Day and night spatial risk factors for Manhattan bicycle-vehicle collisions using risk terrain modeling

Purpose Although temporal and environmental factors influencing bicycle–vehicle collisions have been studied, few analyses examine both simultaneously. This study asked which spatial risk factors contribute to bicycle–vehicle collisions during the day versus at night and how these patterns differ across time periods. Methods Using risk terrain modeling (RTM), we analyzed police-reported bicycle–vehicle collisions in Manhattan and assessed 66 environmental features. Relative risk values (RRVs) were generated using RTMDx. RTM risk maps were produced for daytime and nighttime collisions, and spatial overlaps were compared using hot spot analysis comparison (HSAC). Results Grocery stores were the most persistent risk factor, with RRVs of 4.343 (day) and 4.433 (night). Leading pedestrian interval signals also had strong spatial associations (RRVs: 3.542 day; 3.131 night). Daytime-specific risks were elevated near truck routes (RRV = 1.619) and bakery/ice cream/coffee shops (RRV = 2.871). Additional features retained in the nighttime model included bars/nightclubs (RRV = 1.831) and tattoo/body piercing parlors (RRV = 1.846). HSAC identified 70 consistently high-risk cells across both time periods. Temporal shifts were evident: 50 cells changed from daytime hot to nighttime cold spots, while 31 showed the reverse. An additional 322 nighttime hot spots emerged from previously nonsignificant areas. Conclusion Environmental influences on bicycle–vehicle collisions vary by time of day, supporting time-specific safety interventions. This study demonstrates the application of RTM beyond crime analysis to transportation injury prevention and provides actionable insights for public health, planning, and transportation safety.