Abstract: The study aims to study the relationship between green space pattern and temperature according to the established scale. The pattern of green space was calculated by green space pattern index, the index was determined by patch size, boundary characteristics and spatial characteristics, and inversion temperature was obtained by remote sensing images. The results showed that when the patch number was above 400 and the dispersion of average patch area and patch area changed within a certain range, the temperature could drop. When the average patch boundary complexity was kept within a certain range, the temperature could drop. The results of spatial location also showed that when the average index and the average distance changed within a certain range, the temperature could drop. The results have guiding significances for the construction of urban green space.