奇异值分解(SVD)实现地震波场分离与去噪新思路

依据不同性质的地震信号（反射波、折射波、直达波、面波、VSP上\下行波、多次波、随机干扰等）之间在运动学、视速度和相干性上的差异,借助某种数学变换、SVD分解与重构联合、时域与频域结合的方式,通过这些间接的处理手段,把要提取的目标信号或要剔除的干扰信号转换到一种相干性更好的空间域中,再进行SVD分解与重构,最充分利用SVD滤波技术特点,实现地震波场分离与去噪,而不是直接对信号进行SVD分解与重构来实现地震波场分离与去噪,这样做可有效地避免以往对SVD波场分离与去噪技术应用空间狭窄、有效信号损失严重等缺陷性.     

Simultaneous Denoising and Interpolation of Seismic Data via the Deep Learning Method

Utilizing data from controlled seismic sources to image the subsurface structures and invert the physical properties of the subsurface media is a major effort in exploration geophysics. Dense seismic records with high signal-to-noise ratio (SNR) and high fidelity helps in producing high quality imaging results. Therefore, seismic data denoising and missing traces reconstruction are significant for seismic data processing. Traditional denoising and interpolation methods rarely occasioned rely on noise level estimations, thus requiring heavy manual work to deal with records and the selection of optimal parameters. We propose a simultaneous denoising and interpolation method based on deep learning. For noisy records with missing traces, we adopt an iterative alternating optimization strategy and separate the objective function of the data restoring problem into two sub-problems. The seismic records can be reconstructed by solving a least-square problem and applying a set of pre-trained denoising models alternatively and iteratively.We demonstrate this method with synthetic and field data.     

SVD与EMD联合去噪方法在地震勘探数据处理中的研究与应用

将基于倾角扫描的奇异值分解与经验模式分解法相结合应用到地震资料随机噪声压制中。首先利用经验模式分解法消除部分噪声,增强地震道有效信号的相关性,再利用奇异值分解对地震信号进行同相轴自动追踪,截取小时窗数据体,并进行同相轴拉平处理,经SVD计算小时窗数据中心点的值来代替计算样点的值,最终实现随机噪声的压制。理论模型试算和实际资料处理表明,本文提出的EMD-SVD方法简单易行,比单一的SVD方法去噪效果更显著有效地消除了地震资料中的随机噪声,提高了地震资料的信噪比,并改善了叠加剖面的质量。     

基于CNN和Transformer耦合网络的低剂量CT图像重建方法

在投影角度个数不变的情况下,降低每个角度下的射线剂量,是一种有效的低剂量CT实现方式,然而,这会使得重建图像的噪声较大。当前,以卷积神经网络（CNN）为代表的深度学习图像去噪方法已经成为低剂量CT图像去噪的经典方法。受Transformer在计算机视觉任务中展现的良好性能的启发,本文提出一种CNN和Transformer耦合的网络（CTC）,以进一步提高CT图像去噪的性能。CTC网络综合运用CNN的局部信息关联能力和Transformer的全局信息捕捉能力,构建8个由CNN部件和一种改进的Transformer部件构成的核心网络块,并基于残差连接机制和信息复用机制将之互联。与现有4种去噪网络比较,CTC网络去噪能力更强,可以实现高精度低剂量CT图像重建。      

基于同步大地电磁时间序列依赖关系的噪声处理

本文从信号与系统的角度讨论了同步大地电磁时间序列信号之间的依赖关系,选取高信噪比的时间序列信号作为先验数据,用最小二乘法估算依赖关系;结合参考道的数据,合成本地道含噪声时段的数据,最后用合成数据替代噪声段数据,组成新数据,从而在时域中去除大地电磁噪声.西藏地区高信噪比实测数据的试算结果表明,无论电场还是磁场,信号之间的依赖关系是相对稳定的,只与先验数据的长度有关,与时间无关;虽然不同参考点之间的依赖关系不同,但都可以精确合成本地点数据,与参考点地下电性结构和参考距离无关.仿真实验显示,去噪后的信号与原始信号基本一致.实测数据处理结果表明,该方法可以有效去除强噪声干扰,抑制中高频段的近场源效应,同时保留了微弱的有效信号,保证了处理结果的正确性.最后针对方差比方法无法识别的方波噪声,提出了一种简单的平移方法,成功去除了持续时间大于窗口长度的方波噪声;将该方法与远参考技术结合,可以有效抑制近场源噪声干扰,获得光滑连续并且可信的测深资料.     

