Ask A Scientist

How can you predict when a volcano is going to erupt?

Asked by: Keith Gardner
School: Owego Elementary School
Grade: 6
Teacher: Chris Mahon
Hobbies/Interests: Tennis, soccer, skiing, hockey and collecting stones.
Career Interest: Inventor

Answer from Richard Naslund

Professor of geology, Binghamton University

Ph.D. School:
University of Oregon

Research area:
Volcanology, crystallization of magmas, ore deposits, chemistry of the Earth.

Travel to geologically interesting places, scuba diving and raising tropical fish, hiking in the woods.

Wife, with five children

Web page:

Predicting a volcanic eruption is a political as well as a scientific exercise. The consequences of failing to give adequate warning are obvious. In 1902 when Mt. Pelee on the island of Martinique began to get restless, the authorities encouraged people to stay in town because of an upcoming election. The resulting eruption killed 28,000 people. In 1985 when Nevado del Ruiz volcano in Columbia began to erupt, authorities were warned but instead told people there was nothing to worry about. The town of Armero was inundated with mudflows and 23,000 people died. It is easy to put the blame on the "authorities" but it is difficult to second guess their decision if we don't know the details of the information that they were given. Was the scientific advice consistent, or did they hear a variety of opinions? Was the timeframe of the danger clearly presented? Although it is not as obvious, false warnings are bad as well. We have all heard the story of the boy who cried wolf. There are too many volcanoes in too many populated regions for there to be an evacuation every time a volcano gets active. In 1976, when the volcano La Soufiere on the island of Guadeloupe became active, there was a general panic fueled by the press and a small group of scientists who had little experience with active volcanoes. As a result, 72,000 people living near the volcano were ordered to evacuate. This caused a severe financial hardship, and caused a serious disruption to the lives of not only those 72,000 but to the other 200,000 people on the island whose lives were disturbed by the sudden influx of refugees. The volcano did not erupt, and four months after they were forced to flee, the citizens were allowed to return to their homes and put their lives back together. One thing we learned on Guadeloupe was that in order to make good predictions we need to know what a volcano has done in past eruptions, and how a volcano acts when it is not going to erupt. If we go up to the throat of a volcano and find hot gas coming out at 500 degrees, should we panic? It depends on whether this is normal activity for this volcano or if this is a sudden change. When scientists think a volcano near a populated area is likely to become active, they set up a volcano monitoring program. Over 90% of eruptions are preceded by volcanic earthquakes. This is a good indicator of eruption but we would still miss about 10%, and many volcanic earthquake events do not lead to eruptions, so we would have a lot of false alarms. Monitoring temperature is a simple, low-tech method that is often used, but it is not very reliable. Temperatures near the surface of volcanoes are controlled by a complex set of variables including rainfall, fracture pattern, and increased activity. Temperature can rise, fall, or remain unchanged before an eruption. Tiny changes in elevation and slope are used extensively at some volcanoes, and are generally accurate. At Kilauea in Hawaii, such measurements allow scientist to tell when and where eruptions will occur and is used to get the cameras all set up in advance, before new vents open up. Measuring minor changes in the gravitational field or the magnetic field has been used successfully at some volcanoes. Chemical changes in volcanic gases or volcanic hot springs can also be used with success as a prediction tool if the previous history of a volcano is known. Volcanic prediction is not an exact science, however. In 1993, seven of the worlds premier volcanologists were sampling gases at Galeras volcano, in Colombia, when it erupted killing six of the scientists and three tourists who came along to see how scientists predict eruptions. The best prediction tool for a volcano is given by the previous history of that volcano. We can often accurately predict eruptions at volcanoes that have frequent eruptions, but it is difficult to predict eruptions at volcanoes with no or few historic eruptions. There are over 10,000 recently active volcanoes so we can't monitor them all. In addition, since 1950 three 'inactive' volcanoes have erupted resulting in the deaths of 4000 people. Just last week, Four Peaks Volcano in the Aleutian Islands of Alaska, which was thought to be 'inactive' because it had not erupted in the last 10,000 years, began a new phase of eruption. Fortunately, Four Peaks Volcano is very remote and is not a danger to people on the ground. It is, however, a big danger to commercial aircraft, which can lose power if their jet engines suck in too much volcanic ash.

Last Updated: 3/1/17