Ground Motion Prediction using the Virtual Earthquake Approach
Marine Denolle, Stanford University
Wednesday, April 10, 2013 at 10:30 AM
- Building 3, Room 3240 (main USGS conference room)
- Annemarie Baltay
Ground motion prediction is a key element of seismic hazard assessment. One approach to estimate ground motion is to predict intensity based on empirically developed ground motion prediction equations. With a shortage of data for large events at short distances, seismologists increasingly would like to be able to use physics based simulations of ground shaking. Ambient noise seismology can contribute to ground motion prediction by constraining wave-propagation from virtual sources. We compute the Earth impulse responses such that we extract path effects, both elastic and anelastic. We then directly predict long period surface-wave ground motion using the Virtual Earthquake Approach (VEA). We correct the responses due to a single force acting at the surface, to the responses due to a double-couple force system at depth (virtual earthquake) using our understanding of the surface-wave excitation at the source. We validate the predicted ground motion against moderate earthquakes in southern California, and demonstrate the predictive value of the VEA. We extend point-source ground motion predictions to M7+ scenario earthquakes on the southern San Andreas Fault by using virtual earthquakes for stations along the fault and the representation theorem. We use this approach to test the waveguide-to-basin amplification effect in Los Angeles that occurs in large-scale simulations of San Andreas earthquake ruptures through the San Gorgonio Pass.