Inverse Physics-based Modeling of the 2016 Mw 6.2 Tottori Earthquake

The 2016 Mw 6.2 left-lateral strike-slip Tottori earthquake occurred in the central part of the Tottori Prefecture in the Chugoku region in western Japan. Published rupture models inferred either from geodetic or seismic data exhibit significant discrepancies. In this study, we perform a so-far missing dynamic rupture simulation to improve the understanding of the event. We utilize 3D finite-difference staggered grid code FD3D_TSN to simulate the dynamic rupture propagation assuming the classic linear slip-weakening friction law on a planar vertical fault. Synthetic seismograms are calculated using the representation theorem by convolving the obtained slip rates with Green's functions precalculated in 1D velocity models acquired for each station from a 3D model.

We utilize the fd3d_tsn_pt code to perform a dynamic source inversion with spatially variable prestress and friction parameters, formulated in a Bayesian framework, employing the Parallel Tempering MCMC approach to sample the posterior distribution of the model parameters. We use local low-frequency seismic waveforms (up to 0.6-1.2 Hz) and GNSS static coseismic displacements. We obtain posterior model samples with complex rupture propagation, discuss the inferred heterogeneous rupture parameters in a statistical sense, and compare them with previously published models. We evaluate correlations and trade-offs among kinematic and dynamic rupture parameters. We find that the slip-weakening distance increases in average linearly with distance from the hypocenter, which can be interpreted as an apparent feature substituting the effect of (umodeled) significant off-fault yielding. We also try to include the slip-strengthening (SS) area at shallow depths and find that it is an unnecessary feature for this event.