CSEG / EAGE Talk Series
The EAGE/CSEG/GAGGS talk series event is organized by the University of Calgary European Association of Geoscientists and Engineers (EAGE) / Canadian Society of Exploration Geophysicists (CSEG) Student Chapter, and the Graduate Association of Geology and Geophysics Students (GAGGS) in the Department of Geoscience.
We intend to offer our fellow students the opportunity to interact with experts from various fields in Geoscience and hear from them about the latest advances in our challenging and ever changing field. These talks will also provide the students with a vision of how the industry works and how projects are planned and developed.
The talks in the Winter 2016 semester will be conducted on Mondays or Tuesdays from 4:00-4:45 pm in the Science B building SB-142 at the University of Calgary. Here is our upcoming schedule.
Monday, March 6, 2017 – U of C – SB142 – 4:00pm
Is there a role for deep learning in geophysics?
Neural networks have been used for some time in geophysics to quantitatively predict rock properties from seismic data. Recently there has been tremendous progress in the field of machine learning thanks to a powerful new technique called deep learning. Applications of this are showing up in everyday life including self-driving cars, image recognition, translation, recommender systems and expert systems such as Watson. The key technologies that have enabled this progress to be made are deep learning, big data and increased computing speed largely due to graphic-processing video cards. This paper explores what is meant by deep learning and how we might modify our geophysical algorithms to incorporate these concepts.
This paper examines the supervised learning problem of how to establish nonlinear statistical relationship between the known well control and the seismic at these well locations. This relationship is then used along with the seismic to predict the rock properties between the well locations. Neural networks are often used to construct the nonlinear relationships. The paper starts off by reviewing how to establish these linear and multi-linear relationships, then nonlinear relationships using multi-layer neural networks and then discusses the extra complications of using deep neural networks.
Jon Downton is a Senior Research Advisor with CGG GeoSoftware where his primary focus is developing software in HampsonRussell, Jon has worked as a reservoir geophysicist, research geophysicist and research manager. His work has been focused on reservoir geophysics and the seismic processing associated with this. His current research interests are in the areas of machine learning, inversion, and seismic anisotropy. Jon obtained his Ph.D. from the University of Calgary in 2005 and his B.Sc. in Geophysics from the University of Alberta in 1985. Jon is a member of the CSEG, SEG, EAGE and APEGA and is a past president of the CSEG.
Monday, March 20, 2017 – Nexen Annex Discovery Theatre (+15level) – 4:45pm
Dynamics of Fault Activation by Hydraulic Fracturing in Overpressured Shales
Dr. David Eaton,
Department of Geoscience, University of Calgary
Fluid-injection processes such as disposal of saltwater or hydraulic fracturing can induce earthquakes by increasing pore pressure and/or shear stress on faults. Natural processes, including transformation of organic material (kerogen) into hydrocarbon and cracking to produce gas, can similarly cause fluid overpressure. Here we document two examples from western Canada where earthquakes induced by hydraulic fracturing are strongly clustered within areas characterized by pore-pressure gradient in excess of 15 kPa/m. By contrast, despite extensive hydraulic-fracturing activity associated with resource development, induced earthquakes are virtually absent in the same formations elsewhere. Monte Carlo analysis indicates that there is negligible probability that this spatial correlation developed by chance. A detailed analysis was undertaken within a 400-km2 region in Alberta, Canada where uniquely comprehensive data characterize dynamic interactions between well completions at 6 drilling pads (Bao and Eaton, 2016). Seismicity is strongly clustered in space and time, exhibiting spatially varying persistence and activation threshold. The largest event (ML 4.4) can be reconciled with a previously postulated upper bound on magnitude, only if the cumulative effect of multiple treatment stages is considered. Induced seismicity from hydraulic fracturing reveals contrasting signatures of fault activation by stress effects and fluid diffusion. Patterns of seismicity indicate that stress changes during operations can activate fault slip to an offset distance of > 1 km, whereas pressurization by hydraulic fracturing into a fault yields episodic seismicity that can persist for months.
Professor Dave Eaton received his B.Sc. from Queen’s University in 1984 and M.Sc. and Ph.D. from the University of Calgary in 1988 and 1992. He rejoined the University of Calgary in 2007 after an 11-year academic career at the University of Western Ontario. His postdoctoral research experience included work at Arco’s Research and Technical Services (Plano, Texas) and the Geological Survey of Canada (Ottawa). He is presently co-director of the Microseismic Industry Consortium, a novel, applied-research geophysical initiative dedicated to the advancement of research, education and technological innovations in microseismic methods and their practical applications for resource development. In addition to microseismic monitoring and induced seismicity, his current research is also focused on the lithosphere-asthenosphere boundary beneath continents. He has over 130 peer-reviewed publications.