Network Distributed Paralled Processing of Pre-Stack Layer-Stripping Reverse-Time Migration

  • Author(s): Ruey-Chyuan Shih, Ta-Heng Hsiuan, Yih-Hsiung Yeh, and Kuo-An Lin
  • DOI: 10.3319/TAO.1993.4.3.243(T)
  • Keywords:
  • Citation: Shih, R.-C., T.-H. Hsiuan, Y.-H. Yeh, and K.-A. Lin, 1993: Network Distributed Paralled Processing of Pre-Stack Layer-Stripping Reverse-Time Migration. Terr. Atmos. Ocean. Sci., 4, 243-256, doi: 10.3319/TAO.1993.4.3.243(T)

Pre-stack layer-stripping reverse-time migration is successful for structures with steep dips and strong velocity contrasts. This technique uses reverse-time method to migrate seismic sections through constant or smoothly varying velocity layers, one layer at a time. The final migration result is composited from the individual layer's images. The most important advantage of the pre-stack layer-stripping reverse-time migration method is that it makes interpretation become a step of migration. However, although super computers have made computation less of a problem, applying pre-stack layer-stripping reverse-time migration to a large seismic model still requires significant computation expense.
With the help of modern computer network techniques, we have suc­cessfully developed a cheap and high efficient strategy to perform pre-stack layer-stripping reverse-time migration. This technique uses several networked workstations to share the works of migration. Migration is distributed to dif­ferent machines through network and all machines share the same data file. The networked distributed parallel processing technique greatly increases the computational efficiency, reduces the computational cost, and shares compu­tational resources available. Comparing the performance of migration on 6 clustered VAX workstations and a CONVEX C220 mini super computer, the total computer time on both systems are almost identical. Using the network distributed parallel processing technique, we may develop time consuming data processing algorithm with small computer systems. This technique is particularly important for 3-D data processing.

Read 4026 times
© 1990-2033 Terrestrial, Atmospheric and Oceanic Sciences (TAO). All rights reserved.

Published by The Chinese Geoscience Union