What’s underneath Hawaii’s volcanoes just got a whole lot clearer

Hawaii’s volcanoes have been erupting a lot lately. Kīlauea, the Big Island’s most active volcano, has been oozing lava off and on since September 2021. That activity follows the volcano’s largest eruption of the past two centuries in 2018. Thirty-five kilometers away, Mauna Loa began erupting in November 2022 for the first time in 38 years.

With five active volcanoes of various ages, Hawaii experiences frequent eruptions and earthquakes, and its volcanoes are among the most intensely studied in the world. The Hawaiian Volcano Observatory, operated by the US Geological Survey (USGS), comprises 200 sensors on the Big Island transmitting data around the clock that are analyzed by a team of 30. But despite the wealth of observations, seismologists and volcanologists have still been debating some fundamental properties of the hot spot, particularly how and through what geologic structures magma is transported from the mantle.

It has been hypothesized that a mantle fault zone acts as a conduit to the surface for the magma. The zone would probably need to have many planar discontinuities along which masses of rock are displaced. The collection of discontinuities would likely be sufficient to form a mechanically weak zone through which magma could move toward the surface.

Now a team from Caltech has found evidence consistent with the fault-zone hypothesis. PhD student John Wilding, postdoc Weiqiang Zhu, and their advisers Zachary Ross and Jennifer Jackson analyzed data from nearly half a million earthquake events previously collected from dozens of instruments over three and a half years. In a new paper , they describe a deep, widespread plumbing system underneath Hawaii that links Kīlauea and Mauna Loa.

Seismic picking

One way to see the geologic structures below Earth’s surface is from seismic signals. The measurements collected from the Hawaiian Volcano Observatory’s seismographs provide a continuous record of ground motion. Researchers identify earthquakes in a seismogram by identifying the time at which a location first records P (or primary) waves, the fastest of the types of seismic waves produced by an earthquake, and S (or shear) waves. The seismogram below shows the magnitude of ground motion associated with the arrival of different seismic waves after an event.

The data set of seismic waves analyzed by Wilding and colleagues were recorded by the Hawaiian Volcano Observatory Network between November 2018 and May 2022. During that time, in 2019, a sudden swarm of earthquakes occurred near Kīlauea at a depth of 30–35 km, which is consistent with that of the hypothesized fault zone.

Rather than handpicking each of the millions of earthquake arrivals in the catalog, the researchers, led by Zhu, used a deep-learning model to quickly and accurately do the job. Like other AI algorithms, it was trained with a large data set—in this case, from 700 000 earthquakes in California—that did not include the seismic measurements analyzed in the study. The model picks the arrival of P and S waves that herald the start of an earthquake with an accuracy comparable to that of a human analyst. From that model run, the team amassed an earthquake catalog of 480 000 events across Hawaii, shown on the map below.

Picking the P-wave and S-wave arrivals provides only the time at which the seismic energy reached the detecting instrument, not the source of those waves. To pinpoint where each earthquake started, Wilding and his colleagues used a second algorithm to calculate the phase travel times of the seismic waves. The result is a distribution of earthquake locations. The uncertainty analysis showed that the origins of more than 90% of the earthquakes could be located with a vertical and horizontal precision of at most 2 km.

Pāhala swarm

The largest fraction of the analyzed earthquakes originated 36–51 km below Pāhala, a town in the southeast region of the island of Hawaii. Wilding and his colleagues determined that most of the earthquakes were associated with what they call a sill complex—a network of horizontal sheets of magma that have intruded between older, existing rock layers. The mechanism by which migrating magma triggers the earthquakes remains an open question.

The figure below shows the various earthquake clusters that illuminate the sill complex in cross-section view. (This animation shows the three-dimensional picture.) When looking at the regional scale, the clusters link several of Hawaii’s volcanoes, including Mauna Loa and Kīlauea, to form an interconnected web of seismic activity.

Previous research had suggested a possible transport path for magma beneath the volcanoes, given their sometimes synchronous activity, but the new paper’s results are the first to conclusively identify a continuous rock structure akin to the hypothesized fault zone that connects Kīlauea and Mauna Loa to the 2019 Pāhala swarm. Christina Neal, director of the USGS Volcano Science Center, says the work “is like putting on new glasses to see the architecture of possible magmatic pathways.”

With the plumbing of the sill complex in better view, old coincident events like the Pāhala swarm start to make more sense. The reanalyzed catalog includes a week in July 2019 of long-period earthquakes 30 km beneath the summit of Kīlauea; before and after that, the rate of events surged beneath Pāhala, 25 km to the west. “We think we are observing the resulting magma unrest through all these distal, contemporaneous swarms,” says Wilding. The rate of earthquakes in the eastern rift zone of Kīlauea, for example, increased from an average of about 15 events per week to as many as 400 per week during a major eruption in December 2020. After that eruption, the seismicity around Mauna Loa also steadily increased over several months.

That so much new information was found beneath one of the most-studied volcanic regions in the world suggests that other old seismic data could yield additional discoveries. In another recent paper that used a novel computational technique to analyze old data, researchers studying the active Yellowstone supervolcano found a meager amount of melted rock at depth, not currently enough to fuel a massive eruption. Provided there’s enough instrumentation across other volcanically active regions, it may be possible to identify new magmatic structures in those locations too.