Loading...
Thumbnail Image
Publication

Marchenko imaging for 2D and 3D complex structures—with field applications to sub-salt and sub-basalt imaging

Jia, Xueyi (Alex)
Research Projects
Organizational Units
Journal Issue
Embargo Expires
Abstract
Seismic imaging is a geophysical technique that uses elastic waves to form images of geologic formations in the subsurface. Seismic imaging has become the most reliable diagnostic tool for modern hydrocarbon exploration and production. Conventional imaging methods, however, rely on the single-scattering assumption, which requires the recorded seismic data to not include multiples—waves that are reflected more than once in the subsurface before reaching the receivers. While the surface related multiples can be effectively suppressed with the Surface Related Multiples Elimination (SRME) method, the elimination of the internal multiples—multiples that are not surface related—remains challenging with current seismic processing techniques. The traditional workflow to mitigate the artifacts associated with internal multiples involves 1) predicting internal multiples, and 2) subtracting them from the acquired seismic data. This workflow requires accurate horizons of the multiple generators and a labor-intensive adaptive subtraction, which is usually performed in a least-squares sense and may damage primary events when primaries and multiples interfere. The Marchenko framework used in this dissertation is based on inverse problems in quantum physics. This framework consists of two steps. The first step is Marchenko redatuming, which allows one to use surface seismic reflection data to retrieve seismic responses (Green’s functions) between arbitrary points in the subsurface to the acquisition surface. The second step is Marchenko imaging, which utilizes the Green’s functions retrieved by Marchenko redatuming for imaging. These two steps provide a solution for resolving the issues associated with internal multiples and producing multiple-free images, without requiring horizons of multiple generators or performing adaptive subtraction. For my PhD research, I develop and investigate the 2D and 3D Marchenko framework for field data deployment and application. I elucidate the specific requirements for the two inputs of the Marchenko method: the seismic reflection data acquired on the earth surface and a background velocity model for estimating first arrivals from subsurface locations to the surface. To make the standard surface seismic data (which can be sparsely sampled in practice) useful for Marchenko redatuming, I consider forward interpolation methods to convert sparse surface data to densely and uniformly sampled data that corresponds to an equal number of co-located sources and receivers at the acquisition surface. I show that the background velocity model does not need to be known in great detail since the first-arriving wave needed by the Marchenko method is mostly determined by its travel time. A smooth velocity model is sufficiently accurate for such estimations. I demonstrate that the combination of Marchenko redatuming and imaging is robust with respect to erroneous velocity models. I extend the Marchenko redatuming algorithm to 3D seismic data by reformulating the Marchenko-type equations in 3D Cartesian coordinate system and develop an efficient 3D numerical implementation, in which I resolve the associated computational optimization and memory issues. Despite the idealized assumptions and the specific requirements for the input data within the Marchenko framework, I obtained two successful field data applications of the Marchenko method for imaging complex subsurface areas and propose a practical and effective workflow for processing streamer field data. With a Gulf of Mexico field dataset, I show that discontinuities along true reflectors—resulting from the destructive interference between primaries and the internal multiples due to salt layers—is resolved by the Marchenko method, which produces a clean and continuous sub-salt image. With an offshore Brazil dataset, I show that the artificial or nonphysical interfaces—resulting from the internal multiples that are generated by volcanic intrusion layers in the overburden—are adequately eliminated by performing Marchenko imaging.
Associated Publications
Rights
Copyright of the original work is retained by the author.
Embedded videos