RANDOM PATTERNS OF OCCURRENCE OF EXPLOSIVE ERUPTIONS AT COLIMA VOLCANO, MEXICO

Analysis of the patterns of eruption occurrences may improve our understanding of volcanic processes. In this paper, the available historical data of an individual volcano, Colima in Mexico, are used to classify its eruptions by size using the Volcanic Explosivity Index (VEI). The data shows that, if eruptions are only taken into account above a certain VEI level, the stochastic process associated with the explosive volcanic events can be represented by a non-stationary Poisson point process, which can be reduced to a homogeneous Poisson process through a transformation of the time axis. When eruptions are separated by VEI values, the occurrence patterns of each magnitude category can also be represented by a Poisson distribution. Analysis of the rate of occurrence of all eruptions with VEI greater than 1 permits the recognition of three distinct regimes or rates of volcanic activity during the last 430 years. A double stochastic Poisson model is suggested to describe this non-stationary eruptive pattern of Colima volcano and a Bayesian approach permits an estimation the present hazard.