A dynamic traffic forecasting application on the Amsterdam beltway

This article presents some theoretical developments that have resulted in a dynamic traffic forecasting procedure that is feasible both from a technical viewpoint (data availability) and from a practical viewpoint (data preparation). The procedure consists of a dynamic origin-destination (OD) matrix estimation module and a dynamic traffic assignment module. The OD-estimation module is an extension of traditional (static) OD-estimation methods, i.e. production-attraction models combined with the use of a deterrence function. To make the procedure computationally feasible an efficient parameter estimation method has been provided. To test the combined OD-estimation dynamic assignment model, data were collected continuously during 3 weeks at 141 locations on the beltway of Amsterdam. As an alternative to the proposed procedure, historical averages have been compiled from all observed data. Comparisons between true, predicted and averaged data show that a lot of effort must be investigated in specifying OD-demand and network characteristics in order for the new method to be competitive with historic averages for traffic forecasting. Exceptional circumstances such as severe incidents, however, are reported better with the dynamic forecasting procedure.