This research presents a series of projects that contribute to the understanding of how traffic variability affects the measurement and application of annual average daily traffic (AADT). AADT is the most fundamental traffic statistic in transportation engineering. It is defined as the number of vehicles expected to use a facility on an average day. However, traffic is known to experience periodical fluctuations over time;​ these periodicities are location-specific. This underlying variability in time and space can be lost when calculating and reporting AADT.

This research comprises four research projects. The first evaluates the effectiveness of multiple AADT formulations using simulated data loss scenarios. It finds that a relatively new methodology, proposed by the Federal Highway Administration in the United States, removes a small, systematic bias (0.1%) from the existing calculation convention and reduces the width of the 95% confidence interval by 0.5%. The second project provides a method for measuring and reducing the error produced during the assignment step of the AADT estimation process. It applies this method to a case study, finding that the novel assignment method reduces errors by 2.5% on average. The third project explores the use of unconventional traffic data sources (passively-collected vehicle probe data) in tandem with conventional sources. The research finds that speed-based probe data are most closely correlated with truck-specific volume data, specifically around urban centres and along major trade routes. In the studied data, the Pearson correlation coefficient reached 0.9 at some sites. The final project tests the sensitivity of grade crossing design and regulation to predicted fluctuations in traffic. The results show that daily variations in traffic can cause sites to be apparently over- or under-designed for a day or group of days, when compared to regulatory standards. Moreover, they show that within-day variations can be used to express more detailed grade crossing exposure estimates than the daily averages that are used in current regulations.

On aggregate, the research finds that, while AADT estimates are convenient to calculate and ubiquitously applied, there is a need to better disclose the source data and methodologies used to produce AADT estimates to avoid misuse and false assumptions about comparability. Further, AADT summarizes the traffic at a site into a single average volume, which fails to express the known periodical traffic variability at a site.