Testing the Validity of Shear-Wave Splitting Measurements in the Presence of Scattering Using Synthetic Waveform Modelling

Shear-wave splitting (SWS) measurements are a useful tool for studying temporal variation of seismic anisotropy, which in turn can be used as a proxy for changes in stress around fault-lines, volcanoes, magma movements, and eruptive activity. Seismic scattering has been identified in several cases as a source of uncertainty in SWS, whereby multiple arrivals from the scattering of seismic waves off of heterogeneous bodies can alter the SWS caused by anisotropy. Additionally, changes in the properties of seismic scatterers, or small variations in wave path can mimic or obscure temporal changes in anisotropy. These sources of uncertainty put limitations on our confidence in SWS measurements in environments where scattering is likely, and may contribute to errors in otherwise useful measurements of SWS and anisotropy.

In this project, we aimed to quantitatively measuring the effects of scatterers and related sources of error on SWS measurements by using computer simulations of seismic wave propagation through the crust. The main objectives of this research were to determine the degree to which SWS measurements are affected by scatterers and investigate the impact of small differences in path or source position on measurements. Additionally, we also wanted to evaluate whether using averaging techniques over measurements at multiple stations could improve the quality of results in the presence of scattering. We also explored how groups of aligned dikes could induce anisotropic behavior in isotropic models, and the impact introducing additional noise into synthetic seismograms had on the uncertainty of scattered SWS measurements.

The primary tool used to conduct this investigation was the SpecFem3D software package, which uses the spectral-element method to simulate seismic wave propagation, and to create synthetic waveforms with controlled source mechanisms, wave paths, velocity models and anisotropy. Scattering was introduced by including heterogeneous bodies modeled on a large variety of sources. These synthetic seismograms were processed with the MFAST software package to generate splitting measurements for each set of seismograms. These data-sets of splitting measurements were then used to study how the splitting properties vary with small changes in model parameters, and to examine the effectiveness of different averaging techniques.

Using these methods, we found that scattering impacts both the uncertainty of SWS measurements, and also alter the perceived anisotropy of the material. This was most pronounced when the heterogeneities used to induce scattering were of a similar size to the dominant wavelengths of the S-wave. We investigate the effects of series of aligned dikes on shear-waves passing through an otherwise isotropic medium. Although we were limited by the resolution achievable with the model sizes we could use, in some cases we were able retrieve measurements from multiple stations with directions consistent with the orientation of the dikes.

We also tested the impacts of introduced background noise on synthetic waveforms generated with and without scatterers. When measurement are restricted to having difference between back-azimuth and fast polarization of ≤ 20 degrees, we found that we could retrieve good measurements for SNR greater than 4:1. This could be improved by averaging measurements from multiple stations, where we could get correct measurements for SNR as small as 2.5 in a scattered model. Finally, we studied the impacts of small changes in source position on SWS measurements. We found evidence that these small changes can produce pronounced changes in fast direction measurements, especially when the ray-paths for these measurements intersected the edges of heterogeneous bodies.

Our findings improve our understanding of the sources of uncertainty in SWS measurements. We have also shown that computer simulations can be a useful tool for quantitatively measuring the impacts of scattering on SWS measurements, and we investigated and highlighted phenomena that we can identify as sources of error. Improving our understanding of the factors that affect SWS means that we can be more confident when using SWS as tool to understand other parts of the earth system, such as tectonic stresses, fluid migration, and temporal changes in anisotropy associated with magma movement and volcanic activity.