Amid intensifying urban heat island (UHI) effects, urban green spaces play a vital role in regulating city temperatures and mitigating overheating. Taking the main urban area of Nanjing as a typical case study, this research quantitatively and systematically evaluates the spatial heterogeneity of green space cooling effects across urban functional zones and explored their underlying mechanisms through an integrated approach combining remote sensing data, multi-source geographic datasets, and interpretable machine learning. By integrating an eXtreme Gradient Boosting (XGBoost) regression model with SHapley Additive exPlanations (SHAP) and Partial Dependence Plots (PDP), we revealed and quantified how landscape factors affect green space cooling intensity and to clarify their influence mechanisms. The results showed three main aspects: First, there were significant differences in the cooling intensity of green spaces in different functional zones. Among them, the public service zone had the highest cooling intensity, mainly ranging from 0.02℃ to 0.79℃; the commercial zone had the lowest, ranging from 0.02℃ to 0.41℃; the industrial zone and the residential zone were respectively distributed between 0.02℃ to 0.71℃ and 0.01℃ to 0.42℃. Second, individual landscape factors demonstrated varied effects on the cooling intensity of green spaces. The patch size emerged as the dominant influencing factor in all urban functional zones, although the degree of its effect varied considerably among them. Additionally, other critical landscape variables-such as tree height, the spatial pattern of water bodies surrounding the green spaces, and the distribution of impervious surfaces around the green spaces-were also found to be significant factors affecting the cooling intensity. Third, the combination of certain landscape factors can produced a significant positive synergistic cooling effect under specific conditions. For instance, optimizing the configuration of green space patch area and shape, green space patch area and surrounding openness, tree height and impervious surface boundary density, etc., can effectively enhanced the cooling intensity of green spaces. For different functional zones, green space construction strategies that were tailored to local conditions and optimized through coordinated approaches are proposed. The key measures include controlling the scale of green spaces within an appropriate range, planting medium and tall-canopied trees, enhancing the connectivity of green spaces, and reasonably regulating the proportion of impermeable water surfaces and spatial openness in their surroundings. In conclusion, the strategies and findings presented in this study make meaningful contributions towards optimizing urban green infrastructure, effectively regulating urban thermal environments, and facilitating the sustainable development and construction of beautiful, livable cities.