STATISTICAL ANALYSIS OF DAY-TO-DAY VARIATIONS IN REAL-TIME TRAFFIC FLOW DATA

In the absence of intelligent vehicle-highway system technologies, commuters tend to select their routes through a congested network primarily on the basis of expected average link travel times. For this average to be representative of the current day, it is essential that the traffic conditions be relatively similar each day. However, if the traffic conditions vary considerably from one day to the next, the historical information will be insufficient for commuters to find the optimum routes through the network, and the provision of real-time traffic information could provide major benefits. Furthermore, simulation is becoming an important tool in evaluating different traffic control strategies. As a result it has become more and more important not only that the average typical traffic conditions be established but also that the upper and lower bounds of these average conditions be estimated. Consequently, two related issues are examined : the spatial and temporal magnitude of the variability in traffic conditions during typical nonincident conditions, and the magnitude of this variability during incident conditions. It was shown that in the absence of incidents, the temporal and spatial variations in traffic conditions were very similar for weekdays but varied considerably relative to the typical conditions during weekends. Major incidents, however, were found to alter drastically the average recurring conditions, thus creating a window of opportunity for achieving travel benefits by using dynamic data in real time.