PREDICTING URBAN ARTERIAL ORIGIN-DESTINATION TRIP TABLES

One major requirement in the process of managing and controlling the flow of traffic in urban arterials is knowledge of the origin-destination trip table. The elements of such tables represent the proportion of traffic originating from a particular origin and destined to different destinations. Estimating urban arterials' origin-destination tables ahead of time provides traffic engineers with the necessary information to select the appropriate management and control strategy. This study aims to estimate urban arterial origin-destination trip tables by specifying and estimating a multinomial logit model. It is hypothesized that each driver located either on an on-ramp or arterial's main line has a set of exits from which to choose. As such, the origin-destination problem is formulated for the first time as an "exit-choice" problem. Two other methods were specified to estimate the origin-destination trip table. These methods are ordinary least squares and constrained least squares. To estimate the specified models, time-dependent traffic counts were obtained for each exit and entrance for one major urban arterial in Amman, Jordan's capital. In addition, the actual choice of exits were recorded for a random sample of drivers. Estimation results clearly show the multinomial logit model provided the least sum of squared difference between the observed and estimated trip table. Furthermore, the logit model satisfied the conditions namely that each element of the trip table is positive and that the sum of all proportions from an entrance to all other exits add up to unity. Ordinary least squares estimates failed to satisfy the above conditions and provided very high sum of squared differences. Although constrained least squares estimates satisfied the above conditions, they still provided high sum of squared differences when compared with the multinomial logit model estimates. The developed "exit-choice" method can be used easily by city traffic engineers to estimate the origin-destination trip table since it requires limited traffic-related information. As such, this method is appropriate for developing countries with limited financial and technological resources.