Unraveling urban thermal comfort dynamics: Multi-scale interactions with urban building and green space landscapes in Beijing

Urban thermal comfort, influenced by complex interactions among different urban landscapes, plays a critical role in shaping local residents’ well-being. This study takes the core area of Beijing as its case study. A localized feels-like temperature (FLT) for assessing thermal comfort was estimated based on multi-type & continuous meteorological observations. On this basis, multiple 2/3D landscape metrics of surrounding buildings and green spaces at different spatial buffer units (50∼900m) were estimated for establishing multi-scale regression models, in order to explain their roles on seasonal/diurnal FLT. Firstly, this study revealed significant seasonal and diurnal variations in FLT, with the highest levels observed in summer (daytime: 28.43 °C; nighttime: 24.26 °C) and the lowest in winter (daytime: 3.82 °C; nighttime: 0.30 °C). Secondly, Lasso regression models identified scale- and season-dependent roles of landscape patterns. At smaller scales (50∼100m), 3D landscape metrics (e.g., average height and volume density of buildings) dominated FLT variations, exacerbating thermal stress in winter but mitigating daytime heat via shading in summer. Larger scales (≥300m) highlighted 2D configuration metrics (e.g., patch complexity), where fragmented building layouts enhanced cooling through ventilation. Green spaces exhibited weaker daytime effects due to limited shading but improved nighttime comfort via evapotranspiration. Notably, model performance was stronger in cold seasons (R 2 > 0.7) and nighttime cases, while summer daytime models showed reduced significance, which has been confirmed in the comparative analysis with the LST data and fully reflected the conflicting interactions on thermal comfort among air temperature, humidity, wind speed, and urban landscapes. These findings emphasize the need for multiscale adaptive urban planning to optimize thermal comfort, prioritizing building morphology adjustments at local scales and green space connectivity at broader scales. The study provides actionable insights for mitigating thermal stress in high-density urban regions. • Multi-scale modeling quantifies urban landscape on seasonal/diurnal thermal comfort. • Building morphology outperformed than green space in disturbing local FLT. • 3D landscapes work best at site-scale (50–100m), while 2Ds function at larger scales. • Urban landscapes show heightened FLT regulation efficacy during winter and nighttime.