The original version of this story appeared in Quanta Magazine.
In the summer of 1991, Pinatubo, a volcano in the Philippines, self-destructed. The eruption started on June 12, and three days later it culminated in a tremendous explosion. By the time pyroclastic flows—incandescent avalanches of molten rock and gas—tumbled down its sterilized slopes, Pinatubo’s peak had been obliterated and replaced by a 2.5-kilometer-wide chasm.
The eruption killed more than 800 people, mainly because roofs, weighed down by rain-saturated ash, collapsed. But it could have been so much worse: About 250,000 people, across multiple cities and a sprawling US Air Force base, lived in the volcano’s shadow. When Pinatubo started convulsing and belching steam in April of that year, scientists from the US and the Philippines deployed an array of instruments that tracked the volcano’s inner tumult.
“We didn’t know much about that volcano, and so there was this really rapid geological assessment. And the assessment said, ‘Oh, crap, when this thing erupts, it only erupts big,’” says Mike Poland, current scientist in charge at the US Geological Service’s Yellowstone Volcano Observatory. “And that became the basis for a forecast.”
By early June, ash and lava were escaping Pinatubo’s flanks, and an evacuation was ordered, just a few days before the cataclysmic hammer fell. It was, in other words, a very close call.
Those scientists saved countless lives, but their forecast was more of an educated guess than it might have appeared. It was nothing like a weather forecast; they couldn’t say with anything resembling certainty that on June 12 an explosive eruption was going to occur, nor could they predict the evolution of that eruption.
With very few exceptions, this imprecision is true of all well-monitored volcanoes. But volcanology, as a field, has made great leaps since Pinatubo blew its top. The instrumentation is more advanced, machine learning has made interpreting data far more efficient, and scientists have a much better understanding of the magmatic plumbing that drives volcanism. That’s prompted me—as a professionally trained volcanologist who now writes a lot about the field—to wonder: How close are we to forecasting volcano behavior the way we forecast the weather?
In this award-winning image, photographer Alberto Garcia captured a truck fleeing the cataclysmic eruption of Mount Pinatubo in the Philippines.
Today, we know that a storm of a certain magnitude will fall on a specific city in a few days’ time. Will scientists ever be able to say that a week from now, a certain volcano has an 80 percent chance of erupting in a particular way—with lava gushing, with a certain explosive force, with pyroclastic flows that will travel down its western flank? I asked around, and I found both skepticism and a surprising degree of optimism. “The short answer—otherwise I wouldn’t be doing this—is yes,” says Diana Roman, a volcanologist at Carnegie Science in Washington, DC.
Though sky watchers have anticipated the weather for millennia, contemporary scientific prediction of weather is a recent invention: The first mathematical equations grounding these models were derived at the start of the 20th century. Today, meteorologists can take a pandemoniac system—Earth’s atmosphere, oceans, and landforms—and make accurate forecasts up to two weeks into the future.
Weather affects more people than volcanism—namely, everyone, all the time—but some 800 million people live within 100 kilometers of an active volcano, and some (very rare) eruptions can also affect the entire planet. Both weather and volcanism are complex systems that we want to understand, but the problems they present for forecasting are different.
“The big difference between [volcanoes] and the weather forecasting is the weather is always happening,” says Jenni Barclay, a volcanologist at the University of Bristol in England. The atmosphere is perpetually visible and measurable to meteorologists. “Even they would say they need more observations.” Magma, on the other hand, resides kilometers below Earth’s crust, and at most, active volcanoes erupt once every few decades.
Each volcano is also unique. The architecture of the subterranean pathways that funnel magma to the surface, the chemistry of the magma, the cadence of eruptions, and the assortment of eruption styles differ from place to place. And eruptions don’t have just one trigger. The temperature and pressure of the magma reservoir, the weakness of the enclosing rock, the gas and crystal content, the depth of the magma, the regional motion of tectonic plates—these factors all contribute to whether a paroxysm happens or fizzles out.
“Geology is chaotic,” says Marius Isken, a geophysicist at the GFZ Helmholtz Center for Geosciences in Potsdam, Germany. But there is order buried in the chaos. Can we find it?
A volcanologist from the National Institute of Geophysics and Volcanology in Italy installs a gravitmetry station to measure magma movement within Sicily’s Mount Etna.
