Computational aspects of Gaussian beam migration
The computational efficiency of Gaussian beam migration depends on the solution of two problems: (1) computation of complex-valued beam times and amplitudes in Cartesian (x,z) coordinates, and (2) limiting computations to only those (x,z) coordinates within a region where beam amplitudes are significant. The first problem can be reduced to a particular instance of a class of closest-point problems in computational geometry, for which efficient solutions, such as the Delaunay triangulation, are well known. Delaunay triangulation of sampled points along a ray enables the efficient location of that point on the raypath that is closest to any point (x,z) at which beam times and amplitudes are required. Although Delaunay triangulation provides an efficient solution to this closest point problem, a simpler solution, also presented in this paper, may be sufficient and more easily extended for use in 3-D Gaussian beam migration. The second problem is easily solved by decomposing the subsurface image into a coarse grid of square cells. Within each cell, simple and efficient loops over (x,z) coordinates may be used. Because the region in which beam amplitudes are significant may be difficult to represent with simple loops over (x,z) coordinates, I use recursion to move from cell to cell, until entire region defined by the beam has been covered. Benchmark tests of a computer program implementing these solutions suggest that the cost of Gaussian hewn migration is comparable to that of migration via explicit depth extrapolation in the frequency-space domain. For the data sizes and computer programs tested here, the explicit method was faster. However, as data size was increased, the computation time for Gaussian beam migration grew more slowly than that for the explicit method.
- Research Organization:
- Colorado School of Mines, Golden, CO (United States). Center for Wave Phenomena
- Sponsoring Organization:
- USDOE; USDOE, Washington, DC (United States)
- DOE Contract Number:
- FG02-89ER14079
- OSTI ID:
- 7200248
- Report Number(s):
- DOE/ER/14079-17; CWP-127; ON: DE92018641
- Country of Publication:
- United States
- Language:
- English
Similar Records
MOOSE Framework Meshing Enhancements to Support Reactor Analysis
Computational methods for improving the resolution of subsurface seismic images. Progress report, September 15, 1991--September 14, 1993
Related Subjects
58 GEOSCIENCES
99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE
SEISMIC WAVES
IMAGE PROCESSING
ALGORITHMS
BENCHMARKS
EXTRAPOLATION
GAUSS FUNCTION
MIGRATION
WAVE PROPAGATION
FUNCTIONS
MATHEMATICAL LOGIC
NUMERICAL SOLUTION
PROCESSING
422000* - Engineering- Mining & Underground Engineering- (1980-)
580000 - Geosciences
990200 - Mathematics & Computers