Comparative Evaluation of Passey ΔLogR, Statistical Regression, and Machine-Learning Models for Total Organic Carbon Estimation in parts of Gongola Basin, Nigeria

Geochemical methods for TOC determination required for source rock evaluation during exploration is accurate, but costly, time-consuming, and labor intensive, leading to alternative approaches. For the present study, a total of 829 laboratory measured TOC are available from Well A and Well B. This research evaluated and compared the efficacy of petrophysical, statistical, and advanced machine learning models for predicting Total Organic Carbon (TOC) in well A and well B in Gongola Basin using R2 and RMSE. The data was divided into 70% for training and cross-validation and tested on the remaining 30% percent. The results show that Tree-based machine learning models outperformed both Passey overlay and multilinear regression. The accuracy of Passey overlay, multilinear regression and regularized linear regression models of Lasso (L1) Regression, Ridge (L2) Regression, and LASSO LARS were low because they were based of linear relationship. This linear relationship is weak to very weak for well A and very weak to moderate for well B as observed from the study. Among the five high-performing tree-based models, Gradient Boosted Tree demonstrated the highest predictive accuracy score of 0.99 and RMSE value of 0.03 in all the feature groups in Well B except feature group 3 that yielded 0.98 and RMSE value of 0.04. It was followed in descending order of performance by Decision tree, extra tree, XGBoost and Random Forest. These models overcame poor linear relationship limitations by dynamically mapping non-linear relationship between TOC and log signatures / rock physics parameters. Global feature importance was evaluated by aggregating the magnitude of the SHAP values. Neutron porosity (NPHI), by resistivity (RES), and Lambda-Rho are the top three features considered by most of the models which geologically implies high concentration of hydrogen rich kerogen, thermal maturity of the source rock and variability in TOC being related to lithofacies changes. All the models predicted TOC better in well B than well A probably because well B exhibited better log-to-TOC relationship.

Machine learning, Total Organic Carbon, Gongola Basin and Passey ΔLogR

1. Akande, S. O., Ojo, O. J., Erdtmann, B. D., & Hetenyi, M. (1998). Paleoenvironments, source rock potential and thermal maturity of the Upper Benue rift basins, Nigeria: implications for hydrocarbon exploration. Organic Geochemistry, 29, 531–542. [Google Scholar] [Crossref]

2. Alshakhs, M., & Rezaee, R. (2017). A New Method to Estimate Total Organic Carbon (TOC) Content, an Example from Goldwyer Shale Formation, the Canning Basin. The Open Petroleum Engineering Journal, 10, 118-133. doi:10.2174/1874834101710010118. [Google Scholar] [Crossref]

3. Abiodun, K., Saleh, A. S., Eze, S., & William, S. (2018). “Source Rock Analysis Using Well Logs In Western Niger Delta,” Journal of Applied Geology and Geophysics, Volume 6, Issue 2 Ver. I (Mar. – Apr. 2018), PP 70-85. [Google Scholar] [Crossref]

4. Bolandi, V., Kadkhodaie, A., & Farzi, R. (2017). “Analyzing organic richness of source rocks from well log data by using SVM and ANN classifiers: a case study from the Kazhdumi formation, the Persian Gulf basin, offshore Iran,” Journal of Petroleum Science and Engineering, vol. 151, pp. 224–234. [Google Scholar] [Crossref]

5. Evans, J. D. (1996). Straight forward statistics for the behavioral sciences. Brooks/Cole Publishing. [Google Scholar] [Crossref]

6. Benkhelil, J. (1989). The origin and evolution of the Cretaceous Benue Trough (Nigeria), Journal of African Earth Sciences. 8, 251-282. [Google Scholar] [Crossref]

7. Dike, E.F.C., 1993. Stratigraphy and structure of the Kerri-Kerri Basin, Northeastern Nigeria. Journal of Mining and Geology, 29(2), 77–93. [Google Scholar] [Crossref]

8. Hamis, M. B., Yandoka, B. M. S., and Usman, M. B. (2024). Facies distribution, paleoenvironment and hydrocarbon potential of Kanawa Member of Pindiga Formation, Gongola Basin, Northern Benue Trough, Nigeria. Discover Geoscience, 2(1), Article 46. https://doi.org/10.1007/s44288-024-00052-y [Google Scholar] [Crossref]

9. Manaf, P. E., Suparno, S., Haris, A., Usman, A., & Riyanto, A. (2017). Organic Shale Analysis using Geochemical Data and Seismic Attributes: Case Study of Talang Akar formation, South Sumatera Basin. AIP Conference Proceedings, 1862(030178), 1-5. doi:10.1063/1.4991282 [Google Scholar] [Crossref]

