AI-Enhanced Reservoir Characterization for Sustainable Geological CO₂ Storage

Abstract

Geological carbon dioxide (CO₂) storage is a promising strategy for reducing greenhouse gas emissions and supporting long-term energy-system decarbonization. However, the effectiveness and safety of CO₂ storage depend strongly on accurate reservoir characterization, particularly the understanding of geological heterogeneity, porosity, permeability, fluid distribution, and subsurface transport processes. This study examines the application of artificial intelligence (AI) and machine learning (ML) to enhance reservoir characterization and improve the prediction and management of geological CO₂ storage.
AI-based approaches can integrate diverse geological, geophysical, petrophysical, and monitoring datasets to identify complex reservoir patterns and improve estimates of storage capacity and injectivity. Their integration with numerical and reactive-transport modelling further enables efficient prediction of CO₂ migration, trapping mechanisms, pressure evolution, and long-term reservoir behavior. AI-assisted real-time monitoring can also support early detection of leakage risks and dynamic optimization of injection operations. The study highlights the potential of physics-informed and hybrid AI frameworks to improve reliability, computational efficiency, and sustainability in geological CO₂ storage.