Environmental stressors associated with human activities such as excessive heat (especially the urban heat island effect) or air pollution are mostly concentrated in urban areas. Ongoing global climate change is expected to increase the duration, intensity, and frequency of heatwaves has been projected, with human activities playing a critical role. Understanding the impact of human activities and their connection with extreme heatwaves in urban areas remains an open challenge. The motivation behind this work stems from the difficulty in quantifying the propagation of extreme heatwaves among cities. 

The concept of complex networks has been applied to examine and interpret relationships and interconnections between different components of a climate system in recent years. The basic concept behind a climate network lies in a collection of components that are meaningfully connected. In this work, we treat 520 U.S. urban areas with temperature records as nodes to construct the climate network. The network’s connectivity, represented by directed links, is evaluated through nonlinear interactions based on causal inference. This method helps distinguish direct dependencies from indirect ones and common drivers in complex systems. Network analysis offers a new insight into the intrinsic dynamics of heatwave propagation from a system-level perspective.

A specific criterion to define a heatwave event is used to detect extreme heatwaves. Temperature records during summertime for the last four decades (1979 to 2021) are used, and 12 heatwave events are selected to construct directed causal networks. The structures of causal networks reveal critical information, such as whether a city acts as a heat sink or source during extreme events—determined by examining the incoming and outgoing links of a city node. An intriguing finding of this work is that cities acting as hubs with numerous links are typically major metropolitan areas with large populations. For instance, during a heatwave event, New York City is interlinked with other cities through 477 incoming links (IN) and 292 outgoing links (OUT). This means that New York receives “heat information” from 477 U.S. cities (indicating that NYC can act as a heat sink) and spreads out “heat information” to 292 cities, contributing to heatwave propagation (making NYC a heat source). More interestingly, these complex interactions include long-range connections, which are important in atmospheric circulation, often referred to as teleconnections.

Moreover, an analysis of the topology of causal network reveals that cities can be ranked regarding their overall importance in heatwaves propagation (“centrality”). This analysis also shows that cities with large population functions as hubs or supernodes in the network of heatwave propagation. Examples of such urban hubs include clusters around New York City, Chicago, and San Francisco.

Given the importance of large cities in regulation of heatwaves, we are eager to understand the impact of human activities in large cities on extreme heatwave propagation. In this work we seek to uncover relationships between heatwave causality and population metrices (population totals and density) by focusing on 53 major U.S. cities with populations over 200,000. Our findings reveal an intriguing phenomenon: a positive relationship exists between heatwave causality and population metrices for those large cities, with population totals exhibiting a stronger positive correlation with causality than population density. This can be attributed to the greater variability in population totals among large U.S. cities, which may better reflect the compounding impacts of human activities such as waste heat emissions. Big cities with large populations do matter in the occurrence of extreme temperatures!

In summary, we try to decipher the patterns of urban heatwave propagation through a complex network approach with causal inference. Our findings highlight several key insights: (1) The spreading of extreme heatwaves among U.S. urban areas exhibits a distinct hub-periphery structure, along with long-distance teleconnections via causal pathways; (2) Urban hubs such as New York and Chicago act as pacemakers in the occurrence of extreme heatwaves; (3) This study reveals a positive relationship between causality and population during heatwaves, marking a pioneering effort to elucidate the causal mechanisms behind extreme heatwaves and the potential influence of human activities on such events in the U.S. We believe that findings in this work can offer valuable guidance for urban planners, policy makers, and stakeholders to mitigate extreme heat in cities. In addition, different criteria or metrics regarding the impacts of heatwaves in urban areas can be considered in future studies—we still have a long way to go to fully understand and mitigate extreme heatwaves.