Improved Local Wavenumber Methods in the Interpretation of Potential Field Data

We present two new potential-inversion methods for estimating the depth and the nature (structural index) of the source, which use various combinations of different forms of local wavenumbers and the information about the horizontal location to estimate individually the depth and the nature of a magnetic source. The improved local wavenumber methods only use the horizontal offset and vertical offset of local wavenumbers to estimate the depth and the structural index of the source, so they yield more stable results compared with the results obtained by current methods that require the derivatives of local wavenumbers. Tests conducted with synthetic noise-free and noise-corrupted magnetic data show that the proposed methods can successfully estimate the depth and the nature of the geologic body. However, our methods are sensitive to high-wavenumber noise present in the data, and we reduced the noise effect by upward continuing the noise-corrupted magnetic data. The practical application of the new methods is tested on a real magnetic anomaly over a dike whose source parameters are known and the inversion results are consistent with the true values.     

Interpretation of the magnetic field from a sphere in the wavenumber domain

A Fourier transformation of the magnetic field from a magnetized sphere allows a particularly simple interpretation of the parameters of the sphere. The depth to the centre and total magnetic moments of the sphere are related to the intercept and slope of the power spectrum. The horizontal centre and direction of magnetization are related to slope and intercept of the phase spectra in two perpendicular, horizontal directions. Examples with artificial data contaminated by various noise components are presented.     

Direct interpretation of self-potential anomalies produced by a vertical dipole

In the present paper a new method is proposed for the quantitative interpretation of self-potential anomalies which are produced by a vertical dipole. First the mathematical expression of the wavenumber spectrum of the self-potential anomaly is deduced. It is pointed out that at relatively high wavenumbers the behavior of the amplitude spectrum is controlled by the closer to the surface pole at depth h. On the other hand, the “width” of the amplitude spectrum depends on the depth h and the dipole length L.Making a proper mathematical transformation of the amplitude spectrum, and applying the least squares method, it is possible to calculate the depth to the upper pole. The dipole length may then be calculated, by solving numerically a characteristic algebraic equation, as long as the “width” of the amplitude spectrum has been previously defined.The proposed method is applied on a well known self-potential profile from Greece. The calculated parameters of the polarized body are in good agreement with real data. Experimentation with synthetic models in which random noise was introduced, showed that this method gives reliable results if the noise amplitude is not more than 20% of the signal amplitude. It is clearly more efficient than the methods which are based on the model of the point pole or the dipole with a small length. It can also give good results if the horizontal extensions of the polarized body are not more than a few tenths of the depth of the upper pole. If the polarized body is tilted, the depth of the upper pole can be calculated with satisfactory accuracy.The direct interpretation method which is proposed in the present paper, may be useful in mineral exploration, and particularly if the target of interest is the detection of massive sulfide mineralization.     

一种基于频率域偏振分析提高三分量地震资料信噪比的方法

三分量地震资料包含着丰富的运动学和动力学信息,各种随机噪声的存在严重降低了资料的信噪比,给有效信息的提取带来了一定的困难.本文在引入多窗谱分析方法的基础上,通过对谱密度矩阵特征问题的求解提取三分量地震记录的偏振参数,实现频率域偏振分析;根据所求得的偏振参数以及随机噪声和有效信号的偏振特性差异,设计相应的自适应滤波器,通过该滤波器的滤波作用,最终实现压制随机噪声的目的.     

Estimating magnetic dike parameters using a non‐linear constrained inversion technique: an example from the Särna area,west central Sweden

