During the next 15 to 20 years, the primary methods available to manage the nation's freeways will involve increasing the average occupancy of vehicles using the freeways. Because existing vehicle occupancy measurement is typically project specific and sporadic and uses different collection procedures, it is often difficult to evaluate the effectiveness of programs intended to increase the average vehicle occupancy (AVO). The research discussed in this paper was designed to address the lack of consistent and ongoing AVO measurement in the Seattle area. Because a literature search revealed that little had been published concerning the methodology of collecting vehicle occupancies, the research attempted to determine what factors lead to inaccurate or unusable data and how much data are necessary for accurate counts. A number of observation sites, including six freeway sites, were selected. Three people counted occupancies in the same lane at the same time for 111 15-min periods. Portable computers were used for data collection. The observations were time stamped to match the counters' observations on a vehicle-by-vehicle basis. By examining the agreement or disagreement among the counters, researchers determined the error levels and related them to a variety of factors, including weather, time-of-day, light levels, speed, observation location, counter comfort, vehicle weaving, and length of time counting. Significant relationships between error rates and these factors were used to identify elements that influence the accuracy of a counting program. The research used the error rates to estimate the statistical accuracy of an AVO sampling procedure. The study then examined the cost and administrative feasibility of implementing a continuous automobile vehicle occupancy counting program.
Abstract