地球信息科学学报 ›› 2017, Vol. 19 ›› Issue (4): 437-446.doi: 10.3724/SP.J.1047.2017.00437

• 地球信息科学理论与方法 •    下一篇

高性能GIS研究进展及评述

左尧1,2(), 王少华1,2,3,*(), 钟耳顺1,3, 蔡文文1,2   

  1. 1. 北京超图软件股份有限公司,北京 100015
    2. 超图地理信息技术研究所,北京 100015
    3. 中国科学院地理科学与资源研究所,北京 100101
  • 收稿日期:2016-08-24 修回日期:2017-01-20 出版日期:2017-04-20 发布日期:2017-04-20
  • 通讯作者: 王少华 E-mail:zuoyao@supermap.com;wangshaohua@supermap.com
  • 作者简介:

    作者简介:左 尧(1989-),男,山西阳泉人,硕士,研究方向为地理信息系统软件技术。E-mail:zuoyao@supermap.com

  • 基金资助:
    国家测绘公益项目(201512015);北京市科技专项(Z151100003615012、Z141101004414011)、中国科学院国防科技创新基金项目(CXJJ-14-M13);中国科学院重点部署项目(KZZD-EW-07-01-001);国家科技支撑计划项目(2013BAC03B00);资源与环境信息系统国家重点实验室自主研究项目(088RAC00YA);北京市优秀人才项目(201500002685XG242);全国博士后国际交流计划(20150081);朝阳区博士后基金项目

Research Progress and Review of High-Performance GIS

ZUO Yao1,2(), WANG Shaohua1,2,3,*(), ZHONG Ershun1,3, CAI Wenwen1,2   

  1. 1. SuperMap Software Co. Ltd., Beijing 100015, China
    2. SuperMap GIS Technology Institute, Beijing 100015, China
    3. Institute of Geographic Sciences and Nature Resources Research, CAS, Beijing 100101, China
  • Received:2016-08-24 Revised:2017-01-20 Online:2017-04-20 Published:2017-04-20
  • Contact: WANG Shaohua E-mail:zuoyao@supermap.com;wangshaohua@supermap.com

摘要:

互联网技术的发展使地理信息技术得到了前所未有的发展和应用,地理信息计算呈现出计算速度快、运行效率高、应用多样化的发展特征。而随着计算机硬件性能飞速提升,传统的GIS数据处理方式并不能与之匹配,各种缺陷与弊端逐渐显现,亟待更高效的数据处理方式。目前,以并行集群计算技术和分布式网络技术为代表的高性能计算的出现,为这些问题的解决带来了新思路,并逐渐发展形成了新一代的多核并行高性能计算系统。当前,如何利用新型硬件体系结构带来的计算能力,研究新一代高性能GIS计算系统,解决现在所面临的时空数据密集和计算密集问题成为重要挑战。高性能计算是基于一组或几组计算机系统组成的集群,通过网络连接组成超级计算系统以加强数据处理、分析计算性能的一种技术。在实际应用中,逐渐形成Hadoop,Spark和Storm 3大主流分布式高性能计算系统,它们三者各具优缺点。本文从高性能GIS算法、并行GIS计算、内存计算和众核计算4个方面梳理、归纳总结了高性能GIS的技术体系,分析了每类高性能GIS技术特征,综合分析、评述了近年来高性能GIS的研究进展,并对高性能GIS未来发展进行展望,为更完备、高效的高性能GIS体系的建立、发展和应用提供参考。今后,并行GIS计算、高性能计算模式和分布式存储仍然是GIS技术领域发展的重要方向,通过高性能GIS系统可有效地解决时空数据密集、计算密集和网络通讯密集等问题,大大提升GIS地理分析效率。

关键词: 高性能GIS, 高性能GIS算法, 并行GIS计算, 内存计算, 众核计算, GIS云计算

Abstract:

The development of Internet technology has given birth to the explosive growth of various information in recent years. The traditional data processing method cannot matched with the rapidly improving performance of computer hardware and more efficient methods are needed to process the numerous data. High performance computing technologies, including Parallel cluster computing technology and distributed network technology, bring hints to the solution of these problems. In practice, there are three major distributed computing systems, namely Hadoop, Spark and Storm. Hadoop improves computational performance by introducing MapReduce distributed computing framework, while Spark make full use of computer memory to store data based on Resilient Distributed Datasets(RDD), which has a more rapid reading and writing functions of data . The Storm does not directly collect data. It realizes the data transmission and processing using network nodes. Nowadays, how to take advantage of the improvement of computational performance brought by the development of new hardware architecture to solve the long existing data intensive, computational intensive and communication intensive problems has become a topical issue in the field of GIS studies. In this paper, reviewing current research progress of high performance GIS, we examine and discuss about the algorithm of high performance GIS, parallel GIS computing, memory computing and core computing and give some prospective on the future development of high performance GIS, which provide a reference for the development of high performance GIS system. In addition, the development of the Internet technology and cloud computing is continuously boosting the popularity of GIS cloud computing and big data technology. In this context, domestic and foreign GIS platform vendors have launched their own cloud GIS platform, such as ArcGIS10.4 developed by ESRI and SuperMap 8C by SuperMap, to give support to cross-platform, parallel computing, 64-bit computing, distributed systems and other technologies.

Key words: high performance GIS, high performance GIS algorithm, parallel GIS computing, memory computing, core computing, GIS cloud computing