European Seismological Commission, 32nd General Assembly, Sept. 6-10, 2010, Montpellier, France
Ameri G., Emolo A., Pacor F.
To investigate the performances of simulation techniques in near-fault strong-motion modeling and prediction from past and future events is an important issue in the perspective of using synthetic seismograms for seismic hazard applications. In this work, we compute ground motions for the 1980, magnitude 6.9, Irpinia earthquake (southern Italy) and scenario events using three different simulation codes, two of which are based on the stochastic approach and one on a broadband integral-composite method.
First, we synthesize the near-fault strong motion recordings and evaluate the capability of the simulations to reproduce the main features (amplitude and frequency content) of observed data. The local site amplification is included in the synthetic time-series either by means of HVSR-based amplification functions or of 1D theoretical transfer functions based on soil profiles The results show that the adopted modeling techniques are able to reproduce fairly well the observed peak values and spectral ordinates and confirm that the high-velocity/low-acceleration character of the Irpinia earthquake can be ascribed to the characteristics of the source combined with the crustal attenuation properties. The next step of the study is the generation of ground motion from scenario events, associated to the Irpinia fault, in order to quantify the variability in ground motion prediction for future events. We construct a large set of rupture models for various positions of the nucleation point, final slip
distributions, and rupture velocity values and compute synthetic accelerograms, for generic hard-rock sites, at a grid of virtual receivers, densely distributed around the fault. The calculated peak ground acceleration and velocity are then statistically treated to evaluate the median predicted ground motion and the total, inter-scenario and intra-scenario variabilities. The median values are consistent over the three simulation methods. On the other hand, the components of variability show significant differences, depending on the approximations, made in each simulation method, in describing the source and wave propagation processes.
Finally, the synthetic median curves are compared with some recent empirical ground motion prediction equations. Good agreement is found for P GA and P GV at short distances, while for distances larger than about 10 km, synthetic P GAs decay faster than the empirical predictions.