Seismic data are frequently contaminated by incoherent random noise and coherent surface waves; both severely degrade subsurface data quality. While random noise is statistically irregular and uncorrelated, surface waves exhibit structured regularity in both temporal and spatial domains. Their coherent low-frequency propagation makes them fundamentally distinct from random noise, posing significant challenges to simultaneous suppression. This letter presents a deep adaptive signal denoising network (DASDNet) designed for random and ground-roll simultaneous suppression using weakly supervised training on synthetic clean-noisy pairs. The architecture incorporates short-time Fourier transform (STFT) and discrete wavelet transform (DWT) for multiscale time-frequency decomposition. A fuzzy-logic-based encoder is used to mitigate feature uncertainty, complemented by a time-frequency regularization module that preserves structural coherence. A unified loss function across time, frequency, and spatial domains guides robust training. DASDNet demonstrates remarkable generalization capability to both synthetic and field-recorded datasets, achieving robust performance without reliance on labeled field data or manual post hoc parameter tuning. Experimental results show that DASDNet improves signal-to-noise ratio (SNR) by 6.86 dB, reduces mean squared error (mse) by 4433, and increases structural similarity index (SSIM) by 0.22, outperforming other methods. Qualitative analysis further demonstrates its advantage in preserving primary fidelity while suppressing noises.

Ground roll is a type of coherent noise with low frequency, low velocity, and high amplitude, which masks useful signals and decreases the quality of subsequent seismic data processing. It is a challenge for traditional signal processing methods to separate useful signals effectively when the ground roll and useful reflected signals overlap seriously in the low-frequency band. We develop a supervised-learning-based framework with soft attention residual learning mechanisms for suppressing the ground roll noise. To reduce the cost of manual labeling, the 2D patching technique is used to segment large-scale seismic data into a large number of small-scale patches for training. Our network includes a multibranch attention block that uses multiple branches with different kernel sizes to extract waveform features at different scales from input noisy patches. Then, we use the soft attention mechanism to select and fuse the feature maps of different branches. Our network can achieve encouraging ground roll attenuation performance by using a small number of training samples, which is demonstrated by synthetic and field data examples. Compared with one traditional method and two advanced deep-learning frameworks, our network has better abilities in preserving low-frequency useful signals and removing ground roll.

Salt body segmentation is a critical part of structural interpretation and oil and gas exploration for subsalt reservoirs. Existing automatic salt body segmentation techniques mostly use supervised learning strategies. It is challenging to generate a large number of labels by manual labeling, especially for 3-D salt bodies. Here, we propose a self-supervised learning (SSL) framework called Salt3DNet, for 3-D salt body segmentation. This framework is divided into two stages: pretraining and fine-tuning of downstream tasks. In the pretraining stage, we use the Barlow twins (BTs) method to pretrain the encoder and reduce redundancy in a contrastive learning manner to learn high-level data representations. In the fine-tuning stage, we construct two encoders to reconstruct 3-D seismic data and segment salt bodies in a multitask collaborative learning way. The encoder and decoder are composed of the 3-D fully convolutional DenseNet and soft attention mechanism, where the latter represents the selective kernel block (SKB) with multiple kernels of different sizes. Salt3DNet calculates the correlation matrix of features from different perspectives in the pretraining stage and makes it close to the identity matrix to obtain a more prosperous feature representation. Then, Salt3DNet uses a limited number of labeled samples for training. According to the evaluation metrics, the proposed network has demonstrated promising salt segmentation performance in 3-D SEG advanced modeling (SEAM) synthetic data and $F3$ block real seismic data. In addition, the proposed network is demonstrated to have higher prediction accuracy than state-of-the-art salt segmentation frameworks through ablation experiments.

