño" in Nature (volume 321, pages 827-832), and made a daring claim: their model indicated an El Niño was coming. The 1986-87 event duly arrived. It was the first successful dynamical forecast of El Niño — proof, against considerable skepticism, that the coupled system carried predictability months ahead, and that seasonal forecasting was a real science rather than a hope. Every modern ENSO prediction system is, in lineage, a descendant of the Zebiak-Cane model.

Eyes in orbit

The third pillar arrived from space. In 1992, NASA and France's CNES launched TOPEX/Poseidon, a radar altimeter that measured the height of the sea surface to an accuracy of about 4 cm. Sea level is the ocean's heat gauge: warm water stands taller. From orbit, TOPEX/Poseidon could watch Kelvin waves — the vast, slow ripples that carry warm water eastward along the equator at an El Niño's onset — and map the heat building in the western Pacific before an event.

The system faced its great test in 1997, and passed it spectacularly. The TAO array and TOPEX/Poseidon together observed the 1997-98 super El Niño from first stirrings to final decay — the first event ever watched start to finish — while models flagged it months in advance, enabling the first large-scale preparedness campaigns. The altimeter's false-color images of the swollen warm Pacific became the public face of the El Niño of the century. TOPEX/Poseidon's mission has since been carried forward by Jason-1, Jason-2, Jason-3 and Sentinel-6.

The routine miracle

Today the machinery built between 1985 and the late 1990s operates so smoothly it is easy to forget it exists. NOAA's Climate Prediction Center and Columbia's IRI issue monthly ENSO outlooks, synthesizing dozens of models. The Oceanic Niño Index — the three-month running SST anomaly that defines El Niño's official strength — has been NOAA's operational standard since 2003, computed retroactively back to 1950 so that every event since mid-century sits on a common scale.

That scale is how we now say, with precision, that 1997-98 peaked at +2.4, that 2015-16 set the record at +2.8, that 2023-24 reached +2.1. The very language of El Niño history is a product of this infrastructure.

The deeper achievement is measured in what no longer happens. Since 1982-83, no significant El Niño has arrived unannounced. The 1997-98, 2015-16 and 2023-24 events were each forecast months ahead, tracked in real time, and met with warnings that reached governments and farmers alike. Forecasts cannot abolish drought or flood — the losses of those years prove it — but they have permanently abolished surprise. For a phenomenon that spent ten millennia striking without warning, from the Moche floods to the Victorian famines, that is as close to taming as science gets.