Modeling Large-Scale Dust Storms
Background Research
A model for the emission of PM10 dust has been constructed (Draxler et al., 2001)
using the concept of a threshold friction velocity which is dependent on surface roughness. Surface
roughness in turn was correlated with geomorphology or soil properties for Kuwait, Iraq, part of Syria,
Saudi Arabia, the United Arab Emirates and Oman. A dust emission rate was computed from each cell when
the local wind velocity exceeded the threshold velocity for the soil characteristics of that emission
cell. The dominant mechanism for the PM10 dust input model is "sand-blasting". The emitted
material was dispersed and transported using a modified Lagrangian particle-puff model (Draxler and
Hess, 1998) using gridded meteorological data fields. Computations were made for the period of August
1990 through August 1991. The model calculated air concentrations from mid-May through mid-July, the
period of the most frequent and intense dust storms. These calculations were compared with the
measured data.
The model predicted about the right number of dust events over Kuwait (events occur 18% of
the time). The model results agreed quantitatively with measurements at four locations in
Saudi Arabia and one in Kuwait for one major dust event (>1000 µg/m3). However,
for smaller scale dust events (200 - 1000 µg/m3 ), especially at the coastal
sampling locations, the model substantially over-predicted the air concentrations. Part of
the over-prediction was attributed to the entrainment of dust-free air by the sea breeze, a
flow feature not represented by the large-scale gridded meteorological data fields used in
the model computation. Another part of the over-prediction was the model's strong sensitivity
to threshold friction velocity and the surface soil texture coefficient (the soil emission factor),
and the difficulty in accurately representing these parameters in the model. A comparison of the
model predicted PM10 spatial pattern with the TOMS satellite aerosol index (AI) yielded
a spatial pattern covering a major portion of Saudi Arabia that was quite similar to the observed
AI pattern.
Operational Implementation
The initial research model configuration was only suited for use over Kuwait and Saudi Arabia.
The model was restructured to use the default HYSPLIT land-use characteristics file defined on a
one degree global grid. A preprocessing program was developed that accessed this file over any
selected modeling domain to create a HYSPLIT input control file such that each emission point
entry corresponded with a "desert" land-use grid cell. The desert cell was assumed to correspond
with the "active sand sheet" category defined for the Kuwait simulations. The calculations then
proceeded as before, such that particle emissions only occurred from locations in which the
local friction velocity exceeded the threshold friction velocity. The Kuwait specific PM10
flux equation was replaced by a more generic relationship given by Westphal et al. (1987).
The two 10 day animations, for dust events in April 2001 and
March 2002, represent the daily model particle positions at 0600 UTC
superimposed over the TOMS aerosol index for that day. Particle emissions occurred automatically
during the period based upon land-use and friction velocity. Satellite and model results may not
always correspond due to errors in the model simulation as well as the fact that the TOMS AI may
be obscured by clouds and tends to be more representative of dust at higher levels in the
atmosphere rather than near the ground. However, regardless of the caveats, the match between
model results and measurements is at times quite striking, especially for the event of April
2001 as the dust cloud moved across the Pacific.
Current Experimental Operation
Daily forecasts of long range dust transport from northern Africa to the United States
are produced using a combination between a lagrangian and a eulerian model. The HYSPLIT model
simulates the emissions, transport, dispersion, and deposition of dust particles for an area
covering Northern Africa and Southern Europe. This model is driven by the GFS meteorological
forecast with a horizontal resolution of 0.5x0.5 degrees. The global model is initialized
every hour by dividing the lagrangian particle masses provided by HYSPLIT by the volume of
the grid in which they reside. The global simulation is driven by the GFS meteorological
forecast with a horizontal resolution of 1x1 degrees. This three-dimensional
transport-dispersion global model includes wet and dry deposition of the dust particles.
These settings take advantage of the higher resolution used by HYSPLIT to calculate the
regional emissions of dust and the reduction in computation time required to run the
global simulation. The global dust forecasting system produces 72-hour forecasts of
surface and column integrated PM10 concentrations.
References
Draxler, R.R, Gillette, D.A., Kirkpatrick, J.S., Heller, J., 2001, Estimating PM10 Air Concentrations from Dust Storms in Iraq, Kuwait, and Saudi Arabia, Atmospheric Environment, Vol. 35: 4315-4330
Draxler, R.R. and Hess, G.D., 1998. An Overview of the HYSPLIT_4 Modelling System for Trajectories, Dispersion, and Deposition. Aust. Meteor. Mag. 47, 295-308.
Westphal, D.L., Toon, O.B., Carlson, T.N., 1987. A two-dimensional numerical investigation of the dynamics and microphysics of Saharan dust storms. J. Geophys. Res., 92, 3027-3029. |
