Integrating Seismic Inversion with Machine Learning for Porosity Cube Estimation: A Synergistic Approach

Abstract This study investigates porosity prediction by integrating seismic inversion with machine learning, employing chi angle-based and Acoustic Impedance (AI) cube-based techniques to estimate porosity values. A systematic approach involving data collection, preprocessing, and seismic inversion is utilized to extract rock property information crucial for accurate subsurface characterization. Advanced machine learning models are developed and rigorously trained on diverse, relevant datasets, with porosity estimated using both methodologies. The efficacy of these methods is thoroughly evaluated through cross-validation and comparison with well-log data, ensuring robust validation. Results indicate that the chi angle-based method excels in capturing subtle, intricate patterns for precise porosity predictions, particularly in complex geological environments, while the AI cube-based method provides robust and reliable predictions across diverse geological settings. A detailed comparative analysis highlights the complementary nature of these two techniques, showing how their integration enhances porosity characterization, providing deeper insights than traditional methods alone. This innovative approach offers valuable insights for subsurface exploration and reservoir management, emphasizing the significant benefits of combining seismic inversion with machine learning for improved porosity prediction, enhanced accuracy in reservoir assessment, and more informed decision-making in exploration and production strategies.