Reducing Uncertainty In The Estimation Of Heavy Truck Vehicle Miles Traveled Vmt

Download or read book Reducing Uncertainty in the Estimation of Heavy Truck Vehicle Miles Traveled (VMT) written by Stephen Lamptey and published by . This book was released on 2004 with total page 248 pages. Available in PDF, EPUB and Kindle. Book excerpt: A variety of transportation related applications such as safety, geometric and pavement design of roadways depend on reliable estimates of heavy truck VMT. Truck VMT estimation methodologies used by state DOTs fall under two broad groups; the non-traffic count based and the traffic count based methods. The latter is the most preferred among state DOTs as it utilizes actual data of vehicle movement on a road segment. Resource constraints, however, make it impactical for the monitoring of all road sections of interest continuously throughout the year. State DOTs therefore maintain a traffic count program comprising of permanent counts where data are collected all year round and short-term counts usually collected for periods up to 48 hours. The short-term counts do not represent an average annual daily count. Conversion of the short-term count data to annual average daily estimates are achieved by applying adjustment factors developed from the permanent count data to account for temporal variations. It has been observed that trucks do not exhibit the same temporal variation patterns as passenger vehicles, however, the current practice by a majority of state DOTs involve the use of adjustment factors derived from aggregate volume data not truck volume which fail to adequately explain the temporal variations in truck traffic resulting in biased annual average daily truck traffic (AADTT) estimates. This research utilizes Automatic Traffic Recorder (ATR) data for the rural primary roadways in Iowa to compare three AADTT estimation methods; truck adjustment factor, annual truck percentge and count specific truck percentage methods. An assessment of the acuracies of the 3 methods is made using the estimates of prediction error obtained from cross-validation. Pairwise comparison fo the methods is done by using the non-parametric bootstrap statistical analysis.