A drone hovering over molten magma

Volcano secrets unlocked with robots and drones

Exactly 200 years after the biggest recorded volcanic eruption in history, scientists are using robots and UAVs to unlock the secrets of today’s volcanoes.

Two hundred years ago this year, Mount Tambora erupted on Sumbawa, a remote island in the south of Indonesia. The eruption began on 5 April 1815 and reached its climax five days later. A series of smaller steam-driven eruptions continued for the next three years as magma heated ground and surface water.The island lost all its vegetation, most of its animal life and around 10,000 people. Many tens of thousands more perished in the famine and disease epidemics that hit the surrounding islands in the aftermath. No one knows the exact number of deaths, but it is thought to be between 50,000 and 100,000.

This was the largest eruption in recorded history. It was a ‘7’ on the Volcanic Explosivity Index, one place below the apocalyptic Yellowstone eruption of 640,000 BC, with four times as much energy as the 1883 Krakatoa explosion. The Tambora eruption shot 150 cubic kilometres of rock, dust and gas into the atmosphere. It caused tsunamis locally and affected the weather as far away as London and New York. When it finally stopped, one-third of the great volcano that once stood 4,440m had become rubble and ash.

Mount Tambora is still active; there have been two small, non-explosive eruptions in 1967 and 2011, and Indonesia’s Geological Disaster Mitigation and Volcanology Centre monitors the volcano.

Monitoring a volcano is one thing. Predicting when an eruption might happen in time to evacuate the island of its 1.3 million inhabitants is more difficult.

“We are better at predicting volcano eruptions than we are predicting earthquakes, but not as good as we are at predicting the weather,” says volcanologist Clive Oppenheimer from Cambridge University. “You can’t stop a volcano going off, but if you know more about what’s going on under the ground you can take measures that reduce the impact.”

According to the US Geological Survey there are about 1,500 potentially active volcanoes worldwide, many of them along the Pacific Rim in what is known as the Ring of Fire. Around 500 of these have erupted in historical time. The biggest recent eruption took place in Chile in June 2011. The Puyehue-Cordón Caulle ash cloud reached over 3,600m and 353 people died.

Fluid dynamics

Volcano monitoring experts start from the assumption that magma, which is molten rock beneath the volcano, will move before it erupts, and this will give off signs which are detectable. Rising magma causes earthquakes as it ascends through cracks in the Earth’s crust. Temperatures around the volcano increase and the volcano starts to release gas with higher sulphur content. Changes to the surface of a volcano, known as volcano deformation, can also provide clues about what is happening deep below the surface.

However, according to Clive Oppenheimer, it’s not always easy to distinguish between a volcano that’s about to erupt and one that’s just a bit restless. “The signs are the same, either way,” he says. “What you think are signs that the volcano is coming to life might just mean that magma is moving around underground. Most times the rock will just freeze in place and not make it to the surface; 90 per cent of the time that’s what will happen.”

Getting at the hidden data

To predict with any sort of certainty when a volcano might erupt, scientists need more precise data about what goes on underground, in deep crevices, in some of the most inhospitable terrain on the planet.

“Only when we improve our understanding of how volcanoes erupt - and volcanoes can erupt in many different ways - can we start to ask if there are events, signals or signs that we can monitor, which we do not already observe,” says Nasa’s Carolyn Parcheta.

Parcheta and her colleagues at Nasa’s Jet Propulsion Lab have designed wall-climbing robots that can descend into the fissures in volcanoes - ground cracks from which the magma erupts - to look for the information.

Parcheta says that Nasa’s VolcanoBot is the first robot to go inside fissures and map the geometry to centimetre-accuracy. “The fissures have been made by magma in the past, but presently lie cold and dormant,” she says. “Analysing them helps us understand how the magma behaves.”

VolcanoBot contains a temperature sensor, a short-range distance sensor that helps it brake, an accelerometer and wheels embedded with microspines to gain traction against rock. Previous volcano robots were larger and heavier, were designed to walk on the surface, or were used as aerial platforms that document the volcanoes from above.

“Other monitoring technology that documents underground structures exists, but the resolution is too coarse to see the fissure conduits in detail,” she says. “We also tried to put lidar into these cracks, but didn’t get far because lidar works on a line-of-sight principle. It hits a rock and bounces back and can’t see around even a small overhang. The new robot can get past overhangs and see further into the cracks.”

Others are trying something similar. Professor Keiji Nagatani from Tohoku University in Japan has spent the last ten years developing robots for use in monitoring volcanoes. Nagatani calls them Clover robots. He tested them on Mount Asama last year. An unmanned aerial vehicle (UAV) drops the robot close to the volcano and it moves into place, controlled by an operator using GPS 3km away. They put a second robot closer to the first to act as a signal relay if the first robot moves to where there is no 3G signal. The robots have instruments to collect rock and dust samples.

In 2014, filmmaker Sam Cossman sent in UAVs to take close-up footage of Marum Crater, a volcano in the Pacific Ocean that was actually spewing lava. The drones perished in the 2,000°C heat, but not before the attached GoPro cameras took and sent back thousands of photos of the crater. “You’re monitoring a massive area, many kilometres across with sheer vertical drops, loose rubble and rock falls,” Cossman says. “We use drones to find the right route and to monitor the direction of the plumes. The drone can then take photos from the unaffected areas as the plumes shift. Usually, visibility is minimal. With a drone you are more likely to get a gasless view.”

