Article
Environmental Sciences
Zhiyuan Gu, Xintao Chai, Taihui Yang
Summary: Low frequencies are crucial for FWI in retrieving long-scale features and reliable subsurface properties. However, they are often missing due to limitations in seismic acquisition. This study presents a DL-based LFR approach using a bridge-shaped 3D CNN to transform high frequencies into low frequencies. The experiments show that our approach accurately reconstructs low frequencies and outperforms 2D CNN in precision and low-frequency energy.
Article
Chemistry, Analytical
Thomas Robins, Jorge Camacho, Oscar Calderon Agudo, Joaquin L. Herraiz, Lluis Guasch
Summary: Ultrasound breast imaging is a promising alternative to conventional mammography, offering successful imaging of dense breast tissue without exposing women to harmful radiation. Full-waveform inversion (FWI) technology can achieve high-resolution images using low frequencies, but most USCT systems lack these frequencies.
Article
Geochemistry & Geophysics
Wei Zhang, Jinghuai Gao, Zhaoqi Gao, Hongling Chen
Summary: The article presents a new approach to FWI based on adjoint-driven deep learning, using a fully convolutional network to achieve high-resolution inversion of subsurface velocity. It addresses the issues of ill-posedness, nonlinearity, and cycle-skipping that are common in traditional methods.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Geochemistry & Geophysics
Jian Sun, Kristopher Innanen, Tianze Zhang, Daniel Trad
Summary: Full waveform inversion (FWI) is a state-of-the-art method for imaging subsurface structures and physical parameters with seismic data, but it faces challenges in implementation and use. The implicit full waveform inversion (IFWI) algorithm, designed with deep neural representations, shows improved convergence and the ability to capture high-resolution subsurface structures. Although uncertainty analysis is not fully solved, IFWI addresses it meaningfully by approximating Bayesian inference. Numerical experimentation suggests that IFWI has a strong capacity for generalization and is suitable for multi-scale joint geophysical inversion.
JOURNAL OF GEOPHYSICAL RESEARCH-SOLID EARTH
(2023)
Article
Geochemistry & Geophysics
Shihang Feng, Youzuo Lin, Brendt Wohlberg
Summary: Seismic full-waveform inversion is a powerful imaging technique in exploration geophysics, but faces computational challenges and limitations in accuracy due to the initial velocity models. This study presents a multiscale data-driven FWI method based on fully convolutional networks, which significantly improves accuracy and reduces computation time through training on synthetic subsurface velocity models.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Xiugang Xu, Peng Guo, Jidong Yang, Wende Xu, Siyou Tong
Summary: The cycle-skipping problem is a major obstacle in full-waveform inversion (FWI), which can be mitigated using a multiscale inversion scheme. This study explores the possibility of compensating low-frequency components in seismic records and applying them to FWI for accurate subsurface velocity model construction. Numerical experiments validate the effectiveness of the proposed method.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Computer Science, Artificial Intelligence
Vahid Honarbakhsh, Hamid Reza Siahkoohi, Mansoor Rezghi, Hamid Sabeti
Summary: In this paper, a deep learning method called SeisDeepNet is proposed to determine high resolution velocity models in complex geology settings. Compared to traditional full waveform inversion, this method shows higher efficiency and better preservation of model details.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Geochemistry & Geophysics
Qili Zeng, Shihang Feng, Brendt Wohlberg, Youzuo Lin
Summary: This article presents an efficient and scalable encoder-decoder network for 3-D FWI, which utilizes group convolution and invertible layers to improve the reconstruction performance of 3-D high-resolution velocity maps. Experiments demonstrate that the proposed method has advantages in computational cost and memory footprint compared to traditional approaches.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Zhaoqi Gao, Wei Yang, Chuang Li, Feipeng Li, Qingzhen Wang, Jicai Ding, Jinghuai Gao, Zongben Xu
Summary: In this article, a new nonlinear FWI method is proposed to mitigate the initial model dependence problem. It utilizes a nonlinear operator based on FCEO and U-Net, and employs self-supervised learning to train U-Net. The numerical examples demonstrate the effectiveness of the proposed method.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Computer Science, Artificial Intelligence
Meng Suo, Dong Zhang, Haiqi Yang, Yan Yang
Summary: Full waveform inversion (FWI) has potential for quantitative ultrasound bone imaging, but suffers from cycle skipping, local minima and high computational costs. In this study, we propose an improved dual-encoder-based Unet with high-frequency feature enhancement (DEFE-Unet) for ultrasonic bone quantitative imaging, which obtains more detailed information and achieves competitive results with less computational cost compared to FWI.
