HPCwire recently published a story about work by scientists at the National Center for Atmospheric Research (NCAR). The story features research funded through the Regional & Global Model Analysis (RGMA) program area's A Cooperative Agreement to Analyze variability, Change and Predictability in the Earth System (CATALYST) project.

By HPCwire

As wildfires blaze around the globe, the particles and chemicals in smoke can affect distant climate patterns. Using simulations, climate scientists at the National Center for Atmospheric Research (NCAR) discovered the Australian bushfires of 2019 and 2020 that scorched 46 million acres triggered a La Niña cooling event in the tropical Pacific Ocean, thousands of miles away.

Outside of regular seasons, El Niño and La Niña cycles are the most pronounced variations in global climate patterns. They originate in the tropical Pacific Ocean and demonstrate how the ocean and atmosphere interact.

“Every four years or so, on average, we get an El Niño event where things warm up in the tropical Pacific, and it gives us a response in the climate system across the globe, which leads to, for example, wet or dry years in California,” says John Fasullo, a scientist at NCAR and lead author of a May Science Advances paper on the subject.

El Niño and La Niña dominate the water cycles of Australia, South America and the U.S. Southwest. The Southwest has been dry since 2020 largely because of La Niña, but the region also can be influenced by Pacific Ocean temperature cycles or external factors like volcanic eruptions.

“Fires are so big and prolonged,” Fasullo says, “that now under the right circumstances, they can actually influence the unfolding of El Niño and La Niña events. It’s like a Rube Goldberg of chain reactions: The fires go off, and winds carry their smoke to the Southern Ocean, where the smoke interacts with clouds to make them brighter and longer-lived. And these effects, in turn, cool the tropical Pacific and increase the odds of an ensuing La Niña event.”

Wildfire smoke has been a surprising blind spot, until now considered external to the climate system. Fires have been hard to account for because of the many processes that govern their variability.

Fasullo and NCAR colleagues led by principal investigator Gerald Meehl are currently running two related simulation DOE Office of Science-sponsored projects: the Smoothed Biomass Burning Large Ensemble to explore the wildfire emissions’ effects in climate models (E3SM2) over a 100-year period, and the Seasonal-to-Multiyear Large Ensemble (SMYLE) to predict El Niños and La Niñas and other low-frequency variations over a decade. For the first project, the NCAR team has been allocated 260,000 hours on the Perlmutter-CPU system at the National Energy Research Scientific Computing Center; for the second, 300,000 hours, also on Perlmutter. Both are through the ASCR Leadership Computing Challenge program.

A large ensemble is an analog to a weather forecast, which shows all possible tracks a hurricane can take. “These hurricane forecasts are derived from multiple weather forecasts that encompass the range of natural noise within the system,” Fasullo explains. “While we can’t be sure if the hurricane is going to take any individual line, we’re confident it’s likely to be somewhere within the plume.” The same holds true with climate: Small differences from one simulation to the next can provide alternative outcomes that fall inside a range.

A large ensemble requires supercomputers that run simulations on century scales rather than a couple weeks. And since the team may want to run 40 different simulations of 150 years, they need the fastest supercomputers available today: exascale systems.

Exascale systems also will boost the models’ resolution. But if the physics are wrong, resolution doesn’t matter. “It’ll require a bit of balancing act between high-resolution exascale simulation and lower-resolution simulations aimed at developing the physics of the models,” Fasullo says. “So we’ll try to develop some of these tools first, such as realistically depicting the emissions of wildfires from land in lower-resolution models, where we can run lots of experiments to see what works and what doesn’t, and then eventually move to higher resolution.”

The team’s work spotlights the immediate need to integrate wildfires into the climate system. “We’ve treated it, historically, as an external forcing on the climate system much the way that the sun is,” Fasullo says. “And in reality, wildfire is a highly coupled component of the climate system. We need to get started representing it that way.”