Research On Seismogenic Structure Of Haicheng Fault And Upper Mantle Anisotropy In Western Sichuan

The Haicheng area is located in the transition zone between the Liaohe Basin and the Liaodong uplift,and is a region where small and medium earthquakes occur frequently in Liaoning Province.The Haicheng fault is the main seismogenic structure in this area and is the only Holocene hidden fault.The risk of major earthquakes is still high.Due to the dense population and special geographical location,the Haicheng area has long lacked good seismic data.At present,there is no research on the small-scale details of the seismogenic structure of the Haicheng fault system.The results of different tomography researchs are different in the velocity structure of the Haicheng area,and there is no research of waveform modeling to invert the velocity structure.The observation quality of the Liaoning Seismic Network has improved significantly in the past decade,providing reliable catalog data and waveform data,which is conducive to our study of the velocity structure and seismogenic structure of the Haicheng earthquake zone.In this paper,the differential evolution algorithm is used to invert the waveform data of the 2008 Haicheng ML4.8 and 2012 Gaizhou ML4.8 earthquakes,and an accurate crustal one-dimensional velocity model in the western Liaodong Peninsula is obtained.Based on the crustal velocity model obtained from the inversion,we used the seismic phase travel time data of the seismic catalog to relocate the earthquakes that occurred on the Haicheng fault in the past decade.Combined with the data of terrain and focal mechanism,we have the following new understanding of the Haicheng fault system:(1)The waveform inversion was used to obtain a one-dimensional crustal velocity model in the Haicheng area.There is a low-velocity layer in the middle crust,which may be related to the upwelling of the mantle.(2)The whole Haicheng fault can be divided into east and west sections with an interval of about 2 km.The two sections have the same dip direction but different dip angles.(3)From a small scale,there are two groups of echelon sub-faults arranged in different steps in the Haicheng fault system.The cause may be related to regional stress changes.The western Sichuan area is part of the Sichuan-Yunnan rhombic block.The western side is adjacent to the Qinghai-Tibet Plateau,the eastern side is adjacent to the Sichuan Basin,and the southern and northern sides are the South China block and Songpan-Ganzi block respectively.Therefore,the seismic activity in the western Sichuan area is strong and the fault structure is complex.Because there are few long-term stations in the western Sichuan area,the anisotropy results obtained by shear wave splitting in the western Sichuan area are relatively sparse.Whether the coupling of the crust-mantle motion mode in this area needs to be further studied.In this paper,the teleseismic SKS waveform received by the dense mobile Sichuan array is used to perform shear wave split analysis,and the cross-correlation method and tangential energy minimization method are used to solve the problem.The high-quality shear wave split measurement results are obtained.The understanding of the upper mantle anisotropy of western Sichaun is as follows:(1)The direction of the fast axis of most stations is basically near north-south direction,which is inconsistent with the movement direction of the crustal GPS horizontal velocity field.The crust and mantle in this area are decoupled.(2)The direction of the anisotropic fast axis of some stations is closely related to the direction of regional fault.(3)The overall delay time difference between different regions may be related to the propagation distance of shear waves and regional structure control.In this paper,some methods and improvements proposed in the research process of inversion velocity structure,seismic relocation and shear wave splitting are summarized as follows:(1)Waveform inversion of one-dimensional velocity structure based on differential evolution algorithm.This nonlinear global optimization method does not rely on a initial model,and the parameters converge faster and the results are stable.(2)The clustering method based on spectral clustering can effectively distinguish the linear clusters that are close in distance,and can improve the accuracy of cluster clustering before double-difference relocation.(3)The quality detection method of the shear wave splitting result based on the deep neural network can directly output the quality type of the original waveform,which greatly improves the data processing efficiency of the shear wave splitting process.