Machine learning supervised algorithms for gas hydrate identification and saturation estimation in marine reservoirs using well log data: A case study of NGHP-01-19B

Abstract

Gas hydrates are increasingly recognized as a significant unconventional energy resource and a key factor in marine geohazards and the global carbon cycle. However, accurately identifying and quantifying hydrate-bearing formations remains challenging due to complex geophysical signatures and heterogeneous distribution. This study evaluates twelve supervised machine learning (ML) algorithms for two key tasks: Classification of hydrate-bearing layers and regression-based estimation of hydrate saturation, using well log and pore-water geochemical data from Site NGHP-01-19B. Two physically independent labeling frameworks are employed: One based on Archie’s law using resistivity (1350 samples, 29% hydrate-bearing), and another based on a three-phase velocity model (890 samples, 25% hydrate-bearing). A diverse set of models, including tree-based ensembles (Decision Tree, Random Forest, GBDT, XGBoost, LightGBM, CatBoost, Bagging, AdaBoost), kernel methods (SVM, SVR), instance-based learning (KNN), neural networks (MLP), and Gaussian Process models (GPR, GPC), are systematically compared using cross-validation and grid search. Ensemble methods consistently performed best in classification, with AdaBoost and GBDT, achieving test accuracies above 0.94 (Archie) and 0.98 (velocity-based). For regression, GPR delivered the most accurate hydrate saturation estimates (R² > 0.99), while GBDT and Random Forest provided a strong balance of accuracy and computational efficiency. Notably, depth below seafloor (TDEP), though not a direct geophysical input, significantly enhanced model performance by acting as a proxy for stratigraphic and thermodynamic conditions. Group-based validation confirmed that random-sample splitting overestimates performance due to depth-wise autocorrelation, highlighting the importance of geologically informed model assessment. Overall, the consistent performance of ML models across both labeling schemes and input feature sets underscores their robustness and transferability, supporting their use as a reliable toolset for offshore gas hydrate reservoir characterization.

Keywords

Gas hydrate / Machine learning algorithm / Classification / Regression / Well log data / Archie’s law / Marine geohazards / Global carbon cycle

References

[1] Ahmad AB, Tsuji T. 2021. Machine learning for automatic slump identification from 3D seismic data at convergent plate margins. Marine and Petroleum Geology, 133, 105290. doi: 10.1016/j.marpetgeo.2021.105290.

[2] Archie GE. 1942. The electrical resistivity log as an aid in determining some reservoir characteristics. Transactions of the AIME, 146(1), 54-62. doi: 10.2118/942054-G.

[3] Bastia R. 2006. An overview of Indian sedimentary basins with special focus on emerging east coast deepwater frontiers. The Leading Edge, 25(7), 818-829. doi: 10.1190/1.2221359.

[4] Bastia R, Radhakrishna M, Das S, Kale AS, Catuneanu O. 2010. Delineation of the 85 E ridge and its structure in the Mahanadi Offshore Basin, Eastern Continental Margin of India (ECMI), from seismic reflection imaging. Marine and Petroleum Geology, 27(9), 1841-1848. doi: 10.1016/j.marpetgeo.2010.08.003.

[5] Bayar B, Stamm MC. 2018. Constrained convolutional neural networks: A new approach towards general purpose image manipulation detection. IEEE Transactions on Information Forensics and Security, 13(11), 2691-2706. doi: 10.1109/TIFS.2018.2825953.

[6] Bharali B. 1991. A brief review of Mahanadi delta and the deltaic sediments in Mahanadi basin. Memoir Geological Society of India, 22, 31-49.

[7] Breiman L. 1996. Bagging predictors. Machine Learning, 24(2), 123-140. doi: 10.1007/BF00058655.

[8] Breiman L. 2001. Random forests. Machine Learning, 45(1), 5-32. doi: 10.1023/A:1010933404324.

[9] Buitinck L, Louppe G, Blondel M, Pedregosa F, Mueller A, Grisel O, Niculae V, Prettenhofer P, Gramfort A, Grobler J, Layton R, Vanderplas J, Joly A, Holt B, Varoquaux G. 2013. API design for machine learning software: Experiences from the scikit-learn project. arXiv, 1309.0238. doi: 10.48550/arXiv.1309.0238.

