Joint Inversion For Source Parameters Of Moderate Earthquakes

Moderate earthquakes with magnitude between5-6.5strike frequently at plate boundaries, also they are not rare in intraplate regions. These events could bring damages to human settlements with low-level seismic fortification. Moreover, when these earthquakes happen to be provoked by active faults directly underneath the metropolises, or the source is extremely shallow, it could result in great hazard and large casualties, even in developed countries. According to the Omori law and Gutenberg-Richter law, generally a M7event is accompanied with moderate aftershocks. The possible secondary damages in the disaster area should not be ignored. Therefore, rapid access of source parameters for these events is crucial for rescue operation as well as seismology studies such as regional tectonics, rheological behavior and stress state of rocks.There are two levels of source parameters. The first is origin time, location and intensity represents the temporal/spatial locations and energy scale for earthquakes under the point source assumption, these are the basic information of sources. With more in-depth analysis of seismic waveform data, more detailed description of earthquakes sources, such as the geometric characteristic of fault, slip distribution, and source time functions, can be revealed. In Chapter1of this dissertation, we will review the development of earthquake sources model, and discuss their applicability for moderate events. Also the history of source parameters inversion methods will be introduced. Centroid moment tensor model will be emphasized since its widely application, together with some comparison between different approaches.We choose CAP (Cut and Paste) method to invert moderate earthquake sources. In Chapter2, we developed the CAPjoint method for joint inversion from both local and teleseismic waveforms. We introduced the basic theory of CAP method and structure and manual of the software, as well as some illustrations in different regions. We verified the accuracy and efficiency by compare the results with other agencies. Also we discussed the application range of joint inversion and reliability under the situation of sparse networks. This software has been published online and it’s open source.Chapter3provided an example of CAPjoint application:we inverted the source parameters of2010M6.4Kaohsiung, Taiwan earthquake with seismic data from global broadband seismic networks and Broadband Array in Taiwan for Seismology. The joint inversion result showed that the best double-couple parameters are:nodel I:strike 317°, dip31°,rake51°; node Ⅱ:strike181°,dip66°,rake111°,which is a thrust type with left-lateral component. The focal depth is24.0km and moment magnitude Mw is6.24. The relatively deep centroid depth is correspond with the deep local Moho and complex tectonic and stress history. We used bootstrap static method to analyze the results with resampling. The results showed that the sensitivities of different parameters shows different patterns with local or teleseismic data, which support the necessity of joint inversion.Focal depth is a crucial source parameters. It’s very important to understand the seismogenic mechanics and early access of earthquake damages. At present, most parameters such as focal mechanisms and seismic moment are reliable from earthquake catalogs, such as GlobalCMT and NEIC. However, the precision of focal depth is still a big problem. In Chapter4, we analyze the possible sources of centroid depth misfit from teleseismic waveform inversion, taking2013Lushan earthquake as example. We explored the possible influence factors by two aspects:the completeness of data and inversion parameters. Through analyzing the inversion result under different parameters such as source duration and band filter, we found that these parameters used in inversion have obvious impact on the inverted centroid depth, which is more sensitive than other source parameters. In particular, widely used long period waveform filter by current catalogs could introduce large bias of centroid depth inversion, although it suppresses the effect of small structure in the propagation media, which make the inversion more stable for other source parameters. Thus, we must be careful in weighting such issue.Seismological methods play a key role in studying source parameters. However, besides the issue of centroid moment mentioned in Chapter4, using seismology approach can hardly get high precision horizontal locations as well as the ruptured fault discriminated from the conjugate one. Geodetic methods have been applied in many research fields of seismology. Geodetic data can be divided into two categories:the real-time dynamic data which has the ability to record the seismic waveforms, such as high-rate GPS signals; and the static displacement records, such as InSAR. These data have been widely applied in a series of seismological researches. There are some studies showing that different structure models can affect the inverted slip distribution along finite faults, which can be ignored when the differences are small, so the isotropic half-space assumption is widely used to improve the computational efficiency. However, in Chapter5, we demonstrated that the very slow near-surface shear velocity structures cause substantial bias on earthquake source inversion if simple crustal models are used. We found that up to a factor of2overestimate of the seismic moment could be generated in regions like Bohai basin. Therefore, the very slow sub-surface velocity has to be taken into account before accurate source inversion can be applied.Source parameters inversion studies using InSAR data have been developed rapidly in recent years. In1993, Massonnet et al. derived surface displacement deformation of an earthquake from the InSAR data for the first time. The subsequent works demonstrate that the inversion applied by InSAR technology has multiple advantages over seismology:precise horizontal locations (Ground Truth), accurate discrimination of true ruptured fault, etc. As an important method to update the current catalogs, InSAR data produced a lot of new source parameters for existing earthquakes. However, when we checked these results and compared with the parameters derived by seismology studies, we can discover that some InSAR-generated parameters have low accuracy:surface deformation are more affected by shallower slip, which make the centroid depth systematically shallower, also the dip and rake angles are tend to approximate to the shallower part of the faults. Therefore a joint inversion including seismic data is necessary to get more precise parameters. In Chapter6, we compared the synthetic waveforms generated by InSAR inversion with true recorded data of seismic stations and found disagreement in the SH waves at multiple teleseismic stations and local surface waves, which implied that the InSAR results has some difference with seismological source parameters. We demonstrate a joint inversion method of inverting InSAR with both local and teleseismic data based on neighbourhood algorithm, which utilized the constrain of seismology data to get more accurate inversion results. The comparison of synthetic waveforms showed that our joint inversion improved the precision of parameters compared with InSAR inversion and GlobalCMT catalogs. In the future, this method can be incorporate more geodetic and surface observation data, such as high-rate GPS, to improve the inversion accuracy furthur.