空间异质模型在城市扩展模拟中的应用与评估

以南京江宁区为例，基于遥感影像及其他数据，利用空间扩展模型和地理加权逻辑回归模型对研究区1999-2010年间的城镇用地扩展的影响因素进行有效检验与分析，探索其空间异质性特征.结果表明：1) 空间扩展模型与地理加权逻辑模型分别可解释研究区历史时期57%和64%的城镇用地转变，二者的解释能力在logistic模型基础上 (51%) 得到显著提升，有力地辩驳了全局回归模型的系数稳定性假设；2) 较之于空间扩展模型，地理加权逻辑回归模型更能有效刻画城市增长影响因素的空间非平稳性特征，更能够适用于研究区的城镇用地扩展模拟研究；3) 邻近变量中的DIS2INDU，DIS2ORD及DIS2CENT，邻域变量中的DENURBAN以及政策变量中的DENLANDP是研究区城镇用地扩展的关键驱动因素.

Application and Evaluation of Spatial Heterogeneity Model in the Simulation of Urban Expansion

Based on remote sensing images and GIS data, we applied spatial expansion method and geographically weighted logistic regression (GWLR) model to explore the driving forces of urban expansion in Jiangning District of Nanjing from 1999 to 2010. Special attention was paid to spatial heterogeneity of key factors behind urban growth. The results show that 1) the spatial expansion method and GWLR model explained 57%, 64% of the urban land change of the area under study over the period from 1999 to 2010. The two methods both enhanced explanatory power in comparison with the global regression model (51%) and confirmed the spatially non-stationary process of urban growth; 2) compared with spatial expansion method, GWLR model is more effective in characterizing the spatial variations of the underlying factors of urban growth and therefore, it can be well applied to explore the mechanisms of urban expansion and policy making, as evidenced in this research; and 3) among various explanatory factors, DIS2INDU, DIS2ORD (proximity variables), DIS2CENT (neighborhood variables) and DENURBAN (policy variables) are the key driving factors of urban expansion in the area under investigation.