To improve the atmospheric environment by optimizing urban morphology, this study develops a random forest (RF) model to investigate the influence of urban morphology on PM_2.5 variations via the relative importance of urban morphology and the nonlinear response relationship between urban morphology and PM_2.5. Two indices—reduction range (C_↓) and rate (C_˅) of PM_2.5 concentrations—are defined to evaluate the temporal variations of PM_2.5. Results show that RF models are more accurate and perform better than multiple linear regression models, with R² ranging from 0.861 to 0.936. Five out of nine urban morphological indicators have the most significant contribution to PM_2.5 reduction. For each indicator, the nonlinear response relationship shows similar trends in general, despite of the difference at the higher pollution level. Building evenness index and water body area ratio have a similar response such that C_↓ and C_˅ sharply increase and tend to be stable when they reach at 0.05 and 8 %, respectively. With the increase in vegetated area ratio, the change of C_↓ presents an inverted V-shape trend with the turning point of about 20 %; however, the change of C_˅ greatly differs from the pollution level. A higher density of the low-rising buildings with one to three floors will lead to a small reduction rate but a greater reduction range of PM_2.5. Floor area ratio values generally show a negative and nonlinear influence on C_↓ and C_˅. This study provides useful implications for planners and managers for PM_2.5 reduction through neighborhood morphology optimization.

为了通过优化城市形态改善大气环境，本研究开发了随机森林（RF）模型，通过城市形态的相对重要性以及城市形态与 PM _2.5之间的非线性响应关系，研究城市形态对 PM _2.5变化的影响。定义了两个指标——PM _2.5浓度的减少范围 (C _↓ ) 和速率 (C _˅ )来评估 PM _2.5的时间变化。结果表明，RF 模型比多元线性回归模型更准确，性能更好，R ²范围从 0.861 到 0.936。九个城市形态指标中有五个对 PM _{2.5 的}减少贡献最大。对于每个指标，非线性响应关系总体上表现出相似的趋势，尽管在较高的污染水平上存在差异。建筑平整度指数和水体面积比也有类似的反应，C _↓和_C˅分别达到0.05%和8%时急剧增加并趋于稳定。随着植被面积比例的增加，C _↓的变化呈现倒V型趋势，拐点约为20%；然而，_C˅的变化与污染程度有很大不同。一到三层的低层建筑密度越高，PM _{2.5 的}减少率越小，但减少范围越大。容积率值通常对 C _↓和 C _˅有负面的非线性影响。这项研究为规划者和管理者通过邻里形态优化减少PM _2.5提供了有用的启示。