Andrew John Coxford

Bike share ridership and identifying the impact of infrastructure on moderating fluctuations

Global bike share ridership has seen a meteoric rise in recent years, driven by growing demand for low-carbon, affordable, and active urban transportation. Despite these benefits, bike sharing’s susceptibility to adverse weather conditions severely limits its potential for year-round useability. This study aims to answer how ridership fluctuates based on various weather conditions, and how fluctuations are affected by commuting behaviour and cycling infrastructure. An improved understanding of these factors will be valuable in informing weather-predictive bike share redistribution and necessary advancements for weather resilient infrastructure.

A large body of literature exists on weather’s relationship to cycling, with most analysis being performed using regression modelling. While the overall impacts of most weather conditions and the effects of commuting are well understood, there is clear variation based on the study’s location. This suggests the need for further individualised analysis in each city. There is also a distinct research gap on the ability of infrastructure to moderate ridership fluctuations from weather, which requires further investigation.

To accomplish this, bike share trip data was aligned with the relevant hourly weather records to create year-long databases for Toronto, Montreal, New York, and London. A combination of linear and nonlinear regression modelling was then performed in each city to assess the impact of temperature, precipitation, wind speed, snow depth, and humidity on bike share ridership. Subsequent regressions were performed on subsets of data to determine any changes in impact between commuter and recreational users. Similarly, a final set of regressions was performed on a sample of rides occurring entirely on protected bike lanes to assess changes in sensitivity from the overall ridership base.

As expected, higher temperatures were found to increase ridership while strong winds, precipitation, snow, and humidity decreased bike share use. Several city-specific tolerances were also identified, indicating that cyclists may adapt to the weather conditions most common in their own city. Both commuting behaviour and cycling infrastructure were generally shown to reduce the impacts of weather, serving to normalise cycling as feasible mode of all-weather transportation. Some factors, most notably humidity, show little to no improvement from protected lanes, suggesting areas of improvement within each city’s existing infrastructure system.