Assessing the effect of snowfall and cold temperature on a commuter highway traffic volume using several layers of statistical methods

Abstract

Presented in this paper is an application of the parametric and non-parametric statistical modelling technique to assess and diagnosis the impact of snowfall and cold temperature on classified vehicle volume variations with five years of Weigh-in-Motion (WIM) data recorded at the commuter type highway nearby the City of Leduc in Alberta, Canada. The influence of the winter driving conditions on classified vehicle volumes travelling in mixed traffic stream is examined. The first phase involves developing parametric dummy-variable regression models to relate and then quantify the variations of daily traffic volumes associated with snowfall and temperature. The second phase involves utilizing a non-parametric Chi-square test combined with the Binomial and the Friedman test. Two levels of statistical tests commonly revealed that the total traffic and passenger car volumes are influenced by both snowfall and temperature, but the truck traffic volumes are not affected significantly. The results of this study may contribute to presenting a new statistical methodology for observing changes in highway traffic in winter. It can also provide important insights for highway agencies to develop efficient traffic monitoring programs throughout the year.