Geoscience Reference
In-Depth Information
8.4
The Uncertainties in the Simulated Dust Mass Budget
An illustration of the impact of the weak constraints in dust modelling has been
provided by the AeroCom model intercomparison exercise (Textor et al. 2006 ,
2007 ; Huneeus et al. 2011 ; see also Chap. 9 ) . The aim of AeroCom is to analyze
current global aerosol simulations based on harmonized diagnostics and to provide
an evaluation of various models based on a model intercomparison and a comparison
of these models to selected observations of aerosol properties (Textor et al. 2006 ).
Among other aerosol species, comparisons for dust modelling are included. Most
of the models participating in this intercomparison exercise used parameterization
for deposition similar to those presented in the previous sections. The output
of this intercomparison exercise are freely distributed ( http://nansen.ipsl.jussieu.
fr/AEROCOM ) and used in the following to illustrate the present state of the
uncertainties remaining in global dust models.
Fourteen research groups provided dust simulations to AeroCom. In the so-called
Experiment A (Textor et al. 2006 ), each model worked in its nominal conditions, i.e.,
with its own parameterizations for emissions, transport, deposition, size distribution,
etc. However, for some of these simulations, no information was provided on the
size distribution prescribed in the simulations, while for some other models, the size
distribution does not account for a supermicron mode. Since size distribution is a
key factor for dust deposition, the set of models for the intercomparison is restricted
to the seven models with a relatively similar size distribution. Figure 8.5 ashows
the simulated emissions in Mt/year for these models. It confirms that in global dust
models, simulated dust emissions have large uncertainties as indicated by a ratio
of
4.1 between the lowest (
975 Mt/year) and the highest emissions (
4000
Mt/year).
Surprisingly, despite such a large difference in dust emissions, the models are
more or less in agreement in terms of dust load (Fig. 8.5 b), since the ratio between
the lowest (
29 Mt) is only
1.7, i.e., significantly lower than that for dust emissions. This low uncertainty in
the simulated atmospheric dust load can be explained by the best observational
constraints on this parameter (see Sect. 8.1 ), against which dust models are usually
tuned. The adjustment of dust models with respect to the atmospheric dust load
induces compensative effects: the simulated dust life time exhibits a similar range
of variation between models than that observed for dust emissions (4.1) and models
having the highest (lowest) emissions have the shortest (longest) lifetime (Fig. 8.5 c).
To go further in the model evaluation, the AeroCom group proposed an additional
Experiment B, for which the emission strength, the initial dust size distribution
and the injection height were imposed to be the same for all models. Figure 8.6
reports the simulated annual dry and wet deposited mass (in Mt year 1 )forthe
six models that ran both experiments A and B (note that two models that had
performed Experiment A but were omitted from the comparison because of their
dust size distribution are reintroduced here, the size distribution now being identical
for all models). For Experiment B, the differences (expressed in terms of ratio)
17 Mt) and the highest atmospheric dust load (
Search WWH ::




Custom Search