Thanh-Son Pham

SUMMARY Determining the seismic moment tensor (MT) from the observed waveforms with available Earth's structure models is known as seismic waveform MT inversion. It remains challenging for small to moderate-size earthquakes at regional scales. First, because shallow isotropic (ISO) and compensated linear vector dipole (CLVD) components of MT radiate similar long-period waveforms at regional distances, an intrinsic ISO-CVLD ambiguity impedes resolving seismic sources at shallow depths within the Earth's crust. Secondly, regional scales usually bear 3-D structures; thus, inaccurate Earth's structure models can cause unreliable MT solutions but are rarely considered a theory error in the MT inversion. So far, only the error of the 1-D earth model (1-D structural error), apart from data errors, has been explicitly modelled in the source studies because of relatively inexpensive computation. Here, we utilize a hierarchical Bayesian MT inversion to address the above problems. Our approach takes advantage of affine-invariant ensemble samplers to explore the ISO-CLVD trade-off space thoroughly and effectively. Station-specific time-shifts are also searched for as free parameters to treat the structural errors along specific source–station paths (2-D structural errors). Synthetic experiments demonstrate the method's advantage in resolving the dominating ISO components. The explosive events conducted by the Democratic People's Republic of Korea (DPRK) are well-studied, and we use them to demonstrate highly similar source mechanisms, including dominating ISO and significant CLVD components. The recovered station-specific time-shifts from the blasts present a consistent pattern, which provides a better understanding of the azimuthal variation of Earth's 2-D structures surrounding the events’ location.