2006年Mw 6.1台東地震引起淺層地殼速度隨時間變化和其成因探討

Abstract

The empirical Green’s function (EGF) which approximately corresponds to an impulse response can simply be retrieved from cross-correlation functions of seismic noises between two receivers in the assumption of diffusive wave field. It will increase much more available data than the traditional seismology underlying the view point that the noise is either source or receiver and is known as Passive Imaging Interferometry. In the study, we measure relative velocity variations from the coda part of recovered EGF, observe the temporal evolution, and have discussion on possible mechanisms leading to the changes of physical properties in the crust. The 2006 Mw6.1 Taitung earthquake occurred in the southern Longitudinal Valley and the collision suture zone between the Eurasian plate and the Philippine Sea plate and caused significant coseismic deformation. To detect potential variations in subsurface seismic velocities associated with internal strain change before and after the quake, we construct the GF from auto-correlation function (ACF) and cross-correlation function (CCF) from one-day long seismic noise after normalization. The stacking is applied to produce the long-term GF (Reference GF, RGF) and short-term GF (Current GF, CGF), respectively, for enhancing the coherent phases in EGFs. We use both stretch and moving-window cross-spectrum analysis to detect the relative time shifts of the late-arriving coda waves between CGF and RGF. A sudden drop in relative velocity perturbations is observed in the high –frequency (2-8 Hz) ACFs for the station about 7 km south of the epicenter where the two perpendicular faults may intersect and result in locally maximum coseismic slip deformation and widely-distributed aftershocks. The sustained damage in the regions of fault junction and slip concentration probably cause such a significant velocity disturbance. The CCFs (0.1-0.9 Hz) between pairs of stations also show moderate velocity decrease for the shorter interstation paths traversing the south and southeast of the epicenter, the area subject to the static volumetric strain with coseismic dilatational deformation. Plus, the post-seismic estimation (after October) increase to the pre-seismic value (before April), which shows positive relations between the regional precipitation.