To enhance the thermal regulation efficiency of blue-green infrastructure, this study investigated the influence of blue-green landscape structure on land surface temperature (LST) across three spatial contexts: the main urban area, the urban development area, and the entire municipality. Remote sensing images acquired on September 18 and 19, 2022 were used to derive LST values and classify land cover categories. The correlation between LST values and all commonly used landscape metrics were statistically quantified. Principal component regression analysis was employed to identify the dominant factors influencing LST under different spatial contexts and to reveal the underlying mechanisms. The results showed thatSboth water bodies and green spaces exhibited significant “cooling island effect”, with the cooling intensity of water bodies (8.96°C–9.34°C) being significantly stronger than that of green spaces (4.44°C–5.47°C).S Overall, the independent explanatory ability of the landscape metrics for LST variation followed the pattern: water bodies > green spaces, landscape composition > spatial configuration, patch-level > landscape-level > class-level, and the main urban area > the urban development area > the entire municipality. The dominant factors influencing LST varied across spatialScontexts. In the main urban area, the key factors were: the percentage of water body area (PLAND_W), water body patch density (PD_W), effective mesh size of green spaces (MESH_G), and edge density of green spaces (ED_G). Together, these four metrics explained 82.4% of the LST variation. In the metropolitan development area, the dominant factors were: contrast-weighted edge density of water bodies (CWED_W), percentage of water body area (PLAND_W), mean proximity index of green spaces (SIMI_MN_G), and percentage of green space area (PLAND_G), collectively explaining 59.2% of the LST variation. In the entire municipality, the five dominant landscape metrics related to blue-green spaces only explained 35% of the LST variation. When the combined effects of other landscape elements were considered, the cooling effect of green spaces was remarkably weakened or suppressed. Water bodies and construction land jointly played a dominant role in impacting thermal environment. The findings indicate that the blue-green spaces exhibit a "cold island effect", and the regulation function of blue-green infrastructure in the thermal environment shows a distinct context effect. Targeted spatial allocation and structural optimization of blue-green landscapes based on specific matrix conditions can enhance the cooling efficiency of blue-green infrastructure.