Numerical Dust Models - Mineral Dust: A Key Player in the Earth System - page 210

underestimated dust emission fluxes. The spread of model results for the surface
concentration, mass load and optical thickness is about one order of magnitude,
leading to differences in model estimates of exported dust mass of a factor of 30.
Similarly, Uno et al. ( 2006 ) find a range of an order of magnitude differences for
modelled dust concentration from nine regional models comparing two East Asian
dust events.
In general, dust events that are controlled by large-scale meteorological features
that are well represented by meteorological models, for example, frontal passages
or Sharav cyclones passing through northern Africa in spring (see Chap. 6 ) , are
well represented in regional-scale dust models. In contrast, these regional models
have problems to represent dust emission, transport and deposition events that are
controlled by small-scale atmospheric features like moist convective events, which
are usually parameterized in regional models. Microscale events like dust devils
causing dust emissions cannot be represented directly in such models, and therefore
their contribution to the atmospheric dust load remains unclear.
In spite of the problems in the accurate determination of dust emission areas
and reproducing the meteorological processes responsible for dust emission, model
studies of individual dust events have been successful in simulating the spatio-
temporal distribution of dust transported far downwind, for example, in cases of
transport from the Sahara towards Europe (e.g. Pérez et al. 2006 ). For regional dust
forecast models, an accurate simulation of the meteorology may be more important
than a precise placement of active dust source areas.

9.8 Global Dust Models

Global dust models are used to investigate the role of dust in the climate system.
While such models reproduce the large-scale regional and seasonal patterns of the
atmospheric dust loads, the results of such models may deviate considerably from
observations at individual locations (Huneeus et al. 2011 ). The global aerosol model
intercomparison study AeroCom (Textor et al. 2006 ) was developed as a platform
to evaluate the performance of global aerosol models. In the first AeroCom phase,
aerosols computed with either offline transport models and GCMs were considered.
Within AeroCom, the simulated distribution and radiative effects of dust as part of
natural and anthropogenic aerosol components (dust and sea salt particles, sulphate,
organic carbon and black carbon) were compared for 16 global models. In one series
of experiments, each model used their own individual aerosol emissions (Exper-
iment A), while in Experiment B the models computed the aerosol cycles using
harmonized emission fields (Textor et al. 2007 ). The results were compared and
evaluated with observational data, including concentrations of individual species
at the surface, as well as optical thickness data from the global AERONET sun
photometer network and derived satellite products (see Chap. 7 ) . Kinne et al. ( 2006 )
evaluated differences in optical properties that lead to diversities in aerosol radiative
forcing by the AeroCom models. With respect to dust, the multi-model median for