Simultaneous Multi-Phase Traveltime Inversion For Complex(Including Anisotropic) Media Using Irregular Model Parameterization

Seismic traveltime tomography is one of the important methods to understand the internal structure of the earth.However,in actual situations,the distribution of seismic stations is extremely uneven,and earthquakes mostly occur at the edges of each plate,which causes the uneven distribution of seismic source locations.Traditional tomographic imaging methods are mostly based on regular grids(or cells).Such a model parameterization is unable to be differentiated according to the distribution of the uneven data.On the other hand,in the exploration of mountainous oil and gas fields,irregularly undulating surface(or irregular underground interface)needs to be accurately processed,otherwise the forward modeling(ray tracing)algorithm is difficult to ensure the computational accuracy required in the inversion.In response to these situations,we in this paper extend the traditional traveltime tomography algorithm based on regular grids(or cells)to the one based on irregular grids(or cells)to better solve the above practical problems.In the 2D case,the triangular cell is used for model parameterization,and while in the3 D case,the tetrahedral cell is used for model parameterization.In the inversion method,we mainly discusses the inversion algorithm of conjugate gradient method to solve the constrained and damped least squares problem.Based on irregular model parameterization in2D/3D complex media,we developed the traveltime tomographic method by using first arrivals,the joint inversion method by combining later reflected arrivals,and the simultaneous inversion method of dual(velocity and reflector geometry)model parameters.Subsequently,we extended the above algorithm to 2D/3D anisotropic media,and completed: 1)Simultaneous inversion of elastic(or Thomsen)parameters in anisotropic TI(VTI,HTI,TTI)media with direct arrival times;2)Simultaneous inversion of of elastic(or Thomsen)parameters in anisotropic TI(VTI,HTI,TTI)media by combining later arrival times.Aiming at the different sensitivity of the partial derivative of each elastic(or Thomsen)parameter in traveltime in anisotropic media,we in this paper adopt the method of normalizing the sensitive kernel function to ensure that the parameters are balanced updated simultaneously.On the other hand,in order to overcome the multiple solutions in the simultaneous inversion of q SV waves,a multistage inversion strategy is proposed to avoid the ambiguity problem in the simultaneous inversion.The results of the forward and inversion numerical simulations show that the parameterization of the irregular grids(or cells)model can ensure the computational accuracy of the forward modeling in the complex model,and is suitable for effective imaging of the complex model,including irregularly undulating surface(or underground irregular interface).Finally,in order to test the practicability of the inversion algorithm under the parameterization of the irregular model,we carried out the inversion imaging of the actual traveltime data,including:(1)the 3D traveltime tomography in the loess area A using direct P arrival times;(2)the 3D traveltime tomography in the loess area B using direct P arrival times;(3)3D P-wave velocity structure of Shanxi and its nearby region using natural earthquakes.The results show that the inversion algorithm in this paper has strong practicability.