基于小波变换大地电磁时间序列的极化分析及应用

发布时间：2018-04-05 08:50

本文选题：大地电磁　切入点：极化分析　出处：《中国地质大学(北京)》2015年硕士论文

【摘要】：20世纪90年代以来大地电磁测深法发展迅速,并被广泛应用于区域断裂构造、流体运动、板块构造以及大陆动力学等方面的研究中。在实测数据处理以及阻抗的计算和分析中,信号中的噪声是一个不可忽视的问题。实践证明,在工业区、市区和农业区附近采集的大地电磁时间序列极易受到人文噪声的干扰。为获得稳定可靠的阻抗估计,检测和分离人文噪声便成为了现代大地电磁法必须要解决的问题。多数人文噪声源在空间属性中对于大地电磁场来说是局部的,它所产生的电磁信号是在特定方向上线性极化的。与之相反,广泛存在于自然界中的大地电磁场信号由于其场源的多样性,所以理论上在天然场中观测得到的大地电磁信号是非线性极化的。这就是极化分析应用于大地电磁数据处理的基础。论文首先基于地球物理小波库,编写MATLAB程序从而实现了极化分析算法。并利用此算法针对大地电磁信号中常见的几种噪声,研究了它们的极化属性特征。其过程主要是：首先将大地电磁时间序列中常见的几种噪声添加到受干扰较少的实测数据中,进行极化分析。然后选择含有噪声的多个实测数据段进行极化分析。最后得到了脉冲干扰和矩形干扰等大地电磁信号中的典型噪声的极化属性特征。发现了脉冲干扰和矩形干扰往往会引起极化椭圆率的零值异常,在极化属性图中可以看到明显的楔形零值区域；在相应时频区域内,引起极化角围绕一个固定值变化的现象,相位差也会出现类似的现象。在椭圆率方面,脉冲干扰引起的楔形异常区域往往体现为“窄长”型,而矩形干扰引起的异常区往往表现为“矮宽”型。此后,进一步根据这些噪声的极化属性特征(主要依靠极化椭圆率的特征)来对大地电磁时间序列中的脉冲干扰和矩形干扰进行时域定位,并使用了大量的实测数据来验证时域定位算法的可靠性。最后根据得到的噪声时域位置以及噪声存在类型,依次对相应位置处的时频谱进行处理,继而经过小波反变换得到去除噪声后的信号。通过大量实验数据和实测数据的检验,证明了极化分析方法应用在大地电磁实测数据中进行噪声定位及信噪分离的可行性。
[Abstract]:Magnetotelluric sounding has been developed rapidly since 1990s and has been widely used in the study of regional fault structure fluid movement plate tectonics and continental dynamics.In the calculation and analysis of the measured data and impedance, the noise in the signal is a problem that can not be ignored.It has been proved that magnetotelluric time series collected in industrial area, urban area and agricultural area are easily disturbed by human noise.In order to obtain stable and reliable impedance estimation, detection and separation of human noise has become a problem that must be solved in modern magnetotelluric method.Most of the human noise sources are local to the magnetotelluric field in their spatial properties, and the electromagnetic signals generated by them are linearly polarized in a particular direction.On the contrary, the magnetotelluric field signals widely existing in nature are nonlinear polarized because of the diversity of their field sources.This is the basis of polarization analysis applied to magnetotelluric data processing.Firstly, based on geophysical wavelet library, MATLAB program is written to realize polarization analysis algorithm.The polarization properties of magnetotelluric signals are studied by using this algorithm.The main process is as follows: firstly, some common noises in magnetotelluric time series are added to the measured data with less interference, and polarization analysis is carried out.Then, several real data segments with noise are selected for polarization analysis.Finally, the polarization properties of typical noises in magnetotelluric signals such as pulse interference and rectangular interference are obtained.It is found that pulse interference and rectangular interference often lead to zero anomaly of polarization ellipticity, and the wedge-shaped zero value region can be seen in the polarization attribute map, and the polarization angle changes around a fixed value in the corresponding time-frequency region.The phase difference will also appear similar phenomenon.In terms of ellipticity, the wedge-shaped anomaly caused by pulse interference is usually of "narrow length" type, while the anomalous region caused by rectangular interference is often characterized by "short width" type.Then, according to the polarimetric properties of these noises (which mainly depend on the characteristics of polarization ellipticity), the pulse interference and rectangular interference in magnetotelluric time series are located in time domain.A large number of measured data are used to verify the reliability of the time domain localization algorithm.Finally, according to the noise time domain position and the noise existence type, the time spectrum at the corresponding position is processed in turn, and then the signal after removing noise is obtained by wavelet inverse transform.The feasibility of using polarization analysis method to locate noise and separate signal and noise in magnetotelluric data is proved by testing a large number of experimental data and measured data.
【学位授予单位】：中国地质大学(北京)
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.325

【参考文献】