Abstract： Subsurface fractures within many carbonates and unconventional resources play an important role in the storage and movement of fluid. The more reliably the detection of fractures could be performed, the more finely the reservoir description could be made. In this paper, we aim to propose a method which uses two important tools, a fractured anisotropic rock physics effective model and AVAZ (amplitude versus incident and azimuthal angle) inversion, to predict fractures from azimuthal seismic data. We assume that the rock, which contains one or more sets of vertical or sub-vertical fractures, shows transverse isotropy with a horizontal axis of symmetry (HTI). Firstly, we develop one improved fractured anisotropic rock physics effective model. Using this model, we estimate P-wave velocity, S-wave velocity and fracture weaknesses from well-logging data. Then the method is proposed to predict fractures from azimuthal seismic data based on AVAZ inversion, and well A is used to verify the reliability of the improved rock physics effective model. Results show that the estimated results are consistent with the real log value, and the variation of fracture weaknesses may detect the locations of fractures. The damped least squares method, which uses the estimated results as initial constraints during the inversion, is more stable. Tests on synthetic data show that fracture weaknesses parameters are still estimated reasonably with moderate noise. A test on real data shows that the estimated results are in good agreement with the drilling.
