Dust Production Mechanisms - Mineral Dust: A Key Player in the Earth System - page 111

dotted lines), the two representations do not differ much. The largest differences
correspond to the finest dust fraction (<0.3 m) in the number size distribution
and the coarse dust fraction (>20 m) in the volume size distribution. This clearly
highlights the need for documenting the full size spectra of mineral dust at emission.

5.5.3 Models Versus Observations

While several conceptual approaches for dust emission have been developed, a
main limitation for their improvement is the availability of experimental data to
test and validate these different models. Most of the available datasets are total mass
dust emission fluxes measured as a function of U * (Gillette 1977 ; Nickling 1983 ;
Nickling and Gillies 1993 ; Nickling et al. 1999 ; Gomes et al. 2003 ; Rajot et al.
2003 ). Several of them include simultaneous measurements of the saltation and
dust emission fluxes, which limit some uncertainty in the simulation of the saltation
fluxes. All these comparisons lead to the conclusion that the developed models are
able to reproduce the measured total dust emission flux, after adjustment of the main
input parameters. Indeed, the soil properties required as input parameters in most of
the models are generally not provided in the experimental datasets. As a result, there
is always a combination of input parameters that allows reproducing the measured
dust fluxes. This leads to the conclusion that measurements of the dust emission
fluxes in terms of total mass only are not sufficient to constrain these models and
in particular the physical processes they aim to reproduce. From this point of view,
a key point is to test the capability of the physically explicit dust models to predict
the size distribution of the emitted dust.
This analysis has lead to the implementation of field campaigns dedicated to the
measurements of size-resolved dust emission fluxes such as recently made during
the JADE (Japanese Australian Dust Experiment; Ishizuka et al. 2008 ) and AMMA
(African Monsoon Multidisciplinary Analysis; Sow et al. 2009 ) programmes.
The size-resolved dust emission fluxes measured during AMMA show significant
differences between weak and intense dust events, with a much higher contribution
of the finest modes for the latter (Sow et al. 2009 ). The higher relative contribution
of the finest dust fraction to the total dust size distribution for high U * is in
agreement with the sandblasting models of Alfaro and Gomes ( 2001 )andShao
( 2004 ). However, the airborne dust size distributions measured during JADE (Shao
et al. 2011 ) appear to be independent of the wind speed. Such an invariant behaviour
is reproduced by Kok's model ( 2011b ), with a constant propagation length and
assuming a typical fully dispersed soil size distribution for soils from arid and
semi-arid regions. However, to reproduce the measurements corresponding to the
intense dust event from Sow et al. ( 2009 ) properly, a different adjustment of the
propagation length is required, that was attributed to changes in surface properties
between the different events (Kok 2011b ). Part of the differences between the
models can be due to the representation of the saltation processes. Kok ( 2011b )
argues that the dependence of the dust size distribution on U * in the models of