For 600 years Mount Pinatubo slept. On 2 April it exploded and has been erupting ever since. As soon as the volcano showed signs of rousing, the evacuation began. Eventually some 64 000 Filipinos and thousands of Americans from nearby Subic Bay Naval Base and Clark Air Base left their homes.
Despite the ferocity of the eruption – thousands of tonnes of pumice and ash have been ejected – the death toll is around 300. From this point of view the eruption has been a ‘success’ for volcanologists. Without their warnings and hazard maps, many more people would have died. Scientists had been able to predict explosions and areas threatened by flows of mud, ash and lava.
Advertisement
Yet some aspects of the eruptions are still unpredictable; most of the 40 people killed when Fugen-dake, a peak on Mount Unzen in Japan erupted last month, died because a well-known phenomenon, an avalanche of ash, rock and gas, did not behave as scientists had expected it to.
One reason why predictions were so effective was the quick reaction of the Philippine Institute of Volcanology and Seismology. Pinatubo’s reawakening was announced in early April by an explosion northeast of the summit and the opening of new vents, or fumaroles. Within days, the institute had a seismometer on the mountain and 2000 people were evacuated.
The institute also invited a team of volcanologists from the US Geological Survey to set up equipment on the slopes of the volcano and around the crater’s edge. The team brought seismometers and tiltmeters among other instruments. Outwardly, it looked like a typical volcano watch, one of scores in progress around the world.
‘But it was different, because of the software,’ says Chris Newhall, the volcanologist who headed the USGS team. Seismic data from seven locations on Pinatubo were transmitted continuously to a base a few kilometres away. There, personal computers combined the data using a program developed by Willie Lee of USGS.
‘It was incredibly powerful,’ says Newhall. ‘In real time we could look at the spectral content (frequency range) of quakes, plot them, locate them, examine them in three dimensions. This kind of analysis previously would have taken much much longer.’ The long-distance telemetry had other benefits. ‘Without it,’ notes Newhall, ‘we might have had some dead colleagues.’
From this data, the scientists read patterns of seismic waves of low frequency, between 1 and 2 hertz. Quakes of this frequency warn of magma close to the surface and explosive eruptions, says Newhall.
The Americans also set up tiltmeters, which employ lasers to measure ground movements as small as a few centimetres. They detect inflation and deformation of the volcano’s surface. The team found that movement was originating between one and several kilometres below the surface and that it correlated closely with the source of the quakes – another sign that magma was rising ready for an eruption.
They also employed a technique relatively new to volcano watching; sulphur dioxide detectors. Known as correlation spectrometers, these were used like telescopes to observe ultraviolet radiation which measures levels of SO2 in gases emitted from vents.
In mid-May the volcano was emitting about 500 tonnes of SO2 daily. Two weeks later, emissions had grown 10 times – a sign that magma was on its way up. When magma rises, the pressure confining SO2 in the magma lessens and gas is then released, in the way carbon dioxide fizzes out of an open bottle of champagne.
However, the emissions then decreased, an event that was seen as even more ominous. By 4 June, the team were picking up only 280 tonnes of SO2 daily. They attributed this drop in emissions to a blockage, which suggested that pressure was building up underground. The team reckoned an explosive eruption was imminent.
On 5 June, the volcanologists issued a level-3 alert, warning of a large eruption within a fortnight. By 7 June, the seismic signals, tilting and explosions at the summit led them to upgrade this to level-4, an eruption possible within 24 hours. The major phase of eruption finally began on 12 June. The eruption has continued intermittently since then, and there are no signs that the danger is abating.
During May, the scientists had hurriedly prepared a map of the hazards expected from the volcano. Using geological clues left by previous eruptions and the topography of the peak, volcanologists are now able to predict what kind of material will be emitted (lava, rock fragments or ash) and where and how it will move.
Rock ejected from volcanoes varies in how far and fast it travels and in the danger it poses. For example, lava flows from volcanoes such as Pinatubo rarely travel more than a few kilometres downslope from vents and move at walking speed. Tephra is less predictable – rock fragments varying in size from ash to huge blocks are propelled into the air. Small fragments are carried by the wind and fall according to their size and weight. Ash kills directly by asphyxiation and indirectly, by accumulating on roofs in layers thick enough to make buildings collapse. Many of the deaths around Pinatubo came from collapsing buildings.
But these eruptions have more insidious and unpredictable elements. Both Pinatubo and Unzen exist because a slab of the Pacific plate, which forms the ocean floor, is sliding under the Philippines plate. This produces lavas containing a high proportion of silica, which makes them viscous. They can sustain high gas pressure before they explode, but when they ‘go’, they do so with great force. Because the magmas originate above the descending slab of ocean crust, they contain water and gases such as carbon dioxide and oxygen, that escape explosively as the molten rock nears the surface.
Such explosions produce a plume of ash which spreads into the stratosphere. Ash falling on the flanks of the volcano gives rise to one of the greatest hazards of these eruptions – lahars, which are also known as mudflows. Unstable slopes of fine mud, rain and shallow earthquakes can trigger mudflows that carry off anything in their way.
The slopes of Mount Unzen are known to be vulnerable to mudflows. Barriers have been built high on the volcano to lower the potential for damage downhill. But, as a safeguard, last month, vulnerable river valleys were evacuated when heavy rain made mudflows more likely.
