How can artificial intelligence support high-resolution geological interpretation in one of the Middle East’s most prolific and complex deltaic reservoirs? This study delivers a detailed characterisation of the Lower Cretaceous Zubair Formation across the Sabiriyah and Raudhatain fields in North Kuwait, based on an extensive multi-well dataset comprising core sedimentological logs, borehole image data, and wireline logs. The approach is grounded in a consistent workflow that integrates robust sedimentological and sequence stratigraphic techniques, applied systematically across both fields, with advanced mathematical models, offering new levels of precision in reservoir characterisation.
A hierarchical sedimentological framework was developed to define lithotypes, borehole image facies, depositional packages, and genetic elements (GEs). This framework facilitated the correlation of key stratigraphic surfaces and depositional cycles, revealing the vertical and lateral heterogeneity of a fluvially dominated, weakly tidally influenced delta system. The analysis addresses key reservoir questions, including: What are the dominant depositional and architectural elements of the Zubair Formation? How do their geometries and spatial organisation control sandbody connectivity and fluid flow?
To complement the geological interpretation with a focus on uncored/non-imaged intervals, a supervised Artificial Intelligence (AI) model, trained on core-calibrated data, was used to predict genetic elements pragmatically grouped by lithological dominance. The model achieved up to 92% overall accuracy, with >97% confidence for sand- and mud-prone elements, while transitional facies showed expected uncertainty due to subtle log responses. Crucially, it delivers probabilistic outputs that enable uncertainty quantification and contribute to more reliable reservoir characterisation at the field scale.
This study demonstrates how a scalable and transferable workflow, grounded in geoscientific rigour and enhanced by AI, can deliver detailed, predictive insights into reservoir architecture, continuity, and heterogeneity. The consistent application of this integrated approach across a large dataset offers clear value for field development and reservoir management in complex subsurface settings.
Charlaftis, D., Kostic, B., Seksaf, A., Smith, R., Abd El-Aziz, S., Ramalingam, R., Chao, C. 2025. Integrated sedimentological, stratigraphic, and AI-based characterisation of the Lower Cretaceous Zubair Formation: deciphering geological complexities in Middle East deltaic systems. 2nd Edition AAPG/EAGE Petroleum Systems of the Middle East GTW, Kuwait City, Kuwait.
Keywords: Zubair Formation, Deltaic Systems, Reservoir Characterisation, Core-calibrated Supervised Machine Learning, Probabilistic Facies Prediction, Uncertainty Quantification.
