Deep-learning-based seismic data interpolation: A preliminary result

Geophysics ◽  
2019 ◽  
Vol 84 (1) ◽  
pp. V11-V20 ◽  
Author(s):  
Benfeng Wang ◽  
Ning Zhang ◽  
Wenkai Lu ◽  
Jialin Wang
Keyword(s):  
Deep Learning ◽  
Seismic Data ◽  
Field Data ◽  
Training Data ◽  
Low Rank ◽  
Seismic Survey ◽  
Data Interpolation ◽  
Spatial Direction ◽  
Time Space ◽  
Dense Data

Seismic data interpolation is a longstanding issue. Most current methods are only suitable for randomly missing cases. To deal with regularly missing cases, an antialiasing strategy should be included. However, seismic survey design using a random distribution of shots and receivers is always operationally challenging and impractical. We have used deep-learning-based approaches for seismic data antialiasing interpolation, which could extract deeper features of the training data in a nonlinear way by self-learning. It can also avoid linear events, sparsity, and low-rank assumptions of the traditional interpolation methods. Based on convolutional neural networks, eight-layers residual learning networks (ResNets) with a better back-propagation property for deep layers is designed for interpolation. Detailed training analysis is also performed. A set of simulated data is used to train the designed ResNets. The performance is assessed with several synthetic and field data. Numerical examples indicate that the trained ResNets can help to reconstruct regularly missing traces with high accuracy. The interpolated results in the time-space domain and the frequency-wavenumber ([Formula: see text]-[Formula: see text]) domain demonstrate the validity of the trained ResNets. Even though the accuracy decreases with the increase of the feature difference between the test and training data, the proposed method can still provide reasonable interpolation results. Finally, the trained ResNets is used to reconstruct dense data with halved trace intervals for synthetic and field data. The reconstructed dense data are more continuous along the spatial direction, and the spatial aliasing effects disappear in the [Formula: see text]-[Formula: see text] domain. The reconstructed dense data have the potential to improve the accuracy of subsequent seismic data processing and inversion.

scholarly journals A Natural Images Pre-Trained Deep Learning Method for Seismic Random Noise Attenuation

2022 ◽  
Vol 14 (2) ◽  
pp. 263
Author(s):  
Haixia Zhao ◽  
Tingting Bai ◽  
Zhiqiang Wang
Keyword(s):  
Deep Learning ◽  
Seismic Data ◽  
Field Data ◽  
Noise Suppression ◽  
Signal To Noise Ratio ◽  
Random Noise ◽  
Natural Images ◽  
Training Data ◽  
Learning Approaches ◽  
Learning Method

Seismic field data are usually contaminated by random or complex noise, which seriously affect the quality of seismic data contaminating seismic imaging and seismic interpretation. Improving the signal-to-noise ratio (SNR) of seismic data has always been a key step in seismic data processing. Deep learning approaches have been successfully applied to suppress seismic random noise. The training examples are essential in deep learning methods, especially for the geophysical problems, where the complete training data are not easy to be acquired due to high cost of acquisition. In this work, we propose a natural images pre-trained deep learning method to suppress seismic random noise through insight of the transfer learning. Our network contains pre-trained and post-trained networks: the former is trained by natural images to obtain the preliminary denoising results, while the latter is trained by a small amount of seismic images to fine-tune the denoising effects by semi-supervised learning to enhance the continuity of geological structures. The results of four types of synthetic seismic data and six field data demonstrate that our network has great performance in seismic random noise suppression in terms of both quantitative metrics and intuitive effects.

