首页|期刊导航|China Geology|Deep learning framework with hybrid loss function for seismic velocity inversion of natural gas hydrate reservoirs

Deep learning framework with hybrid loss function for seismic velocity inversion of natural gas hydrate reservoirsOA

中文摘要

Natural gas hydrate(NGH)has attracted increasing attention as a promising unconventional energy resource owing to its high volumetric storage capacity,yet its development is accompanied by significant greenhouse gas risks.Therefore,accurate reservoir characterization is vital for marine resource exploration and sustainable development.Full waveform inversion(FWI)offers high-resolution imaging,yet suffers from heavy computation,sensitivity to initial models and non-uniqueness.Recent deep learning(DL)methods improve efficiency and accuracy,nevertheless,still struggle with clear boundary extraction and multi-level semantic representation.The authors propose a novel deep architecture(SC-UNeXt)designed to learn a mapping from seismic records to velocity models,which integrates a U-Net backbone,ConvNeXt residual blocks,spatial-channel squeeze-and-excitation attention and pixel shuffle up-sampling.Furthermore,a hybrid loss integrating mean squared error,multi-scale structural similarity,and perceptual discrepancy simultaneously optimizes pixel-wise accuracy,structural fidelity,and semantic consistency.Comprehensive tests on both synthetic NGH data and the 3D SEG/EAGE marine overthrust model with NGH demonstrate that SC-UNeXt outperforms FWI and advanced DL methods in boundary delineation,structural preservation,noise robustness,and computational efficiency.These results highlight SC-UNeXt as a reliable tool for high-resolution seismic characterization of NGH reservoirs,thereby supporting sustainable exploration and risk assessment of marine hydrate resources.

Zhen-wei Guo;Jun Lei;Bo-chen Wang;Yan-yi Wang;Yan-jun Chen;Jian-xin Liu

School of Geosciences and Info-physics,Central South University,Changsha 410083,China Key Laboratory of Non-Ferrous Resources and Geological Hazard Detection,Changsha 410083,China Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring,Ministry of Education,Changsha 410083,ChinaSchool of Geosciences and Info-physics,Central South University,Changsha 410083,China Key Laboratory of Non-Ferrous Resources and Geological Hazard Detection,Changsha 410083,China Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring,Ministry of Education,Changsha 410083,ChinaSchool of Geosciences and Info-physics,Central South University,Changsha 410083,China Key Laboratory of Non-Ferrous Resources and Geological Hazard Detection,Changsha 410083,China Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring,Ministry of Education,Changsha 410083,ChinaSchool of Geosciences and Info-physics,Central South University,Changsha 410083,China Key Laboratory of Non-Ferrous Resources and Geological Hazard Detection,Changsha 410083,China Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring,Ministry of Education,Changsha 410083,ChinaSchool of Geosciences and Info-physics,Central South University,Changsha 410083,China Key Laboratory of Non-Ferrous Resources and Geological Hazard Detection,Changsha 410083,China Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring,Ministry of Education,Changsha 410083,ChinaSchool of Geosciences and Info-physics,Central South University,Changsha 410083,China Key Laboratory of Non-Ferrous Resources and Geological Hazard Detection,Changsha 410083,China Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring,Ministry of Education,Changsha 410083,China

天文与地球科学

Natural gas hydrateClean energyCombustible iceUnconventional energySeismic velocity inversionDeep learningHybrid loss functionU-Net-like architectureFull waveform inversionGas hydrate explorationRisk assessment

《China Geology》 2026 (3)

P.502-518,17

jointly supported by the Natural Science Foundation of China(42574195)。

10.31035/cg2025273

评论