基于GWLR模型的不同栅格尺度下城市用地空间扩张驱动力研究——以徐州市为例

Driving Force of Urban Land Spatial Expansion Under Different Grid Scales based on GWLR Model——A Case Study of Xuzhou

城市用地扩张驱动力存在空间异质性,从局部视角分析城市用地扩张驱动力对深入认识城市用地扩张过程及城市规划与管理具有重要意义。然而,在采用栅格数据分析城市用地扩张驱动力时,栅格尺度往往会影响驱动力研究的可靠性。因此,本文以徐州市中心城区为例,基于遥感影像解译数据,在分析研究区城市用地扩张特征基础上,分别采用30m×30m、60m×60m、90m×90m、120m×120m和150m×150m分辨率的栅格数据,运用能够反映驱动力空间异质性的地理加权Logistic回归模型(GWLR)对2005~2015年研究区城市用地空间扩张的驱动力进行分析,通过对比不同栅格尺度回归模型拟合效果确定最优栅格尺度,并分析最优栅格尺度下城市用地扩张驱动力。结果表明:GWLR模型可解决全局模型难以克服的空间异质性问题,GWLR模型结果优于Logistic回归模型;研究区城市用地空间扩张驱动力的最佳栅格尺度为90m×90m;2005~2015年研究区城市扩张为外延式扩张和飞地式扩张,主要沿交通干线向对外交通轴线及水系方向扩张;其城市扩张最主要受到高速公路、高速出口、一般公路、区中心和水系的距离的影响,到高速公路的距离对于研究区东部地区影响较强,到区中心和高速公路出口的距离在南部地区影响较强,到一般公路、水系的距离在北部地区影响较强,其余各变量也在不同的区域对城市用地扩张产生不同程度的影响。

The driving forces of urban land expansionare spatially heterogeneous,it is of great significance to analyze the driving force of urban land expansion from a local perspective for further understanding the process of urban land expansion,as well as urban planning and management.However,when the grid data are used to studythe driving forces of urban land expansion,grid scale often affects the results'reliability.Therefore,taking the central city in xuzhou as an example,this paper used the remote sensing image interpretation data to analyze the characteristics of urban land expansion in the study area.Then,the grid data for resolution of 30m×30m, 60m×60m, 90m×90m, 120m×120m and 150m×150m respectively, were used to analyze thedriving forces of urban spatial expansion of the study area from 2005 to 2015 by using the geographic weighted logistic regression model (GWLR), which can reflect the spatial heterogeneity of driving forces. By comparing the results under different grid scale, the optimal grid scale wasdetermined and the driving forces of urban land expansion under the optimal grid scale was analyzed. The results show that the GWLR model can solve the problem of spatial heterogeneity which is impossible to be solved by the global model, and the GWLR model is superior to the logistic regression model. The optimal grid scale of the driving forces of urban land spatial expansion is 90m×90m in the study area. The urban land expanded in types of extension and enclavefrom 2005 to 2015, and new urban expanded along the external transportation lines and water system. Urban expansion is mainly affected by the distancestoexpressway, to expressway exits, to main roads, to district center and to water area. The distance to expressway has a strong impact inthe eastern part of the study area, the distances to district center and to expressway exit havestrong impactsinthe southern part, and the distances to main roads and to water area havestrong impactsinthe northern part. Other variables also have different degrees of influence on urban land expansion in different locations.