Tomographic velocity model building for pre-stack depth migration

Abstract

It is now commonly accepted that prestack depth migration is the best method available for accurate imaging. However, this technique is only used in relatively extreme circumstances. The implementation of pre-stack depth migration is restricted by relatively high costs and turnaround times, as well as its sensitivity to errors in the interval velocity/depth model. For a pre-stack depth migration project to be successful, a reliable interval velocity/depth model is essential. Multiple iterations of pre-stack depth migration are often required to achieve a sufficiently reliable model. Obviously, this iterative nature contributes significantly to higher costs and turnaround times. Reflection tomography has been considered, in recent years, as a potential method for finding a reliable velocity model more easily. This method has also been held back due to its own generally high costs and turnaround. Reflection tomography normally requires interpretation of many reflectors on pre-stack data. This is time consuming and prone to error. Also, in efforts to simplify the overall procedure, the inversion itself has sometimes been compromised. A reflection tomography algorithm has been developed that does not require manual picking of events and does not compromise the inversion process. The scheme depends on a method of tracing reflected raypaths that does not require reflector definition and that allows automatic picking to be successful. It also utilises entropy constraints in a subspace inversion with stages of decreasing model space smoothing. The aim of this inversion is to help ensure that reliable velocity models are created and that local minima are avoided. This approach to reflection tomography is quite automated and has been applied successfully to many datasets from Australia and around the world. Two examples from offshore Australia demonstrate that the velocity models resulting from this reflection tomography algorithm clearly improve the results of pre-stack depth migration.