Anisotropic Earth model building and some sources of uncertainty in the results

Abstract

Anisotropic earth model building (EMB) is a challenging task: even when we use best-quality modern workflows with non-seismic data and information to better constrain the problem, results are inherently non-unique. Methods for quantifying the uncertainty of earth models for seismic imaging exist and their successful application has been demonstrated in the past. All of them assume that the available model is a close representation of the true earth and is accurate enough, i.e. it explains at least all available seismic and borehole data. In addition, they rely on some extra knowledge and information for the area under investigation to be brought in to form some priors. This abstract discusses and illustrates the EMB sensitivity and uncertainty associated with: (1) the absence of enough complementary to surface seismic measurements; (2) inaccuracy in salt geometry and subsalt velocities, and (3) Q-compensation methods and parameterization. It emphasizes the need to validate earth models thoroughly before conducting uncertainty analysis and, if needed, to further update the models and ensure all limitations and assumptions of the data conditioning and EMB validation are factored into the priors for further uncertainty analysis.