Urban green spaces provide not only critical ecological functions but also serve as essential public resources, supporting daily physical activity and promoting public health. Understanding how green space characteristics influence residents’ physical activity is crucial for improving the effectiveness of urban green infrastructure and achieving health-related planning goals. This study integrates map-based exercise trajectory data from a popular fitness app with high-resolution urban green space vector data, reconstructing approximately 65,000 resident exercise trajectories across Beijing. The analysis reveals a strong resident preference for large, spatially compact green spaces, with a non-linear, inverted U-shaped relationship between NDVI and exercise intensity, suggesting that moderate vegetation levels are most attractive. Spatial autocorrelation tests and spatial econometric models further indicate that physical activity exhibits significant spatial clustering. Local vitality in green spaces tends to influence surrounding areas through a neighborhood spillover effect. By integrating behavioral big data with fine-grained spatial analysis, this approach provides a more detailed and reliable means of capturing human-environment interactions. To further examine spatial dependence, we distinguish between spatial error and lag effects, proposing an "ecological–behavioral" positive feedback mechanism. This mechanism highlights how the active use of green spaces may jointly enhance social participation and ecological performance. Based on these findings, we propose three policy directions: (1) refining green space design and management to align with local environmental and behavioral characteristics, ensuring that interventions are context-sensitive; (2) promoting neighborhood-level coordination to leverage spatial spillovers for community engagement; and (3) developing an integrated green space network, with attention to spatial connectivity and cross-district collaboration. Collectively, these strategies?underscore?the?importance of aligning?ecological quality with human activity patterns?to achieve?healthier, more resilient cities. This research contributes a novel methodological approach to evaluating the effectiveness of urban green space and offers new insights into the integration of ecological and behavioral data for evidence-based urban planning.