Nonlinear force analysis of urban morphology and building heat emission based on multi-scale microclimate prediction

To predict the spatiotemporal impacts of urban planning on microclimates, previous studies developed air temperature (AT) models based on urban morphology but often overlooked anthropogenic heat emissions from buildings (AHEb) and lacked cross-scale evaluation. This study analyzes data from 35 sites across five blocks in Xi'an, China, focusing on the block, 50 m, and 20 m buffer scales. Seven morphological indices are calculated using Ladybug in Rhino's Grasshopper, including three 3D indicators—sky view factor, mean building height (MBH), and floor area ratio (FAR)—and four 2D indicators—building cover ratio (BCR), green cover ratio (GCR), impervious surface area ratio (ISAR), and water body ratio. AHEb is simulated via URBANopt in EnergyPlus. Partial least squares and Bayesian-optimized random forest models are developed to predict microclimate at block and point-buffer scales. The nonlinear and scale-dependent effects of AHEb and morphological parameters are assessed, and thresholds for key planning indicators are identified. Results show that GCR and ISAR are the most influential across all scales and should be prioritized in planning. Maintaining GCR at 40 %–60 % (50 m) and 45 %–70 % (20 m) yields significant cooling in high-density areas. These strategies enhance cooling and mitigate adverse design impacts. The influence of BCR and FAR diminishes at smaller scales, while MBH shows the opposite trend. AHEb ranks 4th–6th in importance across scales, less impactful than key 2D and 3D indicators. These findings enhance understanding of how urban morphology and building heat emissions affect microclimates and support climate-adaptive design.