The Simulation And Inversion Of Cross-hole Electromagnetic Fields

Crosshole electromagnetic wave detection technique becomes one of the popular techniques because its low cost, high efficiency, high accuracy, high detecting depth, and sensitive to pore fluid in underground medium. Nowadays, it is widely used in the area, such as oil and gas field, nuclear waste disposal, groundwater, and solid mine detection, and becomes the leading edge of geophysics.Forward and inversion problems are the two main research directions of Crosshole electromagnetic wave detection. Forward can optimize the parameter settings of equipment, and can improve the ability of equipment. Meanwhile, the effect of crosshole electromagnetic tomography relies on the results of forward which is the basis of inversion. Compared to forward, the development of inversion seems slow. The progress of technology brought more challenges to inversion. These provided the space of research to this dissertation.On the basis of existing literature, this dissertation analyzed the technical problems of FDTD, on forward aspect. In this part, the dissertation mainly focused on absorbing boundary conditions, and also deduced the difference scheme of FDTD in cylindrical coordinates. While on inversion, the dissertation brought the technique of wavelet adaptive signal processing to full waveform inversion, and discussed the advantages and disadvantages of it.The dissertation can be divided into five parts:In chapter one, the dissertation introduced the history of crosshole electromagnetic wave detection technique, the popular numerical simulation and inversion methods, and also discussed their advantages and disadvantages.In chapter two, from the fundamental equations of electromagnetic, the dissertation analyzed the FDTD in loss medium, in cylindrical coordinates. Also, the dissertation provided a 3D-FDTD program, and adopted generalized perfectly matched layer (GPML) as absorbing boundary condition to absorb reflective wave.In chapter three, through the introduction of Born iteration and Frèchet derivative, the dissertation introduced the basic theories of full waveform inversion method, and also showed the derivation of formulas. After constructing misfit function, the dissertation used conjugate gradient optimized method to solve it. In addition, the dissertation proposed an adaptive LMS filter and Kalman filtering deconvolution, based on dyadic wavelet, to de-noising and update source wavelet.In chapter four, the dissertation tested the methods brought above. The dissertation used a series of synthetic models to test FDTD code, compared to analytical results, and analyzed the diffusive rules and influential factors of crosshole electromagnetic wave. On inversion, the dissertation used synthetic data to test full waveform inversion method, comparing to CT.In chapter five, the dissertation drawled conclusions and the deficiencies of the whole research.The innovations of the dissertation are as follows:Firstly, wrote a 3D-FDTD program in cylindrical coordinates, and adopted generalized perfectly matched layer (GPML) as absorbing boundary condition. It is proved by synthetic modeling that it is correct and accurate, and is suitable for being the modeling method for inversion.Secondly, because full waveform inversion was suffered a lot from the noises and estimated value of source waveform in wild, the dissertation combined adaptive LMS filter and Kalman filtering deconvolution with full waveform inversion, and put forward the concept of adaptive full waveform inversion.