Stochastic variability of mass flux in a cloud
resolving simulation
Jahanshah Davoudi
Thomas Birner, Norm McFarlane and Ted Shepherd
References
Brenda G. Cohen and George C. Craig, 2004: Quart. J. Roy. Meteor. Soc.
130
, 933-944.
Brenda G. Cohen and George C. Craig, 2006: J. Atmos. Sci.
63
, 1996-2004.
Brenda G. Cohen and George C. Craig, 2006: J. Atmos. Sci.
63
, 2005-2015.
Convective parameterization
Many clouds and especially the processes within them are
subgrid-scale size
both horizontally and vertically and thus must be parameterized.
This means a mathematical model is constructed that attempts to assess their
Parameterization Basics
Arakawa
&
Schubert 1974
Key equilibrium assumption:
Fluctuations in radiative-convective equilibrium
For convection in equilibrium with a given forcing, the mean mass flux
should be well defined.
At a particular time, this mean value would only be measured in an
infinite
domain
.
For a region of
finite size
:
What is the magnitude and distribution of variability?
Main assumptions
Assume:
1.
Large-scale constraints
- mean mass flux within a region
h
M
i
is given in terms
of large scale resolved conditions
2.
Scale separation
- environment sufficiently uniform in time and space to
average over a large number of clouds
3.
Weak interactions
- clouds feel only mean effects of total cloud field( no
organization)
Large scale constraint
h
M
i
is determined by the requirement that the convection balance the large
scale forcing when averaged over a large region.
h
m
i
is
not
necessarily a function of large scale forcing
Observations suggest that
h
m
i
is independent of large scale forcing
Response to the change in forcing is to change the
number
of clouds.
h
m
i
might be only sensitive to the initial perturbation triggering it and the
dynamical entrainment processes.
mass flux of individual clouds are statistically un-correlated :
P
M
(
n
) =
P rob
{
N
[(0
, M
]) =
n
}
=
(
λM
)
n
e
−
λM
n
!
n
= 0
,
1
,
· · ·
given
λ
= 1
/
(
h
m
i
) =
h
h
M
N
i
i
is fixed.
Poisson point process implies:
P
(
m
) =
1
h
m
i
e
−
m hmiThe total Mass flux for a given
N
Poisson distributed plumes is a
Compound point process
:
M
=
N
X
i
=0
Predicted distribution
So the Generating function of
M
is calculated exactly:
h
e
tM
i
=
e
−
Λ
e
Λ
G
(
t
)
(1)
G
(
t
) =
h
e
tm
i
Λ =
h
N
i
Therefore the probability distribution of the
total mass flux
is exactly given by:
P
(
M
) =
P
(
M
) =
„
h
N
i
h
m
i
«
1
/
2
e
−h
N
i
M
−
1
/
2
e
−
M/
h
m
i
I
1
2
„
h
N
i
h
m
i
M
«
1
/
2
!
All the moments of
M
are analytically tractable and are functions of
h
N
i
and
h
m
i
.
h
(
δM
)
2
i
h
M
i
2
=
2
h
N
i
h
M
r
|
N
i
=
(
N
+
r
−
1)!
(
N
−
1)!
h
m
i
r
Estimates
In a region with area
A
and grid size
∆
x
≫
L
where
L
the
mean cloud
spacing
is:
L
= (
A/
h
N
i
)
1
/
2
= (
h
m
i
A/
h
M
i
)
1
/
2
Assume latent heat release balance radiative cooling
S
,
rate of Latent heating
≃
Convective mass flux
×
Typical water vapor mass mixing
ratio
q
l
v
q
h
M
A
i
=
S
Estimate:
S
= 250
W m
−
2
,
q
= 10
gkg
−
1
and
l
v
= 2
.
