Ground-motion simulation for the M_W6.1 Ludian earthquake on 3 August 2014 using the stochastic finite-fault method

The stochastic finite-fault simulation method was applied to synthesize the horizontal ground acceleration seismograms produced by the M_W6.1 Ludian earthquake on August 3, 2014. For this purpose, we produced first a total of 200 kinematic source models for the Ludian event, which are characterized by the heterogeneous slip on the conjugated ruptured fault and the slip-dependent spreading of the rupture front. The results indicated that the heterogeneous slip and the spatial extent of the ruptured fault play dominant roles in the spatial distribution of ground motions in the near-fault area. The peak ground accelerations (PGAs) and 5%-damped pseudospectral accelerations (PSAs) at periods shorter than 0.5 s estimated on the resulting synthetics generally match well with the observations at stations with Joyner-Boore distances (R_JB) greater than 20 km. The synthetic PGVs and PSAs at periods of 0.5 s and 0.75 s are in good agreement with predicted medians by the Yu14 model (Yu et al., 2014). However, the synthetic results are generally much lower than the predicted medians by BSSA14 model (Boore et al., 2014). Moreover, the ground motion variability caused by the randomness in the source rupture process was evaluated by these synthetics. The standard deviations of PSAs on the base-10 logarithmic scale, Sigmalog10(PSA), are closely dependent on either the spectral period or the R_JB. The Sigmalog10(PSA) remains a constant approximately 0.55 at periods shorter than 0.1 s, and then increase continuously up to ~0.13 as the period increases from 0.1 to 2.0 s. The Sigmalog10(PSA) values at periods of 0.1‒2.0 s show the downward tendency as the R_JB values increase. However, the Sigmalog10(PSA)​​​​​​​ values at periods shorter than 0.1 s decrease as the R_JB values increase up to ~50 km, and then increase with the increasing R_JB. Furthermore, we found that the ground-motion variability shows the significant dependence on the azimuth.