Article
Geochemistry & Geophysics
Aaron Davis
Summary: Airborne geophysical surveys often collect data with sample spacing distances that are much smaller than between-line separations. Interpolating grids and maps from these surveys can lead to aliasing and artifacts such as boudinage or string-of-beads effects when crossing flight lines. This study proposes a novel gridding method using a nonstationary nested anisotropic gridding scheme to address the boudinage effects. The method is validated using synthetic data and case histories from real airborne geophysical surveys, demonstrating improved interpretation confidence.
Article
Computer Science, Artificial Intelligence
Muhammad Ismail, Changjing Shang, Jing Yang, Qiang Shen
Summary: This article proposes a sparse data-based approach for image super-resolution using ANFIS interpolation. By splitting the training dataset into sufficient and sparse subsets, different techniques are applied accordingly. The experimental evaluation shows positive results for the proposed method.
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Geochemistry & Geophysics
Fantong Kong, Francesco Picetti, Vincenzo Lipari, Paolo Bestagini, Xiaoming Tang, Stefano Tubaro
Summary: The irregularity and coarse spatial sampling of seismic data can significantly impact the performance of processing and imaging algorithms. To address this issue, the study proposes a seismic data interpolation method based on the deep prior paradigm, utilizing a convolutional neural network as a prior to solve the interpolation inverse problem. The approach effectively leverages a multiresolution U-Net with 3-D convolution kernels, taking advantage of correlations in seismic data cubes at different scales in all directions. Numerical examples demonstrate the effectiveness and promising features of the proposed approach.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Mathematics, Applied
Thomas C. H. Lux, Layne T. Watson, Tyler H. Chang, Yili Hong, Kirk Cameron
Summary: Advancements in data availability have led to more accurate models in all scientific fields. Interpolation has advantages in handling high-dimensional approximation problems, as demonstrated in this paper with a novel and insightful error bound for linear interpolation.
NUMERICAL ALGORITHMS
(2021)
Article
Geosciences, Multidisciplinary
Nicola Piana Agostinetti, Alberto Malinverno, Thomas Bodin, Christina Dahner, Savka Dineva, Eduard Kissling
Summary: In this study, a novel approach for automatic weighting of data in geophysical inverse problems is presented, based on a trans-dimensional algorithm. The approach is applied to seismic event location in mines and achieves consistent results compared to a more standard method. Our approach outperforms standard seismic monitoring approaches when limited information is available on local seismic structure.
GEOPHYSICAL RESEARCH LETTERS
(2023)
Article
Environmental Sciences
Chengdong Xu, Jinfeng Wang, Maogui Hu, Wei Wang
Summary: This study introduces a Spatial-Temporal Point Interpolation based on Biased Sentinel Hospitals Areal Disease Estimation (STPI-BSHADE) method to address the issue of incomplete air quality datasets in environmental studies. The method quantifies the spatial association of variables to obtain estimates of missing data, and compares its accuracy with other statistical methods, demonstrating its superiority.
ENVIRONMENT INTERNATIONAL
(2022)
Article
Engineering, Multidisciplinary
Vikram Bhamidipati, Loukas F. Kallivokas, Gregory J. Rodin
Summary: This paper focuses on the inverse problem of detecting and localizing defects in finite lattices. By applying polarization dipoles at the ends of defective bars, the problem of defective lattice is transformed into a problem on the pristine lattice. Therefore, the inverse problem of localizing lattice defects is reduced to an inverse source problem.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2022)
Article
Acoustics
Manuel Hahmann, Samuel A. Verburg, Efren Fernandez-Grande
Summary: Sound field analysis methods utilize local representations to characterize and reconstruct complex sound fields, reducing model discrepancies and using data-driven approaches for suitable models. The use of dictionary learning and principal component analysis demonstrate the potential for modeling diverse sound fields based on their local and statistical properties.
JOURNAL OF THE ACOUSTICAL SOCIETY OF AMERICA
(2021)
Article
Computer Science, Artificial Intelligence
Jing Yang, Changjing Shang, Ying Li, Fangyi Li, Liang Shen, Qiang Shen
Summary: This article introduces a new ANFIS learning approach that uses an evolutionary process and interpolation to generate high-performing ANFIS models in cases of data shortage.
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2022)
Article
Computer Science, Interdisciplinary Applications
Chudong Pan, Xiongjie Deng, Zhenjie Huang
Summary: This study proposes a parallel computing-oriented method for dealing with long-time duration force identification problems. The method partitions the problem into sub-problems, solves them in parallel considering unknown initial conditions, and fuses the identified results. Numerical simulations show that the method can effectively identify dynamic forces in long-time durations and save computing time.