5
×
10
6
Jkg
−
1
gives
h
M
i
/A
= 10
−
2
kgs
−
1
m
−
2
h
m
i
=
wρσ
with
w
≃
10
ms
−
1
and
σ
≃
1
km
2
gives
h
m
i ≃
10
7
kgs
−
1
hence
L
≃
30
km
δM
M
=
√
2
∆
L
x
Simulations with a ’cloud resolving’ model
Resolution:
2km
×
2km
×
90 levels
Domain:
96 km
×
96 km
×
30 km
Boundary conditions:
doubly periodic, fixed SST of 300 K
Forcing:
An-elastic equations with fully interactive radiation
Stochastic variability of up-draught M and N
0 0.002 0.004 0.006 0.008 0.01 0.012 124.5 125 125.5 126 126.5 127 127.5 M(t) (x 10 10 Kg /sec) t(days) z=4989.53 m 0 2 4 6 8 10 12 14 16 124.5 125 125.5 126 126.5 127 127.5 N(t) t(days) z=1071.34 m 0 0.0002 0.0004 0.0006 0.0008 0.001 0.0012 0.0014 124.5 125 125.5 126 126.5 127 127.5 mc (t)(x 10 10 Kg /sec) t(days)Auto-correlation of up-draught M
-0.2
0
0.2
0.4
0.6
0.8
1
0
10
20
30
40
50
60
70
<M(t+
τ
)M(t)>
|
τ
(hours)|
z=2976.26 m
z=6907.73 m
z=9900.00 m
z=12900.00 m
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
0.5
1
1.5
2
2.5
3
<M(t+
τ
)M(t)>
|
τ
(hours)|
z=2976.26 m
z=6907.73 m
z=9900.00 m
z=12900.00 m
Auto-correlation of up-draught N
-0.2
0
0.2
0.4
0.6
0.8
1
0
10
20
30
40
50
60
70
<N(t+
τ
)N(t)>
|
τ
(hours)|
z=2976.26 m
z=6907.73 m
z=9900.00 m
z=12900.00 m
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
0.5
1
1.5
2
2.5
3
<N(t+
τ
)N(t)>
|
τ
(hours)|
z=2976.26 m
z=6907.73 m
z=9900.00 m
z=12900.00 m
Auto-correlation of up-draught
m
c
-0.2
0
0.2
0.4
0.6
0.8
1
0
10
20
30
40
50
60
70
<m
c(t+
τ
)m
c(t)>
|
τ
(hours)|
z=2976.26 m
z=6907.73 m
z=9900.00 m
z=12900.00 m
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
0.2
0.4
0.6
0.8
1
<m
c(t+
τ
)m
c(t)>
|
τ
(hours)|
z=2976.26 m
z=6907.73 m
z=9900.00 m
z=12900.00 m
Comparison of short de-correlation time
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
0.5
1
1.5
2
Auto-correlation
|
τ
(hours)|
w
M
N
-0.2
0
0.2
0.4
0.6
0.8
1
0
10
20
30
40
50
60
70
Auto-correlation
|
τ
(hours)|
w
M
N
Mean characteristics
0
5
10
15
20
25
0
0.001
0.002
0.003
0.004
0.005
0.006
z (Km)
<M> (Kg m
-2s
-1)
0
5
10
15
20
25
0
1
2
3
4
5
6
z (Km)
<N>
Variance scaling
0
2
4
6
8
10
12
14
0
2
4
6
8
10
12
(<M
2>-<M>
2)/<M>
22/<N>
0
2
4
6
8
10
12
14
0
5
10
15
20
25
<N>
σ
2NThe scaling of the variance of total mass flux and number of active grids in different heights with the
Craig and Cohen prediction:
h
(
δM
)
2
i
h
M
i
2
=
2
h
N
i
Skewness scaling
0
2
4
6
8
10
12
0
1
2
3
4
5
6
7
8
9
z (Km)
〈
(M -
〈
M
〉
)
3〉
⁄
〈
M
〉
36
⁄
〈
N
〉
2The prediction:
h
(
δM
)
3
i
h
M
i
3
=
6
h
N
i
2
Altitude variability of total Mass flux PDF
z
= 1517
m
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02 0 20 40 60 80 100 120 140 160 M P(M;z) Measured PDF Free 〈 M 〉 and 〈 N 〉P(M;z) with measured 〈 M 〉 and 〈 N 〉
z
= 2976
.
26
m
0 0.005 0.01 0.015 0.02 0.025 0 20 40 60 80 100 120 140 M (Kg M−2 s−1) P(M;z) Measured PDF Free 〈 M 〉 and 〈 N 〉P(M;z) with measured 〈 M 〉 and 〈 N 〉
0 2 4 6 8 10 12 14 16 18 20 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 N P(N;z) Measured PDF Fit 〈 N 〉=5.288
Poisson PDF with measured 〈 N 〉=5.28
0 2 4 6 8 10 12 14 16 18 20 0 0.05 0.1 0.15 0.2 0.25 N P(N;z) Measured PDF Fit 〈 N 〉=3.5
P(M,z) and P(N,z)
z
= 4989
.
