Predicting a volcano's next move is a difficult task, particularly in a place like Canada where most of them are in remote and inhospitable places. Still, over the last century much has been learned about eruptions through careful research and monitoring on volcanoes throughout the world, and in particular on Hawaii, Mt. Pelee (West Indies), St. Vincent (West Indies), Mt. St. Helens, Mt. Pinatubo (Phillipines), Mt. Unzen (Japan), Montserrat (West Indies) and Mt. Etna (Italy). From these eruptions, scientists have developed two rules of thumb that are useful in eruption forecasting at any volcano. The first rule is that every volcano tends to have its own pattern of behaviour, in terms of its eruption style, magnitude and frequency, so that its next eruption is likely to be similar to its past eruptions. This means that by studying a volcano's past, we can start to predict its future. The second rule is that volcanoes normally do not erupt without providing weeks, months or years of warning signs. The most common warning signs are clusters of small, shallow (<15 km deep) earthquakes, most too small to be felt. This rule means that by monitoring volcanoes at least for earthquakes, we can determine how active their underground "plumbing system" is, and how close they are to erupting. In 1991, at Mount Pinatubo, Philippines, monitoring precursory activity provided enough warning of the eruption to save the lives of tens of thousands of people.

In Canada, even though some volcanoes pose a significant threat to local communities and any sizable eruption would affect the economy of western Canada, volcano monitoring and research is limited. Because no large eruptions have occurred in Canada in the last few hundred years and most of our volcanoes are currently in remote locations, volcano monitoring is a lower priority than dealing with the hazards of earthquakes, landslides and tsunamis. However, as for earthquakes, a future volcanic eruption in Canada is inevitable and it is very likely to have a serious impact on people. This impact is becoming ever more likely as our population increases and development spreads. For these reasons the Geological Survey of Canada, with much help from Canadian university scientists, has been gradually building a baseline of knowledge on the state of our volcanoes. In addition, we are continually improving our ability to monitor the volcanoes in order to forecast impending activity.

Over the last 50 years, scientists at the Geological Survey of Canada and at Canadian universities have documented the past behaviour of a number of Canadian volcanoes. We now have a robust database telling us where the volcanoes are, how they tend to erupt and which are most active. The Catalogue of Canadian Volcanoes lists a small portion of that database. Unfortunately we don't yet know enough about the frequency of eruptions to predict which volcanoes are most likely to erupt next, and what their likely impact will be. Work in this area is ongoing, and you can find out more in the section Volcanology in the GSC.

Currently no Canadian volcanoes are monitored sufficiently to allow us to determine how active their magma systems are. The network of seismographs, which exists to monitor and understand tectonic earthquakes, is too far from our volcanoes to give us an accurate picture of what is happening beneath them. If a Canadian volcano becomes highly restless, the existing seismic network will probably detect the increase in activity, but in most areas this detection will come in advance only for potentially large eruptions, or when the eruption is on the brink of starting -- or has already started.

The threat from volcanoes outside of Canada seems much greater than the threat from Canadian volcanoes. In part, this appearance is due to the lack of monitoring data on our volcanoes. However, it is certain that volcanoes in Alaska, Washington, Oregon, and California have erupted considerably more frequently in historic time than those in Canada. The American volcanoes are carefully monitored by the United States Geological Survey, and a plan is in place to alert Canadian authorities of eruptions that may send ash into Canada.

The most typical signs of an impending eruption are shallow earthquakes, ground deformation, increased volcanic gas release, increased fumarole vigour and temperatures, and minor landslides and rockfalls. Most of these start to occur months to weeks before the eruption. Days before the eruption starts, other activities occur such as minor steam explosions, and the opening of cracks near the eruptive vent.

There are a wide range of methods for early detection of pre-eruptive volcanic phenomena. A network of at least three seismographs (Figure 52 and Figure 53) surrounding a volcano is needed to accurately locate the small earthquakes occurring in the subsurface. The seismographs are also useful for detecting at a distance events such as rockfalls, landslides, mudflows resulting from melting snow, or small steam blasts. By accurately locating the earthquakes, it's possible to crudely trace the path of the magma toward the surface.