Quality factor Q is an important parameter that accounts for the amplitude dissipation and phase distortion of seismic waves propagating in the Earth's interior. Q estimation with improved accuracy benefits nonstationary seismic inversion, seismic imaging, fluid identification, and so on. Usually, logarithmic spectral ratio (LSR) is widely used to estimate Q based on vertical seismic profile (VSP) and poststack data. However, LSR is very sensitive to noise, and the effect of normal moveout (NMO) distorts the spectrum of the stacked seismic data, leading to an inferior Q estimation result. To weaken the effect of NMO and enhance the accuracy of Q estimation, we expand an improved LSR method in the zero-offset traveltime-local slope (e.g., t(0 )- p) domain and propose a robust prestack Q estimation method based on common midpoint (CMP) gathers. The proposed method incorporates warped mapping (WM) and shaping regularization to stabilize it during Q estimation in the case of low signal-to-noise ratio (SNR). Additionally, we incorporate nonzero-offset information for Q estimation, which weakens the strong dependence on zero-offset information during prestack Q estimation. Compared with the single-gather prestack Q estimation methods (SPQEM), we make the most of the spatial coherence between the adjacent CMP to eliminate the unexpected noise-related outliers during spectral division for improved robustness and accuracy. Numerical examples are used to validate the superior performance of the proposed method, even in the presence of strong ambient noise.

We have introduced a new PP and PS joint inversion method based on the analytical solution of the 1D elastic wave equation (ASEW). The ASEW can simulate the phase changes of PP and PS waves, so it has the ability to simplify or avoid the PP and PS registration and optimize the inversion process. In addition, compared with the conventional Zoeppritz-based methods, the ASEW can not only simulate the primary reflections, but also the wave propagation effects, such as transmission loss, internal multiples, and converted waves, etc. These propagation effects are difficult to estimate and eliminate and treated as primary reflections in Zoeppritz-based methods, which can degrade the inversion results, and reduce the accuracy of reservoir characterization. Synthetic and field data examples demonstrate the effectiveness and feasibility of the proposed PP and PS joint inversion method.

As an important elastic parameter, seismic acoustic impedance is usually obtained through poststack inversion. However, there are usually two problems that limit the quality of the inversion results. First, conventional inversion methods typically use regularization terms to enhance the stability of the inversion results, and effective regularization terms are particularly important for accurately inverting seismic impedance. Second, most inversion methods adopt a trace-by-trace inversion strategy, resulting in poor lateral continuity when connecting the inversion results of all traces into a 2-D profile, especially for processing noisy data. To address these two problems, we propose a structure-oriented minimum entropy stabilizer for acoustic impedance inversion that enhances the lateral continuity of the inversion results while restoring the blocky structures of the strata and improving the resolution of the inversion results. The stabilizer consists of a structure-oriented regularization (SOR) operator and the minimum entropy norm. The SOR operator is constructed using the local dip estimated from the seismic data by the plane-wave destruction (PWD) algorithm and constrains the inverted impedance along the structural trend, making it more consistent with geological features. The minimum entropy norm restores the blocky structures and enhances resolution by imposing sparse constraints on the temporal and spatial derivatives of the impedance. Based on synthetic and field seismic data, we compare the inversion results of the proposed method with those of conventional Tikhonov regularization and total variation (TV) regularization methods. The results show that the proposed method exhibits superior performance, especially in processing noisy data.

As one of the important research topics in the field of natural language processing, sentiment analysis aims to analyze web data related to COVID-19, e.g., supporting China government agencies combating COVID-19. There are popular sentiment analysis models based on deep learning techniques, but their performance is limited by the size and distribution of the dataset. In this study, we propose a model based on a federal learning framework with Bert and multi-scale convolutional neural network (Fed_BERT_MSCNN), which contains a Bidirectional Encoder Representations from Transformer modules and a multi-scale convolution layer. The federal learning framework contains a central server and local deep learning machines that train local datasets. Parameter communications were processed through edge networks. The weighted average of each participant’s model parameters was communicated in the edge network for final utilization. The proposed federal network not only solves the problem of insufficient data, but also ensures the data privacy of the social platform during the training process and improve the communication efficiency. In the experiment, we used datasets of six social platforms, and used accuracy and F1-score as evaluation criteria to conduct comparative studies. The performance of the proposed Fed_BERT_MSCNN model was generally superior than the existing models in the literature.

