1. science
2. physics

Computational Modelling of
Aerosol Cycle
An Integrated Environmental Modelling System and Its
Applications
Dr. Yaping Shao
CEMAP, School of Mathematics
The University of New South Wales
Sydney, Australia
Tel: 61 2 9385 5746; Fax: 61 2 9386 7123
email: [email protected]
Aerosols
– Aerosols are small particles suspended in air.
The sizes of aerosols range between 0.1 - 20
microns;
– Aerosol sources include natural and human
induced ones.
Aerosol Research: Climate and Weather
• Directly, aerosols affect atmospheric radiation
budget through scattering and absorbing;
• Indirectly, aerosols modify the optical properties
and lifetimes of clouds;
• Dust (global emission):
~ 3000 Mt/yr.
• Sea salt:
~ 1300 Mt/yr.
• Dust (mean column load):
~ 65 mg m-2
• Sea salt:
~ 7 mg m-2
Dust storm in Africa: 27 July 1998, Algeria and Mali
A severe dust storm over China (16 April 1998)
Dust clouds seen from satellite picture (14 April 1998)
Aerosol Research: Air quality
• Aerosols cause air-quality hazards in populated
areas, e.g., Beijing;
• Many contaminants which pose significant risks to
human health and the environment are found or
associated with dust, including metal, pesticides,
dioxins and radionuclides.
A severe dust storm (acknowledgement)
Land-Use Sustainability
• In agricultural areas, soil erosion depletes fine
particles which are rich in organic matters and soil
nutrients. This leads to land degradation;
• Wind erosion also reduces water, resulting in
desertification.
A dust storm in an agricultural area
Soil Erosion
Melbourne 08-02-1983 dust storm:
Nutrient content in soil particles < 44 microns
Site
Soil Type
Wamberra Box Creek Montarna Sandy
Sand
Clay Loam Loam
Total N (%)
0.226
0.16
0.153
Totol P (%)
0.038
0.029
0.034
N enrichment
ratio
19
2
2
P enrichment
ratio
5.7
1.9
2.4
Mass fraction
0.003
0.11
0.06
Melbourne 08-02-1983 dust storm
Total loss of top soil
M=2 million tonnes
Total loss of N
M*0.0017=3400 tonnes
Total loss of P
M*0.000055=110 tonnes
Cost of fertilizer
(N:P:K=32:10:0)
0.37 dollars
Cost of N
3.9 million dollars
Cost of P
0.4 million dollars
Mineral Aerosol Cycle
– Entrainment: atmosphere and land-surface
interactions; multi-disciplinary;
– Transport: atmospheric circulation;
atmospheric boundary layers; turbulence; two
phase flow problem
– Deposition: turbulent diffusion; clouds and
precipitation.
Integrated Environmental Modelling
– How can such complex environmental
problems be simulated and predicted?
– Computational environmental modelling: the
integration of dynamic models with spatially
distributed data
– Atmosphere-land surface interactions
– Air quality
– Aerosol cycle
– Land surface hydrology and salinity
Framework I
Computational Environmental
Modelling System (CEMSYS_3)
– Atmospheric prediction model (HIRES): highresolution limited-area; nested in GCM, selfnested; 3rd order upwinding and semilagrangian schemes; clouds and radiation.
– Land surface (ALSIS): Soil moisture,
temperature; fluxes of energy, mass and
momentum;
– Aerosol cycle: entrainment, transport and
deposition.
– Air quality, etc
Framework of CEMSYS_3 (partial)
Physical processes involved in wind erosion
Particle Motion
– Saltation:
hop motion
of sand
particles;
– Suspension:
small
particles can
remain
suspended
once
airborne.
Friction velocity & threshold
friction velocity
• The capability of wind to cause erosion is
quantified by surface friction velocity,
u*,depending on wind speed and surface
roughness
• The ability of the surface to resist erosion is
quantified by threshold friction velocity u*t,
depending on soil texture, compactness, moisture
content and surface coverage
• Modeling u*t is difficult
Entrainment of Coarse Particles
• Balance of aerodynamic, gravity and cohesive
forces, fa, fg and fi, determines the entrainment;
• For coarse particles, fa overcome fg and fi;
• Friction velocity u* measures aerodynamic forces;
• Threshold friction velocity u*t measures retarding
forces.
• Shao-Lu model for u*t is

d
u*t  f (Re * )( p gd 
)

