PVM based 3-D Kirchhoff depth migration using dynamically computed travel-times: An application in seismic data processing

摘要

Seismic depth migration is an image reconstruction technique used to generate realistic images of the Earth's interior from surface recordings of sound waves reflected by buried geological structures. Migration algorithms that have been developed for digitally recorded seismic data are computationally intensive and are typically implemented on massively parallel architecture. In this paper we describe a highly accurate method of computing seismic travel- times and study the feasibility of a PVM-based SPMD parallelization of a migration algorithm based on this method. The parallel algorithm developed in this study is coarse grained and utilizes a simple geometrical decomposition of the input problem. We show that near linear speedup can be achieved on a homogeneous cluster and the algorithm can be ported to a variety of platforms, i.e. clusters of Sun workstations, Cray Y-MP. nCUBE-2, Cray T3E and SGI Origin 2000. The performance of the scheme on different computational platforms ulti- mately depends on a number of different factors, e.g., machine architecture, machine speed, network configuration etc. For a simple test case that was studied in this research, the per- formances of modest PVM clusters were found to be encouraging. This underscores the ad- vantages of adopting a strategy based on parallelizing computationally intensive tasks and also demonstrates the suitability of the travel-time scheme for developing coarsely parallel applications.