Improving AADT Estimation Accuracy of Short-Term Traffic Counts Using Pattern Matching and Bayesian Statistics

Abstract

The importance of reliable estimates of travel demand for effective planning, design, and management of roads and facilities is well known by transportation engineers. A review of current short-term traffic monitoring practices shows that most transportation agencies simply use road functional class as the criteria to assign short-term traffic counts (STTCs) to permanent traffic counter (PTC) factor groups, or they commit significant resources to implement other data intensive methods, such as regression analysis. The improved methods described in this study estimate average annual daily traffic (AADT) with higher accuracy using all historical counts collected to date for a short-term counting site to create its seasonal traffic pattern and assign it to a PTC or a PTC group without imposing additional data collection cost. Two pattern-matching methods, and their combination with Bayesian statistics, are proposed and tested using PTC data from Alberta, and their results are compared to the Federal Highway Administration (FHWA) method. Study results show that, compared to the FHWA method, the proposed methods reduce the 95th percentile of the absolute percent AADT estimation errors (P95) by 0.5 to 31.9 when applied to different testing sites.