“Who then beheld the figures of the clouds
Like blooms secluded in the thick marine?”
W allace Stevens (1879–1955)
3D Cloud Properties & Climate
Robert.F.Cahalan@nasa.gov
Head, Goddard Climate & Radiation Branch
Proj Scientist: 3DWG (I3RC), THOR, and SORCE
Observations
Radiative Transfer
Convection
Bell
Davis
Marshak
McGill
Oreopoulos
Ridgway
Varnai
Wen
Wiscombe
…
M. Budyko
(1920-2001)
Theme …
1. Moisture Evap, Precip, Runoff
2. Energy Albedo, Atmos, Sfc
3. Feedback Ice (+), Cloud (-)
4. Global Warming & Life
Decades …
1940’s – 50’s
1950’s – 60’s
1960’s – 70’s
1970’s – 90’s
Budyko’s Bucket: Soil Capacitance, Physical and Bio-physical
“…Assuming Earth’s albedo as α s = 0.33, we find short-wave radiation absorbed …167 kcal/cm2/yr.”
Quantity …
Qo
α
s
Qo
α
(1- s) Qo
Atmos Abs
Sfc Abs

Budyko …
Kiehl & Trenberth…
333
342 (+9)
111
107
2
222
X 1.3327 222W/m
(-4)
235 (+13)
55
67 (+12)
167
168 (+ 1)
“The precision
of these data is
of importance
in the study of
climate.”
Stratus, deep convection, cirrus …
We observe many cloud types …
Morphological diversity of marine stratocumulus clouds
“Who then beheld the figures of the clouds
Like blooms secluded in the thick marine?”
W allace Stevens (1879–1955)
Cloud vortices seen by MISR
Von Karman vortex street near Jan Mayen Island
6 June 2001
Time interval from 70º forward to
70º backward view: 7 minutes
“Look when the clouds are blowing
And all the winds are free:
In fury of their going
They fall upon the sea.”
Frederick W illiam Henry Myers (1843–1901)
D. Diner
Future: CloudSat radar will see cloud drops
(not just rain drops like TRMM)
• with complementary
measurements from
other cars on “the Atrain”:
• - CALIPSO: lidar
• - PARASOL: polarized
radiances (French)
• - Aqua, Aura: last
great multi-instrument
EOS platforms
===== Act II, Scene 1
I3RC
phase
Imag e o f sc e ne
Sc e ne desc riptio n
S imple array o f clou d
slabs
250 m
18
I3RC
Phases
and
Cases
1
2
18
2
18
2
Status
Comp leted
250 m
X-Z field observe d by
cloud ra dar
Comp leted
X-Y fields of optica l a nd
geom e trica l cloud
thicknesses
Comp leted
3D str a toc umulu s fields
from Larg e Ed dy
S imula tions
Comp leted
3D c umulu s fields fr om
Large Edd y S imula tions
Comp leted
3D clou d field from
combin e d
MODIS /MIS R/ AST ER
observation s , usin g
comple x surface
In
progress
2
Now in phase 3
3
Sprea ding of lidar pulses
inside clouds
In
progress
I3RC codes demonstrated high accuracy and flexibility
===== Act II, Scene 2
Cloud fraction
is insufficent
for radiation,
and cloud
overlap is an
inadequate
“band-aid”
... none of
the 1D
schemes
work well
Validation of THOR geometrical
cloud thickness retrievals
THOR
System
QuickTimeª and a
Graphics decompressor
are needed to see this picture.
1000
Flight altitude:
8060 m
5020 m
900
7320 m
8540 m
7-8
16 6
11
15
800
42
3
5
12
700
9
THOR retrieval
(m)
14
600
Cahalan
et al. (2005)
13
10
1
Image from DOE ARM program
500
500 600 700 800 900 1000
THOR + ARM estimate (m
Remote Sensing depends on 3D Radiative Transfer,
while Climate Models depend on 3D Cloud Structure
Independent Column
Plane-Parallel
small pixels
have ICA
error;
large ones
have PP
biases
Retrieved properties of cloud and
aerosol depend upon 3DRT …
αϖεραγεδ
ιλλυµινατεδ
σηαδοωεδ
270
275
280
285
Brightness
290Temperature (K)
295
10
12
14
effective radius (
16
18
20
µ µ)
Effect of scattering angle on the
retrieval of cloud droplet
effective radius (from Vant Hull
et al., 2006).
Observed aerosol reflectance vs 3D
modeled reflectance at the same
points assuming constant aerosol
optical depth (from Wen et al.,
2006).