I imagine volcanoes as orchestras composed of hundreds of different instruments. Forecasting eruptions isn’t about hearing the music. We already do that: Seismometers sense the cracking of rock as magma ascends; ground sensors and satellites can track shifts in the crust, indicating where magma is flowing; gas detectors reveal when magma rises to shallow depths, depressurizes, and emits noxious fumes.
The challenge comes in knowing how the symphony will develop to a climax, long before it gets underway. Today, at the most comprehensively monitored volcanoes, the best that volcanologists can normally offer is not prediction but a form of acute caution. Often, alert systems—including those used by the US Geological Survey—notify the public if a volcano is exhibiting heightened or escalating unrest. But that doesn’t mean an eruption is imminent. “Only 50 percent of volcanic unrest that looks like it’s going to be an eruption ends up in an eruption,” says Jessica Johnson, a geophysicist at the University of East Anglia in England.
On the other hand, some volcanoes prefer to ambush us, even when smothered in instrumentation. Pockets of highly pressurized water trapped just below the surface can be heated by adjacent bodies of magma. If that pocket ruptures, a dangerous steam explosion follows, which can then unleash imprisoned magma. This type of eruption often occurs with no discernible warning signs, and it’s like a land mine going off next to a buried mountain of dynamite.
Small eruptions of the Soufrière Hills volcano on the Caribbean island of Montserrat preceded destructive large eruptions in 1997.
More predictive detail can come if a volcano has been studied over the course of several eruption cycles. At certain peaks, such as Italy’s Stromboli and Etna volcanoes, which regularly spout fountains of lava, scientists can confidently forecast an outburst. “We have systems that can tell us that in a few hours, the volcano will erupt,” says Maurizio Ripepe, a geophysicist at the University of Florence.
Using seismology and ground deformation measurements, scientists at other volcanoes, including Hawaii’s Kīlauea and those on Iceland’s Reykjanes Peninsula, can track magma migrating underground with such staggering precision that they know exactly where it will emerge as lava, to within an hour or so. But such precise forecasts are “relatively unusual,” says Tom Winder, a volcano seismologist at the University of Iceland. These are frequently active volcanoes, unlikely to produce a major explosive event, and people in surrounding communities generally know to be wary of them. In most other cases, the earliest warning times—perhaps an hour or so before the eruption—aren’t always enough to get people to safety.
Forecasting eruptions is a big ask, because volcanoes cannot be reduced to simple models. They’re baroque geologic beasts with hidden, labyrinthine plumbing. Twenty years ago, during my first year as a geoscience undergraduate, a lecturer told me that predicting when and where the next major eruption would take place was a pipe dream—the implication being that volcanoes are far too idiosyncratic and mercurial to have much in common with one another. That comment felt off even then. After all, they are all vessels of immense pressure and heat. Their schematics may differ. But molten rock flows through all of them, and eventually, something cracks, breaks, and explodes.
There were many signals that Mount St. Helens was going to erupt in 1980, but the form of the eruption was unexpected. The event led to the development of more sophisticated monitoring systems in the United States.
Everyone I spoke to agreed that scientists still need to crack a vital piece of the volcano forecasting puzzle. “We don’t even fully understand the underlying physics,” Roman says. What causes a magma reservoir to transition from a stable state to catastrophic failure?
“They have to have shared physics,” she says. If those underlying equations can be discovered, perhaps we can apply them to all volcanoes and output values that tell us, with high accuracy, when the next eruption is due and what its shape may be.
Scientists have identified some of these governing equations, but they only apply after eruptions have begun. Using more than a century of observations, researchers have largely derived the physics of volcanic hazards—particularly lava flows and pyroclastic flows. For examples, the Navier-Stokes equations, which describe how fluids of all kinds move, have been successfully applied to both of these hazards, while the heat equation reveals how and when these volcanic fluids cool down. Today, they allow experts to predict, for specific volcanoes, where outpourings will emerge, how far the different kinds of flows will reach, and how quickly it will all happen.
This work saves lives, but it’s a fraction of the forecasting dilemma. Using our weather analogy, this is like saying, “Once the rain starts to fall, we can forecast what watersheds might flood,” Poland says. Knowing when the storm will start requires getting at the subsurface physics of magma reservoirs.
For now, eruption warnings are based on recognizing patterns in measurable geophysical signals, such as an escalation of seismic activity, that precede eruptions. But correlation isn’t enough for prediction if the patterns aren’t consistent, which is often the case. “What we’re trying to do is looking at the causative relationships there … to understand the physics,” Johnson says. “If you understand what those patterns mean, [then] when those patterns change, we’re not that stuck.”
