Long-Term Probabilistic Analysis of Future Explosive Eruptions

Abstract

The analysis of the patterns of explosive eruption occurrences helps volcanologists and civil defense authorities to make quantitative estimations of the levels of hazard. An important first step in an assessment of volcano hazards is to classify the eruptions by size or magnitude, and to recognize the statistical distributions of the eruption occurrences and repose times between them. For a single volcano, such distributions may be stationary or time dependent. In the latter case, the volcano under study may have different regimes or rates of activity and their distribution should be looked for. The current and future hazard level or probability of eruption occurrence in a certain magnitude range may thus be estimated by Bayesian methods. This type of analysis could prove helpful in assigning priorities for volcano surveillance during repose periods, in decision making during volcanic crises, and in land-use planning. When large populations of volcanoes are considered, as in the global case, the distribution of explosive eruption occurrences seems to approach a stationary pattern, at least for longer times, during which the rate of occurrence of a given class of eruptive magnitude is inversely proportional to the volcanic energy released by eruptions in that magnitude range.