The nonlinear climatological impacts of urban morphology on extreme heats in urban functional zones: An interpretable ensemble learning-based approach

Abstract

Extreme heat (EH) is leading to multifaceted urban challenges. Despite increasing explorations on urban morphology affecting EH, its varied impacts on EH across different climatic zones (CZs) and urban functional zones (UFZs) involving substantial samples remain undefined. Therefore, we developed an interpretable ensemble learning model to dissect the climatological effects of urban morphology, including building morphological and landscape environmental metrics, on EH across varied UFZs. Taking China's 279 cities as testbeds, our findings highlight heterogeneous impacts are heavily contingent upon CZs and UFZs types. The horizontal factors including vegetation coverage, waterbodies, and building density (BD) are more prominent in Temperate and Tropical CZs, whereas the vertical encompassing building height (BH) and floor area ratio (FAR) are more significant in Arid and Cold CZs. Moreover, cooling and warming factors varied across CZs and UFZs. Vegetation, waterbodies, and BH are dominant cooling factors in Cold, Arid, and Temperate CZs, exerting more pronounced inverted-U trends in commercial and residential UFZs with diverse and high-rise forms. BD and FAR are stronger warming factors in public service and industrial UFZs characterizing massive impervious surfaces. Intriguingly, due to minimal water movement and weak energy interactions, a limited number of scattered waterbodies act as warm islands in Arid CZs. Notably, waterbody cooling distance is shorter in Tropical CZs, and BD and FAR stand out as warming impacts in commercial and industrial UFZs. These heterogeneities contribute to a deeper understanding of mechanisms driving EH and provide evidently tailored landscape and urban planning decision-making across CZs and UFZs types.