From California mudslides, to Australian droughts, to empty fishing nets off the South American coast, El Niño is a catastrophic worldwide weather phenomenon that emerges every few years, leaving devastation in its wake. While meteorologists have up until now had only limited ability to predict when the next such event will occur, a research team made up of Israeli and German scientists recently discovered a way to predict El Niño’s events – with high of accuracy – a full year in advance.
“There are over fifty climate models that have been used to try to predict El Niño events, but they provide warning at relatively low accuracy, and only about six months ahead of time,” says Prof. Shlomo Havlin, of Bar Ilan University. “In our new approach, which uses climate network analysis to reveal the evolution of network links in the Pacific Ocean, we were able to double the advance warning time to one year, while providing a predictive accuracy of an unprecedented 75 percent.”
Get our weekly highlights directly in your inbox!Sign up
Seeing the bigger picture
According to Prof. Hans Joachim Schellnhuber, a co-author of the study, the researchers’ predictive model benefits from seeing the bigger picture while focusing on one specific factor: temperature. “Large-scale warming events occur when temperature changes build up over time and affect one another,” he says. “It is like an orchestra of 200 musicians playing together. If the different regions in the Pacific are rather playing their own tunes, like soloists, no El Niño develops. On the other hand, when there’s a ‘harmony’ building up – a harmony which collapses when the El Niño event finally arrives – this serves as a warning, and a very accurate one at that.”
Prof. Armin Bunde, who contributed to the study, describes how the team detected data strongly predictive for a 2014 El Niño event already in September 2013 – a prediction that was later proven to be correct.
“Predictions by other, much bigger models wobbled up and down and as late as November 2014 gave a likelihood of only 58 percent that an El Niño will arrive,” Bunde recalls. “In contrast, the new and early forecast was stable over the whole period before the event and provided a significantly higher probability of 75 percent.” Bunde points out that the US National Oceanic and Atmospheric Administration only recently declared El Niño’s arrival – the event started last year, but it has to last for some time to be officially recognized. Japan’s weather bureau saw the conditions fulfilled in December last year.
Applying a multi-facted approach
As Prof. Havlin explains, the researchers’ predictive method combines mathematics, physics – and history. “We used weather data dating back to 1980, examining temperature changes in sites throughout the Pacific, and characterizing the network-based interactions between these individual sites,” he says. “Correlating this temperature data with the weather events occurring during this time period, we were able to positively identify the network characteristics that were present – at least 75 percent of the time – when an El Niño event broke out the following year.”
The next step for the scientists was to put their model to work, and accurately predict the onset of an El Niño event in the future. The researchers’ success was the basis for their recent publication.
“Our results indicate that this new predictive method is more accurate and reliable than other models,” Havlin says. “This is something very important for societies wishing to plan ahead, and minimize El Niño’s devastating effects.”
The scientists’ findings were recently published in an article entitled “Very Early Warning of Next El Niño” in the Proceedings of the National Academy of Sciences (PNAS). Prof. Shlomo Havlin is a former President of the Israel Physical Society and a faculty member in Bar-Ilan University’s Department of Physics. Prof. Hans Joachim Schellnhuber co-authored the study and is director of Germany’s Potsdam Institute for Climate Impact Research. Prof. Armin Bunde is a theoretical physicist at Justus-Liebig-Universität Gießen (JLU) in Germany, who led the study together with Havlin and Schellnhuber.