[10] Cao Y, Miao QG, Liu JC, Gao L. 2013. Advance and prospects of AdaBoost algorithm. Acta Automatica Sinica, 39(6), 745-758. doi: 10.1016/S1874-1029(13)60052-X.

[11] Chen TQ, Guestrin C. 2016. XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785-794. doi: 10.1145/2939672.2939785.

[12] Chong L, Collett TS, Creason CG, Seol Y, Myshakin EM. 2024. Machine-learning application to assess occurrence and saturations of methane hydrate in marine deposits offshore India. Interpretation, 12(1), T63-T75. doi: 10.1190/INT-2023-0056.1.

[13] Chong L, Singh H, Creason CG, Seol Y, Myshakin EM. 2022. Application of machine learning to characterize gas hydrate reservoirs in Mackenzie Delta (Canada) and on the Alaska north slope (USA). Computational Geosciences, 26(5), 1151-1165. doi: 10.1007/s10596-022-10151-9.

[14] Collett TS, Riedel M, Cochran JR, Boswell R, Kumar P, Sathe AV. 2008. Indian continental margin gas hydrate prospects: Results of the Indian National Gas Hydrate Program (NGHP) Expedition 01. doi: 10.7916/D8-NF9W-CZ91.

[15] Collett TS. 2002. Energy resource potential of natural gas hydrates. AAPG Bulletin, 86(11), 1971-1992. doi: 10.1306/61EEDDD2-173E-11D7-8645000102C1865D.

[16] Collett TS, Boswell R, Cochran JR, Kumar P, Lall M, Mazumdar A, Ramana MV, Ramprasad T, Riedel M, Sain K, Vishwanath K. Sathe AV, 2014. Geologic implications of gas hydrates in the offshore of India: Results of the National Gas Hydrate Program Expedition 01. Marine and Petroleum Geology, 58, 3-28. doi: 10.1016/j.marpetgeo.2014.07.021.

[17] Collett TS, Ladd J. 2000. Detection of gas hydrate with downhole logs and assessment of gas hydrate concentrations (saturations) and gas volumes on the Blake Ridge with electrically resistivity log data. In:Proceedings of the Ocean Drilling Program, Scientific Results, 164. College Station, TX, Texas A & M University. https://www.academia.edu/download/70992860/227b90d630a066a2f498f64fac64e6a1c3f7.pdf.

[18] Collett TS, Riedel M, Boswell R, Presley J, Kumar P, Sathe A, Sethi A, Lall MV, NGHP Expedition scientists. 2015. Indian National Gas Hydrate Program Expedition 01 report. Reston, US Geological Survey. https://pubs.usgs.gov/publication/sir20125054.

[19] Dangwal V, Sengupta S, Desai AG. 2008. Speculated petroleum systems in deep offshore Mahanadi Basin in Bay of Bengal, India. International Conference & Exposition on Petroleum Geophysics, Hyderabad, India, 1-7. https://spgindia.org/2008/001.pdf.

[20] Dickens GR. 2003. Rethinking the global carbon cycle with a large, dynamic and microbially mediated gas hydrate capacitor. Earth and Planetary Science Letters, 213(3-4), 169-183. doi: 10.1016/S0012-821X(03)00325-X.

[21] Drineas P, Mahoney MW, Cristianini N. 2005. On the Nyström method for approximating a Gram matrix for improved kernel-based learning. Journal of Machine Learning Research, 6(12), 2153-2175. http://www.jmlr.org/papers/volume6/drineas05a/drineas05a.pdf.

[22] Drucker H, Burges CJ, Kaufman L, Smola A, Vapnik V. 1996. Support vector regression machines. Advances in Neural Information Processing Systems, 9, 155-161. https://proceedings.neurips.cc/paper/1996/hash/d38901788c533e8286cb6400b40b386d-Abstract.html.

[23] Fofonoff NP. 1985. Physical properties of seawater: A new salinity scale and equation of state for seawater. Journal of Geophysical Research: Oceans, 90(C2), 3332-3342. doi: 10.1029/JC090iC02p03332.

[24] Fuloria RC, Pandey RN, Bharali BR, Mishra JK. 1992. Stratigraphy, structure and tectonics of Mahanadi offshore basin. Visesa Prakasana-Bharatiya Bhuvaijñanika Sarveksana, 29, 255-265. http://pascal-francis.inist.fr/vibad/index.php?action=getRecordDetail&idt=6375318.

