Abstract: Prestack depth migration based on wave equation is an efficient tool for seismic imaging in area with complex topography and geology characteristics. In practical application, however, it is confronted with two difficulties of enormous data volume and huge computation cost. Therefore, based on the PC Cluster and Message Passing Interface (MPI) program development environment, Job-by-Application master-slaver scheme was designed to implement the parallel computing of wave equation prestack depth migration. This method solved the problem of dynamic load balancing. Job registration and analysis approach used in this method settled the problem of fault-tolerance processing. The result with both theoretical model data and field data showed that this parallel program was correct, the running of the program was stable, and the efficiency was desirable. The Job-by-Application master-slaver scheme saved 23.68% time by using Job-by-average-allocation master-slaver scheme in processing one of 3D seismic data in eastern China.

Key words: prestack depth migration, parallel computing, job-by-application master-slaver scheme, dynamic load balancing, fault-tolerance processing