Structural interpretation plays a vital role in petroleum exploration and production workflow (e.g. seismic interpretation, geological modeling, well placement, etc.). Conventionally, this has been achieved by manually picking faults on seismic lines which are labor-intensive and time-consuming especially in complex geological areas. More recently, Machine Learning techniques have proposed a lot of advances but they still remain a considerable challenge due to the requirement of time and huge datasets to train the software.
Hereby using PaleoScan™, we propose an innovative, data-driven workflow for detecting and extracting faults. The fault set comprised of 1400 faults was achieved from a cropped Chrysalids 3D seismic volume (10gb~ 700km2) (Figure 1) in a reasonable manner of time (45 mins to 1 hour for 1GB of seismic data). Hence, the time cycle for fault picking and PaleoScan™ Seismic Interpretation workflow was significantly reduced.