Quentin Corlay successfully passed his viva with a thesis on detection of geobodies in 3D seismic with ML

Congratulations to Quentin Corlay with his most successfull viva as the result of 4 years of hard work as a NERC CDT in Oil and Gas PhD student supervised by prof. V. Demyanov, Dr Dave McCarthy (BGS), Dr. D. Arnold.
The examiners – Prof Guillaume Caumon (Uni Lorraine) and Dr Uisdean Nicholson (HWU) have assessed the work – Detection of Geobodies in 3D Seismic using Unsupervised Machine Learning – and found it of very highly quality and worthy of the Heriot-Watt McFarlaine medal nomination.

The thesis proposes  a novel, automated method for detecting geobodies in 3D seismic reflection data, helping to reduce interpreter bias and speed up seismic interpretation. A seismic geobody refers to a geometrical, structural, or stratigraphic feature, such as a channel, turbidite fan, or igneous intrusion. Geobodies are subtle seismic features, hard to pick, and their detection is challenging to automate due to their complex 3D geomorphology and diversity of shapes. Nevertheless, the detection and delineation of these structures are essential for improving the understanding of the subsurface as well as building a variety of conceptual models.
In his approach, Quentin, can rapidly interpret large volumes of 3D seismic volume using point cloud-based segmentation to identify geobodies of interest, even complex stratigraphic features like lobes and channels. By converting the 3D seismic cube into a 3D seismic point cloud, we reduce the volume of data to analyse, which in turn speeds up the detection process. The method allows the selection of a specific geobody and can retrieve geobodies based on their similarity to exploration targets of interest.
The method has been applied successfully to two modern 3D seismic datasets unknown at the time of the development (Falkland Basins) and two types of geobodies: fans and sill intrusions. We demonstrate that our method can scan through a large 3D seismic volume and automatically retrieve likely fan and sill geobodies in a very efficient manner. This approach can be used to scan through large volumes of 3D seismic, looking for a wide variety of geobodies.

Quentin has shared the finding of his PhD at a EAGE and IAMG conferences, NERC CDT annual conferences and of course with the GeoDataScience Group Industry Knowledge Share Days that provided insights into his work.

The seismic auto-interpretation tech developed by Quentin is open for new applications and collaboration with industry and universities.

Congratulations to Quentin and best wishes for his further career.