Two new methods for estimating trip matrices from traffic counts

This chapter describes different methods for estimating traffic models and trip matrices. Most conventional methods assume that either traffic counts are error-free, deterministic variables or they use a simplified traffic assignment model. Without these general assumptions, the methods often demand prohibitive calculation times. In order to overcome these problems in practice, two matrix-estimation methods are formulated and discussed in this chapter. Both methods are able to handle traffic counts with inconsistencies and uncertainties. The estimated trip patterns reflect the route choice patterns given by traffic assignment models following the Method of Successive Averages (MSA), including Stochastic User-Equilibrium (SUE). The first method, for Single Path Matrix Estimation (SPME), is easy to implement as it utilises existing implementations of traffic assignment models. SPME is, however, very heuristic in nature and utilises only the counts along the optimal path between each zone-pair. Therefore, a second method, Multiple Path Matrix Estimation (MPME) is also presented. MPME utilises all counts along paths according to MSA and, thus, reflects the route choice better than SPME. However, MPME is more difficult to implement, An examination of the methods was carried out on several full-scale cases, where both SPME and especially MPME gave very reasonable results. For practitioners, the methods are most promising in cases where an old trip matrix needs to be updated for use in sketch-plan models, or as a pivot matrix in larger traffic models. The final part of the chapter deals with the promising prospect for cost-efficient traffic modelling in practice, when using the methods presented in the study.