Student Webinar by Diana Salazar Florez

Identification of faults and fractures in seismic data is crucial both at a large-scale, when understanding migration pathways in basin analysis, and at a small-scale, when characterizing reservoirs either for production or injection of fluids, for example, in enhanced oil recovery (EOR), geothermal energy, and carbon capture and storage (CCS). Failing to correctly identify faults and fractures in the rock may have consequences from losing capital, or control of fluid injected, to even triggering earthquakes. Seismic data and interpretation can help to optimize and enhance the visualization of faults and fractures in the subsurface by applying seismic attributes. Recent workflows have also started to implement machine learning techniques such as convolutional neural networks, probabilistic neural networks, and self-organizing maps to accelerate or facilitate this task. Most of these techniques rely in the integration of several seismic attributes at the same time, which are selected and calculated by the geoscientist, which makes them have an important role in the process, both for the initial steps as well as to validate the results performed by the machine. In this lecture, we summarize some of the most recent techniques that are being applied in academia and industry to identify faults in seismic data, and we review some examples found in the literature.