Objective Due to the advancement of urbanization in China, a series of severe public health problems have ensued, and chronic diseases have become one of the major diseases that cause death among urban residents. Green space in urban parks has various functions, such as reducing air pollution, regulating temperature and humidity, and reducing noise, which can provide residents with ideal and comfortable places for physical activity. Promoting physical activity in urban parks has functional continuity and public benefits and is an integral part of the healthy city concept. However, previous studies on promoting physical activity in urban parks have primarily focused on the correlation between park characteristics and physical activity behavior, and most of them are single-dimensional feature measures. The direct correlation between park green space features and physical activity ignores the complexity of their pathways. Therefore, constructing a multidimensional assessment system of park green space features and clarifying the green space-physical activity pathway have both theoretical value and practical support for health behavior-oriented urban green space planning and layout.

Methods This study took two sample plots in Nanjing with significant differences in greenness level indicators as research objects (Xiamafang Park and Yanziji Park), recruited 56 subjects to participate in a 2-hour free physical activity experiment, recorded the type and duration of the activity, and measured the simultaneous data of park green space features in the sample plots in real-time, and finally modeled and analyzed all the data of the spatial unit of the physical activity. Among them, the park green space feature system is divided into two-dimensional spatial features and three-dimensional visual features. The two-dimensional spatial features include usability features (NDVI), organizational features (connectivity value, integration degree), and accessibility features (distance from the network of entrances and exits and vertical distance from external roads), and the three-dimensional visual features are the visibility features (green visibility and depression degree). Secondly, this study used the independent t-test of 5% significance to analyze the difference between the green space characteristics of each raster sample site of Yanziji Park and Xiamafang Park, respectively, to explore the differences in the green space characteristic dimensions of the parks in the two sample sites. We also calculated the degree of explanation of the physical activity indexes by the green space characteristics of the two sample parks through the multivariate linear regression model. The partial least squares structural equation modeling (PLS-SEM) was utilized to link the multidimensional park green space characteristics with physical activity representations and to clarify the potential influence pathways between park green space characteristics and physical activity. In the model, the independent variable was park green space characteristics, the dependent variable was physical activity characteristics, and the mediator variables were noise, temperature and humidity index, and air pollution, respectively. Finally, based on the mechanism of “park green space-physical activity” under different levels of green volume, we explore the methods and optimization approaches of urban park green space planning to promote physical activity.

Results This study found that the correlation between the multi-dimensional park green space characteristic indexes would be affected by the level of green volume of the green space, the two-dimensional indexes of usability characteristics (NDVI), and the three-dimensional indexes of visibility characteristics (green visibility, depression) in the green space with high green volume showed a significant positive correlation; in the green space with low green volume, the environmental organization characteristics and the accessibility characteristics positively affected the noise perception and thus positively promoted the physical activity; the temperature and humidity index was the most critical mediator variable in the green space affecting physical activity, and was not affected by the difference in the green volume of sample plots. The temperature and humidity index was the most critical mediating variable in green space affecting physical activity and was not affected by differences in the amount of green space in the sample sites.

Conclusion The findings of this study provide empirical evidence that park green space features influence physical activity behavior through environmental stressors, and the differences and similarities in the pathways of action are analyzed in relation to different green spaces. First, an index system coupling two-dimensional spatial features and three-dimensional visual features is investigated to provide a more scientific and comprehensive method for assessing park green space features. In addition, the temperature and humidity indices in the parallel mediation model are all shown to be more critical paths, indicating that microclimate comfort greatly influences the physical activity representations of the active population. The guidance and shaping of green space need to consider the composite perception of the population on the green visibility, the degree of depression, and the microclimate at the same time, and the environmental optimization of the green space should pay attention to the principle of the green visibility of the moderate principle. In addition, there is a significant positive correlation between noise and the frequency of physical activity in urban green spaces, and the noise feeds back to the attractive characteristics of the space. At the same time, the convergence of people makes the space produce a higher frequency of activity stay and pass through. Finally, the study’s results provide ideas for the subsequent investigation of the association mechanism between green space and physical activity in urban parks and provide strategies for health-oriented urban green space planning and design.