In this paper, we describe a non‐linear constrained inversion technique for 2D interpretation of high resolution magnetic field data along flight lines using a simple dike model. We first estimate the strike direction of a quasi 2D structure based on the eigenvector corresponding to the minimum eigenvalue of the pseudogravity gradient tensor derived from gridded, low‐pass filtered magnetic field anomalies, assuming that the magnetization direction is known. Then the measured magnetic field can be transformed into the strike coordinate system and all magnetic dike parameters – horizontal position, depth to the top, dip angle, width and susceptibility contrast – can be estimated by non‐linear least squares inversion of the high resolution magnetic field data along the flight lines. We use the Levenberg‐Marquardt algorithm together with the trust‐region‐reflective method enabling users to define inequality constraints on model parameters such that the estimated parameters are always in a trust region. Assuming that the maximum of the calculated g_zz (vertical gradient of the pseudogravity field) is approximately located above the causative body, data points enclosed by a window, along the profile, centred at the maximum of g_zz are used in the inversion scheme for estimating the dike parameters. The size of the window is increased until it exceeds a predefined limit. Then the solution corresponding to the minimum data fit error is chosen as the most reliable one. Using synthetic data we study the effect of random noise and interfering sources on the estimated models and we apply our method to a new aeromagnetic data set from the Särna area, west central Sweden including constraints from laboratory measurements on rock samples from the area.     

Continuous wavelet transform,theoretical aspects and application to aeromagnetic data at the Huanghua Depression,Dagang Oilfield,China

We use the continuous wavelet transform based on complex Morlet wavelets, which has been developed to estimate the source distribution of potential fields. For magnetic anomalies of adjacent sources, they always superimpose upon each other in space and wavenumber, making the identification of magnetic sources problematic. Therefore, a scale normalization factor, a^?n, is introduced on the wavelet coefficients to improve resolution in the scalogram. By theoretical modelling, we set up an approximate linear relationship between the pseudo‐wavenumber and source depth. The influences of background field, random noise and magnetization inclination on the continuous wavelet transform of magnetic anomalies are also discussed and compared with the short‐time Fourier transform results. Synthetic examples indicate that the regional trend has little effect on our method, while the influence of random noise is mainly imposed on shallower sources with higher wavenumbers. The source horizontal position will be affected by the change of magnetization direction, whereas the source depth remains unchanged. After discussing the performance of our method by showing the results of various synthetic tests, we use this method on the aeromagnetic data of the Huanghua depression in central China to define the distribution of volcanic rocks. The spectrum slices in different scales are used to determine horizontal positions of volcanic rocks and their source depths are estimated from the modulus maxima of complex coefficients, which is in good accordance with drilling results.     

Noise transfer in variable‐depth streamer deghosting

Single‐component towed‐streamer marine data acquisition records the pressure variations of the upgoing compressional waves followed by the polarity‐reversed pressure variations of downgoing waves, creating sea‐surface ghost events in the data. The sea‐surface ghost for constant‐depth towed‐streamer marine data acquisition is usually characterised by a ghost operator acting on the upgoing waves, which can be formulated as a filtering process in the frequency–wavenumber domain. The deghosting operation, usually via the application of the inverse Wiener filter related to the ghost operator, acts on the signal as well as the noise. The noise power transfer into the deghosted data is proportional to the power spectrum of the inverse Wiener filter and is amplifying the noise strongly at the notch wavenumbers and frequencies of the ghost operator. For variable‐depth streamer acquisition, the sea‐surface ghost cannot be described any longer as a wavenumber–frequency operator but as a linear relationship between the wavenumber–frequency representation of the upgoing waves at the sea surface and the data in the space–frequency domain. In this article, we investigate how the application of the inverse process acts on noise. It turns out that the noise magnification is less severe with variable‐depth streamer data, as opposed to constant depth, and is inversely proportional to the local slant of the streamer. We support this statement via application of the deghosting process to real and numerical random noise. We also propose a more general concept of a wavenumber–frequency ghost power transfer function, applicable for variable‐depth streamer acquisition, and demonstrate that the inverse of the proposed variable‐depth ghost power transfer function can be used to approximately quantify the action of the variable‐depth streamer deghosting process on noise.     

汶川地震前后四川区域GPS时序特征演变及统计分析

为探索地震事件对GPS坐标时间序列的长周期影响,对汶川地震前后四川GPS观测网络长约10年的解算成果进行了多参数模型噪声特征分析.基于最大似然估计方法和频谱特性分析,提取了地震前后各测站坐标序列中的噪声分量,使用Λ-统计检验,得出"白噪声+闪烁噪声"模型可以作为四川GPS区域观测网络的最优噪声组合模型,同时,地震事件使得地震前后GPS噪声分量中的白噪声、闪烁噪声、随机游走噪声等发生显著改变,说明传统谱噪声分析中简单地将地震数据拼接在一起并进行统一处理的模式并不可取;使用共模误差分析方法、区域速度场变化趋势等信息对地震前后噪声模型的改变成因进行了初步的物理解析.     

