Geoscience Reference
In-Depth Information
described next in this section. From the comparison between the reference
run and the sensitivity experiments, a simple representation of convective-
scale transport in cloud-resolving simulations is proposed in the last part
of the section. The conclusions of this study are summarized in Sec. 4.
2. Model Setup and Experimental Design
The dust transport model used here is based on a regional meteorological
model, the Advanced Regional Prediction System (ARPS),
11
which was
developed by The University of Oklahoma Center for Analysis and Prediction
of Storms. The emission and transport processes of dust aerosol have been
built into the ARPS model.
12
The model is configured in an idealized way as in
the study of Takemi
et al.
6, 7
in order to focus on the fundamental dynamics of
convective dust transport under a fair weather condition. The meteorological
model includes full physics parameterizations, i.e. cloud microphysics,
13
subgrid-scale (SGS) turbulence mixing,
14
land-surface physics, and radiative
transfer.
15
In the dust module, the vertical dust flux at the surface is
determined as the fourth power of friction velocity,
16
and the atmospheric
transport is computed by the velocities obtained from the atmospheric model.
The threshold of the friction velocity for dust emission is set to be 60 cm s
−
1
.
The dust property is represented as a mixing ratio, and a single size bin of
1.0-
µ
mradiusisassumed.
In order to perform high-resolution simulations that would explicitly
resolve boundary-layer eddies, a large-eddy simulation (LES) model of
Deardorff
14
is used for the parameterization of SGS turbulence mixing. The
turbulence length-scale depends on the stability, and has the same value for
both the horizontal and the vertical directions.
A high-resolution simulation with the horizontal grid spacing (∆
x
)of
100 m and the vertical grid spacings of 20-240 m (85 levels) is conducted in
a mesoscale domain of 40 km (east-west, the
x
-axis)
×
10 km (north-south,
the
y
-axis)
11 km (vertical, the
z
-axis). Although the spacing of 100 m
seems to be relatively larger for LES, recent studies using this grid size
have been successful in representing both shallow and deep convection;
17, 18
therefore, we consider that the 100 m grid is sucient for simulating both
shallow and deep convection and the associated dust transport. A periodic
condition is imposed at all the lateral boundaries, and the upper boundary
is a rigid lid with a Rayleigh-type damping layer above the 9 km height.
This 100 m grid simulation is referred to as the control.
×
