Road safety performance measures and AADT uncertainty from short-term counts
Related Experiment Video
Updated: Mar 14, 2026
Evaluation of an Exclusive Spur Dike U-Turn Design with Radar-Collected Data and Simulation
Published on: February 1, 2020
Craig Milligan1, Jeannette Montufar2, Jonathan Regehr2
1University of Manitoba Department of Civil Engineering, E1-327, 15 Gilson Street, Winnipeg, MB, R3T 2N2, Canada; Fireseeds North Infrastructure, 200-135 Innovation Drive, Winnipeg, MB R3T 6A8, Canada.
Objective:
The objective of this paper is to enable better risk analysis of road safety performance measures by creating the first knowledge base on uncertainty surrounding annual average daily traffic (AADT) estimates when the estimates are derived by expanding short-term counts with the individual permanent counter method.
Background:
Many road safety performance measures and performance models use AADT as an input. While there is an awareness that the input suffers from uncertainty, the uncertainty is not well known or accounted for.
Method:
The paper samples data from a set of 69 permanent automatic traffic recorders in Manitoba, Canada, to simulate almost 2 million short-term counts over a five year period. These short-term counts are expanded to AADT estimates by transferring temporal information from a directly linked nearby permanent count control station, and the resulting AADT values are compared to a known reference AADT to compute errors. The impacts of five factors on AADT error are considered: length of short-term count, number of short-term counts, use of weekday versus weekend counts, distance from a count to its expansion control station, and the AADT at the count site.
Results:
The mean absolute transfer error for expanded AADT estimates is 6.7%, and this value varied by traffic pattern group from 5% to 10.5%. Reference percentiles of the error distribution show that almost all errors are between -20% and +30%. Error decreases substantially by using a 48-h count instead of a 24-h count, and only slightly by using two counts instead of one. Weekday counts are superior to weekend counts, especially if the count is only 24h. Mean absolute transfer error increases with distance to control station (elasticity 0.121, p=0.001), and increases with AADT (elasticity 0.857, p<0.001).
Implications:
These results can support evidence-based risk analysis of road safety performance measures that use AADT as inputs. Analytical frameworks for such analysis exist but are infrequently used in road safety because the evidence base on AADT uncertainty is not well developed.
