Research On Modeling Spatiotemporal Trajectories And Spatiotemporal Proximity Analysis For Time Geographic Analysis

Time geography is a powerful framework for studying human activities under various space-time constraints and has served as a fundamental analytical framework for many scientific fields,such as transportation planning,accessibility analysis,human mobility analysis and etc.With advances in positioning and wireless communication techniques,collecting the spatiotemporal trajectories data of vehicles and travellers becomes more convenient in urban cities.Massive trajectories data are generated by various positional sensors,including floating car data,mobile phone records,smart card data,social media check-ins,etc.However,the rapid increase in the volume,diversity,and intensity of spatiotemporal data poses a significant challenge to contemporary GIS platforms with respect to the storing,modeling and operation of spatiotemporal trajectories data under the framework of time geography.Since conventional spatiotemporal data model and methods can not meet the demands of modeling,processing and analysising spatiotemporal big data,it's necessary to study the methods of quality assessment,data cleaning,modeling and analysis for spatiotemporal trajectories data.Aiming to model urban spatiotemporal trajectories data,floating car dat(FCD)has been choosen as the main data source and methods of quality assessment,data cleaning,modeling and spatiotemporal operations of massive FCD are introduced in this paper.The main contributions of this paper are listed as follows.(1)A multi-criteria dynamic programming map-matching(MDP-MM)algorithm is proposed for data cleaning of FCD.First,the large candidate routes issue of multiple hypothesis technique is addressed by the multi-criteria dynamic programming technique in which route evaluation score,network topology and speed limitation are three main criteria for determining the candidate routes at each GPS point.Using this technique,the number of candidate routes at each GPS point is reduced to the number of candidate locations within the error region of the GPS point,and the best matching route is guaranteed.In addition,two useful techniques are developed to improve the path finding procedure.Second,the procedure of MDP-MM algorithm is designed for matching large-scale low-frequency FCD.Finally,to demonstrate the applicability of the proposed MDP-MM algorithm,two case studies using real FCD are carried out.The results indicate that the proposed MDP-MM algorithm is superior to existing FCD map matching algorithms in both matching accuracy and computational performance.(2)A comprehensive data quality assessment model for FCD is proposed.First,traditional quality assessment models of spatial data are summarized,and the limitations of traditional quality assessment models are discussed.Second,in order to meet the demand of various applications of FCD,9 data quality assessment elements for FCD are introduced,including positional accuracy,sampling frequency,trajectory completeness,attribute validity,trip validity,coverage,timeliness,spatial interaction and mobility.Moreover,the mathematical methods of quality assessment indices are detailed to quantitatively evaluate the quality of FCD.Finally,a case study using real FCD is carried out,and the 9 quality elements of FCD are assessed in detail.The results demonstrate the validity of proposed FCD quality assessment model.(3)A spatiotemporal data model for network time geographic analysis is proposed.First,time geographic entities(space-time path,space-time prism,space-time station,space-time lifeline)and relations(intersection and bundling)in the planar space are introduced.Second,based on proposed compressed linear reference(CLR)technique,the proposed spatiotemporal data model can transform network time geographic entities in three-dimensional(x,y,t)space to two-dimensional CLR space.Network time geographic entities can be stored in conventional spatial databases by utilizing the proposed spatiotemporal data model.Spatiotemporal operations and queries for network time geographic entities in CLR space can be directly implemented by efficient spatial operations and index structures.Finally,to validate the proposed spatiotemporal data model,a case study is performed using large-scale datasets of space-time paths and prisms.The case study indicates that the proposed spatiotemporal data model is effective and efficient for storing and querying large-scale datasets of network time geographic entities.(4)A method of space-time buffering for spatiotemporal proximity analysis is proposed.First,the conception of space-time buffer of trajectories is proposed in this paper.Space-time buffering is a natural extension of conventional spatial buffering operation to space and time dimensions.Based on the space-time buffering operation,spatiotemporal proximal trajectories can be determined efficiently by considering space and time dimensions simultaneously.Second,to generate space-time buffer of trajectories,the space-time buffering algorithm in network space is proposed and the CLR technique is applied to implement space-time buffer in classical spatial databases.Finally,a case study is performed using large-scale datasets of trajectories to validate the applicability of the proposed space-time buffer.The case study indicates that the proposed space-time buffering operation is an effective and efficient method for spatiotemporal proximity analysis of trajectories data.