Full-waveform tomography combining body and surface waves to characterize liquefaction hazards

Abstract

Full waveform tomography for near-surface applications has garnered increased attention in recent years due to increasing computational capabilities and rapid developments in the continued search for hydrocarbon sources. Full waveform tomography attempts to solve an inversion problem whereby the entirety of the seismic waveforms measured at a site are matched with waveforms acquired from numerical simulations of wave propagation in a subsurface model. One of the strengths of the method is the ability to incorporate all types of waves into the inversion. With careful placement of measurement locations, combined recordings of body waves and surface waves can be exploited to improve the inversion results. One near-surface engineering application where a combined body and surface wave full waveform inversion (FWI) technique can potentially improve resolution capabilities is for evaluation of liquefaction triggering. Liquefaction of granular soils refers to the loss of shear strength caused by rapid dynamic loading as encountered in earthquake events. The evaluation of liquefaction triggering typically involves site characterization at pre-determined locations with standard penetration tests (SPT) and cone penetration tests (CPT) to estimate the resistance of the subsurface soils against liquefaction. Geophysical measurements from either downhole, crosshole, or surface-wave testing methods can be used to augment this SPT/CPT information. Since boreholes or CPT soundings are often already present at a site being investigated for liquefaction hazards, a receiver array could be simultaneously deployed at the surface and within the subsurface to capture both surface and body waves generated by a single source. This study uses numerical simulations to examine how the accuracy and resolution of FWI can be improved with combined body and surface wave measurements within the context of characterizing liquefaction triggering. The results demonstrated that the spatial extent of liquefaction is better estimated using the combined full waveform approach when compared to full waveform of only surface measurements and when compared to interpolation between localized in-situ measurements.