Tuesday, 13 August, 2024 / Edition 20

School starts at different times around the United States, and our kids are heading back on Thursday. A tradition in Colorado (maybe elsewhere in the country too?) is to visit a water park just before school starts as a last little bit of fun and freedom before the kids buckle down and get back to the important task of focused learning. That’s what’s on the docket for me today: a day of fun in the sun with my kiddos!

Speaking of fun (but not sun.. we are not covering solar this week! 😎), drilling deeper into the mantle sounds like an absolute blast, and you know you’ve signed up for a good time when convolutional neural networks are involved 🤓 Let’s dive in!

Sarah Compton

Editor, Enspired

Boldly Drilling Where No Drill Has Gone Before

Vadim Sadovski/Shutterstock.com

Yes…you were supposed to read that title with the voice of Jean-Luc Picard in your head, while visualizing the Enterprise shooting off at light speed.

As geoscientists, we’re not used to being able to hold important pieces of the mantle in our hands at all. But recently, a team with the International Ocean Discovery Program (IODP) got more mantle than it bargained for: over a kilometer of core!

The drilling operation:

• Researchers aimed for a location where they knew the crust would be thin to maximize their chance of getting the most mantle.

• The team only expected roughly 200 meters of core based on what had previously been extracted. A 600 percent return isn’t too shabby.

• Although sampling the mantle is a hot topic, (literally and figuratively), there could be economic implications, too: the team also collected samples of micro-organisms to study the reactions that produce hydrogen and other molecules.

• Folks interested in the abiogenic theory of oil generation now find themselves with the best opportunity ever to study the upper mantle directly.

While new technology didn’t play a role here: this innovative discovery was possible using processes that might sound familiar:

• Research identified a location with the best chance of success.

• The best tool for the job was selected.

• Perhaps most important: some flexibility was built into the plan so that the team wasn’t forced to stop when they hit the “anticipated” core amount of 200 meters. They were able to continue drilling when it proved easier and faster than originally thought, and the results were truly Earth-shattering.

The future of this work is uncertain as NSF funding has run out, meaning the $72 million per year required to commission the research vessel JOIDES Resolution has dried up.

A message from AAPG Academy

Webinar Starting Soon

Cracking the Code: Use AI to Identify Hydrocarbon Accumulations in Noisy & Sparse Data

Register now to discover how: 

• Advanced AI deep learning techniques can optimize drilling operations in sedimentary volcanic basins

• You can reduce costs and operational risks by implementing these optimization techniques

• To implement techniques that will reshape and sharpen your approach to seismic interpretation

Geoscience Uses for AI

Boy Anthony/Shutterstock.com

While AI is having a bit of a “knock off the pedestal” moment in the investment world, geoscientists are finding some great uses for it that directly transfer to oil and gas.

A study from Scientific Reports caught my eye and integrates remote sensing, petrology, and field geology to identify lithological units.

What they did: Deep learning and convolutional neural networks (CNNs) integrated with old-school field geology to better identify lithological units.

Breaking it down: CNN is a type of deep learning algorithm that uses three-dimensional data for image classification and object recognition tasks.

Geophysicists might be most familiar with these algorithms, since they commonly support or run many oil and gas geophysical workflows.

• Automatic fault recognition algorithms can use CNN to train on synthetic seismic record data sets—or actual seismic data sets—to construct intelligent fault recognition models and automatically identify parameters, including the possibility of fault existence and dip angle.

• Phan and Sen combined several deep learning and AI algorithms, including CNNs, to perform pre-stack seismic inversions.

• Applications to well logs include predicting physical parameters and lithofacies categories.

The use of CNN in field work and mapping got me thinking about its applications for thin section work, point counts especially.

The bottom line: It’s good to keep our heads up and pay attention to work being done across all aspects of geoscience, not just what’s going on in petroleum. As demonstrated in my first newsletter, cross-pollination of ideas is often a driver of new technological and innovative discoveries!

👍 If you enjoyed this edition of Enspired, consider supporting AAPG's brand of newsletters by forwarding to a friend or colleague and signing up for our other newsletters here.

➡️ Was this email forwarded to you? Sign up for Enspired here.

✉️ To get in touch with Sarah, send an email to editorial@aapg.org.