多方法多信息电磁联合约束反演方法研究进展

Progress on joint inversion of electromagnetic constraints combined with multiple methods and multiple information

电磁法在油气勘探、深部矿产资源调查、地热地下水勘查、地球深部结构成像以及水下军事目标侦察等方面都发挥着重要作用。文中系统地总结了多种信息和多方法的电磁联合约束反演研究进展,从方法应用的角度介绍了研究现状,着重分析了不同电磁联合约束反演方法目标函数的构建形式及存在的问题和技术难点。在地震数据等已知信息较多的地区,利用先验信息构建不同类型约束函数,通过约束反演提高盐岩、火成岩发育区构造成像精度,降低反演的非唯一性。在已采集天然源大地电磁(MT)数据和人工源电磁(CSEM)数据的地区,可利用这两类方法各自的优势开展联合反演,发展顺序约束的联合反演及基于统一目标函数的多方法联合反演。CSEM法与MT法在反演深度上具有互补性,覆盖整个研究深度,同时可提高目标反演精度,减少反演的非唯一性。文中针对这一类联合反演的目标函数和权重因子的选择进行讨论。根据实际应用效果分析,目前最重要、最有潜力的联合反演是电磁数据与地震数据的联合反演,这种联合反演方式经历了从地震约束电磁反演到基于物理模型的联合反演和基于交叉梯度方法的联合反演。文中分析了速度与电阻率模型之间难以构建合理联系这一难点,说明目标函数中交叉梯度项和物理模型约束项的权重因子设置是目前开展地震—电磁联合约束反演的关键和难点。最后,笔者提出:将电磁联合反演与人工智能算法、三维可视化分析相结合;采用大规模并行计算等新方法、新技术的发展方向;发展联合采集系统,为多物理场联合反演提供更有效的数据体。

Electromagnetic exploration plays an important role in oil and gas exploration,deep mineral resources sur-vey,geothermal groundwater exploration,imaging of deep earth structures,and underwater military target reconnais-sance.This paper systematically summarizes the research progress on joint inversion of electromagnetic constraints combined with multiple methods and information,introduces the research status from the perspective of method ap-plication,and focuses on the construction form,existing problems,and technical difficulties of the objective function for different inversion methods.In areas with more known information such as seismic data,different types of con-straint functions are constructed by utilizing priori information to improve the tectonic imaging accuracy in saline and igneous rock development areas and reduce the inversion non-uniqueness.In areas where the magnetotelluric(MT)and controlled source electromagnetic(CSEM)data have been collected,the two methods can be employed to carry out joint inversion with their respective advantages.Various forms such as joint inversion with sequential constraints and joint inversion based on a unified objective function are developed.Additionally,CSEM and MT make up for each other in-depth inversion to improve the target inversion accuracy and reduce non-uniqueness.This paper dis-cusses the selection of objective functions and weighting factors for this type of joint inversion.According to actual ap-plication effects,the most important and potential joint inversion is the inversion of electromagnetic and seismic data.Joint inversion has gone through from seismic constrained electromagnetic inversion to physical model-based and cross-gradient method-based joint inversion.This paper also analyzes the difficulty in constructing a reasonable con-nection between velocity and resistivity models and explains that the weighting factor setting of velocity and resis-tivity models in the objective functions is still the key and difficult points.Finally,this paper proposes to combine electromagnetic joint inversion,artificial intelligence algorithms,and three-dimensional visualization analysis.It also points out that large-scale parallel computation and other new methods and technologies can be adopted in the future,and joint acquisition systems can be developed to provide more effective data volumes for the multi-physical field joint inversion.