Urban Functional Area Classification And Hot Spot Detection Based On Multi-source Spatial Data

Urban functional classification and urban hot spot distribution are the key factors to describe the spatial structure of a city.With the development of economy and society and the acceleration of urbanization,the population and resources are converging to the city.The structure of the city has changed greatly compared with the past,and the division of function in different regions is more and more obvious.The areas of specific function in the city attract a large number of population inflow and becomes the hot spot of the city.At the same time,it also causes local traffic congestion,environmental pollution,land shortage and repeated construction.In order to alleviate the above urban problems,the planning department needs to understand the functional structure of urban area and the concentration area of human activities to carry out urban construction and resource allocation more effectively.However,the existing methods of urban function classification are not applicable at extracting the data from the scenes with more mixed functional areas due to insufficient data mining and utilization.Moreover,the boundary of hot spot areas extracted by the existing urban hot spot detection method is fuzzy,and the hot areas of different levels are not distinguished,which makes it difficult for the decision-makers to accurately and intuitively understand the status quo of urban development.In this paper,Futian district of Shenzhen city is selected as the experimental area,and multi-source spatiotemporal data are comprehensively used to fully explore the correlation between the data and the spatial information contained in the data.By introducing the adjacency information of basic functional units,more accurate functional area classification and urban hot spot detection are realized.This study is divided into three parts: correlation analysis of multi-source spatial data,classification of urban functional areas and urban hot spot detection.Firstly,this paper carry out conduct statistical analysis and spatial coupling analysis on nighttime light images,point of interest data,floating vehicle trajectory data and microblog check-in data,and conduct principal component analysis on data with high repeatability to reduce dimension,to provide more effective feature vectors for subsequent research.Secondly,this paper extracted the geographical elements distribution,human activity intensity information and human transfer mode of urban functional areas from multi-source spatial data.In this way,category characteristics and spatial characteristics are combined effectively.Then,we used Dirichlet Multinomial Regression model to calculate the theme subject probability calculation model,and obtained urban function classification result by clustering and interpretation.Finally,the nighttime light images,the floating car trace data and the micro-blog check-in data are fused into three kinds of data that can reflect the intensity of human activities,and the basic unit of urban heat research is determined by using the objectoriented segmentation method.Then we used the local spatial autocorrelation analysis algorithm and the geographical weighted regression model to detect the main center with the strongest global heat and the sub-center with strong local heat in the city.Finally,combining with the functional area classification of the region,the location entropy is calculated to analyze the functional structure of the hot region,and the reasons for the regional attraction are discussed.In order to verify the feasibility of this paper’s functional area classification and urban hot spot detection method,this paper carried out verification experiments on the accuracy of functional area classification,the effectiveness of spatial feature construction and the hot spot detection effect.The experimental results show that the data fusion method and space characteristic information adopted in this paper,can effectively improve the performance of function classification.The proposed hot spot detection method can accurately extracted the main center and sub-center of the city,and the extraction results have stronger interpretability,which can help the planning departments to carry out urban construction.