Framework for analysing reliability and information degradation of demand matrices in extended transport networks

This paper examines a methodological framework for improving the estimation of current and future O-D demand matrices. The problem of O-D matrix estimation is investigated for the case of extended urban transport networks, where the topological complexity and high variability of the prevailing traffic conditions result in the rapid degradation of the information concerning the underlying O-D demand patterns. The paper aims to contribute to the development of a set of analytical tools for interpreting the loss of the resulting O-D matrix reliability and the extent and sources of the information degradation. The suggested framework treats in an appropriate way the short-term systematic variations of prior demand information and, hence, increase the consistency and predictability of the within-day time-dependent O-D matrices. In addition, it takes into account the long-term dynamics underlying the degradation of O-D information by means of equilibrium analysis of the evolving O-D flows over a series of day-of-the-week. In this way, changes in the reliability thresholds of current O-D matrices may be estimated to enhance the predictability of daily demand flows.