Research On The Identification Method Of Urban Functional Area Based On Multi-source Spatiotemporal Data

In the process of expansion and development of modern cities,under the influence of regional functional differentiation,the activities of urban residents will present specific spatial and temporal distribution rules,and the characteristics of regional functional areas can be reflected by the spatial and temporal characteristics of residents' travel.The traditional classification methods of urban land use types are mainly land use thematic map,field survey statistics and expert rating based on remote sensing images.Most of these methods start from the macro level and are difficult to reflect the social,economic and cultural characteristics of urban land in detail,or have high cost and long cycle,and are not time-sensitive.In recent years,with the development and popularization of GPS positioning technology,people can quickly and accurately obtain the geographical location and moving trajectory of the research object.Trajectory by studying the floating car data mining city residents travel behaviors and characteristics become a hot topic of data mining in recent years,and can through the motion law of floating car to probe the space distribution features of human activity,and its use and urban land feature extraction and identification of urban functional areas become provides a new method for scientific monitoring,urban planning.This paper mainly USES the drops a taxi order data of Chengdu,Baidu maps of POI data and multi-source data such as road network data,combining block classification methods of Gaussian Mixture Model and characteristics of the number and area of the POI block classification method,the main city of Chengdu,which can identify the features of the final respectively from the aspects of time and space to analyze,and use the decision tree for functional area accuracy is verified.The main research of this paper is as follows:(1)Block classification based on the spatial and temporal characteristics of passengers getting on/off.In this paper,considering the boarding passengers in different blocks can reflect the law of travel,the use of getting some time series as a travel characteristics,based on Gaussian Mixture Model and the correlation coefficient method for clustering points up and down the space and time of the vehicle,to extract the features of different blocks of the city structure,the final results obtained including rural tourism,live,mixed functions,business,shopping,transportation hub,landmark,a total of 7 different functional blocks.(2)Block classification based on Voronoi diagram.This article uses the Voronoi diagram of POI area as the weight of the structure model of the POI weighted basis,divided into building POI weighted structural model,make up the different POI space area function division deviation of different factors,thus different blocks in urban function classification and include vehicle services,industrial park,catering mainly of the mall,shopping mall,park scenic area of residential,commercial hybrid functions,commercial residential areas,a total of 7 different functional blocks.(3)Comparison and identification of urban functional areas.To combine mobile trajectory data and POI data,this paper respectively USES comparative analysis method and the decision tree classification method,combined with the plan of Chengdu,the results of different data to determine urban functional areas to fusion,forecasts the city function and verify its accuracy,the final analysis found that this method have a high recognition of residential area,business district and medical area in cities.To sum up,this paper uses movement trajectory data and POI data to identify the urban functional area of Chengdu,and prospects the future research methods.