Research On Elastic Wavefield Decomposition

Elastic wavefield decomposition,as one of the most important technologies in seismic data processing,utilizes the difference of P-and S-wave's physical properties such as polarization,velocity and propagation direction to conduct decomposition,then we can get pure P-and S-wave,which can avoid crosstalk between those different wave modes.As for VSP(Vertical Seismic Profile)data,down-going P-wave,down-going Swave,up-going P-wave and up-going S-wave can be obtained.This dissertation,aiming at wavefield extrapolation and VSP wavefield,mainly studies decomposition based wavefield extrapolation and decomposition based parametric inversion,and validity of the methods are verified by synthetic records and real seismic data.Firstly,this dissertation studies finite difference and ray tracing in forward modeling,which builds synthetic data for wavefield decomposition.Three decomposition based methods,i.e.,Helmholtz decomposition method,equivalent equation method and vectorial wavefield decomposition method in wavenumber domain,are introduced for wavefield extrapolation.Decomposition results of the three methods are shown and analyzed.Model tests show that,vectorial wavefield decomposition in wavenumber domain is better than the other two methods.In the study of decomposition based parametric inversion,the conventional method is based on local optimization or global optimization inversion.This dissertation proposes that the two methods should be effectively combined to achieve the goal of both accuracy and efficiency.Also,the inversion method for vertical wel s is extended to inclined wel s.Finally,the decomposition based parametric inversion is applied to synthetic data,real offshore and onshore VSP data,and satisfactory decomposition results are obtained.