Research On Adaptive Seismic Phase Arrival Picking And Spatiotemporal Rupture Process Inversion Of Source

Seismic source locating and spatiotemporal rupture inversion using seismic waveform data have important theoretical significance for understanding the mechanism of earthquake occurrence and analyzing regional seismicity.Fast and accurate earthquake locating and earthquake intensity maps based on rupture information play important guiding roles in earthquake prediction and early warning,disaster prevention,emergency rescue and other practical work.Based on the phases(P-wave and S-wave)recorded by seismic stations,using the multistation locating method to infer the spatial coordinates of the initial rupture point is a common source locating method.However,the limited picking accuracy of seismic phase and the uncertainty of velocity structure model often lead to low accuracy of source location.In order to improve the picking accuracy of seismic phases,a variety of automatic phase picking methods have been proposed to reduce the picking error caused by the manual subjectivity.At present,most picking methods seldom consider the different characteristics of signal-to-noise ratios of the local and teleseismic waveforms,making the phase picking lack sufficient stability.In terms of locating methods,due to that different earthquakes have different types of available observation data,how to select appropriate locating methods based on seismic observations is an urgent problem in earthquake research.The inversion of the spatiotemporal rupture process of the source based on seismic data is an effective way to quantitatively understand the kinematic characteristics of an earthquake.Aiming at the problem of unstable solution caused by many unknown parameters in the model,the finite fault inversion method manually constrains the maximum rupture speed,sets the sub-fault source time function,etc.,improving the inversion stability by reducing the number of unknown parameters.However,improper constraints will cause the slip model to deviate from the actual rupture process,and affect the effectiveness of earthquake prevention and mitigation and post-earthquake rescue work.Therefore,how to set appropriate constraints on the basis of minimizing artificial constraints is a major problem in investigating the rupture process of seismic sources.In this paper,the research on improving the locating accuracy of seismic source has been carried out from two aspects:improving the accuracy of seismic phase picking and selecting appropriate locating methods.In order to improve the accuracy of seismic phase picking,a new adaptive picking method based on the sliding window is proposed to analyze the characteristics of the waveform spectrum variation,and the first arrival of the seismic phase can be accurately picked up by identifying feature mutation time.The paper uses theoretical and actual seismic wave data to carry out picking experimental tests comprehensively.The experimental results of theoretical and actual seismic phase picking show that this method can effectively improve the accuracy and stability of phase picking.For selecting locating method,in consideration of the coupling relationship between the source location and the velocity structure,the double-difference tomography method which jointly inverses for the velocity model and the source location is introduced to improve the locating accuracy.In order to verify the effectiveness of this method,taking the 2019 Ridgecrest earthquake sequence as the research object,the seismic events are relocated based on the abundant near-seismic data in Southern California area.An accurate three-dimensional velocity model of the studied area is obtained while improving the accuracy of source location.For earthquakes that lack nearseismic data,such as the 2017 Mw7.7 Commander earthquake and its aftershocks,the teleseismic double-difference tomography method is introduced,extending the type of observation data from near-seismic range to tele-seismic range,to obtain accurate seismic source locations and regional three-dimensional velocity model.This paper takes the source locating as the research object,and analyzes the influence of different factors on the locating accuracy of the source from the aspects of the quality of the phase arrival data and the locating method.Since the pick-up accuracy of the P-wave and Swave phases directly determines the accuracy of the source location,this paper innovatively proposes a new type of seismic phase picking method based on various automatic and semiautomatic phase picking methods.The adaptive seismic phase picking method analyzes the changes of the spectral characteristics of the seismic signal in a specific time window,and accurately identifies the sudden change in frequency energy as the seismic phase arrival,finally improving the picking accuracy of the seismic phase arrival.The phase picking results of simulated waveforms and actual seismic waveforms show that this newly proposed seismic phase picking method has high efficiency and robustness while improving the accuracy of seismic phase picking.This paper introduces the back-projection method to obtain the earthquake rupture process by directly and linearly operating the teleseismic waveform.This method does not need to calculate Green's function,nor does it need to add constraints such as sub-fault source time function,maximum rupture velocity,etc.,which improves the efficiency.In order to verify the effectiveness of this method,the 2017 Mw7.7 earthquake that occurred in the Comandor Island area is taken as the research object,different frequency bands have been utilized to back project the temporal and spatial distribution of coseismic radiation energy to analyze the frequency-dependent rupture behavior of this earthquake.To further improve the objectivity of imaging results,the source region was divided into three sub-regions based on its unique geological location and regional tectonic characteristics,and the aftershock calibration back-projection method has been used to correct the P-wave travel time table of each sub-area.On basis of the calibrated time table,the coseismic high-frequency energy radiation process closer to the actual rupture process can be obtained by back-projection method,and the rupture speeds at different stages have also been estimated.The backprojection results based on aftershock calibration show that the entire rupture process can be divided into three stages.The area where the super-shear rupture occurs has a good spatial consistency with the tectonic stress mutation zone.The change of tectonic stress reflects the heterogeneity of the velocity structure to a certain extent,resulting in the rupture speed variation.In order to overcome the problem that the aftershock correction back-projection method cannot quantitatively obtain the spatiotemporal slip distribution on the fault plane,this paper introduces the iterative deconvolution stacking inversion method to develop a rupture inversion system.This system effectively combines the finite fault inversion method and the back-projection method,extracting the sub-fault source time function from the regional waveform to obtain the spatiotemporal rupture process on the fault plane.This method does not need to preset the shape of the sub-fault source time function,the maximum rupture speed,etc.,which improves the objectivity and efficiency of the inversion result.In order to evaluate the inversion ability of the system,based on the abundant strong motion data in the California region and the velocity structure model that is derived by Tomodd method,this paper inversed for the rupture process of the Mw6.4 foreshock and Mw7.1 mainshock in the Ridgecrest earthquake sequence in July 2019 to quickly obtain the rupture models of the two earthquakes.To verify the reliability of the inversion system,for the Mw7.1 mainshock,the rupture process of the earthquake has been derived based on the same set of waveform data using different filtering frequency ranges and velocity models to analyze the influence of the inversion parameters on the rupture model.The research results show that the obtained rupture models are basically same,indicating that the inversion system has strong stability.