In this paper, the spatial-temporal characteristics of green travel behavior in public bike-sharing system are studied. From the vector perspective, a coarse graining method of green travel direction is proposed, the methods and steps of analyzing the spatial characteristics and temporal window of green travel behavior are established, which opens up new horizons in the study of green travel behavior, develop the existing coarse granulation methods and temporal network theory, and provides a new way of route optimization and high-frequency task window recognition for the dynamic rebalancing scheme. At the same time, we conduct an empirical analysis by the travel data of the public bike-sharing system in Nanjing from March 20, 2017 to March 26, 2017, and verify the dynamic spatial characteristics and timevarying of green travel behavior based on vector perspective.

Introduction

As a sustainable, efficient, economical, flexible and environmentally friendly transportation mode, public bike has become an increasingly powerful way to alleviate the last-mile problem. Around the world, public bike-sharing system has been operated in about 1500 cities at present (Ahmadreza et al., 2017; Haider et al., 2018), and green travel behavior has been widely concerned by scholars (Si et al., 2019). Travel data records green travel behavior in time and space, and contains a large amount of potentially valuable information. The improvement of data acquisition and analysis of public bike system promotes the research of green travel behavior. In order to capture the characteristics of green travel behavior, travel speed (Jensen et al., 2010), travel time (Jappinen et al., 2013), the gendered travel behavior (Zhao et al., 2015), travel type (Bordagaray et al., 2016), trip chains (Zhang et al., 2018) and spatial structure (Boss et al., 2018) are studied. At the same time, the relationships between green travel behavior and site selection (Wang et al., 2016), the sustainability and efficient functioning of cities (Bullock et al., 2017), dynamic repositioning (Zhang et al., 2017), built environments (Liu and Lin, 2019), job accessibility (Pritcharda et al., 2019) are analyzed according to the spatial-temporal characteristics of the data. Spatial-temporal characteristics are the eternal research theme of the objective world. Although there are many literatures on green travel behavior, the researches focus on spatial-temporal characteristics by travel data are limited, and there are some imperfections: 1) spatial characteristics are mostly based on statistical pattern (Zhao et al., 2015; Wang et al., 2016; Zhang et al., 2018; Boss et al., 2018), and lack of dynamic analysis from the vector perspective; 2) temporal characteristics are mostly based on artificial time windows with equal intervals (Wang et al., 2016; Bordagaray et al., 2016; Pritcharda et al., 2019), and lack of high frequency temporal windows which are based on dynamic spatial characteristics to describe green travel behavior with tidal phenomena (Chardon et al., 2016).