| Abstract: | A semi-empirical approach using fore- or after-shockrecords as Green's functions is applicable to thesimulation of strong ground motion, however suchrecords are obviously not available for predictionpurposes. Thus we have predicted ground motion fora hypothetical large earthquake from other minorevents by adopting a distance correction based ongeometrical spreading. Another difficulty inprediction is fault modeling. Surface traces weresimplified as fault models 27, 46, 55, and 77 km inlength. Further, the actual fault rupture may beinhomogeneous, so an asperity distribution isassumed. This asperity model assumes thatdislocation and stress drop are double than theaverage values. Although, the near field term isneglected in our simulation, no significantdifference was seen in the motions estimated byindividual models for periods up to 2.0 seconds. This indicates that the dependence of source size issmall for strong motion, perhaps as a result of therandom summation of high-frequency phases. |
