Bayesian inversion of microseismic moment tensors with 3D velocity models and characteristic waveforms

Abstract

Microseismic (MS) source moment tensor (MT) plays an essential role in understanding dynamic disasters such as rock bursts in underground mines. However, the commonly used full-waveform MT inversion result can be affected by the adopted simple velocity model and MS waveform tails. To handle this, we proposed a source MT Bayesian inversion method through a three-dimensional (3D) velocity model and characteristic waveforms. Firstly, broadband Green's functions (10–70 Hz) were constructed using the spectral element method within a 3D velocity model, which realized a high-frequency up to 77Hz waveform modeling in a mining region for the first time. Then, a short-term average/long-term average (STA/LTA) technique was employed to pick seismic phase arrivals, enabling the extraction of characteristic waveform segments surrounding these arrivals. Finally, an MT Bayesian waveform inversion method was developed using the Markov Chain Monte Carlo (MCMC) technique. Three typical synthetic events and five blasting events were adopted to validate the proposed method. Synthetic results show that an average cross-correlation coefficient is more than 90 % when the signal-to-noise ratio (SNR) is larger than 10, representing a 5 % accuracy increase compared with that using a homogeneous velocity model. The MTs for the five blasting events achieved the average waveform fitting accuracy of 85.20 % and an isotropic (ISO) component of more than 50 %, showing satisfactory inversion results. Furthermore, two typical MS events were analyzed. The MS event 2 near the fault F350 shows a 26.43 % compensated linear vector dipole (CLVD) component with strike and dip angles aligning with the fault orientation. In contrast, the MS event 1 near fault F310 reveals a more complex focal mechanism affected by mining disturbances. It shows that mining disturbances can impact the CLVD component of MS events and offer valuable insights into mining-induced fault slips.