Multi-component seismic modeling and robust pre-stack seismic waveform inversion for elastic anisotropic media parameters

Consideration of azimuthal anisotropy, at least to an orthorhombic symmetry is important in exploring the naturally fractured and unconventional hydrocarbon reservoirs. Full waveform inversion of multicomponent seismic data can, in principle, provide more robust estimates of subsurface elastic parameters and density than the inversion of single component (P wave) seismic data. In addition, azimuthally dependent anisotropy can only be resolved by carefully studying the multicomponent seismic displacement data acquired and processed along different azimuths. Such an analysis needs an inversion algorithm capable of simultaneously optimizing multiple objectives, one for each data component along each azimuth. In this dissertation, I propose a novel multiobjective methodology using a parallelized version of NSGA II for waveform inversion of multicomponent seismic data along two azimuths. The proposed methodology is also an improvement of the original NSGA II in overall computational efficiency, preservation of population diversity, and rapid sampling of the model space. Next, the proposed methodology is applied on wide azimuth and multicomponent vertical seismic profile (VSP) data to provide reliable estimation of subsurface anisotropy at and near the well location. Prestack waveform inversion was applied to the wide-azimuth multicomponent VSP data acquired at the Wattenberg Field, located in Denver Basin of northeastern Colorado, USA, to characterize the Niobrara formation for azimuthal anisotropy. By comparing the waveform inversion results with an independent study that used a joint slowness-polarization approach to invert the same data, we conclude that the waveform inversion is a reliable tool for inverting the wide-azimuth multicomponent VSP data for anisotropy estimation. Last but not least, an anisotropic elastic three-dimensional scheme for modeling the elastodynamic wavefield is developed in order to go beyond the 1D layering assumption being used in previous studies. The method uses quadratic tetrahedral elements to discretize the model in space, Newmark family of time integration scheme to discretize in time, and a perfectly matched layer as the absorbing boundary condition. Various preconditioned conjugate gradient approaches were tested and implemented in the algorithm to improve the numerical efficiency of the finite element solver and the entire methodology was optimized to run efficiently in a parallel computing architecture. By computing synthetic seismic responses for a homogeneous medium and for a layered anisotropic medium and comparing results with other analytical and numerical solutions, the validity of the methodology was also verified for numerical accuracy.