Environmental Engineering Reference
In-Depth Information
resolution cells, as is done in the usual CTM modelling, we applied the proposed
emission scheme. We thus model pollutant concentrations implicitly over the sub-
grid areas assigned as 'roads', 'residential', parks and 'industrial'. The implicit
calculation of concentration variability is compared to the explicitly resolved one
at all resolutions.
2. Results
The comparison showed that for primary emitted species (NO
2
and PM10) the
emission scheme is able to reproduce concentrations over the sub-grid environments
that are very close to the concentrations modelled explicitly at 1 km resolution.
The variability represented by the implicit calculation reaches the 90% of the
variability resolved by the model at 1 km. For secondary species (i.e. ozone), the
implicit calculation underestimates more significantly the variability resolved by
the model at 1 km (~70% of the 1 km variability is represented). However, the
model is able to dissociate the fast ozone depletion in the traffic environment from
a state closer to the photo-stationary equilibrium over the residential micro-
environment.
The stability of the calculation, with regards to the model time-step, was tested
and it was shown that modelled variability was not affected by this parameter.
This means that inside each micro-environment modelled concentrations reach a
steady-state before getting mixed to the grid-averaged value. We also tested the
effect of the wind speed ant the intensity of turbulence on modelled variability. It
was shown that in both cases (implicit and explicit calculation) modelled variability
behaved in the same manner towards changes in wind speed and the boundary
layer height.
Finally the emission scheme was used in a real-case simulation during the
summer 2006 (1 June-31 August) over the city of Paris and model results were
compared with measurements. This validation over real data was focused in the
comparison of modelled concentrations over the 'traffic' or the 'residential' environ-
ments with measurements at stations co-existing inside the same grid-cell and
characterized by the local air-quality network as 'traffic' or 'background'
respectively. First of all we show that for primary species (NO
2
and PM10) the
recombination of sub-grid concentrations to a 'grid-averaged' value provides
practically the same concentration as the 'usual' model calculation. This means
that any the extracted information on concentration variability is a clear 'gain'
since the mean concentration remains unaltered. For ozone the difference between
the recombined mean concentration and the 'standard' grid-averaged value can go
up to 10%. This is due to the non-linear aspect of ozone chemistry.
The comparison of modelled results with measurements showed that the
emission scheme is able to generate concentration fluctuations around the grid-
averaged value that are in all cases closer to the measurement at the corresponding