Time Slicing Origin-Destination Matrices Using Mobile Network Data, Link Counts and Vehicle Probe Data

Origin-Destination (OD) matrices are essential inputs to both traffic planning and management applications. A time-sliced OD matrix, which reflects traffic patterns over different times of the day, is important for understanding peak-hour traffic and for managing congestion effectively. In this paper, we develop a method that uses mobile network data, link counts, and vehicle probe data, within a data-driven network assignment (DDNA) process, to estimate a time-sliced OD matrix from an average day OD matrix. The method is evaluated on empirical data for Norrkoping, Sweden, demonstrating that the method provides link flow estimates that are consistent with link flow observations for both training and test sets while maintaining structure from the mobile network data and the transport planning model. Compared to benchmark approaches, the proposed method yields significantly improved results for both training and test sets. The resulting matrices are suitable for more detailed traffic analysis and can serve as input to dynamic traffic models for planning and management applications.