Abstract: Using the monitoring data from the automatic weather stations of Baoding，Wangdu and Zhengding on Jingshi freeway from
December 2007 to November 2009，this paper analyzed the characteristics of road surface temperature． Results show that the variation
of mean road surface temperature was very similar to the diurnal variation of air temperature． This two temperature almost at the same
time reached a very identical minimum about one and a half hours after sunrise，but the road surface temperature reached maximum one
to two hours earlier than the highest air temperature． The air temperature always shows a significant positive correlation to the road surface
temperature． But the total cloud cover，low cloud amount，dew － point temperature，visibility and relative humidity show some opposite
correlations to maximum and minimum road surface temperature，and one of the correlations was important and another was not
obvious． Establishment of the prediction model contained variety of weather factors using the multiple regression for the maximum and
minimum road surface temperature both in winter and summer． The model works well with the temperature ranging from 40 ℃ to 60 ℃
in summer． The predictions for the road surface temperature below － 5 ℃ in winter show a slightly higher value，but the prediction error
less than 2 ℃ accounted for 80%．