3d crs analysis: a new data-driven optimization strategy for the simultaneous estimate of the eight stacking parameters

Abstract

We devised a data-driven strategy for the simultaneous estimate of the eight CRS traveltime attributes, solving a global non-linear minimization problem without the need of computing gradients. The essential elements in this Eni research and development activity are the following: a conjugate-direction method supported by well known convergence properties and an iterative line-search implementing the strong Wolfe-Powell rule for the control of the steplength. The resulting algorithm can reach very good solutions in presence of many local minima.