LAUSR

Abstract Cities are the hotspots of global human–environment interactions, and their sustainable development requires proactive strategies to mitigate and adapt to emergent environmental issues. Nevertheless, most of the existing studies and strategies are based on specific (and often singular) environmental processes, and their efficacy is largely undermined by their heavy dependence on locality. Here we present a novel modeling framework for urban studies to capture spatial connectivity and teleconnection among cities in response to different environmental stressors. For illustration, a generic message-passing-based algorithm is used to identify spatial structures among U.S. cities. Urban structures are analyzed under two types of environmental stressors, i.e., extreme heat and air pollution, based on remotely sensed land surface temperature data during short-term heat wave events and a yearlong remotely sensed aerosol optical depth dataset, respectively. Results show that U.S. cities are clustered as locally and regionally connected groups, while the hub–periphery organization manifest via environmental similarity and atmospheric transport under both event-scale meteorological extremes and long-term environmental stressors. The physics-driven urban agglomeration reveals that cities are multilevel interconnected complex systems rather than isolated entities. The proposed framework provides a new pathway to shift goal- or process-based urban studies to system-based global ones.