[25] Goodfellow I, Bengio Y, Courville A. 2016. Deep feedforward networks. Deep Learning, 1, 161-217.

[26] Hancock JT, Khoshgoftaar TM. 2020. CatBoost for big data: An interdisciplinary review. Journal of Big Data, 7(1), 94. doi: 10.1186/s40537-020-00369-8.

[27] Harris JR, Grunsky EC 2015. Predictive lithological mapping of Canada’ s north using Random Forest classification applied to geophysical and geochemical data. Computers & Geosciences, 80, 9-25. doi: 10.1016/j.cageo.2015.03.013.

[28] Jagannathan CR, Ratnam C, Baishva NC, Gupta UD. 1984. Geology of the offshore Mahanadi basin. Petroleum Asia Journal, 6(4), 101-104. https://www.osti.gov/etdeweb/biblio/6776553.

[29] Jana S, Ojha M, Sain K, Srivastava S. 2017. An approach to estimate gas hydrate saturation from 3-D heterogeneous resistivity model: A study from Krishna-Godavari Basin, Eastern Indian offshore. Marine and Petroleum Geology, 79, 99-107. doi: 10.1016/j.marpetgeo.2016.11.006.

[30] Karpatne A, Ebert-Uphoff I, Ravela S, Babaie HA, Kumar V. 2018. Machine learning for the geosciences: Challenges and opportunities. IEEE Transactions on Knowledge and Data Engineering, 31(8), 1544-1554. doi: 10.1109/TKDE.2018.2861006.

[31] Ke GL, Meng Q, Finley T, Wang TF, Chen W, Ma WD, Ye QW, Liu TY. 2017. LightGBM: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems, 30. https://proceedings.neurips.cc/paper/2017/hash/6449f44a102fde848669bdd9eb6b76fa-Abstract.html.

[32] Kuremoto T, Kimura S, Kobayashi K, Obayashi M. 2014. Time series forecasting using a deep belief network with restricted Boltzmann machines. Neurocomputing, 137, 47-56. doi: 10.1016/j.neucom.2013.03.047.

[33] Ma XL, Ding CD, Luan SL, Wang Y, Wang YP. 2017. Prioritizing influential factors for freeway incident clearance time prediction using the gradient boosting decision trees method. IEEE Transactions on Intelligent Transportation Systems, 18(9), 2303-2310. doi: 10.1109/TITS.2016.2635719.

[34] Malinverno A, Kastner M, Torres ME, Wortmann UG. 2008. Gas hydrate occurrence from pore water chlorinity and downhole logs in a transect across the northern Cascadia margin (Integrated Ocean Drilling Program Expedition 311). Journal of Geophysical Research: Solid Earth, 113(B8), B08103. doi: 10.1029/2008JB005702.

[35] Marjanovic M, Kovacevic M, Bajat B, Vozenilek V. 2011. Landslide susceptibility assessment using SVM machine learning algorithm. Engineering Geology, 123(3), 225-234. doi: 10.1016/j.enggeo.2011.09.006.

[36] Max MD, Johnson AH. 2019. Economic characteristics of deepwater natural gas hydrate. In: Max MD, Johnson AH (eds). Exploration and Production of Oceanic Natural Gas Hydrate, 57-94. Cham, Springer International Publishing. doi: 10.1007/978-3-030-00401-9_2.

[37] Michelucci U. 2024. Fundamental Mathematical Concepts for Machine Learning in Science. Cham, Springer International Publishing. doi: 10.1007/978-3-031-56431-4.

[38] Milkov AV. 2004. Global estimates of hydrate-bound gas in marine sediments: How much is really out there? Earth-Science Reviews, 66(3-4), 183-197. doi: 10.1016/j.earscirev.2003.11.002.

[39] Milkov AV, Sassen R. 2003. Preliminary assessment of resources and economic potential of individual gas hydrate accumulations in the Gulf of Mexico continental slope. Marine and Petroleum Geology, 20(2), 111-128. doi: 10.1016/S0264-8172(03)00024-2.

[40] Mohapatra P. 2006. Sequence stratigraphic approach for identification of hydrocarbon plays in Mahanadi offshore basin. 6th International Conference & Exposition on Petroleum Geophysics, 672-677. https://www.academia.edu/download/76798365/231.pdf.

