Inside Yellowstone’s Fiery Heart

Beneath the steaming geysers and bubbling mud pots of Yellowstone National Park lies one of the world’s most closely watched volcanic systems. Now a team of geoscientists has uncovered new evidence that sheds light on how this mighty system may behave in the future—and what might keep it from erupting. The findings were recently published in Nature.

A team of researchers from Rice University, University of New Mexico, University of Utah, and The University of Texas at Dallas have discovered a sharp, volatile-rich cap of magma just 3.8 kilometers beneath Yellowstone’s surface. This cap acts like a lid, helping to trap pressure and heat below it. Using innovative controlled-source seismic imaging and advanced computer models, their findings suggest that the Yellowstone magma reservoir is actively releasing gas while remaining in a stable state.

The research, led by Rice’s Chenglong Duan and Brandon Schmandt along with collaborators, provides new insight into how magma, volatiles and fluids move within Earth’s crust. The project was supported by the National Science Foundation (NSF) and supercomputing resources at the Texas Advanced Computing Center (TACC). 

“For decades, we’ve known there’s magma beneath Yellowstone, but the exact depth and structure of its upper boundary has been a big question,” said Schmandt, professor of Earth, environmental and planetary sciences. “What we’ve found is that this reservoir hasn’t shut down—it’s been sitting there for a couple million years, but it’s still dynamic.”

Previous studies suggested the top of Yellowstone’s magma system could lie anywhere from 3 to 8 kilometers deep—an uncertainty that left geologists debating how the magma system today compares with conditions before prior eruptions.

That changed after Schmandt conducted a high-resolution seismic survey in the northeastern part of the caldera. A 53,000-pound vibroseis truck—typically used for oil and gas exploration—essentially generated tiny earthquakes to send seismic waves into the ground. These waves reflected off subsurface layers and were recorded at the surface, revealing a sharp boundary at about 3.8 km depth.

“The motivation behind my research is to advance structural seismic imaging beyond the limits of conventional travel-time methods,” said Duan, a postdoctoral research associate. “Using a wave-equation imaging technique I developed during my Ph.D. for irregular seismic data, we made one of the first super clear images of the top of the magma reservoir beneath Yellowstone caldera.”

“Seeing such a strong reflector at that depth was a surprise,” Schmandt said. “It tells us that something physically distinct is happening there—likely a buildup of partially molten rock interspersed with gas bubbles.”

To better understand what causes this signal, Duan and Schmandt modeled various rock, melt and volatile combinations. The best match they determined is a mixture of silicate melt and supercritical water bubbles within a porous rock matrix resulting in a volatile-rich cap with about 14% porosity, half of which is occupied by fluid bubbles.

The NSF-funded Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program awarded Schmandt allocations on the Stampede3 supercomputer at TACC. Stampede3 helped his team run high-frequency elastic wave propagation simulations for many different scenarios of magma and bubbles within the pore space of an elastic medium. 

"The use of Stampede3 helped us avoid the need to make less accurate assumptions like ray theory for evaluating seismic reflection properties. ACCESS made it easy to get an allocation and identify that Stampede3 was a good fit for the project's computing needs," Schmandt said.

The high bandwidth memory-enabled nodes on the supercomputer worked well for the team's seismic waveform studies. "We are using Stampede3 for ongoing work on 3-D simulations for magma reservoirs that would be difficult to try otherwise," Schmandt added.

As magma rises and decompresses in volcanic systems, gases like water and carbon dioxide exsolve from the melt, forming bubbles. In some cases, these bubbles can accumulate, increasing buoyancy and potentially driving explosive eruptions.

But present conditions at Yellowstone appear to tell a different story.

“Although we detected a volatile-rich layer, its bubble and melt contents are below the levels typically associated with imminent eruption,” Schmandt said. “Instead, it looks like the system is efficiently venting gas through cracks and channels between mineral crystals, which makes sense to me given Yellowstone’s abundant hydrothermal features emitting magmatic gases.”

Schmandt likened the system to “steady breathing” with bubbles rising and releasing through the porous rock—a natural pressure-release valve that lowers eruption risk.

Getting these results was anything but easy. The research team not only completed the field survey during the COVID-19 pandemic, but they also had to coordinate the project within a busy and carefully protected national park. This meant they could only operate the heavy vibroseis truck at night and only from designated roadside turnouts. 

More than 600 seismometers were temporarily deployed to record the vibroseis truck signals, then recovered a few weeks later. Collaboration with University of Utah professor Jamie Farrell, a Yellowstone geophysics expert and seismic network operator, was essential to making this unusual survey possible, Schmandt said.

Processing the data proved just as difficult. Yellowstone’s complex geology—known for scattering seismic waves—produced noisy data that were initially hard to interpret. But with persistence and many discussions with Schmandt, Duan said he kept going, refining his approach again and again until the numbers finally told a clear story.

“The challenge was that the raw data made it almost impossible to visualize any reflection signals,” Duan said. “We used the STA/LTA function to enhance coherent seismic reflections, and this was the first time we had innovatively applied STA/LTA data within the wave-equation imaging algorithm.”

Duan said that just like traversing the rocky landscape of Yellowstone, tenacity is key for navigating its mysteries underground.

“When you see noisy, challenging data, don’t give up,” Duan said. “After we realized the standard processing was not going to work, that’s when we got creative and adapted our approach.”

Even sensitive field observations from hundreds of seismometers are not enough to give scientists answers to questions about the subsurface. "We need to compare our observations to model predictions to understand them," Schmandt said. "Numerical wave propagation codes on high performance computing systems are the main way we can predict seismic data in a physically accurate manner."

By identifying this sharp, volatile-rich cap beneath Yellowstone, Schmandt’s team has established a new benchmark for monitoring the volcano’s activity. Future research could attempt to detect any shifts in melt content or gas accumulation that may serve as early warning signs of unrest.

Beyond Yellowstone, the study offers broader insights into onshore subsurface imaging with potential applications not only for volcano monitoring but also for carbon storage, energy exploration and hazard assessment.

“Being able to image what’s happening underground is important for everything from geothermal energy to storing carbon dioxide,” Schmandt said. “This work shows that with creativity and perseverance, we can see through complicated data and reveal what’s happening beneath our feet.”

Adapted from a Rice press release by Alexandra Becker.