地理栅格数据并行访问技术研究与实现

发布时间：2018-04-16 09:39

本文选题：地理计算 + 地理栅格数据　；参考：《国防科学技术大学》2012年硕士论文

【摘要】：随着对地观测技术的长足发展，海量地理空间数据日益增长，地理计算也越来越呈现出数据密集型和计算密集型的特点。如今以并行计算集群、多核处理器为代表的新型硬件设施逐渐成为主流，它们为复杂地理计算提供了高性能的计算能力。然而，I/O性能的增长并没有跟上计算性能的增长步伐，从而限制了对海量空间数据的处理能力，成为影响复杂地理计算整体性能的瓶颈，，所以对并行地理计算中I/O问题的研究具有重要的意义和价值。本文围绕“面向地理栅格数据的并行访问技术”这一主题，从以下几个方面展开了深入研究。首先，通过研究地理计算的特点，分析并行处理中数据I/O模式的读写时间、数据划分时间和通信时间之间的关系，建立了地理计算中数据I/O的代价模型，并通过实验分析比较了分发收集式和并行I/O两种代价模型的特点，为设计和实现地理栅格数据处理的并行访问方法提供了理论和实验依据。其次，分析现有并行文件存储模型的特点及其适用条件，针对复杂地理计算任务的特点，设计面向地理栅格数据的并行访问模式，设计了支持多源地理栅格数据的抽象数据模型，并实现了地理栅格数据并行访问接口，为复杂地理计算算法的开发提供支持。再次，基于并行访问模式，提出了面向地理栅格数据的并行I/O框架，设计实现了地理栅格数据并行访问算法，提供了一种复杂地理计算算法中数据并行I/O的编程模式，并通过实验验证了并行I/O模型能够提高地理计算算法的整体性能。最后，论文基于并行集群环境，实现了地理栅格数据并行I/O访问技术在数字地形分析和地图可视化中的应用，并通过实验验证了算法的正确性和高效性。综上所述，通过对面向地理栅格数据的并行访问技术进行研究，有效地改善了地理栅格数据的访问效率，为大幅度提升并行地理计算的整体性能提供技术支撑。
[Abstract]:With the rapid development of Earth observation technology, the massive geospatial data is increasing day by day, and the geographical calculation is more and more data-intensive and computation-intensive.Nowadays, new hardware facilities, represented by parallel computing clusters and multi-core processors, are gradually becoming the mainstream. They provide high performance computing power for complex geographic computing.However, the growth of I / O performance has not kept pace with the increase in computing performance, which limits the processing capacity of massive spatial data and becomes a bottleneck affecting the overall performance of complex geographic computing.Therefore, the study of I / O problem in parallel geographic computing is of great significance and value.This paper focuses on the topic of "parallel access technology for geo-grid data", which is studied from the following aspects.First of all, by studying the characteristics of geographic computing, the paper analyzes the relationship among the reading and writing time, data partition time and communication time of data I / O mode in parallel processing, and establishes the cost model of data I / O in geographic computing.The characteristics of the two cost models of distributed collection and parallel I / O are analyzed and compared, which provides a theoretical and experimental basis for the design and implementation of parallel access methods for geographic grid data processing.Secondly, the characteristics of the existing parallel file storage models and their applicable conditions are analyzed. According to the characteristics of complex geographical computing tasks, a parallel access mode for geographic grid data is designed.The abstract data model supporting multi-source geographic raster data is designed, and the parallel access interface of geographical raster data is implemented, which provides support for the development of complex geographic computing algorithm.Thirdly, based on the parallel access mode, a parallel I / O framework for geo-grid data is proposed, and a parallel access algorithm for geo-raster data is designed and implemented, which provides a programming model of data parallel I / O in complex geographic computing algorithm.Experimental results show that the parallel I / O model can improve the overall performance of the algorithm.Finally, based on the parallel cluster environment, the paper realizes the application of the parallel I / O access technology in the digital terrain analysis and map visualization, and verifies the correctness and efficiency of the algorithm through experiments.To sum up, through the research of the parallel access technology for geographic grid data, the efficiency of geographical grid data access is improved effectively, and the technical support is provided for improving the overall performance of parallel geographic computing.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP338.6

【参考文献】