Seismic images are only as good as the velocity models used to produce them. As we move from "easy oil" to "difficult oil" targets in sub-salt, sub-basalt and deep complex areas, we can no longer build the simple isotropic models of the past. To fully leverage the potential of new data types (e.g. wide azimuth and long offsets), we have to move to anisotropic imaging with vertical or tilted transversely isotropic (VTI or TTI) models in all geological provinces. Incorporating anisotropy increases our ability both to focus the seismic data and to accurately position the reflectors for drilling decisions. While these goals are achievable with anisotropic models, they are only met when geology information and data from boreholes are intimately incorporated into velocity model building from the very start. We discuss several different approaches for anisotropic model parameter estimation and we illustrate some of the possible strategies with examples from the Gulf of Mexico and West Africa.
Anisotropic depth imaging with VTI or TTI models has become the dominant industry practice in recent years. However deriving all the parameters needed to describe a transversely isotropic medium, throughout a 3D model, suitable for depth imaging is far from trivial. A TTI model requires five parameters: symmetry-axis velocity (VP0), Thomsen parameters e and d, and two angles describing the tilt of the symmetry axis. Over the last decade, we have developed many methods and techniques for deriving anisotropic parameters and building and updating VTI and TTI models for depth imaging. We have organized them in multiple workflows that enable us to pursue flexible approaches, optimally using all the information available in any situation. The three case studies included in this paper illustrate the importance of having a broad portfolio of tools and techniques that allow the design of fit-for-purpose model building strategies in areas with or without good quality well data control.
For all of the examples, we build anisotropic models using variations of the generalized workflow described by Zdraveva and Cogan (2011) and compare the results against images and models from previous imaging efforts with isotropic or simple regional VTI models. Because many anisotropic models will fit a single surface-seismic data set, we evaluate the final model correctness not only on image focusing, reduction of residual curvature, and ties to well data, but also on the geological, rock-physics and geomechanical plausibility of the model and image.
Because surface-seismic data alone do not constrain all anisotropic parameters, an important step of any anisotropic model building workflow is to evaluate Thomsen's parameters and build local anisotropic models around wells where additional information is available. Examples of such techniques include:
1D layer-stripping modeling and inversion with well data.
Localized tomography with well data (Bakulin et al., 2010a and 2010b).
Tomography with uncertainty analysis (Bakulin et al., 2009).
Trial-and-error scenarios in combination with 3D tomographic inversion with quick feedback loop.
