Update of the BC Hydro Subduction Ground-Motion Model using the NGA- Subduction Dataset

Abstract

An update to the BCHydro ground-motion model for subduction earthquakes has been developed using the 2018 PEER NGA-SUB dataset. The selected subset includes over 70,000 recordings from 1880 earthquakes. The update modifies the BCHydro model to include regional terms for the VS30 scaling, large distance (linear R) scaling, and constant terms, which is consistent with the regionalization approach used in the NGA-W2 ground-motion models. A total of six regions were considered: Cascadia, Central America, Japan, New Zealand, South America, and Taiwan. Region- independent terms are used for the small-magnitude scaling, geometrical spreading, depth to top of rupture (ZTOR ) scaling, and slab/interface scaling. The break in the magnitude scaling at large magnitudes for slab earthquakes is based on thickness of the slab and is subduction-zone dependent. The magnitude scaling for large magnitudes is constrained based on finite-fault simulations as given in the 2016 BCHydro model. Nonlinear site response is also constrained to be the same as the 2016 BCHydro model. The sparse ground-motion data from Cascadia show a factor of 2–3 lower ground motions than for other regions. Without a sound physical basis for this large reduction, the Cascadia model is adjusted to be consistent with the average from all regions for the center range of the data: M = 6.5, R = 100 km, VS30 = 400 m/sec. Epistemic uncertainty is included using the scaled backbone approach, with high and low models based on the range of average ground motions for the different regions. For the Cascadia region, the ground-motion model is considered applicable to distance up to 1000 km, magnitudes of 5.0 to 9.5, and periods from 0 to 10 sec. The intended use of this update is to provide an improved ground-motion model for consideration for use in the development of updated U.S. national hazard maps. This update ground-motion model will be superseded by the NGA-SUB ground-motion model when they are completed.