Abstract
A road icing prediction system is under development to predict several hours in advance the presence or absence of an iced road by making use of the road surface temperature measured by the optical fiber sensor laid underneath the road and the weather forecast data of that area. In order to upgrade the icing prediction accuracy, the method of utilizing the weather forecast data has been improved and integrated into a prototype system. The system has been further used in carrying out a field experiment on an actual road, with the predicted results closely corresponding with the actual icing status, proving the validity of the system.