5
m
0 0.005 0.01 0.015 0.02 0.025 0.03 0 20 40 60 80 100 120 140 M (Kg M−2 s−1) P(M;z) Measured PDF Free 〈 M 〉 and 〈 N 〉P(M;z) with measured 〈 M 〉 and 〈 N 〉
z
= 6907
.
73
m
0 0.005 0.01 0.015 0.02 0.025 0.03 0 50 100 150 200 250 M (Kg M−2 s−1) P(M;z) Measured PDF Free 〈 M 〉 and 〈 N 〉P(M;z) with measured 〈 M 〉 and 〈 N 〉
0 2 4 6 8 10 12 14 16 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 N P(N;z) Measured PDF Fit 〈 N 〉=2.6
Poisson PDF with measured 〈 N 〉=2.7
0 5 10 15 20 25 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 N P(N;z) Measured PDF Fit 〈 N 〉=2.45
P(M,z) and P(N,z)
z
= 8701
.
13
m
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 50 100 150 200 250 M (Kg M−2 s−1) P(M;z) Measured PDF Free 〈 M 〉 and 〈 N 〉P(M;z) with measured 〈 M 〉 and 〈 N 〉
z
= 9900
.
00
m
0 0.005 0.01 0.015 0.02 0.025 0.03 0 50 100 150 200 250 300 350 M (Kg M−2 s−1) P(M;z) Measured PDF Free 〈 M 〉 and 〈 N 〉P(M;z) with measured 〈 M 〉 and 〈 N 〉
0 5 10 15 20 25 30 35 40 0 0.05 0.1 0.15 0.2 0.25 N P(N;z) Measured PDF Fit 〈 N 〉=2.9
Poisson PDF with measured 〈 N 〉=3.6
0 5 10 15 20 25 30 35 40 45 50 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 N P(N;z) Measured PDF Fit 〈 N 〉=1.7
P(M,z) and P(N,z)
z
= 11400
.
00
m
0 0.005 0.01 0.015 0.02 0.025 0.03 0 50 100 150 200 250 300 350 400 450 500 M (Kg M−2 s−1) P(M;z) Measured PDF Free 〈 M 〉 and 〈 N 〉P(M;z) with measured 〈 M 〉 and 〈 N 〉
z
= 12900
.
00
m
0 0.005 0.01 0.015 0.02 0.025 0 100 200 300 400 500 600 700 M (Kg M−2 s−1) P(M;z) Measured PDF Free 〈 M 〉 and 〈 N 〉P(M;z) with measured 〈 M 〉 and 〈 N 〉
0 10 20 30 40 50 60 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 N P(N;z) Measured PDF Fit 〈 N 〉=0.95
Poisson PDF with measured 〈 N 〉=3.4
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 N P(N;z) Measured PDF Fit 〈 N 〉=0.5
Conditional average
h
M
|
N
i
0 2 4 6 8 10 12 14 16 18 20 0 0.5 1 1.5 2 2.5 3 3.5 < M | N>/<m> N/<N> z=1516.91 m <N>*x 0 2 4 6 8 10 12 14 16 18 0 1 2 3 4 5 6 < M | N>/<m> N/<N> z=4989.53 m <N>*x -5 0 5 10 15 20 25 30 35 40 0 2 4 6 8 10 12 < M | N>/<m> N/<N> z=8403.16 m <N>*x -10 0 10 20 30 40 50 60 0 2 4 6 8 10 12 14 16 < M | N>/<m> N/<N> z=11400.00 m <N>*xh
M
|
N
i
=
h
m
i
N
Cross-correlation of M and N
0
2
4
6
8
10
12
14
16
0.7
0.75
0.8
0.85
0.9
0.95
1
z (Km)
<(M-<M>)(N-<N>)>/
σ
M
σ
N
The prediction:
h
(
M
− h
M
i
)(
N
− h
N
i
)
i
σ
M
σ
N
=
1
√
2
Low order mixed moment
0
2
4
6
8
10
12
14
16
0
1
2
3
4
5
6
7
z (Km)
<(M-<M>)
2
(N-<N>)>
⁄
σ
M
2
σ
N
1
⁄
<N>
The prediction:
h
(
M
− h
M
i
)
2
(
N
− h
N
i
)
i
σ
2
M
σ
N
=
1
√
N
Fat tails of marginal PDF of active grids