Pitfalls of 1D inversion of small-loop electromagnetic data for detecting man-made objects

In order to interpret field data from small-loop electromagnetic (EM) instruments with fixed source–receiver separation, 1D inversion method is commonly used due to its efficiency with regard to computation costs. This application of 1D inversion is based on the assumption that small-offset broadband EM signals are insensitive to lateral resistivity variation. However, this assumption can be false when isolated conductive bodies such as man-made objects are embedded in the earth. Thus, we need to clarify the applicability of the 1D inversion method for small-loop EM data. In order to systematically analyze this conventional inversion approach, we developed a 2D EM inversion algorithm and verified this algorithm with a synthetic EM data set. 1D and 2D inversions were applied to synthetic and field EM data sets. The comparison of these inversion results shows that the resistivity distribution of the subsurface constructed by the 1D inversion approach can be distorted when the earth contains man-made objects, because they induce drastic variation of the resistivity distribution. By analyzing the integrated sensitivity of the small-loop EM method, we found that this pitfall of 1D inversion may be caused by the considerable sensitivity of the small-loop EM responses to lateral resistivity variation. However, the application of our 2D inversion algorithm to synthetic and field EM data sets demonstrate that the pitfall of 1D inversion due to man-made objects can be successfully alleviated. Thus, 2D EM inversion is strongly recommended for detecting conductive isolated bodies, such as man-made objects, whereas this approach may not always be essential for interpreting the EM field data.     

基于曲波变换的航空电磁数据调平方法研究

航空电磁法由于高效和高精度的特点广泛应用于地质填图、矿产资源、地下水、及环境与工程等勘查.然而,航空电磁系统处于动态环境,噪声影响严重,航空电磁数据处理至关重要.航空电磁数据噪声除随机成分外,还包括有各种效应引起的畸变,数据去噪需要依据噪声特征进行处理.航空电磁数据调平是航空电磁数据处理中至关重要的步骤,它能有效去除数据中由飞机飞行条件变化导致系统状态变化而产生的异常.传统的调平方法由于效率较低、易产生数据畸变等受到限制.为了克服这些局限性,我们提出一种基于曲波变换的数据调平方法.该方法得益于曲波变换多尺度和多方向性特征,可以有效地提取数据中的调平误差并予以去除.与此同时,利用该方法我们可以对非规则测区数据进行直接调平,无需进行测区分割,显著提高调平效率和普适性.为了检验本文曲波变换调平方法的有效性,我们将其应用于理论数据以及在爱尔兰Waterford地区实测的航电数据调平.实验结果表明该方法有效地去除调平误差的同时很好地保留有用信号.     

Comparisons of wavelets, contourlets and curvelets in seismic denoising

Seismic synthetic records represent wave-front components which contain abundant damageable directional information about important geologic substances. However, the information is often polluted by random noises. Unlike coherent noise, random noise may be incoherent in space and time, and sometimes unpredictable. One of the main tasks in seismic denoising is to eliminate the random noise and useless interferential wave components while preserving or recovering the important seismic features. In this paper, we investigate multi-resolution (MR) methods including wavelets, contourlets and curvelets for seismic denoising of random noise. Discussions and interpretations for these methods are presented in detail to evaluate their performances. Furthermore, a combination scheme of wavelets and curvelets is applied to seismic random denoising by solving an l₁ norm optimization problem. The combined scheme aims at taking advantage of the respective merits of wavelets and curvelets in order to obtain better effects. Experimental results indicate that the directional wavelets such as contourlets and curvelets are prominent for seismic profiles containing textural features and the combination method shows promising performances than individual transforms.     

Analysis and interpretation of anomalous conductivity and magnetic permeability effects in time domain electromagnetic data: Part I: Numerical modeling

Time domain electromagnetic (TDEM) response is usually associated with eddy currents in conductive bodies, since this is the dominant effect. However, other effects, such as displacement currents from dielectric processes and magnetic fields associated with rock magnetization, can contribute to TDEM response. In this paper we analyze the effect of magnetization on TDEM data. We use a 3-D code based on finite-difference method, developed by Wang and Hohmann [Geophysics 58 (1993) 797], to study transient electromagnetic field propagation through a medium containing bodies with both anomalous conductivity and anomalous magnetic permeability. The remarkable result is that the combination of anomalous conductivity and permeability within the same body could increase significantly the anomalous TDEM response in comparison with purely conductive or purely magnetic anomalies. This effect has to be taken into account in interpretation of TDEM data over electrical inhomogeneous structures with potentially anomalous magnetic permeability.     

Geophysical aspects of man-made electromagnetic noise in the earth—A review

The paper gives a summary of geophysical aspects of man-made electromagnetic noise in the Earth as follows:

- EM distortion effects of man-made constructions below and over the Earth's surface defined as ‘passive-noise’,

- field observation of EM disturbances due to ‘active’ man-made sources,

- EM source mechanism of some important active sources from a geophysical point of view,

- efforts in order to improve the signal-to noise ratio by instrumental, methodological and dataprocessing ways,

- application of man-made EM noise for geophysical prospecting.