ENGINEERING WITH COMPUTERS
(2022)
Article
Computer Science, Interdisciplinary Applications
Youssef El Seblani, Elyas Shivanian
Summary: This paper presents a suitable method for treating partial derivative equations, specifically the Laplace equation with Robin boundary conditions. The approach used is a nodal Hermite meshless collocation technique, incorporating radial basis functions to obtain shape functions and applying Hermite interpolation technique to impose boundary conditions directly, known as MRPHI. Trustworthy results were obtained through examples demonstrating the effectiveness of the method.
ENGINEERING WITH COMPUTERS
(2021)
Article
Geochemistry & Geophysics
Dario Grana, Mingliang Liu, Mohit Ayani
Summary: The text discusses the use of numerical models to predict the distribution of carbon dioxide in deep saline aquifers and depleted reservoirs, and how geophysical data can be used to monitor and predict CO2 saturation. It introduces a new geostatistical inversion approach that combines stochastic optimization methods for predicting the saturation of CO2.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Engineering, Electrical & Electronic
Brendt Wohlberg, Przemek Wozniak
Summary: The algorithm presents a new method for estimating the Point Spread Function in wide-field astronomical images with extreme source crowding, utilizing convolutional sparse representations and avoiding the need to detect and localize individual point sources. In experiments involving simulated astronomical imagery, it significantly outperforms recent alternative methods.
IEEE SIGNAL PROCESSING LETTERS
(2021)
Article
Geochemistry & Geophysics
Yaojun Wang, Guiqian Zhang, Ting Chen, Yu Liu, Bingxin Shen, Jiandong Liang, Guangmin Hu
Summary: In this study, a data and model dual-driven seismic deconvolution method based on error-constrained joint sparse representation is proposed. The features of borehole reflectivity and surface seismic data are combined through joint dictionary learning, and the relationship between seismic waveforms and reflectivity is captured by sparse coefficients. This method effectively improves the resolution and accuracy of deconvolution.
Article
Mathematics, Applied
Zhaoying Wei, Guangsheng Wei
Summary: We consider the inverse spectral problem for the Sturm-Liouville problem and provide a method for the unique reconstruction of the potential.
Article
Geochemistry & Geophysics
Yangkang Chen, Sergey Fomel, Ray Abma
Summary: This study proposes a joint inversion framework to simultaneously separate blended sources and invert for the shot time. The method iteratively inverts the shot-time vector using a Gauss-Newton method and then applies a traditional deblending framework. Results from synthetic and field data examples show the effectiveness of the proposed approach and the significance of the low-frequency component in the nonlinear inversion of the shot-time vector.
Article
Geochemistry & Geophysics
Wei Chen, Yapo Abole Serge Innocent Oboue, Yangkang Chen
Summary: Seismic denoising often results in signal leakage, which traditional methods either ignore or try to decrease at the cost of residual noise. This paper proposes a robust dictionary learning and sparse coding algorithm to retrieve the leaked signals, by using a Huber-norm sparse coding model and iterative solving. Synthetic and field data examples demonstrate the effectiveness of the proposed method.
Article
Geochemistry & Geophysics
Omar M. Saad, Sergey Fomel, Raymond Abma, Yangkang Chen
Summary: We propose an unsupervised deep learning framework for simultaneous denoising and reconstruction of 3D seismic data without prior information and labels. Iterative reconstruction and POCS algorithm are used, with patching technique and attention mechanism to enhance learning capability. Synthetic and field examples demonstrate the superiority of the proposed method over benchmark methods.
Article
Geochemistry & Geophysics
Liuqing Yang, Sergey Fomel, Shoudong Wang, Xiaohong Chen, Wei Chen, Omar M. Saad, Yangkang Chen
Summary: Distributed acoustic sensing (DAS) is used to acquire seismic data due to its high-density and low-cost advantages. However, the acquired DAS data is often masked by complex noise, resulting in a low signal-to-noise ratio. In this study, a fully convolutional neural network with dense and residual connections is proposed to attenuate complex noise in DAS data. The network is trained using labeled data generated from an integrated framework and can effectively recover weak hidden signals.
Article
Geochemistry & Geophysics
Yangkang Chen, Alexandros Savvaidis, Sergey Fomel, Omar M. Saad, Yunfeng Chen
Summary: Passive seismic source location imaging is of great importance in a wide range of scientific and engineering research fields. The proposed RFloc3D method leverages machine learning techniques to accurately locate passive seismic sources based on traveltime information. The method demonstrates real-time location capability and the inclusion of S-wave arrivals greatly reduces depth errors.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Yangkang Chen, Alexandros Savvaidis, Sergey Fomel, Yunfeng Chen, Omar M. Saad, Hang Wang, Yapo Abole Serge Innocent Oboue, Liuqing Yang, Wei Chen
Summary: Distributed acoustic sensing (DAS) is a promising technology for high-resolution multi-scale seismic investigation. However, DAS data are often affected by strong noise. In this study, we propose a novel denoising framework that combines multiple methods to suppress specific types of noise. Through the application of a bandpass filter, a structure-oriented median filter, and a carefully designed dip filter, the SNR of DAS data is dramatically improved.
