Urban Functional District Extraction And Analysis From Multi-source Data

Urban functional districts refer to residential land,commercial land,industrial land,public management and public service facilities as well as other functional zones gradually formed in the process of urban development.The unified coordination and reasonable layout of urban functional districts contribute to improving urban land use efficiency,optimizing the resource allocation,realizing balanced urban development,and improving the overall strength of the city.Identifying different types of functional districts and studying their spatial distribution patterns and interaction laws are of great significance for managing the urban spatial structure and establishing the scientific and reasonable policy of urban planning.Traditional urban functional district extraxtion and analysis mainly used land use maps,questionnaire surveys and other data,resulting in the singleness of data and insufficient analysis,and considering more about urban physical attributes than urban social attributes.Due to the limitation of the data source and data scale,the related research in the early stage did not make full use of the present spatiotemporal data analysis technology,and can not meet the needs of studying urban functional district in the era of big spatiotemporal data.To solve the problems mentioned above,this study combined physical attributes and social attributes to extract urban functional districts based on multi-source data,and made full use of big spatiotemporal data analysis techniques auch as trajectory mining to provide technical support for the healthy urban development and reasonable palnning.This study used multi-source data,such as high-spatial-resolution remote sensing imagery,building contour data,Open Street Map(OSM)data,taxi trajectory data,and point of interest(POIs),to mine urban physical attribute information and socioecomomic attribute information for extracting the functional attributes of urban local regions and analyzing the interaction and distribution patterns of urban elements.The main research contributions are as follows:(1)Analysis unit construction to unify the research dimension of multi-source dataBased on the OSM road network,this study used taxi trajectory data to update urban road information,and proposed a weighted mean head/tail breaks method to classify pixels of taxi trajectory rasterize image.Then referring to remote sensing imagery and online maps,this study built the analysis unit to solve the dimentional disunity of remote sensing imagery analysis and social sensing data analysis.(2)Urban land use classification from high-spatial-resolution remote sensing imageryHigh-spatial remote sensing data reflects the physical attributes of urban surface,including spectral features,shape features,texture features,spatial geometry features,etc.This study used the aforementioned features to classify urban land cover.Then based on the classification results,landscape patterns of each analysis units were mined.Assisted by the building function,urban land use classification was accomplished.(3)Urban functional district identification and spatial interaction analysis based on taxi trajectory dataThe pick-up and drop-off points of taxi trajectory provide massive continuous socioeconomic activity information of urban residents,and implies socioeconomic attribute information of urban areas.This paper used pick-up and drop-off data to analyze spatiotemporal patterns of residents’ daily activities and extracted multiple spatiotemporal features to identify urban functional districts based on the best combination of features.Based on the identification results,this study used taxi origin/destination flows to visualize the spatial interaction between functional districts,in the typical time period during working days and weekends.Taking Xudong Commercial District and Zhongnan Road Commercial District as examples,this study compared the spatial interaction intensity and distance between them and other functional districts.(4)Spatial distribution pattern analysis from POIsPOIs is the external representation of current land use in each urban functional district.The density distribution and area distribution of POIs in each functional districts have been analyzed to explore its land use distribution.