Figure 52. Seismograph and video surveillance station A seismograph and video surveillance station on Montserrat (West Indies). To the left is the video camera and radio antenna, and dimly visible at far left is a dark box which contains the seismometer. The data is sent by radio to the observatory. The station is powered by an array of solar panels (foreground), which are shown here being cleaned of a thin deposit of ash by Jim McMann, the helicopter pilot The ash blocks the sun and causes the station to lose power. (Photo by M. Stasiuk (Geological Survey of Canada)) larger image [JPEG, 27.6 kb, 400 X 260, notice]

Figure 53. Seismogram A seismogram on its drum at the Montserrat Volcano Observatory. The drum and attached paper rotates beneath a blue-ink pen, which moves in proportion to the ground motion experienced by the seismometer that feeds the pen its data. The large amplitude signals visible on the paper record are mostly ground vibrations resulting from the passage of pyroclastic flows and rockfalls from the lava dome, not actual earthquakes. Such paper records are catalogued as visual records, but the data is also typically collected, stored and analyzed in digital format. (Photo by M. Stasiuk (Geological Survey of Canada)) larger image [JPEG, 22.8 kb, 400 X 262, notice]

As the magma approaches the surface, it must push aside the rock and this causes the ground surface to shift and swell. Normally the amount of ground deformation varies from millimeters to meters, especially close to the eruption site. Ground deformation can be measured using traditional methods of surveying, such as levelling, theodolite surveying, or total station surveying (electronic distance measurement with an infra-red laser plus accurate angle measurement), as well as tilt meters (Figure 48, 49, 50, 51). A relatively new method that is currently being widely applied is surveying by very accurate Global Positioning System (GPS) methods (Figure 49). There is also a newer method for detecting ground deformation that is being developed, which uses satellite measurements, called "InSAR" (Interferometric synthetic aperture radar).

Figure 48. Optical theodolite An optical theodolite being used on Montserrat in 1996 to measure the growth of the lava dome (in distance, partially covered in meteoric and ash clouds). A theodolite measures horizontal and vertical angles extremely accurately, and is a common surveying tool. By siting on points on the dome from at least two locations, they can be accurately located in 3 dimensions. Repeated measurements then can define the movement of the dome. (Photo by M. Stasiuk (Geological Survey of Canada)) larger image [JPEG, 33.7 kb, 250 X 390, notice]

Figure 49. Optical theodolite A high-precision GPS (Global Positioning System) instrument being used in Dominica in 1999 to monitor volcanic unrest. Such instruments can provide point locations with accuracies of about half a centimetre, and can detect ground deformation, or volcano swelling, related to pressure changes inside the volcano. (Photo by L. Millar) larger image [JPEG, 35.6 kb, 260 X 390, notice]

Figure 50. Performing a survey Rob Watts (U. Bristol) on Montserrat (West Indies) performing a survey on the growing lava dome (out of view) using infrared laser range-finding binoculars. The binoculars can measure locations to about 3 km range and to an accuracy of a few metres, sufficient for crude estimates of dome growth. (Photo by M. Stasiuk (Geological Survey of Canada)) larger image [JPEG, 21.6 kb, 400 X 259, notice]

Figure 51. Using an EMD George Skerritt of the Montserrat Volcano Observatory, using an EDM (electronic distance measurement) combined with a digital theodolite, to measure ground deformation near the growing lava dome on Montserrat, West Indies. The EDM uses an infrared laser and requires specially-made reflectors as targets, which have to be installed. Provided these can be installed, measurements can be made to millimetric accuracy from ranges of about 5 km. This technique requires clear weather and the target reflectors must be kept clear of ash and debris, a significant task when they are placed close to an actively erupting vent. (Photo by M. Stasiuk (Geological Survey of Canada)) larger image [JPEG, 24.3 kb, 400 X 261, notice]