On the hazard map made for Pinatubo, all the rivers leading from the volcano are identified as high risk. Lahars now pose the greatest threat, according to Ronaldo Arboledo, of the Philippines Institute. ‘We have already lost three seismographs,’ he said, ‘and nearly lost a field team when a flow chased them down the mountain.’ There is now an early warning system; civil defence teams watch the mudflows and report by radio to a control centre if areas ‘downstream’ need evacuating.
More unpredictable still are pyroclastic flows. These fast-moving avalanches of ash, pumice and fragments of rock, at temperatures up to 800 °C, were the main fear for volcanologists in the Philippines, and the main killer in Japan. They form in a variety of ways and move with such force that they can surge down one mountainside and up another. Predicting this behaviour reliably has proved impossible so far.
The pyroclastic flows from Unzen were triggered by the collapse of a dome of lava built up near the summit of the volcano. This dome, at least 44 metres high and 110 metres across, collapsed on 24 May with the first of the explosions and pyroclastic flows that menaced the town of Kamikobi.
Another pyroclastic flow on 3 June accounted for most of the deaths caused by the Unzen eruption. Part of the cloud of ash and debris seems to have branched off the main flow unexpectedly. It overwhelmed volcanologists who were watching it. This single flow killed, amongst others, Maurice and Katia Krafft, who made a career out of observing and filming volcanoes in action, and Harry Glicken, formerly field assistant to David Johnston, the geologist killed in the 1980 eruption of Mount St Helens.
The danger from pyroclastic flows is not limited to any particular phase in volcanic development. Laurence Fink, of the University of Arizona, describes one example of an eruption in Guatemala in 1973: ‘Lava had travelled about 2 kilometres when it began to move down a steep slope at the head of a valley. The front of the flow slipped off, exposing the hot interior which exploded to produce a pyroclastic flow.’
This belated explosion was perhaps triggered by the ground over which the lava flows. Fink identifies it as a type of hazard often neglected during an eruption; conventional hazard maps may not be directly helpful. ‘There can be no warning at all,’ he says. ‘Predicting this sort of collapse is a two step process, first predicting lavas and then figuring out where they might collapse.’
Despite their failings, hazard maps are invaluable when a volcano threatens. Thousands of people were evacuated around both volcanoes on the basis of warnings covering specific areas. These charts are becoming more sophisticated with each well-monitored eruption, and as new techniques identify the results of ancient eruptions.
One is palaeomagnetism, measuring the magnetic field remaining in the rocks from their formation, to distinguish between mudflows and pyroclastic avalanches. Large rocks, or clasts, embedded in finer material indicate that the rock formed in some sort of flow. If the magnetic orientation of the clasts match each other and the surrounding rock, they probably cooled at their final resting place; they were deposited while hot, in a pyroclastic flow. If the clasts’ magnetism is randomly aligned, they moved around after cooling – the mark of a mudflow.
Hazard maps have also profited from new computer models that use topography, the geological record and the state of an active volcano to predict its behaviour. Michael Sheridan, of the University of Buffalo in New York, applied such a model to predict the course of a pyroclastic eruption at Colima in Mexico, in April (This Week, 4 May).
Another new technique being tested at Colima, by Charles Connor, a geophysicist at Florida International University, measures the temperature of fumaroles. Connor says that as more gas spews from these openings, the temperature rises. One end of a 25-metre fracture at Colima has been fluctuating by 80 °C daily. He is trying to correlate this rhythm with seismic readings.
Hazard maps form one of the projects in the UN’s International Decade for Natural Disaster Reduction. In this research programme, scientists have identified risk mapping and ways to diminish effects, cheap monitoring and emergency response plans as important. Another crucial part of the emergency response is communication. The measures taken around Pinatubo were notable because they avoided mistakes that cost thousands of lives less than a decade ago.
In 1985, scientists drew up a hazard map for Nevado del Ruiz, a volcano in Columbia. It was delivered to towns near the volcano and even published in local newspapers. The map specifically warned of mudflows, but to no effect. When the volcano erupted mudflows killed more than 20 000 people.
One of the problems apparent in Colombia was that volcanologists cannot predict the exact timing of an eruption. Each volcano is different: the key to prediction lies in being on the spot early, to study its rhythm. This message is reinforced by John Guest, a volcanologist experienced in monitoring eruptions. ‘When a volcano erupts, you need to know what is normal,’ he says. ‘Seismic monitoring is important as is measuring ground deformation, but there is no point in doing either without baseline studies.’
A network of seismometers across the country can warn that a volcano is stirring, as happened for Unzen last year. But seismometers are not cheap, and accurate monitoring of the progress of an eruption demands a network of perhaps 20 seismometers. Guest estimates that a single array would cost hundreds of thousands of pounds.
This is fine for Japan, but a struggling, developing country like the Philippines was forced to rely on American money and researchers. Forming international pools of equipment and teams of volcanologists is one practical step that could help poorer nations.
With initiatives like this volcanic hazards can be predicted, planned for and avoided. And the planning must include education at all levels; countries at risk need trained staff, people living near volcanoes should understand the warnings offered, and local administrators know how to act on them. Only in this way, will we understand and mitigate the effects of these awesome natural hazards.