Ai Seismic Interpretation: Challenges, Solutions and Applications

Summary Artificial Intelligence assisted solutions are reshaping the ways of seismic interpretation. AI solutions to challenging and time-consuming interpretation tasks can yield results of completeness and accuracy that are far out of the reach of most interpreters. The first real success of applying AI into the E&P workflows was the detection of faults in the seismic cube. A complete extraction of all horizons using AI, however, proved to be a much bigger challenge: a new solution had to be developed that is entirely aware of faults and geobodies and understands the interplay between these features and the horizons it is analysing. These AI solutions allow different workflows and improved analyses to be undertaken. AI networks enable the fast, accurate and unbiased analysis of dozens of volumes, allowing the human interpreter to focus on comparing and interrogating the results to understand their significance. This presentation will discuss some of the new workflows: how the best AI network can be selected for a particular task, how AI enhances our understanding of the structural uncertainty in a reservoir compartmentalised by faults, and how AI can preserve the knowledge of geoscientists and transfer to and use it in different projects and data sets.