New warning system enables prediction of extreme rainfall events in the Andes

Research study conducted by Brazilian and German scientists applied the methodology of complex networks to meteorological data obtained by satellites.

By José Tadeu Arantes

Agência FAPESP – Torrential rains accompanied by floods, mudslides and fatalities have occurred in the Andean region. These extreme events will tend to increase in magnitude and frequency as a result of global climate change.

Exacerbating the effects is the poverty of the affected area, with its precarious roads and its human dwellings built dangerously close to the edges of mountain passes. The scenario is highly similar to that of the Himalayan region, where more than 10,000 people lost their lives as a result of the catastrophic rainfalls of 2013.

Now, a powerful warning system, capable of providing up to two days of advance notice and a high degree of accuracy, is available and can easily be used by weather forecasting centers.

Fruit of the collective efforts of Brazilian and German scientists who applied the complex networks technique to meteorological data obtained by satellite, the system was reported in the article “Prediction of extreme floods in the eastern Central Andes based on a complex networks approach,” published in the on-line journal Nature Communications October 14, 2014.

The research is part of the thematic project, "Dynamical phenomena in complex networks: fundamentals and applications," coordinated by Elbert Einstein Nehrer Macau, principal investigator at the National Institute for Space Research (INPE), in Brazil, and by Jurgen Kurths, from Humboldt University, in Germany. The project is funded by FAPESP and the German Research Foundation (DFG).

The two-nation climatology team analyzed approximately 50,000 high-resolution weather data time series recorded over the past 15 years and made available by the U.S. Space Agency (NASA) and the Japan Aerospace Exploration Agency. The team determined that after emerging in the region of Buenos Aires, Argentina, large convective systems moved northwest towards the Andes, where, days later, they caused torrential rains. Surprisingly, these extreme events spread in a direction opposite to that of the winds, which moved southward.

“Simple correlations between data time series are made every day in many fields of science – for example, correlations between series of temperatures recorded at specific points in the ocean and time series of precipitation over the continent. The methodology we developed is much more complex and complete, allowing us not only to correlate huge amounts of data but also to establish cause and effect relationships between the phenomena observed,” said Henrique de Melo Jorge Barbosa, professor at the Physics Institute of the University São Paulo and one of the authors of the article that appeared in Nature Communications.

The thematic project also has applications in other fields that range from photonics to neurology.

"Flying river"

“What surprised us was discovering that the shift in rainfalls occurs in the opposite direction to the ‘flying river’.”

The expression “flying rivers” was popularized by José Antonio Marengo Orsini, a researcher at the National Center for Natural Disaster Monitoring and Alerts (CEMADEN) and a member of the United Nations (UN) Intergovernmental Panel on Climate Change (IPCC). It refers to large masses of vapor-filled air, often accompanied by clouds, propelled by the winds.

The “flying river,” which originates in the Atlantic Ocean and is loaded with additional moisture after interacting with the Amazon Rainforest, can transport volumes of water of the same order of magnitude as the earth’s largest river, the Amazon (200,000 cubic meters per second). Blocked by the 4,000 meter high wall formed by the Andes mountain range, this moist air mass shifts southward to the Central-Western, Southeastern and Southern regions of Brazil and northern Argentina.

“We initially thought that the time sequence of the rainfalls would match the direction of the flying river. In other words, we thought that increased rainfall over the Amazon Forest would cause more intense rain in the south. It seemed obvious. However, by using the new methodology, we determined that the extreme rain events spread, not from north to south, but from south to north. We did the calculations several times and found that, in reality, extreme rainfall events spread in the direction opposite to that of the flying river,” Barbosa said.

The researchers had to refer back to atmospheric physics and meteorology to understand why this was the case. “Actually, the direction of the shift in precipitation depends on the direction of the shift in instability, from south to north, and not from the direction of the shift of the moist air mass, from north to south. And this direction in the shift of the cold front is determined by a low-pressure system that forms in the north of Argentina,” Barbosa explained.

Based on this discovery, the researchers were able to establish a set of rules that can be used to predict the occurrences. Typically, after a peak in convection and rain in the north of Argentina, certain specific pressure and wind circulation conditions cause the rainfall to move to the Andes and the Amazon region. “With this rule, we set up our forecast system. And compared with traditional systems, we experienced a much larger degree of accuracy. In El Niño years, when extreme events were stronger and more frequent, our system presented a 90% margin of accuracy. For the other years, accuracy was at least 60%,” he said.

“We were not able to predict when the first rainfalls would occur in the north of Argentina. But once they did occur, and once certain pressure and wind conditions were determined, we could say, with a high degree of accuracy, that extreme rains would occur in the Andes region two days later.”

There is still a long way to go before this system can become a tool used by civil defense entities, but the most difficult step, from the scientific standpoint, has already been taken. “The tool is available. And people who would like to use it do not have to understand complex networks or re-analyze the 15 years of data that we used. They just need to follow our recipe,” Barbosa concluded.