Porosity prediction with Bi-LSTM network for deep methane reservoirs

Abstract

Quantitative prediction of petrophysical parameters, such as porosity, is crucial for the evaluation and development of coalbed methane (CBM) reservoirs. However, conventional methods based on linear assumptions and empirical formulas often fall short due to the strong heterogeneity of coal seams, complex lithologies and structures, and the highly non-linear relationship between seismic elastic parameters and reservoir properties under deep-buried conditions. While machine learning techniques have shown promise in petrophysical prediction, many existing approaches struggle to effectively capture long-range dependencies within sequential log data. This study proposes a deep learning-based method that integrates comprehensive input feature selection with a bidirectional long short-term memory (Bi-LSTM) network incorporating dropout regularization for enhanced petrophysical parameter prediction. The proposed method is designed to fully exploit the non-linear mapping between seismic elastic parameters (e.g., P-wave velocity, S-wave velocity, density, elastic impedance) and petrophysical parameter (porosity). By combining the bidirectional contextual learning capability of Bi-LSTM, the model effectively captures feature relationships within depth sequences. Comparative analysis against a fully connected neural network and a standard LSTM network demonstrates the superiority of the proposed method. The analysis also reveals the optimal feature combination and network parameter setting (sequential length, sampling interval, etc.). Results indicate that the Bi-LSTM model achieves a significant improvement in prediction accuracy, outperforming other models, and demonstrating better generalization capability in blind well tests. The method provides a reliable and effective tool for quantitative reservoir characterization, offering substantial potential for application in deep CBM exploration.

Keywords

Deep coalbed methane / Porosity prediction / Deep learning / LSTM network

Funding

We appreciate the support provided by the National Key Research and Development Program of China (2024YFC3015802), the National Natural Science Foundation of China (42574178 and 42374128) and the Jiangsu Provincial Science and Technology Plan Project (Natural Science Foundation of Jiangsu Province, BK20252046).

References

[1] Pan J, Ge T, Liu W, et al. Organic matter provenance and accumulation of transitional facies coal and mudstone in Yangquan, China: Insights from petrology and geochemistry. J Nat Gas Sci Eng. 2021; 94:104076. doi:10.1016/j.jngse.2021.104076

[2] Hou X, Liu S, Zhu Y, Yang Y. Evaluation of gas contents for a multi-seam deep coalbed methane reservoir and their geological controls: In situ direct method versus indirect method. Fuel. 2020; 265:116917. doi:10.1016/j.fuel.2019.116917

[3] Gong F, Cheng J, Wang G, Peng S, Zhang Z. The effect of lamination on elastic anisotropy of primary coals under confining pressure: Experiment and theoretical modelling. Geophys Prospect. 2025; 73(4):1228-1242. doi:10.1111/1365-2478.70009

[4] Gong F, Huang A, Kang W, et al. The influence of lamination and fracture on the velocity anisotropy of tectonic coals. Geophysics. 2024; 89(6):MR355-MR365.doi:10.1190/GEO2024-0033.1

[5] Khaksar A, Griffiths CM. Porosity from sonic log in gas-bearing shale sandstones: Field data versus empirical equations. Explor Geophys. 1998; 29(4):440-446. doi:10.1071/EG998440

[6] Liu X, Shao G, Yuan C, Chen X, Li J, Chen Y. Mixture of relevance vector regression experts for reservoir properties prediction. J Petrol Sci Eng. 2022; 214:110498. doi:10.1016/j.petrol.2022.110498

[7] Wang P, Chen X, Wang B, Li J, Dai H. An improved method for lithology identification based on a hidden Markov model and random forests. Geophysics. 2020; 85(6):IM27-IM36. doi:10.1190/GEO2020-0108.1

[8] Sang K, Yin X, Zhang F. Machine learning seismic reservoir prediction method based on virtual sample generation. Petrol Sci. 2021; 18(6):1662-1674. doi:10.1016/j.petsci.2021.09.034

[9] Guo Q, Ba J, Luo C. Seismic rock-physics linearized inversion for reservoir- property and pore-type parameters with application to carbonate reservoirs. Geoenergy Sci Eng. 2023; 224:211640. doi:10.1016/j.geoen.2023.211640

[10] Luo C, Ba J, Guo Q. Probabilistic seismic petrophysical inversion with statistical double-porosity Biot-Rayleigh model. Geophysics. 2023; 88(3):M157-M171. doi:10.1190/GEO2022-0288.1

[11] Wang P, Cui Y, Zhou L. Multi-task learning for seismic elastic parameter inversion with the lateral constraint of angle-gather difference. Petrol Sci. 2024; 21(6):4001-4009. doi:10.1016/j.petsci.2024.06.010

[12] Zhao L, Nasser M, Han D. Quantitative geophysical pore-type characterization and its geological implication in carbonate reservoirs. Geophys Prospect. 2013; 61:827-841. doi:10.1111/1365-2478.12043

[13] Song L, Yin X, Zong Z, Jiang M. Semi-supervised learning seismic inversion based on Spatio-temporal sequence residual modeling neural network. J Petrol Sci Eng. 2022; 208:109549. doi:10.1016/j.petrol.2021.109549

[14] Wu X, Jiang G, Wang X, et al. Prediction of reservoir sensitivity using RBF neural network with trainable radial basis function. Neural Comput Appl. 2013; 22:947-953. doi:10.1007/s00521-011-0787-z