SEISMOLOGICAL RESEARCH LETTERS
(2023)
Article
Geochemistry & Geophysics
Xintao Chai, Taihui Yang, Hanming Gu, Genyang Tang, Wenjun Cao, Yufeng Wang
Summary: Deep learning has made remarkable progress in geophysics. However, the traditional supervised learning framework faces the problem of limited or unavailable labels in seismic data applications, and it cannot generate physically consistent results. Therefore, we provide an open-source package for geophysics-steered self-supervised learning in seismic deconvolution. Experimental results show that this approach outperforms the traditional trace-by-trace method in terms of accuracy and spatial continuity.
GEOPHYSICAL JOURNAL INTERNATIONAL
(2023)
Article
Geochemistry & Geophysics
Liuqing Yang, Shoudong Wang, Xiaohong Chen, Omar M. Saad, Wanli Cheng, Yangkang Chen
Summary: Ground roll noise seriously affects the useful reflection signals in seismic data and decreases the signal-to-noise ratio. We propose a fully convolutional framework called GRDNet to attenuate ground roll in land seismic data. The testing results show that GRDNet can effectively reduce ground roll noise and preserve useful reflection signals.
SURVEYS IN GEOPHYSICS
(2023)
Article
Geochemistry & Geophysics
Yangkang Chen, Alexandros Savvaidis, Sergey Fomel
Summary: Passive seismic denoising is often done using band-pass filters, but this can be problematic when signal and noise have the same frequency. This study presents a new data-driven denoising method based on adaptive sparse transform. The method is flexible and can be applied to any passive seismic monitoring project.
SEISMOLOGICAL RESEARCH LETTERS
(2023)
Article
Computer Science, Artificial Intelligence
Omar M. Saad, Wei Chen, Fangxue Zhang, Liuqing Yang, Xu Zhou, Yangkang Chen
Summary: In this paper, a fully convolutional DenseNet method for automatic salt segmentation is proposed, with a squeeze-and-excitation network used as a self-attention mechanism to extract the important information related to the salt signals. The method demonstrates robust performance when applied to new datasets using transfer learning and a small amount of training data.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Liuqing Yang, Shoudong Wang, Xiaohong Chen, Wei Chen, Omar M. Saad, Xu Zhou, Nam Pham, Zhicheng Geng, Sergey Fomel, Yangkang Chen
Summary: This article proposes a deep learning model combining convolutional and recurrent networks for predicting reservoir permeability and porosity. By training a combination of nonlinear and linear modules, the mapping between logging data and reservoir parameters is established. Optimization algorithms and a self-attention mechanism are used to improve prediction accuracy. Testing on blind wells from different regions demonstrates the accuracy and robustness of the proposed method.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Geochemistry & Geophysics
Huijian Li, Bo Liu, Xu Liu, Abdullatif A. Al-Shuhail, Sherif M. H. Mahmoud, Yangkang Chen
Summary: In this article, a novel frequency-independent CFS (FiCFS) method is proposed for signal attenuation estimation with higher adaptability. It is proved mathematically that the centroid frequency and variance are frequency insensitive for arbitrary frequency bands, and the FiCFS method is derived based on this property. Experimental results validate the adaptability, noise immunity, and reliability of the proposed method.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Yuanpeng Zhang, Hui Zhou, Mingzhu Zhang, Yufeng Wang, Bin Feng, Meng Liang
Summary: The establishment of a proper initial subsurface model is crucial for improving the quality of seismic inversion. This study proposes a novel structurally constrained modeling method (SCMM) that incorporates geological rules into the initial model. Synthetic and field data tests demonstrate that SCMM outperforms the traditional method in terms of convergence property and accuracy of inversion result.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Omar M. Saad, Yunfeng Chen, Daniel Siervo, Fangxue Zhang, Alexandros Savvaidis, Guo-chin Dino Huang, Nadine Igonin, Sergey Fomel, Yangkang Chen
Summary: We propose a compact convolutional transformer (CCT) for earthquake phase arrival picking, which outperforms other deep learning methods in terms of picking accuracy. The pre-trained model achieves higher accuracy on independent datasets, such as Japanese, Texas, and Instance.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Hagar S. Elsayed, Omar M. Saad, M. Sami Soliman, Yangkang Chen, Hassan A. Youness
Summary: A novel deep-learning method called ConvMixer network is proposed for automatic earthquake location. The ConvMixer network achieves high accuracy in estimating the hypocenter locations and outperforms benchmark methods ResNet, AlexNet, MobileNet, and Xception.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)