… accounting for 3DRT alters
observed correlation of cloud and
aerosol
ASTER image of a sample scene, and
MODIS cloud optical thickness product
for the same scene. Cloud coverage is
53%, the average cloud optical thickness
is 12. Aerosol optical thickness is near
0.2 at 0.47 µm and near 0.1 at 0.67 µm
wavelength. Black rectangles highlight
areas selected for detailed analysis and
the black dots identify pixels used in
operational MODIS aerosol retrievals.
1D cloud-aerosol relation from above scene,
and 3D relation for same scene. Aerosol
optical thicknesses were corrected using 3D
effects estimated from 3D Monte Carlo
simulations. Mean droplet effective radius of
each 10 km area was corrected by excluding
1 km-size pixels where strong local
brightness temperature variability indicated
large 3D-related retrieval uncertainties.
===== Act III
Can dynamical cloud models be
driven by 3d radiation?
Cloud resolving models have tall narrow
columns that (so far) exclude horizontal
radiative fluxes.
3DRT produces hot and cold spots.
Simulations of nocturnal stratocumulus
driven by 3DRT infrared cooling have
shown enhanced convection (Kagan et al.)
Daytime convective run with 3DRT heating
W. O’Hirok, UCSB
WRF 3DRT
non-hydrostatic
2D – 400/512 columns x 80 layers
250 m layer and column resolution
240 minute simulation
3 second time step
5 minute radiation time step
shortwave - 3D*
longwave -1D RRTM
microphysics - water, ice, graupel, snow, rain (Lin et al.)
surface layer physics - Monin-Obukov
turbulence – TKE scheme*
ocean/soil surface layers - thermal diffusion
open/periodic boundaries
initial temperature perturbation
* modified for air-surface exchange
Two experiments comparing generated cloud
fields using *3D and ICA radiation schemes
A. Deep convection over water - periodic boundary conditions for
dynamics and radiation.
B Deep convection over land - open x boundary for dynamics, periodic
boundary for radiation.
3D radiation on 2D field
*
periodic boundary 35 min
periodic boundary 55 min
periodic boundary 110 min
periodic boundary 195 min
periodic boundary 220 min
Periodic boundary
15
domain average condensate
(liquid, ice, snow, graupel,rain)
0 - 240 min
3D
ICA
height (km)
height (km)
0 - 80 min
3D
ICA
15
cloud fraction
(liquid, ice)
0
.0
g kg-1
.25
0
.0
fraction
.40
Domain Average
Albedo
(3D/ICA)
Transmission
(3D/ICA)
Atm. Heating
(3D/ICA)
1.6
1.6
1.2
1.0
1.0
1.0
0.4
0.4
0.8
Time 13:00
17:00
13:00
17:00
13:00
17:00
Domain average ratios (3D/ICA)
2.0
accumulated
precipitation
1.0
0.0
2.0
latent Heat
1.0
0.0
5.0
vertical wind
variance
1.0
0.0
1.2
atmospheric
absorption
0.0
0.8
0
60
120
simulation time (m)
240
cloud condensate (g/kg)
ICA
13:40
8.0
SW heating rate (K/hour)
ICA
4.0
2.4
2.0
1.6
1.2
2.0
0.8
1.0
0.4
0.2
0.5
0.1
0.
3D
8.0
4.0
2.0
0.0
3D
2.4
2.0
1.6
1.2
0.8
1.0
0.4
0.2
0.5
0.1
0.
0.0
14:40
ICA
3D
cloud condensate
cloud condensate
6 g/kg
4.1 g/kg
SW heating rate
SW heating rate
1.2 K/hour
vertical velocity
2.4 K/hour
vertical velocity
+11 m/s
-20 m/s
-19 m/s
0
+20 m/s
km
100
0
km
100
Summary
･ R em ot e sen si n g & m od eli n g of clou d s r equ i r es 3D
T r a n sfer .
･ P u bli c 3D R T cod es a r e n ow a v a i la ble t h a t a r e
R a dia tiv e
Accurate, Precise, Diverse, Flexible, Rapid, Scalable, Understandable.
･ A p p li ca t i on s ex t en d bey on d clou d s to
vegetation, terrain, ice/snow, … 3DRT causes components to interact.
･ D y n a m i ca l clou d m od eli n g w i t h 3D R T i s n ow i n i t s i n fa n cy .
･ C on v ect i v e clou d m od els a r e beg i n n i n g t o a p p ly bot h ex a ct
a n d a p p r ox i m a t e 3D R T .
“Joy in looking and comprehending is
nature’s most beautiful gift.”
Albert Einstein (1879-1955)