[15] Ahmadi MA, Ebadi M, Shokrollahi A, Majidi SMJ. Evolving artificial neural network and imperialist competitive algorithm for prediction oil flow rate of the reservoir. Appl Soft Comput. 2013; 13(2):1082-1098.doi:10.1016/j.asoc.2012.10.009

[16] Zerrouki AA, Aifa T, Baddari K. Prediction of natural fracture porosity from well log data by means of fuzzy ranking and an artificial neural network in Hassi Messaoud oil field, Algeria. J Petrol Sci Eng. 2014; 115:78-89. doi:10.1016/j.petrol.2014.01.011

[17] Cao J, Yang J, Wang Y, Wang D, Shi Y. Extreme learning machine for reservoir parameter estimation in heterogeneous sandstone reservoir. Math Probl Eng. 2015; 2015:287816. doi:10.1155/2015/287816

[18] Zou C, Zhao L, Xu M, Chen Y, Geng J. Porosity prediction with uncertainty quantification from multiple seismic attributes using Random Forest. J Geophys Res Solid Earth. 2021;126:e2021JB021826. doi:10.1029/2021JB021826

[19] Elkatatny S, Mahmoud M, Tariq Z. New insights into the prediction of heterogeneous carbonate reservoir permeability from well logs using artificial intelligence network. Neural Comput Appl. 2018; 30(9):2673-2683. doi:10.1007/s00521-017-2850-x

[20] Wu X, Shi Y, Fomel S. FaultNet3D: Predicting fault probabilities, strikes, and dips with a single convolutional neural network. IEEE Trans Geosci Remote Sens. 2019; 57(11):9138-9155. doi:10.1109/TGRS.2019.2925003

[21] Wang P, Xu H, Peng Z, Wang Z, Yang M. Application of data augmentation based on generative adversarial network in impedance inversion. J Seismic Explor. 2023; 32(2):155-168.

[22] Behnia AMO, Reza M, Ali M. A new approach for seismic inversion with GAN algorithm. J Seismic Explor. 2024; 33(3):1-36.

[23] Suraj P, Omer S, Aditya N, et al. Model fusion with physics-guided machine learning: Projection-based reduced-order modeling. Phys Fluids. 2021; 33(6):067123. doi:10.1063/5.0053349

[24] Xu M, Zhao L, Gao S, Zhu X, Geng J. Joint use of multiseismic information for lithofacies prediction via supervised convolutional neural networks. Geophysics. 2022; 87(5): 151-162. doi:10.1190/GEO2021-0554.1

[25] Gao S, Xu M, Zhao L, Chen Y, Geng J. Seismic predictions of fluids via supervised deep learning: Incorporating various class-rebalance strategies. Geophysics. 2023; 88(4):185-200.doi:10.1190/GEO2022-0363.1

[26] Yu S, Ma J. Deep learning for geophysics: Current and future trends. Rev Geophys. 2021;59:e2021RG000742. doi:10.1029/2021RG000742

[27] Wang Y, Niu L, Zhao L, et al. Gaussian mixture model deep neural network and its application in porosity prediction of deep carbonate reservoir. Geophysics. 2022; 87(2):59-72. doi:10.1190/GEO2020-0740.1

[28] Wu H, Wu R, Zhang P, Huang Y, Huang Y, Dong S. Combined fluid factor and brittleness index inversion for coal-measure gas reservoirs. Geophys Prospect. 2022; 70:751-764. doi:10.1111/1365-2478.13172

[29] Liu J, Zhao L, Xu M, Zhao X, You Y, Geng J. Porosity prediction from prestack seismic data via deep learning: Incorporating a low-frequency porosity model. J Geophys Eng. 2023; 20(5):1016-1029. doi:10.1093/jge/gxad063

[30] Zhang J, Liu Z, Zhou Y, Ai H, Han H. Joint inversion method of rock physics based on hunger games search correction and Bi-LSTM. IEEE Trans Geosci Remote Sens. 2024; 62:5914310.

[31] Sun Y, Pang S, Zhang J, Zhang Y. Porosity prediction through well logging data: A combined approach of convolutional neural network and transformer model (CNN-transformer). Phys Fluids. 2024; 36(2):026604. doi:10.1063/5.0190078

[32] Tao B, Zhou H, Chen L, Liu B, Wang R, Liu X. Porosity prediction based on stochastic modeling and facies-controlled dataset constrained by seismic attribute. IEEE Geosci Remote Sens Lett. 2025; 22:1-5. doi:10.1109/LGRS.2025.3580778

[33] Ashraf M, Robles WRQ, Kim M, Ko YS, Yi MY. A loss-based patch label denoising method for improving whole-slide image analysis using a convolutional neural network. Sci Rep. 2022; 12:1392. doi:10.1038/s41598-022-05001-8

[34] Gao Z, Li C, Yang T, Pan Z, Gao J, Xu Z. OMMDE-Net: A deep learning-based global optimization method for seismic inversion. IEEE Geosci Remote Sens Lett. 2021; 18:208-212. doi:10.1109/LGRS.2020.2973266

[35] Hochreiter S, Schmidhuber J. Long short-term memory. Neural Comput. 1997; 9:1735-1780. doi:10.1162/neco.1997.9.8.1735