Estimation of Dynamic Traffic Densities for Official Statistics

Abstract

Traffic density is one of the important variables to identify traffic states. Traffic management and traffic control require real-time estimation of traffic density as an input for large spatial and temporal coverage of the road network. Statistics offices responsible for the production and publishing of official traffic statistics for the benefit of policy makers often do not consider the use of this rich information source for a variety of reasons. The objective of this study is to theoretically link current attempts to exploit real-time monitoring information from the domain of dynamic traffic management with the usual practice of producing traffic statistics by statistics offices. Specifically, the study proposes a theoretical model for lane-level traffic density estimation for official statistics based on the use of data from the Global Positioning System (GPS) and loop detectors. Currently, the adoption of GPS data is seriously hampered because few vehicles are equipped with GPS transponders. By combining loop detector and GPS data, one can benefit from the relative quality of both data sources. In the current model, the definition of traffic density is applied on the basis of the cumulative traffic counts from loop detectors during the travel time of GPS vehicles that pass by the loop detectors. Furthermore, estimators are presented of the mean traffic density in the population by upscaling to the relevant parts of the road network and certain time periods. Statistical weighting strategy is applied for both temporal upscaling and spatial upscaling.