• Title, Summary, Keyword: pseudo-Hessian matrix

Search Result 6, Processing Time 0.037 seconds

Frequency domain elastic full waveform inversion using the new pseudo-Hessian matrix: elastic Marmousi-2 synthetic test (향상된 슈도-헤시안 행렬을 이용한 탄성파 완전 파형역산)

  • Choi, Yun-Seok;Shin, Chang-Soo;Min, Dong-Joo
    • 한국지구물리탐사학회:학술대회논문집
    • /
    • /
    • pp.329-336
    • /
    • 2007
  • For scaling of the gradient of misfit function, we develop a new pseudo-Hessian matrix constructed by combining amplitude field and pseudo-Hessian matrix. Since pseudo- Hessian matrix neglects the calculation of the zero-lag auto-correlation of impulse responses in the approximate Hessian matrix, the pseudo-Hessian matrix has a limitation to scale the gradient of misfit function compared to the approximate Hessian matrix. To validate the new pseudo- Hessian matrix, we perform frequency-domain elastic full waveform inversion using this Hessian matrix. By synthetic experiments, we show that the new pseudo-Hessian matrix can give better convergence to the true model than the old one does. Furthermore, since the amplitude fields are intrinsically obtained in forward modeling procedure, we do not have to pay any extra cost to compute the new pseudo-Hessian. We think that the new pseudo-Hessian matrix can be used as an alternative of the approximate Hessian matrix of the Gauss-Newton method.

  • PDF

Construction the pseudo-Hessian matrix in Gauss-Newton Method and Seismic Waveform Inversion (Gauss-Newton 방법에서의 유사 Hessian 행렬의 구축과 이를 이용한 파형역산)

  • Ha, Tae-Young
    • Geophysics and Geophysical Exploration
    • /
    • v.7 no.3
    • /
    • pp.191-196
    • /
    • 2004
  • Seismic waveform inversion can be solved by using the classical Gauss-Newton method, which needs to construct the huge Hessian by the directly computed Jacobian. The property of Hessian mainly depends upon a source and receiver aperture, a velocity model, an illumination Bone and a frequency content of source wavelet. In this paper, we try to invert the Marmousi seismic data by controlling the huge Hessian appearing in the Gauss-Newton method. Wemake the two kinds of he approximate Hessian. One is the banded Hessian and the other is the approximate Hessian with automatic gain function. One is that the 1st updated velocity model from the banded Hessian is nearly the same of the result from the full approximate Hessian. The other is that the stability using the automatic gain function is more improved than that without automatic gain control.

Pseudo-multiscale Waveform Inversion for Velocity Modeling

  • Yang Dongwoo;Shin Changsoo;Yoon Kwangjin;Yang Seungjin;Suh Junghee;Hong Soonduk
    • Proceedings of the KSEEG Conference
    • /
    • /
    • pp.159-162
    • /
    • 2002
  • We tried to obtain an initial velocity model for prestack depth migration via waveform inversion. For application of any field data we chose a smooth background layered velocity model (v=v0 + k x z) as an initial velocity model. Newton type waveform inversion needs to invert huge Hessian matrix. In order to compute full Hessian matrix arising from full aperture data and full illumination zone, we meet insurmountable difficulties of paying astronomical computing cost. For the layered media, approximate Hessian emerging from single shot aperture data can be used repeatedly for split spread source configuration. In our work of using this Hessian characteristic of layered media we attempted to obtain the approximate velocity model as close as possible to the true velocity model in first iteration.

  • PDF

Depth Scaling Strategy Using a Flexible Damping Factor forFrequency-Domain Elastic Full Waveform Inversion

  • Oh, Ju-Won;Kim, Shin-Woong;Min, Dong-Joo;Moon, Seok-Joon;Hwang, Jong-Ha
    • Journal of the Korean earth science society
    • /
    • v.37 no.5
    • /
    • pp.277-285
    • /
    • 2016
  • We introduce a depth scaling strategy to improve the accuracy of frequency-domain elastic full waveform inversion (FWI) using the new pseudo-Hessian matrix for seismic data without low-frequency components. The depth scaling strategy is based on the fact that the damping factor in the Levenberg-Marquardt method controls the energy concentration in the gradient. In other words, a large damping factor makes the Levenberg-Marquardt method similar to the steepest-descent method, by which shallow structures are mainly recovered. With a small damping factor, the Levenberg-Marquardt method becomes similar to the Gauss-Newton methods by which we can resolve deep structures as well as shallow structures. In our depth scaling strategy, a large damping factor is used in the early stage and then decreases automatically with the trend of error as the iteration goes on. With the depth scaling strategy, we can gradually move the parameter-searching region from shallow to deep parts. This flexible damping factor plays a role in retarding the model parameter update for shallow parts and mainly inverting deeper parts in the later stage of inversion. By doing so, we can improve deep parts in inversion results. The depth scaling strategy is applied to synthetic data without lowfrequency components for a modified version of the SEG/EAGE overthrust model. Numerical examples show that the flexible damping factor yields better results than the constant damping factor when reliable low-frequency components are missing.

Covariance Estimates of GLMs under Model Misspecification

  • Yoon, Heewon;Jeong, Kwang Mo
    • Journal of the Korean Data Analysis Society
    • /
    • v.20 no.5
    • /
    • pp.2177-2187
    • /
    • 2018
  • In analyzing data set we usually assume a tentative model that seems to be valid. But the assumed model may be incorrect in respect of distributional assumption, regression relationship, and so on. The incorrect model usually causes larger variance estimates than the naive variance estimates that are given as the inverse of Hessian matrix. In this paper, we discuss robust variance estimates that are more valid in the presence of model misspecification. Through a Monte Carlo study we investigate the improvement of robust variance estimates against the naive estimates in respect of attaining nominal coverage probabilities for the regression coefficients of generalized linear models. We consider logistic GLMs for binomial responses versus beta-binomial or zero-inflated binomial responses, and also loglinear GLMs for Poisson counts versus negative binomial or zero-inflated counts responses. In the presence of model misspecification, the robust covariance estimator appears to be desirable in attaining nominal significance level.

Time-domain Seismic Waveform Inversion for Anisotropic media (이방성을 고려한 탄성매질에서의 시간영역 파형역산)

  • Lee, Ho-Yong;Min, Dong-Joo;Kwon, Byung-Doo;Yoo, Hai-Soo
    • 한국지구물리탐사학회:학술대회논문집
    • /
    • /
    • pp.51-56
    • /
    • 2008
  • The waveform inversion for isotropic media has ever been studied since the 1980s, but there has been few studies for anisotropic media. We present a seismic waveform inversion algorithm for 2-D heterogeneous transversely isotropic structures. A cell-based finite difference algorithm for anisotropic media in time domain is adopted. The steepest descent during the non-linear iterative inversion approach is obtained by backpropagating residual errors using a reverse time migration technique. For scaling the gradient of a misfit function, we use the pseudo Hessian matrix which is assumed to neglect the zero-lag auto-correlation terms of impulse responses in the approximate Hessian matrix of the Gauss-Newton method. We demonstrate the use of these waveform inversion algorithm by applying them to a two layer model and the anisotropic Marmousi model data. With numerical examples, we show that it's difficult to converge to the true model when we assumed that anisotropic media are isotropic. Therefore, it is expected that our waveform inversion algorithm for anisotropic media is adequate to interpret real seismic exploration data.

  • PDF