Tens of thousands of people have been killed in volcanic eruptions over the past few years. To mitigate future disasters Hazel Rymer says we need to understand and recognise the signs of an impending eruption. It is not only the people living on the flanks of active volcanoes who are affected by eruptions. In 1815, Tambora volcano in Indonesia erupted and poured 50 cubic kilometres of lava on to the ground (enough to cover the whole of the United Kingdom to a depth of 20 centimetres).

In just 36 hours it also injected ash and gases 40 kilometres high into the stratosphere. As the ash was dispersed around the earth, it shaded the planet from the sun and more than 92,000 people died from famine resulting from crop failures in the "year without a summer". Disease and economic ruin followed this event as far away from the volcano as India and Europe. The worldwide darkness was at best depressing and is thought to have inspired Mary Shelley's novel Frankenstein. More recently, other eruptions (for example, Pinatubo, 1991) have been found to reduce global temperatures and these affect global climate and the ability of farmers to grow their crops. This in turn affects the global economy.

What are volcanoes anyway? Peter Francis in his recent book Volcanoes defined them as "the manifestation at the surface of a planet or satellite of internal processes through the emission at the surface of solid, liquid, or gaseous products." This definition covers the eruptions of sulphur, SO2, ammonia, water and other substances at non-terrestrial volcanoes as well as the more familiar rocky volcanoes here on Earth. Even on the earth though, volcanoes take on a huge variety of forms and eruptive characteristics. Volcanoes are a window into the internal working of the planet. They are the most efficient way in which the earth can lose its heat energy, and on earth they erupt mainly silica based rocks and gases such as steam, CO2, SO2, and others.

The theory of plate tectonics can be used to explain the geographical distribution of volcanoes. Broadly, the surface of the Earth is divided into seven large and several smaller regions, or plates. These plates move about relative to each other and at their edges where they are in contact with one another, they can move apart from, collide with, or move past each other. Plates form the outer "rocky" shell of the Earth and extend from the surface to depths of up to about 125 kilometres. Volcanoes form the link between the surface and the hot convecting material beneath the plates, and are found mostly at the edges of plates.

³ÉÈËVRÊÓƵ

Volcanoes can be classified simply by the type of plate boundary on which they occur, as this dictates the chemistry of the rocks erupted and therefore the characteristics of the eruptions. For example, Iceland lies astride the mid-Atlantic ridge, a chain of submarine volcanoes erupting at the diverging boundary between the North American and Eurasian plates. Most Icelandic volcanoes erupt basaltic lava. This is erupted at 800 to 1,000 degrees centigrade and can form spectacular rivers of fast flowing molten rock. A small change in the gas content of the rock erupted makes it much more viscous, and rather than a river of fire, a thick oozing wall of clinker is developed with a red, molten interior. At a distractive plate boundary, where the more heavy, dense plate is forced under the lighter plate, explosive volcanoes are developed on the overriding plate. These structures typify the western coast of the Americas and the rest of the "Pacific rim of fire."

Having defined volcanoes and their eruptions, the question "can we predict eruptions?" becomes more difficult to answer. Do we mean by an eruption a single explosion, or a continual oozing of molten material taking place over months, or even years? There are 500 to 600 active volcanoes in the world, and every year about 60 may erupt in some way or another. The vast majority are not monitored in any way at all. Volcanoes such as Kilauea on the big island of Hawaii, though, are relatively well understood. It is located not on a plate boundary, but in the middle of the Pacific plate, over a "hot spot" or "plume". A hole has been "burnt" through the plate and formed a chain of volcanoes as the plate has moved due to continental drift over the stationary plume. Scientists have measured everything from thermal heat output, to ground deformation, gravity and even electrical resistivity at the volcano to seek out precursors to eruptions. This "drive-in" volcano has been very receptive to such measurements, as it is almost continuously active with lava flows erupting from its summit and flanks. It has been possible to predict accurately in terms of time and location the occurence of a new lava flow, but this activity is not typical of the more dangerous, explosive eruptions.

³ÉÈËVRÊÓƵ

Explosive eruptions of incandescent ash and blocks, and floods caused by remobilization of mud and snow melt, are hazards at volcanoes on destructive plate boundaries. These volcanoes are steeper, more rugged and inaccessible than their inter-plate or constructive plate boundary counterparts. The vast majority of volcanoes on land are of this explosive type. (There are many more volcanoes erupting continuously under the sea, but they pose no immediate hazard). Most volcanoes are not monitored in any way at all, and it is the unwatched, explosive ones that pose the greatest threat. Before an eruption, magma (molten rock) rises up beneath the edifice, and is often stored there, or within the volcano itself before escaping at the summit or flank. In the past, there was no reliable method for detecting this rise of magma, unless boiling mud pools started to appear at the surface, or the ground moved dramatically. Ground deformation is a classic precursor to volcanic activity, although it is by no means fool proof. Deformation can be measured using various surveying methods, and the new Global Positioning System (GPS) using satellite receivers placed at strategic places over a volcano can detect movements of a centimetre or more. Satellites can gather thermal images of volcanoes, so that when an eruption has taken place its progress can be monitored.To predict activity though, methods that detect underground magma movements or other precursors are needed. Small changes in the acceleration due to gravity at the surface occur before eruptions, also the level of seismicity (small local earthquakes), the magnetic field and resistivity of the ground show variations. If there were enough instruments to monitor a particular volcano comprehensively, then an impending eruption could be forecast. However, knowing that an eruption will take place sometime between now and in 10, 20 even 100 years time is not helpful to local inhabitants.

A collaborative project between the Volcano Geophysics Group at the Open University, the National Group for Volcanology (GNV, Italy) and others has enabled several continuously recording instruments to be set up on the edifice of Mount Etna, Europe's most active volcano. Previous work during the 1991/93 eruption showed the sorts of geophysical precursors to expect (gravity change, seismicity), but the rate of magma rise has not been evaluated before due to a lack of continuous date. By correlating the rate of change of geophysical parameters with the rate of magmatic and volcanic processes, we hope to develop a reliable system for eruption prediction. The aim of this ambitious project will be the facility to "dial up" any of the geophysical instruments deployed on the volcano through the Internet so that real-time evaluation of changes in the internal plumbing system of the volcano can be made from the safety of an office.

This is particularly relevant given the recent deaths of volcanologists "in the line of duty". In order to make measurements at active volcanoes, it is necessary to take some calculated risks. The use of high tech communications and remotely operated equipment should reduce the amount of time that volcanologists need to spend in hazardous areas making measurements in the future. It must be emphasised that volcanologists are not "fool hardy" or irresponsible, rather they are trying to do their job in the only way possible at present.

We cannot prevent volcanic eruptions from happening, but by trying to understand how volcanoes work, we are moving towards a better chance of anticipating important eruptions and so mitigating some of the disastrous effects of volcanic activity.

³ÉÈËVRÊÓƵ

Hazel Rymer is a Royal Society university research fellow based at the Open University.