Towards the Development of Fully Assisted Fault Interpretations Leveraged by Cutting-Edge Attributes and Cloud Computing

This paper presents an automated fault interpretation workflow that leverages the latest advancements in fault imaging, machine learning and cloud computing to extract and visualise fault surfaces from seismic volumes. By following this approach, it is intended to address the limitations shown by traditional manual fault interpretations that could be inaccurate and time-consuming. The first component of the workflow involves the generation of fault imaging volumes utilising machine learning models or semblance-oriented attributes. The second component entails the extraction of fault surfaces from the fault probability volumes by implementing a network analysis technique that establishes connections between nearby fault samples that possess similar attributes, thereby constructing individual fault surfaces. Cloud-based computing is used to parallelise and scale up computationally intensive processes. This enables efficient and faster workflow execution, irrespective of the size or complexity of the dataset. The efficacy of this automated approach is evaluated with real datasets, from offshore Australia and The North Sea. The results demonstrate the ability of the automated process to extract complex fault geometries faster and with increased accuracy. This represents a significant advancement in the field of subsurface structural interpretation, as it can offer a systematic solution to the challenges around fault interpretation.