Using multi-source GPS data to characterize multiday driving patterns and fuel use in a large city region

Abstract

The paper describes the use of GPS data obtained from both commercial and project-specific sources to examine the travel behavior and fuel consumption patterns of drivers over a three-day period in Gauteng Province, South Africa. Data for commercial (truck and light delivery vehicle) traffic are obtained from a commercial fleet management provider, which continuously tracks the movements of 42,000 vehicles. Data for private car users come from a panel of 720 drivers, whose multiday driving activity is tracked using mobile passive GPS loggers. We analyze and compare the driving behavior of the two driver populations in terms of total distance travelled, spatial patterns (e.g. the amount of travel on different road types) and temporal variations (e.g. variations across time of day and across multiple days). The detailed nature of GPS data also permits the estimation of fuel consumption at a very disaggregate level (by link and time of day), and the identification of differences between user groups, which have significant implications for transport and energy policy. We introduce a new indicator, the recovery ratio, to assess the relationship of fuel use to distance travelled on different classes of roads, to help identify equity distortions across user groups. Lastly, we comment on research needs related to the collection and integration of GPS data from multiple sources for model calibration and program evaluation.