Numerical seismogram computations for inhomogeneous media using a short, variable length convolutional differentiator
- Flinders Univ. of South Australia, Adelaide (Australia). School of Earth Sciences
A short convolutional differentiator (CD) for computing second spatial derivatives in the acoustic wave equation is presented. This differentiator is obtained by tapering the inverse Fourier transform of the band-limited Fourier spectrum of the second-derivative operator. This new filter has been applied to seismogram computations for inhomogeneous media and results are compared with the conventional high-order finite-difference (FD) and Fourier schemes. The operator can be progressively shortened at the model edges to reduce boundary artifacts. The CD method is superior to the conventional FD operator and comparable with the Fourier method in accuracy but faster to run. A strategy to reduce computation time by 20%, which exploits the localized nature of the operator, is given. The method is illustrated using simple 2D models.
- OSTI ID:
- 6054228
- Journal Information:
- Geophysical Prospecting; (United States), Journal Name: Geophysical Prospecting; (United States) Vol. 41:6; ISSN GPPRAR; ISSN 0016-8025
- Country of Publication:
- United States
- Language:
- English
Similar Records
Efficient convolutional sparse coding
Synthesis of body wave seismograms from point sources in anisotropic media