Modeling the Probability of a Volcanic Event

Three models developed at the CNWRA to describe spatial and temporal patterns in basaltic volcanism have been applied to the probability of volcanic eruptions occurring in the Yucca Mountain region. These nonparametric models include a spatio-temporal nearest-neighbor model that treats volcanism as a marked spatio-temporal point pattern process, a kernel method that treats volcanism as a point process within a defined spatial and temporal bandwidth, and a hybrid nearest-neighbor kernel model that combines several aspects of the other two models.

Probability density functions are estimated directly from the location and timing of discrete volcanic events using each of these three methods. Researchers have found that these near-neighbor models are sensitive to patterns generally recognized in cinder cone distributions. The resulting probability surfaces are continuous, rather than consisting of the abrupt changes in probability prevalent in spatially homogeneous models. A further advantage is that these probability surfaces can be readily compared to other geologic data, such as the distribution of faults, that may influence volcano distribution.

Published in the Summer 1995 issue of Technology Today®, published by Southwest Research Institute. For more information, contact Joe Fohn.