The Study Of Data-driven Interbed Multiple Suppression Method

Multiple suppression is one of the problems that are long-standing and not resolved well in seismic exploration. Multiples include surface-related multiples and interbed multiples. At present, most study on multiple suppression is aimed at the surface-related multiples, and its theory and algorithm are very mature. Surface-related multiple suppression has been a conventional process flow in the seismic data processing. With increasing sophisticated technology of surface-related multiple suppression and the increasing requirement for accurate imaging on complex blocks and complex reservoirs, the study on the method of interbed multiple suppression has become a hot research on multiples.In this paper, we first make a detailed description on the generation, characteristics, classification and identification methods of multiples, and we reviews the three methods of multiple suppression based on wave equation. Then we describe the theory and realization process of surface-related multiple suppression in detail. On this basis we expand the surface-related multiple suppression to the interbed multiple suppression. This method needs seismic data acquired through the geometry in which shot points are located on the surface and receiver points are located on the multiple generators or receiver points are located on the surface and shot points are located on the multiple generators. In other words, this method needs wavefield continuation. In this paper, we get the required data directly from the measured seismic data to circumvent wavefield continuation. This method is completely data-driven without knowing the model information. According to different implementations of interbed multiple suppression, it can be divided into boundary-related method and pseudo boundary-related method. The key to this method is the identification and extraction of multiple generator interface or pseudo boundary. After getting the interbed multiple prediction method, we study multiple subtraction methods. We derive the least square method based on minimum energy and use it for a simple application. Aiming at the defects of the least square method, this paper introduces two ways of improvement and uses these three ways to test seismic data for analysis on their advantages and disadvantages. Finally we use model data and Pluto data to test the method in this paper and prove the effectiveness and practicability of the method used by this paper.