SAFF model proposed in this study allows to forecast the road surface temperature along a road network using the DSM and heat balance method. The accuracy of the SAFF model was identified by comparing observed thermal mapping data and forecasted results at 631 selected stretches (each with a length of 20m) of KASUGA-WADAYAMA road (12.6km) in Hyogo prefecture.
 The SAFF model had 58% and 86% of spatial accuracy rate for the calculation of T_s with an error of less than 1°C and 2°C based on all road stretches, respectively. The mean absolute error of Ts over the entire route and calculation term of 45 hours was 1.5°C and the error was smaller during the nighttime than the daytime. CPU time with an algorithm to combine mountain elements and to remove invisible mountain elements from calculation routine was reduced to 1/23 of that without the algorithm. Consequently, the SAFF model made it possible for one hour forecast of the spatial variation in dry road surface temperature over the entire KASUGA-WADAYAMA road.

View full abstract