地理科学 ›› 2015, Vol. 35 ›› Issue (6): 674-682.doi: 10.13249/j.cnki.sgs.2015.06.674

• • 上一篇    下一篇

城郊聚落景观的集聚特征分析方法选择研究

刘焱序(), 王仰麟(), 彭建, 袁媛, 马晶, 魏海   

  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:2014-03-12 修回日期:2014-06-22 出版日期:2015-06-20 发布日期:2015-06-20
  • 作者简介:

    作者简介:刘焱序(1988-),男,陕西西安人,博士研究生,研究方向为综合自然地理与景观生态。E-mail:liuyanxu@pku.edu.cn

  • 基金资助:
    国家自然科学基金重点项目(41330747) 资助

Selection of Different Clustering Algorithms for Settlement Landscape Aggregation in Suburb

Yan-xu LIU(), Yang-lin WANG(), Jian PENG, Yuan YUAN, Jing MA, Hai WEI   

  1. Laboratory for Earth Surface Processes of the Ministry of Education, College of Urban and Environmental Sciences, Peking University, Beijing 100871, China
  • Received:2014-03-12 Revised:2014-06-22 Online:2015-06-20 Published:2015-06-20

摘要:

高度集聚是城郊聚落景观最明显的空间格局特征,是郊区城市化的直观反映。针对目前景观集聚程度研究空间计算方法缺乏筛选、不具可比性等问题,通过4种集聚空间算法在不同角度分析典型区域的城郊聚落景观集聚特征,并在西安市长安区作以简单应用。结果表明:① 景观聚集度适于区分出同类连续大斑块和不同类破碎小斑块,核密度适于宏观上集聚组团的识别,空间关联算法适于空间定位具体要素点的集聚特征,Ripley’s L函数适于识别空间距离以确定搜索半径;② 根据核密度计算结果,从研究区聚落景观中提取出3个大型组团,分别命名为“政府商业中心聚落组团”“沣渭新区聚落组团”和“旅游度假区聚落组团”,其划分模式符合研究区各经济板块的未来发展方向。

关键词: 城郊, 聚落景观, 集聚算法, 尺度

Abstract:

The spatial pattern of high level concentration on settlement landscape in suburban is the most obvious characteristics of urbanization in the outskirts. The present spatial agglomeration methods for landscape are usually used without selection, and they are also lack of comparability. Therefore, this study proposes 4 different analysis of spatial clustering algorithms in a typical area to describe the agglomeration of settlements. And then a simple application in this typical area which realized by one of the 4 analyses is shown. The results reflect that: 1) The landscape aggregation index (CONTAG) is adapted to distinguishing similar continuous large patches and small fragments which are in different types, and the search radius is relatively limited; nuclear density algorithm is adapted to identifying the macro clustering groups, and it’s not usually used in a small spatial scale; spatial correlation algorithm is suitable for spatially locating the specific elements, and the spatial gradual changes are not seen; Ripley's L function is suitable for recognizing different spatial scales to determine the search radius, but the result is rather a chart than a map; 2) Based on the results of the nuclear density algorithm, three large groups are draw on the settlement landscape in the study area, respectively named "government business center settlement groups" "Feng Wei new settlement group" and "Tourism Resort settlement group", the division mode is consistent with the direction of future development for each economic sector in the study area. Landscape planning itself is a subjective process and may not have the only correct process. Quantification of spatial agglomeration is one of the important ways to study the settlement geography from qualitative methods to quantitative methods. Therefore, calculation results may clearly reflect the rationality of the planning, which is currently in need in the landscape research. The quantitative calculation cannot replace the qualitative description, and the diversity of methods leads to a more perfect planning. It goes without saying that more practicable and objective spatial agglomeration algorithms are looked forward to be applied in the settlement landscape studies.

Key words: suburban, settlement landscape, clustering algorithm, scale

中图分类号: 

  • P901