Abstract: The relationships of land surface temperature (LST) with the land use types were analyzed in Shenyang urban districts.The analyzing processes are as follows.First,the LSTs of Shenyang urban districts and within the third ring road were calculated using Landsat TM data in August 2010.Then,the land use of the studying area were classified as building land,green land,farmland,abandoned land,road and water body six types using visual interpretation and supervised classification methods.Finally,the city center was regarded as an origin point,and the relationships between the land use types and the LSTs at different circles distancing from the origin point in the city were analyzed using multivariate linear regression and partial least-square regression methods.The corresponding modeling results were verified as well.The results show that the mean LSTs of building land,green land,farmland,and road and the central urban district decrease with the increasing of the distance from the city center origin point.At the same level within buffering zones,the LSTs of building land and road are higher,and that of water body is the lowest.The "booth pie" developing in a city makes that the percentage of building land and road areas gradually reduced with the increasing of the distance from the city center origin point.Both of building land and road enhance the heat island effect.The main land use types are building land in the studying area,followed by road.Both of multivariate linear regression and partial least square regression methods perform very well in predicting LST.Comparatively,the partial least square regression method is more ideal to analyze the relationships between land use and LST in Shenyang.

Key words: Land surface temperature, Land use, Multivariate linear regression method, Partial least square regression method