[41] Murphy KP. 2012. Machine Learning: A Probabilistic Perspective. Cambridge, MIT Press. https://books.google.com/books?hl=en&lr=&id=RC43AgAAQBAJ.

[42] Nguyen B, Morell C, De Baets BD. 2016. Large-scale distance metric learning for k-nearest neighbors regression. Neurocomputing, 214, 805-814. doi: 10.1016/j.neucom.2016.07.005.

[43] Nguyen H, Bui XN, Bui HB, Cuong DT. 2019. Developing an XGBoost model to predict blast-induced peak particle velocity in an open-pit mine: A case study. Acta Geophysica, 67(2), 477-490. doi: 10.1007/s11600-019-00268-4.

[44] Pearson CF, Halleck PM, McGuire PL, Hermes R, Mathews M. 1983. Natural gas hydrate deposits: A review of in situ properties. The Journal of Physical Chemistry, 87(21), 4180-4185. doi: 10.1021/j100244a041.

[45] Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Blondel M, Prettenhofer P, Weiss R, Dubourg V. 2011. Scikit-learn: Machine learning in Python. The Journal of Machine Learning Research, 12, 2825-2830.

[46] Prokhorenkova L, Gusev G, Vorobev A, Dorogush AV, Gulin A. 2018. CatBoost: Unbiased boosting with categorical features. Advances in Neural Information Processing Systems, 31. https://proceedings.neurips.cc/paper/2018/hash/14491b756b3a51daac41c24863285549-Abstract.html.

[47] Raschka S, Mirjalili V. 2019. Python Machine Learning: Machine Learning and Deep Learning with Python, Scikit-learn, and TensorFlow 2. Birmingham, Packt Publishing Ltd. https://books.google.com/books?hl=en&id=sKXIDwAAQBAJ.

[48] Rumelhart DE, Hinton GE, Williams RJ. 1986. Learning representations by back-propagating errors. Nature, 323(6088), 533-536. doi: 10.1038/323533a0.

[49] Sastri VV, Venkatachala BS, Narayanan V. 1981. The evolution of the east coast of India. Palaeogeography, Palaeoclimatology, Palaeoecology, 36(1-2), 23-54. doi: 10.1016/0031-0182(81)90047-X.

[50] Shankar U, Riedel M. 2014. Assessment of gas hydrate saturation in marine sediments from resistivity and compressional-wave velocity log measurements in the Mahanadi Basin, India. Marine and Petroleum Geology, 58, 265-277. doi: 10.1016/j.marpetgeo.2013.10.007.

[51] Singh H, Seol Y, Myshakin EM. 2021. Prediction of gas hydrate saturation using machine learning and optimal set of well-logs. Computational Geosciences, 25(1), 267-283. doi: 10.1007/s10596-020-10004-3.

[52] Sloan ED. 1991. Natural gas hydrates. Journal of Petroleum Technology, 43(12), 1414-1417. doi: 10.2118/23562-PA.

[53] Smola AJ, Schölkopf B. 2004. A tutorial on support vector regression. Statistics and Computing, 14(3), 199-222. doi: 10.1023/B:STCO.0000035301.49549.88.

[54] Spangenberg E. 2001. Modeling of the influence of gas hydrate content on the electrical properties of porous sediments. Journal of Geophysical Research: Solid Earth, 106(B4), 6535-6548. doi: 10.1029/2000JB900434.

[55] Su Z, Cao YC, Wu NY, Chen DF, Yang SX, Wang HB. 2012. Numerical investigation on methane hydrate accumulation in Shenhu area, northern continental slope of South China Sea. Marine and Petroleum Geology, 38(1), 158-165. doi: 10.1016/j.marpetgeo.2012.06.005.

[56] Su Z, Huang L, Wu NY, Yang SX. 2013. Effect of thermal stimulation on gas production from hydrate deposits in Shenhu area of the South China Sea. Science China Earth Sciences, 56(4), 601-610. doi: 10.1007/s11430-013-4587-4.

[57] Sultan N, Cochonat P, Foucher JP, Mienert J. 2004. Effect of gas hydrates melting on seafloor slope instability. Marine Geology, 213(1-4), 379-401. doi: 10.1016/j.margeo.2004.10.015.

