【作者】 龚玉蓉；

【导师】 汤井田；

【作者基本信息】 中南大学 ， 地球探测与信息技术， 2011， 博士

【摘要】 大地电磁测深(Magnetotelluric,MT)野外施工方便,成本低廉,探测深度大,不受高阻层屏蔽影响,对低阻层分辨能力强,广泛的应用在能源和资源勘探中。然而,静态效应使得频率域电磁测深数据的解释复杂化、引起深度推测的误差。本文在研究分析国内外各种静态效应校正方法的基础上,在863计划项目(2007AA06Z134)资助下,利用二维小波理论对三维大地电磁测深数据中静态效应的压制进行了研究。本文先对论文的研究背景与现状进行了探讨,总结了静态效应压制研究方法,阐述了小波变换的发展和起源。小波变换良好的时频分析特性,为信号消噪,构建滤波器提供了非常有效的工具。在此基础上本文做了以下土作：首先描述了大地电磁测深数据的静态效应的物理原因以及其特征,研究了二维静态效应与三维静态效应之间的区别,给出了三维大地电磁静态效应的压制方法。其次,从小波变换的基本原理出发,基于小波分析的去噪原理研究了二维小波分析在三维大地电磁测深数据中的静态效应压制的应用。利用二维小波变换的分解与重构算法,良好的时频特性,引入去噪因子合理的选择阈值函数以及阈值准则,通过对含有三维静态效应的地电模型仿真以及实测数据进行处理,表明了二维小波分析能够有效的从三维大地电磁测深数据中分离出静态效应,再利用二维小波分析的重构算法可以很好的重构压制了静态效应的卡尼亚视电阻率。实验表明,该方法能够遵循静态效应消除的原则,很好的体现原始模型的层状特征,由小三维体产生的静态效应能够得到很好的压制。然后,对小波包分析进行了研究,小波包分析比小波变换具有更好的时频特性,它可以对小波变换没有加以分析的高频部分继续进行细分,呈现更好的细节分析能力。基于小波包这个优良的特性,提出了基于二维小波包分析在三维大地电磁测深数据中静态效应压制的方法。利用这种方法对三维大地电磁测深数据中的静态效应进行压制,首先应用最小熵标准进行最佳小波包基的选择,确定小波包基函数对三维的卡尼亚视电阻率进行处理,分离出其中的高频部分,即静态效应；然后重构。通过低频重构能够很好的体现地电模型的层状结构,同时,因为小波包的提取信号细节的能力很强,在需要利用静态效应时,可以用小波包来精确的提取二维、三维静态效应,重构引起静态效应的不均匀体细节。通过以上的研究,表明了二维小波分析和二维小波包分析具有良好的时频局部分析特性,通过合理的选择小波(包)基以及分解的层数能够有效地分析处理大地电磁测深数据中的静态效应,将这两种方法用在了实测数据的静态效应消除中,从结果分析,利用二维小波(包)可以很好的抑制地表不均匀体的垂向拉长。最后讨论了三维大地电磁测深数据的静态效应压制中算法的不足之处,并对下一步的研究工作的展开提出了一些建议。更多还原

【Abstract】 Magnetotelluric (MT) Sounding, with the advantages of convenient wild construction, the lower cost, the deeper depth of invertigation, and being free from the shielding of resistive formation, is widely used to energy exploration. The existence of static effects in the frequency domain electromagnetic sounding makes the Magnetotelluric sounding data interpretation complicated and causes errors in the depth of speculation. Based on the study of domestic and international various correction method of static effects,this paper, studied the wavelet theory to process the static effects of three-dimensional Magnetotelluric Sounding data using two dimension under the help of 863 project(2007AA06Z134).First the paper discussed background of the research and current situation, explained the research on the suppression method of static effect, and state wavelet transform and wavelet neural network development and origins. The good frequency analysis characteristics of the wavelet transform provides a very effective tool for noise cancelling and constructing filter. The main work is as follows:Firstly, it describes the physical cause and the characteristic of Magnetotelluric sounding data static effects, discusses the difference between two-dimensional and three-dimensional static effects, gives the methods of suppressing the static effect of three-dimensional MT sounding.Secondly, the paper from the basic principle of wavelets transform, in view of the denoising principle of wavelet analysis, researches the application of two-dimensional wavelet in the suppression static effects of three-dimensional Magnetotelluric data. It makes use of the two-dimensional wavelet transform decomposition and reconstruction algorithms, its good frequency analysis characteristics, to introduce the denoising factor to choose threshold function and the threshold criteria reasonably, and then by simulating the three-dimensional earth electricity model and the measured data that have static effect, shows that the two-dimensional wavelet transform analysis can separate the static effect from three-dimensional magnetotelluric sounding data. Then, using the two-dimensional wavelet analysis to reconstruction algorithm can get the Cagniard apparent resistivity which is suppressed the static effect. Experiments show that the method can reflect the layered characteristics of the original model easily according to the principle of eliminating static effect, and the static effect produced by the small three-dimensional subject can be well suppressed.In addition, the paper studies on the wavelets packet. The wavelet packet transform’s frequency characteristics is better than wavelet transform’s. It can analyse the high-frequency parts of signal which was not processed by the wavelet transform, showing the better detail analysis ability. Based on the fine features of wavlet packet, we proposed the methods to suppress the static effect that the two-dimensional wavelet packet analysis in the three-dimensional MagnetoTelluric data. Using this method to oppress the static effect of three-dimensional magnetotelluric sounding data, firstly, we apply the minimum entropy criteria to choose the best wavelet packet basis function, processing the cagniard apparent resistivity curse, and then separate out the high-frequency part, which is called static effect. Then reconstruct it and get the apparent resistivity. Because wavelet packet detailed analysis shows a better ability, when needed, use this feature to extract the two-dimensional and three-dimensional magnetotelluric sounding data suppression of the static effects, and reconstruct the heterogeneous body details that causes the static effects.The above study shows that two-dimensional wavelet analysis and two-dimensional wavelet packet have the good characteristics of part analysis, by a reasonable choice of wavelets (packet) and the the number of decomposition layers, using the two methods to static effects cancelling, it have a good result to process the static effect of magnetotelluric sounding data by means of the two-dimensional wavelet analysis.Finally, the thesis discusses the algorithm deficiencies of the system that static effect suppressing of three-dimensional magnetotelluric sounding data, and proposes some suggestions about the further research work.更多还原