Cossman is currently making a 3D reconstruction of Marum’s lava lake with these images. He thinks video can be a valuable addition to existing volcano monitoring research. “If you can work out the crater’s volumetric mass, you can start to understand how much energy is needed for an eruption,” he says.

Scientists have tried to work out changes in the release of carbon dioxide and sulphur dioxide from a volcano. They believe this will give an early indication of any unrest beneath the surface.

Active volcanoes release gas with high sulphur content through hot fumaroles, active vents and porous ground surfaces. The gases escape as magma rises toward the surface when it erupts and even as it cools and crystallises below ground.

“We measure gas because gases come from magmas,” says Professor Andrew McGonigle from Sheffield University. “A volcano’s explosivity is down to what extent the gas is trapped in magmas. By measuring gases, we gain an idea of how potentially blocked up the volcano might get and whether it’s likely to explode.”

Predicting an explosion

McGonigle adds that although volcanologists have been making gas measurements for a long time, previous techniques have only allowed them to be taken once a day at most, or just before and after an eruption. This, he argues, doesn’t provide enough data for experts to understand what is going on with a potentially erupting volcano.

“When a volcano explodes, you get a huge spike every ten minutes or so,” he says. “Gas release only lasts for a few tens of seconds. To get a real handle on volcanoes’ gas loss characteristics you need to measure every second.”

McGonigle uses digital cameras that can sense UV radiation to monitor volcanic gas. He explains that volcanic gases inside craters absorb in the UV. “We end up with a camera image from which we can work out how much gas is being released from the volcano,” he says. A small UAV with sensors flies into the gas stream to work out the chemical composition of the gas. “There’s an electro-chemical sensor that measures sulphur dioxide and an infrared analyser that measures carbon dioxide,” McGonigle adds.

Taking readings at this resolution enables McGonigle to look at trends in gas release. “We can understand explosions in a way not possible before, because we get a picture of gas release through explosive episodes,” he says. “Previously you’d take one measurement before the eruption and another after.”

From this McGonigle is also hoping to develop models that explain what is happening to the fluid dynamics within the magma column when the volcano is not exploding and when it’s about to explode. “We want to characterise when a volcano switches from periodic gas release to potential pressurisation and then eruption,” he says.

Early stages

This application of the technology is new, even if the technology itself has been around for some time. Sheffield’s UAVs and sensors are available commercially, as are the cameras, which are adapted from cameras designed for the amateur astronomy market. “There aren’t a lot of volcanologists, so you don’t often get technology designed specifically for the volcanology market,” he says. “Our work is for research purposes, to show others what it’s possible to do.”

Keiji Nagatani is in the process of upgrading his Clover robots. Sam Cossman wants to take his drones and cameras to make 3D videos of other volcanoes. Carolyn Parcheta’s robots are also at prototype stage. She’s used them to create a 3D map of a fissure in Kilauea, an active Hawaiian volcano, and is currently designing a smaller robot that can store data on board and therefore get deeper into a fissure. Nasa hopes that it will one day be able to use robots like these on the Moon or Mars. “We’re still trying to understand what it all means,” McGonigle says. “We’re collecting data during eruptions to capture traces that we can look for the next time there’s an eruption.”

Clive Oppenheimer thinks it will be a while before robots and drones are widely used in volcanoes around the world, though. He remembers using a robot to explore a volcano in Antarctica way back in the 1980s. The robot had only gone three metres when the cable snapped. It turns out it was too cold. “Volcanoes are rough and dirty places,” he says. “You need robust equipment that can run off a car battery. The technology must be smaller and rugged.”

Oppenheimer adds that predicting when earthquakes might happen is important, but that it’s also important to find better ways of monitoring what goes on after the volcano has erupted. “Think about where the ash cloud is going, whether it will affect aviation, and who will get a lot of fallout,” he says. “It happens over too large an area for ground observations; sometimes the cloud spreads out over 1,000km or more. We need satellites to monitor this.”

He would also like to see the world more prepared for the very rare and very large eruptions. “We know from the fossil records that there have been eruptions a hundred times bigger than anything we’ve seen, even Mount Tambora,” he says. Scientists are investigating the sites of some of these super-volcano eruptions to see if they can find clues to how they work.

Oppenheimer isn’t sure whether the threat of volcanoes is taken seriously enough. “Eruptions tend to focus interest and funding follows.” Examples are the Mount St Helens eruption in 1980, Montserrat in 1997 and the 2010 Iceland eruption that affected aviation in the northern hemisphere. “Focus tends to move away very quickly on to other things though,” he adds.

He says it’s extremely unusual for a volcano to erupt without the people who monitor it seeing something. As long as the people are there, that is. “The problem is most volcanoes in the world aren’t monitored,” he says. “Many erupt without warning, simply because people weren’t looking.”

Picture credits: Sam Cossman

Recent articles

Info Message

Our sites use cookies to support some functionality, and to collect anonymous user data.

Learn more about IET cookies and how to control them