The paper is based on world-wide EM noise survey studies published mainly in geophysical journals.     

基于小波熵与相关性相结合的小波模极大值地震信号去噪

小波模极大值去噪算法中将高频小波系数全部当做噪声处理, 忽略了高频小波系数中仍含有的有用信息, 从而导致了模极大值传播点错选现象以及计算出的噪声方差中仍含有用信息． 针对这些问题, 提出了小波熵与相关性相结合的小波模极大值去噪算法． 将高频小波系数进行相关处理, 确定有效信号的位置; 将最大尺度上的高频小波系数划分成若干个小区间, 计算各区间小波熵; 以小波熵最大区间的高频小波系数的平均值作为噪声方差, 根据Donoho提出的阈值公式计算最大尺度上的阈值; 经阈值比较得到的模极大值点位置与相关处理得到的有用信息的位置进行比较, 保留相同位置的模极大值, 剔除位置不同由噪声引起的模极大值点; 采用即兴(Adhoc)算法逐级搜索各尺度上的模极大值, 并用交替投影算法进行重构． 该算法实现了阈值的自适应选取, 并有效解决了去除错选模极大值传播点的问题． 将本算法和传统去噪方法用于仿真信号处理中, 经对比分析验证了本算法的有效性．      

基于3MNet去噪网络的快速EPRI成像

电子顺磁共振成像（EPRI）是一种先进的氧成像技术。当前EPRI的瓶颈问题是扫描速度过慢,其原因是每个角度下的投影信号需要被重复采集几千次,以压制随机噪声。一种实现快速扫描的方法是减少投影信号的重复采集次数,然而这又使得投影信号信噪比降低,重建出来的图像噪声较大。为有效压制重建图像中的噪声,本文提出一种基于多通道、多尺度、多拼接的（3MNet）图像去噪网络,以实现高精度快速EPRI成像。该网络由3个子网络构成。第1个子网络是基于注意力机制的卷积网络,其输出的特征图像与输入图像拼接以构成后端网络的输入;第2个子网络是3通道卷积网络;第3个子网络是多尺度卷积网络。实验结果表明,本文提出的3MNet网络可以实现EPRI图像的高精度去噪,进而实现快速成像。     

音频大地电磁法强干扰压制试验研究

压制非相关噪声已经有多种比较成熟的技术方法,如多次叠加、远参考、Robust阻抗估计等,但这些方法对压制矿集区存在的强电磁噪声(通常属于相关噪声)基本没有效果.本文在前期提出的针对矿集区大地电磁强噪声压制的数学形态滤波基础上,开展了音频大地电磁法强干扰压制的试验研究.在四川西昌某AMT测点附近,布置了大功率电磁发送站,通过接地导线向地下供20A的方波电流,同时在供电期和间歇期用凤凰公司MTU-5A仪器观测4个水平分量的电磁场.在室内采用数学形态滤波和阈值法相结合的手段在时间域对含有人工源强干扰的电磁场数据进行处理,以视电阻率对比、频谱和极化方向等参数对处理效果进行评价.结果表明:结构元素恰当的数学形态滤波可以对人工源强干扰进行识别、去除,阈值法则可以消除形态滤波后的脉冲干扰,二者结合可以有效地压制AMT中的强干扰噪声,提高数据可靠性.本文结果对AMT实际资料处理具有重要的实用和参考价值.     

The results of marine electromagnetic sounding with a high-power remote source in the Kola Bay in the Barents Sea

The first Russian six-component seafloor electromagnetic (EM) receivers were tested in an experiment carried out in Kola Bay in the Barents Sea. The signals transmitted by a remote high-power ELF source at several frequencies in the decahertz range were recorded by six receivers deployed on the seafloor along the profile crossing the Kola Bay. Although not all the stations successfully recorded all the six components due to technical failures, the quality of the data overall is quite suitable for interpretation. The interpretation was carried out by the three-dimensional (3D) modeling of an electromagnetic field with neural network inversion. The a priori geoelectrical model of Kola Bay, which was reconstructed by generalizing the previous geological and geophysical data, including the data of the ground magnetotelluric sounding and magnetovariational profiling, provided the EM fields that are far from those measured in the experiment. However, by a step-by-step modification of the initial model, we achieved quite a satisfactory fit. The resulting model provides the basis for introducing the corrections into the previous notions concerning the regional geological and geophysical structure of the region and particularly the features associated with fault tectonics.     

基于小波变换的大尺度岩体结构微震监测信号去噪方法研究

为将小波去噪方法应用于大尺度岩体结构微震监测信号的去噪研究,首先在MATLAB环境下进行仿真,验证了使用Symlet6小波进行小波去噪的可行性;利用4种自适应阈值规则对含噪信号进行去噪对比,结果表明4种阈值去噪后的信号在均方差较小的情况下都极大地提高了信号的信噪比,有效地去除了噪声,对不同的含噪信号,无偏似然原则阈值去...     

