Ground motion models (GMMs) are used to assess seismic demands given parameters descriptive of source, path, and site conditions. The current models used in California were developed from a global database, and incorporate path and/or site adjustments representing the statewide average. My research is concerned with improving ground motion predictions in the Sacramento-San Joaquin Delta region of northern California. The Delta is home to vital infrastructure which includes over 1,770 km of levees that serve as a conduit for approximately two-thirds of the state’s drinking-water supply. As such, levee damage from earthquakes is potentially catastrophic. Seismic hazards are controlled mostly by nearby sources, however distant larger sources are also impactful. Seismic waves from these sources cross many physiographic provinces to reach the Delta, hence complex path effects deviating from the statewide average are expected. Furthermore, much of the Delta is underlain by soft peaty-organic soils, which are absent from databases used to develop global GMMs. These factors can be accounted for by customizing GMMs for the Delta subregion, particularly the path and site response components.

The approach followed in this research was to first develop a California ground motion database with particular emphasis on northern California and the Delta subregion. Path effects were modeled to allow for different anelastic attenuation in each of nine distinct physiographic domains, which has a stronger physical basis than prior cell-specific methods. The regionally customized path model was used with the ground motion data to derive observation-based site amplification at 36 sites in and around the Delta. Based on those results, I then developed two subregional site response models. The first model relates small-strain amplification with the time-averaged shear wave velocity in the upper 30 m (VS30), and can be applied for all sites in the Delta and immediately surrounding areas. The second model incorporates additive terms to the VS30-based model to account for site resonance effects and additional levels of amplification; the independent variables are derived from microtremor horizontal-to-vertical spectral ratios (mHVSR). Collectively, these regionally-calibrated models significantly reduce bias and variability for ground motion predictions in the Delta region.