Developing Cold Region Winter Weather Traffic Models and Testing Their Temporal Transferability and Model Specification

Abstract

It is well known from past studies that inclement winter weather conditions such as heavy snowfall and harsh temperature affect the variation of traffic volume. In order to quantify the change in traffic volume caused by winter weather conditions, a winter weather model explaining the relationship between weather conditions and traffic volume has been generally developed. However, none of the studies in the past have evaluated the performance of the developed model in terms of temporal transferability. In addition, there is no research showing that model specification for predicting correct traffic under winter weather conditions can be optimized for different types of vehicles. This paper presents an application of dummy-variable regression to develop winter weather traffic models. In an effort to fine-tune winter weather model specification by the vehicle class, two model validation tests were conducted. First, a temporal transferability test is conducted to determine whether the model developed is suitable to be used regardless of time. Second, a model specification test is conducted to determine whether the winter weather model specification initially structured is the most appropriate format or not. To pursue this, 6 years of traffic and weather data sets from weigh-in-motion (WIM) and weather stations were collected at Highway 2A in Alberta, Canada. The study results indicate that a dummy-variable winter weather traffic model that includes a temperature variable shows good performance in terms of accuracy and temporal transferability for passenger cars. In contrast, the naïve winter weather model specification that excludes a temperature variable is preferable in predicting truck traffic volume. It is very interesting to note that even if vehicles are traveling in the same traffic stream under the same environmental conditions, different model specifications are preferable for different vehicle types to predict more accurate traffic in winter season.