Abstract:

For dealing with the problem of ocean data's interactive visualization, this paper proposes an interactive visualization platform architecture based on cloud computing. It inserts GPU, MPI parallel computing into MapReduce mechanism of Hadoop to realize the parallel processing of large-scale ocean environment data sets, such as data retrieval, data extraction, data interpolation, analysis of characteristics' visualization, so that the massive data's visualization can be processed in the remote interactive way. The experiment results show that the visualization platform architecture can combine MPI, GPU with MapReduce effectively to meet the needs of ocean data's interactive visualization.

Key words: ocean environment information, cloud computing, GPU, visualization