Abstract: Under the background that urban heat island affects urban thermal comfort, blue-green space has become an important way to effectively mitigate urban heat island effect. Selecting 18 river riparian buffer in Shanghai, the marginal effect of various indices of waterfront green spaces on the distribution of LST was quantitatively analyzed by the boosted regression tree(BRT) model. The results showed that: the cooling island effect of green spaces was the strongest in the urban periphery, followed by the suburb, and the worst was in the central area. In the aspect of spatial morphological factors of green spaces, the index of Fv and area were the dominant impacting factors on the LST. The marginal effect threshold of Fv, area, landscape shape index and albedo were 0.3, 50 hm², 2 and 0.16, respectively. In the aspect of spatial structure factors among blue spaces and green spaces, the influence of the index of water surface ratio and the distance to riverbank were larger on the LST. The higher water surface ratio is, the more obvious the cooling island effect is. The marginal effect threshold of the distance to riverbank and cohesion value were 400 m and 75/100, respectively. The study can provide useful guidance for the development of blue-green space coping with global climate change.