Abstract
A frozen road surface increases traffic accidents during the winter season. Hence, information on easily-frozen road sections and their specificities are required to prevent traffic accidents. Frozen road surfaces are determined by equipment measuring road surface temperatures. However, there are limitations in investigating the entire road network. Therefore, it is imperative to develop new methods that effectively determine road surface freezing risks. Meteorologically, road surfaces are frozen when the actual temperature cools down to the dew point temperature. Under this condition, there is likely to be frost if relative humidity reaches 100% and frozen road surfaces as the temperature gets lower. Meteorological characteristics give us an alternative to a direct measurement road surface temperature to estimate risks of road surface freezing. Based on the clues, the relationship between severity of traffic accidents and temperature changes is empirically investigated using Paju weather data. The results reveal that as the temperature gets lower and changes in current temperature are relatively small, the severity of traffic accidents become higher. In addition, the same is true when the difference between current temperature and the dew point temperature is relatively small, as it increases possibilities of road surface freezing. Future studies must investigate how current temperature and the dew point temperature affect road surface freezing and thereby establish a time-space scope to estimate possible road surface freezing sections using only weather and road material type data. This would provide invaluable information for predicting and preventing frozen road accidents based on weather patterns.