Study On New Method Of Small Scale Seismic Velocity And Structural Imaging

With more and more abundant seismological observation means and more and more intensive observation stations,the boundary between traditional exploration seismological method and natural seismological method is becoming more and more blurred.With the development of observation methods,the traditional seismic imaging methods based on ray theory gradually show disadvantages such as low computational efficiency and poor resolution.This paper focuses on some basic problems in body wave imaging methods,such as improving the computational efficiency,improving the accuracy of imaging results,stabilizing the imaging results and preventing from entering local minima.It includes the following four aspects:1.For the ray theory based tomography method,when the inversion grid is fixed,its computing efficiency decreases linearly with the increase of the number of geophones.In practice,the inversion grid is often increased with the increase of the number of geophones,which not only reduces the operational efficiency,but also makes the inversion process more unstable.Travel time tomography based on the adjoint state method can solve the problem of operation efficiency and stability.In order to prevent the inconsistent normal directions outside the undulating surface from leading to unreasonable calculation results,we set the geophone within the model region rather than on the boundary.In the process of theoretical derivation,we directly add an imaginary infinite velocity boundary to the theoretical boundary,on which the gradient of time field can be zero.However,in the actual calculation,because it is outside the calculation boundary,this operation is not involved,and the stability and accuracy of calculation are not affected.Finally,this method is applied to active source tomography in Sanjiang area.A clear fault boundary and high resolution velocity structure of the upper crust are obtained.It provides seismological basis for the study of rock structure and metallogenic mechanism of Sanjiang fold belt.2.The basis of the travel time tomography method based on the functional equation in the previous section was derived under the assumption of the high frequency approximation of the wave equation.The travel time tomography based on the wave equation is closer to the more real physical process.However,in the traditional wave equation-based travel time tomography,the accuracy of the seismic source wavelet directly determines the accuracy of the travel time measurement,and then affects the final inversion results.We develop a wave equation traveltime imaging method independent of source wavelet,which converts the absolute traveltime measurement from the seismic source to the geophone into the relative traveltime measurement from the adjacent reference geophone to all the geophones,thus avoiding the problem of the inaccuracy of the final traveltime measurement from the source wavelet.We also discuss the active source observation mode and the passive source observation mode respectively,and compare the method with the double difference method.It is concluded that the method is a special case of the double difference method.We have studied various cases of shear rupture source.For the case of source wavelet position,the accurate travel time residual cannot be obtained based on wave equation travel time cross correlation method.Finally,we applied this method to land data collected in Beiya ore concentration area of Sanjiang metallogenic belt in Yunnan.The inversion velocity model shows the effectiveness of the method and indicates that the source of the gold deposit is located in the east of the orebody.It provides some geophysical indication for deep ore prospecting in the future.3.The gradient of full waveform inversion can be decomposed into a smooth tomographic component(low wave number part)and a relatively sharp imaging component(high wave number part).Most waveform inversion processes need to start with a smooth velocity model and gradually update the details to avoid the well-known periodic jump problem.Scattering Angle filtering can be solved from two aspects of wave field and gradient.On the one hand,the field can be operated,that is,explicitly solving each forward and backward wave field solving the propagation direction of the wave field at each time or frequency,and then solving the scattering Angle of the wave field.Or use the travel time field to calculate the travel time table of each grid point in space.Then according to the direction of the gradient of the travel time field as the propagation direction of the wave field,the Angle between the two propagation directions is solved as the scattering Angle of the wave field.On the other hand,gradients can be manipulated directly by filtering them directly in the wavenumber domain.In this paper,a binomial expansion method is proposed to deal with wave fields from two aspects,which can enhance the imaging conditions of energy mode and obtain smoother gradients in theory.In the aspect of processing gradient directly,we improve the singular value problem of filter factor.Finally,we apply this strategy to a Marine seismic survey data,and we get a more detailed velocity structure based on the original superposition velocity.It provides a certain data support for the next step of geological drilling.4.All the above tomographic imaging studies are based on the information of the first arrival.In actual observation,the subsequent scattered waves also exist in a lot of geophysical information.The interface information can be obtained by stacking them according to the travel time field.However,due to incomplete observation system,random noise interference,inaccurate velocity model and other reasons,there will be inconsistent results of using different waveforms to offset.In order to suppress the difference between P-wave scattering imaging and S-wave imaging,we adopt a joint imaging condition in the phase domain.The effectiveness of the proposed method is proved by the test of synthetic model.In order to better display the structural discontinuities surrounding the San Andres Fault Depth Observation(SAFOD)in Southern California,we extended the generalized Radon Transform(GRT)method.We used dense observatory stations to record abundant local microearthquakes and used a combination of scattered P and scattered SH waves to image the vertical fault near SAFOD.We use this hybrid imaging condition to extract the common features of scattered P and SH images.Although the combined imaging may lose some resolution compared to the P and S imaging results separately,it can produce more reliable results when using passive seismic imaging.Our final joint imaging results show that the results based on the passive source of natural earthquake are consistent with the surface fault characterization.The fault structure of the Parkfield area in San Andreas was characterized.