Abstract:Base on the analysis of the hourly ground surface temperature differences between new automatic weather station and old one in 7 national weather stations of Shanxi in 2014, it is found that the existing observation methods and equipment make large errors of surface temperature. The data consistent rate is only 93.6 7%, while the outoftolerance rate reaches 35.70%, and errors are more evident during the daytime. Through an indepth analysis on the factors that affect the ground temperature observations, combined with the test results, it is concluded that the solar radiation intensity has little effect on the ground temperature observation errors, and the observation field soil, observation method, sensors, and equipment maintenance have a greater impact on errors. Therefore, it is suggested that the infrared remote sensing method should be used to replace the existing observation method of platinum resistance buried directly.