Unveiling the Coupled Spatiotemporal–Network Resilience of Urban Taxi Systems under Recurrent Daily Peak-hour Disturbances

Urban taxi systems are continuously exposed to recurrent peak-hour demand surges that impose sustained operational stress, while existing resilience studies have predominantly focused on rare extreme disruptions. This study develops a coupled spatiotemporal–network resilience framework to investigate how urban taxi systems respond to routine, high-frequency peak-hour disturbances. Based on two weeks of high-resolution taxi trip data collected in Zhuhai, eight representative daily disturbance events are identified through a data-driven detection approach capturing distinct commuting and leisure travel demand regimes. Spatiotemporal resilience and network resilience are assessed separately considering resistance and adaptability at both node and system levels. These dimensions are subsequently integrated through an entropy-weighted coupling method to reveal multidimensional resilience states. Results show pronounced scenario-dependent resilience patterns. Structured commuting peaks maintain moderate operational performance despite limited structural flexibility, whereas leisure-oriented disturbances exhibit higher network robustness and greater route diversity, facilitating adaptation despite more dispersed operational demands. For all identified events, resilience is determined not by total demand volume, but by spatiotemporal distribution of trips and the alignment between operational performance and network topology. The coupled framework further uncovers scale-dependent effects. Spatiotemporal resilience exhibits significant spatial variations across disturbance scenarios, whereas network resilience remains comparatively stable owing to persistent structural properties. High-demand nodes often exhibit stronger integrated resilience owing to structural centrality, while system-level resilience is governed by demand dispersion and flow organization. By explicitly linking spatiotemporal operations to network topology under recurrent daily disturbances, this study lays a diagnostic foundation for scenario-adaptive resilience management in routine urban transportation operations.