Research On The Identification Of Large Urban Functional Areas Considering Spatial Semantic Characteristics

The rapid development of urbanization in China has an irreversible impact on the land use and so on in the regional scope,urbanization leads to the rapid changes of land use,population density and environmental pollution,under the influence of frequent transactions of modern social and economic activities,rapid human mobility and rapid development of transportation,the urban spatial structure becomes more and more complex,which brings great challenges to urban planning and resource allocation,and the reasonable layout of urban functions is of The rational layout of urban functions is of great significance to urban development,and the correct identification of urban functional areas is particularly important.Whether from single data to data application of spatial information,from spatial information into textual information and additional feature vectors to make the zoning nature of spatial distribution characteristics more significant,research on urban functional zoning has made greater progress.To address the limitations of urban functional zones based on POI data,firstly,the traditional urban functional zones are mostly divided manually,which has the problems of slow update frequency and high economic cost;secondly,the research has explored the methods of automatic urban functional zone division using remote sensing images and Point of Interest(POI)data,but the classification accuracy still needs to be further improved;thirdly,the research results have been used to further improve the classification accuracy.Third,some research results only transformed the name text information in POI into feature vectors,but did not consider the role of their spatial distribution characteristics on zoning.Therefore,in this study,five cities,Beijing,Chengdu,Guangzhou,Shanghai and Wuhan,are used as the study area,and the study area is dissected by 1000 m,500m and 200 m grids,respectively,to obtain the grids data at different resolutions,and to build a recognition framework for large-area urban functional area classification that takes into account spatial semantic features and the GeoHash+Doc2Vec model: first Firstly,the POI and AOI data are obtained by Python crawling technique,and the POI is used as the data source for extracting spatial and semantic features,and the AOI is used as the data source for validation samples.Secondly,the spatial feature vectors of POI are extracted by GeoHash algorithm,and the high-dimensional feature semantic vectors of POI type are extracted by PV-DM model,and the spatial and semantic vectors are fused and weighted to obtain the feature vectors of each study unit,and then the study units are clustered and analyzed based on Random Forest algorithm,and the final results of urban functional area identification are obtained.A total of six types of functional areas are identified,including residential land,public administration and public service facilities,commercial and service land,industrial land,road and transportation facilities,and green space and square land.Finally,the details of different functional areas in each city are further explored.The research results show that the overall accuracy of the urban functional area identification framework based on GeoHash+DocVec model proposed in this thesis reaches 71.82%,the Kappa coefficient is 0.62,which indicates that the method is feasible and can effectively identify the urban functional areas.The method of this study has good repeatability and generalizability,and the detailed urban functional area identification and delineation map will provide more support for urban planning and environmental decision-making,as well as help to conduct quantitative and objective research on urban functional areas.