Zeyu Zhao

Zeyu Zhao

Auto-generated profile, based on publicly available information

Zeyu Zhao

Institution of Theoretical and Applied Geophysics, School of Earth and Space Sciences Peking University Beijing China

Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
22 September 2021 in Geophysical Journal International
Verified
Journal Article
Closed access
PDF file
ZGZhicheng Geng
Zhicheng Geng
ZZZeyu Zhao
Zeyu Zhao
YSYunzhi Shi
Yunzhi Shi
XWXinming Wu
Xinming Wu
SFSergey Fomel
Sergey Fomel
MSMrinal Sen
Mrinal Sen
SUMMARY: Subsurface velocity model building is a crucial step for seismic imaging. It is a challenging problem for conventional methods such as full-waveform inversion (FWI) and wave equation migration velocity analysis (WEMVA), due to the highly nonlinear relationship between subsurface velocity values and seismic responses. In addition, traditional FWI and WEMVA methods are often computationally expensive. In this paper, we propose to apply a deep learning technique to construct subsurface velocity models automatically from common-image gather (CIG) volumes. In our method, pairs of synthetic velocity models and CIG volumes are generated to train a convolutional neural network. Our proposed network achieves promising results on different synthetic data sets. The training performance of several commonly used loss functions is also studied.

16 December 2020 in Geophysics
Verified
Journal Article
Closed access
ZZZeyu Zhao
Zeyu Zhao
MSMrinal K. Sen
Mrinal K. Sen
Traditional full-waveform inversion (FWI) methods only render a “best-fit” model that cannot account for uncertainties of the ill-posed inverse problem. Additionally, local optimization-based FWI methods cannot always converge to a geologically meaningful solution unless the inversion starts with an accurate background model. We seek the solution for FWI in the Bayesian inference framework to address those two issues. In Bayesian inference, the model space is directly probed by sampling methods such that we obtain a reliable uncertainty appraisal, determine optimal models, and avoid entrapment in a small local region of the model space. The solution of such a statistical inverse method is completely described by the posterior distribution, which quantifies the distributions for parameters and inversion uncertainties. To efficiently sample the posterior distribution, we introduce a sampling algorithm in which the proposal distribution is constructed by the local gradient and the diagonal approximate Hessian of the local log posterior. Our algorithm is called the gradient-based Markov chain Monte Carlo (GMCMC) method. The GMCMC FWI method can quantify inversion uncertainties with estimated posterior distribution given sufficiently long Markov chains. By directly sampling the posterior distribution, we obtain a global view of the model space. Theoretically speaking, statistical assessments do not depend on starting models. Our method is applied to the 2D Marmousi model with the frequency-domain FWI setting. Numerical results suggest that our method can be readily applied to 2D cases with affordable computational efforts.

28 January 2016 in Advanced Optical Materials
Verified
Journal Article
Closed access
PDF file
XLXiong Li
Xiong Li
Mingbo PuMingbo Pu
Mingbo Pu
YWYanqin Wang
Yanqin Wang
XMXiaoliang Ma
Xiaoliang Ma
YLYang Li
Yang Li
XLXiangang Luo
Xiangang Luo

14 November 2016 in Optics Express
Verified
Journal Article
Open access
PDF file
ZZZeyu Zhao
Zeyu Zhao
YLYunfei Luo
Yunfei Luo
NYNa Yao
Na Yao
WZWei Zhang
Wei Zhang
CWChangtao Wang
Changtao Wang
XLXiangang Luo
Xiangang Luo
Plasmonic lens imaging with some resolution enhancement methods are investigated in this paper, mainly by physical modeling and numerical simulations. The imaging model is based on the refined optical transfer function with extra reflection in imaging space and measured in variant magnetic and electric field components. The influences of structured light illumination and mask patterns' modifications are considered as well. As experimental demonstrations, L-shaped slits pattern with a half-pitch of 60 nm is successfully imaged with 50 nm air distance, by using plasmonic cavity lens lithography and off-axis illumination with 365 nm wavelength light. This study is believed to provide the model and methods for the design of high resolution plasmonic lens employed in nano lithography and optical storage etc.

14 August 2022 in Journal of Geophysical Research: Solid Earth
Verified
Journal Article
Closed access
ZZZeyu Zhao
Zeyu Zhao
MSMrinal K. Sen
Mrinal K. Sen
BDBertrand Denel
Bertrand Denel
DSDong Sun
Dong Sun
PWPaul Williamson
Paul Williamson

1 October 2018 in Geophysics
Verified
Journal Article
Closed access
ZZZeyu Zhao
Zeyu Zhao
MSMrinal K. Sen
Mrinal K. Sen
We have developed a fast image-domain target-oriented least-squares reverse time migration (LSRTM) method based on applying the inverse or pseudoinverse of a target-oriented Hessian matrix to a migrated image. The image and the target-oriented Hessian matrix are constructed using plane-wave Green’s functions that are computed by solving the two-way wave equation. Because the number of required plane-wave Green’s functions is small, the proposed method is highly efficient. We exploit the sparsity of the Hessian matrix by computing only a couple of off-diagonal terms for the target-oriented Hessian, which further improves the computational efficiency. We examined the proposed LSRTM method using the 2D Marmousi model. We demonstrated that our method correctly recovers the reflectivity model, and the retrieved results have more balanced illumination and higher spatial resolution than traditional images. Because of the low cost of computing the target-oriented Hessian matrix, the proposed method has the potential to be applied to large-scale problems.