Construction And Application Of Seismic Data Processing Grid Platform

The amount of seismic data acquisition and computing is growing rapidly with the development and application of the new seismic technology in the field of seismic exploration. The purpose of this dissertation is to make full use of the high-performance resources of the industry and increase the scale of seismic data parallel processing by using grid technology. The grid platform for seismic data processing solves the key problems in wide area network environment, including heterogeneous resource sharing and cooperation, distributed parallel task scheduling,and distributed data management.Firstly, the characteristics of seismic data structure and data processing are analysed. The result shows that common shotpoint gather based wave equation prestack depth migration is coarse-grained task. Globus Toolkit4 based seismic data processing grid architecture was established. Then the mechanism of standard grid service and message passing between grid nodes was studied.Distributed parallel processing of seimic data need to select multiple grid resources. Grid resource select framework was presented. A multi-objective optimization mathematical model was established to select multiple grid resources. And the model was solved by using Genetic Algorithm. Resource monitoring component provided grid resources information for selecting resources and helped grid users monitor the grid resources.A three layer scheduling model is presented according to the coarse-grained data processing task. The application level scheduler was responsible for data partitioning, scheduling and automatic parallel processing. Divided load theory was used in data partitioning. A probing based data partition strategy was designed and implemented. The result of Fourier finite difference prestack depth migration of Marmousi model validated the PDP strategy on the condition that resources list was given.A distributed seismic data management framework was presented to support parallel processing in grid environment. Metadata model for distributed grid data and higher data management services including data replication, data location and deletion services, and portal for data management were designed and implemented.Finally, application grid platform was deployed in a production environment. A kirchoff integral pre stack time migration program was migrated to the grid platform. The result of processing production data shows the validity, stability and efficiency of the grid platform. The application grid platform has the advantage of being able to realize the sharing and collaboration of distributed, heterogeneous high performance computing resources. The grid platform could be used to integrate the industry high performance esources, to improve resource utilization and further expand the scale of seismic data parallel processing.