The tricky science of tracking and predicting volcanic eruptions

It was just after 3pm on 13 November 1985 when the Colombian volcano Nevado del Ruiz erupted. Within minutes, four deadly rivers of clay, ice and molten rock raced down its flanks, destroying towns and villages. 

More than 23,000 people died, making it the second deadliest volcanic disaster in the 20th century – outranked only by the 1902 eruption of Mount Pelée in the Caribbean, which killed 30,000. 

There had been mini-eruptions and earthquakes in the run-up to the Colombian event, but while scientists noted such rumblings, they had no way of knowing whether they were just minor tantrums or harbingers of something worse.

Since then, the science of eruption forecasting has come a long way. In 1991, 75,000 people were evacuated prior to the massive explosion of the Mount Pinatubo on the Philippine island of Luzon. In 2010, 70,000 were moved out of harm’s way before Indonesia’s Mount Merapi erupted. 

That’s not to say forecasting has become infallible. In 2014, Mount Ontake in Japan erupted unexpectedly, killing 57 people. And in many areas people live in the shadows of dangerous volcanoes that are not monitored at all. 

{%recommended 4475%}

But new methods of remote forecasting, combined with powerful computer models, promise to be a game changer. 

Around the world are an estimated 1,550 active volcanoes. Most signal their vitality with just an occasional rumble, around 20 are non-stop fumers that don’t erupt – and about 50 explode each year. A handful of these are big enough to cause problems. 

To forecast big blasts, scientists measure fumes emanating from within the rocks. Changes in the ratio of carbon dioxide to sulfur dioxide can be a clue to restlessness down below. When magma starts to move upwards, carbon dioxide, being less soluble, bubbles out first. It’s followed by a belch of sulfur dioxide as the magma nears the surface. 

For decades, the only way to measure these gases involved walking up the slopes towards the crater, or swooping past in an aircraft – both risky activities.

Since 2005, though, an international group of researchers has been developing instruments to monitor the target gases remotely and continuously. Known as the Network for Observation of Volcanic and Atmospheric Change (NOVAC), group members use portable low-cost spectrometers that analyse gas concentrations based on how sunlight is absorbed as it passes through the volcanic plume. 

A type of spectrometer measures changing levels of sulfur dioxide and can be installed kilometres downwind of active vents or on aircraft and satellites, allowing continuous monitoring. 

At present, some 35 volcanoes around the world are watched this way. 

Another type of spectrometer known as a multi-GAS analyser continuously measures the ratio of carbon dioxide to sulfur dioxide. It is installed right on the edge of volcanic craters, inside the plume.

In Costa Rica, these instruments are now successfully sniffing out tell-tale bad volcano breath, providing a valuable early warning service. Early in 2014, Maarten de Moor, from the country’s Volcanic and Seismic Observatory, installed gas sensors on Turrialba, a volcano that threatens the capital city of San José, which lies just 30 kilometres to its west. 

Around six months later, an eruption kicked off. Prior to each ejection, de Moor and his colleagues saw a sharp increase in the carbon-sulfur ratio.  “It is a really promising result and a huge step forward for eruption forecasting,” de Moor says.

So far, the activity of Turrialba has been small, but de Moor is worried. “The last large eruption on Turrialba was in 1864,” he says. “The ash deposits suggest that it started with small eruptions, like those we are seeing now.”

The little disturbances, he continued, gave way to an enormous outburst – dubbed “Strombolian” in the jargon of the discipline, a reference to an ultra-active volcano on the island of Stromboli, off the coast of Sicily in the Tyrrhenian Sea. Such a powerful eruption from Turrialba would devastate the surrounding terrain, potentially killing thousands and crippling Costa Rica’s economy.

The change to the carbon/sulfur ratio picked up by NOVAC’s spectrometers, though, turns out not to be a reliable early warning signal in every case. It was not recorded, for instance, on Turrialba’s neighbour, Poas, before its most recent eruption.

The explanation for its absence concerns the acidic lake that fills Poas’s crater. The lake absorbs sulfur dioxide while allowing carbon dioxide to pass through, resulting in a markedly different gas profile. As pressure built and an eruption became imminent, the lake became super-saturated with sulfur dioxide, meaning the excess gas passed through into the atmosphere. This produced a different, but equally telling, change in the ratio, a clear warning that trouble was afoot.

Deciphering this signal from Poás was a milestone, de Moor says, since many of the world’s most unpredictable and explosive volcanoes – including Nevado del Ruiz and Mount Ontake in Japan – have crater lakes. 

The two Costa Rican volcanoes underscore that “there is no one size fits all” eruption signal, he adds. 

The key to successful prediction is to combine gas and classic seismic monitoring, as well as deploying new techniques that reveal whether the volcano is actually swelling with magma. 

A satellite’s GPS can monitor the movement of a volcano’s surface. Volcanologist James Hickey at the University of Exeter in the UK used this type of data to generate a computer model of what was happening underneath Sakurajima, an active volcano in Kyushu, Japan. 

Sakurajima’s last major eruption took place in 1914, killing 58 people and causing a massive flood in the nearby seaside city of Kagoshima. Its magma chambers have been refilling since, causing minor eruptions virtually every day.

Hickey and his colleagues incorporated the area’s topography and underlying rock types into their model, along with very precise GPS measurements of surface movement, to gauge just how fast the magma was replenishing. Their results, published in Scientific Reports last September, indicate the tank needs roughly 130 years to fill. 

“In other words […] enough magma might be stored in the next 30 years for an eruption of the same scale as one in 1914,” Hickey says. 

That finding prompted the Kagoshima City Office to review its evacuation plans. Meanwhile, Hickey is developing similar models for volcanoes in Ecuador and the Lesser Antilles in the Caribbean. 

But even with all the high-tech advances, people in the poorest parts of the world are still at risk. Despite the efforts of NOVAC, right now there are still too few experts to analyse every volcano’s halitosis and generate the models that reveal what is going on deep underground. “We hope to interest more people in coming to do this kind of work,” de Moor says. 

In many countries with significant populations living in volcanic danger zones, there is barely any monitoring at all. Indonesia and the Philippines top the list for populations most at threat, according to a 2015 United Nations report.

But at least in Colombia, 30 years after the devastation, villagers living under the menacing shadow of Nevado de Ruiz are placing their hopes in science. 

Continuous gas monitoring instruments were installed in the volcano’s vent last year and scientists are schooling themselves in how to read warning signs. 

With luck, they’ll have sussed it out before she blows again.