they want to know when it will blow

After 11 volcanologists died during the eruption of the Galeras Volcano in Colombia in 1993, Charles Connor, a long time volcano researcher and chairman of USF’s Department of Geology, decided it was time to develop a better method for predicting volcanic eruptions. 

With the help of colleagues from the University of Bristol in England and volcano researchers from Hawaii, Connor formulated a new model, which he said he hopes will save lives as well as teach volcanologists more about when and why volcanoes erupt. 

Connor said the new model is referred to as the log-logistics model and is a revised edition of the old materials failure model. The traditional way to use probability forecasts was to examine the rate of volcanic activity, as well as the frequency of eruptions, and then apply a random process in order to forecast when the next volcanic eruption will occur. But Connor said this model did not take into account other significant activities occurring in the conduit — the upper part of the volcano — such as certain pressures and the factors that influence those pressures. 

“When we took that into account, we discovered that we could more accurately forecast volcanic eruptions, at least in some cases,” Connor said. 

Connor and his colleagues came up with this new log-logistics model after realizing that there are competing processes occurring in the conduit of the volcano that increase and decrease the pressure. Pressure increases in the upper part of the volcano due to the intrusion of magma into the conduit, Connor said. Pressure can also decrease due to fractures developing in the volcano, which allow gas and pressure to escape, Connor said. 

The log-logistics model is able to forecast volcanic eruptions better by accounting for these competing processes that build or reduce pressure. In addition, there are a number of ways to monitor these processes. One is by using instruments that monitor the amount of gas fluctuation occurring in the volcano. Another is the rate at which the volcano deforms and a final way is by taking into account the number of earthquakes that occur at the volcano, Connor said. 

Although the new forecast model is being used to help volcanologists better predict volcanic eruptions, Connor said it would not be used to forecast all volcanic eruptions,  “so it’s not a magic bullet that will solve all our problems,” Connor said. 

The new model will only be used to predict explosive eruptions that occur in the upper part of the volcano, he said. These particular types of volcanoes are characterized by a steady fluctuation and accumulation of magma in the upper part of the volcano.

Connor added that the accumulation of magma increases the pressure inside the volcano causing it to explode. Volcanoes that emit this type of eruption include Galeras, Montserrat in the Lesser Antilles and Unzen located in Japan.  

The log-logistics model is being tested on the Colima volcano in Mexico by one of Connor’s graduate students, Kristin Martin. 

Martin said working on volcanoes is a dangerous business, so any improvement in the ability to forecast eruptions will help researchers. She added that this research would help to benefit society in the long run. 

“People living in this city (Colima) want to know what is happening at the volcano because it is right outside their door. Hopefully our model will help improve the understanding of the dangers,” Martin said. 

Connor also said he agrees with Martin’s objective of being able to inform those most affected by volcanic activities. 

“People who live near volcanoes want to know when they erupt, so any model we can develop that more closely ties what we know about the physics of volcanic eruptions to the probability that they are going to happen will help reduce hazards for people living near volcanoes,” Connor said.