Tuesday, 5 August, 2025 / Edition 70

My softball career started behind the plate. Suddenly, though, I started getting worse at my literal one job and couldn’t catch the ball. Unbeknownst to my parents and me, my eyesight had slowly gone downhill, and I couldn’t see the ball, which makes it very hard to catch. An eye exam and a pair of glasses later, I was back to being a steel trap for wayward pitches. Improving what I could see improved my execution.

Our science often prevents us from properly observing different processes, but this week I cover one way geoscientists are improving that as well as a way standard diesel trucks can be easily upgraded with a bit of a bolt-on feature that is redefining “hybrid.” Let’s dig in.

Sarah Compton

Editor, Enspired

Using AI to Refine Seismic Observations and Fault Behavior

Lane V. Erickson/Shutterstock

Setting the stage: Despite decades of seismic monitoring, we still don’t fully understand the mechanisms behind the spatial and temporal distribution of earthquakes.

Scientists using AI can quickly and effectively analyze large data sets like those we have in instrumentally dense, seismically active areas such as California, China, or (in this instance) Yellowstone.

What they did: The Yellowstone caldera is both highly active and closely monitored, so a group of scientists from the University of Western Ontario turned it into their natural lab:

• They used deep learning techniques on a 15-year continuous seismic data set.

• The process was divided into two sequential steps: continuous records were analyzed to find seismic phases, which were then associated into grouped events.

• The team used a 3D S-wave velocity model, consistent with the presence of partial melt in the shallow crust, to build a high-resolution catalogue.

• Nearest-neighbors was used to identify all swarm-like sequences and to systematically characterize their movement through space and time, sometimes months, but sometimes years, and even over a decade.

Look what I found: The biggest result from the work is an incredible 86,000 events detected, a ten-fold increase.

With more detectable events, it was discovered there was movement along both established faults and new “rough” faults.

Swarms were found to be more of the norm, and while exact triggering mechanisms and relationships require more study, it is thought rapid fluid injection and pressure diffusion could be main causes.

Bringing it home: A big takeaway from this for me is the application of a new technique, deep learning, to the abundance of old data. The researchers noted that more work like this is needed in the future (hint, hint, fellow geos!).

Oil and gas applications: My mind immediately went to the numerous applications for oil and gas, even beyond analyses of microseismic data during completions, although I don’t know how often old data is being reviewed using newer AI techniques.

Seeking answers: We have questions about earthquake triggering in our own industry, but we also need to understand the behavior of pre-existing faults in our reservoirs and which are sealing vs communicative.

The fact this research was done on Yellowstone, a location of high heat and fluid flow in the US, means it could also have implications for geothermal energy. Understanding the dynamics of heat sources and country rock could help with efficiently utilizing the system.

Dig deeper: Go here to read the paper.

Sponsored

Harnessing Machine Learning for Enhanced Seismic Interpretation

Discover how ML-driven fault and horizon interpretation is delivering measurable improvements in subsurface imaging across complex geological settings including the North Sea, Gulf of Mexico, and offshore Brazil.

Plug-n-play Hybrid Long-haul Equipment

Vitpho/Shutterstock

Upgrade fatigue: For the generations that have had to swap records for cassette tapes, tapes for CDs, CDs for MP3 players, and finally just pay for music subscriptions on our phones, the idea of switching from gas to EVs may evoke some…feelings.

Yeah…no: Take those feelings and apply them to entire commercial diesel truck fleets and you get “hard pass” vibes.

At least one company sees you and has come up with a solution that doesn’t require a full swap, just an add-on.

Plug-n-play: Revoy has come up with a battery pack that fits between the trailer and cab of a Class 8 semi:

• Instead of waiting for the batteries to recharge, the packs are switched out in a process the company claims takes around five minutes, which is on par (maybe faster?) with filling a diesel tank.

• Swap stations will have clean bathrooms, places to grab food, and other amenities truckers are used to at typical truck stops.

• If the planned route is longer than the battery pack allows, the pack can be dropped off, and the truck can continue on its own diesel power.

• The battery packs come equipped with a motor that drives the wheels, providing an electric push to the cabs a-la the way trains have combined diesel and electric motors.

• The battery packs also have sensors and auto-intelligence corrections to prevent rollovers, jackknifing, and flips.

• The wheels don’t just “go,” but they have brakes to help with “stopping” as well.

Coming soon? The company had a rollout on Arkansas-Texas routes in October last year with further progressions planned this year, and given the media push, I’d expect to hear more news about that soon.

Justifiable questions: The X account The All Day Trucker had plenty of skeptics bring up some of the same thoughts I had regarding additional length, weight — which takes away from payload capacity — and maneuverability.

Some comments I don’t agree with include built-in obsolescence.

Crystal ball time: It might be the product manager in me, but I clearly see how this product can/will evolve into a fully-automated, self-driving electric cab.

Foot in the door: The market penetration method is also brilliant because electric Class 8 trucks are expensive, and the company claims this solution is at least break-even.

A link in the chain: The product’s physical placement is between the trailer and diesel cab, and it’s easy to see how the cab, not this battery pack, will be the item that gets disconnected in the chain during future iterations.

The company has not said as much, that I could find, but much like Teslas are collecting data for their self-driving feature, I would be shocked if this company is not doing the same thing.

Student becomes the master: And what better way for the machine to learn than shadowing a real driver?

Design for the win: The innovative product design is what gets my attention here, as it sidesteps two of the biggest issues with EVs: charge time (just swap it out) and range.

Geoscientists beware: I don’t see the EV revolution taking out fossil fuel-powered vehicles in the next five years, but innovations like this certainly make the horizon feel a bit closer and demand us geoscientists maintain skillsets that are transferable to all energy sectors.

To learn more about Revoy, go here, here, and here.

👍 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 or suggest a topic for Enspired, send an email to editorial@aapg.org.