Over the past two decades, the prediction of pavement surface temperatures in winter has evolved into a maturing field of applied meteorology. Used in refining winter road treatment strategies, these predictions have enormous economic value and implications for motoring safety. Much of the early development was done in Northwestern Europe, but road maintenance agencies (RMAs) across North America are now adopting this technology.
Faced with dwindling resources, RMAs must demonstrate the cost effectiveness of new initiatives. Verification studies are necessary to quantify the net benefit of pavement temperature forecast services and to demonstrate their importance within the framework of road weather information systems (RWIS). RMAs are also interested in establishing a standardized suite of performance measures to aid in evaluating weather information services. To be effective, such performance measures must account for regional factors, including climate conditions. Hitherto, verification studies have concentrated on results achieved in European climates.
The Ottawa Regional Weather Centre has been providing pavement temperature forecasts operationally for a number of RMAs for several years. Forecasts were based on output from a simple point heat balance model with no statistical corrections. During the winter of 1997/98, services were provided for a sufficient number of sites to conduct a preliminary verification study. The Ottawa Centre is collaborating with RMAs to develop a relevant set of performance measures. Measures under consideration include: a simplified signal detection technique for forecast crossings of critical threshold temperatures; temporal errors in predictions of threshold crossings; and absolute errors in forecast pavement temperatures for various ranges of observed pavement temperature.
This study provides preliminary results which may be useful in defining the performance measures for pavement temperature forecasting for eastern North America. These results will also assist in the determination of the value of RWIS to RMAs.