Chaveste, Alvaro, Rada, Fabian, Chaparro, Thomas, Patel, Kailesh, and Alan Mcinally. “Practical Machine-Learning Lithofacies Prediction and Porosity Modeling for Faster Deepwater Exploration Decisions.” Paper presented at the ADIPEC, Abu Dhabi, UAE, November 2026. doi: https://doi.org/10.2118/236164-MS
Introduction
Deep-water exploration is commonly constrained by sparse well control, complex depositional architecture, and high uncertainty in predicting reservoir presence and quality. In these settings, seismic data provide the principal subsurface constraint, and seismic-driven workflows are essential for reducing exploration risk and predicting reservoir properties such as porosity, clay volume, and fluid-related behavior. We evaluate an integrated workflow that combines machine-learning-based seismic interpretation with geostatistical porosity modeling for stacked oil-bearing Campanian sandstone reservoirs offshore Suriname. Multi attribute seismic classification at the sampling-interval scale, using Self-Organizing Maps (SOM), is calibrated with a discrete architectural-elements log that groups sand, shale, and mass-transport-complex (MTC) facies. This calibration generates both a discrete lithofacies volume and lithofacies probability volumes. These products are then used as geologically meaningful soft conditioning data in a geostatistical framework for porosity modeling. The workflow begins with seismic and well-data conditioning to ensure consistent sampling and reliable calibration. Six instantaneous seismic attributes are selected using Seismic-To-Well Mutual Information (SWMI, patent pending), which ranks attributes according to their statistical relationship with well logs. The selected attributes are classified using SOM at the 2 ms seismic sampling interval, producing high-resolution volumes that capture reservoir connectivity and depositional heterogeneity in thin-bedded deep-water systems. Cross-tabulation between SOM neurons and the lithofacies log enables the prediction of sand, shale, and MTC facies, along with their associated probabilities. Integration of these seismic-derived facies and probability volumes with effective porosity as hard well data within a Sequential Gaussian Simulation framework produces porosity models that honor well observations while following seismic-driven depositional trends. Machine-learning-derived geobodies are also extracted to delineate channel and lobe geometries. By making lithofacies and porosity prediction easier to interpret and faster to update, the workflow aims to improve geological consistency, support identification of higher-quality reservoir zones, and provide a practical basis for prospect evaluation and exploration decision-making in deep-water settings.
Keywords
Machine Learning, Deepwater Exploration, Lithofacies Prediction, Porosity Modeling, Seismic Attribute Analysis, Self-Organizing Maps (SOM), Seismic-to-Well Mutual Information (SWMI), Geostatistical Modeling, Reservoir Characterization, Offshore Suriname
Citation
Chaveste, Alvaro, Rada, Fabian, Chaparro, Thomas, Patel, Kailesh, and Alan Mcinally. “Practical Machine-Learning Lithofacies Prediction and Porosity Modeling for Faster Deepwater Exploration Decisions.” Paper presented at the ADIPEC, Abu Dhabi, UAE, November 2026. doi: https://doi.org/10.2118/236164-MS
1. Chaveste, A., R. Roden, and T. Smith, 2025. A novel, fast, and efficient machine learning approach for lithofacies prediction: A lithology identification case study in the Niobrara Formation. The Leading Edge, 44, no. 9, 728–737.
https://doi.org/10.1190/tle44090728.1
2. Chopra, S., and K. Marfurt, 2007. Seismic Attributes for Prospect Identification and Reservoir Characterization. SEG Geophysical Development Series, no. 11.
https://doi.org/10.1190/1.9781560801900
3. Deutsch, C. V., and A. G. Journel, 1997. GSLIB: Geostatistical Software Library and User’s Guide, 2nd ed. Oxford University Press.
4. Leal, J., R. Jerónimo, F. Rada, R. Viloria, and R. Roden, 2019. Net reservoir discrimination through multi-attribute analysis at single sample scale. First Break, 37, no. 9, 77–86.
https://doi.org/10.3997/1365-2397.n0058
5. Pedregosa, F., G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. VanderPlas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and É. Duchesnay, 2011. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12, 2825–2830.
https://doi.org/10.48550/arXiv.1201.0490
6. Roden, R., T. Smith, and D. Sacrey, 2015. Geologic pattern recognition from seismic attributes: Principal component analysis and self-organizing maps. Interpretation, 3, no. 4, SAE59–SAE83.
https://doi.org/10.1190/int-2015-0037.1
7. Shannon, Claude E., 1948. A Mathematical Theory of Communication. Bell System Technical Journal, 27, no. 4, 623–656.
https://doi.org/10.1002/j.1538-7305.1948.tb00917.x
8. Smith, T., and S. Treitel, 2011. Introduction to self-organizing maps in multi-attribute seismic data. Presented at the Geophysical Society of Houston.
9. Vinnels, J., and H. Kombrink, 2023. The Suriname-Guyana Basin.
https://geoexpro.com/the-suriname-guyana-basin