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Engineering, Multidisciplinary
Mingyu Yu, Fei Cheng, Jiangping Liu, Daicheng Peng, Zhijian Tian
Summary: Tunnel seismic detection methods are important for tunnel engineering, but current methods often lack accuracy in acquired geological information and physical properties. This study applies a frequency-domain acoustic full-waveform inversion method and discusses the influence of frequency group selection strategy and tunnel observation system settings on inversion results. Improved strategies are proposed to enhance resolution in imaging tunnel structure and physical parameters.
Article
Geochemistry & Geophysics
Weiguang He, Guanghui Hu, Bing Zhang
Summary: The conventional least squares full waveform inversion (FWI) lacks coherency along the spatial and temporal axes. This study proposes an optimal matching function that represents each receiver using extended vector attributes and calculates a cost matrix to determine receiver pairs. The algorithm avoids point-by-point comparison and reduces computation, and has been extensively validated in multiple models.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Hongyu Sun, Yen Sun, Rami Nammour, Christian Rivera, Paul Williamson, Laurent Demanet
Summary: Full-waveform inversion relies on low-frequency data but is limited by the availability of a good initial model. This paper proposes a method for extrapolating low frequencies for field data by training a CycleGAN with unpaired images of field seismic data and synthetic low-frequency data. The inverted velocity model using only extrapolated low frequencies is comparable to the tomography model, demonstrating the effectiveness of the method in mitigating the cycle-skipping problem and mapping fine Earth structures.
GEOPHYSICAL JOURNAL INTERNATIONAL
(2023)
Article
Geochemistry & Geophysics
Dirk Philip van Herwaarden, Michael Afanasiev, Solvi Thrastarson, Andreas Fichtner
Summary: We propose a new approach to full-waveform inversion that allows for continuous assimilation of growing data volumes without the need to reinvert all data. Specifically designed for seismological applications, our method utilizes a dynamic mini-batch stochastic L-BFGS to sequentially add new data while maintaining convergence and consistency in model fit measurement.
GEOPHYSICAL JOURNAL INTERNATIONAL
(2021)
Article
Geochemistry & Geophysics
Yuchen Jin, Yuan Zi, Wenyi Hu, Yanyan Hu, Xuqing Wu, Jiefu Chen
Summary: This study introduces a robust progressive learning algorithm based on deep learning, combining physics-guided FWI and data-driven deep learning technology to extrapolate low-frequency data for high efficiency and accuracy in subsurface imaging.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Geosciences, Multidisciplinary
Jinwei Fang, Hui Zhou, Qingchen Zhang, Hanming Chen, Pengyuan Sun, Jianlei Zhang, Liang Zhang
JOURNAL OF APPLIED GEOPHYSICS
(2020)
Article
Geochemistry & Geophysics
Jinwei Fang, Hanming Chen, Hui Zhou, Qingchen Zhang, Lide Wang
Summary: This paper presents a 3-D elastic full-waveform inversion (FWI) method using a temporal fourth-order finite-difference (FD) approximation. By solving a new elastic equation and introducing a novel inversion procedure, the accuracy of the inversion results is improved.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Geosciences, Multidisciplinary
Jinwei Fang, Hui Zhou, Jun Zhang, Qingchen Zhang, Shengdong Liu, Bo Wang
Summary: The study introduces a novel 3D elastic crosstalk-free multisource full-waveform inversion method (MFWI) that effectively avoids the impact of crosstalk noise on inversion quality and enhances the robustness of the algorithm by using a multi-source encoding strategy and phase-sensitive detection algorithm to decompose multi-source wavefields.