叠前地震资料噪声衰减的小波域方法研究

讨论了叠前地震记录中随机噪声的模拟方法,采用定性分析与Monte Carlo仿真相结合的方法,研究了在物理小波为基本小波的变换域该类随机噪声局部小波功率谱的统计特性,得出：其局部小波功率谱服从χ22分布.在此基础上,利用假设检验给出了在给定置信度下确定有效信号在小波变换域内能量分布空间的方法,通过在有效信号能量分布空间重建信号来压制噪声.给出了模型及实际地震资料算例,证明了本文提出的方法的有效性.     

Use of the Hilbert transform to interpret self-potential anomalies due to two-dimensional inclined sheets

Self-potential anomalies due to two-dimensional inclined sheet of finite depth extent are analysed from their horizontal and vertical derivatives via the Hilbert transform. The depths to the top and bottom of the sheet and the inclination are obtained by means of simple mathematical expressions. The method is applied to a theoretical example and to a field S.P. anomaly over a sulphide deposit in the Kalava fault zone of Cuddapah basin, India. The presence of random noise is analysed for various levels and its effect is discussed. This interpretation process can be automated for all practical purposes by simple programming.     

基于卷积神经网络的井中分布式光纤传感器地震数据随机噪声压制新技术

分布式光纤传感器(distributed fiber-optical acoustic sensor,DAS)是一种快速发展的具有巨大应用前景的地震勘探检波器技术.实际DAS地震资料往往会受到大量强能量随机噪声的干扰,通常表现为低信噪比(signal-to-noise ratio,SNR).这一现象给接下来的成像、反演以及解释带来了巨大的困难,因此如何压制DAS地震资料中的随机噪声并提高其SNR成为一个有待解决的技术问题.卷积神经网络(convolutional neural network,CNN)已经被证明是一种有效的噪声压制工具.通常情况下,CNN需要一个理论纯净地震数据集来优化网络,这极大地限制了CNN在DAS地震资料处理中的应用.在本文中,我们采用正演模拟的方法来构建理论纯净DAS地震数据集,通过正演模型的参数多样化增强数据集的真实性,从而获得适合DAS地震资料随机噪声压制的CNN去噪模型.此外,在网络结构方面,我们利用泄漏线性整流单元作为CNN的激活函数增强训练后模型对微弱有效信号的恢复能力;在训练过程中,通过能量比矩阵调节噪声片和有效信号片之间的SNR,增强CNN去噪模型对于不同SNR的DAS地震数据的适应性.模拟和实际实验均表明本文提出的这种正演模型驱动的卷积神经网络(forward-model-actuation convolutional neural network,FMA-CNN)能够有效地压制DAS随机噪声同时完整地恢复有效信号.

     

一种神经网络改进小波的地震数据随机噪声去除方法

地震资料的有效信号反射弱,且易受多次波的影响,不可避免地存在随机噪声干扰。提出一种基于神经网络改进小波的地震数据随机噪声去除方法,采用神经网络模型,识别出随机噪声信号,对该信号进行小波包分解,获取多类别随机噪声信号,采用级联BP神经网络模型提取出多类别随机噪声信号,实现地震数据的随机信号压制。实验结果显示,这种改进小波方法对地震数据随机噪声信号的去噪效果较好,在复杂沉积地质结构被探测介质的地震数据随机噪声压制方面具有较强的适用性。     

自适应非局部均值地震随机噪声压制(英文)

非局部均值滤波是一种基于图像信息冗余的去噪方法,其认为图像自身的有效结构具有一定的重复性,而随机噪声则不具备这一特点,通过利用图像本身的自相似性来达到压制随机噪声的目的,是一种全局的去噪方法。本文把这一思想引入地震数据随机噪声压制中,针对传统非局部均值滤波计算量过大的问题,文章采用分块非局部均值的方式来减少计算量;针对滤波参数选取会影响非局部均值滤波效果的问题,提出一种简单的自适应滤波参数地震数据分块非局部均值算法。模型和实际数据处理结果表明:相对于传统的去噪算法(如f-x反褶积),该方法在压制随机噪声的同时对有效信号保护地更好,具有更高的保真度,更有利于后续的处理和解释工作。     