De-aliased seismic data interpolation using a deep learning-based prediction-error filter

Geophysics ◽  
2021 ◽  
pp. 1-63
Author(s):  
Wenqian Fang ◽  
Lihua Fu ◽  
Shaoyong Liu ◽  
Hongwei Li
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Seismic Data ◽  
Prediction Error ◽  
Training Data ◽  
Training Dataset ◽  
Data Interpolation ◽  
Filter Methods ◽  
The Neural Network ◽  
Trained Network

Deep learning (DL) technology has emerged as a new approach for seismic data interpolation. DL-based methods can automatically learn the mapping between regularly subsampled and complete data from a large training dataset. Subsequently, the trained network can be used to directly interpolate new data. Therefore, compared with traditional methods, DL-based methods reduce the manual workload and render the interpolation process efficient and automatic by avoiding the selection of hyperparameters. However, two limitations of DL-based approaches exist. First, the generalization performance of the neural network is inadequate when processing new data with a different structure compared to the training data. Second, the interpretation of the trained networks is very difficult. To overcome these limitations, we combine the deep neural network and classic prediction-error filter methods, proposing a novel seismic data de-aliased interpolation framework termed PEFNet (Prediction-Error Filters Network). The PEFNet designs convolutional neural networks to learn the relationship between the subsampled data and the prediction-error filters. Thus, the filters estimated by the trained network are used for the recovery of missing traces. The learning of filters enables the network to better extract the local dip of seismic data and has a good generalization ability. In addition, PEFNet has the same interpretability as traditional prediction error-filter based methods. The applicability and the effectiveness of the proposed method are demonstrated here by synthetic and field data examples.

Deep learning-driven velocity model building workflow

2019 ◽  
Vol 38 (11) ◽  
pp. 872a1-872a9 ◽  
Author(s):  
Mauricio Araya-Polo ◽  
Stuart Farris ◽  
Manuel Florez
Keyword(s):  
Deep Learning ◽  
Seismic Data ◽  
Model Building ◽  
Ground Truth ◽  
Velocity Model ◽  
Training Data ◽  
Quality Data ◽  
Generative Adversarial Network ◽  
Data Set ◽  
Velocity Models

Exploration seismic data are heavily manipulated before human interpreters are able to extract meaningful information regarding subsurface structures. This manipulation adds modeling and human biases and is limited by methodological shortcomings. Alternatively, using seismic data directly is becoming possible thanks to deep learning (DL) techniques. A DL-based workflow is introduced that uses analog velocity models and realistic raw seismic waveforms as input and produces subsurface velocity models as output. When insufficient data are used for training, DL algorithms tend to overfit or fail. Gathering large amounts of labeled and standardized seismic data sets is not straightforward. This shortage of quality data is addressed by building a generative adversarial network (GAN) to augment the original training data set, which is then used by DL-driven seismic tomography as input. The DL tomographic operator predicts velocity models with high statistical and structural accuracy after being trained with GAN-generated velocity models. Beyond the field of exploration geophysics, the use of machine learning in earth science is challenged by the lack of labeled data or properly interpreted ground truth, since we seldom know what truly exists beneath the earth's surface. The unsupervised approach (using GANs to generate labeled data)illustrates a way to mitigate this problem and opens geology, geophysics, and planetary sciences to more DL applications.

Random noise attenuation using f-x regularized nonstationary autoregression

Geophysics ◽  
2012 ◽  
Vol 77 (2) ◽  
pp. V61-V69 ◽  
Author(s):  
Guochang Liu ◽  
Xiaohong Chen ◽  
Jing Du ◽  
Kailong Wu
Keyword(s):  
Seismic Data ◽  
Random Noise ◽  
Geological Structure ◽  
Least Square ◽  
Noise Attenuation ◽  
Spatial Direction ◽  
Random Noise Attenuation ◽  
Time Space ◽  
Prediction Technique ◽  
Domain Prediction

We have developed a novel method for random noise attenuation in seismic data by applying regularized nonstationary autoregression (RNA) in the frequency-space ([Formula: see text]) domain. The method adaptively predicts the signal with spatial changes in dip or amplitude using [Formula: see text] RNA. The key idea is to overcome the assumption of linearity and stationarity of the signal in conventional [Formula: see text] domain prediction technique. The conventional [Formula: see text] domain prediction technique uses short temporal and spatial analysis windows to cope with the nonstationary of the seismic data. The new method does not require windowing strategies in spatial direction. We implement the algorithm by an iterated scheme using the conjugate-gradient method. We constrain the coefficients of nonstationary autoregression (NA) to be smooth along space and frequency in the [Formula: see text] domain. The shaping regularization in least-square inversion controls the smoothness of the coefficients of [Formula: see text] RNA. There are two key parameters in the proposed method: filter length and radius of shaping operator. Tests on synthetic and field data examples showed that, compared with [Formula: see text] domain and time-space domain prediction methods, [Formula: see text] RNA can be more effective in suppressing random noise and preserving the signals, especially for complex geological structure.

