When a volcano erupts, the ground vibrates. Scientists who study volcanoes refer to the continuous ground motion as volcanic tremor. Unfortunately, the mechanism that causes these vibrations isn’t well understood. 

In a new paper published in Geochemistry, Geophysics, Geosystems, a team led by University of Potsdam doctoral student Alea Joachim analyzed seismic data from the 2021 Geldingadalir eruption in Iceland. They compared the data with images and videos captured via drones they flew above a churning lava lake. They found that peak tremor corresponded with bursting bubbles, which suggests that near surface processes — bubbling gas in the lava lake or spattering of the molten rock — may generate tremor at this volcano. 

A ruckus on the Reykjanes Peninsula  

The Reykjanes Peninsula juts from the southwestern side of Iceland into the Atlantic Ocean. The peninsula is the onshore extension of the Reykjanes Ridge, part of the mid-Atlantic ridge that pushes North America west and Eurasia east. As a result, volcanic activity and seismicity are part of life for denizens of this region. 

Here, several volcanic systems trend southeast-to-northwest. They are, from east to west, the Hengill, Brennisteinsfjöll, Krýsuvík, Fagradalsfjall, Svartsengi, and Reykjanes volcanic systems. These systems have been episodically active for thousands of years. Eruptive episodes typically last about 400 to 500 years. Intervals between episodes span about 1,000 years. Only one system appears to be active at any given time. 

The Fagradalsfjall system is young, and was only recently identified as distinct from the others. Its last eruption was more than 6,000 years ago. On the other hand, the most recent eruption on the Reykjanes Peninsula occurred 781 years ago. And so, the 2021 eruption may signal the start of a new and long-lived event. 

More than a year prior to the eruption, in December of 2019, seismicity increased in the vicinity and lasted for several weeks. Seismicity continued in January of 2020 at the Svartsengi volcanic system, located west of Fagradalsfjall, with the Svartsengi geothermal field inflating and deflating. 

On February 24, 2021, an intense earthquake swarm struck Geldingadalir Valley, with around 39,500 earthquakes — the largest, a magnitude 5.6 event. This signaled intrusion of a nine-kilometer-long dyke. It is above this dike that a fissure opened on March 19, 2021, thus beginning an eruption that lasted six months. The Fagradalsfjall system has since erupted two more times.

Tremor

Around the world, scientists use tremor for real-time monitoring of volcanoes. Upticks in tremor can signal an impending eruption. But interpreting tremor is rarely simple. 

Tremor can persist for minutes, days, months, or even years. It can occur before, during or after eruptions. Frequencies range from 0.5 to 10 Hertz, although some volcanos experience higher frequencies. Tremor can be either continuous or episodic. 

“Tremor generation is not yet fully understood,” Joachim says. Fountaining or ash plumes can generate tremor. Subsurface processes like bubbles coming together and growing can do so as well. If the heat from an intrusion causes hydrothermal boiling of shallow aquifers, tremor can result. Tremor is also observed at lava lakes and has been correlated with fluctuations in lake level — for instance, at Kı̄lauea in Hawai’i and Nyiragongo in the Democratic Republic of the Congo.

In Iceland, volcanic tremor began with the March 19 eruption, and continued unceasingly until May 1, 2021. Then, episodic tremor commenced, with a distinct start-and-stop pattern. A total of 8,698 of these tremor episodes were recorded. 

By May, the volcano’s vent in this study had grown large enough to house a lava lake — a remarkable volcanic phenomenon. Of the approximately 1,350 potentially active volcanoes worldwide (excluding undersea volcanoes associated with spreading ridges), only a handful have (or have recently had) long-term lava lakes — for instance, Kı̄lauea, Erta’ Ale in Ethiopia, Nyamuragira and Nyiragongo in the Democratic Republic of the Congo, Mount Erebus in Antarctica, and Villarrica in Chile. “Lava lakes are rare on Earth, so their formation depends on the plumbing system,” Joachim says. “There must be a constant magma supply.”

Joachim and colleagues took a two-pronged approach to investigating the Geldingadalir eruption and its unusual lava lake. They captured drone footage of the roiling lava, with a total of 947 photos and 18 videos recorded on June 8, 2021 over the course of two hours. They also explored seismic data at station MUKA, located 1.8 kilometers from the vent. 

