Anisotropic Parameter Estimation from PP and PS Waves in 4-Component Data

Abstract

The estimation of anisotropic parameters in the shallow subsurface becomes increasingly important for 4C seismic data processing in order to obtain accurate images in both time and depth domain. I focus on two approaches to evaluate anisotropy in seismic data: using P-wave data and PS-converted (C-wave) data. To gain better insight into the accuracy and sensitivity of anisotropic parameters to for instance layering and compaction gradients, I undertake numerical modelling studies and verify the results with full-wave modelling as well as findings from the real data from a 4C data set from the Alba field. The focus of this thesis is on vertical transverse isotropy (VTI) which widely occurs in marine sediments and cannot be neglected in seismic processing. P-wave data alone cannot constrain the vertical velocity and the depth scale of the earth model for a VTI medium. Therefore, the joint inversion of non-hyperbolic P- and converted wave (C-wave) or S-wave data from long offsets has been suggested. I carried out a detailed analysis of the resolution and accuracy of non-hyperbolic moveout inversion for P-, S- and C-waves for a single VTI layer in two parts. First, I introduce the concept of the inherited error delta inh as a measure of the possible resolution of the moveout approximations for the different wave types. The range of this error stays constant regardless of the magnitude of the anisotropic parameter for each wave type. Second, I analyse the accuracy of non-hyperbolic moveout inversion. I find that for anisotropy parameter eta the error of estimation from C-wave data is in most cases about half that from P-wave data. Inversion of non-hyperbolic S-wave moveout data does not resolve the anisotropy parameter due to the presence of cusps in the data. The study is then extended to a multilayered medium considering only P- and C-waves. The results confirm the findings from the single layer case. Furthermore, I investigate phase effects on parameter estimation for P- and C-waves. It is suggested that eta estimated from C-wave data gives a better description of the anisotropy found in a medium than the eta values picked from P-wave data. To verify the above findings near-surface effects are studied on the 4C data from the Alba field and accompanied by a full-waveform modelling study. I find that the picked eta values from P-wave data are distinctly larger than the eta values from C-wave data and also larger than the eta values from VSP data. The full-wave modelling study shows that picked eta values from P-wave data may account for influence of structure such as velocity gradients in the near-surface and are influenced by high velocity ratios and phase reversals. Finally, I have carried out an integrated analysis of the Alba 4C data to demonstrate how seismic anisotropy can be estimated from 4C seismic data and how such information can be used to improve subsurface imaging. The results are presented in two parts. The first part deals with non-hyperbolic moveout analysis for estimating anisotropic parameters to gain improved stacked sections. The second part describes migration model building and final imaging. The models are evaluated by comparison with VSP data results and with a synthetic modelling study for three events of the overburden. The evaluation confirms that the anisotropy parameter obtained from C-wave moveout corresponds better with the VSP data than the values directly estimated from P-wave data.