基于空间突变特征的城市边缘区提取方法

城市边缘区位于城市与乡村之间的过渡交接地带,既是城市扩张的前沿,也是城乡建设和用地置换中最具活力的地区。准确识别城市边缘区的空间范围一直是城市空间结构研究的核心问题,有助于从城乡对比的角度来衡量城市化程度。本文以城市边缘区的空间突变特征为切入点,基于格网尺度评价构建基于多源数据的城市边缘区特征识别指标,然后采用小波变换检测进行特征值突变点群的识别,并利用基于Delaunay的自动边长阈值的边界提取算法识别突变点群内外边界,从而实现了一种基于空间突变特征的城市边缘区提取方法。最后,以江阴市作为研究区进行了实证分析,并将本文方法提取结果与通过信息熵模型和土地利用程度综合指数模型提取城市边缘区的经典方法提取结果进行对比,经典方法所提取的城市边缘区结果更为混乱分散,而本文结果更为完整客观。将本文方法提取结果与使用土地利用数据与行政区划统计年鉴数据构建城市边缘区识别指标进行突变检测的不同指标方法提取结果进行对比,二者重叠度达88.03%,体现了本文方法的正确性,而从局部细节分析来看,本文结果更符合实际情况。为了更好地验证本文方法的有效性,利用景观格局指数对本文方法和不同指标方法提取的城市建成区、城市边缘区和乡村腹地的范围进行检验：从斑块类型层级指数分析, 2种方法划定的区域都具有典型的空间特征;而从景观层级指数分析,本文识别出的边缘区所计算的斑块密度、最大斑块指数、景观分离度、景观破碎化指数和香农多样性指数均高于对比方法,而蔓延度和香农均匀度均低于对比方法,说明本文识别的城市边缘区范围内景观破碎化程度和异质性更高,景观分布不均匀,社会经济条件更复杂,从而证明了本方法的有效性,尤其适合于非闭合环状的城市边缘区的提取。

Identification Method of Urban Fringe Area based on Spatial Mutation Characteristics

As a transition zone between the city and the countryside, the urban fringe area is not only a geographical space affected by both of the regions, but also an area shrouded in conflicts of interest. The rapid development of urbanization witnesses tremendous changes the urban spatial structure is undergoing. Therefore, studying the spatial scope of the urban fringe area is conducive to the assessment of the current situation of urban development, and is further significant for policy formulation, population management, and resource allocation in the urban fringe area. Thanks to the development of remote sensing and geographic information technology, the types, quality, and accuracy of geospatial data that are applied to depict the characteristics of the urban fringe area have been significantly enhanced. Considering this, this paper takes the spatial mutation characteristics of the urban fringe area as a starting point, and a method for mutation point groups detection, combining multi-indices fusion and wavelet transform, is adopted to distinguish the spatial extent of urban fringe area based on the optimal results by grid-scale evaluation. And then we use the Delaunay triangulation automatic edge length threshold to extract the boundary of the mutation point groups and to obtain the spatial range of the urban fringe area. Empirical analysis is conducted taking Jiangyin City as the research area. The main experimental steps are as follow: firstly, several basic data are selected, containing land use data, road data, night light data, and service-oriented POI data according to the analysis of the characteristics of the urban fringe area. These multi-source data are then standardized based on grids and entropy weighting method for weight determination. In this way, the eigenvalues of the discriminant index of the urban fringe area are calculated. Secondly, based on the spatial mutation characteristics, the wavelet transform is employed to extract the mutation point groups on the eigenvalue sequence, which can effectively improve the discrimination accuracy of the mutation point group. It is suitable for the non-closed circular urban fringe area and for avoiding the influence of human subjective factors. Then, the algorithm based on the Delaunay triangulation automatic edge length threshold is utilized to extract the boundary of the mutation point groups and to obtain the spatial range of the urban fringe area, which can provide a reference for optimizing the urban spatial layout. Finally, the extraction results in this paper are compared with those obtained by the classical methods, like the information entropy model and the comprehensive index model of land use degree. It is apparent to see that the results from classical methods are more chaotic and scattered, while the results in this paper are more complete and objective. Comparing the extraction results of this paper with the extraction results of different index methods employing land use data and administrative division statistical yearbook data to construct urban fringe identification indicators for mutation detection, this study discovers the overlap between them is 88.03%, which testifies the factualness of this method. In terms of the analysis of local details, the results of this paper are more in line with the actual situation. To verify the effectiveness of the method proposed in this paper, the landscape pattern indices are adopted to test the range of urban built-up area, urban fringe area and rural hinterland extracted by the method of this paper and the other different index method respectively. Considering the patch class size landscape pattern indices, the areas delineated by two methods are following the spatial characteristics. Meanwhile, the value of data calculated in the urban fringe area identified in this paper is all higher than the counterpart method given the landscape size landscape pattern indices, the patch density, maximum patch index, landscape separation degree, landscape fragmentation index, and Shannon diversity index. The spread and Shannon uniformity, however, are both lower than the comparison method. It can be indicated that the fragmentation and heterogeneity of the landscape in the urban fringe area identified in this paper is higher, the landscape distribution is uneven, and the socio-economic conditions are more complex, thus proving the effectiveness of this method is especially suitable for the extraction of non-closed circular urban fringe area.