El Niño: Predicting Strength and Onset a Year in Advance ✎

Original: "Forecasting the magnitude and onset of El Nino based on climate network"
arXiv:1703.09138 · 2017-03-27 · arXiv non-exclusive · independent analysis · 2 min
A climate network in the Niño 3.4 region predicts the onset and strength of El Niño about a year ahead.

An orchestra falls out of tune before a storm, then plays in unison again. In the equatorial Pacific, within a rectangle stretching from 5°S to 5°N and from 120°W to 170°W, water temperatures usually change in concert, like musicians playing the same part. This area is called the Niño 3.4 region. When that coherence breaks, the orchestra falls into discord, and a few months later El Niño arrives: the warm phase of the climate cycle that reshapes weather from Australia to Europe.

Scientists built a climate network from 105 points in this region. Each point is a location where sea surface temperature is measured daily. Invisible threads stretch between points: if temperatures at two places vary similarly, the thread is strong; if each follows its own course, the thread snaps. It turns out that before El Niño, the threads weaken and the orchestra loses its harmony. This discord can be quantified by a number called the forecast index FI. When FI rises, disorder in the region grows.

Rising disorder in the Niño 3.4 region precedes El Niño by a year, and the peak of that disorder is proportional to the future strength of the event.

Tests on four datasets showed that the fraction of correct onset predictions ranged from 0.81 to 0.95, with false alarm rates from 0.09 to 0.24. The average lead time was 6.9 to 12.2 months. For strength prediction, the correlation between predicted and observed magnitude reached 0.76 on one dataset. This means that the height of the disorder peak can tell you in advance how powerful the El Niño will be.

Previously, reliable prediction of El Niño strength more than six months ahead was impossible. Existing methods either predicted only the onset or required data from the entire Pacific. The new approach gets by with temperatures from a single small region, making it computationally light and applicable to long observational records. It is as if a conductor could predict, from the sound of one violin section, how loudly the whole orchestra will crash a year later.

Such forecasts are not just for curiosity. Farmers in India and Australia learn in advance whether to expect drought. Coastal cities in Peru prepare for floods. Dam operators decide how much water to store. And the orchestra, once out of tune, begins to play in unison again, and now we hear the opening notes much earlier.

The FI index uses only temperature data from a small region of the Pacific Ocean, yet it predicts an event that affects weather worldwide. The strongest El Niño since 1948 (2014–2016) was predicted by this method a year before it began.

Concepts
El Niño La Niña Oceanic Niño index NOAA CPC ERSST forecast ENSO Multivariate ENSO Index spring predictability barrier interannual variability relative ONI climate network seasonal cycle Westerly wind burst monsoon onset NOAA PMEL / TAO forecasting equatorial wave propagation drought probabilistic forecast