A new seismic tomography model using a cutting-edge imaging approach that harnesses supercomputing provides an updated view of Yellowstone’s magma reservoir system
Yellowstone’s magma reservoir comes into sharper focus
Yellowstone Caldera Chronicles is a weekly column written by scientists and collaborators of the Yellowstone Volcano Observatory. This week's contribution is from Ross Maguire, Assistant Professor in the Department of Geology at the University of Illinois, and Brandon Schmandt, Associate Professor in the Department of Earth and Planetary Science and the University of New Mexico.
Seismic tomography, analogous to CT scanning in medical imaging, allows us to make geological inferences from 3D variations of seismic wave speed in the subsurface. This technique has been commonly applied to investigate magmatic systems because low-wave-speed anomalies may indicate regions of partial melt in the crust and mantle. But how much magma is in the subsurface? Is it concentrated in melt-rich structures such as sills (tabular horizontal magma bodies), or broadly distributed in a “crystal mush”? What are the implications for the eruptive life cycle of a volcano?
At Yellowstone, numerous studies have tried to address these questions using high-resolution images of the subsurface developed using seismic tomography. Although differences exist between studies, a low-wave-speed region at middle-to-upper crustal depths (5–15 km, or 3–9 mi) is thought to represent Yellowstone’s silicic magma reservoir. Typically, this anomaly is associated with seismic wave speed reductions of less than 10% compared with the surrounding crust, which suggests that the magma reservoir is composed of around 10% melt.
Thus far, most Yellowstone tomography studies have been based on minimizing the difference between observed and predicted travel times of seismic waves using a theoretical approximation. In this approach, it is assumed that the travel time of a seismic wave is sensitive only to the wave speed along an infinitely narrow path connecting the source and receiver. In reality, however, seismic waves have a finite wavelength and are sensitive to the wave speed in a volume around the path, as opposed to a narrow region. Additionally, the method does not accurately account for the effects of the subtle deflection of seismic waves that is caused by low-wave-speed bodies like magma reservoirs. This characteristic can mask the seismic signature of magma reservoirs, making them difficult to accurately image with this form of seismic tomography.
For decades, seismologists have had their sights set on using more accurate wave-propagation physics in tomography, although this has been challenging because it typically requires computationally intensive numerical methods to accurately simulate 3D wave propagation in complex Earth models. One method for addressing this challenge is to accurately simulate synthetic seismograms, which could allow for development of a tomography model capable of matching every wiggle in the observed seismic waveforms. In other words, it can take further advantage of the richness of information contained in seismic waves. Thus, the technique is often referred to as “full waveform” tomography, although in practice these models typically still rely on fitting specific types of seismic waves. Only with relatively recent advances in supercomputing has full-waveform tomography been possible for regional or global scale applications, but applications of full-waveform tomography to imaging volcanic systems have been few.
In a research paper published on December 2, 2022, in the journal Science, the full-waveform tomography method was applied to create a new seismic wave speed model of Yellowstone’s crustal magmatic system. The model utilized over 20 years of broadband seismic data from both temporary and permanent seismic networks deployed in Yellowstone and the surrounding region. The new images show a stronger seismic wave speed anomaly associated with Yellowstone’s silicic magma reservoir, with wave speed reductions of greater than 30% compared to the surrounding crust. Additionally, compared to previous tomographic images, the peak anomaly appears to be slightly shallower, at roughly 5 km (3 mi) depth (compared to 7–10 km, or 4-6 mi, depth from past studies). The shallower anomaly depth overlaps with estimates of the storage depths of previously erupted Yellowstone rhyolites derived from studies of rock and crystal chemistry. Based on the minimum seismic wave speed in the Yellowstone anomaly, the peak melt fraction within the magmatic system appears to be 16–20 %, assuming that melt is uniformly distributed at mineral grain boundaries.
The new results do not show new accumulation of magma, nor do they signal that magma is on the move or that the magma chamber has changed in recent years. Rather, through using a modern imaging approach, it is now possible to see a sharper picture of what was already there—rather like getting a better lens or imaging sensor for a camera, and thus being able to take a higher-resolution picture. The fundamental view of the Yellowstone magma reservoir system has not changed—it is a mostly solid crystalline mush, and shows no indication of preparing for any sort of eruption. But our view of this system is coming into increasingly better focus thanks to improvements in computing capabilities, theoretical models, and data.
As supercomputers continue to grow more powerful, seismologists will be able to tackle more ambitious problems with full-waveform tomography, which promises to continue to improve our understanding of magmatic systems at Yellowstone and beyond. This will be a particularly powerful approach when combined with dense deployments of seismic instruments that dramatically improve data coverage—an approach that has also been attempted at Yellowstone, with data that seismologists are currently examining—so stay tuned!