Reservoir-Scale Heterogeneity in CO2 Geological Storage: A Review of Characterization, Mechanisms, Prediction, and Management with Future Perspectives

Abstract The large-scale deployment of CO2 geological storage imposes stringent requirements on injectivity, subsurface space utilization, and long-term safety, making reservoir-scale heterogeneity a critical constraint on engineering feasibility and storage reliability. However, existing studies have primarily focused on individual heterogeneity elements, localized storage processes, or specific spatial scales, while systematic understanding of the coupling among multiple structural types, cross-scale transfer, and their impacts throughout the storage lifecycle remains limited. Therefore, this study synthesizes experimental, modeling, field, and ML studies, distinguishing their evidence sources and validation limits, to summarize the origins, types, characterization methods, and quantitative indicators of heterogeneity and to compare approaches for geological modeling, probabilistic prediction, and engineering regulation. The reviewed literature indicates the influence of geological architecture and connectivity on preferential migration and plume distribution. Quantitative effects on pressure, capacity, and trapping remain dependent on site conditions and model assumptions. Based on the identified limitations, recommendations are proposed in eight areas: evaluation standards, multisource data integration, structural fidelity, scale correction, probabilistic prediction, dynamic updating, zonal injection, and intelligent regulation. These recommendations are aligned with the major challenges identified in this review, covering standardized evaluation, structural representation, uncertainty-aware prediction, adaptive management, and engineering regulation throughout the storage lifecycle. This study reveals the differentiated roles and dual effects of heterogeneous structures and provides a basis for reservoir screening, capacity and risk assessment, injection optimization, and long-term management.