Seismological Modeling

Abstract

AbstractThe precise assessment of the long‐term risk of major earthquakes is required for setting insurance rates, for establishing appropriate building codes, and for allocating resources to emergency management. The locations of earthquake epicenters’ together with the times at which they occur, form a spatiotemporal point pattern. If, in addition, their magnitudes (on the Richter scale) are available, then one has a marked spatiotemporal point pattern. For risk assessment, one shall assume that this pattern is a realization of a marked spatiotemporal point process, a random mechanism for generating the locations and times of events as well as their marks. The statistical approach to earthquake risk assessment depends on whether one is interested in assessing risk at a specific site (e.g. Los Angeles), or in assessing risk as a function of location. For the former, one shall consider models for the marginal temporal point pattern formed by the times of seismological events. Methods for fitting parametric models are described, and then the use of fitted models for assessing risk is considered. To assess risk as a function of location, the marginal spatial point pattern is considered. Here, a nonparametric approach is described for estimating the intensity of earthquake epicenters as a function of location.