Abstract: The differentiation characteristics of landscape pattern affect the urban thermal environment. In this study, temperature characteristics of nine types of landscape in a national forest city, Longquan City, were analyzed by temperature inversion method and spatial analysis. The landscape pattern analysis method was used to explore the correlation of landscape metrics and the thermal environment from 1 km to 3.5 km. The results showed that the high and sub-high temperature zones of Longquan City were distributed in northeast-southwest, mainly composed of urban and rural residential areas. The low and sub-low temperature zone were mainly distributed in the northwest and southeast areas, mainly composed of public welfare forests. By calculating the mean land surface temperature of each landscape type in the area below 700 m above sea level, the temperature of coniferous forest, broadleaf forest, conifer-broadleaf forest, bamboo forest and water was relatively low, whereas that of shrub land, other forest land, cultivated land and construction land was relatively high. Through the analysis of landscape pattern and thermal environment, it was found that the class pattern index was more practical than the landscape pattern index. The correlation between thermal environment effect and construction land distribution reached 0.835, coniferous forest land, broadleaf forest land, coniferous-broadleaf forest land and water were the second, up to -0.5 to -0.4. The cooling effects of different forest types vaied across different spatial scales. Broadleaved forests and coniferous-broadleaved forests were more conducive to cooling at large scales. The larger the area and volume stock of forest land, the more likely it had the lowest land surface temperature.