AI-Driven Environmental Assessment and Optimization of CCUS: From Subsurface Modeling to Risk Management

Geological storage of CO₂ in depleted reservoirs, deep saline aquifers, and other subsurface formations is a critical pathway toward global carbon neutrality. Yet deploying storage at scale raises pressing environmental concerns, from the risk of CO₂ leakage into groundwater and surface ecosystems, to induced seismicity, to the challenge of ensuring long-term containment integrity across decades. Addressing these concerns requires accurate prediction of coupled thermal, hydraulic, mechanical, and chemical (THMC) processes in heterogeneous geological media, robust environmental monitoring, and decision-support tools that balance storage efficiency against environmental safety.

Traditional physics-based simulations of subsurface CO₂ behavior, while accurate, are computationally prohibitive for the real-time monitoring and scenario analysis that environmental risk management demands. Recent advances in artificial intelligence, including physics-informed neural networks, deep learning surrogates, and generative models, offer promising pathways to overcome these limitations, enabling faster-than-real-time prediction of storage performance, early detection of anomalous behavior, and multi-objective optimization that explicitly accounts for environmental constraints.
This Research Topic invites original research, reviews, and perspective articles that advance the integration of AI with physics-driven modeling to improve the environmental safety, sustainability, and management of CO₂ geological storage. We are particularly interested in work that bridges computational innovation and environmental application, demonstrating not only modeling accuracy but also how these tools inform environmental risk decisions, regulatory compliance, or storage site management.

Topics include, but are not limited to:
• AI-driven environmental risk assessment for CO₂ storage site selection, including leakage pathway prediction and groundwater impact modeling
• Physics-informed machine learning for simulating coupled THMC processes with application to storage integrity and containment assurance
• Surrogate and reduced-order models for real-time environmental monitoring, early warning, and adaptive management of storage operations
• Multi-objective optimization frameworks that jointly consider storage capacity, injection efficiency, environmental safety, and cost under uncertainty
• Data-driven approaches to long-term storage performance prediction, including caprock integrity, fault reactivation risk, and induced seismicity assessment
• Integration of subsurface modeling with surface-level environmental monitoring data (e.g., soil gas flux, groundwater chemistry, remote sensing) for comprehensive risk characterization
• AI-assisted regulatory and decision-support tools for environmental compliance in large-scale CCUS deployment
Scope boundaries: Submissions should demonstrate clear environmental relevance. Papers focused primarily on hydrocarbon production optimization, enhanced oil recovery without quantified environmental outcomes, or pure AI methodology benchmarks without direct application to environmental challenges in CO₂ storage are outside the scope of this topic.