Spatial Heterogeneity and Nonlinear Effects of 2D/3D Grey–Green Urban Morphology on Day–Night Heat Exposure Among Older Adults and Children: Evidence from Hangzhou, China

Global warming and urbanisation have intensified urban heat exposure, with older adults and children particularly vulnerable. However, day–night variations, age-group disparities, and nonlinear responses to two-dimensional (2D) and three-dimensional (3D) grey–green morphology remain underexplored. Using ten urban districts in Hangzhou, this study integrated land surface temperature, population, land use, and canopy height data into a 1500 m grid database. Population-weighted heat exposure was quantified for older adults and children. An OLS–MGWR–CatBoost–SHAP (Shapley additive explanations) framework identified spatial heterogeneity, key drivers, nonlinear response ranges. Results show: (1) Heat exposure exhibits spatial clustering and is higher during the day. High-risk areas are concentrated in the central built-up districts, especially Shangcheng and Gongshu, while lower exposure occurs in western mountainous and greenbelt areas, waterfront zones, and peripheral ecological spaces. (2) 3D urban morphology is associated with exposure disparities: building coverage ratio (BCR) is positively associated with exposure, mean building height (BH) shows stronger nighttime associations, particularly for children, and average canopy height (ACH) shows stronger negative nighttime associations. (3) 2D grey–green patterns do not follow a simple “grey spaces increase heat and green spaces reduce heat” rule. Branched, bridged, and fragmented grey spaces are associated with higher heat exposure, while green-space cooling is related to connectivity, integrity, and canopy height. In high-explanatory-power areas, BCR shows a rapid-response range of 0.01–0.04, while BH and ACH show response changes at 7–13 m and 7.5–10 m, respectively. These responses mainly characterise western high-explanatory-power areas covering 8.5–20.2% of the corresponding vulnerable populations, reflecting conditional responses within these populations; central districts remain priorities for citywide heat-risk identification rather than direct application of these morphology-based responses. This study provides evidence for heat-risk management, grey–green spatial optimisation, and health-equity planning, thereby contributing to climate-resilient and sustainable urban development.