Urban park green spaces serve as vital carbon sinks and help mitigate atmospheric carbon dioxide concentrations. However, research into how landscape characteristics infl uence carbon dioxide mitigation and inform design strategies remains insuffi cient. This study examines six extensive parks of varying sizes within Chengdu’s central urban district. Through fi eld monitoring of CO2 concentrations in park green spaces, the study analyzes diurnal variation patterns. It also explores quantitative relationships between landscape features—such as park area, perimeter, shape index, proportion of surface types, and canopy closure—and CO2 reduction levels. The fi ndings reveal three main results: (1) CO2 concentrations within parks are signifi cantly lower than in external environments, with reductions ranging from 4.2 to 11.7 ppm, and show an inverted U-shaped trend throughout the day. (2) During winter, CO2 reduction in park green spaces shows a signifi cant positive correlation with park perimeter and shape index. There is a nonlinear relationship with area, with a threshold eff ect observed at 35~40 hm2. (3) The proportion of hard surfaces and buildings is signifi cantly negatively correlated with CO2 reduction, while the proportion of water is signifi cantly positively correlated. Specifi cally, a 0.1 decrease in hard-surface coverage proportion results in a 9-ppm reduction, and a 0.1 increase in water coverage proportion yields a 3-ppm reduction. These fi ndings provide quantitative support and strategic guidance for developing low-carbon urban parks, maximizing the ecological benefi ts of green spaces, and advancing sustainable urban development.