Like many analysts who use Palisade software, Willy Aspinall is faced with a mountain of data. But he is also faced with a mountain of molten lava. A British physicist who specializes in risk analysis of earthquakes and volcano eruptions, Willy has an ongoing assignment on the Caribbean island of Montserrat-to keep tabs on the Soufriere Hills volcano and the risks it poses to the inhabitants of the island. This means continuous modeling, with myriad variables. Not surprisingly, Willy, as a long-time Palisade customer, relies on @RISK and PrecisionTree.
In 1995 the Soufriere Hills volcano became active, venting steam à la the notorious Mount Pinatubo. It continued seething for two years and started to produce what geophysicists refer to as a "dome." Over this time, inhabitants of the island grew desensitized to potential dangers of eruption, and when the dome collapsed explosively in 1997, some people on the island were unprepared. Twenty of them died. As Willy observes, "It's a relatively small volcano, but it's on a very small island," where evacuation must be taken seriously.
The Soufriere Hills volcano remains active, and it remains a hazard. Four times a year Willy visits the Montserrat Volcano Observatory to meet with an international team of volcano experts. He uses @RISK in combination with PrecisionTree to simulate risks to individual villages on Montserrat under various volcanic scenarios. In addition Willy also uses these Palisade tools for a more unusual purpose-to quantify the variability of expert opinion provided by the scientists. He models their "subjective probabilities" as the experts try to anticipate the behavior of the volcano and the ramifications of its eruptive behavior patterns.
Government officials in charge of safety on Montserrat have to rely on expert opinion on fast-breaking events because they, and the scientists, don't have the luxury of waiting for more data to come in to undertake full analyses. Unfortunately, Willy notes, volcanic activity is intrinsically unpredictable, so "all evidence-even expert testimony--is imperfect." In the case of the Montserrat volcano, his task is to quantify all the uncertainties that accompany each assessment. At this point the decision process becomes very complex, with a typical event tree blooming with as many as two hundred branches. For this work, Willy says, PrecisionTree's capacity to enter probability distributions at any juncture, rather than single best estimate values, is invaluable for obtaining meaningful results.
Although most Caribbean volcanoes in recent centuries have tended to erupt briefly and violently, the Montserrat volcano has had a slow buildup and, so far, has not produced a really big "blow." The eruptive activity associated with the dome growth, however, has been quite dangerous for a tiny island, including in its repertoire a re-run of the well-known 1980 Mount St. Helens blast, but on a smaller scale. As the present eruption continues, the volcano falls in the top 15% of long-lived dome-building eruptions known to volcanologists. The emphasis of Willy's work is therefore as much on long-term planning as it is on short-term hazard mitigation. The proven attributes of @RISK and PrecisionTree are just as vital for this work as they were earlier for the urgent business of risk assessment when the eruption started.