Abstract EN:

Exploring the solid Earth for hydrocarbons, as social needs, is one of the main tasks of seismic imaging. As a domain of the modern geophysics, the seismic imaging by full waveform inversion (FWI) aims to improve and refine imaging of shallow and deep structures. Theoretically, the FWI method takes into account all the data gathered from subsurface in order to extract information about the physical parameter of the Earth. The kernel of the FWI is the full wave equation, which is considered in the heart of forward modeling engine. The FWI problem is represented as a least-squares local optimisation problem that retrieved the quantitative values of subsurface physical parameters. The seismic images are affected by the manifested anisotropy in the seismic data as anomalies in travel time, amplitude and waveform. In order to circumvent mis-focusing and mis-positioning events in seismic imaging and to obtain accurate model parameters, as valuable lithology indicators, the anisotropy needs to integrated in propagation-inversion workflows. In this context, the aim of this work is to develop two dimensional FWI for vertically transverse isotropic media (VTI). The physical parameters describing the Earth are elastic moduli or wave speeds and Thomsen parameter(s). The forward modeling and the inversion are performed entirely in frequency domain. The frequency-domain anisotropic P-SV waves propagation modelling is discretized by the finite element discontinuous Galerkin method. The full waveform modelling (FWM) is performed for VTI and titled transverse isotropic (TTI) media by various synthetic examples. The gradient of the misfit function is computed by adjoint-state method. The linearized inverse problem is solved with the quasi-Newton l-BFGS algorithm, which is able to compute an estimated Hessian matrix from a preconditionner and few gradients of previous iterations. Three categories of parametrization type are proposed in order to parametrize the model space of the inverse problem. The sensitivity analysis on acoustic VTI FWI method is preformed by studying the partial derivative of pressure wave field and the grid analysis of least-squares misfit functional. The conclusions inferred from the sensitivity analysis of least-squares misfit functional. The conclusions inferred from the sensitivity analysis are verified by FWI experimental on a simple synthetic model. The anisotropic parameter classes that can be well retrieved by VTI FWI are recognized. Furthermore, the acoustic VTI FWI is applied on the realistic synthetic Valhall benchmark for a wide-aperture surface acquisition survey. The anisotropic acoustic and elastic FWI are performed on the three components of ocean bottom cable (OBC) data sets from Valhall oil/gas field that is located in North Sea.