Assessing the Impacts of the COVID-19 Pandemic on Subway Ridership and on the Interactions with other Transportation Modes

Abstract

The COVID-19 pandemic has led governments to implement restrictive policies which have caused unprecedented effects on transportation systems. This paper assesses which measures had more impacts on subway daily ridership in Montreal (Quebec, Canada) and on the interactions between modes using time series approaches. Change point detection methods, based on regression structure and Bayesian posterior probabilities, are first applied to automatic fare collection (AFC) data available from January 2019 to December 2021. Nine breakpoints (or ten phases) are found and linked to the COVID-19 timeline of the city. The impacts are then quantified by phase, and their variability is analyzed by day type and period. The evolution of the daily and weekly patterns in subway usage is also examined using time-frequency wavelet analysis. Finally, changes in correlations between the subway ridership and the use of three other transportation modes (cycling, private car, and carsharing) are modeled using interrupted time series models with autoregressive errors. The results reveal that lockdown implementations had a negative, immediate but decreasing impact on subway use, while release measures combined with transit-specific policies (such as free distribution of masks) led to a gradual recovery. The impacts varied in time, but traditional 5-day-a-week peak hour travel declined the most. The use of the other modes, as well as their system-level interactions with the subway, were also affected. Potential modal shifts were highlighted. Such findings provide practitioners and planners with useful insights into the COVID-19 pandemic impacts on mobility.

Spatio‐temporal patterns of the impact of COVID‐19 on public transit: An exploratory analysis from Lyon, France

“…Among all the subjects that have been raised in relation to COVID-19, resilience is widely studied. Deschaintres et al (2022) studied daily time series of PT demand and identified breakpoints, corresponding to the implementation of health policies. At a more disaggregated level, they compared different periods of the day and types of days.…”