基于CA-SVM模型的莆田市城市扩张特征

【目的】利用城市用地扩张模型分析城市用地系统特征、演变过程与机制。【方法】利用支持向量机的非线性特性来定义元胞自动机的非线性转换规则,计算出各个元胞与城市中心、商业中心、大小道路、地质地貌控制边界等距离的区域空间变量,并求出这些变量对每个元胞转变为城市概率的贡献,建立基于CA-SVM的城市扩张模拟模型,以揭示莆田市城市用地扩张规律。【结果】构建的CA-SVM模型在定义分类超平面的同时考虑最小化结构风险和最大距离间隔,使用该模型在莆田市城市扩张模拟中的逐点对比精度达到83%以上。莆田市城市用地扩张的主要方向为东北、东南和南部沿海地区,且呈现出沿交通主干道的轴向扩张模式,在港口和码头等沿海交通网络节点处形成城市扩张的跳跃式布局。【结论】构建的CA-SVM模型可定量地反映各空间要素在研究区城市格局所起的作用,继而完成分析城市发展的主要规律。

Characteristic of urban expansion based on CA-SVM model

Abstract: 【Objective】The analysis of urban expansion simulations is an effective way to reveal the characteristics, mechanisms, and evolutionary processes of urban land use systems. 【Method】Nonlinear transformation rules defined by nonlinear classification of Support Vector Machine(SVM)for the Cellular Automata(CA)are used to calculate regional-spatial variables, such as the distance between cellular and the center of Putian City, business centers, different types of roads, and geological and geomorphological control boundaries. The contribution of these variables toward the transformation of each cellular to the urban development is then calculated. A CA-SVM-based urban expansion simulation model is established to reveal the trend of urban land expansion in Putian City. 【Result】The classification hyperplane in the SVM-CA model is defined considering the minimization of structural risk and maximum distance interval. Results indicate that the accuracy of the point-by-point comparison of this model to simulate the urban expansion is up to 83%. Urban expansion is mainly concentrated in the northeast, southeast, and southern coastal regions of Putian City. The city also exhibits a pattern of axial expansion along the main roads. A spatial salutatory layout can be seen at traffic network nodes along the coastal line, harbors and wharfs, etc. 【Conclusion】CA-SVM model can reflect quantitatively the role of each spatial element in the urban pattern of the study area, and further to analyze the major rule of city.