Environmental Engineering Reference
In-Depth Information
modify grid-averaged concentrations leading to biased model results (Krol, 2000).
A first approach may be a deterministic downscaling to higher resolution. It has
been shown that uncertainties, including emission and meteorological input,
accumulate at high resolutions and model error may increase with resolution
(Valari and Menut, 2008). Sub-grid scale parametrizations have been proposed to
account for the impact of the unresolved scale on Reynolds average fields (Vinuesa
and de Arellano, 2005; Galmarini, 2007). However, these applications may model
concentration variability but they do not provide direct information on what the
'real' concentration would be over specific areas inside grid-cells.
In the present study we propose a computational scheme that allows the
estimation of concentrations over micro-environments inside grid-cells, where
specific emission activity dominates. Fractional land-use data were used to divide
grid-cell areas into source-specific emission surfaces (e.g. roads, residential districts,
parks, etc.). Furthermore, we calculated the contribution of each different emission
activity in the grid-averaged flux by using emission data at sub-grid scale. The
separate contributions are then mapped on the sub-grid emission surfaces creating
a set of emission scenarios. Each scenario is accounted for individually during the
CTM simulation leading to a set of concentrations for all of model species.
Modelled concentrations, under the influence of each scenario represent concent-
rations over sub-grid surfaces. The variability of pollutant concentrations around
the commonly modelled grid-averaged concentration can be expressed as the
standard deviation of the concentrations over the considered micro-environments.
The described methodology is based on the consideration than urban chemistry
is faster than the mixing of emissions over grid-cell surfaces. Emitted species over
each source-specific surface are therefore, not diluted in the entire grid-cell
volume. Air mass is transported from the neighboring cells inside each micro-
environment, where it comes in contact with local emissions. Chemical transform-
ations take place before emissions are mixed. This treatment is particularly
advantageous in what concerns fast chemical reactions such as NO oxidation by
O
3
. At the urban environment high NOx emissions are released over the traffic
network. Under the common CTM assumption of instant and homogeneous
mixing O
3
consumption in the O
3
+ NO reaction is overestimated because grid-cell
surface is larger than the corresponding emission surface.
The method was validated in an idealized case. It is commonly argued that
1 km is the lower limit of validity for CTMs. In practice, however, and due to the
high computational time and uncertainties in input data, the large majority of
urban air-quality models run at lower resolution (~3 km). We created a simulation
domain of 1 km resolution centered around an emission domain typical of the
urban environment. We considered that at 1 km grid-cell size emission surfaces
may be represented by a single activity such as traffic, residential, parks or
industrial. A first simulation is ran over the domain resolving explicitly emissions
variability at 1 km. The obtained variability in modelled concentrations is thus
considered to be the maximal variability that a CTM can represent over a city.
At a second stage, we run a simulation over the same domain but at lower
resolutions (3, 6 and 12 km). Instead of averaging emission fluxes at the coarser
