Despite the progress in reducing emissions from stationary and mobile sources, the overall air pollution levels in urban areas have not improved significantly over the last two decades. In particular, traffic emissions are still a major problem, since the success of various implemented control measures has been compensated by the pronounced increase of miles traveled. Consequently, dispersion modeling of traffic emissions in urban areas has been one of the main research areas over the last few years. A variety of advanced dispersion models have been developed that are able to resolve wind-driven, microscale flow and dispersion phenomena. Such phenomena, like for example a vortex rotating in street canyons, are typically observed for roof-level wind velocities above a threshold value of the order of 2-3 m/s and strongly affect the mean and turbulent transport of pollutants released inside the urban canopy. The street ventilation is then controlled by the interaction between the microscale flow structures and the urban boundary layer flow above roof level.

However, despite the progress in reproducing canopy flow patterns in urban dispersion models, previous studies have shown that urban pollution forecasts are still not very accurate. Significant concentration overestimations are a major problem, especially for low wind speed conditions. The fact that only wind-driven ventilation processes are considered, but turbulent motions mechanically generated by traffic are typically ignored or seen as being of only peripheral importance, is one of the possible reasons for this deficiency. Although several recent studies indicated that these motions play an important role in the dispersion of traffic emissions, their significance for urban dispersion modeling has been a matter of dispute.

The present paper summarizes the findings obtained over the duration of the TRAPOS research network, regarding the importance of traffic-produced turbulence (TPT) for operational dispersion modeling. Improved TPT parameterizations and their implementation in practical models will be discussed. Finally, an extensive analysis of urban concentration datasets will be presented, which clearly shows that TPT parameterizations significantly improve concentration predictions for the worst air pollution episodes.