[58] Tian DM, Yang SX, Gong YH, Geng MH, Li YH, Hu G. 2023. A comparative study of machine learning methods for gas hydrate identification. Geoenergy Science and Engineering, 223, 211564. doi: 10.1016/j.geoen.2023.211564.

[59] Vapnik V. 1998. The Support Vector Method of Function Estimation. In: Suykens JAK, Vandewalle J. Nonlinear Modeling: Advanced Black-Box Techniques. Boston, Springer, 55-85. doi: 10.1007/978-1-4615-5703-6_3.

[60] Wang JJ, Lau HC. 2020. Thickness of gas hydrate stability zone in permafrost and marine gas hydrate deposits: Analysis and implications. Fuel, 282, 118784. doi: 10.1016/j.fuel.2020.118784.

[61] Wang XJ, Wu SG, Lee M, Guo YQ, Yang SX, Liang JQ. 2011. Gas hydrate saturation from acoustic impedance and resistivity logs in the Shenhu area, South China Sea. Marine and Petroleum Geology, 28(9), 1625-1633. doi: 10.1016/j.marpetgeo.2011.07.002.

[62] Williams CK, Rasmussen CE. 2006. Gaussian Processes for Machine Learning. Cambridge, MIT Press. https://www.newton.ac.uk/files/seminar/20070809140015001-150844.pdf.

[63] Wood AB. 1941. A Textbook of Sound. New York, MacMillan.

[64] Wyllie MRJ, Gregory AR, Gardner GHF. 1958. An experimental investigation of factors affecting elastic wave velocities in porous media. Geophysics, 23(3), 459-493. doi: 10.1190/1.1438493.

[65] Xu HF, Jiao ZR, Zhang ZR, Huffman M, Wang QS. 2021. Prediction of methane hydrate formation conditions in salt water using machine learning algorithms. Computers & Chemical Engineering, 151, 107358. doi: 10.1016/j.compchemeng.2021.107358.

[66] Yuan HM, Wang Y, Wang XC. 2021. Seismic methods for exploration and exploitation of gas hydrate. Journal of Earth Science, 32(4), 839-849. doi: 10.1007/s12583-021-1502-3.

[67] Yuan T, Hyndman RD, Spence GD, Desmons B. 1996. Seismic velocity increase and deep-sea gas hydrate concentration above a bottom-simulating reflector on the northern Cascadia continental slope. Journal of Geophysical Research: Solid Earth, 101(B6), 13655-13671. doi: 10.1029/96JB00102.

[68] Zhao XB, Chen XJ, Huang Q, Lan ZJ, Wang XG, Yao GQ. 2022. Logging-data-driven permeability prediction in low-permeable sandstones based on machine learning with pattern visualization: A case study in Wenchang A Sag, Pearl River Mouth Basin. Journal of Petroleum Science and Engineering, 214, 110517. doi: 10.1016/j.petrol.2022.110517.

[69] Zhu LQ, Zhang C, Wei Y, Zhang CM. 2017. Permeability prediction of the tight sandstone reservoirs using hybrid intelligent algorithm and nuclear magnetic resonance logging data. Arabian Journal for Science and Engineering, 42(4), 1643-1654. doi: 10.1007/s13369-016-2365-2.

[70] Zhu LQ, Zhou XQ, Zhang CM. 2021. Rapid identification of high-quality marine shale gas reservoirs based on the oversampling method and random forest algorithm. Artificial Intelligence in Geosciences, 2, 76-81. doi: 10.1016/j.aiig.2021.12.001.

[71] Zhu LQ, Wei JG, Wu SG, Zhou XQ, Sun J. 2022. Application of unlabelled big data and deep semi-supervised learning to significantly improve the logging interpretation accuracy for deep-sea gas hydrate-bearing sediment reservoirs. Energy Reports, 8, 2947-2963. doi: 10.1016/j.egyr.2022.01.139.

[72] Zhu LQ, Zhou XQ, Sun J, Liu YR, Wang JC, Wu SG. 2023. Reservoir classification and log prediction of gas hydrate occurrence in the Qiongdongnan Basin, South China Sea. Frontiers in Marine Science, 10, 1055843. doi: 10.3389/fmars.2023.1055843.

[73] Zuo RG. 2017. Machine learning of mineralization-related geochemical anomalies: A review of potential methods. Natural Resources Research, 26(4), 457-464. doi: 10.1007/s11053-017-9345-4.