Simultaneous seismic data interpolation and denoising with a new adaptive method based on dreamlet transform

Wang, Benfeng; Wu, Ru‐Shan; Chen, Xiaohong; Li, Jingye

doi:10.1093/gji/ggv072

Cited by 123 publications

(16 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Methods based on time-frequency denoising [20,21] form a large class of seismic denoising techniques. In this approach, noisy time series are first transformed into the timefrequency domain using a time-frequency transform, such as a wavelet transform [22,23,24,25,26,27], Short Time Fourier Transform (STFT) [28], S-transform [29], curvelet transform [30,31,32], dreamlet transform [33], contourlet zhuwq@stanford.edu transform [34], shearlet transform [35], empirical mode decomposition [36,37,38,37,39,40], etc. The resulting timefrequency coefficients are modified (thresholded) to attenuate the coefficients associated with noise and to find an estimate of the signal coefficients.…”

Section: Introductionmentioning

confidence: 99%

Seismic Signal Denoising and Decomposition Using Deep Neural Networks

Zhu

Mousavi

Beroza

2019

IEEE Trans. Geosci. Remote Sensing

282

100

View full text Add to dashboard Cite

Denoising and filtering are widely used in routine seismic-data-processing to improve the signal-to-noise ratio (SNR) of recorded signals and by doing so to improve subsequent analyses. In this paper we develop a new denoising/decomposition method, DeepDenoiser, based on a deep neural network. This network is able to learn simultaneously a sparse representation of data in the time-frequency domain and a non-linear function that maps this representation into masks that decompose input data into a signal of interest and noise (defined as any nonseismic signal). We show that DeepDenoiser achieves impressive denoising of seismic signals even when the signal and noise share a common frequency band. Our method properly handles a variety of colored noise and non-earthquake signals. DeepDenoiser can significantly improve the SNR with minimal changes in the waveform shape of interest, even in presence of high noise levels. We demonstrate the effect of our method on improving earthquake detection. There are clear applications of DeepDenoiser to seismic imaging, micro-seismic monitoring, and preprocessing of ambient noise data. We also note that potential applications of our approach are not limited to these applications or even to earthquake data, and that our approach can be adapted to diverse signals and applications in other settings.

show abstract

Section: Introductionmentioning

confidence: 99%

Seismic Signal Denoising and Decomposition Using Deep Neural Networks

Zhu

Mousavi

Beroza

2019

IEEE Trans. Geosci. Remote Sensing

282

100

View full text Add to dashboard Cite

show abstract

“…; Xue, Ma and Chen ; Wang et al . ; Yu, Ma and Osher ) and so on. Of the numerous methods, however, very few can handle non‐uniformly sampled data.…”

Section: Introductionmentioning

confidence: 98%

Curvelet reconstruction of non‐uniformly sampled seismic data using the linearized Bregman method

Zhang

Diao

Chen

et al. 2019

Geophysical Prospecting

View full text Add to dashboard Cite

A B S T R A C TSeismic data reconstruction, as a preconditioning process, is critical to the performance of subsequent data and imaging processing tasks. Often, seismic data are sparsely and non-uniformly sampled due to limitations of economic costs and field conditions. However, most reconstruction processing algorithms are designed for the ideal case of uniformly sampled data. In this paper, we propose the non-equispaced fast discrete curvelet transform-based three-dimensional reconstruction method that can handle and interpolate non-uniformly sampled data effectively along two spatial coordinates. In the procedure, the three-dimensional seismic data sets are organized in a sequence of two-dimensional time slices along the source-receiver domain. By introducing the two-dimensional non-equispaced fast Fourier transform in the conventional fast discrete curvelet transform, we formulate an L1 sparsity regularized problem to invert for the uniformly sampled curvelet coefficients from the non-uniformly sampled data. In order to improve the inversion algorithm efficiency, we employ the linearized Bregman method to solve the L1-norm minimization problem. Once the uniform curvelet coefficients are obtained, uniformly sampled three-dimensional seismic data can be reconstructed via the conventional inverse curvelet transform. The reconstructed results using both synthetic and real data demonstrate that the proposed method can reconstruct not only non-uniformly sampled and aliased data with missing traces, but also the subset of observed data on a non-uniform grid to a specified uniform grid along two spatial coordinates. Also, the results show that the simple linearized Bregman method is superior to the complex spectral projected gradient for L1 norm method in terms of reconstruction accuracy.

show abstract

“…Sparse transform based approaches first transform seismic data to a sparse domain (Sacchi & Ulrych 1995;Wang et al 2015b;Chen 2017;Huang et al 2017b), then apply soft thresholding to the coefficients, finally transform the sparse coefficients back to the time-space domain. Widely used sparse transforms are Fourier transform (Pratt et al 1998;Naghizadeh & Sacchi 2011;Zhong et al 2014;Wang et al 2015a;Li et al 2016d;Shen et al 2016), curvelet transform (Candès et al 2006;Hermann et al 2007;Neelamani et al 2008;Liu et al 2016e;Zu et al 2016), seislet transform Chen & Fomel 2015a;Gan et al 2015b,c;Liu et al 2015;Gan et al 2016c;Liu et al 2016f), seislet frame transform Gan et al 2015aGan et al , 2016b, shearlet transform (Kong et al 2016;Liu et al 2016a), Radon transform (Sacchi & Ulrych 1995;Xue et al 2014;Sun & Wang 2016;Xue et al 2016bXue et al , 2017, different types of wavelet transforms (e.g.…”

Section: Introductionmentioning

confidence: 99%

Seismic noise attenuation using an online subspace tracking algorithm

Zhou

Zhang

et al. 2020

Geophysical Journal International

View full text Add to dashboard Cite

SUMMARY We propose a new low-rank based noise attenuation method using an efficient algorithm for tracking subspaces from highly corrupted seismic observations. The subspace tracking algorithm requires only basic linear algebraic manipulations. The algorithm is derived by analysing incremental gradient descent on the Grassmannian manifold of subspaces. When the multidimensional seismic data are mapped to a low-rank space, the subspace tracking algorithm can be directly applied to the input low-rank matrix to estimate the useful signals. Since the subspace tracking algorithm is an online algorithm, it is more robust to random noise than traditional truncated singular value decomposition (TSVD) based subspace tracking algorithm. Compared with the state-of-the-art algorithms, the proposed denoising method can obtain better performance. More specifically, the proposed method outperforms the TSVD-based singular spectrum analysis method in causing less residual noise and also in saving half of the computational cost. Several synthetic and field data examples with different levels of complexities demonstrate the effectiveness and robustness of the presented algorithm in rejecting different types of noise including random noise, spiky noise, blending noise, and coherent noise.

show abstract

Simultaneous seismic data interpolation and denoising with a new adaptive method based on dreamlet transform

Cited by 123 publications

References 47 publications

Seismic Signal Denoising and Decomposition Using Deep Neural Networks

Seismic Signal Denoising and Decomposition Using Deep Neural Networks

Curvelet reconstruction of non‐uniformly sampled seismic data using the linearized Bregman method

Seismic noise attenuation using an online subspace tracking algorithm

Contact Info

Product

Resources

About