Stochastic point-source modeling of ground motions in the Cascadia region

A stochastic model is used to develop preliminary ground motion relations for the Cascadia region for rock sites. The model parameters are derived from empirical analyses of seismographic data from the Cascadia region. The model is based on a Brune point-source characterized by a stress parameter of 50 bars. The model predictions are compared to ground-motion data from the Cascadia region and to data from large earthquakes in other subduction zones. The point-source simulations match the observations from moderate events (M < 7) in the Cascadia region. The simulations predict a steeper attenuation than observed for very large subduction events (M ??? 7.5) in other regions; motions are overpredicted near the earthquake source and underpredicted at large distances (>100 km). The discrepancy at large magnitudes suggests further work on modeling finite-fault effects and regional attenuation is warranted. In the meantime, the preliminary equations are satisfactory for predicting motions from events of M < 7 and provide conservative estimates of motions from larger events at distances less than 100 km.