Full Waveform Tomography to Address Challenges with Surface Wave Dispersion Information Caused by Significant Stochastic Variability of Subsurface Stiffness

Surface wave methods have increased in popularity as a means to acquire rapid and accurate shear wave velocity ( VS) profiles for engineering applications. Use of the multichannel analysis of surface waves (MASW) method, in particular, has proliferated due to multiple factors, including the ease with which strong signal-to-noise ratio can be achieved and the increased dispersion resolution offered by multichannel acquisitions. However, typical MASW processing to extract dispersion information assumes the surface waves propagate through a layered model. Errors can arise when significant lateral variability is present in the underlying stratigraphy as encountered in certain geologic settings such as residual deposits. This study investigated the effects of such variability on the dispersion information acquired with MASW. In particular, a spatially-correlated Gaussian random field was used to model the natural fluctuations in stiffness introduced by depositional processes, which differs from the approach in other studies where more specific anomalous features have been explored. Numerical modeling was subsequently performed to simulate surface wave propagation in the representative geotechnical site condition. The recovered surface waves were used to develop a subsurface stiffness profile using a dispersion-based pseudo-2D MASW approach and a tomographic approach using full waveform inversion (FWI). The results demonstrate that considerable natural spatial variability significantly complicates interpretation of dispersion information in two primary ways: (1) uncertainty can arise regarding what the dispersion curve exactly quantifies since none of the underlying VS profiles nor the average VS profile are obtained; and (2) the dispersion images exhibit evidence of false depth-related dispersion information indicative of multiple “fundamental” modes from the superposition of multiple stratigraphic units. The FWI procedure that bypasses extraction of dispersion information was found to better recover the underlying subsurface conditions when compared to the pseudo-2D MASW results at the cost of additional computational efforts.