10. Nwojiji C.N., Osterloff P., Okoro A. U., Ukeri P. O. (2013). Palynostratigraphy and Age of the Sequence Penetrated by the Kolmani River 1 Well in the Gongola Basin, Northern Benue Trough, Nigeria. Journal of Geosciences and Geomatics. Vol. 1(1), pp. 15-21; DOI:10.12691/jgg-1-1-3 [Google Scholar] [Crossref]

11. Obaje, N. G., Ulu, O. K. and Petters, S. W. (1999). Biostratigraphic and geochemical controls of hydrocarbon prospects in the Benue trough and Anambra basin, Nigeria: Nigerian Association of Petroleum Explorationists Bulletin, v. 14, p. 18–54. [Google Scholar] [Crossref]

12. Obaje, N. G., Wehner, H., Scheeder, G., Abubakar, M. B. and Jauro, A. (2004). Hydrocarbon Prospectivity of Nigeria’s Inland Basins: From the Viewpoint of Organic Geochemistry and Organic Petrology. AAPG Bulletin, Vol. 88, No. 3, pp. 325-353. [Google Scholar] [Crossref]

13. Passey, Q. R., Creany, S., Kulla, J. B., Moretti, F. J., & Stroud, J. D. (1990). A Practical Model for Organic Richness from Porosity and Resistivity Log. American Association of Petroleum Geologists Bulletin, 74(12), 1777-1794. [Google Scholar] [Crossref]

14. Petters, S. W. (1982): Central West Africa Cretaceous – Tertiary benthic foraminifera and Stratigraphy, Paleontolographica Abtelung, A, 179, 1–104. [Google Scholar] [Crossref]

15. Siddique, A. B., Alvee, A. M. M., Ana, P. D., Jahan, L. N., & Hashan, M. (2025). Tree-based, boosting, and stacked models for accurate prediction of total organic carbon from conventional well logs. Energy Reports, 13, 4491–4513. https://doi.org/10.1016/j.egyr.2025.04.021 [Google Scholar] [Crossref]

16. Sun, J., Dang, W., Wang, F., Nie, H., Wei, X., Li, P., Zhang, S., Feng, Y., and Li, F. (2023), Prediction of TOC Content in Organic-Rich Shale Using Machine Learning Algorithms: Comparative Study of Random Forest, Support Vector Machine, and XGBoost, May 2023, Energies 16(10):4159, DOI: 10.3390/en16104159 [Google Scholar] [Crossref]

17. Tan, M., Song, X., Yang, X., and Wu, Q. (2015). “Support-vector-regression machine technology for total organic carbon content prediction from wireline logs in organic shale: a comparative study,” Journal of Natural Gas Science and Engineering, vol. 26, pp. 792–802. [Google Scholar] [Crossref]

18. Rui, J., Zhang, H., Zhang, D., Han, F., & Guo, Q. (2019). “Total organic carbon content prediction based on support-vector regression machine with particle swarm optimization,” Journal of Petroleum Science and Engineering, vol. 180, pp. 699–706. [Google Scholar] [Crossref]

19. Wang, P., Chen, Z., Pang, X., Hu, K., Sun, M., & Chen, X. (2016). “Revised models for determining TOC in shale play: example from devonian Duvernay shale, western Canada sedimentary basin,” Marine and Petroleum Geology, vol. 70, pp. 304–319. [Google Scholar] [Crossref]

20. Wu, Q., Pang, H., Zhang, B., Jiang, F., Wu, L., Chen, J., Ma, M., & Huo, X. (2025). Application of shale TOC prediction model using the XGBoost machine learning algorithm: A case study of the Qiongzhusi Formation in central Sichuan Basin. Carbonates and Evaporites, 40(1), Article 8. doi.org [Google Scholar] [Crossref]

21. Zarboski, P., Ugodunlunwa, F. Idornigie, A., Nnabo, P and Ibe, K. (1997). Stratigraphy and structure of the Cretaceous Gongola basin, Northeastern Nigeria. Bulletin des Centres de Recherches Exploration-Production. ElF-Aquitaine 21, 153-185. [Google Scholar] [Crossref]

22. Zhao, P., Ma, H., Rasouli, V., Liu, W., Cai, J., & Huang, Z. (2017). “An improved model for estimating the TOC in shale formations,” Marine and Petroleum Geology, vol. 83, pp. 174–183. [Google Scholar] [Crossref]

23. Zhao, Z., Zhong, G., Diao, F., Ding, P., & He, J. (2026). A feature engineering and XGBoost framework for prediction of TOC from conventional logs in the Dongying Depression, Bohai Bay Basin. Geosciences, 16(1), 44. doi.org [Google Scholar] [Crossref]