We put forward a method to invert surface magnetotelluric resistivity(MMR) data which can improve the discrimination of anomaly bodies and obtain the absolute conductivity.The response of magnetotelluric resistivity is replaced by the quasi-direct current low frequency magnetic field.The electric field of the frequency domain Helmholtz equations is solved by a 3D staggered-grid finite difference method for MMR modeling.The calculation of the total magnetic field has two parts which are the normal magnetic field (including the source) and anomalous magnetic field.Y-component is selected for inversion study combined with the characteristics of surface MMR data.The nonlinear conjugate gradient (NLCG) method is used to perform 3D inversion combined with multi reconstruction to improve the discrimination of anomaly bodies and conduct the multi-reconstruction inversion.The impressing method is introduced in starting model construction in which the auxiliary model is placed at a shallow depth and the meta model is located at a larger depth.Inversions of synthetic and field data show that (1) we can get conductivity directly instead of the ratio from low frequency magnetic data inversion by frequency domain Maxwell equation.(2) Using the impressing method to combine the starting model and conducting multi-reconstruction inversion can improve the discrimination of anomaly bodies by inverting surface MMR data and confirm their depths.Meanwhile,it does not lose the discrimination of anomaly bodies at the shallow depth.(3) Influence of the non-uniform conductor at the surface discrimination of the anomaly body at depth is less in 3D surface MMR data inversion combined with the impressing method.(4) Model variation at the end of last reconstruction inversion can be used to determine impressing depth.Differences between variations at the end of two adjacent reconstruction inversions can be used to determine if the iterative reconstruction is ended.Due to the similarity between static magnetic field and gravitational field in math,this inversion methodology proposed can be used in the inversion of surface data in the potential field data (such as the gravitational field).%本文提出了能提高异常体分辨能力,同时得到绝对电导率的地面磁电阻率数据三维反演方法.磁电阻率响应用准直流的低频磁场代替;数值模拟由频率域电场满足的Helmholtz方程出发,采用三维交错网格有限差分法;长直导线源作为发射源,其中源的计算包含在背景场中;结合地面磁电阻率数据各分量的特点,选择y分量进行反演研究;反演采用三维非线性共轭梯度反演技术,为了提高异常体的深度分辨能力,进行迭代重构反演;用印模法对初始模型进行重构,采用的是辅模型在浅部,元模型在深部的组合方式.从合成数据和实际数据的反演结果可以得到以下的认识:(1)由频率域麦克斯韦方程组出发,低频磁场数据反演可以直接得到电导率,而不是相对电导率之比;(2)采用印模法组合初始模型,进行迭代重构反演,可以提高地面磁电阻率数据反演对异常体的分辨能力,确定埋深位置,同时不会丧失对于浅部异常体的分辨能力;(3)在结合印模法的地面磁电阻率数据三维反演中,深部异常体的分辨能力受地表不均匀导电体影响较小;(4)确定印模深度可以采用上一次重构反演结束时的模型变化量,通过相邻两次重构反演结束时的模型变化量之差来确定迭代重构是否终止.因为静磁场与重力场在数学上的相似性,本文的反演方法可以被运用到重力场等位场的地面数据的反演中.
展开▼
