Search for: All records

Abstract Urban air quality and temperature are closely linked through coupled physical and chemical processes. However, most existing evidence relies on correlation-based associations lacking directionality, or process-based models whose causal pathways depend on model structure and parameterizations. Here we apply a nonlinear causal inference method to quantify directed coupling between near-surface air temperature and three major air pollutants (PM_2.5, ozone, and NO₂) across 481 U.S. urban areas using 14 years of daily data. We identify distinct diurnal and seasonal regimes of temperature–pollution coupling not captured by linear correlation. Temperature–PM_2.5and temperature–O₃coupling strengthens in summer, whereas temperature–NO₂coupling intensifies in winter. Ozone shows the most consistent causal structure among all pollutants, with temperature dominant in roughly 80% of urban areas in both seasons. PM_2.5exhibits balanced and spatially heterogeneous coupling, while NO₂shifts from mixed behavior in summer to pronounced temperature dominance in winter. Across pollutants, linear correlations frequently overestimate coupling strength, especially for winter NO₂. As the first continental-scale causal assessment of urban temperature–pollution interactions in the U.S., this study offers a data-driven complement to process-based modeling. The identified pollutant-specific sensitivities and their regional, diurnal, and seasonal variability provide new insight for understanding and managing urban heat stress and air quality. 

Urban heat is a growing concern especially under global climate change and continuous urbanization. However, the understanding of its spatiotemporal propagation behaviours remains limited. In this study, we leverage a data-driven modelling framework that integrates causal inference, network topology analysis and dynamic synchronization to investigate the structure and evolution of temperature-based causal networks across the continental United States. We perform the first systematic comparison of causal networks constructed using warm-season daytime and nighttime air temperature anomalies in urban and surrounding rural areas. Results suggest strong spatial coherence of network links, especially during nighttime, and small-world properties across all cases. In addition, urban heat dynamics becomes increasingly synchronized across cities over time, particularly for maximum air temperature. Different network centrality measures consistently identify the Great Lakes region as a key mediator for spreading and mediating heat perturbations. This system-level analysis provides new insights into the spatial organization and dynamic behaviours of urban heat in a changing climate.