Investigating hourly volume proportions on freeways considering hour, day of week and month

Date of Completion

January 2001


Engineering, Civil




Hourly proportions (or time of day factors) are often needed for estimating highway link hourly volumes, which are important for models in air quality estimation, vehicle crash prediction and transportation planning. Hence, it is a common task for researchers or engineers in transportation to estimate hourly proportion models. This task, however, is very challenging because there are many factors that affect the hourly proportions. These factors in general include: (1) geometric and operational features, socio-economic characteristics and land use patterns associated with the highway network, and (2) temporal factors, including hour, day of week and month. The temporal factors in the second group contrast from those in the first group further in that they have distinct cyclic effects. ^ These factors should all be taken into account in an hourly proportion model. However, factors in the second group are generally not considered or considered in a highly approximate way in previous models. This is because, in order to consider these factors, the number of model parameters can become extremely large and beyond the capacities of existing model building tools and computers. Consequently, only annual average hourly proportions are estimated using the previous models, which in turn result in annual average hourly volume estimates. In many cases, the accuracy of the estimated hourly volumes cannot satisfy the rapidly increasing demands of more detailed and accurate vehicle emissions, accident prediction, and transportation planning models. ^ This work is intended to learn how to obtain more accurate hourly proportion estimates. This is carried out by first thoroughly investigating the effects of hour, day of week and month using Analysis of Variance (ANOVA) statistical procedure. The issue of primary concern here is to identify whether these factors interact with each other. Procedures are then established to group these factors. This includes the use of Comparison of Means statistical procedure (along with an engineering criterion) and a grouping algorithm developed ad hoc. The purpose of these exercises is to reduce the model size and make it feasible to include hour, day of week and month, as well as other factors, in hourly proportion models. In addition, hourly proportion models are estimated considering only hour, day of week and month. The hourly proportions estimated using these models can be used to estimate volumes for a specific hour, day of week and month in a year at a highway location, which in turn can significantly improve the accuracy of the hourly volume estimates. ^