Distributed acoustic sensing (DAS) technology has been widely used in seismic exploration to acquire high-quality data due to its noteworthy advantages, such as high coverage, high resolution, low cost, and strong environmental friendliness. However, the seismic signals acquired in DAS are often masked by various types of noise (e.g., high-frequency random, high-amplitude erratic, horizontal, and coupled noise), which seriously decreases the signal-to-noise ratio. We develop a fully connected neural network with dense and residual connections to attenuate various complex noises in real DAS data. The network is designed to learn the features of useful reflection signals and remove various noises in an unsupervised way, therefore enjoying the convenience of label-free processing. Our network uses several encoders and decoders to compress and reconstruct the abstract waveform features, respectively. Each encoder/decoder consists of one dense block with stacked fully connected blocks (FCBs). To transfer the shallow-level features to the deep level for reuse, we add the skip connections with one FCB between the corresponding encoders and decoders. Our method provides encouraging results when applied to synthetic and real DAS data sets. Compared with several traditional and advanced deep-learning methods, our method can more effectively attenuate strong noise and better extract hidden signals.

Time-lapse seismic exploration, as an effective means of reservoir dynamic monitoring, plays a vital role in obtaining subsurface media variation. Misleading travel -time differences will occur due to differences in environmental conditions, acquisition methods, and processing techniques between multiple periods of seismic data. To eliminate the travel -time discrepancy, we introduce the Dynamic Time Warping (DTW) method to realize the optimal matching of seismic angle gathers. Based on the effective variation information of different period seismic records, we propose time-lapse seismic difference inversion by elastic wave equation via scattering theory to yield the difference in elastic parameters of the formation and realize quantitative characterization of reservoir alteration. Our method adopts WKBJ approximates the inhomogeneous background, which has higher inversion accuracy, especially under complex geological conditions. Here, we carried out numerical simulation experiments and practical data applications on the proposed method, which proved that the proposed method allows for higher resolution inversion results and more precision characterization of stratigraphic changes than traditional methods.

To perform joint PP/PS amplitude-variation-with-angle inversion or attribute analysis, we usually map converted-wave (PS) data to the compressional wave (PP) time domain to guarantee that the reflection events from similar reflectors have the same two-way traveltimes. Mapping PS data to the PP time domain leads to a nonstationary characteristic of the PS data. Nonstationary here means varying spectral content of the wavelet in PS data, which will degrade inversion results. We develop a zero-phase time-frequency adaptive shaping filter to filter the nonstationary PS data to be stationary. This adaptive filter can correct the distorted wavelets in the PS data without introducing numerical noise and artifacts. In addition, we use the dynamic time warping algorithm to estimate the time shifts between the PP and PS data and the local seismic attribute estimation method to estimate the subsurface [Formula: see text] ratios from the time shifts. The estimated [Formula: see text] ratios will be used to calculate the time- and location-varying wavelets by the Fourier scaling theorem, and then the adaptive shaping filter is constructed. Based on the synthetic and field data examples, we determine the better performance of the adaptive shaping filter over the conventional shaping filter in attenuating numerical noise and artifacts.

Accurate estimation of reservoir parameters (e.g., permeability and porosity) helps to understand the movement of underground fluids. However, reservoir parameters are usually expensive and time-consuming to obtain through petrophysical experiments of core samples, which makes a fast and reliable prediction method highly demanded. In this article, we propose a deep learning model that combines the 1-D convo- lutional layer and the bidirectional long short-term memory network to predict reservoir permeability and porosity. The mapping relationship between logging data and reservoir parameters is established by training a network with a combination of nonlinear and linear modules. Optimization algorithms, such as layer normalization, recurrent dropout, and early stopping, can help obtain a more accurate training model. Besides, the self-attention mechanism enables the network to better allocate weights to improve the prediction accuracy. The testing results of the well-trained network in blind wells of three different regions show that our proposed method is accurate and robust in the reservoir parameters prediction task.

Distributed acoustic sensing (DAS) is a novel and fast-developing seismic acquisition technology, which enjoys many advantages compared with traditional geophones. However, DAS data often suffer from severe and diverse types of noise with varying amplitudes, resulting in a low signal-to-noise ratio (S/N) and making the extraction of hidden signals a challenging task. Therefore, exploring a high-efficiency and high-generalization denoising method is crucial for improving the S/N of DAS data and subsequent processing. We develop a dense connection network with the kernel-wise attention mechanism to denoise complex and diverse noise (e.g., high-amplitude erratic, high-frequency, random, and horizontal noise) on real DAS data sets. We use an integrated denoising framework that is suitable for attenuating DAS noise to generate labels for network training. Our network consists of five types of blocks, i.e., convolutional, dense, transition down, transition up, and selective kernel blocks (SKB). In particular, the SKB is used to fuse multiscale features by weighting, thereby improving denoising accuracy. The computational efficiency and denoising performance are further augmented by using a patching method to segment the DAS data and generate many small-scale patches. Our network is trained on a small DAS data set and tested on the synthetic and field data from vastly different geographic areas. The comparisons of our network with three state-of-the-art deep-learning-based benchmark models demonstrate more robust performance and superior signal extraction ability.