JOURNAL OF APPLIED GEOPHYSICS
(2022)
Article
Geosciences, Multidisciplinary
Honglei Shen, Gang Tian, Chunhui Tao, Hanchuang Wang, Jinwei Fang
Summary: The study confirms the significance of multi-frequency acquisition in enhancing the resolution of subsurface targets. The joint deconvolution approach and weighted matched filter contribute to better noise control and resolution enhancement, as demonstrated by both synthetic modeling and field data examples.
JOURNAL OF APPLIED GEOPHYSICS
(2022)
Article
Environmental Sciences
Jinwei Fang, Ying Shi, Hui Zhou, Hanming Chen, Qingchen Zhang, Ning Wang
Summary: This paper proposes a high-precision elastic reverse-time migration (ERTM) method based on high-order accuracy wavefields and vector-based imaging conditions, which can improve the accuracy and simulation effect of ERTM imaging. Experimental results show that this method can generate high-quality ERTM images and improve imaging locations and quality.
Article
Geosciences, Multidisciplinary
Tao Huang, Qingchen Zhang
Summary: In this paper, a hybrid sparsity-constrained multi-source encoding algorithm is proposed to overcome the problems faced by the conventional source-encoding algorithm. By utilizing harmonic wavelets as encoding operators, the algorithm can decompose the blended multi-source wavefields into each individual single-source wavefield based on the orthogonality of trigonometric functions. The application of combined-sparsity constraint and convolution-norm misfit function improves the robustness and convergence rate of the algorithm, leading to rational and feasible results as demonstrated by synthetic data examples.
JOURNAL OF APPLIED GEOPHYSICS
(2022)
Article
Geosciences, Multidisciplinary
Jinwei Fang, Lanying Huang, Ying Shi, Hanming Chen, Bo Wang
Summary: This paper proposes a vector-based 3D ERTM using the high-order accuracy SGFD method in time to obtain high-accuracy images of subsurface seismic structures.
FRONTIERS IN EARTH SCIENCE
(2023)
Article
Geochemistry & Geophysics
Jinwei Fang, Hui Zhou, Yunyue Elita Li, Ying Shi
Summary: Time-domain elastic full waveform inversion (FWI) is used to recover high-resolution subsurface properties for structural imaging and lithologic identification. A new method is proposed to eliminate the limitation of wavelet spectra in time-domain FWI, by using a refined seismic wavelet with normalized amplitude and accurate phase for wave propagation in the time domain. The proposed method enhances the low-wavenumber reconstruction of velocity models and avoids local minima in FWI.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Qingqing Li, Qingchen Zhang, Qizhen Du, Shijun Cheng
Summary: Full-waveform inversion (FWI) is a promising method for estimating high-resolution subsurface parameters, but it relies heavily on the accuracy of the initial model and is vulnerable to local minimum issues. This study proposes an efficient EFWI method that combines crosstalk-free multisource elastic FWI (MS-EFWI) and a well-guided initial model-building algorithm to improve the inversion results.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Yanan Ran, Qingqing Li, Li-Yun Fu, Qizhen Du, Qingchen Zhang
Summary: This work proposes a new wave-equation-based Q inversion methodology to evaluate more accurate underground Q values in local domain. This approach is applicable to both the early arrivals and reflected waves, and improves the inversion accuracy of Q by using different types of waves.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Jinwei Fang, Liang Zhang, Hui Zhou, Shengdong Liu, Bo Wang, Wenjie Chen
Summary: In this study, a denoising workflow based on multi-scale sparse dictionary learning is proposed to effectively remove noise from seismic data. The method is validated on synthetic and real datasets and shows promising results in signal recovery.
JOURNAL OF SEISMIC EXPLORATION
(2022)
Article
Geochemistry & Geophysics
Yang YuChen, Fang JinWei, Wang Ning, Li SongLing
Summary: The article introduces an efficient simultaneous source inversion method along with three inversion strategies: real wavefield inversion, imaginary wavefield inversion, and complex wavefield inversion, to improve the accuracy and efficiency of waveform inversion.
CHINESE JOURNAL OF GEOPHYSICS-CHINESE EDITION
(2021)
Article
Geochemistry & Geophysics
Liang Zhang, Liguo Han, Ao Chang, Jinwei Fang, Pan Zhang, Yong Hu, Zhengguang Liu
JOURNAL OF SEISMIC EXPLORATION
(2020)