基于自组织映射的北京主体功能区识别研究

主体功能区的划分对于我国构建高效、协调、可持续发展的国土开发格局具有重要意义。本文利用自然环境(NDVI、坡度等)和人类社会活动(夜间灯光、POI点等)两类数据,构建了12个因子,采用Self-Organizing Map(SOM)神经网络和层次聚类法,对北京城市功能区进行分类,并结合经典功能区分类和Google Earth目视验证进行了主体功能区的划分。结果表明,由于SOM神经网络能够保持输入数据的拓扑关系及非线性的特征,基于SOM神经网络的层次聚类法具有较强的通用性及抗干扰能力,对于多源数据的城市用地分类有较大的应用潜力。同时表明,北京市主体功能区由中心城区向外呈优化开发、重点开发、限制开发的环形分布特征,各区之间呈"点-轴-面"协调发展的空间结构特征。

Identification Research of Major Functional Areas Based on Self-organizing Map Method of Beijing

Identifying the major functional area is significant to construct an efficient and sustainable development land development pattern in China. In this paper, by natural environment and human activities datasets, we extracted 12 factors and constructed the i-dentification model to identify the major functional areas of Beijing, and verified the result by Google satellite maps through visual in-terpretation method. The research shows that, the identification model has strong versatility and anti-interference capacity, due to SOM neural network' s advantage of maintaining the topological relationship and better performance in dealing with nonlinear datasets, which has great potential for urban land use classification based on multi-source datasets. Meanwhile, the distribution of the major functional areas in Beijing have a circular characteristic, from the center to the outside of the city is the optimized development area, the key development area and the restricted development area respectively. In addition, there is a spatial structure of "point-axis-plane" coordinated development among the functional areas.