4. Conclusions
Diffusive BTX samplingwas proved an efficient
technique for revealingthe spatial variability of air
pollution in a complex roadside environment in central
Paris. The air circulation within Av. Leclerc gave rise to
relatively high pollutant concentrations on the leeward
side of the street. Nevertheless, winds parallel or nearparallel
to the street axis induced the highest roadside
concentrations, while perpendicular winds generally
reduced pollution levels (especially on the windward
side of the street). Observations within Av. Leclerc
showed that there was no substantial reduction in
ambient benzene concentrations with height above
second floor level on either side of the street.
Although there was no marked seasonal variation in
the observed street-level values, higher pollutant concentrations
were mainly associated with periods of
relatively low southerly winds, which probably transported
polluted air masses from Bd. Pe´riphe´rique
towards Pl. Basch. Concern was created over the longterm
NO2 and benzene concentration averages, which
seemed to exceed the annual EU limit values at all
roadside samplinglocations. However, the BG benzeneconcentrations remained below the EU limit value
duringthe same time period.
The AIRPARIF monitoringstation in Pl. Basch has
recorded the highest CO and NOx concentrations during
recent years in the region of Paris. The AIRPARIF
values were generally much higher than the concentrations
observed within the asymmetric canyon segment of
Av. Leclerc duringan intensive 1-week monitoring
period. Given the sitingof the permanent monitoring
station and the pronounced spatial variability of air
pollution in its vicinity, it can be concluded that the
measurements from this site do not give a representative
picture of air quality in the surroundingarea and are
thus inappropriate for population exposure studies.
The relatively simple dispersion models tested in this
study failed to convincingly reproduce the observed
roadside benzene concentrations. Two key parameters
not properly treated within these models were the
differential height of the canyon walls and the urban
vegetation. Parametric models should be adequately
calibrated when applied to complex street canyons.
More sophisticated modellingapproaches may be also
considered when assessingair quality in this kind of
locations.
Finally, the experimental methodology described in this
paper may be seen as a practical approach to urban air
quality management that avoids costly monitoring as well
as excessive reliance on models. More field campaigns of
this kind would help to assess the representativeness of air
quality monitoringstations and revise the existingsiting
criteria for roadside receptors