Rapid Generation of Synthetic Seismograms in Layered Media by Vectorization of the Algorithm
HAWAII UNIV HONOLULU INST OF GEOPHYSICS
Pagination or Media Count:
The theory and practice of computing synthetic seismograms in layered media are now quite well understood. In applications, these methods can still be unsatisfactorily slow. It may be necessary to work with models having a large number of microlayers. In seeking to determine the variation of material properties in a sedimentary pile, we may start with some generalized reflection data set, containing reflected and refracted signals produced by an impulsive point source and recorded at a number of sensors in the water layer at various offsets from the source. Any true inversion procedure, regardless of details, must compare the observed data with synthetic data for a sequence of models in order to obtain and characterize models which fit the data within some prescribed limits. We thus seek a forward modeling procedure for arbitrary layered models which is fast enough that many repetitions are possible in a reasonable time. The procedure described in this paper is based on a reorganization of the inside loops of a conventional reflectivity algorithm to permit vectorization. Sedimentary rock.