Cancelling directional EM noise in magnetoteilurics1

The prospecting of densely urbanized areas by the measurement of magnetic and electric natural fields is severely hampered by electromagnetic (EM) noise. Active man-made EM noise sources can generally be considered fixed in space, thus affecting the magnetotelluric (MT) signals of a measuring site mainly along their polarization directions. Taking advantage of the impulsive nature of polarized EM noise, a time-domain directional noise cancelling (DNC) technique is proposed. The comparison of noisy data with data predicted, using a low noise reference signal or with data interpolated whenever no reference is available, allows the detection of the most likely noise sources with prevailing directional patterns using a Bayes's criterion. The DNC approach is general and can be adapted, depending on the reference signal used (single-site or remote-reference). In field data, hodograms of the prediction residuals basically confirm the directional noise model assumed in DNC. An example is presented in which the DNC technique has been applied to a single-site MT survey carried out in northern Italy, where the signal was heavily corrupted by noise with prevailing directional properties due to the densely urbanized area. MT apparent resistivities and phases obtained at the site of the survey before and after DNC are presented and discussed.     

基于压缩感知的加权MCA地震数据重构方法

地震数据规则化重构是地震资料处理十分重要的基础性工作.压缩感知理论打破了香农采样定理的制约,利用信号在某个变换域的稀疏特性重构出完整的信号,在地震数据重构领域得到了很好的应用.深反射地震剖面大都布置在地质构造比较复杂的区段,复杂的地质构造使深反射地震剖面上的波阻特征复杂,采用单一稀疏变换不能最有效地表征数据的内部结构特征.MCA(形态成分分析)方法将信号分解为几种形态特征区别明显的分量来逼近数据的内部复杂结构,但是对各成分简单的叠加仍然无法有效地描述复杂构造数据的各种特征.结合两种方法的优点,本文提出了一种新的基于压缩感知的重构算法框架,在MCA方法的基础上对各稀疏字典进行加权,在迭代中不断更新各个稀疏字典的权值系数,对信号内部的各种特征进行最优描述,从而实现对信号的高质量重构.模型测试和实际资料处理结果表明:基于压缩感知的加权MCA方法不仅可以对地质构造复杂的地震数据进行高效的插值重建,而且可以很好的消除空间假频.     

基于迭代去噪的多源地震混合采集数据分离

多源地震混合采集采用随机线性编码方式同时激发多个震源,检波器连续接收地震信号,获得波场混叠的炮记录.该采集技术能够显著提升采集效率和成像质量,其实现关键在于炮分离,即将相互混叠的多源波场数据彼此分离,获得传统采集的单炮记录.最小平方算法只能得到伪分离记录,不能去除混叠噪声.在高混合度的混叠数据中,混叠噪声的能量往往数倍于有效信号,炮分离难度倍增.但在伪分离记录中,该噪声在除共炮点道集以外的其它时域均为随机分布.本文研究提出了一种多时域组合迭代去噪的炮分离技术:通过运用多级中值滤波与Curvelet阈值迭代去噪算法,在不同时域根据混叠噪声特性采用相应的去噪手段,并设计迭代算法优化炮分离结果.实际资料处理结果证明:将本方法应用于高混合度的混叠数据,无论是分离质量还是计算效率,都有明显提升.     

重建地下介质相对衰减的分布:新的成象方法

在传统的连续波电磁层析成象方法中,影响线性方程组的解的重要参量--波源的辐射功率因子k,常常难以确定或根本无法确定,使方法的应用受到严重限制.本文提出了新的电磁层析成象方法,其线性方程组与k无关,能用于传统方法无法应用的情况.新方法重建的地下介质相对衰减的二维分布图象,比传统方法重建的绝对衰减的二维分布图象具有更好的分辨力和稳定性,更有利于地质解释.新方法已成功应用于8个不同地区电波勘探资料的处理.     

含噪声数据反演的概率描述

根据贝叶斯理论给出了对含噪声地球物理数据处理的具体流程和方法,主要包括似然函数估计和后验概率计算.我们将数据向量的概念扩展为数据向量的集合,通过引入数据空间内的信赖度,把数据噪声转移到模型空间的概率密度函数上,即获得了反映数据本身的不确定性的似然函数.该方法由于避免了处理阶段数据空间内的人工干预,因而可以保证模型空间中的概率密度单纯反映数据噪声,具有信息保真度高、保留可行解的优点.为了得到加入先验信息的后验分布,本文提出了使用加权矩阵的概率分析法,该方法在模型空间直接引入地质信息,对噪声引起的反演多解性有很强的约束效果.整个处理流程均以大地电磁反演为例进行了展示.