Answers are grounded in this paper and cite their sources.
Prolonged hot seasons, intense solar radiation, hot and humid conditions, rapid urbanization, and intensive development exacerbate subtropical heat risks. Yet few block-scale studies integrate global and local importance, nonlinear responses, and interactions between urban features and land surface temperature (LST) within a multidimensional framework. Using 6,542 road-enclosed urban blocks in Shenzhen, we analyzed 22 urban features in five categories using exploratory spatial data analysis, XGBoost–SHAP, and GW-XGBoost. The results were as follows: (1) summer LST was generally higher in western and central than in eastern Shenzhen, with significant positive spatial autocorrelation (Global Moran’s I = 0.642); (2) NDBI, NTL, and DEM had the highest global predictive contributions, while BSC and HBR were also relatively important. NDBI showed relatively continuous areas of high local importance, whereas BSC and HBR areas were more dispersed; and (3) predictive contributions increased continuously for NDBI, initially increased then decreased for NTL, and declined overall for DEM, while BSC showed a non-monotonic response and HBR a nonlinear decline that levelled off. Joint contributions between NDBI and NTL, BSC, or HBR changed direction across paired-variable levels; DEM–POP and NTL–PD showed conditional dependence. These findings can inform sensitive-range identification, coordinated urban-feature configuration, and spatially differentiated interventions for block-level thermal environments in subtropical cities.
No takes yet. Share an insight, caveat, or question.
Hu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: