Understanding the urban form-air quality relationship is essential to mitigate intra-urban air pollution and to improve urban ecology. However, few studies considered urban form and meteorological factors integratively and analyzed their synthetic effects on air pollution. We investigate how to model the integrated effects on the spatiotemporal distribution of PM2.5 in the Atlanta metropolitan area to improve the understanding of the urban form-air quality relationship. Two groups of models are developed: one uses urban form only and the other uses wind-direct urban form. Relative humidity, wind speed, and temperature are included as control variables. Both linear (Multiple Linear Regression) and nonlinear models (Random Forest and Artificial Neural Network) are constructed and tested with both tenfold cross-validation and field PM2.5 data obtained from a portable device, AirBeam2. Random Forest overall outperforms other models suggesting that the urban form-air quality relationship is most likely to be nonlinear. Additionally, the group using wind-direct urban form outperforms the other group and the contribution of the same urban form metrics differs in different wind sections proving that meteorological factors and urban form have synthetic effects on PM2.5. Finally, the patch density, dominance, and aggregation of roads and vegetation, demonstrate higher attribute significance than other urban form metrics. Urban planners, practitioners, and policymakers need to carefully consider not only the spatial configuration of roads and vegetation but also the local climate patterns to minimize intra-urban air pollution effectively.

中文翻译：

---

整合气象因素以更好地理解城市形态-空气质量关系

了解城市形态-空气质量关系对于减轻城市内部空气污染和改善城市生态至关重要。然而，很少有研究综合考虑城市形态和气象因素并分析它们对空气污染的综合影响。我们研究了如何模拟亚特兰大大都市区 PM2.5 时空分布的综合影响，以提高对城市形态-空气质量关系的理解。开发了两组模型：一组仅使用城市形态，另一组使用直风城市形态。包括相对湿度、风速和温度作为控制变量。线性（多重线性回归）和非线性模型（随机森林和人工神经网络）都是用十倍交叉验证和现场 PM2.5 构建和测试的。5 从便携式设备 AirBeam2 获得的数据。随机森林总体上优于其他模型，表明城市形态-空气质量关系最有可能是非线性的。此外，使用直风城市形态的组优于另一组，并且相同城市形态指标在不同风段的贡献不同，证明气象因素和城市形态对 PM2.5 具有综合影响。最后，道路和植被的斑块密度、优势和聚集显示出比其他城市形态指标更高的属性重要性。城市规划者、实践者和政策制定者不仅需要仔细考虑道路和植被的空间配置，还需要仔细考虑当地的气候模式，以有效地减少城市内的空气污染。随机森林总体上优于其他模型，表明城市形态-空气质量关系最有可能是非线性的。此外，使用直风城市形态的组优于另一组，并且相同城市形态指标在不同风段的贡献不同，证明气象因素和城市形态对 PM2.5 具有综合影响。最后，道路和植被的斑块密度、优势和聚集显示出比其他城市形态指标更高的属性重要性。城市规划者、实践者和政策制定者不仅需要仔细考虑道路和植被的空间配置，还需要仔细考虑当地的气候模式，以有效地减少城市内的空气污染。随机森林总体上优于其他模型，表明城市形态-空气质量关系最有可能是非线性的。此外，使用直风城市形态的组优于另一组，并且相同城市形态指标在不同风段的贡献不同，证明气象因素和城市形态对 PM2.5 具有综合影响。最后，道路和植被的斑块密度、优势和聚集显示出比其他城市形态指标更高的属性重要性。城市规划者、实践者和政策制定者不仅需要仔细考虑道路和植被的空间配置，还需要仔细考虑当地的气候模式，以有效地减少城市内的空气污染。