Impact of road surface temperature and condition on the risk of winter vehicle collisions

Pavement surface temperature and moisture condition are very important factors that influence road safety during the wintertime. Yet the literature lacks any serious research to determine their effects. Knowledge of pavement temperature is also necessary to prepare effective road winter maintenance operations.; This thesis examined the effect of pavement surface temperature and moisture condition on the risk of vehicle collisions during wintertime using 2001/2002 data from the City of Ottawa. Analysis included collisions involving fatalities, injury, PDO, single-vehicle, rear-end, and other impact types. During peak and off-peak periods, collision frequencies on wet pavement had their maximums at -1°C and remained relatively high around this temperature. The collision frequencies on dry surface had no specific pattern with changing temperature. A pavement moisture risk factor was calculated as the ratio between collision rate per hour of exposure on wet surface and that on dry surface at each pavement temperature. Results concluded that the risks of collision on wet surface were higher than those on dry surface for all categories and were higher during the off-peak period than during the peak period at all categories but the single-vehicle collisions. The increase in collision risk ranged from 12% for rear-end collisions during peak period to 106% for single-vehicle collisions, also during peak period.; Empirical Bayes probability method was developed as a powerful tool to determine the hazardous pavement surface temperature and moisture condition combinations. Results concluded that at 95% confidence level, driving on dry pavement surface during wintertime was not hazardous at any temperature and that moisture on pavement surface increased the hazardous driving condition particularly when the surface temperature was at or below the freezing mark.; In addition, statistical models that can be used in the decision-making process for winter maintenance operations were developed to predict pavement surface temperature from weather variables. Advanced statistical tests were performed to detect multicollinearities and autocorrelation. The final version of the models included lag dependant variables.; A case study confirmed the model accuracy and applicability. The predicted values of the surface temperature were dependent on the accuracy of the air temperature and dew point forecasts.