Accuracy of distance-based travel time decomposition in probe vehicle systems

Probe vehicle data are commonly used for areawide measurements of travel time in road networks. In this context, travel times usually refer to fixed edges of an underlying (digital) map. That means, measured travel times have to be transformed into so-called link travel times first. This paper analyzes a common method being applied for solving this task (distance-based travel time decomposition). It is shown that, in general, its inherent imprecision must not be neglected. Instead, it might cause a serious misinterpretation of data if potential errors in context of travel time decomposition are ignored. For this purpose, systematic as well as maximum deviations between "decomposed" and "true" link travel times are mathematically analyzed. Three examples ranging from pure theory to real world confirm the analytical findings and underline the problems resulting from distance-based travel time decomposition at local level, i.e. at individual intersections