Logistics heat islands in California’s inland empire: A geographically weighted machine learning approach

The rapid expansion of logistics and warehouse infrastructure has transformed the land and thermal landscapes of Southern California’s Inland Empire. While traditional urban heat research emphasizes residential density and impervious surfaces, this study identifies a distinct logistics heat island phenomenon associated with large-scale warehouse clusters. Using parcel-level warehouse data from 2001to 2024, satellite-derived land surface temperature, and local climate records, we quantify how industrial land-use intensification reshapes urban heat exposure. A Geographically Weighted Random-Forest Regression framework integrates spatial heterogeneity, nonlinearity, and interaction effects to explain localized heat dynamics. Results show that vegetation loss remains the dominant driver of surface warming, but warehouse concentration amplifies heat particularly in neighborhoods with higher poverty rates, larger Hispanic populations, and more elderly residents. These heat shadows extend beyond parcel boundaries, revealing spatial spillovers from logistics land use. The findings conceptualize warehouses as climate infrastructures that redistribute thermal risk, extend environmental injustice, and challenge existing urban resilience paradigms. Methodologically, this study demonstrates how spatially explicit machine learning can reveal place-based interactions among land use, vegetation fragmentation, and social vulnerability. The results highlight the need for equitable climate adaptation and regulatory frameworks that integrate industrial land governance, green buffers, and cooling infrastructure into sustainable urban planning.