Abstract: Exploring the effects of greenspace landscape pattern on thermal environment is of great significance to improving urban ecological environment. Based on Landsat remote sensing data in Xiamen, Zhangzhou and Quanzhou of southern Fujian, landscape ecology and spatial autocorrelation theory were used to quantify spatial distribution characteristics of greenspace landscape and land surface temperature. Further, we used bivariate spatial autocorrelation and spatial autoregressive model to analyze the spatial relationships. The results showed that both of greenspace and land surface temperature had significant spatial autocorrelation. In the inland areas with high elevations, greenspace was significantly concentrated with relatively low land surface temperature, while in urban areas, towns and large areas of cultivated land, greenspace was less distributed and relatively high land surface temperature. Greenspace landscape pattern had stronger effects on the land surface temperature in Zhangzhou and Quanzhou than in Xiamen. The percentage of landscape, largest patch index, aggregation index and mean patch area of greenspace were negatively correlated with land surface temperature, whereas patch density was positively correlated with land surface temperature. The correlation between edge density, mean shape index and land surface temperature is uncertain. Spatial lag model and spatial error model performed better in explaining the spatial relationship between greenspace landscape pattern and land surface temperature.