Abstract: Eight greenspace types were extracted from QuickBird image, and land surface temperature was estimated from Landsat 8. The cooling effect of various greenspace types was compared by scoring method, and 4 patch features of greenspace were stepwise regressed to the cooling degree and the cooling distance respectively. The results show that forest has the best cooling effect, followed by the greenspace in parks and residential districts, and the combination of greenspace and water body also could cool its surrounding areas. It is found that the perimeter, area and the vegetation coverage ratio in greenspace are weakly and positively related to the cooling degree and distance. The stepwise regression models for all and each greenspace types suggest that the augment of vegetation covering ratio in greenspace could bring some effective cooling effect on the surrounding areas, while the geometric features of greenspace patch do little on the LST of the surrounding areas.