A Study Of The Urban Functional Zoning Method From The Multi-source Heterogeneous Data Fusion Analysis Perspective

Various urban elements are laid out according to their different functions.Some urban elements are concentrated in certain urban areas,which form different urban functional regions.Cities are complex systems composed of multiple layers of activity and interactions across various urban domains.As the essential feature of a city,urban function emphasizes the order of internal space and the connection with external space,which is the driving force of urban development.With the interaction of the “CitySpace-Human” ternary structure,cities are ever increasing and developing each day,and changes in cities bring about updating of the data source types and data acquisition methods.Many studies have demonstrated the feasibility of using multi-source databased methods to discover urban functional regions.Urban research faces difficulty in achieving the completeness of data due to the limitation of urban data access and copyrights.How to best improve data descriptions through heterogeneous data fusion is the focus and main difficulty in the field of urban data mining.Reasonably designing urban functional units is the basis of studying urban functional zoning.The existing functional units lack interpretability because of their single purpose,being well regularized and ignoring spatial-temporal associations.It is therefore important to guarantee good urban functional zoning results with appropriate units and granularity.In addition,most of the existing research places more emphasis on the static characteristics of urban functions,reflecting the state of a city at a certain moment,and there is a lack of empirical research on functional connectivity and spatiotemporal interaction.It can be seen that research on the mixed function layout of urban structures and the spatial-temporal variation characteristics of urban functions are future research trends in urban science.From the perspective of multi-source heterogeneous data fusion analysis,this paper comprehensively uses graph model,association rule mining,topic models,regional eigenvector,and so on,to carry out the research of urban functions from three aspects: basic functional units,mixed function layout and urban functional zoning methods.The main content of our study and technological innovations are as follows.(1)The main research theories and technical means of urban functional zoning were summarized.On this basis,this paper detailed three functional zoning methods: clustering-based methods,image object-recognition-based methods and topic model methods.The theoretical and operational aspects of the three methods were analysed and discussed from the aspects of the data used to drive them,regional feature vectors,and their advantages and disadvantages.(2)In the aspect of constructing functional units,this paper proposed a novel parcel grouping method to access this issue,which considers both the urban morphology and the urban functional connectivity.Infiltration behaviors of urban components provides a basis for exploring the correlation between morphology mechanism and functional connectivity of urban.We measured the infiltration behaviors among adjacent parcels and concluded that the occurrence of infiltration behaviors often appears in the form of groups,which indicated the practical significance of parcel grouping.Our method employed two parcel morphology indicators–the similarity of the line segments and the compactness of the distribution.The line segment similarity was used to establish the adjacent relationship among parcels and the compactness was used to optimize the grouping result to obtain a satisfactory visual expression.In our study,constrained Delaunay triangulation,Hausdorff distance,and graph theory were employed to construct the proximity,delineate the parcel adjacency matrix,and implement the grouping of parcels.We applied this method for grouping urban parcel data of Beijing and verified the rationality of grouping results based on the quantified results of infiltration behaviors.Our method proved to take a good account of infiltration behaviors and satisfied human cognition,compared with a k-means++ method.(3)In terms of urban function composition,this study discussed the characteristics of mixed function layouts and provided a reasonable description of mixed functions through functional co-location pattern mining and functional time-varying characteristic mining.The spatial proximity of various POI elements was identified.In doing so,we obtained size-2 and size-n co-location patterns using the participation index,Bron-Kerbosch algorithm and condensed tree pruning algorithm.This paper tentatively introduced urban time-heat data into the analysis of mining functional timevarying characteristics.From the perspective of document similarity,co-location patterns,time stamps and heat values were coupled into mode strings and used as word terms,and the similarity among city states at different times was discovered through the LDA model.(4)With a discussion of the urban functional time-varying features,this paper developed an urban functional zoning method based on a dynamic topic model(DTM)and used Tencent location data and POIs as data sources.This method took into account the cumulative effect of urban functions in multi-temporal areas and improved the interpretability of urban region semantics.Compared with the TF-IDF and NMF methods,our method obtains better results in the discovery of urban functions.Furthermore,fine-grained blocks,as functional units,may lose semantic information to some extent.Therefore,we embedded the UPG method into the original algorithms,which further improved the above algorithms and yielded better functional zoning results.This paper focuses on addressing urban function layout and an analysis of spatiotemporal characteristics and conducts a series of experiments on the construction of urban functional units,mixed function layouts and urban functional zoning methods considering time-varying features.This study can effectively obtain reasonable urban functional zoning results and further spur the application of functional zoning methods in field planning,which has important scientifically significant and social value for developing urban functional zoning theories,urban planning,environmental monitoring and other related fields.