A Multiscale Method Based On Multi-source Data For Urban Spatial Structure Analysis And Its Application

Profound changes have taken place in urban scale and spatial structure of different cities during China's rapid urbanization.It has become a difficult and urgent issue in the field of geographic information science to scientifically and accurately analyze the present situation and evolution law of China's urban spatial structure at different spatial scale based on multi-source data.All of these analysis can enrich the research content of spatial analysis and provide reference for the spatial planning of land space.Taking physical urban area as basic study unit,based on the support by geographic information system and remote sensing technology,we improved methods for the identification of multi-scale urban spatial structure in China by using multi-source data such as nighttime light data,Tencent migration data,points of interest,urban road network data and Sina micro-blog check-in data.Firstly,the long time sequence and consistent DMSP-OLS nighttime light data were generated by using Normalized Difference Vegetation Index(NDVI)and Sigmoid model.We improved the objectedbased thresholding algorithm to extract the built-up area based on the spatiotemporally enhancing time-series nighttime light data.Then,the urban extents of physical urban area were identified.Secondly,Infomap community detection algorithm and K-shell decomposition were combined to obtain the spatial scope of different urban agglomerations in China based on human migration that tracked by Tencent.Thirdly,the rank-size model was used to detect morphological polycentricity in Chinese urban system and different urban agglomerations on the basis of the identified physical urban areas.Fourthly,relying on the concept and method of the physical urban area,we presented a reliable and efficient method that used multi-source geospatial big data,including Points of Interest,Weibo check-in maps and OpenStreetMap road network,to identify main center and subcenters of 294 Chinese cities at metropolitan area and central city scale.Fifthly,a synthetic polycentric index was used to analyze the development and spatial characteristics of the spatial structure of Chinese cities at metropolitan area and central city scale.Finally,suggestions for the future development of spatial structures at various scales were proposed based on the comprehensive analysis of the development of urban spatial structure in China.The main results and conclusions of this thesis are as follows:(1)The method of spatial structure identification of different scales was improved and a multi-scale research workflow of urban spatial structure was constructed with the support of multisource data and related technologies,which enriches the content and methods of spatial analysis research.(2)Normalized Difference Vegetation Index(NDVI)and Sigmoid model were used to improve the accuracy of nighttime light data and estimate the missing data in 2014 and 2015.The improved objected-based thresholding method can ensure the time-series continuity of the extracted physical urban area.(3)With the turning point in 2010,the polycentric urban development at national scale has gradually improved.The polycentric characteristics of different urban agglomerations were obvious,and the equilibrium of different centers within the urban agglomeration was enhanced.The spatial structure of urban agglomerations is generally developing toward a polycentric urban structure.(4)The polycentricity of spatial structure at metropolitan area scale was relatively low.Although some cities were polycentric nowadays with the help of new towns,monocentricity was more common than polycentricity at metropolitan area scale.The main centers still played a key role in metropolitan area nowadays.General dispersion was the most common urban structure at central city scale,the internal development of most Chinese central cities was balanced.However,some of Chinese megacities showed mature polycentric spatial structure at central city scale.