Threshold effects of green space exposure around residences and workplaces on the duration of leisure walking among urban residents

Leisure walking is one of the most important physical activities, and green space exposure is regarded as an important influencing factor of leisure walking. However, previous studies have paid limited attention to the relative importance of each green space exposure indicator to leisure walking, and the threshold effects of green space exposure and the spatial heterogeneity of residences and workplaces. Based on a survey conducted in Shanghai between 2018 and 2019, consisting of 1049 participants, this study used a gradient boosting decision tree approach to explore the threshold effects of exposure to green space around residences and workplaces on the duration of leisure walking. The results show that around both residences and workplaces, among green space exposure indicators, distance to the nearest park is the most important predictor for the duration of leisure walking. Around residences, distance to the nearest park is the second most important predictor and its ranking is only lower than age. Around workplaces, distance to the nearest park is the sixth important predictor, followed by the area of greenspace and the supply-demand ratio of greenspace, ranking seventh and ninth, respectively. Around both residences and workplaces, distance to the nearest park has negative associations with leisure walking when it is lower than 650 m. Beyond this threshold, distance to the nearest park has a trivial contribution to the duration of leisure walking. Value of NDVI around residences, when between 0.10 and 0.14, is positively correlated with the duration of leisure walking. When the supply-demand ratio of greenspace around workplaces is less than 0.2, it has a positive association with the duration of leisure walking. More green areas and parks are positively related to the duration of leisure walking without clear thresholds. With regard to the attributes of the built environment, around both residences and workplaces, land use diversity positively correlates with the duration of leisure walking, while distance to the city center has a negative association with the duration of leisure walking. Population density around residences and the number of transit stops around workplaces are negatively related to the duration of leisure walking with certain thresholds. These findings provide evidence for planning and designing green spaces around residences and workplaces to promote leisure walking.