Status and prospects of joint inversion of geophysical electromagnetic data

Wei Laonao; Li Yongji; Yin Changchun; Su Yang; Liu Yunhe

doi:10.19975/j.dqyxx.2024-058

Abstract：Geophysical electromagnetic data inversion is a typical underdetermined problem.Due to the limit-ations of observational data,individual geophysical methods often suffer from significant non-uniqueness and un-certainty,making it challenging to provide accurate and stable interpretations of subsurface geological structures.To address this issue,joint inversion techniques have emerged as a key research direction in geophysical explora-tion.By integrating the advantages of different geophysical methods,joint inversion enhances the resolution and re-liability of inversion results.In recent years,with the rapid advancement of computational technologies and novel numerical algorithms,various joint inversion approaches have been proposed and widely applied.These primarily include empirical coupling methods based on petrophysical properties,structural coupling methods utilizing spatial correlation constraints,and prior information-constrained methods based on Bayesian theory and fuzzy clustering,and so on.The core concept of these methods is to exploit the complementary information from different geophysi-cal techniques to optimize the joint inversion objective function through petrophysical relationships,spatial gradi-ent constraints,or probabilistic modeling.This reduces solution non-uniqueness and enhances the characterization of subsurface geological structures.Joint inversion techniques have demonstrated significant success in applica-tions such as mineral exploration,energy resource detection,geological mapping,and deep structural studies. Despite their promising potential,joint inversion techniques still face several challenges in practical applica-tions.These challenges include the rational construction of relationships between physical fields,the optimization of regularization strategies,the improvement of computational efficiency,and the handling of complex terrain ef-fects.Future research in joint inversion will focus on the following aspects:(1)leveraging artificial intelligence and data-driven methods to learn nonlinear mappings between petrophysical parameters from large-scale training data-sets,thereby improving inversion speed and accuracy;(2)integrating multi-source data and prior information with-in a probabilistic inversion framework to provide uncertainty quantification and enhance the reliability of inversion results;(3)employing multi-resolution optimization strategies,wherein a coarse-scale inversion captures the over-all structure before progressively refining the model to improve computational efficiency and mitigate local mini-ma issues;and(4)integrating seismic,gravity,magnetic,and electromagnetic data to enhance inversion robustness,while incorporating real-time monitoring data to better capture subsurface dynamic processes. With advancements in high-performance computing,artificial intelligence,and novel geophysical observation technologies,joint inversion methods are expected to play an increasingly crucial role in resource exploration,sub-surface structure detection,and geological hazard monitoring,providing higher-resolution and more accurate sub-surface imaging techniques for Earth science research.