Over the Pacific Ocean, giant seesaws swing back and forth. Warm water pushes against the coasts of South America, and El Niño begins; a few years later, the ocean surface cools, and the seesaw tips the other way, into La Niña. These oscillations last for years and reshape cloud cover, rainfall, and wind across an entire continent. Down below, Brazil and Chile are ramping up solar power plants so fast that generation forecasts can't keep up with the weather.
To understand how the phases of the ocean seesaw affect sunlight, scientists gathered data from 31 weather stations: 25 in Brazil's solar belt and 6 in Chile. They selected three full years: La Niña, neutral, and El Niño. For each series, they applied detrended fluctuation analysis (DFA): the signal is integrated, divided into windows of varying sizes, a local polynomial trend is subtracted, and the root mean square deviation is calculated. The slope of the fluctuation-versus-window-size relationship yields the self-similarity exponent α.
The results were striking. During the neutral period, almost everywhere in Brazil, α was above 0.5 on scales longer than 10 days, meaning the series are persistent. El Niño significantly increased α compared to neutral and La Niña periods, especially in the northern part of the country. In Chile, all cities showed antipersistence on scales from 27 hours to 10 days, while on larger scales during La Niña, two southern cities transitioned from a subdiffusive to a persistent regime. The city of Coxim in Brazil showed the largest spread: from antipersistence during El Niño to subdiffusion during La Niña and persistence in the neutral year.
These swings hit energy planners hard. Measurements taken during an El Niño or La Niña year cannot be directly extrapolated to the station's decade-long service life without adjustments. Before this work, it was unknown that ocean phases change not only the mean radiation values but also the very nature of its variability. Now it's clear: certified output estimates can be systematically off if you don't account for which way the seesaw was tipping during the measurement year. For designing and financing solar plants in Brazil and Chile, this is not an abstraction but a concrete risk.
The city of Coxim in Brazil switched regimes three times in three years: antipersistent during El Niño, subdiffusive during La Niña, and persistent in the neutral year. Such variability is comparable to a person who in some years predictably repeats their habits and in others behaves chaotically.