金属矿山地地区地震勘探随机噪声的波动方程模拟

消减随机噪声是目前陆地地震勘探数据处理的关键问题之一,分析随机噪声的产生机制及特征是对其进行有效压制的先决条件.本文针对中国南方山地金属矿区的勘探环境,根据随机噪声中包含的自然噪声和人文噪声的发声机理分别确定其噪声源函数,以波动方程作为噪声传播模型对山地地区随机噪声进行建模,将随机噪声作为一个综合波场,并且与实际噪声记录进行比较.随机噪声记录作为时空域的二维随机过程,分别对模拟噪声和实际噪声记录的时间域波形(振动图)特征包括频谱、功率谱密度,相空间轨迹图,统计量特征(能量分布,累积分布,均值,方差,峰度,偏度),和空间域波形(波剖面)特征包括波数谱和统计量特征进行比较,对比结果显示在时空域模拟噪声和实际噪声都有基本相同的性质,证明了本文对随机噪声模拟方法的可行性,为进一步研究随机噪声时空域传播特性以及噪声消除奠定理论基础.     

基于分块字典学习理论的地震数据去噪

随着油气勘探观测环境愈发复杂,采集的地震数据常常掺杂各种噪声信号,导致勘探目标引起的有效微弱信号被覆盖,严重影响高精度的地震勘探数据解译,因而有效的压制地震勘探数据噪声显得越发重要。本文采用字典学习策略,将复杂地震数据进行分块,通过分块数据的字典学习获取字典原子,构建高精度的字典学习地震数据稀疏表示,通过两次迭代更新字典原子,进行数据去噪。将本文的字典学习算法应用于含随机噪声的模拟数据和实测地震勘探数据处理,验证该算法的可行性及有效性。结果表明,本文算法有效去除了随机噪声,保留了有效信号同相轴,提高了信噪比,可为复杂含噪地震数据的去噪处理提供新的技术手段。     

基于非平稳相似性系数的构造导向滤波及断层检测方法

不连续地质体（如断层）的自动检测一直以来都是叠后地震数据解释中的关键问题之一,尤其在三维情况中尤为重要.然而,大多数边缘检测和相干算法都对随机噪声很敏感,随机噪声衰减是叠后地震数据解释的另一个主要问题.针对构造保护去噪和断层检测问题,本文基于非平稳相似性系数完善一种构造导向滤波方法并且提出一种自动断层检测方法,形成了一套匹配的处理技术.该构造导向滤波既能够有效地衰减随机噪声又可以很好地保护地震资料中的断层等信息不被破坏,增强地震剖面中弯曲、倾斜同相轴的连续性.根据地震数据局部倾角走向,利用相邻道构建当前地震道的预测,通过预测道的叠加得到参考道,计算预测道与参考道之间的非平稳相似性系数可以设计出数据驱动的加权中值滤波.另一方面,预测道与原始道之间的非平稳相似性系数能够用于带有断层指示性的相干分析.这两种方法都基于构造预测和非平稳相似性系数,但是使用不同的调节参数和处理方案.理论模型和实际数据的处理结果证明了本文提出构造导向滤波和断层检测方法的有效性.     

基于DMFF-Net的地震随机噪声压制新技术

地震勘探是油气和矿产资源开发领域使用最为广泛的物探方法之一.由于采集条件的限制,地震记录中通常混杂有大量的随机噪声,导致勘探资料普遍信噪比(Signal-to-Noise Ratio,SNR)较低,这严重影响有效信号辨识的精度,为后续反演、解释等工作带来巨大挑战.此外,地震勘探随机噪声通常具有非平稳、非高斯和与信号存在频带混叠等复杂特性,导致传统方法在处理复杂勘探记录时,消噪性能可能发生退化.针对复杂勘探随机噪声消减问题,本文提出了一种新型的双层多尺度特征融合去噪网络(Double-layer Multi-scale Feature Fusion Denoising Network,DMFF-Net).该网络具有多尺度网络结构,利用多分支模块提取勘探数据不同尺度和不同分支的潜在特征,提升网络对于勘探记录复杂特征的学习能力.同时,采用跳跃连接实现浅层和深层信息的融合,提升网络对微弱信号的恢复能力.模拟和实际资料处理结果表明,相较传统地震勘探资料消噪方法而言,DMFF-Net可以更加有效地压制随机噪声,完整恢复有效信号,显著提升地震资料信噪比,在信号保幅性和微弱信号恢复能力方面更具优势.

