Synthetic stochastic ground motions compatible with the Chilean seismic hazard

This paper develops a record-based stochastic ground motion model for the Chilean Subduction zone that generates ground motions compatible with the seismic hazard (represented by a local Ground Motion Predictive Equation GMPE) for both interplate and intraslab mechanism and considering rock soil conditions. The stochastic ground motion is obtained by the modulation of white-noise sequence in both time and spectral domain applying non-stationary filters. Compatibility is obtained by a proper tuning of predictive relationships, which offers a relation between seismological parameters and the driven parameters in the non-stationary filters. This process demands an intensive computational optimization problem where the parameters in the predictive relationships are found such that the stochastic ground motions generated match with the GMPE. The computational burden is decreased by the use of a Kriging metamodel which provides a high-fidelity approximation of the mean response spectra for different natural structural periods and the predictive relationship parameters. Once the predictive relationships are adjusted, the results are validated by comparing the prediction of mean spectral acceleration response from the GMPE with the prediction using the stochastic ground motion model. The Kriging model is built once and could be used to repeat the optimization for new GMPEs and predictive relationships (as a result of an updating process to incorporate new important seismic events) within a relatively low computational cost. In addition, a post-adjustment methodology is proposed to improve the results, using a scaling and spectral matching technique.

Automated summary

This paper presents a record-based stochastic ground motion model for the Chilean Subduction zone, taking into account interplate and intraslab mechanisms as well as rock soil conditions. By modulating white-noise sequence in both time and spectral domain, the model generates ground motions compatible with a local Ground Motion Predictive Equation. The study involves an intensive computational optimization process to adjust predictive relationships and match the stochastic ground motions with the GMPE.