Operational Dust Prediction - Mineral Dust: A Key Player in the Earth System - page 226

Fig. 10.1 Time series of modelled 10M winds in Djougou, 9.7 N, 1.6E ( top ), and corresponding
dust surface concentrations in Rome, 41.9 N, 12.5E ( bottom ). Each coloured line represents
successive forecasts of a single day starting 4 days in advance ( top ); and a forecast lead time
( bottom ), respectively
low amount of wind observations available to constrain the meteorological analysis
driving the simulation. Moreover, nearly all large-scale models and regional models
do not have the capability to resolve convective-scale phenomena (e.g. Reinfried
et al. 2009 ) and are therefore missing potentially important emission sources (see
Chap. 6 for more details). In the last few years, a good degree of accuracy in the
prediction of dust at the synoptic scale and in some cases at the regional scale has
been achieved, thanks to model improvements and in some cases data assimilation
to the point that the information can be offered to forecasters as guidance.
The issue of predictability is illustrated with an example in Fig. 10.1 ,which
presents time series of modelled 10 m wind speed and dust surface concentrations.
For the wind speed, here presented for Djougou and modelled with WRF (Menut
et al. 2009 ), each coloured line represents a forecast of 4 days. The corresponding
days are superimposed and show the spread from one forecast to the following. The
wind speed values range from 1 to 6 m/s, and the differences between each forecast
do not exceed 1 m/s. This small variability is not important to those interested in the
weather forecast. However, this variation becomes important when this wind speed
is used to calculate dust emission and transport. The dust concentrations over Rome
after long-range transport from Africa, modelled with CHIMERE, are presented in