A Ground Motion Prediction Model for Deep Earthquakes beneath the Island of Hawaii

Until recently, no ground motion prediction model was available for deep (>20 km) Hawaiian earthquakes, including the 2006 M6.7 Kiholo Bay earthquake. We developed such a model based on the stochastic point-source model. Strong motion data from the 2006 event and 15 other deep Hawaiian earthquakes of M3.3 to M6.2 were inverted using a nonlinear least-squares inversion of Fourier amplitude spectra to estimate stress drops for input into the stochastic modeling and for the few larger events (M ≥ 5.0), to calibrate the ground motion prediction model. The ground motion model is valid for M3.5 to M7.5 over the Joyner-Boore (R_JB) distance range of 20 km to 400 km and are for 5%-damped horizontal spectral acceleration at 27 periods from PGA (0.01 s) to 10.0 s. The shallow site condition assumed for the model is soil and weathered basalt with a mean V_S30 of 428 m/s.