PEGASOS PROJECT: ESTIMATION OF THE MEDIAN, NEAR FAULT GROUND MOTION IN SWITZERLAND - Scientific Report n.5

We perform some numerical simulations to assess the median ground motion for a M = 6:5 event at short distances from the fault. We take into account the 1-D Northern Switzerland model (CHN) and two fault mechanisms, i.e. vertical strike-slip and 45± dip-slip fault. In order to evaluate the ground motion variation with both distance and azimuth, seismograms are computed along ¯ve arrays of receivers at di®erent distances from the fault in the distance range from 1 km to 25 km.. We focus on the median ground motion. For this reason, the source parametrization follows di®erent criteria from those used in our previous study (Priolo et al., 2002), whose goal was to assess the extreme motion. Here, we de¯ne seismic moment (or slip) distri- butions having medium stress drop level and feature no sharp and localized asperities. Furthermore, instead of evaluating a single critical case, we need to build-up a robust statistics. Therefore, we take into account a large number of simulations for which we let to vary the slip distribution and rupture propagation (i.e., the aleatory parameters). Simulations are performed using EXWIM 2.1 method. Version 2.1 improves version 2.0, used in our previous study, in that it implements a hybrid low-frequency-deterministic, high-frequency-stochastic approach. The need of using a hybrid deterministic-stochastic approach comes from few observations (Liu and Helmberger, 1994) and a common ac- ceptance among authors (for instance, Pitarka et al, 2000; Madariaga, 2002) of the fact that the signal looses its coherence in the high-frequency band (e.g., for f > 2 Hz) as an e®ect of the wave¯eld propagation through the earth. Any modelling method should take into account this aspect correctly (Madariaga, 2002). In particular, the high-frequency stochastic character of the seismogram is relevant when one wants to simulate realistic, highly likely situations, as we try to do here. However, the use of a full deterministic ap- proach (Priolo et al., 2002) is still justi¯ed to evaluate extreme cases in which the signal coherence persists in the high frequency band (Zahradnik, 2002b). In this report, we ¯rst summarize the improvement of the new version of EXWIM, then we describe the setting of the numerical simulations, summarize the results obtained, and draw some conclusions.