Differentiated Effects of Public Transport Accessibility on Park Utilization: Evidence from Crowdsourced Data in Singapore

Parks provide essential health and environmental benefits in cities. Park utilization refers to the extent to which residents use parks for recreational activities. It is substantially influenced by public transport accessibility. Existing research examines this relationship as a single outcome without distinguishing effects across different utilization types. This study investigates the differentiated effects of public transport accessibility on park utilization across Singapore. We integrated public transport stop data, park characteristics, and crowdsourced mobility data. Public transport accessibility was measured by stop density and nearest stop proximity within 500 m catchments around parks, differentiated by transport mode. Park utilization was quantified along two dimensions: routine utilization intensity, derived from GPS activity traces, and organized utilization frequency, identified from geotagged social media imagery using computer vision. Hotspot analysis and regression modeling at the park and district levels examined the effects of transport accessibility on each dimension. At the park level, higher public transport accessibility significantly predicted organized utilization frequency but not routine utilization intensity. Routine utilization was instead associated with park characteristics such as trail network density. At the district level, central and eastern districts formed significant utilization hotspots, while peripheral districts with limited transport accessibility showed persistent underutilization. These findings reveal that public transport accessibility selectively enables park utilization, with a stronger effect on organized than routine utilization. This distinction suggests that accessibility assessments in urban transport planning should account for different utilization dimensions to develop more targeted strategies for enhancing equitable and effective park utilization.