Operational Dust Prediction - Mineral Dust: A Key Player in the Earth System - page 248

sites, and the polygon areas are used as the observation weights. Since the polygons
will naturally be smaller in data-dense areas, observations in these areas receive
lower weights than those in data-sparse areas. To prevent observations in very data-
sparse areas receiving higher than reasonable weights, the polygon edges are limited
to a maximum radius. This is currently set to a value which results in a maximum
polygon area of 1 % of the total area being scored.

Case Study Evaluation

An exhaustive comparison of model outputs against other models and observations
can reveal weaknesses of individual models, provide an assessment of uncertainties
in simulating the dust cycle and give additional information on sources for potential
model improvement. For this kind of study, multiple and different observations are
combined to deliver a detailed idea of the structure and evolution of the dust cloud
and the state of the atmosphere at the different stages of the event. Observations
detailed in Sect. 10.5.2 are usually complemented with strictly meteorological
observations such as wind speed and direction at the surface and wind profile within
the atmospheric boundary layer.
Multiple case studies concerning a single model can be found in the literature
(e.g. Pérez et al. 2006b ; Heinold et al. 2007 ; Cavazos et al. 2009 ). On the other hand,
inter-comparisons of multiple models simulating the same event are described by
Uno et al. ( 2006 ) and Todd et al. ( 2008 ). Both studies reveal the ability of models
to reproduce the onset and duration, but not the magnitude of a given dust event.
Furthermore, even though the models were able to reproduce surface measurements,
large differences existed among them in processes such as emission, transport and
deposition. Shao et al. ( 2003 ) evaluated not only the model performance to simulate
a specific dust event but also the model capacity to predict the event for different
lead times. The authors found that the predicted quantities agreed well with the
observations. Many global aerosol models have also been evaluated over extended
time periods as part of the AeroCom project (see Chap. 9 ) .

10.6 Conclusion

Dust numerical prediction is a growing area of research with many operational
applications. In the last few years, many centres have started activities to provide
dust forecasts to interested stakeholders, who range from solar energy plant
managers to health and aviation authorities, from policymakers to climate scientists.
There is also a growing interest in understanding how dust impacts the general
circulation of the atmosphere through its radiative effects which could help in
improving numerical weather prediction and projections of climate change. Dust
forecast models have reached a high degree of complexity and can provide useful
information to forecasters. Some factors limiting the accuracy of the models are