Uncertainty estimation of moment tensor solutions for the 2017 Mw 5.5 Pohang earthquake using hierarchical Bayesian inversion method

The Mw 5.5 Pohang earthquake in 2017 came into the spotlight in that it was an earthquake potentially triggered by a hydraulic stimulation in an enhanced geothermal system (EGS) project. One of the noticeable features is high non-double couple (non-DC) terms, which may indicate complexities in the source process and potential relationship with injections of water. It has been well-documented that the non-DC components sensitively vary based on the assumed conditions and data noise in the moment tensor inversion. However, testing the instability of non-DC parts is difficult in the conventional inversion methods due to the effect of regularization. In this study, we apply a hierarchical Bayesian method incorporated with the linear inversion method to better estimate the uncertainty of MT parameters. Our approaches are divided into three parts: the type of assumed data noise, allowed time shift, and frequency bands. Our results show relatively consistent patterns of source mechanisms, but the isotropic component is highly sensitive to different settings. In particular, it exceeds 20% at the lowest frequency band we used, while it remarkably decreases close to zero in other frequency bands. These findings highlight an unstable nature of non-DC components, suggesting a careful interpretation of the isotropic part as it expects volumetric changes attributed to the injected water.