摘要
针对当前大规模全球科学数据可视化中存在的单机可视化数据量有限、从底层开发并行可视化系统难度大等问题,该文基于分布式环境和VisIt,提出了一种简便、开放而又有效的大规模全球科学数据可视化方法。介绍了VisIt的体系结构及运行机制,给出了自定义数据的并行可视化方法;并基于NCEP数据集及全球空间格网,在小规模集群环境下实现了小粒度适应性球体退化八叉树格网(SDOG)下的全球大气温度场的并行可视化。VisIt的并行可视化性能测试结果表明:通过增加计算节点,VisIt能有效摆脱传统单机可视化对数据量的限制,可实现大规模全球科学数据的并行可视化。
Visualization is essential for the scenario representation of massive global scientific data, and is of great importance for global change researches and Earth System Science. Referring to the deficiency and week-capability of current visualization methods with single PC on global scientific data and to the great difficulty to develop initially a parallel visualization system, this paper proposed a Vislt-based convenient, open and efficient visualization method under the usual distributed computer environment. The architecture of Visit was introduced, and the parallel visualization principle of Visit was analyzed based on the cluster structures. The Visit-based methods for parallel visualization of global scientific data were proposed. Finally, based on Global Spatial Grid and NECP datasets, taken the temperature field data of global atmosphere being a case, this paper realized its visualization with adaptive Spheroid Degenerated Octree Grid. Simultaneously, experiments on the visual ability of Visit were carried out on the laboratory cluster. The results showed that Visit can efficiently get rid of the restraint on data size of the single PC, and the relationship between data size and compute node was analyzed. The method introduced in this paper is of promising future.
出处
《地理与地理信息科学》
CSCD
北大核心
2012年第1期24-28,共5页
Geography and Geo-Information Science
基金
国家自然科学基金重点项目(40930104)
国家863子课题(2011AA120302)
关键词
VISIT
全球科学数据
全球空间格网(GSG)
并行可视化
适应性SDOG
大气温度场
Visit
large scale global scientific data
Global Spatial Grid(GSG)
parallel visualization
adaptive Spheroid Degenera- ted Octree Grid(SDOG)
atmosphere temperature field