A numerical model named AZUR has been developed by IFP-EDF-LISA to simulate
regional episodes of photochemical pollution. This model requires data on
pollutant emissions defined at space and time scales corresponding to one s
quare kilometre and one hour. In addition to the amount of emitted pollutan
ts, the knowledge of their chemical composition is of prime interest for co
rrectly representing air chemistry through detailed modules such as MOCA. I
t is therefore important to know the emitter type in order to assign an app
ropriate chemical composition to various pollutant flows. IFP has developed
a set of methods whose aim is to determine an inventory of traffic-related
pollutants (CO, SO2, particles, VOCs and NOx). They are based on the knowl
edge of the vehicle density and average traffic flow at the site of each me
asurement point. Density and speed data were converted into pollutant flows
using emission factors (in grams of pollution per km travelled) as functio
ns of speed. The emission factors are specific to each vehicle category. Se
parate counts for each category enabled us to assign, as previously mention
ed, a specific chemical composition (in particular for VOCs) to emissions i
n each category. But very few data are available regarding the VOC emission
speciation as a function of different vehicle types. Moreover, these data
are very often associated with a specific driving cycle, which does not nec
essarily correspond to cycles used for determining emission factors. The pu
rpose of this paper is to evaluate the impact of these uncertainties in VOC
speciation. Air quality in the Paris area will be simulated using the AZUR
model with various VOC distributions. These distributions, taken from the
literature, are either measurements carried out using specific driving cycl
es or overall values. Ozone profiles will be analysed and compared in both
urban and rural areas to assess the sensitivity of results to the VOC distr
ibution related to road traffic.