A story in images

In the drone footage collected discontinuously over the two-hour study period, the team observed the lava lake rise and fall five times. In each of these episodes, the lake level increased and then overflowed, spilling lava over the crater rim. Then, the lake level fell and the cycle began anew. To quantify the changing lake levels, the team used photographs. (It’s likely there were more than five cycles, but the drones were not flying continuously for the entire two-hour study period.)

The team also used color, taken from the red channel of the drone video footage, as a way to track just how much of the lake’s surface either crusted over with a cool, dark gray layer or boiled in a blaze of orange and yellow and red. These changing colors serve as a proxy for temperature. “The brighter the color, the hotter the lava,” Joachim says. The higher the temperature, the more vigorous the lava lake. By tracking how the color changed through time, Joachim and her colleagues charted how the diameter of the region of roiling increased and decreased as the lake level rose and fell. 

Focusing on the fifth cycle, they found that at the start, the surface of the lake appeared flat and dark; the diameter of boiling was low. As the diameter of the region of bright, hot lava increased on the surface, it reached up to 30 meters across. Where the lava boiled, Joachim observed bursting bubbles and spattering lava flung from the vent. The rise in diameter was gradual, its decrease sharp. 

Notably, as the diameter decreased, the drone footage showed that spattering and bursting bubbles remained high. Even more curious: As the lake level dropped, diameter continued to increase before beginning its downward trend. 

Seismic analysis

To analyze the tremor, the team focused on only one of two nearby seismic stations. Station NUPH is unburied and experiences wind noise; these data were not used. Instead, they relied on data from station MUKA, which is buried, thus minimizing noise. These data are available through the NSF NGF data archive.

In the two hours of June 8, 2021 during which the drones captured images of the volcano, the seismic data recorded 12 cycles of episodic tremor. The team specified the start and stop of each tremor episode, and calculated their durations and repose times. Repose time is the difference between the end of one episode and the start of the next.

They found that the duration and repose times of each episode were slightly different. The duration of tremor episodes were, on average, about 5 minutes, whereas the repose times were about 7 minutes. A complete cycle of tremor, then, was about 12 minutes long.

Putting it all together

The first question to address is whether lake level modulates tremor. The peak of the lake level — when it reaches its highest elevation — always preceded the peak of volcanic tremor. This suggests that the answer is no, the rise and fall of the lava lake does not modulate tremor. Moreover, the authors observed that the lava lake consistently overflowed before peak tremor was recorded. 

Further supporting this finding is the shape of these curves. The lake level rose slowly and fell quickly. On the other hand, tremor either increased and decreased at similar rates or rose quickly and fell gradually, opposite to the behavior of lake level.

A detailed look at diameter of the boiling region derived from red-channel information tells us that diameter and tremor amplitude reached their maxima at about the same time. Interestingly, as diameter dropped, spattering and bubbles bursting remained high. 

“We tried to quantify [the bubble bursts], but it became impossible at a certain point,” Joachim says, noting that the sheer number of bubbles and how they overlapped stymied these efforts. “Nevertheless, we can say with certainty that the number of bubbles and the variety of bubble sizes increased with increasing lake level and activity.”

Bubbles point to process

When lava lakes rapidly rise and fall, this is known as “gas-pistoning.” When this process has been observed elsewhere, like at Kı̄lauea, small bubbles burst on the lake surface and increase in size and number to release gas. As the gas is released, the system drains and lava flows back into its conduit. Because the authors observed the infrequent rise of small bubbles during the slow upward rise of the lava lake, this supports the hypothesis of gas-pistoning during the 2021 Gedingadalir eruption. If this hypothesis is correct, this would be the first observation of gas-pistoning in an Icelandic eruption.

However, the authors noted that they still observed bubbles bursting, releasing their gas, not only at peak lake level, but also as the lake level dropped. In this way, bubbles can still explain the slow drop in tremor.

Though they cannot exclude tremor generated at depth, they note that because tremor amplitude is linked to the vigorously boiling area, “it seems more likely that tremor is generated by surface activity like bursting bubbles or spattering.” 

“Lava lake fluctuations do not directly cause tremor,” Joachim emphasizes. “Rather, it is more due to the degassing processes.”