Study And Application Of Computing Grid Based On Seismic Migration Imaging

This paper is based on combination of two important areas, seismic migration and computing grid, and aims to integrate multiple clusters to provide enormous computing power for pre-stack migration process.Firstly, this paper designed multiple layer tree-like multi-clusters grid architecture according to feature of seismic data processing especially migration imaging. In this structure, multi-clusters are connected and integrated by high speed network and grid technology to become a super computing platform. And each cluster in grid can use different hardware architecture, different operation system and resource management software.Next, this paper researched grid data replica strategies, grid meta scheduler and meta scheduling policy etc. key technologies of computing grid. Fast Cascading replica strategy was put forward in terms of data replica by considering data replica's creation, removement etc. The experimental results of grid simulator show Fast Cascading strategy improved data transfer efficiency between grid nodes. GridSched meta scheduler model, two stage queue scheduling algorithm based on load balancing, dependent task scheduling algorithm are brought forward to increase jobs execution efficiency and enhance the usability. Then this paper raised p-v fast migration algorithm. The p-v fast migration reduces executing time of computing phase shift angle, square root and sin/cos operations by means of using look-up table.At last pre-stack migration method in multi-clusters grid is studied, implemented and tested. The results show that multi-clusters grid improves processing power and efficiency of computing platform, also reduces processing time by combined exits computing system using grid technology.