Clinoform interpretation for stratigraphic features utilizing machine learning methodology
Mikael Kvalvaer, Clifford Kelley, Abdul Rahman Rahbi, Johannes Stammeijer, Nadeem Balushi · 2020
Petroleum Development Oman (PDO) is doing exploration work in the Habshan region. To obtain a better understanding of the subsurface opportunities much time and resources is used to map a large clinoform system across a platform area. For scale, it has taken one geoscientist about 6 months to map 20% of the Habshan clinoform system. The hypothesis is that with a tailored machine learning technology the same high quality interpretation can be ‘learned’ and achieved with a huge time-saving. With success of the pilot, PDO will have a prototype which shows that it will be possible to map the remaining part of the platform area in a time-efficient manner, allocating more time for map analysis, evaluation and maturation of any future opportunities. Clinoform features are common stratigraphic geometries encountered in carbonate and siliciclastic and basin margin data around the world. They typically present multiple challenges in their interpretation. Understanding the geology and the stratigraphic record defined by their depositional history and environment can be key in the pursuit of hydrocarbons. Gaining this knowledge will help sweet-spotting areas and defining prospect for exploration, reducing the risk in these complex depositional environments. We offer a Machine Learning methodology to analyse and interpret these stratigraphic features. Our problem challenge was; can machine learning tools automatically track and identify stratigraphic surfaces and features in clinoform systems. Overcoming and producing acceptable interpretations despite all of these limitations, was the target. Presentation Date: Tuesday, October 13, 2020 Session Start Time: 1:50 PM Presentation Time: 2:15 PM Location: 361A Presentation Type: Oral