Seismic signal enhancement based on the low-rank methods

Based on the fact that the Hankel matrix constructed by noise-free seismic data is low-rank, low-rank approximation (or rank-reduction) methods have been widely used for removing noise from seismic data. Due to the linear-event assumption of the traditional low-rank approximation method, it is difficult to define a rank that optimally separates the data subspace into signal and noise subspaces. For preserving the most useful signal energy, a relatively large rank threshold is often chosen, which inevitably leaves residual noise. To reduce the energy of residual noise, we propose an optimally damped rank-reduction method. The optimal damping is applied via two steps. In the first step, a set of optimal damping weights is derived. In the second step, we derive an optimal singular value damping operator. We review several traditional low-rank methods and compare their performance with the new one. We also compare these low-rank methods with two sparsity-promoting transform methods. Examples demonstrate that the proposed optimally damped rank-reduction method could get significantly cleaner denoised images compared with the state-of-the-art methods.     

Random noise attenuation via the randomized canonical polyadic decomposition

Tensor algebra provides a robust framework for multi-dimensional seismic data processing. A low-rank tensor can represent a noise-free seismic data volume. Additive random noise will increase the rank of the tensor. Hence, tensor rank-reduction techniques can be used to filter random noise. Our filtering method adopts the Candecomp/Parafac decomposition to approximates a N-dimensional seismic data volume via the superposition of rank-one tensors. Similar to the singular value decomposition for matrices, a low-rank Candecomp/Parafac decomposition can capture the signal and exclude random noise in situations where a low-rank tensor can represent the ideal noise-free seismic volume. The alternating least squares method is adopted to compute the Candecomp/Parafac decomposition with a provided target rank. This method involves solving a series of highly over-determined linear least-squares subproblems. To improve the efficiency of the alternating least squares algorithm, we uniformly randomly sample equations of the linear least-squares subproblems to reduce the size of the problem significantly. The computational overhead is further reduced by avoiding unfolding and folding large dense tensors. We investigate the applicability of the randomized Candecomp/Parafac decomposition for incoherent noise attenuation via experiments conducted on a synthetic dataset and field data seismic volumes. We also compare the proposed algorithm (randomized Candecomp/Parafac decomposition) against multi-dimensional singular spectrum analysis and classical prediction filtering. We conclude the proposed approach can achieve slightly better denoising performance in terms of signal-to-noise ratio enhancement than traditional methods, but with a less computational cost.     

基于正交秩-1矩阵追踪的天然地震数据重建研究:以加州San Jacinto断层密集地震台阵为例

由于受地理环境和采集成本等因素的影响,采集到的天然地震数据往往呈现不规则和不完整分布,将直接影响到后续的天然地震数据处理效果,因此需要对缺失数据进行重建.本文将一种基于降秩补全理论的正交秩-1矩阵追踪算法(Orthogonal Rank-One Matrix Pursuit,OR1MP)应用于加州San Jacinto断层带的天然地震数据重建.首先将空间数据的每个频率切片进行Hankel预变换,获取具有低秩结构特征的预变换矩阵,缺失地震道和随机噪声会增加数据预变换矩阵的秩,然后运用OR1MP算法进行降秩处理,最后做反Hankel变换,得到频域上的重建数据.OR1MP算法对2D和3D的加州San Jacinto断层带的天然地震数据实验结果表明,OR1MP算法能够有效地增加地震体的峰值信噪比,能较好地实现对天然地震信号的重建.     

基于SW统计量的自适应时频峰值滤波压制地震勘探随机噪声研究

由于金属矿区地震记录中随机噪声性质复杂且信噪比低,常规降噪方法难以达到预期的滤波效果.时频峰值滤波(TFPF)方法是实现低信噪比地震勘探记录中随机噪声压制的有效方法,但其在复杂地震勘探随机噪声下时窗参数优化问题仍难以解决.本文充分利用地震勘探噪声的统计特性,结合Shapiro-Wilk(SW)统计量辨识地震勘探记录中的微弱有效信号,提出基于SW统计量的自适应时频峰值滤波降噪方法(S-TFPF).在S-TFPF方案中,对于有效信号集中区,S-TFPF方法根据信号频率特征,选择有利于信号保持的较短时窗长度;对于噪声集中区,按噪声方差自适应增加时窗长度,增强随机噪声压制能力.S-TFPF应用于合成记录和共炮点记录的滤波结果表明,与传统时频峰值滤波方法相比,S-TFPF方法可以有效抑制低信噪比地震勘探记录中的随机噪声,更好地恢复出同相轴.     

