Full-waveform inversion (FWI) is capable of estimating the elastic properties of subsurface rocks including hydrocarbon reservoirs with high spatial resolution. Orthorhombic models are needed to adequately describe typical fractured reservoirs and formations under nonhydrostatic stress. Most existing applications of FWI to orthorhombic media, however, are limited to models with a horizontal symmetry plane. Here, we develop an elastic FWI algorithm for tilted orthorhombic media (i.e., models with tilted symmetry planes) which are typical, for example, for subsalt exploration in the Gulf of Mexico. We assume one of the symmetry planes to coincide with the underlying reflector so that the orientation of that plane can be estimated from migrated images. Orthorhombic models are described using nine velocity-based parameters: the P-wave (⁠VP0,VP1,VP2⁠) and the S-wave (⁠VS0,VS1,VS2⁠) velocities along the symmetry directions and the P-wave symmetry-plane normal-moveout velocities (⁠Vnmo,1,Vnmo,2,Vnmo,3⁠). To reduce nonuniqueness, FWI is constrained using lithologic information, such as the vertical P- and S-wave velocities and density obtained from well logs. A convolutional neural network (CNN) trained on the available well logs is used to predict the facies distribution from sparse borehole information. The developed algorithm is successfully tested on synthetic 3D wide-azimuth multicomponent reflection data. In particular, FWI is applied to an orthorhombic version of the structurally complex 3D SEG-EAGE overthrust model. The obtained high-resolution parameter fields confirm that our FWI algorithm is capable of estimating the essential elastic parameters from multicomponent surface data and improving migrated images of orthorhombic formations.

You do not have access to this content, please speak to your institutional administrator if you feel you should have access.