The high resolution Radon transform is one of the commonly used methods in seismic data processing. Its inversion usually involves matrix inversion, multiple iterations, hyperparameter selection et al. These factors lead to problems such as a large amount of calculation and a slow convergence rate in the inversion of the Radon transform inversion. Based on the analysis of the low resolution of Radon transform, we propose a high resolution Radon transform inversion method based on one-dimensional Convolutional Neural Network (1-D CNN). This method realizes the mapping from conjugate Radon solution to high resolution by the nonlinear representation ability of CNN. And the principle of improving resolution of series mapping model based on deconvolution principle and parallel mapping model based on residual learning are analyzed. The specific frequency Radon parameter obtained by the above CNN network is restricted to the inversion of other frequency parameters, which avoids the drawbacks of frequency division training. Multiple suppression experiments on synthetic and field data show that the proposed high-resolution Radon transform based on 1-D CNN can suppress multiples with high computational efficiency.

Stratigraphy in the crust is widely anisotropic. Anisotropic parameters play an important role from inversion and migration to stratigraphic interpretation and reservoir characterization. At present, under conventional geophysical methods, whether logging or seismic, do not directly measure anisotropic parameters. That is, it is difficult to obtain anisotropic parameters. However, there is a certain correlation between anisotropy parameters and other kinds of logging data, so that anisotropy parameters can be calculated from other logging curves. In view of the complexity of this relationship, a machine learning approach can be used. So, we propose a deep multiple triangular kernel extreme learning machine optimized by the flower pollination algorithm (FPA-D-MK-ELM), which is used to predict the anisotropy parameters of the strata. The accuracy and stability of the FPA-D-MK-ELM algorithm are verified by comparing the algorithm before and after optimization.

Prestack seismic inversion has emerged as a powerful technique for reconstructing parameters attribute to the subsurface properties and building the geophysical parameter models. However, the inversion algorithms always suffer from spatial blur and low resolution. Total variation (TV) regularization preserves the spatial variation boundary of data by highlighting the sparsity of the first-order difference, which is regarded as an important technical means for image restoration. However, when the data do not change along the spatial grid direction, TV regularization is prone to a staircase effect. In this article, a directional TV (DTV) method is proposed to conduct the prestack amplitude variation with offset/angle (AVO/AVA) inversion. The method consists of three essential steps: estimating the seismic slope attribute from the seismic data, introducing seismic slope attribute to the TV regularization to establish the objective function, and optimizing the objective function by the split-Bregman algorithm. Finally, the conventional and proposed methods are applied to the synthetic and the real seismic data. The comparison of different methods demonstrates that the proposed method is applicable to reveal the detailed subsurface models, alleviate the staircase effect or artifact substantially, and further upgrade the quality of prestack inversion results.

For the large-scale dynamic data stream, incremental learning is an effective and efficient technique and is widely used in machine learning. Incremental dimensionality reduction algorithms have been proposed by many scholars. As an improved canonical correlation analysis (CCA) method based on incremental learning, incremental canonical correlation analysis (ICCA) can effectively deal with the problem of dimensionality reduction of high-dimensional multi-view data stream. However, there is a drawback in this approach that the projection vector must be updated once for each new sample, which consumes a lot of time on the issue of online learning. Aiming at this problem, chunk incremental canonical correlation analysis (CICCA) is proposed in this paper. It can avoid the calculation of sample covariance matrices and process batch data stream directly. The main projection vector is updated each time with the newly added batch sample information, which is used to revise and update the projection vector of the previous step. Further, the other projection vectors are calculated in the orthogonal complement space of the projection vector. Therefore, data can be got from low-dimensional spaces. Experimental results show that the classification performance of CICCA is comparable to CCA and ICCA, but the training time is greatly reduced on synthetic dataset and real dataset.

