AI-Assisted Subtle Faults Characterization Based on the Integrated Seismic Diffraction Imaging and its Application in M Oilfield, Middle East

Abstract Subtle faults play a key role in reservoir characterization. Due to subtle faults in carbonate reservoirs are often below seismic resolution, it is very difficult to identify them by conventional seismic attributes. Seismic diffraction imaging is a technique used in reservoir geophysics to enhance the imaging of subsurface structures. This technique can provide valuable information about the presence and characteristics of subtle faults features that might be missed by conventional methods. To improve the accuracy of subtle fault prediction, an AI-assisted subtle fault prediction technique was proposed based on integrated seismic diffraction imaging. There are 5 key steps: 1) Separated and enhanced diffraction signals, and generated integrated seismic diffraction imaging data, 2) Integrated FMI logging data to select sensitive frequency and azimuth for subtle fault identification; 3) Use AI-assisted subtle fault prediction method to obtain subtle faults efficiently; 4) Combining the geometric characteristics of subtle faults and FMI logging to analysis kinematic characteristics and interfering factors; 5) Optimizing subtle fault prediction results through noise suppression by orientation and dip controlled. The AI-assisted subtle fault prediction technique based on integrated seismic diffraction imaging seismic data has been successfully applied in the M oilfield in Middle East. Compare with the legacy seismic data, integrated seismic diffraction imaging seismic data was improved obviously in subtle fault identification. By integrating the fault orientation information provided by FMI logging, sensitive azimuths of seismic data were chosen for different subtle faults. AI subtle fault identification method based on deep learning can analyze large volumes of seismic diffraction imaging data more efficiently and accurately than traditional manual methods. AI algorithms can be trained to recognize patterns and features associated with subtle faults, allowing for the automatic detection and characterization of these features within seismic data. In the M oilfield, the faults are mainly formed under regional stress of torsion in the W-E direction. Two sets of subtle faults-oriented NW-SE and NEE-SWW developed, with fault dips ranging between 72°–85°. Subtle fault prediction is influenced by noise, but noise interference can be mitigated through azimuth and dip angle control, the accuracy of fault prediction was further improved. The integrated approach combines the high-resolution imaging capabilities of seismic diffraction imaging with the predictive power of AI algorithms, resulting in improved detection, characterization, and understanding of subtle faults in the subsurface. This technique has the potential to enhance reservoir characterization, optimize drilling and production strategies, and mitigate geological risks in various subsurface exploration and production operations.