摘要
Urban heat and air pollution increasingly co-occur in urbanizing regions, yet existing green-space studies remain dominated by vegetation-quantity metrics and single-outcome assessments. It remains unclear whether green-space configurations contribute beyond vegetation quantity to LST, PM2.5, and O₃ regulation across seasons. This study examines associations between green space morphology and land surface temperature (LST), fine particulate matter (PM2.5), and ozone (O₃) in Chengdu, China, using data from 2000 to 2024. Green spaces were classified into seven morphological types using Morphological Spatial Pattern Analysis, and relationships were assessed using year-blocked analysis of variance, rank correlation, explainable machine learning, and land-use scenario simulation. Continuous types, particularly Core, were associated with the lowest LST and PM2.5, whereas fragmented types such as Islet and Branch showed the highest values, with an inter-type temperature difference exceeding 7 °C. This gradient peaked in summer for LST and winter for PM2.5. Ozone showed a seasonal reversal masked at the annual scale: fragmented types were associated with higher O₃ in summer but lower O₃ in winter, consistent with a shift from photochemical production to nitrogen oxide titration. After controlling for elevation, nighttime light intensity, population density, built-up density, and vegetation abundance, morphology accounted for 12–27% of attributed importance, while the negative Core contribution to PM2.5 was directionally consistent across all years. Scenario simulations showed that land-use change alters environmental outcomes primarily through green-space configuration. These findings demonstrate that urban environmental performance depends on both green-space quantity and configuration, and that no single configuration can optimize all environmental dimensions simultaneously.