A framework for estimating origin-destination trip tables from observed traffic data: A quality assessment methodology

As a representation of travel demand in transportation networks, reliable origin-destination (O-D) trip tables are crucial to economical and efficient transportation planning, design, and management. So far, the estimation of O-D trip tables from traffic counts provides a quick, inexpensive, and theoretically-sound alternative to the traditional large-scale, survey-based techniques. The main objective of this study is to develop a framework that rigorously considers several components of the estimation problem, including the input component, the estimator, and the output component, as well as their interaction effects on the estimation results. In this study, several extensions and enhancements have been made on the three components with emphasis on developing a new methodology for assessing the quality of an O-D trip table estimate.; First, the nonlinear Path Flow Estimator (PFE), one of the efficient techniques for estimating O-D trip tables from traffic counts, was reformulated to properly handle measurement errors and inconsistencies of traffic counts. This modification is essential to minimize any systematic bias of the total demand estimate encountered in the original PFE model. Next, the traffic counting location model was used to identify the number of traffic counts and their corresponding locations required to completely observe trips of certain O-D pairs. The results indicate a significant improvement in the quality of the O-D trip table estimate when proper traffic counts are used. With proper selection of traffic counting stations, the number of traffic counts required to observe trips of certain O-D pairs is significantly reduced when compared with the case where traffic counts are subjectively selected. Finally, a new quality assessment method is proposed to quantify the reliability of estimated demands at different spatial levels including O-D pair level, zonal level, and network level. By using the proposed method, confidence intervals of the estimated demands can be constructed to account for the uncertainty due to the multiplicity of O-D trip table estimates as well as the uncertainty due to possible errors of traffic counts. The new developments proposed in this study have been demonstrated with various synthetic networks and applied to a real road network in the City of Irvine, California.