Setting Key Model Parameters for Microscopic Traffic Simulation Using Vehicle Trajectory Data

Microscopic traffic simulation represents one of the most popular tools for analysing and comparing traffic performance in different scenarios of urban mobility. The reproduction of real-world traffic dynamics is its primary requirement for providing time-dependent estimates. This study presents a process to build a microscopic traffic model for an urban area of Athens, developed using high-resolution vehicle trajectories obtained from the pNEUMA dataset. Based on drone-recorded trajectories, a simulation scenario was built, combining a realistic road network model, including traffic light regulation, an estimated traffic demand, and a set of parameters to replicate the observed vehicle behaviour. The modelling process relies on an iterative comparison between simulated outputs and observed vehicle-level trajectory data. The proposed approach evaluates travel time distributions and helps develop an improved model, enhancing its ability to replicate the vehicle’s behaviour. The final calibrated configuration reduced the Wasserstein distance by approximately 55.8% compared with the default SUMO configuration. The results also show that the calibration of microscopic behavioural parameters can substantially affect secondary simulation outputs, including emission estimates relevant for sustainability-oriented traffic analyses.