Evaluation of an emission inventory by comparisons of modelled and measured emission ratios of individual HCs, CO and NOx

Air quality models require emission inventories of individual compounds with high temporal and spatial resolution. These emission inventories are calculated by complex emission models. To evaluate these models, emission ratios of non-methane hydrocarbons (HCs, containing only H- and C-atoms), CO and NOx of the city of Augsburg were determined during two field campaigns in March and October 1998. The measured emission ratios have been compared with emission ratios calculated by an emission model at two accuracy levels. Only C2–C10 HCs were measured quantitatively by the applied techniques. C11–C13 HCs were measured with losses increasing with the number of C atoms, oxygenated and halogenated volatile organic compounds were not measured at all. The direct comparison could be made only for those HCs whose individual emissions can be specified by the model. They represent 21% of the total predicted volatile organic compound emissions and—according to the model—originate to a large percentage from traffic-related emission. The measured HCi/HCsum emission ratios (HCsum Σ(HCi) for C2–C10-HCs and HCi mixing ratio of an individual hydrocarbon) for these compounds agreed within 30% for most compounds. The HCi/HCsum emission ratios measured in Augsburg are very similar to emission ratios from traffic-related sources suggesting that the emission model reflects the traffic emissions quite well. In order to consider all measured HCs for comparison the share of unspecified HCs, which are detectable by the GC system and should therefore be included in the experimental results, was estimated. Taking this into account yields a deviation between modelled and measured monthly mean values of HCi/CO and HCi/NOx of about 40%. The comparison of modelled and measured HCi/CO with HCi/HCsum emission ratios derived from the investigation based on single events confirms this finding: in this case an overestimation by the model of a factor of 1.5–2 was found. Since modelled and measured NOx and CO emissions agreed well within the uncertainty ranges, these deviations can be attributed to overestimation of HC emissions. Unspecified HC emissions mainly originate from solvent mixtures thus the overestimation of their emissions could explain the observed deviations. Regarding one specific HC which originates nearly exclusively from solvent emissions, i.e. n-decane, yields an overestimation of about a factor of 10 by the model.