Crustal And Upper-mantle S-wave Velocity Structure In Southeastern Tibetan And Tengchong Volcanic Region

Joint inversion of receiver function (RF) and surface wave dispersion (SWD) is an important tool for the construction of S-wave velocity structure. The deformations and geodynamic characteristics in Southeastern Tibetan Plateau (SETP) and Tengchong Volcanic area (TCVA), the two hotspots in China geophysical community, are of significance for better understanding the deformation mechanism of Tibetan Plateau (TP) and the volcanism and seismicity in TCVA. On the other hand, the mannual model parameterization strategies in traditional RF inversion methods may cause intolerable misfits between inverted results and the true models. For this reason, the data-driven model parameterization methods are preferred for improving the reliability and resolution of the inverted models. This thesis firstly constructs the 3-D crustal and upper-mantle S-wave velocity structures in SETP and TCVA, and the geodynamic patterns are then discussed according to the velocity structures. For methodology, a new transdimensional Bayesian inversion method is proposed for 1-D S-wave velocity structure based on scattered body waves. The main contents and contributions of this thesis are concluded as following:(1) A crustal thickness and Poisson's ratio model are provided from RF functions by H-k stacking approach based on a large number of RFs recorded by the Yunnan digital seismic network and the temporal array deployed by MIT and Chengdu Institute of Geology and Mineral Resources (CIGMR) in this study. The results show that the crustal thickness changes from 30 km in the south to 62 km in the north, presenting strong lateral variations. The Poisson's ratio shows also strong lateral variations and its spatial distribution has clear partitioned characteristics. The northern part has relative higher Poisson's ratio than that of sourthern area.(2) The 3-D S-wave velocity structure of the crust and uppermost-mantle are constructed from joint inversion of RFs and SWD. The structure presents two continuously distributed low velocity zones (LVZ), which are inferred as two crustal flow channels. The two channels originate from the intersection of Xiao-Jinhe Fault and Xiao-jiang Fault, and then extend to southwest and southeast, respectively. They join together in southern Yunnan. By comparing with the results of crustal anisotropy and maximum shear strain, it can be concluded that the material migration in the crustal flow channels has the main constribution to the crustal anisotropy and significant influence on fault activity. The strength of crustal anisotropy decreases from northern to southern, suggesting the same trend of material migration speed and structural extrusion intensity in the crustal flow channels.(3) In order to explore the existence and spatial distribution of magma chambers, the S-wave velocity structure beneath TCVA is constructed from joint analysis of SWD and RFs recorded by Tengchong Earthquake-Monitoring Net. The results show that there are four magma chambers in the upper-to-middle crust and the magma chambers may also exist in the uppermost-mantle. The crustal magma chambers are not isolated; they connect with each other at different depths. For that reason, the lower crust has no magma chambers, however, the small-scale LVZs may represent the magma conduits, though which the deep materials migrate to the shallower part.(4) This study applies the modified anisotropic H-k stacking approach to study the crustal anisotropy in TCVA and the corresponding magma contributions. Six parameters, namely Vp/Vs, fast propagation direction, the strengths of P-wave and S-wave anisotropy, aspect ratio of intrusion body and melt fraction, are obtained in this method. Our anisotropic results reveal the complexity of the crustal anisotropy and its formation causes. The results are indeed preliminary that need to be improved in the further study.(5) In order to reduce the effect of manual model parameterization and overcome the difficulty in covariance matrix estimation of RFs, the present study proposes a new transdimensional Bayesian inversion method for 1-D S-wave velocity structure based on scattered body waves. This approach avoids the deconvolution procedure step in data processing and measures the difference between the observed radial waveform and the synthetic radial waveform generated by convolving the synthetic RF with observed vertical waveform. The misfit function denotes the direct waveform fit to the observed radial waveforms. The data-driven model parameterization strategy is selected to reduce the effect of artificial disturbance. Its feasibility is clearly illustrated using two synthetic tests having different types of noises added to seismograms. The real site application for obtainning the 1-D S-wave velocity structure underneath Tengchong seismic station shows that the new method has the ability to identify the small-scale velocity anomaly zones comparing with traditional inversion methods.