You are here: Home News In Short New approach in El Niño forecasting potentially doubles the lead-time and helps forecasting its magnitude

New approach in El Niño forecasting potentially doubles the lead-time and helps forecasting its magnitude

06.01.2020 - El Niño, probably the most far-ranging climate phenomena on Earth, is likely to hit again in 2020, as groundbreaking research by PIK and others has shown. Now, PIK researchers also found a new way to improve forecasts regarding its magnitude using data from air and sea surface temperature series.
New approach in El Niño forecasting potentially doubles the lead-time and helps forecasting its magnitude

The Niño 3.4 region. The red circles indicate the 22 nodes in the Niño 3.4 region. The curves are examples ofthe temperature anomaly time series for 3 nodes in the Niño 3.4 region.

Occurring roughly every two to seven years, El Niño (God’s Child) can send torrential rainfalls and empty fishing nets to Peru, droughts to Australia, or changes to the Indian monsoon patterns. Its worldwide impact makes early forecasting of this phenomenon so vital. While significant progress was made in forecasting the time El Niño is likely to hit the world’s shores, reliable forecasts of its magnitude remained limited to about six months in advance. Now, an international team of researchers led by the Potsdam Institute for Climate Impact Research has succeeded in doubling that time, allowing for El Niño forecasts a year in advance and also delivering information about its magnitude with high accuracy.

The new approach draws on the methodology of medical diagnostics: It relies on quantifying the complexity of multiple air and sea surface temperature time series in the eastern equatorial Pacific. The authors call the complexity indicator they derive from spatiotemporal temperature variations “System Sample Entropy” (SysSampEn). They found that the SysSampEn for a given calendar year was strongly and positively correlated with the magnitude of an El Niño event in the following year. Using this correlation, the authors constructed a forecasting index that predicted 9 out of 10 El Niño events between 1984 and 2018, with 3 false positives. For the current year 2020, forecasts show very high probabilities for an El Nino at the end of the year. The SysSampEn approach shows that its magnitude is 1.48±0.25 ℃, indicating a moderate El Nino event.

With a conventional early warning period of at most half a year so far, people in the tropics and subtropics are poorly prepared for the often devastating consequences of "El Niño" at irregular intervals around Christmas. The new approach could be a potential game changer in El Niño forecasting, thus helping to reduce its devastating effects on millions of people across the globe.

Article: Jun Meng, Jingfang Fan, Josef Ludescher, Ankit Agarwal, Xiaosong Chen, Armin Bunde, Jürgen Kurths, Hans Joachim Schellnhuber (2019): New approach in El Niño forecasting potentially double the lead-time and helps forecasting its magnitude. PNAS [DOI:10.1073/pnas.1917007117]

Weblink to the article:

Previous PIK research on El Niño: Josef Ludescher, Armin Bunde, Shlomo Havlin, Hans Joachim Schellnhuber (2019): Very early warning signal for El Niño in 2020 with a 4 in 5 likelihood. arXiv:1910.14642 (find our press release here)

This work was supported by the East Africa Peru India Climate Capacities (EPICC) project. This project is part of the International Climate Initiative (IKI). The Federal Ministry for the Environment, Nature Conservation and Nuclear Safety (BMU) supports this initiative on the basis of a decision adopted by the German Bundestag. Please find more information on the project here

Document Actions

Contact PR

For further information please contact the PIK press office:
Phone: +49 331 288 25 07


Would you like to subscribe to our rss feed?
RSS Feed rss feed

By activating this social media plug-in you agree that your IP address may be forwarded to Facebook or Twitter. More information may be found in our Privacy Policy (section 5.2 Social Media Plug-Ins).