In spite of the decrease of atmospheric pollution from transport following the technological development of vehicles and fuels, there still remains much to do to minimize critical events mainly due to ozone and the emission of very fine particles dangerous for health. CO2 emissions remain the main problem. The transport sector accounts for 26% of the EU's CO2 emissions, road traffic being the most important contributor by far. In particular in the urban areas, they account for more than 90% of the CO2 emissions from the local transport sector.
Concerning safety a reduction of 13% deaths over three years has been achieved in EU25, far from the 50% reduction foreseen by 2010; in addition about 74% of accidents occur on urban roads. Taking into account that the total social cost of the road is equivalent on average to approximately 4.2% of the EU GDP (1.5% for the infrastructure; 1.1% for the environment; 0.5% for the accidents and 1.1% for congestion) and that almost 80% of the European population is living in urbanized areas, it is straightforward to size the problems faced by cities administrators.
The key point to mitigate the impacts of urban traffic is the management of private vehicles flows, acting both on the demand side and the network regulation. This means the adoption of structural policies and set up of technologies (ITS) to implement both sides of the strategy.
In any case the crucial point relays on the capability of monitoring the flows on the network; by necessity measurements, either real time and off line, can be performed only on a very limited number of roads. Therefore, in order to reconstruct the status of flows on the entire network, traffic modeling plays an essential role, either to assess the scenario generated by traffic demand management strategies (off line modeling) or to monitor the efficiency of the network (near real time modeling).
In the presentation a detailed description of a system integrating the on line process from traffic data to pollutant doses on critical receptors will be given, highlighting the still open issues on traffic modeling.