Planning for Low-Carbon Mobility: Unraveling the Spatiotemporal Operational Mechanisms of Bike-Sharing Systems for Proactive Urban Planning

Integrating bike-sharing into low-carbon transport networks requires a granular understanding of spatiotemporal dynamics. This study presents an integrated framework—combining travel behavior analysis, spatial network modeling, community detection, and deep learning—to examine the operational patterns of Washington, D.C.’s Capital Bikeshare system using high-resolution trip data from July to September 2024. Within this observation period, the analysis reveals distinct behavioral patterns: members undertake shorter trips, whereas casual users make longer, more variable trips. Spatially, a pronounced core–periphery structure emerges, with concentrated activity in the central service area and fragmented peripheral connectivity. Community detection identifies spatial modules that provide groupings for planning. Furthermore, a gated recurrent unit (GRU) model is developed for station-level hourly demand forecasting. These findings inform proactive urban planning through targeted infrastructure assessment, predictive rebalancing, and user-tailored mobility policies. By bridging pattern mining and predictive intelligence, this study provides an empirical foundation for planning within the observed system and period, while its applicability to other seasons and cities requires further evaluation.