Spatial Transferability Testing of Dummy Variable Winter Weather Model Using Traffic Data Collected from Five Geographically Dispersed Weigh-in-Motion Sites in Alberta Highway Systems

It has been an engineering practice that highway agencies collect traffic data using highway traffic monitoring techniques such as permanent traffic counts (PTCs) and weigh in motion (WIM). This research used the WIM traffic data collected for $6\mathrm{years}$ from one of six WIM sites installed and operated in the Alberta provincial highway network to develop a winter weather dummy variable model. Five other sites were used for a spatial transferability test of the estimated model. Few past studies have tested empirically whether a winter weather model developed for one site can be transferred spatially to other locations. A goal of this paper is to develop a winter weather dummy variable model using winter season traffic and weather data and then test its spatial transferability by applying the model to other geographically dispersed locations. A total of 16,746,310 vehicular records collected for $6\mathrm{years}$ spanning from 2005 to 2010 at a WIM site on a commuter road near the City of Leduc, Alberta, Canada, were used to calibrate a model. Three vehicle classes such as total traffic, passenger cars, and truck traffic were classified from raw WIM data and used for model development. Using four types of model structures differentiated from the initially developed model for each vehicle class, this research performed spatial transferability. This research shows that the developed dummy variable model can be successfully spatially transferred to the other five WIM sites. More accurate traffic estimates can be made during winter seasons by using other model structures for each vehicle type.