Metro ridership recoverability and the built environment: Exploring station-level non-linear impacts

The COVID-19 pandemic has highlighted the importance of resilience in metro ridership, as it shifted travel choices towards private transport modes, thereby undermining sustainable transportation. Focusing on the recovery phase of transit system disruptions, this study proposes a new measure of ridership recoverability, inspired by the traditional resilience triangle, and employs the XGBoost model to investigate the nonlinear relationships of land use variables, along with accessibility and network structure measures, with metro ridership and its recoverability. Our findings show that bus services around metro stations are positively correlated with both ridership and its recoverability. In contrast, some land use variables have opposite associations with ridership and recoverability. For example, increasing commercial uses promotes ridership but undermines recoverability; whereas mixed land uses dampen ridership but improve recoverability. These insights help better understand the complex interactions within urban transport systems and offer valuable considerations for decision-making in urban rail transit and land use planning.