A more detailed park prescription: How does landscape support human health behavior?

Introduction: The role of urban parks in promoting public health is continuously evolving. For time-pressed urban residents, knowing the specific locations within parks that support various health behaviors can help them use these spaces more effectively for their daily usage. This study attempts to develop an analytical framework at the plot scale, using a 30*30 m grid as the basic unit, to assess the intensity of health behaviors driven by different landscape characteristics within parks. Methods: To investigate these relationships, seven representative urban parks in Shanghai were selected, with 68 standard-sized plots established as sampling sites. Firstly, computer vision-based semantic segmentation was employed to measure the landscape features within these spots. This was combined with systematic behavioral observation and coding to quantify the intensity of three types of health behaviors-exercise, leisure, and social activities-at each measurement spot. Subsequently, regression analysis was used to construct a model defining the relationship between landscape characteristics and health behavior intensity. This model was then applied to predict and visualize the intensity of health behaviors across the entire park area. Results: Results show that sky visibility, pavement coverage, and the presence of rough or uneven ground are significant predictors of the overall intensity of health-related behaviors in the studied plots. More specifically, trees, pavements, rough or uneven ground and resting facilities are closely associated with the intensity of exercise behavior, while shrubs primarily affect the intensity of leisurely behaviors. Mapping the intensity of health-related behaviors in Fuxing Park revealed that spaces for leisurely activities overlap with those for the other two types of health-related activities, whereas exercise and social behaviors exhibit spatial exclusivity. Discussion: Ultimately, the resulting visualizations, which map the distribution and intensity of different health behaviors, thereby serve a dual purpose. Firstly, they enable users to promptly locate areas within parks that are suitable for specific health-promoting activities, thereby helping to prevent potential conflicts between different types of behaviors. Secondly, by establishing clear relationships between plot-level landscape features and observed health behaviors, the framework provides park managers with an evidence-based tool for optimizing the allocation of environmental resources to support diverse recreational needs.