Unravel Faults on Seismic Migration Images Using Structure-Oriented, Fault Preserving and Nonlinear Anisotropic Diffusion Filtering

Abstract

A high-resolution subsurface structural image can provide valuable information for optimizing well placement in geothermal exploration and production. A major challenge in imaging complex structures of geothermal fields is how to clearly reveal faults and fractures on seismic images. Land surface seismic data from geothermal fields are usually very noisy, resulting in noisy and low-resolution seismic migration images and, therefore, it is difficult to accurately detect faults/fractures on seismic images, particularly on 3D seismic migration images. We develop a structure-oriented, fault-preserving and nonlinear anisotropic diffusion filtering technique to greatly improve the quality of seismic migration images to facilitate fault interpretation. We employ a fast explicit diffusion scheme to significantly accelerate the computation of the nonlinear anisotropic diffusion, and employ computed coherence information from a seismic migration image itself to constrain the anisotropic diffusion process in space. The resulting parallel algorithm achieves structure-oriented, fault-preserving filtering accurately within minutes of computational time even for a large 3D seismic image volume. More importantly, the method simultaneously enhances continuous geological layers and discontinuous structures, such as faults and fracture zones, on seismic migration images. We apply our method to the 3D seismic migration image produced using 3D surface seismic data from the Soda Lake geothermal field, and obtain a greatly improved, high-resolution 3D seismic image volume that unravels several major and minor faults in this area, which are mostly invisible on the original seismic migration image. These faults are consistent with existing geological models and can be validated using the existing wells.