Mixed Cadzow filtering method in fractional Fourier domain

Conventional frequency domain singular value decomposition (SVD) filtering method used in random noise attenuation processing causes bending event damage. To mitigate this problem, we present a mixed Cadzow filtering method based on fractional Fourier transform to suppress random noise in 3D seismic data. First, the seismic data is transformed to the time-frequency plane via the fractional Fourier transform. Second, based on the Eigenimage filtering method and Cadzow filtering method, the mixed high-dimensional Hankel matrix is built; then, SVD is performed. Finally, random noise is eliminated effectively by reducing the rank of the matrix. The theoretical model and real applications of the mixed filtering method in a region of Sichuan show that our method can not only suppress noise effectively but also preserve the frequency and phase of effective signals quite well and significantly improve the signal-to-noise ratio of 3D post-stack seismic data.     

Noise suppression for microseismic data by non‐subsampled shearlet transform based on singular value decomposition

The existence of strong random noise in surface microseismic data may decrease the utility of these data. Non‐subsampled shearlet transform can effectively suppress noise by properly setting a threshold to the non‐subsampled shearlet transform coefficients. However, when the signal‐to‐noise ratio of data is low, the coefficients related to the noise are very close to the coefficients associated with signals in the non‐subsampled shearlet transform domain that the coefficients related to the noise will be retained and be treated as signals. Therefore, we need to minimise the overlapping coefficients before thresholding. In this paper, a singular value decomposition algorithm is introduced to the non‐subsampled shearlet transform coefficients, and low‐rank approximation reconstructs each non‐subsampled shearlet transform coefficient matrix in the singular value decomposition domain. The non‐subsampled shearlet transform coefficients of signals have bigger singular values than those of the random noise, which implies that the non‐subsampled shearlet transform coefficients can be well estimated by taking only a few largest singular values. Therefore, those properties of singular value decomposition may significantly help minimise overlapping of noise and signals coefficients in the non‐subsampled shearlet transform domain. Finally, the denoised microseismic data are obtained easily by giving a simple threshold to the reconstructed coefficient matrix. The performance of the proposed method is evaluated on both synthetic and field microseismic data. The experimental results illustrate that the proposed method can eliminate random noise and preserve signals of interest more effectively.     

De-aliased and de-noise Cadzow filtering for seismic data reconstruction

Cadzow filtering is currently considered as one of the most effective approaches for seismic data reconstruction. The basic version of Cadzow filtering first reorders each frequency slice of the seismic data (to be reconstructed) to a block Hankel/Toeplitz matrix, and then implements a rank-reduction operator, that is truncated singular value decomposition, to the Hankel/Toeplitz matrix. However, basic Cadzow filtering cannot deal with the problem of recovering regularly missing data (up-sampling) in the case of strongly aliased energy, because the regularly missing data will mix with signals in the singular spectrum. To solve this problem, it has been proposed to precondition the reconstruction of high-frequency components using information from the low-frequency components which are less aliased. In this paper, we further extend the de-aliased Cadzow filtering approach to reconstruct regularly sampled seismic traces from the noisy observed data by modifying the reinserting operation, in which the high-frequency components are projected onto the sub-space spanned by several singular vectors of the low-frequency components. At each iteration, the filtered data are weighted to the original data as a feedback. The weighting factor is related to the background noise level and changes with iteration. The feasibility of the proposed technique is validated via two-dimensional, three-dimensional and five-dimensional synthetic data examples, as well as two-dimensional post-stack and three-dimensional pre-stack field data examples. The results demonstrate that the proposed technique can effectively interpolate regularly sampled data and is robust in noisy environments.     

