The following paper presents a Bayesian model for seismic hazard mapping. The main features of the model are as follows.
Faults are modeled by deterministically located dipping planes; several planes (trapezoids) can be combined within a source to satisfy geometric constraints. Area sources are modeled by horizontal trapezoids, and line sources at constant depth are modeled by one or a series of straight line segments.
The tectonic model considers a rupture area (plane sources) or a rupture length (line sources) associated with each magnitude event.
The seismicity is modeled in two steps. On each source, events, which occur independently of magnitude, are assumed to follow a Poisson model. Distribution of magnitudes is obtained using a Bernoulli process. Subjective information is incorporated using Bayesian statistical concepts to both the above models to supplement insufficient data.
A lognormal distribution is used to take into account the uncertainty in attenuation. The significant distance for attenuation purposes is chosen as the closest distance from the rupture area (length) to the site.
Finally, to simplify the mapping procedure, this model makes it possible for a number of mapping parameters (governed by core limitations) to be obtained in a single run at the nodes of a whole grid.