Chemistry Reference
In-Depth Information
secondary organic compounds (SOA), which processes continue to be involved in a
high level of uncertainty.
Also special care should be taken to reduce uncertainties on emission data and
measurements. The validation of an aerosol model requires the analysis of the
aerosol chemical composition for the main particulate species (ammonium,
sulphate, nitrate and secondary organic aerosol). To find data to perform this kind
of more complete evaluation is not always easy. The same applies to emissions
data. The lack of detailed information regarding the chemical composition of
aerosols obliges modellers to use previously defined aerosols components
distributions, which are found in the literature. Present knowledge in emission
processes is yet lacunal, especially concerning suspension and resuspension of
deposited particles [
37
].
Moreover, the large diversity of environments and sources of aerosols makes
difficult to obtain a general assessment of models performance for a given model all
over a large region like Europe, simultaneously at urban and regional scales.
Northern Europe undergoes a very different climate than southern Europe, and
sources mostly result from combustion in large cities and industrial areas, while
southern Europe may be significantly influenced by wind-blown dust.
The issue of modelling scales and objective is very important. City-Delta
exercise in which mesoscale models were used to calculate PM concentrations in
urban areas allowed concluding that finer-scale models show better performance
for PM
10
in the cities than large-scale models. However, these improvements
are limited because these models do not generally use small-scale meteorology,
and still have limited vertical resolution [
37
]. A further increase in horizontal
resolution, aiming to reach urban scale, may be necessary to increase further
the skill. Dynamical downscaling, from regional till local scale, could be a way to
increase the simulation resolution, taking into account the influence of large-scale
phenomena in micro-scale.
Improvements of aerosol modelling performance have been focused on data
assimilation of PM and aerosols species. However, they have so far mainly been
indirectly assimilated via aerosol optical depth (AOD), either using satellite data
alone, or in combination with ground-based data. This is confounded by a number
of difficulties due to the complex character of the different constituents of PM.
It is thus not possible to use such assimilation systems yet to assess the PM
concentrations in cities or urban areas. Before using more sophisticated
techniques of data assimilation it is of vital importance that the model is thor-
oughly evaluated and validated, using proper model input data and measurements.
This part of the work should not be underestimated, as it often forms the key to
success in combining monitoring and modelling. Ideally, the model should show
little or no bias as compared to the measurements, i.e., the model should not
underestimate or overestimate concentrations systematically, or as an average, as
compared to the measurements. Most data assimilation methods work best if there
are little or no bias between the model and the measurements. There should also
be a reasonable good time correlation (perhaps 0.5 or higher) between the two
before attempting to use such methods.
