| It is an important issue of modern seismology which used moment tensor concept in focal physics research. With the widely employment of broadband digital seismometers in recent 20 years, recorded waveform’s quality has significantly improved, as well as theories and methods for calculate synthetic seismograph have made considerable progress. These greatly promoted the study of physical processes on the source, especially for regional medium and small earthquakes.Because records on regional stations contain plenty of information about source and regional crust, it can be used to study focal mechanisms of regional medium and small earthquakes. The results help us to understand tectonic stress status in focal area and fault features; it is an important way to find out the progress of rupturing or breaking of rocks.In most recent years, researchers have made remarkable progress in the area of determination of the point source mechanism solutions and limited rupture parameters of moderate or small earthquakes by the usage of seismic moment tensor inversion method. Based on previous researches, this thesis systematically elaborates the moment tensor inversion method for focal mechanisms of regional earthquakes, and two methods for source parameters moment tensor inversion are investigated.In the first chapter, basic concepts of source parameters and moment tensor are stated briefly. Fundamental relationships between fault parameters and moment tensor solution are reviewed. Some features such as eigen values, eigen vectors, etc, and the decomposition of moment tensor with physical meaning are depicted. Subsequently, some moment tensor inversion methods are introduced and categorized, while general steps for moment tensor inversion are presented.In the second chapter, we first review the developing progress and the present status of regional moment tensor inversion, and discuss the trend of evolution in this study field. Then, we give a short, succinct statement of minimum rotation angle of double couple mechanisms, which is known as Kagan angle, and its’calculation and representation. Finally, take the Mw5.4 Feb.26,2010 earthquake in Taiwan offshore area for an example, we discuss the stability of deviatoric moment tensor solutions for regional earthquake using sparse network. The results show that:(1) for data sets from three 3-component stations, the solutions are mostly correct. With the station coverage not great than 30°, the solution’s kagan angle to GCMT result is little than 20°; while the station coverage is greater than 30°, the solution’s kagan angle within 15°. The better station coverage we used, the smaller departure from real solution. (2) Statistics of two-station combination solutions shows that,23.8% solutions’ kagan angle are not more than 15°, and 76.2% within 30°. It implies that combination of two 3-component stations can also retrieve relatively reliable solution in appropriate case,.Based on Langston’s decomposition theory, the 3rd chapter proposes a deviatoric moment tensor inversion method in the light of Minson & Dreger’s formula. Waveform’s envelop function which calculated from Hilbert transform is introduced to stabilize the inversion. The method is successfully tested on synthetic data. Numerical experiments result shows that the method has decent robustness and stabilization. Specifically:(1).While the velocity structure and earthquake location are exactly known, solutions will be accurate and stable (kagan angle<5°, DC% within±5%), despite of high noise level even reach 30% of origin signals. (2) While random phase arrival disturbance are limited in±3s, moment tensor solutions are still entirely stable. About two-thirds of solutions can be considered to be very close to the true solution (kagan angle< 15°, DC% within±5%); About 90% of solutions have a small deviation of the true one (kagan angle<30°, DC% within±10%). It shows that the method has good compatibility of crustal anisotropy and lateral heterogeneities. (3) The results of tests with various crust models show that, velocity model has important effect on inversion result. If 1D model used in general study works is the mean model of true crust structure, inversion can retrieve near-really solution. If model has little difference with the true model (no more than 10% error level), solution is close to the real one; if error level exceed 10% even reach 20%, solution will have notable bias. (4) Location error mainly affect determination of focal depth, accordingly cause spurious non-DC component in final solution. But with the constraint of pure deviatoric moment tensor, fault parameters retrieved from inversion result have little bias with the really one. (5) With the sparse data (3 stations at least) and worse station coverage(less than 60du), we can still get relatively correct solution.By means of a series of tests on different noise level, phase arrival errors, location errors, various crust models and stations distributions; it is proved that the method has decent robustness and stabilization.The method is applied to Oct.24,2010 Zhoukou-Taikang M4.7 event in Henan Province. The optimized solution retrieve by this method is according with that estimated from P wave polarities.In the 4th chapter, an inversion method is proposed to calculate synthetic spectrum of full waveform which fitted best with observations in frequency domain for source mechanism studies. Grid search method is adopted for searching fault parameters by correlating theoretical spectra with observations. Employing particle swarm optimization algorithm, stable and reliable solutions can be obtained by using this method in shorter time. It is shown with numeric test that, in the case of greater location error and worse station coverage, inversion result can get the solutions close to the true mechanism, with the depth near the really focus. Compared with waveform inversion performed in time domain, amplitude spectrum inversion running in frequency domain has faster compute speed and easier instrument response deduction. Because this method needs neither exact origin time, nor artificial alignment of waveforms, intermediate errors are decreased. The method has been applied to local and near regional recordings of the 17 May 2010, ML4.0 event in Bohai Sea. The result is in agreement with that obtained from P wave polarities. In addition, the study also shows that increasing high frequency components in inversion data, resolution on focal depth will be improved. However, by the restrictions of one-dimensional velocity structure model, it may weaken the fitting degree of the synthetic waveforms. We also estimate focal mechanisms for 14 earthquakes with ML≥4 in Shandong Peninsula and it’s adjacent sea area from 2003 to 2010, and have got the optimized DC solutions of those events.In the last chapter, the moment tensor solutions are retrieved for the earthquake swarm which occurred in November to December 2010 in Boshan mining area, Shandong Province, China. With local geological structure analysis, this chapter discusses possible sources and physical causes of the non-DC mechanisms in regional moment tensor solutions of small earthquakes. The results show that DC components in source mechanisms are higher in the beginning stage of the swarm, consist mainly of shear faulting controlled by tectonic stress. The following events have significant non-DC components, indicating tensile faulting. The DC components predominately present normal faulting, and P axes orient near vertically. The slip vectors of the swarm events are relatively stable. Referring to tectonic features near the epicenter, we conclude that the swarm is a result of subordinate faults motion affiliated to Wangmu Mountain fault, and high pressure pore fluid has played a crucial role in the earthquake swarm activities. |
PDF Full Text Request