1e-05 1e-04 0.001 0.01 0.1 1 0 5 10 15 20 P(N;z) N z=1516 m z=2641 m z=4129.88 m 1e-05 1e-04 0.001 0.01 0.1 1 0 10 20 30 40 50 60 P(N;z) N z=4129.88 m z=5830 m z=7219 m z=8403 m z=9600 m z=10800 m 1e-05 1e-04 0.001 0.01 0.1 1 0 10 20 30 40 50 60 P(N;z) N z=10800 m z=12000 m z=13200 m z=14400 m z=15600 mPDF of dM/dz
1e-06 1e-05 1e-04 0.001 0.01 0.1 1 7e-05 5e-05 3e-05 1e-05 0 -1e-05 -3e-05 P(dM/dz) dM/dz (x107 Kg/ms) [0,1] Km 1e-07 1e-06 1e-05 1e-04 0.001 0.01 0.1 1 1e-05 0 -1e-05 P(dM/dz) dM/dz (x107 Kg/ms) [1,4.9] Km 1e-07 1e-06 1e-05 1e-04 0.001 0.01 0.1 1 1e-05 0 -1e-05 P(dM/dz) dM/dz (x107 Kg/ms) [4,9,9.9] Km 1e-06 1e-05 1e-04 0.001 0.01 0.1 1 1e-05 0 -1e-05 P(dM/dz) dM/dz (x107 Kg/ms) [9.9,16] KmRegime change
2
4
6
8
10
12
0
0.001
0.002
0.003
0.004
0.005
0.006
z (Km)
<m> (Kg m
-2
s
-1
),<M> (Kg m
-2
s
-1
),<N> (x10
3
)
<m>
<M>
<N>
z
∼
[1
,
6]
Km: The
h
m
i
is increasing while
h
N
i
is decreasing.
z
∼
[6
,
11]
Km:The
h
m
i
is decreasing while
h
N
i
is increasing.
z
∼
[11
,
15]
Km: The
h
m
i
is increasing while
h
N
i
is decreasing.
Cumulative Probability of
q
l
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 5 10 15 20 25 30 35 P> (ql ;z) ql (x10−6) z=1516.91 m z=2641.41 m z=4129.88 m z=5418.46 m 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 10 20 30 40 50 60 70 80 90 P> (ql ;z) ql (x10−6) z=5830 m z=7812.87 m z=9000.32 m z=10800 m 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 10 20 30 40 50 60 70 80 90 100 P> (q l ;z) ql (x10−6) z=11400.00 m z=12600.00 m z=13800.00 m z=15300.00 mDistribution of CAPE and Level of Neutral Buoyancy
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0
500
1000
1500
2000
2500
3000
P(CAPE)
CAPE (J Kg
-1)
0
0.05
0.1
0.15
0.2
0.25
0.3
10500 11000 11500 12000 12500 13000 13500
P(LNB)
LNB (m)
Size distribution of clouds
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1 2 3 4 5 6 7 8 9 10 P(A) A ( Grid cell) z=2976 m <A>≅ 1 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 1 2 3 4 5 6 7 8 9 10 P(A) A ( Grid cell) z=8403 m <A>≅ 2 0 0.05 0.1 0.15 0.2 0.25 0.3 1 2 3 4 5 6 7 8 9 10 P(A) A ( Grid cell) z=10500 m <A>≅ 3Exponential distribution P(M,z) above LNB
z
= 11400
m
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 50 100 150 200 250 300 350 400 450 500 M (Kg M−2 s−1) P(M;z) Measured PDF Fit 〈 M 〉=0.0017 and 〈 N 〉=1.021 Exponential fit 〈 m 〉=0.0027 Measured 〈 M 〉=0.002 and 〈 N 〉=3.41z
= 12900
.
00
m
0 0.005 0.01 0.015 0.02 0.025 0 100 200 300 400 500 600 700 M (Kg M−2 s−1) P(M;z) Measured PDF Fit 〈 M 〉=0.00066 and 〈 N 〉=0.5818 Exponential fit 〈 m 〉=0.00152 Measured 〈 M 〉=0.000868 and 〈 N 〉=1.82 0 10 20 30 40 50 60 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 N P(N;z) Measured PDF Fit 〈 N 〉=0.95Poisson PDF with estimated cluster 〈 N 〉=3.4/3
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 N P(N;z) Measured PDF Fit 〈 N 〉=0.5