With increasing data availability due to intelligent transportation systems (ITS) deployments, methods for assessing and reporting traffic characteristics and conditions have begun to shift. While previous level of service (LOS) methods were developed for use with limited data, we now have the power to develop and test the use of actual performance measures. Important measures like average speed, travel time, and intersection delay can be used for performance monitoring of the transportation system. These measures are useful for system management, planning and for users. On freeways, such performance measures are often estimated directly using data from inductive loop detectors (e.g., speed, occupancy, vehicle counts). For arterials with numerous signalized intersections, performance measures are more challenging due to more complicated traffic control and many origins and destinations. However, within signalized networks, travel time, speed, and other key performance measures can be obtained both directly and indirectly from sources such as automatic vehicle location (AVL) data. In this paper, we demonstrate how AVL data can be used to characterize the performance of an arterial. First, we extract data from the bus dispatch system (BDS) of the Tri-County Metropolitan Transit District (TriMet), the transit provider for Portland, Oregon. Then, the performance characteristics as described by bus travel on an arterial are compared with ground truth data collected by probe vehicles equipped with global positioning systems (GPS) sensors traveling with normal (non-transit) traffic on the same arterial on the same days. Comparisons are drawn between the two methods and some conclusions are drawn regarding the utility of the transit AVL data. INTRODUCTION Freeway performance characteristics are relatively well-understood. However, arterials are characterized by complicated traffic behavior and many more variables than are associated with freeways. For arterial performance measurement, traffic conditions are often evaluated using test vehicles to collect travel time and delay data (1). However, these travel time and delay studies are limited temporally and spatially, time consuming and expensive. Test vehicles and personnel may be dispatched to collect travel time data for one peak period on only one day. With the increasing intelligent transportation systems (ITS) deployments, the floating probe vehicle technique can play an important role for collecting data in real time. Probe vehicles respond to changes in traffic flow as they traverse the network and can transmit location and travel time data to a traffic management center at frequent time intervals (2). As in the case of a transit fleet, these floating probes may already be in the traffic stream. Most transit automatic vehicle location (AVL) systems are used primarily for managing operations in real time. Past research has used transit AVL data to test possible congestion monitoring and transit information uses (3, 4). BACKGROUND There has been heightened interest in providing performance measures along arterials, both in the context of advanced traffic management system (ATMS) and advanced traveler information system (ATIS). In Portland, Oregon, the Tri-County Metropolitan Transportation District of Oregon (TriMet) provides transit service in the metropolitan area. During weekdays, more than 600 TriMet buses traverse most major arterials during peak periods (5). These vehicles are equipped with a bus dispatch system (BDS) which includes AVL, comprised of differential global positioning systems (GPS), automatic passenger counters, wireless communications, and stop-level data archiving capabilities. The BDS provides a rich source of accurate time and location information. Since the buses are already in the traffic stream, they can be used as probe vehicles for collecting travel time data. The BDS records bus arrival and departure times at each geo-coded stop, and records the maximum instantaneous speed achieved between stops. As a result, TriMet, the Oregon Department of Transportation (ODOT), and the City of Portland are developing plans to use BDS data for ATMS and ATIS purposes (6). The extent to which the travel characteristics of buses are related to those of general traffic is not well understood. Therefore, a comparison of the transit bus data and ground truth data collected by GPS-instrumented passenger vehicles was conducted. Vehicle trajectories–graphs of vehicle location versus time–of both buses and non-transit test vehicles were produced to measure the differences in travel time and speed. To better understand this relationship, hypothetical and pseudo bus analyses were also investigated. Hypothetical buses are defined as the buses traveling non-stop and pseudo buses are buses traveling at the maximum speed recorded for each link. Speed contour plots were also used to observe the precise differences in speed for both types of vehicles traveling along a study corridor. By estimating speeds throughout a road segment at any particular time, speed contours were plotted on a three-dimensional graph using time and location as the xand yaxis respectively. The next section contains a description of the study corridor and the two sources of data used in this study. Bertini and Tantiyanugulchai 3
[1]
J C Oppenlander,et al.
SAMPLE SIZE DETERMINATION FOR TRAVEL TIME AND DELAY STUDIES
,
1976
.
[2]
Herbert S Levinson,et al.
ANALYZING TRANSIT TRAVEL TIME PERFORMANCE
,
1983
.
[3]
Darcy M. Bullock,et al.
DETERMINATION OF SAMPLE SIZES FOR TRAVEL TIME STUDIES
,
1998
.
[4]
William H. Crown,et al.
Statistical Models for the Social and Behavioral Sciences: Multiple Regression and Limited-Dependent Variable Models
,
1998
.
[5]
Steve Callas,et al.
Evaluation of transit operations: data applications of Tri-Met's automated Bus Dispatching System
,
2002
.
[6]
M. Goodchild,et al.
Geographic Information Systems and Science (second edition)
,
2001
.
[7]
William L Eisele,et al.
TRAVEL TIME DATA COLLECTION HANDBOOK
,
1998
.
[8]
J. Dwyer,et al.
Statistical Models for the Social and Behavioral Sciences
,
1983
.
[9]
Richard A. Johnson.
Miller & Freund's Probability and Statistics for Engineers
,
1993
.