NETL researchers will share their internationally recognized geo-data science and artificial intelligence/machine learning (AI/ML) expertise through an invited talk and other sessions during the American Geophysical Union (AGU) Fall Meeting, held in New Orleans, Louisiana, Dec. 13-17.

Kelly Rose, geo-data scientist and technical director for the Lab’s Science-based Artificial Intelligence/Machine Learning Institute (SAMI), will share her group’s work blending geoscience with AI and ML during a session at 1:45 p.m. ET, Wednesday, Dec. 15. This innovative work is advancing energy, environmental justice, climate, data management and resource characterization applications.   

Rose’s invited talk, titled “Improving Prediction of Subsurface Properties Using Geoscience Informed, Multi-Technique, Artificial Intelligence Approach,” will cover how advancements in data science and AI techniques have significant potential to improve analysis and prediction of subsurface properties, even in areas with limited pre-existing data.

“AI can help scientists consume and appropriately find, extract and utilize larger amounts of data and knowledge than a single person or team alone can handle,” Rose said. “Also, these AI tools and capabilities can be utilized to help identify patterns and trends faster and more thoroughly than traditional methods afford. When paired with a geologist or team’s expertise of the subsurface, these AI-informed models provide new ways to explore and analyze Earth science systems. This supervised approach is often key when using AI for complex, uncertain systems like the subsurface.”

NETL’s SAMI-affiliated experts will present several poster sessions throughout the week on geo-science and AI/ML topics, including offshore geohazard and climate modeling; use of AI to forecast energy infrastructure longevity and risks to prevent failures like spills and emissions; data search, transformation and curation to improve access and reuse of subsurface data; AI modeling to improve subsurface reservoir and property predictions for carbon storage, induced seismicity, and other applications.

A full list of SAMI-related sessions follow. Clicking the links will take you to the AGU website to read an abstract and see when each presentation is occurring.

• On the Predictability of Loop Current Eddy Shedding Events and Unexpected Links to the Brazil and Guiana Currents
• AI/ML Integration for Accelerated Analysis and Forecast of Offshore Hazards
• Exploring Subsurface Data Availability on the Energy Data eXchange
• Leveraging Data Ecosystems to Support Earth Science Research for Decarbonization
• Integrating Disparate, Big Data for Applications of Machine Learning and Geospatial Models to Forecast the Remaining Lifespan of Offshore Oil and Gas Infrastructure: Optimizing Production and Mitigating Risk
• Improving Prediction of Subsurface Properties Using a Geoscience Informed, Multi-technique, Artificial Intelligence Approach (Invited)
• Science-based Artificial Intelligence and Machine Learning (AI/ML) Institute (SAMI) — Accelerating Cross-disciplinary AI/ML for Applied Geoscience, Energy, and Environmental Challenges
• Development of Machine Learning Models for Full Field Reservoir Characterization
• Shear Failure Prediction Using a Physics-guided Deep Learning Framework
• Decline Curve Analysis in SACROC Unit Using Long Short-Term Memory (LSTM) Network
• Comparison of Supervised Machine Learning Image Segmentation Algorithms for Petrophysical Characterization of a Saturated Porous Medium
• Reconciling Reservoir Simulation with Rock Physics Models
• Subsurface Characterization using Deep Learning Approaches
• A Framework for Data-Driven Solution of Linear Poroelasticity Using Continuous Conditional Generative Adversarial Networks
• Multiscale Segmentation of Rock Images and Ensemble Approach for Deep Learning Methods
• Deep Learning Assisted Inverse Modeling and Forecasting for Large-Scale Geologic CO₂ Sequestration
• Development and Evaluation of Deep Learning to Predict Multiphase Flow Behavior During Geological CO₂ Sequestration Injection and Post-Injection Periods
• Deep-Learning Seismic Inversion for CO₂ Saturation Monitoring and Uncertainty Quantification
• Geophysical Monitoring for Geologic Storage of CO₂
• Illuminating Faults and Stresses Using Deep Learning-Enabled Seismology for Injection-Induced Seismicity in Decatur, IL
• Optimizing Unconventional Wellbore Stimulation Using a Data Based Model
• Optimization of Stimulated Reservoir Volume (SRV) Prediction Using a Machine Learning (ML)Based Multilevel Feature Selection Technique

The AGU Fall Meeting unites more than 25,000 attendees from more than 100 countries in the Earth and space sciences community to share scientific findings and identify innovative solutions.

Register for the event here, and check back next week for a news story covering NETL’s geologic and environmental systems presentations for the AGU Fall meeting.

NETL is a U.S. Department of Energy national laboratory that drives innovation and delivers technological solutions for an environmentally sustainable and prosperous energy future. By leveraging its world-class talent and research facilities, NETL is ensuring affordable, abundant and reliable energy that drives a robust economy and national security, while developing technologies to manage carbon across the full life cycle, enabling environmental sustainability for all Americans.