What is the best catchment area of bike share station? A study based on Divvy system in Chicago, USA

Bike-sharing systems are becoming more and more popular both in the United States and around the world due to their easy accessibility and environmental friendliness. At the end of 2016, more than 430 bike-sharing programs and nearly 2.3 million bikes were available to the public around the world. In the United States, 123 million trips have been made using shared bikes since 2010. In 2017, the number of trips made using shared bikes has reached 35 million, which is 25\% higher than that of 2016. As an increasingly popular mode of transportation, bike-sharing systems are designed for short-distance trips, which not only improves the efficiency of short-distance travels but also alleviates traffic congestion and air pollution. In addition, interests in bike-sharing systems have been fueled by several cities, including Washington DC, where bike share has quickly become a convenient and popular transportation option. Feasibility is the first to study for parties considering implementing a bike-sharing system, detailed information such as the location of stations, number of shared bikes and especially the appropriate size of catchment area should be estimated precisely by system planners. Afterwards, with such information reasonable policies can be made to ensure the operational efficiency of system and maximize system-wide ridership. However, little information is available on what size of an appropriate catchment area should be used. Due to its simplicity in application, direct demand model is gaining more attention in demand forecasting. In this study, direct demand models were used to explore the best catchment area of the bike-sharing stations in Chicago, USA. In order to find out the best catchment area, multiple radii distance range from 200 m to 800 m were established around stations in the system. Results indicated that model fitting results do not vary among different catchment areas. Thus, for purpose of completing their studies, whatever catchment area is readily available or calculated can be used by direct demand modelers. And more attention should be paid by researchers to explore the appropriate boundaries in the future.