The probabilistic prediction approach can be used not only for obtaining the maximum posterior probability solution but also for uncertainty evaluation. Its prior distribution has a significant impact on the prediction result. An improper prior assumption may lead to prediction deviation. To improve the prediction accuracy of elastic parameters, a Laplace prior with total variation (TV) constraint is introduced in the probabilistic prediction. First, the effect of TV constraint on the probability distribution of elastic parameters is analyzed in detail. Then, two approaches are proposed to handle the cases where the elastic parameters have blocky boundaries and no blocky boundaries: probabilistic prediction scheme for elastic parameters with blocky boundaries and probabilistic prediction scheme with blocky lithology prior constraint. The former imposes a sparse constraint on the elastic parameters, while the latter imposes a sparse constraint on the TV processing lithology. Their posterior probabilities are rederived. Considering that the discrete lithology is more likely to be blocky compared with the continuous elastic parameters, the sparse lithology constraint can handle more general cases. In addition, this approach allows for lithology prediction. The applications of numerical examples and field seismic data verify the feasibility of the proposed approaches.

Instantaneous frequency is an important seismic attribute, which can indicate thin beds and lithofacies boundaries. However, instantaneous frequency attribute is susceptible to noise when it is obtained by the traditional Hilbert transform (HT) method. We propose a robust method for instantaneous frequency estimation. The method first obtains the time-frequency distribution of the seismic signal by inverse spectral decomposition (ISD) and then calculates the analytic signal through the window HT transform. Inverse spectral decomposition achieves high-resolution time-frequency distribution by adding sparse constraint to the corresponding inverse problem, and therefore the noise can be suppressed. The algorithm we choose to solve the mix l(2) -l(1) problem is the fast iterative shrinkage-thresholding algorithm (FISTA). Compared with the traditional iterative least squares (IRLS) algorithm, FISTA can achieve a better computational efficiency. We perform the method on a quadratic frequency modulation (QFM) signal, a synthetic data based on a wedge model and field data sets to demonstrate its performance, compared with the HT method and the time-frequency adaptive filtering method.

Effective random noise attenuation is critical for subsequent processing of seismic data, such as velocity analysis, migration, and inversion. Thus, the removal of seismic random noise with an uncertainty level is meaningful. Attenuating 3-D random noise in a supervised way based on deep learning (DL) is challenging because clean labels are difficult to obtain. Therefore, it is necessary to develop an adaptive unsupervised-based method for random noise attenuation. In this article, we propose a deep-denoising unsupervised learning (DDUL) network to attenuate random noise in 2-D/3-D seismic data. A patching technique is used to split 2-D/3-D seismic data into several patches to be fed into the network, which helps to expand the number of samples for training. We use the fully symmetrical structure of the autoencoder to construct the network. In each corresponding encoder and decoder layer, skip connections are added to enhance the learning of seismic data features. We construct three blocks to extract waveform features in seismic data, i.e., encoder, decoder, and skip blocks. Among them, the skip is connected between the encoder and decoder blocks of each hidden layer. The use of multiple blocks not only improves the network’s ability to extract seismic data features but also solves the problem of excessive training parameters caused by hidden layer stacking. Five 2-D/3-D synthetic and field seismic datasets are used to test the denoising performance of our proposed method. The denoising results demonstrate that our proposed method has good signal-preserving and noise attenuation capabilities in real-world applications.

ABSTRACT As an advanced real‐time monitoring technique, microseismic source‐location imaging provides valuable information during hydraulic fracturing, for example, the development of fracture networks and the effective reservoir reconstruction volume. However, microseismic data always suffer from weak induced energy and susceptibility to noise interference. In the case of a low signal‐to‐noise ratio, it is extremely challenging to perform robust microseismic imaging. Here, we first introduce several state‐of‐the‐art imaging conditions and two hybrid imaging conditions, which are followed by a detailed analysis of the impact of different grouping strategies. Then, we briefly analyse the sensitivity of different imaging conditions to noise using a one‐dimensional signal. Next, several benchmark models, including two‐dimensional Marmousi‐II and three‐dimensional SEG Advanced Modeling, are used as numerical examples for testing the passive‐source imaging algorithms. Finally, three‐dimensional real microseismic data are used to further investigate the impact of the grouping strategy on the imaging. The numerical examples and field data demonstrate the effectiveness of the proposed grouping strategy for the grouped imaging conditions.