Two Faces of El Niño: How Different Patterns of Ocean Heat Rewrite the Map of Droughts and Floods

Original: "Differential imprints of distinct ENSO flavors in global patterns of very low and high seasonal precipitation"
arXiv:1702.00218 · 2017-02-01 · CC BY 4.0 · 3 min
Eastern Pacific and Central Pacific types of El Niño and La Niña affect global precipitation differently, and this is only visible in the statistics of coincidences of rare events.

The same orchestra sounds different when two different conductors stand at the podium. One swings the baton widely, and the whole hall hears his accents; the other barely nods, and only individual instruments respond. So it is in the Pacific Ocean: the warm phase El Niño and the cold phase La Niña are two conductors of global precipitation, but each has two different personalities.

The classic Eastern Pacific type (EP) warms the water off the coast of South America. The Central Pacific type (CP), also called Modoki, shifts the warm pool to the middle of the ocean. Previous studies mostly looked at average precipitation and did not distinguish how the tails of the distribution behave, that is, very wet and very dry seasons. The new work applies an event-based approach to the global precipitation grid for the first time: the authors counted how often extreme seasons coincide in time with a given phase of ENSO.

The event coincidence analysis method is like checking two alarm clocks: if one always rings when the sun rises, and the other only occasionally and on random days, then the first is linked to dawn, while the second is not. Here, instead of alarm clocks, we have rare events: seasons with precipitation above the 80th or below the 20th percentile.

The results were striking. EP El Niño confirmed known links: droughts in Indonesia, the Philippines, and northern Amazonia in autumn, wet conditions in western North America and Ecuador, and dry seasons in northeastern Brazil in spring. CP El Niño loses many of these patterns but reveals new ones: droughts on the eastern coast of Australia, wet conditions in southern Chile and Central Asia in winter, and dry seasons in southeastern Africa in spring. For La Niña, the EP type produces extensive coherent anomalies, wet in Australia and Indonesia, dry in southern Europe and southern Brazil, while the CP type yields only isolated patches: wet in Australia in autumn, dry in Florida in winter, and dry in the UK, Ireland, and western Kazakhstan in spring.

For the first time, it is shown that CP El Niño is accompanied by increased precipitation in Central Asia, and CP La Niña by a deficit there. Previously, this region was considered insensitive to ENSO.

Overall, CP periods produce statistically significant coincidences less often and over smaller areas than EP, and this holds for both phases. The work shows that the distinction between types matters not only for average precipitation but also for the probability of extreme seasons. This strengthens the argument that La Niña should also be divided into two types, although earlier this was controversial based on sea surface temperature correlations.

If the type of a developing El Niño can be determined early enough, it becomes possible to provide advance warning of likely droughts or floods in specific regions. The method can be extended to other climate variables, such as air temperature, soil moisture, river runoff, as well as crop yields and water availability. Then the two conductors will no longer be a mystery: their gestures will be read in advance, before the first chords.

The term "Modoki" for Central Pacific El Niño comes from a Japanese word meaning "similar but different," which accurately reflects its character: it looks like classic El Niño on the surface but behaves differently.

P^{\pm}_{s,i}(t) = \Theta(\pm P_{s,i}(t) \mp p^{\pm}_{s,i})
The Heaviside function equals 1 if the seasonal precipitation total exceeds the 80th percentile (for wet) or falls below the 20th percentile (for dry), and 0 otherwise.
ECR^{\pm}_{s,i,\bullet} = \frac{\sum_t X_{\bullet}(t) P^{\pm}_{s,i}(t-\tau)}{\sum_t X_{\bullet}(t)}
The numerator is the number of years when an ENSO phase of a given type coincided with an extreme season; the denominator is the total number of such phases. The lag τ is 0 for SON and DJF, and 1 for MAM.
Concepts
ENSO El Niño La Niña spring predictability barrier Multivariate ENSO Index El Niño index Southern Oscillation Indian Ocean capacitor effect ERSST NOAA CPC Jin-Timmermann model TAO/TRITON moorings interannual variability Pacific decadal oscillation relative ONI recharge oscillator model equatorial wave propagation Westerly wind burst Warm water volume natural variability