Entrainment of Fine Particles
• The Entrainment
mechanisms for
coarse and fine
particles differ as
the importance of
forces change.
• fg  d3, fa  d2
and fi  d; fi
dominates.
Dust Emission Mechanisms
• Fa, aerodynamic lift. Particles can be lifted
directly by fa, but emission is weak;
• Fb, saltation bombardment. Striking particles
cause local impacts, overcome fi, result in
strong emission;
• Fc, aggregates disintegration. Fine particles
exist as aggregates. Weak events, they behave
as grains. Strong events, they disintegrate.
• Dust-emission rate:
F = Fa + Fb + Fc
Particle-size Distribution
•
•
•
•
•
•
Soil particle size ranges: 0.1 m - 2 m
Gravel:
2000 m < d  2m
Sand:
63 < d  2000 m
Silt:
4 < d  63 m
Clay:
d  4 m
Silt and clay particles are dust.
Particle-size Distributions
• ps(d): sediment particle-size distribution (psd);
• pm(d): in-situ soil psd; minimally dispersed
analysis;
• pf(d): fully-disturbed soil psd; fully-dispersed
analysis.
Model for ps(d)
• Limiting cases
ps(d)  pm(d)
ps(d)  p (d)
f
u u
* *t
u  u
*
*t
Model
ps(d)  pm(d)  (1 )p (d)
f
Weight
  exp[ k(u  u )n ]
* *t
Example of ps(d)
Fractions of Fine Particles
dd
m   pm(d)d
0
dd
   p (d)d
f 0 f
dd
s   ps(d)d
0
dd
c   p (d)  pm(d)d
0 f
m: free dust, lower limit for
dust emission from unit soil
mass;
f: not free dust, released
through saltation impact and
aggregates disintegration,
upper limit for dust emission
from unit soil mass;
s: aerosol in suspension
Theory of Saltation
• Saltation plays a critical role in the process of
dust emission.
• Two quantities are of particular importance,
namely, the streamwise saltation flux, Q, and
the number flux of striking particles, ns
u *t
 3
Q  c 0 u * (1  2 )
g
u*
2
Qg
ns 
mc 0 u * ( U1 cos 1  U 2 cos  2 )
Volume Based Model for Fb
Particle trajectory is (XT, YT) in soil, forms a crater
of volume 
Volume Based Model for Fb
•Trajectory from equation of particle motion;
•cbf: fraction released;
• (1-cb)f : fraction retained;
dX T
  b  YT
dt
0
dt
du p
dv p
dX T
dYT
m
 a x p x  0; m
 a y p y  0;
- u p  0;
- vp  0
dt
dt
dt
dt
Fb (d s )  c b f  s n s
tc
Volume Based Model for Fb
•Particle trajectory is (XT, YT) in soil, forms a crater
of volume ;
•Trajectory from equation of particle motion;
•cbf: fraction released; (1-cb)f : fraction retained;
dX T
  b  YT
dt
0
dt
du p
dv p
dX T
dYT
m
 a x p x  0; m
 a y p y  0;
- u p  0;
- vp  0
dt
dt
dt
dt
Fb (d s )  c b f  s n s
tc
Aggregates Disintegration: Fc
• Aggregates disintegration occurs as they strike
surface.
• Corresponding to ns, the mass flux of particles
striking surface is mns.
Fc(ds) = cc fc m ns
• cc: a coefficient
Total Dust Emission: F
• Divide particles into I size groups, mean di,
increment di; Consider emission of i group
~
F (d , ds )  c~   ns;
b i
b fi b
~
Fc(d , ds )  c~c  m ns
i
ci
c~  c~c  c E
b
 (d )
i
~
c
F(d , ds )  6.5 c Qg
(     m )