Figure 73. The GSC-Resonance COSPEC Simon Young using the GSC-Resonance COSPEC on Montserrat, West Indies. The instrument measures the quantity of sulphur dioxide in the volcanic plume (top). Sulphur dioxide absorbs certain wavelengths of light, and the instrument can measure the amount of this absorption by observing the sunlight passing through the plume. (Photo by M. Stasiuk (Geological Survey of Canada)) larger image [JPEG, 35.1 kb, 400 X 300, notice]

Fumarole temperatures are normally monitored in the field with thermocouples, and water samples from hotsprings are taken for analysis to look for sudden changes in chemistry. Large-scale volcanic gas emissions are monitored normally with an instrument called a "COSPEC", a Canadian invention whose name stands for Correlation Spectrometer and which measures the amount of sulphur dioxide in volcanic gas plumes (Figure 73). Sulphur dioxide is characteristically produced by magma in large quantities and so its presence in a steam plume is a good indicator that magma is near the surface.

Once an eruption has started, the monitoring must continue. This is because the vast majority of eruptions are not short, simple events. Usually an eruption consists of many eruptive events or episodes, spaced out over days, weeks, months, years or even decades. The eruption on the Caribbean island of Montserrat that began in 1995, for example, has gone on almost continously and shows no clear signs of stopping. During a prolonged eruption, the activity usually evolves or cycles through periods of more vigorous -- and hazardous -- behaviour. By continuously monitoring the activity, volcanologists can keep track of these cycles or evolutions and often can develop an ability to provide short-term forecasts of heightened activity which save lives. If possible, all the methods described above are used constantly during eruptions, and a big part of the job becomes maintaining the operation of the instruments despite the activity (Figure 52 and Figure 55). In addition, where possible scientists obtain samples of the most recently erupted magma to monitor its physical and chemical characteristics (Figure 54). These characteristics often change during eruptions and help to understand the volcano's behaviour.

Figure 54. Volcanologists taking samples Volcanologists taking samples from the delta of pyroclastic material (see fig. 6) at the coast on Montserrat (West Indies) in 1996. Collapses of the lava dome regularly sent pyroclastic flows the few kilometers to the coast. Keeping track of the physical properties and chemistry of the erupting lava can provide important insight into the direction that an eruption is taking, so it was important to take such samples whenever possible, and this was the safest place to do so despite the material being a few hundred degrees celsius only a few centimeters below the surface. (Photo by M. Stasiuk (Geological Survey of Canada)) larger image [JPEG, 15.7 kb, 400 X 258, notice]

Figure 55. Volcanologists installing instruments Volcanologists visiting the crater rim early in the eruption of Montserrat (West Indies) in 1996 to install instruments that would radio information on the ground deformation back to the observatory. The fog is acidic volcanic gas emanating from the vent. Close to the vent, the conentration of these gases is sufficient to kill all vegetation; the vertical "rods" are actually the still-rooted trunks of dead palm trees. Scientist in the foreground is Jean-Christophe Komorowski. (Photo by M. Stasiuk (Geological Survey of Canada)) larger image [JPEG, 12.4 kb, 400 X 257, notice]

It can be as difficult to tell when a volcano has stopped erupting as to predict whether it will erupt. The reason for this is that during an eruption, there are often pauses in activity that last for days to months, during which little or nothing appears to be happening. Because the activity can resume, it isn't wise to stop monitoring and let people return to their homes the minute a volcano goes quiet. Even after many months of quiet, when people may have been allowed to return home, monitoring usually continues at least at a basic level (seismicity), just in case another phase of activity starts. In the post-eruption period, monitoring helps mitigate risks typical of recently active volcanoes, such as hot rock avalanches from newly-grown lava domes, steam explosions from hot deposits, or sudden mudflows or floods from rivers dammed by new volcanic deposits. In the Philipines, the region around Mt. Pinatubo continues to suffer from extremely dangerous mudflows as rain and rivers remobilize the huge volumes of ash produced in the 1991 eruption. Most post-eruption hazards persist at volcanoes for decades, effectively requiring at least some level of permanent monitoring.