TY - JOUR

T1 - A nonlinear inversion for the velocity background and perturbation models

AU - Wu, Zedong

AU - Alkhalifah, Tariq Ali

N1 - KAUST Repository Item: Exported on 2020-10-01

PY - 2015/8/19

Y1 - 2015/8/19

N2 - Reflected waveform inversion (RWI) provides a method to reduce the nonlinearity of the standard full waveform inversion (FWI) by inverting for the single scattered wavefield obtained using an image. However, current RWI methods usually neglect diving waves, which is an important source of information for extracting the long wavelength components of the velocity model. Thus, we propose a new optimization problem through breaking the velocity model into the background and the perturbation in the wave equation directly. In this case, the perturbed model is no longer the single scattering model, but includes all scattering. We optimize both components simultaneously, and thus, the objective function is nonlinear with respect to both the background and perturbation. The new introduced w can absorb the non-smooth update of background naturally. Application to the Marmousi model with frequencies that start at 5 Hz shows that this method can converge to the accurate velocity starting from a linearly increasing initial velocity. Application to the SEG2014 demonstrates the versatility of the approach.

AB - Reflected waveform inversion (RWI) provides a method to reduce the nonlinearity of the standard full waveform inversion (FWI) by inverting for the single scattered wavefield obtained using an image. However, current RWI methods usually neglect diving waves, which is an important source of information for extracting the long wavelength components of the velocity model. Thus, we propose a new optimization problem through breaking the velocity model into the background and the perturbation in the wave equation directly. In this case, the perturbed model is no longer the single scattering model, but includes all scattering. We optimize both components simultaneously, and thus, the objective function is nonlinear with respect to both the background and perturbation. The new introduced w can absorb the non-smooth update of background naturally. Application to the Marmousi model with frequencies that start at 5 Hz shows that this method can converge to the accurate velocity starting from a linearly increasing initial velocity. Application to the SEG2014 demonstrates the versatility of the approach.

UR - http://hdl.handle.net/10754/593054

UR - http://library.seg.org/doi/10.1190/segam2015-5846177.1

UR - http://www.scopus.com/inward/record.url?scp=85018982801&partnerID=8YFLogxK

U2 - 10.1190/segam2015-5846177.1

DO - 10.1190/segam2015-5846177.1

M3 - Article

VL - 34

SP - 1292

EP - 1296

JO - SEG Technical Program Expanded Abstracts 2015

JF - SEG Technical Program Expanded Abstracts 2015

SN - 1949-4645

ER -