Deep learning spectral enhancement considering features of seismic field data

Geophysics ◽  
2021 ◽  
pp. 1-60
Author(s):  
Yonggyu Choi ◽  
Yeonghwa Jo ◽  
Soon Jee Seol ◽  
Joongmoo Byun ◽  
Young Kim
Keyword(s):  
Machine Learning ◽  
Seismic Data ◽  
Input Data ◽  
Field Data ◽  
A Priori ◽  
Resolution Enhancement ◽  
Training Data ◽  
A Priori Information ◽  
Spectral Enhancement ◽  
Priori Information

The resolution of seismic data dictates the ability to identify individual features or details in a given image, and the temporal (vertical) resolution is a function of the frequency content of a signal. To improve thin-bed resolution, broadening of the frequency spectrum is required; this has been one of the major objectives in seismic data processing. Recently, many researchers have proposed machine learning based resolution enhancement and showed their applicability. However, since the performance of machine learning depends on what the model has learned, output from training data with features different from the target field data may be poor. Thus, we present a machine learning based spectral enhancement technique considering features of seismic field data. We used a convolutional U-Net model, which preserves the temporal connectivity and resolution of the input data, and generated numerous synthetic input traces and their corresponding spectrally broadened traces for training the model. A priori information from field data, such as the estimated source wavelet and reflectivity distribution, was considered when generating the input data for complementing the field features. Using synthetic tests and field post-stack seismic data examples, we showed that the trained model with a priori information outperforms the models trained without a priori information in terms of the accuracy of enhanced signals. In addition, our new spectral enhancing method was verified through the application to the high-cut filtered data and its promising features were presented through the comparison with well log data.

scholarly journals Can learning from natural image denoising be used for seismic data interpolation?

Geophysics ◽  
2020 ◽  
Vol 85 (4) ◽  
pp. WA115-WA136 ◽  
Author(s):  
Hao Zhang ◽  
Xiuyan Yang ◽  
Jianwei Ma
Keyword(s):  
Deep Learning ◽  
Seismic Data ◽  
Interpolation Method ◽  
Signal To Noise Ratio ◽  
Curvelet Transform ◽  
Block Matching ◽  
Natural Image ◽  
Data Interpolation ◽  
Sampling Locations ◽  
Text Prediction

We have developed an interpolation method based on the denoising convolutional neural network (CNN) for seismic data. It provides a simple and efficient way to break through the problem of the scarcity of geophysical training labels that are often required by deep learning methods. This new method consists of two steps: (1) training a set of CNN denoisers to learn denoising from natural image noisy-clean pairs and (2) integrating the trained CNN denoisers into the project onto convex set (POCS) framework to perform seismic data interpolation. We call it the CNN-POCS method. This method alleviates the demands of seismic data that require shared similar features in the applications of end-to-end deep learning for seismic data interpolation. Additionally, the adopted method is flexible and applicable for different types of missing traces because the missing or down-sampling locations are not involved in the training step; thus, it is of a plug-and-play nature. These indicate the high generalizability of the proposed method and a reduction in the necessity of problem-specific training. The primary results of synthetic and field data show promising interpolation performances of the adopted CNN-POCS method in terms of the signal-to-noise ratio, dealiasing, and weak-feature reconstruction, in comparison with the traditional [Formula: see text]-[Formula: see text] prediction filtering, curvelet transform, and block-matching 3D filtering methods.

Sign in / Sign up

Export Citation Format

Share Document