You are hereEngineering implications of asperity-induced ground motion

Engineering implications of asperity-induced ground motion

Chu-Chuan Peter Tsai
Principal Advisor: 
C. Allin Cornell
Year Published: 
Wed, 01/01/1992 (All day)

A simple model was introduced to reflect local non-homogeneous characteristics along a fault. For strong motion prediction, the important part of these spatial source heterogeneities is the most energetic regions that we refer to here as asperities. The ultimate objective of the research is to explore the engineering implications of asperity-induced ground motion to probabilistic seismic hazard analysis. We focus on strike-slip faults. No site and topographic effects are considered. The seismic waves considered are direct shear waves.

The effects of asperities are first investigated empirically by way of inferred slip maps available for two past earthquakes. We use Andrews' (1980) method to calculate static stress drop distributions in the wave number domain. The proposed model is tested and calibrated by comparing its predictions with strong ground motion observations from the 1979 Imperial Valley and the 1984 Morgan Hill earthquakes. For future earthquake events, simulations of fault geometry and slip distribution are both explored. A concept adopted from fractal geometry is employed to set up procedures for the simulation of samples of the future slip distribution.

The results show that ground motion in the near-field can be adequately predicted using Andrews' method with appropriate calibration factors. The directivity effect of ground motion in the near-field is found negligible for the high-frequency accelerations. To a limited degree, the non-homogeneous source model seems to depict better the variability of the source strength along the fault than the homogeneous source model. The cut-off frequency of ground acceleration at a site may be an important parameter. In the near-field, both non-uniform and uniform source models can produce non-stationary high-frequency ground motions. Peak motions may be not caused by the nearest sections of the fault (even if the uniform source model is considered). Hazard curves are sensitive to the standard deviation of the ground motion prediction model. The dispersion of the prediction model may mask the difference between hazard curves using the proposed asperity model and that using the conventional models.