Non-stationary inversion of seismic data for fracture compliances in azimuthally anisotropic media

Fractures and tectonic settings cause azimuthal anisotropy in reservoirs. Recognizing the fracture model from the seismic data is a useful tool for identifying the productive zone in reservoirs. We applied azimuthal velocity analysis in seismic processing to improve the image quality and to estimate the anisotropic model parameters. Using azimuthal residual moveout analysis, the direction of azimuthal anisotropy in the reservoir was predicted, and it was found that the results are consistent with fracture orientations obtained from image logs in the reservoirs. Bayes’ theorem and a cascaded procedure in least-squares inversion, which matched observed amplitudes to linearized Zoeppritz equations, were used to estimate the elastic moduli as a first step, and the normal and tangential fracture weaknesses were estimated in a second step. Laboratory experiments were carried out on core samples to validate the first-step inversion results. It was found that the propagation wavelets varied in space and reflection time, and so a library of extracted wavelets in the time–frequency domain was used for seismic inversion. Maps of the computed fracture fluid index and estimated fracture weaknesses were used to help to visualize the role of fractures in reservoir productivity, and revealed a consistency with the seismic peak frequency attribute in identifying zones of highly compliant fracture fill. The estimated fracture model demonstrates a good fit with the fractures seen in the available core samples and implies that the fracture fluid index is a useful attribute for determining the productive zones in the reservoir.