摘要
针对大规模栅格数据的空间分析,设计并实现了一种基于MapReduce架构的通用地图代数并行计算方法。该方法能将栅格像素矩阵按数据行分割为多个独立的子矩阵,并在并行节点上使用地图代数的四种算子对来自不同矩阵的像素进行分析计算。栅格数据叠加实验结果表明,该方法具备有效性和可靠性,能降低计算对硬件设备的要求,并提升了计算效率。
To solve the problem of spatial analysis of large-scale raster, this paper designs and implements a parallel calculation method of map algebra based on the MapReduce framework. The method can divide a pixel matrix into multiple independent sub-grid matrixes by data row on the MapReduce, implement- ing parallel calculating on the pixels from different matrixes by four map algebra operators. The result of computational experi- ment in resuperposition of raster data shows that the method has validity and reliability and has reduced the requirements in computing hardware, and also enhanced the efficiency in com- putationing greatly.
出处
《测绘地理信息》
2014年第3期51-55,共5页
Journal of Geomatics
基金
国家自然科学基金资助项目(41074025)