A universal parallel scheduling approach to polyline and polygon vector data buffer analysis on conventional GIS platforms
A universal parallel scheduling approach to polyline and polygon vector data buffer analysis on conventional GIS platforms
复制标题
DOI：
10.1111/tgis.12670
复制
发表时间：
2020-08-04
影响因子：
2.4
通讯作者：
Xie, Zhong
中科院分区：
文献类型：
--
作者：
Guo, Mingqiang;Han, Chengde;Xie, Zhong
随着空间数据量的不断增加，传统的矢量数据缓冲区分析算法已经无法满足快速数据处理的需求，因此引入并行计算来加速矢量数据分析。然而，由于传统的数据格式和矢量缓冲区分析算法，这些GIS平台很难采用并行方法进行缓冲区分析。为了解决这个问题，在这项研究中，提出了一种通用的并行调度方法，在传统的GIS平台上的缓冲区分析。首先，确定了影响缓冲区分析计算时间的关键因素。分析了影响因子与缓冲区分析计算时间之间的关系，生成了计算强度转换函数。然后，通过对网格计算强度的部分计算，分别构造了三角形网格和多边形网格的计算强度网格。利用相应的CIG，一个两阶段的自适应空间分解方法的并行缓冲区分析。首先将空间向量数据集的整个域划分为子域，每个子域的计算强度尽可能相等;其次，将子域内的特征均匀分配到并行缓冲区分析任务中，以实现负载均衡。实验表明，本文提出的方法可以有效地评估和空间表示缓冲区分析的计算强度的折线和多边形。与典型的空间自适应分解方法和常规区域分解方法相比，新方法实现了并行缓冲区分析计算强度的更均衡分解，并获得了近线性的加速比。新方法在三个传统的GIS平台上取得了优异的性能，表明我们的方法是一个有效的并行调度方法，矢量数据缓冲区分析的传统GIS平台。
With the increasing volume of spatial data, conventional vector data buffer analysis algorithms cannot meet the demands of fast data processing, so parallel computing is introduced to accelerate vector data analysis. However, it is difficult for these GIS platforms to adopt parallel approaches to conduct buffer analysis due to their conventional data format and vector buffer analysis algorithms. To address the problem, a universal parallel scheduling approach to buffer analysis on conventional GIS platforms is proposed in this study. First, the key impacting factor of buffer analysis computing time is identified. The relationship between the impacting factor and the buffer analysis computing time is analyzed to generate computational intensity transformation functions. Then, computational intensity grids (CIGs) of polyline and polygon are constructed by partially computing the computational intensity of a lattice. Using the corresponding CIGs, a two-stage adaptive spatial decomposition method for parallel buffer analysis is developed. Firstly, the whole domain of the spatial vector dataset is divided into sub-domains, with the computational intensity of each sub-domain being equal as far as possible; secondly, the features are evenly assigned within the sub-domains into parallel buffer analysis tasks for load balance. The experiments demonstrate that the approach presented in this article can effectively evaluate and spatially represent the computational intensity of buffer analysis for polylines and polygons. Compared with typical spatial adaptive decomposition methods and regular domain decomposition methods, the new approach accomplishes greater balanced decomposition of computational intensity for parallel buffer analysis and achieves near-linear speedups. The new approach achieves excellent performance on three conventional GIS platforms, indicating that our approach is an effective parallel scheduling approach to vector data buffer analysis for conventional GIS platforms.