Urban parks can contribute to significant cooling effect for the city microclimate and, therefore, promote urban residents’ outdoor thermal comfort. As the enhancement of the urban heat island effect, understanding the cooling efficiency of urban park and associated determinants has become critical for maximizing the ecological benefits of urban ecological infrastructure. In this paper, we measured holistically the cooling efficiency of urban parks from a relative and absolute perspective by using the data envelopment analysis (DEA) model and Landsat 8 OLI/TIRS satellite imageries. We adopted two novel cooling effect indices to rank and identify the inefficient park: DEA-based cooling efficiency index for relative efficiency and park cooling intensity index (PCII) for absolute efficiency. A total of 146 urban parks within the Shanghai metropolitan city was selected to conduct the empirical research. The correlation analysis and machine learning techniques (XGBoost) were also used to examine the relationships between the characteristics of parks and two cooling effect indices. The findings revealed that the average cooling efficiency of urban park in summer was approximately 2.7 times higher than that in winter. Park size and vegetation greenness (NDVI) were identified as the dominant factors for park cooling efficiency. Moreover, larger parks with area > 5 ha showed better cooling performance on average. In the current study area, increasing the water body coverage ratio also played an important role on the improvement of cooling effect in both summer and winter. The research findings could provide valuable references for designing urban parks and for urban planning to maximize the cooling benefits and mitigate the UHI effects.

城市公园可以为城市小气候提供显着的降温效果，从而提高城市居民的室外热舒适度。随着城市热岛效应的增强，了解城市公园的降温效率及相关决定因素对于最大化城市生态基础设施的生态效益变得至关重要。在本文中，我们通过使用数据包络分析 (DEA) 模型和 Landsat 8 OLI/TIRS 卫星图像，从相对和绝对角度全面测量了城市公园的冷却效率。我们采用了两个新的冷却效果指数来对低效园区进行排序和识别：基于 DEA 的相对效率冷却效率指数和用于绝对效率的园区冷却强度指数 (PCII)。选取上海大都市内的 146 个城市公园进行实证研究。相关分析和机器学习技术 (XGBoost) 也被用于检查公园特征与两个冷却效果指数之间的关系。结果表明，夏季城市公园的平均降温效率比冬季高约2.7倍。公园规模和植被绿度 (NDVI) 被确定为公园降温效率的主要因素。此外，面积大于 5 公顷的大型公园平均表现出更好的冷却性能。在当前研究区，提高水体覆盖率对夏季和冬季降温效果的提高也起到了重要作用。