     

任意起伏地形下重力异常三维正演及并行计算

为了进一步提高空间-波数域三维重力异常正演算法的适用范围和计算效率,本文采用任意傅里叶变换算法实现了空间-波数域三维重力异常正演,且在NVIDIA CUDA平台上进行CPU-GPU并行加速.任意傅里叶变换算法的基本思想是将二维傅里叶变换转化为两个一维傅里叶变换,一维傅里叶变换积分离散为多个单元积分累加和,离散单元中原函数采用二次插值形函数拟合,求出单元积分的解析表达式.相比现有的傅里叶变换算法,新方法具有采样灵活、积分精度高、计算速度快和傅里叶变换的截断效应小等优势.利用空间-波数域算法的高度并行性,采用CPU并行求解常微分方程,GPU并行计算任意傅里叶变换,实现了CPU-GPU并行加速方案,进一步提升了本文算法效率.利用常密度模型,对比数值解和解析解,结果表明本文算法正确;利用变密度模型对比了任意傅里叶变换算法与高斯快速傅里叶变换算法的计算效率与精度,在相近的数值精度下,本文算法波数选取少,效率高;测试CPU-GPU并行效果,结果表明相比CPU串行算法,CPU-GPU并行算法的计算效率大大提升,千万数量级节点数模型正演仅耗时数秒.最后利用实际地形数据进行三维重力异常场数值模拟,证明了新方法的高效性与实用性,对实现大规模复杂条件下重力异常精细化反演成像与综合解释有重要意义.

     

压制空间非平稳地震勘探随机噪声的ROAD径向时频峰值滤波算法

径向时频峰值滤波算法是一种有效保持低信噪比地震勘探记录中反射同相轴的随机噪声压制方法,但该算法对空间非平稳地震勘探随机噪声压制效果不理想.本文研究空间非平稳地震勘探随机噪声,即各道噪声功率不同的地震勘探随机噪声,其在径向滤波轨线上表征近似脉冲噪声,在径向时频峰值滤波过程中干扰相邻道滤波结果.为了减小空间非平稳随机噪声的影响,本文提出一种基于绝对级差统计量(ROAD)的径向时频峰值滤波随机噪声压制方法.该方法首先根据径向轨线上信号的绝对级差统计量检测空间非平稳地震勘探随机噪声,然后结合局部时频峰值滤波和径向时频峰值滤波压制地震勘探记录中的随机噪声.将ROAD径向时频峰值滤波方法应用于合成记录和实际共炮点地震记录,结果表明ROAD径向时频峰值滤波方法可以压制空间非平稳地震勘探随机噪声且不损害有效信号,有效抑制随机噪声空间非平稳对滤波结果的影响.与径向时频峰值滤波相比,ROAD径向时频峰值滤波方法更适用于空间非平稳地震勘探随机噪声压制.     

An Automatic Method of Direct Interpretation of Residual Gravity Anomaly Profiles due to Spheres and Cylinders

We have developed a least-squares minimization approach to determine the depth and the amplitude coefficient of a buried structure from residual gravity anomaly profile. This approach is basically based on application of Werner deconvolution method to gravity formulas due to spheres and cylinders, and solving a set of algebraic linear equations to estimate the two-model parameters. The validity of this new method is demonstrated through studying and analyzing two synthetic gravity anomalies, using simulated data generated from a known model with different random error components and a known statistical distribution. After being theoretically proven, this approach was applied on two real field gravity anomalies from Cuba and Sweden. The agreement between the results obtained by the proposed method and those obtained by other interpretation methods is good and comparable. Moreover, the depth obtained by the proposed approach is found to be in very good agreement with that obtained from drilling information.