3D Data-Space OCCAM Inversion For Controlled Source Extremely Low Frequency Method

The controlled source extremely low frequency method is an improved electromagnetic prospecting method derived from the integration of MT and CSAMT methods which inherits the advantages and overcomes the defects of both methods. An extremely low frequency transmitter facility has been set up in Central China. Many related researches have been made in China, still the forward modeling and inversion studies regarding this method is preliminary.With regard to the singularity of the transmitter, the total electromagnetic field is decomposed into two parts: the primary and secondary fields which are calculated respectively. We develop an algorithm for calculating the electromagnetic field of horizontal electric dipole and long grounded line considering the whole space consisting of the ionosphere,air and layered stratum,which is used in our evaluation of the primary field. The validity of the 1D algorithm is also tested. Then the characteristics of the electromagnetic field in the axial and equatorial directions of the transmitter under the influence of the ionosphere are well scrutinized.The staggered-grid finite difference method is adopted to obtain the secondary field.Starting from the Maxwell's equations, the second-order partial differential equation concerning the secondary electric field is deduced. And the discretization of the equation is also introduced in detail. Then the differences between the 3D numerical simulations of active and passive source electromagnetic methods are discussed. The efficiencies of the QMR solver when choosing different divergence correction frequencies are also compared. The forward modeling is completed by the combination of the primary and secondary field.The commonly used inversion methods of electromagnetic data are firstly summarized and compared to start the inversion chapter. The regularized 3D inversion is carried out using data-space OCCAM method in this dissertation. The calculation of sensitivity matrix is completed by applying the reciprocal property of electromagnetic field.The 3D inversion results in both scalar and tensor observation scenarios prove the accuracy and stability of our inversion algorithm and the superiority of the tensor observation scheme. The effect of the observation network density and the addition of prior information on the inversion are also studied numerically.The study is summarized and objectively reviewed at last. Several improvements which can be made are listed. And some suggestions with regard to further studies of this method are also given accordingly.