Data-Driven Approach to Eliminate Internal Multiples Based on the Marchenko Method for a Carbonate Field in the Middle East

Abstract

Abstract Due to shale-carbonate alternating successions with a gentle structure, primary reflections are severely interfered by various orders of strong and complex internal multiples from the overburden to reservoir levels in carbonate fields in the Middle East. These factors make the existing techniques ineffective, and still pose various geophysical challenges. This study aims to demonstrate an innovative scheme to eliminate internal multiples based on the Marchenko equation, originated from a basis of the inverse scattering theory. The extension of the Marchenko equation to surface seismic data enables a retrieval of upgoing and downgoing Green’s functions between a virtual source in a given subsurface location and receivers at the surface. Such a key feature in the Marchenko method is capable of identifying and/or eliminating internal multiples in a data-driven fashion, without the need for explicit subsurface models or numerical modelling. The technique estimates all the downgoing reflections from the overburden level, interfering primaries in the target level as internal multiples, which are subsequently subtracted from the input data to derive internal-multiple free data. To thoroughly assess the value of the Marchenko method, a numerical study was conducted. The realistic earth model that incorporates subsurface contexts in a carbonate field in the Middle East was derived to generate synthetic data through finite-difference modelling. The Marchenko-based internal multiple elimination was applied to synthesized data followed by reverse time migration. As true subsurface responses are known here, this trial properly revealed the effectiveness of the algorithm. For a comparison purpose, we also performed another processing flow using conventional internal multiple elimination that makes use of periodicity and velocity discrimination. The Marchenko scheme successfully estimates various types of internal multiples, such as over- and under-corrected multiples and even flat internal multiples, regardless of their periodicities, at once. Despite its ability to effectively suppress internal multiples, the final imaging results indicate that underlaying primary events are still preserved. The technique also leads to notable improvements in the imaging quality as compared to conventional processing flow. In addition to the effectiveness of the Marchenko method, we also describe a dedicated workflow adapted in this study to deal with strong and complex internal multiples. The unique advantage of the Marchenko scheme over existing approaches is its ability to predict internal multiples in a data-driven manner without precisely specifying multiple generators, assuming behaviors of internal multiples, or requiring prior knowledge of subsurface context. The Marchenko method can be also implemented for various ways such as redatuming, target oriented imaging and velocity estimation, source inversion problem, passive seismic, and electromagnetic, which will certainly provide several aspects of additional value.