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。     

Preconditioned prestack plane-wave least squares reverse time migration with singular spectrum constraint

Least squares migration can eliminate the artifacts introduced by the direct imaging of irregular seismic data but is computationally costly and of slow convergence. In order to suppress the migration noise, we propose the preconditioned prestack plane-wave least squares reverse time migration (PLSRTM) method with singular spectrum constraint. Singular spectrum analysis (SSA) is used in the preconditioning of the take-offangle-domain common-image gathers (TADCIGs). In addition, we adopt randomized singular value decomposition (RSVD) to calculate the singular values. RSVD reduces the computational cost of SSA by replacing the singular value decomposition (SVD) of one large matrix with the SVD of two small matrices. We incorporate a regularization term into the preconditioned PLSRTM method that penalizes misfits between the migration images from the plane waves with adjacent angles to reduce the migration noise because the stacking of the migration results cannot effectively suppress the migration noise when the migration velocity contains errors. The regularization imposes smoothness constraints on the TADCIGs that favor differential semblance optimization constraints. Numerical analysis of synthetic data using the Marmousi model suggests that the proposed method can efficiently suppress the artifacts introduced by plane-wave gathers or irregular seismic data and improve the imaging quality of PLSRTM. Furthermore, it produces better images with less noise and more continuous structures even for inaccurate migration velocities.     

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.     

Coherence algorithm with a high-resolution time–time transform and feature matrix for seismic data

Traditional coherence algorithms are often based on the assumption that seismic traces are stationary and Gaussian. However, seismic traces are actually non-stationary and non-Gaussian. A constant time window and the canonical correlation analysis in traditional coherence algorithms are not optimal for non-stationary seismic traces and cannot describe the similarity between adjacent seismic traces in detail. To overcome this problem, a new coherence algorithm using the high-resolution time–time transform and the feature matrix is designed. The high-resolution time–time transform used to replace the constant time window can produce a frequency-dependent time local series to analyse non-stationary seismic traces. The feature matrix, constructed by the frequency-dependent time local series and the related local gradients, defines a new correlation metric that enhances more details of the geological discontinuities in seismic images than does the canonical correlation analysis. Additionally, the Riemannian metric is introduced for related calculations because the feature matrices are not defined in a Euclidean space but rather in a manifold space. Application to field data illustrates that the proposed method reveals more details of structural and stratigraphic features.     

基于鬼波衰减与非平稳多阶差分地震拓频技术的研究与应用

高分辨率地震资料可以达到更精细的井震标定结果,能更清晰地进行构造解释与储层刻画,同时对薄层具有更好的识别能力。为提高地震分辨率,需要对地震数据进行拓频处理。常规地震拓频方法通常在频率域进行,易受到高频噪音影响,降低资料可靠性。本文提出一种基于鬼波处理与多阶差分结合的时域拓频技术,仅通过原始地震道数次积分与差分运算便能拓宽地震频带,积分及差分结果在高斯窗口内进行振幅匹配以保证处理前后振幅一致性。与差分结果进行加权融合,高阶差分设置较小权值以避免高频噪声的影响,提高算法的抗噪性。本理论模型及实际地震资料处理分析结果表明,该方法能有效提高地震资料分辨率。      

Applications of Savitzky-Golay Filter for Seismic Random Noise Reduction

This article utilizes Savitzky–Golay (SG) filter to eliminate seismic random noise. This is a novel method for seismic random noise reduction in which SG filter adopts piecewise weighted polynomial via leastsquares estimation. Therefore, effective smoothing is achieved in extracting the original signal from noise environment while retaining the shape of the signal as close as possible to the original one. Although there are lots of classical methods such as Wiener filtering and wavelet denoising applied to eliminate seismic random noise, the SG filter outperforms them in approximating the true signal. SG filter will obtain a good tradeoff in waveform smoothing and valid signal preservation under suitable conditions. These are the appropriate window size and the polynomial degree. Through examples from synthetic seismic signals and field seismic data, we demonstrate the good performance of SG filter by comparing it with the Wiener filtering and wavelet denoising methods.     

