時間相依走時層析成像反演之探討：以2013年瑞穗地震(ML 6.4)為例

Abstract

Using seismic travel time tomographic method, changes of seismic wave velocity in crust associated with moderate-to-large sized earthquakes could potentially be resolved with the tomographic images before and after the occurrence of an earthquake. However, unequal ray distribution during two time periods can also cause the artifacts in resulting of temporal seismic wave velocity changes. Thus, to selected the two travel time datasets with equal ray distribution, we considered the location distance of each event pair, the differences of epicentral distance and the gap of azimuths for each station-source pair. We then conducted a series of tests to investigate the temporal-spatial resolution of tomographic results, especially for examining how the unequal ray distributions can influence the three-dimensional VP and VS crustal structures. We demonstrated the feasibility of time-dependent tomographic method by applying it to image the velocity changes before and after an earthquake occurred on 2013 October 31, in Rueisuei in eastern Taiwan (ML = 6.4), which was well recorded by the Central Weather Bureau Seismic Network (CWBSN) stations, Taiwan Strong Motion Instrumental Program (TSMIP) and Broadband Array in Taiwan for Seismology (BATS) stations. In our tomographic results, quite different patterns were found between the results of equal and unequal type of ray distributions. Through investigating the checkerboard resolution tests, resolution maps and discrepancy between the checkerboard-like models within two time periods, the reliable region were revealed. In our results, in the source region of the 2013 Rueisuei earthquake, a positive velocity changes (the model difference, ΔM) in VP (~2-10%) appeared surrounding the source location at 15 km depth. This anomaly of velocity changes downwardly extended to ~10-30 km away in the northern region of source at 20 km depth with a tendency in the NNE-SSW direction, which rupture of the fault plane propagated. In the north-eastern region of aftershocks, a negative ΔM (~3-8%) appeared in the north-eastern region at 10 km depth and downwardly stretched to 15 km with decreased amplitude (~1-2%). For ΔM in VS, two negative ΔM (~5-10%) appeared in the north-eastern region and the west side of the source location, and these anomalies downwardly extended to 15 km depth with decreasing in intensity. In the region where most aftershocks located, a strong positive anomaly (~5-10%) was shown with the same trending in the NNE-SSW direction just located at the major slip area. This anomaly downwardly stretched and further localized in the north where there was ~30 km away from the source location at 20 km depth. We suggested that the observed ΔM in the source region were mainly caused by increasing in stress due to the rupture of the mainshock and elastic rebounded of crust. It is worth noting that not only will the distribution of rays affect the results of time-dependent travel time tomographic inversion but also different weighting value of the arrival-pickings from travel time data in the pre-seismic and the post-seismic periods bias the structures in tomographic inversion procedure. When the comparable resolution before and after a moderated-to-large sized earthquake can be achieved with identical ray distribution, the travel time tomographic method is then able to resolve the seismic wave velocity changes.