Comparison of a road traffic emission model (HBEFA) with emissions derived from measurements in the Gubrist road tunnel, Switzerland

Road vehicle emissions contribute significantly to air pollution. These traffic emissions are characterized by emission factors (EFs), which describe the emitted mass of a compound per driven distance. Here, we present a comparison of EFs which were obtained with two different techniques: (i) Trace gas measurements in the Gubrist tunnel, Switzerland in fall 2002 (Colberg et al., 2005. Atmospheric Environment, in press) provided real-world EFs of entire vehicle fleets and, after statistical regression analysis, additionally the EFs of different vehicle classes. Furthermore, a 12 year time series of EFs of measurements in the Gubrist tunnel in Switzerland is being used (Steinemann and Zumsteg, 2003, Bericht US 89-16-11 fur OSTLUFT durch Amt fur Abfall, Wasser, Energie and Luft des Kantons Zurich (AWEL), Switzerland). (ii) Based on dynamometric test measurements and engine maps at different conditions, EFs of single vehicles are combined with a comprehensive road traffic emission model (HBEFA, 1999. Umweltbundesamt Berlin, Bundesamt Mr Umwelt, Wald and Landschaft Bern (published as software on CD-ROM), Infras AG, Bern; HBEFA, 2004. Umweltbundesamt Berlin, Umweltbundesamt Wien, Bundesamt fur Umwelt, Wald and Landschaft Bern (published as software on CD-ROM), Infras AG, Bern). The results of both techniques are processed and if enough information on the vehicle fleet is available, a comparable standard is achieved, which allows the validation of the road traffic emission model. Since two HBEFA versions are being used the improvement of the model can be shown nicely. The results demonstrate good agreement for the trends of the time series measurements. However, a deviation for the absolute EFs of NO, is observed for light- and heavyduty vehicles (LDV and HDV, respectively). The HBEFA, 2004 derived estimates are larger than 50% for LDV and 15% for HDV compared to the EF derived from the tunnel study. The deviation is even more prominent for the CO emissions (75% overestimation for LDV and 130% overestimation for HDV). The VOC emissions could only be compared for LDV. There the deviation is within the confidence interval of the measurements. The strengths and weaknesses of both techniques are revealed by this study. (c) 2005 Elsevier Ltd. All rights reserved.