For the last 11,000 years, the southern Pacific Ocean has cycled between warm El Niño and cold La Niña conditions, driving the climate on both sides of the ocean. But new modelling suggests that these cycles may be interrupted as a world warms under human-induced climate change.
An international team used one of South Korea’s fastest supercomputers to create a series of global climate model simulations. These simulations had unprecedented spatial resolution – of 10 km in the ocean and 25 km in the atmosphere – able to capture small-scale climatic processes like tropical cyclones and instability waves.
“Our supercomputer ran non-stop for over one year to complete a series of century-long simulations covering present-day climate and two different global warming levels,” says co-author Sun-Seon Lee from the IBS Centre for Climate Physics (ICCP) at Pusan National University in South Korea. “The model generated two quadrillion bytes of data; enough to fill up about 2,000 hard disks.”
The resulting study is published in Nature Climate Change.
The simulations add a new piece in the long-standing puzzle of how El Niño and La Niña events (commonly known as the El Niño-Southern Oscillation, or ENSO) will be affected under climate change.
“Two generations of climate scientists have looked at this issue using climate models of varying complexity,” explains Axel Timmermann, director of the ICCP. “Some models simulated weaker; others predicted larger eastern Pacific temperature swings in a future warmer climate. The jury was still out.”
He notes that most of these previous models always produced temperatures in the equatorial Pacific that were colder than observations.
“This prevented them from properly representing the delicate balance between positive and negative feedback processes that are important in the ENSO cycle,” Timmermann explains.
The new model addresses these temperature anomalies, and concludes that increasing CO2 concentrations will weaken the intensity of the ENSO temperature cycle.
The mechanism? Evaporating water will cause future El Niño events to lose heat to the atmosphere more quickly; plus, in the future, the temperature difference between the eastern and western tropical Pacific will be reduced, in turn decreasing the development of temperature extremes during the ENSO cycle.
But the new simulations also show the detailed structure of tropical instability waves that usually hasten the demise of a La Niña event – they are projected to weaken, partially offsetting the two above factors.
“There is a tug-of-war between positive and negative feedbacks in the ENSO system, which tips over to the negative side in a warmer climate,” explains Malte Stuecker, co-author from the University of Hawaiʻi at Mānoa. “This means future El Niño and La Niña events cannot develop their full amplitude anymore.”
However, even though this model predicts long-term weakening in the ENSO cycle, the authors say that El Niño and La Niña-related rainfall extremes will continue to increase in coming years.
But this model is far from the only one addressing these questions.
Earlier this month, Australian scientist Wenju Cai from the Centre for Southern Hemisphere Oceans Research at CSIRO released a paper that reviewed 50 recent models. Overall, they showed that El Niño and La Niña events are expected to increase in frequency and intensity as a result of climate change.
Cai, who was not involved in these new simulations, says that they add necessary complexity and resolution, in particular incorporating ocean meso-scale eddies.
But he cautions that “it is just one model. We need 20 or more models to see if there is an inter-modal consensus. In our series of papers, there are still 20% of models generating a reduction in ENSO.”
He also notes that this new simulation considers the situation after the climate has stabilised, while other models examine how the ENSO system changes as CO2 in the atmosphere is still increasing.
“There are suggestions that ENSO in transient and stabilised climate could be different, including my own work,” Cai says.
But he praises the new methods presented in this work. “We need more of such models with a transient CO2 increase to see if there is inter-model consensus,” he concludes.