地理科学进展 ›› 2012, Vol. 31 ›› Issue (8): 1024-1031.doi: 10.11820/dlkxjz.2012.08.005

• 城市地理 • 上一篇    下一篇

城市家庭居住地选址的空间异质性分析——以美国佛罗里达州橙县为例

宇林军1,2, 孙丹峰2, 彭仲仁3, 张健4   

  1. 1. 中国科学院遥感应用研究所, 北京100101;
    2. 中国农业大学资源与环境学院, 北京100193;
    3. 美国佛罗里达大学区域与城市规划系, 美国盖恩斯威尔32611-5706;
    4. 中国科学院南京地理与湖泊研究所, 南京210008
  • 收稿日期:2012-01-01 修回日期:2012-03-01 出版日期:2012-08-25 发布日期:2012-08-25
  • 通讯作者: 孙丹峰(1971-),男,博士生导师,教授,主要研究方向为土地系统分析理论与方法。E-mail:sundf@cau.edu.cn E-mail:sundf@cau.edu.cn
  • 作者简介:宇林军(1981-),男,博士,主要从事基于3DGIS技术的土地利用和城市发展研究。E-mail:yulinjun81@gmail.com
  • 基金资助:

    国家自然科学基金重点项目(41130526)。

Spatial Heterogeneity Analysis of Household Residential Location Choice: A Case Study in Orange County, Florida, USA

YU Linjun1,2, Sun Danfeng2, PENG Zhongren3, ZHANG Jian4   

  1. 1. Institute of Remote Sensing Applications, CAS, Beijing 100101, China;
    2. College of Resources and Environmental Sciences, China Agriculture University, Beijing 100193, China;
    3. Department of Urban and Regional Planning, University of Florida, Gainesville, FL 32611-5706, USA;
    4. Nanjing Institute of Geography and Limnology, CAS, Nanjing 210008, China
  • Received:2012-01-01 Revised:2012-03-01 Online:2012-08-25 Published:2012-08-25

摘要: 家庭选择居住地的行为天然具有空间性, 因而空间异质性效应是家庭居住地选址建模不可忽视的因素。传统的居住地选址模型基于空间一致性假设, 即假设影响因素对家庭的居住地选择行为的影响在空间上一致, 因而忽略了空间异质性效应。基于多项Logit 模型构建了居住地选址模型, 并在两个空间尺度和5 个子区域中分别应用该模型, 来反映影响因素的影响作用在空间上的变化。以美国佛罗里达州橙县家庭选址为例进行实证研究, 结果表明:家庭居住地选址行为的影响因素在不同的空间位置和空间尺度上具有不同的作用, 因而存在显著的空间异质性。尽管以美国地区为例, 但所得结论对国内案例区研究同样具有借鉴意义。

关键词: 多项Logit模型, 家庭居住地选址, 空间同质性, 空间效应, 空间异质性, 美国佛罗里达州

Abstract: Households’behaviors of choosing residential locations are spatial in nature. Therefore, spatial effects cannot be disregarded when molding household residential location choice. Spatial correlation and spatial heterogeneity are two aspects of spatial effect. Several studies regarding spatial effect in household location choice modeling have been conducted. However, the existing studies mainly focused on spatial autocorrelation effect in household location choices, where the spatial heterogeneity effect was ignored. Conventional residential location choice models are typically based on the assumption of spatial homogeneity, assuming that the impacts of an influencing factor are constant across space. Therefore, spatial heterogeneity effect is ignored in traditional residential location choice models. In this study, Orange County, FL, USA was used as a case study. Census Block Groups (CBG) in the study area were grouped into five sets, where hot spot analysis (Getis-Ord Gi*) in ArcGIS was employed for location alternatives grouping because it enables spatial locations to be geographically grouped based on one non-spatial attribute. Finally, we used seven multinomial logit model based residential location choices models in the five sub-regions of the study area and at two spatial scales (CBG scale and CBG set scale), respectively. The estimation results provided evidences that both location-dependent and scale-dependent variations exist in the impacts of the influencing factors on household residential location choice. At the higher scale (CBG set scale), zonal attributes are the main influencing factors. At the lower scale (CBG scale), however, the interaction variables between household characters and zonal attributes play a leading role. Variable effects vary significantly across regions at the CBG scale. For example, the interactive variable between zonal Black population density and the Black ethnicity dummy variable of a household is only significant in two regions out of all the five regions. Therefore, spatial heterogeneity is a significant characteristic of the impacts of influencing factors on household residential location choice. Although the study area is located in USA, the conclusion is meaningful for future studies which use regions in China as cases.

Key words: Florida, USA, household location choice, multinomial logit model, spatial effect, spatial heterogeneity, spatial homogeneity