Urban thermal comfort has emerged as a critical factor affecting environmental sustainability and social equity, particularly in aging societies undergoing compact urbanization. This study employs machine learning algorithms to investigate how multidimensional urban morphology shapes spatial disparities in thermal environment. Taking Xi'an, China, as a case study, this research integrates multi-source data—including remote sensing imagery, elderly population distribution and 2D/3D urban morphology models—and employs Extreme Gradient Boosting (XGBoost) model to capture the complex, nonlinear relationships between urban form and thermal conditions. Results indicate that over 70 % of urban blocks experience thermal discomfort, with high-density historic districts facing 4.7–7.4 times higher thermal exposure than peripheral new towns. Moreover, thermal discomfort is highly associated with social vulnerability, as evidenced by an 88 % spatial overlap between thermal discomfort blocks and areas with above-average elderly population densities, where the thermal discomfort index is over ten times the citywide average. The AI-driven model indicates that vegetation and water bodies account for over 60 % of the total cooling contribution. However, when NDVI falls below 0.2 or building density exceeds 0.2, thermal discomfort tends to intensify significantly. A “double-edged sword” effect of 3D urban morphology is observed: moderate building height enhances shading and ventilation, yet excessive height (above 32.5 m) worsens heat stress. This research supports climate-responsive urban design and underscores the need to address thermal inequity in aging, high-density cities.
扫码关注我们
求助内容:
应助结果提醒方式:
