Research On Urban Rail Transit Station-level Ridership And Transfer Pattern Considering The Effects Of Built Environment

China is currently undergoing rapid urbanization and motorization,which inevitably brings about a series of urban and traffic problems,such as traffic congestion,environmental pollution,and urban sprawl.The high-capacity public transit system,represented by the urban rail transit,is the key to alleviate these problems.Built environment,as the human-made environment that provides the setting for human activity,is closely related to the mobility of urban dwellers.Due to the difference in urban spatial layout and passengers’ travel purposes,urban rail transit ridership is unevenly distributed at both spatial and temporal scales.The method of capturing the complex pattern of transit ridership influenced by the built environment at different levels is of great importance to the research on trip generation.In addition,intermodal transfers between rail and bus at the rail station have become an important part of station-level ridership.Influenced by the built environment elements around the station,intermodal transfers during different periods exhibit a significantly uneven distribution at the spatial scale.It is important to conduct an in-depth transfer pattern analysis by accounting for multi-dimensional attributes of intermodal transfers.Although the urban rail transit station-level ridership and intermodal transfers have been widely investigated throughout the world,the existing literature on pattern analysis of urban rail transit ridership at a single station and at all stations,and intermodal transfers between metro and bus from both theoretical and empirical perspectives,remains scarce,while considering the effects of built environment.Therefore,this study funded by the National Key R&D Program of China(No.2018YFB1600900),the National Natural Science Foundation of China(No.71771049),and the Jiangsu Province Science Fund for Distinguished Young Scholars(No.BK20200014),focused on urban rail transit ridership and intermodal transfers between metro and bus at different levels considering the effects of built environment.An empirical study was conducted from multi-source datasets including built environment data and smart card data in Nanjing.This study attempted to propose feasible solutions to address the existing research questions by establishing generic modeling frameworks.By unraveling the underlying mechanism of effects of the built environment and pattern of transit ridership and intermodal transfers,this study contributed to enhancing our understanding of the trip generation and behavior of urban public transit.More specifically,this study includes the following contents:First,this study introduced the type and collection method of multi-source datasets including urban built environment data and public transit smart card data,and discussed the attribute and structure of each dataset.This study then analyzed the standardized processing method of heterogeneous datasets and expanded the information of one single dataset by matching and fusing mobility data with other datasets.Based on the processed datasets,this study could further visualize the variability of built environment elements and station-level transit ridership.With the effects of built environment,this study then analyzed the uneven distribution of transit ridership at different levels over space and time.It provided the theoretical and technical support for further research on pattern analysis of urban rail transit station-level ridership and intermodal transfers between metro and bus,and influencing built environment.Second,this study focused on the short-term transit ridership at a specific station.Based on certain built environment element outside the rail station,this study chose the influenced station with the unstable distribution of station-level ridership.Considering the sudden change of the ridership,this study analyzed the variant and complex characteristics of ridership in the time series and proposed a generic framework to model the trend and fluctuation of transit ridership.In specific,this study tested the correlation between squared residuals and verified the residual distribution of short-term ridership at the station level.A hybrid model was built to fit the mean and conditional variance of ridership.The time series pattern of ridership was measured by the modeling coefficient and information impact curve.Compared with the traditional analytical model,the hybrid model could improve the predictions of short-term ridership in terms of accuracy and reliability by incorporating nonlinearity and asymmetry.Third,this study then focused on urban rail transit ridership at all stations.Based on the urban functional layout as well as residents’ commuting and non-commuting pattern,this study systematically investigated the impact of built environment elements on transit travel behaviors.Considering the uneven distribution of ridership over space and time and the layout of the urban rail transit network,this study redesigned the structure and setting of the geographically weighted regression model and used the adaptive distance metric.A novel ridership model was built based on the irregular and complex rail transit network in the real world.Through establishing the multi-level theoretical framework,this study could analyze the pattern of boarding ridership and alighting ridership at multiple spatial and temporal scales.It helped us fully understand the underlying mechanism of the relationship between built environment and travel behaviors of urban rail transit system and offer policy implications for the development of city and transportation.Finally,this study investigated the intermodal trips between urban rail transit and bus.Urban rail transit stations were clustered based on transfer-related built environment surrounding the station to reflect their spatial attributes of built environment.Considering the uneven distribution of intermodal transfers over space and time,this study built ridership cube and structural topic model at both aggregated and disaggregated levels.In specific,this study set time,space,and passengers’ attributes as the dimension of the cube.It helped us visualize the uneven distribution of intermodal transfers for each combination.By redesigning the generative mechanism of the structural topic model,this study incorporated multi-dimensional information into the modeling structure.This process considered each passenger as a unit and calculated the probability of each transfer pattern.After obtaining the prevalence of each transfer pattern,this study could identify the key pattern and measure the correlation between different patterns.Meanwhile,this study also evaluated the role that different age groups played in each transfer pattern,which could analyze how the external land use and internal passengers’ attributes influence the transfer behavior.This study investigated the rail transit station-level transfers with the bus at both aggregated and disaggregated levels,which was helpful to fully understand passengers’ travel behaviors as well as the spatial and temporal pattern of intermodal trips.It provided a useful method for pattern analysis in large-scale intermodal transfer activities and enhanced the integration between urban rail transit and bus.