i
ci
fi b
0
Emission of ith group : Fˆ (d )  dd12 F(d , d) ps(d) d
i
i
I ˆ
Total : F   F(d )
i1 i
Model of Particle Size Distribution
• Emission model requires pm(d) and pf(d).
• Express as sum of J log-normal pdfs with
parameters wj, Dj and j; both for pm(d) and
pf(d) for sand, loam and silty clay.
w
(ln d  ln D )2
J
j •exp j
p(d)  1 
2
d j12
2

j
j
















c
• Model requires ;
• cE fi /: fraction of release;
• cY = 1/7 co, order 0.1.
~
F(d , ds )  c E  (   m)ns
i
fi b
si ps (di )

E
c    
p
(
d
)
fi
f i
pm(d ) Qg
~
i ]
F(d ,ds )  c [(1-)  
(    m)
i
ci
Y
p (d ) u2m b fi
f i *
Quantities Required
•
•
•
•
•
•
•
u*: friction velocity;
u*t: threshold friction velocity for surface;
pm(d): minimally-dispersed psd;
pf(d): fully-dispersed psd;
b, p: bulk soil and particle density;
s: soil drag coefficient;
pys:vertical component of plastic pressure.
Results
Conclusions for Emission Model
• Concept: F is related to Q;
• Mechanisms: saltation bombardment and aggregates
disintegration;
• Models for Fb and Fc;
• Soft soils, Fb dominates;Hard soils, Fc dominates;
• psds are used to eliminate empirical parameters;
• psds modeled using log-normal pdfs;
• Emission rates compare well with observations.
Transport: Lagrangian
• Particles are individuals; Trajectories are
determined by integrating equations of motion;
• Isentropic trajectories on surface of constant
potential temperature;
• Fluid parcel and particle are at height zft-1 = zpt-1 at
t-1, fluid moves to zft, particle to zpt=zft+wtt.
Transport: Eulerian
• Particulate phase is a continuum;
• Particle concentration obeys advection-diffusion type
of conservation equation;
• Kpx: particle eddy diffusivity; Sr: wet and dry removal;
Sc: dry and wet convection; F0: dust flux at surface
c  u c  v c (w  w )c   K c   K c
t z x px x y py y
t x y
  K pz c  Sr  Sc
z
z
(w  wt )c - K pz c  F
z 0
K pz c  0
z
Inertial and Trajectory Crossing
Particle Eddy Diffusivity
Deposition
• Dry-deposition flux
Fd = -wd[c(z)-c(0)]
• c(0), c(z): concentration at surface and reference
level; wd: dry-deposition velocity.
• Single-layer dry-deposition model
wd=-wt+gbb+gbm
• gbb: molecular conductance; gbm: impaction
conductance; fr: ratio of pressure drag to total drag;
wd=-wt+ga[fr ap em+(1-fr)avSc-2/3]
Wet Deposition
Wet deposition is the removal of aerosols by
precipitation. The processes is extremely
complicated, but is commonly calculated using
Fw=w pr0s0c0
s0: scavenging ratio is a function of many
parameters, but ranges from 100 to 2000.
pr0: rain received at the surface;
c0: concentration in rain water .
Example 1: How does the Scheme Work
Comparison with Field Measurements
Land Surface Data
Weather pattern
Feb. 1996
Soil Erosion
Threshold
Friction
Velocity
Friction Velocity
Concentration Cross Section
Total
Suspended
Dust Time
Series
Comparison with Satellite Image
Aerosol
Concentration
Surface Concentration: Birdsvill, Feb. 1996
Higher
Resolution
Higher
Resolution
Summary
• A comprehensively integrated system has been
developed for the simulation and prediction of the
entire mineral dust cycle, from entrainment, transport
to deposition. CEMSYS_3 has a much wider range of
applications;
• I have illustrated how the entire cycle can be modeled.
Each of the modeling components constitutes an
interesting research area. I have concentrated on dust
emission in this talk;
• Coupling dynamic models with spatially distributed
data has enabled the predictions of dust storm events.