Objective

Because of the large uncertainties in both traffic assignment and emission computation, there is a need for uncertainty quantification of the emissions that are computed from a modeled flow. This requires to propagate uncertainties in the traffic assignment model and then in the emission model.

Global sensitivity analysis

The first step we carried out was a global sensitivity analysis of the combined models (traffic assignment and emission model) for the city of Clermont-Ferrand (France). An extremely fast meta-model of the dynamic traffic assignment model LADTA was linked to the emission model Pollemission. As a result, it was possible to perturb all uncertain inputs and general large ensembles of emission simulations.

A global sensitivity analysis consisted in computing Sobol’ indices. Below is an example result where the Sobol’ first-order indexes are shown for various inputs, like demand between certain origins and destination, or the speed limits on the roads.

Such analysis allowed us to identify spatial patterns in the sensitivities, e.g., with the sensitivity of emissions to the proportion of heavy duty vehicles in the vehicule fleet.

Reference

R. Chen, V. Mallet, V. Aguiléra, F. Cohn, D. Poulet, F. Brocheton. Global Sensitivity Analysis of a Road Traffic Emission Model at Metropolitan scale. In preparation.