Study On The Spatiotemporal Characteristics Of Human Convergence And Divergence Based On Massive Mobile Phone Location Data

City is a complex spatial structure formed by the long-term interaction between human and space.Recently,rapid urbanization makes the urban population increasing sharply,which brings great pressure to the management of the city and leads to emerge some urban disease such as traffic jam.The mismatch between human mobility patterns and urban spatial structure is the major reason for these urban problems.A better understanding of relationship between human mobility patterns and urban spatial structure is helpful for urban management and planning,and is meaningful for building sustainable and environmental friendly livable city,so it is necessary to have an in-depth understanding the coupling relationship between human mobility patterns and urban spatial structure for solving urban problems.With the development of information and communication technology(ICT),it becomes possible for collecting large-scale human tracking datasets(e.g.mobile phone data,taxi data).In contrast to conventional travel survey datasets,this type of dataset has large sample size and updates in real-time,especially for mobile phone data,which can continuous track the change of residents' location in the city.Therefore,this dataset contains potential abundant human mobility patterns,which allows us to study human mobility and urban structure in an unprecedented space-time scale.In order to meet the demand of activities,residents move among different places in the city.It exists difference for attracting residents(e.g.the intensity or time of attracting residents)in different areas of the city due to the difference of function,which may lead to human convergent in some areas and divergent in other areas,and the spatial locations of human convergence or divergence is changing over time,forming human space-time dynamic landscape.Therefore,human convergence-divergence is a common and important phenomenon generated by human mobility in the city.Studying on the spatiotemporal characteristics of human convergence and divergence can deepen our knowledge of relationship between human spatiotemporal dynamics and urban spatial structure,and make us understand the potential power factor of human mobility,which is of great significance for traffic management,urban planning and emergency response.This paper focuses on the spatiotemporal characteristics of human convergence and divergence using mobile phone location data,the main objective of the research include identifying spatiotemporal patterns of human convergence and divergence,exploring the relationship between these patterns and urban functional areas;constructing the model for accessing the stability human convergence and divergence as well as quantifying the relationship between stability of human dynamic and urban spatial structure.More specifically,the contents of this paper are:(1)A method is proposed to extract spatiotemporal patterns of human convergence and divergence from vast amounts of mobile phone location data.The proposed method first defines an indicator to represent the intensity of human convergence or divergence,and classifies the intensity into different classes and adds labels to represent the level of intensity of human mobility,then constructs the time series matrix based the labels,finally a x-means clustering method is used to group the matrix into different clusters to identify the areas with similar human convergence-divergence pattern.After that,these spatiotemporal patterns and urban functional areas are put together to analyze the main occurring pattern on each functional land use.(2)Regarding to the stability of human dynamic,this paper propose a model to assess the stability of human convergence and divergence.The model divides the time series of human dynamic into human convergence process(HCP)and human divergence process(HDP),which generates a human convergence and divergence sequence(HCDS).By analyzing the different characteristic of HCPs and HDPs,this study first proposes a model for assessing the stability of HCPs and HDPs,and on this basis further develops a model for assessing the stability of HCDS.Based on mobile phone location data form Shenzhen,China,the proposed model is employed to assess the spatiotemporal stability of people covered by traffic analysis zones(TAZs)and urban public transit system,which provides a reference for urban public traffic management.Finally,we discuss the difference between the proposed assessment model and entropy,finding that the two methods can quantify the stability of human dynamic from two different perspectives and can be used mutually.(3)In order to exploring the influence of urban spatial structure on the stability of human convergence and divergence,this paper utilizes multiple linear regression model(MLRM)to analyzing the relationship between the stability and urban spatial structure.The independent variables are derived from socio-economic,land use and road network which are used to represent the urban spatial structure.By building different MLRMs,we identify the independent variables that have significant impact on the stability of human dynamic,and analyze the relative importance of these variables.We found that the land use mixed entropy has most effects on the stability,the second is the number of population,and the population density and global betweenness centrality have little effect on the stability of human dynamic.