Surface-Related Multiple Elimination, an Inversion Approach

摘要

In seismic processes such as migration and inversion, the input data is assumedto consist of primary reflections only. Therefore, multiple reflections should be removed in one of the pre-processing steps. It is not always realized that the free surface acts as the major multiple generating interface. Therefore, removing the surface-related multiples will solve the multiple problem to a large extend. The thesis starts with an overview of the most popular multiple elimination methods, being predictive deconvolution, velocity filtering and wave field based dereverberation. Next, the newly developed surface-related multiple elimination method is introduced. If the subsurface structures are complex and/or if velocity discrimination is not possible and/or if true pre-stack primary amplitudes are desired, the surface-related method is to be preferred. Furthermore, if the target is rather deep in the section (say below 2s) the multiple problem is often not defined by reverberations in the first layer, but by other surface-related multiples. In that situation, the surface-related multiple elimination method is the only method to be used. The surface-related multiple elimination method is derived as an inversion process on the seismic surface data. The results in the thesis show that the surface-related multiple elimination process (including pre-processing) solves the seismic signal-to-noise problem to a very large extent.