Inversion-based data-driven time-space domain random noise attenuation method

Conventional time-space domain and frequency-space domain prediction filtering methods assume that seismic data consists of two parts, signal and random noise. That is, the so-called additive noise model. However, when estimating random noise, it is assumed that random noise can be predicted from the seismic data by convolving with a prediction error filter. That is, the source-noise model. Model inconsistencies, before and after denoising, compromise the noise attenuation and signal-preservation performances of prediction filtering methods. Therefore, this study presents an inversion-based time-space domain random noise attenuation method to overcome the model inconsistencies. In this method, a prediction error filter (PEF), is first estimated from seismic data; the filter characterizes the predictability of the seismic data and adaptively describes the seismic data’s space structure. After calculating PEF, it can be applied as a regularized constraint in the inversion process for seismic signal from noisy data. Unlike conventional random noise attenuation methods, the proposed method solves a seismic data inversion problem using regularization constraint; this overcomes the model inconsistency of the prediction filtering method. The proposed method was tested on both synthetic and real seismic data, and results from the prediction filtering method and the proposed method are compared. The testing demonstrated that the proposed method suppresses noise effectively and provides better signal-preservation performance.     

Random noise suppression for seismic data using a non-local Bayes algorithm

For random noise suppression of seismic data, we present a non-local Bayes (NLBayes) filtering algorithm. The NL-Bayes algorithm uses the Gaussian model instead of the weighted average of all similar patches in the NL-means algorithm to reduce the fuzzy of structural details, thereby improving the denoising performance. In the denoising process of seismic data, the size and the number of patches in the Gaussian model are adaptively calculated according to the standard deviation of noise. The NL-Bayes algorithm requires two iterations to complete seismic data denoising, but the second iteration makes use of denoised seismic data from the first iteration to calculate the better mean and covariance of the patch Gaussian model for improving the similarity of patches and achieving the purpose of denoising. Tests with synthetic and real data sets demonstrate that the NL-Bayes algorithm can effectively improve the SNR and preserve the fidelity of seismic data.     

Two‐dimensional dictionary learning for suppressing random seismic noise

Dictionary learning is a successful method for random seismic noise attenuation that has been proven by some scholars. Dictionary learning–based techniques aim to learn a set of common bases called dictionaries from given noised seismic data. Then, the denoising process will be performed by assuming a sparse representation on each small local patch of the seismic data over the learned dictionary. The local patches that are extracted from the seismic section are essentially two‐dimensional matrices. However, for the sake of simplicity, almost all of the existing dictionary learning methods just convert each two‐dimensional patch into a one‐dimensional vector. In doing this, the geometric structure information of the raw data will be revealed, leading to low capability in the reconstruction of seismic structures, such as faults and dip events. In this paper, we propose a two‐dimensional dictionary learning method for the seismic denoising problem. Unlike other dictionary learning–based methods, the proposed method represents the two‐dimensional patches directly to avoid the conversion process, and thus reserves the important structure information for a better reconstruction. Our method first learns a two‐dimensional dictionary from the noisy seismic patches. Then, we use the two‐dimensional dictionary to sparsely represent all of the noisy two‐dimensional patches to obtain clean patches. Finally, the clean patches are patched back to generate a denoised seismic section. The proposed method is compared with the other three denoising methods, including FX‐decon, curvelet and one‐dimensional learning method. The results demonstrate that our method has better denoising performance in terms of signal‐to‐noise ratio, fault and amplitude preservation.     

