Assessment of shear wave velocity models in the Southeast Tibet with full-wave simulation

Various velocity models have been built for southeast Tibet with the purpose of revealing the internal dynamics and estimating local seismic hazards. In this study, we use a 3-D full-waveform modeling package to systematically validate three published continental-scale velocity models, that is, Shen2016, FWEA18, and USTClitho1.0, leveraging the ample datasets in Southeast Tibet region. Travel time residuals and waveform similarities are measured between observed Empirical Green’s functions and synthetic waveforms. The results show that the Shen2016 model, derived from traditional surface wave tomography, performs best in fitting Rayleigh waves in the Southeast Tibet, followed by FWEA18, built from full-waveform inversion of long-period body and surface waves. The USTClitho1.0 model, although inverted from body wave datasets, is comparable with FWEA18 in fitting Rayleigh waves. The results also show that all the models are faster than the ground-truth model and show relatively large travel-time residuals and poor waveform similarities at shorter period bands, possibly caused by small-scale structural heterogeneities in the shallower crust. We further invert the time residuals for spatial velocity residuals and reveal that all three models underestimate the amplitudes of high- and low-velocity anomalies. The underestimated amplitude is up to 4%, which is non-negligible considering that the overall amplitude of anomalies is only 5%-10% in the crust. These results suggest that datasets and the inversion method are both essential to building accurate models and further refinements of these models are necessary.