基于DEM的分布式并行通视分析算法研究  

Research on Distributed Parallel Visibility Analysis Algorithm Based on DEM

作　　者：张刚[1] 汤国安[1] 宋效东[1] 杨坤[2] 

机构地区：[1]南京师范大学虚拟地理环境教育部重点实验室,江苏南京210023 [2]南京师范大学计算机科学与技术学院,江苏南京210023

出　　处：《地理与地理信息科学》2013年第4期81-85,共5页Geography and Geo-Information Science

基　　金：国家863计划项目(2011AA120303);江苏省高校自然科学研究重大项目(13KJA170001);江苏省普通高校研究生科研创新计划项目(CXZZ12_0393)

摘　　要：从负载均衡的角度详细分析了数据并行的特征,提出一种通用且有效数据可达的DEM数据划分策略。基于该方法设计了分布式并行通视分析算法,以全国90mSRTM作为数据源,对算法的执行效率进行实验,结果表明:基于海量地形数据进行分布式并行通视分析的计算效率与进程数具有一定的关系。另外,算法的并行性能在一定程度上受到地形数据的影响。该文提出的方法有效地提高了海量数据的通视分析算法的计算效率,动态数据划分方案有望为并行环境下地形分析提供新的思路。As the study area extremely expands and the resolution of DEM constantly improves, the amount of inter-visibility analysis computations reveals exponential growth trend, which needs to meet the demand of real-time response requirement of costumes by parallel computing technique. This paper analyzes the characteristics of data parallelism in detail from the point of load balancing and also puts forward a common and reachable data partition strategy for valid DEM. After designing a distribu- ted parallel visibility analysis algorithm by data partition strategy proposed, this paper also verifies the execution efficiency of the algorithm by 90 m SRTM of China. The result shows that the efficiency of parallel visibility analysis algorithm based on massive terrain data has a certain relationship with the number of processes. In addition, the speed-up ratio of parallel algorithm is partly affected by topographic features. In a word, the proposed approach is effectively improved computational efficiency of visibility analysis by massive data, and its dynamic data partitioning strategy will provide fresh ideas of terrain analysis in parallel environment.

关 键 词：DEM通视分析 分布式并行 加速比 

分 类 号：P208[天文地球—地图制图学与地理信息工程]