A multi-data training method for a deep neural network to improve the separation effect of simultaneous-source data

Within the field of seismic data acquisition with active sources, the technique of acquiring simultaneous data, also known as blended data, offers operational advantages. The preferred processing of blended data starts with a step of deblending, that is separation of the data acquired by the different sources, to produce data that mimic data from a conventional seismic acquisition and can be effectively processed by standard methods. Recently, deep learning methods based on the deep neural network have been applied to the deblending task with promising results, in particular using an iterative approach. We propose an enhancement to deblending with an iterative deep neural network, whereby we modify the training stage of the deep neural network in order to achieve better performance through the iterations. We refer to the method that only uses the blended data as the input data as the general training method. Our new multi-data training method allows the deep neural network to be trained by the data set with the input patches composed of blended data, noisy data with low amplitude crosstalk noise, and unblended data, which can improve the ability of the deep neural network to remove crosstalk noise and protect weak signal. Based on such an extended training data set, the multi-data training method embedded in the iterative separation framework can result in different outputs at different iterations and converge to the best result in a shorter iteration number. Transfer learning can further improve the generalization and separation efficacy of our proposed method to deblend the simultaneous-source data. Our proposed method is tested on two synthetic data and two field data to prove the effectiveness and superiority in the deblending of the simultaneous-source data compared with the general training method, generic noise attenuation network and low-rank matrix factorization methods.     

基于数据增广和CNN的地震随机噪声压制

卷积神经网络(Convolutional Neural Network,CNN)是一种基于数据驱动的学习算法,简化了传统从特征提取到分类的两阶段式处理任务,被广泛应用于计算机科学的各个领域.在标注数据不足的地震数据去噪领域,CNN的推广应用受到限制.针对这一问题,本文提出了一种基于数据生成和增广的地震数据CNN去噪框架.对于合成数据,本文对无噪地震数据添加不同方差的高斯噪声,增广后构成训练集,实现基于小样本的CNN训练.对于实际地震数据,由于无法获得真实的干净数据和噪声来生成训练样本集,本文提出一种直接从无标签实际有噪数据生成标签数据集的方法.在所提出的方法中,我们利用目前已有的去噪方法从实际地震数据中分别获得估计干净数据和估计噪声,前者与未知的干净数据具有相似纹理,后者与实际噪声具有相似的概率分布.人工合成数据和实际数据实验结果表明,相较于F-X反褶积,BM3D和自适应频域滤波算法,本文方法能更好地压制随机噪声和保护有效信号.最后,本文采用神经网络可视化方法对去噪CNN的机理进行了探索,一定程度上解释了网络每一层的学习内容.     

基于有效邻域波场近似的起伏地表保幅高斯束偏移

随着我国陆上地震勘探向复杂地表探区的转移,高精度、适应性强的地震成像方法在地震资料的处理、解释及后续属性分析、储层预测中具有重要意义.本文基于有效邻域波场近似理论发展了一种成像精度更高且适用于复杂起伏地表条件的叠前保幅高斯束偏移方法.在传统水平地表高斯束偏移的基础上,本文根据中心射线附近有效邻域内高斯束表征的近似波场,导出了起伏地表条件下具有相对振幅保持的高斯束偏移公式,并给出了一种精度更高的旁轴射线传播角度计算方法.同现有的高斯束偏移方法相比,本文方法不仅考虑了起伏地表对高斯束走时的线性影响,而且首次引入了由地表高程差异和近地表速度变化引起的二次时差校正项和振幅校正项,使得成像结果更加准确可靠.两个典型模型算例验证了本文方法的正确性和有效性.     

基于分数阶小波域GSM模型的地震信号随机噪声压制方法

地震信号中的随机噪声是一种干扰波,严重降低了地震信号的信噪比,并影响着资料的后续处理和分析.本文根据地震信号中有效信号和随机噪声的差异,结合分数阶B样条小波变换与高斯尺度混合模型提出了一种地震信号随机噪声压制方法.首先利用分数阶B样条小波变换将含噪地震信号映射到最优分数阶小波时频域内,然后对各小波子带系数分别建立高斯尺度混合模型,由贝叶斯方法估计出源地震信号小波系数,最后使用分数阶B样条小波逆变换重构得到降噪后的地震信号.利用本文方法对合成地震记录和实际地震信号进行降噪处理,实验结果表明本文方法能够有效地压制地震信号中的随机噪声,并且较好地保留了有效信号.