Fine particulate matter (PM_2.5) is harmful to human health. Although the relationship between urban land use and PM_2.5 has been studied in recent years, there has been little consideration of the relationship between land use structure and PM_2.5 spatiotemporal patterns at the microscale. Based on mobile monitoring PM_2.5 data and point of interest data, this paper explored their relationship with a classification and regression tree model. The results showed that PM_2.5 exhibits spatiotemporal heterogeneity at the microscale. The neighborhoods’ land use structure can explain 60.4% of the PM_2.5 spatiotemporal patterns. Transportation and ecology are the two most significant land use types that correlated with PM_2.5 spatiotemporal patterns. Fourteen rules of neighborhood land use structures with different land use types are identified land use structure which leads to different spatiotemporal patterns of PM_2.5. The higher the PM_2.5 risk, the stronger the correlation with neighborhood land use structure is. The classification and regression tree model can be effectively used to judge the relationship between neighborhood land use structure and PM_2.5 spatiotemporal patterns. The results provide a basis for developing appropriate measures, based on local conditions, to predict PM_2.5 pollution levels at the microscale, and reduce the risk of neighborhood exposure to PM_2.5.

细颗粒物（PM _2.5）对人体健康有害。尽管近年来研究了城市土地利用与PM _2.5的关系，但在微观尺度上很少考虑土地利用结构与PM _2.5时空格局之间的关系。基于移动监控PM _2.5数据和兴趣点数据，本文探讨了它们与分类和回归树模型的关系。结果表明，PM _2.5在微观尺度上表现出时空异质性。该街区的土地利用结构可以解释PM的60.4％，_2.5时空模式。运输和生态是与PM _2.5时空格局相关的两种最重要的土地利用类型。确定了不同土地利用类型的邻里土地利用结构的14条规则，从而导致了PM _2.5的时空模式的不同。PM _2.5风险越高，与附近土地利用结构的相关性越强。分类回归树模型可以有效地判断邻里土地利用结构与PM _2.5时空格局之间的关系。结果为根据当地条件制定适当措施预测PM _2.5提供了基础在微观上达到污染水平，并降低邻里暴露于PM _2.5的风险。