Sensitivity analysis for seismic source characteristics to probabilistic seismic hazard assessment in Central Apennines (Abruzzo area)

This paper presents the results of an analysis showing the effect of different assumptions of magnitude recurrence and time dependence on probabilistic seismic hazard assessment for the Abruzzo area. The focus is on alternative peak ground acceleration maps for the whole area and disaggregation analysis for three main towns (L’Aquila, Sulmona and Avezzano). The source model of regional relevance has been defined by: 1) definition of primary source faults on a geological-structural basis; 2) drawing of seismogenic boxes with computation of parameters such as maximum magnitude and relative mean recurrence, based on geometrical and kinematic elements; 3) association of a proper seismicity model consistent with the historical information available. The final seismic hazard assessment incorporates a renewal-time, characteristic-magnitude model for the larger earthquakes, with a conventional exponential-time, exponential-magnitude model for the smaller events, depending on the knowledge of the sources. The hybrid method used here seems to be adequate in modelling earthquake occurrences in fault-specific cases; using the surface projection of 3D sources, we approximate the computations to obtain results suitable for engineering applications. The hypotheses for seismicity models are wide ranging, and are all in some way realistic, because they are based on different earthquake catalogue interpretations, rare paleoseismological data, no geodynamic measurement constraints (at least till now). But, the more we are willing to abandon simplistic assumptions, such as the Gutenberg-Richter law or the characteristic-earthquake model, the more accurately we have to model the implications of all the available information. The disaggregation analysis, in particular, shows how valuable the use of the long history of seismic shaking of some localities is, in assessing the need to model a time-dependent recurrence in the analyses.