地理研究 ›› 2011, Vol. 30 ›› Issue (9): 1648-1659.doi: 10.11821/yj2011090009

• 地表过程研究 • 上一篇    下一篇

基于GIS与SOFM网络的中国综合自然区划

黄姣, 高阳, 赵志强, 李双成   

  1. 北京大学城市与环境学院,地表过程分析与模拟教育部重点实验室,北京 100871
  • 收稿日期:2010-10-19 修回日期:2011-04-28 出版日期:2011-09-20 发布日期:2011-09-20
  • 通讯作者: 李双成 (1961-), 男, 河北平山人, 教授, 博士,主要从事地表格局与过程复杂性计算和模拟。 E-mail: scli@urban.pku.edu.cn
  • 作者简介:黄姣 (1986-), 女, 四川绵阳人, 研究生,主要从事地表格局与过程复杂性计算和模拟。 E-mail : facile86@163.com
  • 基金资助:

    国家自然科学基金(40771001、40971052)

Comprehensive physiographic regionalization of China using GIS and SOFM neural network

HUANG Jiao, GAO Yang, ZHAO Zhi-qiang, LI Shuang-cheng   

  1. College of Urban and Environmental Sciences, Laboratory for Earth Surface Processes, Peking University, Beijing 100871, China
  • Received:2010-10-19 Revised:2011-04-28 Online:2011-09-20 Published:2011-09-20

摘要: 综合自然区划一直是中国地理学界的研究核心和热点之一,已有大量的区划方案应用于指导社会生产实践或教学活动中。已有的区划工作主要是基于传统地域划分研究范式,大多采用专家经验集成方法和技术。专家经验与知识的主观性和个体差异性会对区划方案的科学性和客观性产生影响。为了弥补传统区划范式的不足,丰富区划的方法与途径,本文探讨了自组织映射网络(SOFM)在综合自然区划研究中的应用。在GIS技术支持下,秉承传统区划的研究成果,采用温度带、干湿地区和自然区的三级单位系统,分别选取相应的温度指标、水分指标和地形、植被指标,构建和运行不同层次的SOFM网络,将中国陆地区域划分为8个温度带,17个干湿地区和43个自然区,并将区划方案与传统区划方案进行了对比和检验。结果表明,使用基于GIS平台的SOFM网络进行综合自然区划具有划分层次明显、区域分割清晰、客观性强等优点,是对传统区划方法的有力补充和拓展。

关键词: 综合自然区划, GIS, SOFM网络, 中国

Abstract: Comprehensive physiographic regionalization has long been a core issue of physical geography in China. A great number of regionalization themes have been developed and applied as guidelines for regional development and geography teaching. However, these themes mainly use the traditional expertise-experiences-based regionalization methodology, which probably make themselves unreliable due to certain prejudices and different knowledge backgrounds of each individual. In order to overcome this obstacle, and to enrich regionalization research theoretically and methodologically, this paper tries to apply SOFM neural network to the regionalization. Supported by GIS technology and following the traditional three-level-strategy, we construct and operate SOFM neural networks at each level, using temperature factors, moisture factors and supplement factors respectively. Finally, we divide Chinese mainland into 8 temperature zones, 17 moisture regions and 43 natural sub-regions, then compare this scheme with those based on traditional methods. The result shows that based on GIS platform, applying SOFM neural network into comprehensive physiographic regionalization has significant advantages, which is an important supplement and development to traditional regionalization paradigm.

Key words: comprehensive physiographic regionalization, GIS, SOFM neural network, China