Upload
ngonhan
View
216
Download
0
Embed Size (px)
Citation preview
Sea Ice, Climate Change and RemoteSea Ice, Climate Change and Remote
SensingSensing
Prof. David Barber
Canada Research Chair in Arctic System Science
Director, Centre for Earth Observation Science
University of Manitoba
Winnipeg, MB. Canada
www.umanitoba.ca/ceos
ESA Summer School, August, 2006
CEOSCEOS
Lecture outline
1) Arctic Climate Change and Remote Sensing
2) Geophysics, dielectrics and thermodynamics
3) Scattering and emission modeling
CEOSCEOS
Outline of this talk
• The Electro-thermophysical concept
• The complex dielectric constant
• Scattering and emission models
• A few examples
• Conclusions
CEOSCEOS
Three key features of the Arctic:
1) it is cold
2) it is dark
3) it is cloudy
4) it is changing
CEOSCEOS
Review of Sea Ice Microwave
Scattering Theory
• Microwave scattering from Sea Ice is
controlled by three factors:
– 1) The Complex Dielectric Constant
– 2) The inhomogeneities of Scattering
Inclusions
– 3) The Frequency, Polarization and Sensor
Geometry of the SAR
CEOSCEOS
Complex Dielectric
!*=!"+j!#
Ice Type
Ice Thickness
Ice Salinity
Ice Temperature
Snow Depth
.
.
Multi-
frequency
& polarized
EM
Signatures
Forward Approach
Inverse Approach
Freeze onset date Radiative
transfer model
The electromagnetic properties of sea ice
IEEE TGARS, ONR ARI special issue. 36(5): 1750-1763
CEOSCEOSBarber et al. 1998
0-5-10
20
0
30
60
90
Depth
(cm
)
-5-10-15-20 -5-10-15 -5-10-15-20-20
Temperature (°C)
-15-5-10-15-20
Average T°
Diurnal !
Show
twave
Flu
x (
K ) 400
200
0
Snow
Sea Ice
5
Winter ablation 1 ablation 2 ablation 3 ablation 4 ablation 5
0-5-10 -15
Temperature is the control
CEOSCEOS
An electro-thermophysical model of snow
covered sea ice
snow
ice
QH QELd LuKd Ku
K*o
K*B=Q*is
K*1x=Q*1x
Qso
Qs1x
QsB
Qio
Z=1x
Z=2x
Z=Nx
Z=0x
Z=B
QM
Q* Q*
Density (kg m-3)
Salinity (ppt)
Liquid (% by vol.)
Mass
CEOSCEOS
Snow
Sea Ice
0.0
6.0
9.0
12.0
15.0
18.0
Snow
Depth
(!s;
cm
)10.0 20.0 30.0 40.0
Snow Grain Size (mm-2)
-8 -7 -6 -5 -4
Snow Temperature (Ts; °C)
0.00 0.02 0.04 0.06
Snow Brine Volume (Vb; %/100)
0.0 100 200 300 400 500
Snow Density ("s; Kg·m-3)
Grain Size3.0
Snow-Ice
Air-Snow
"s
#°
Vb
Geophysics
Thermodynamics
Energy Flux
Mass Flux
Gas Flux?
CEOSCEOS
The Temporal Evolution of !º
..
-20
-15
-10
-5
Multiyear
First-Year
Winter EarlyMelt
MeltOnset
AdvancedMelt
Pen
dula
r
Funic
ula
r
Pondin
g
Dra
inag
e
!°
(ER
S-1
)
Freeze-up
CEOSCEOS
Coupled sea Ice thermophysical and
dielectrics model.
•The complex dielectric constant
is defined as:
!% = !" + j!#
!" is the permittivity
!# is the loss
CEOSCEOS
Frequency/Polarization and
Sensor Geometry
!°total =
!°ss
+ "as
(#) *
!°sv
(#') + "s(#')
* !°
is + "
si (#") * !°
iv (#")
Snow Surface
Snow Volume
Ice Surface
Ice Volume
!h and L
Ri, Rw, $s, Wv, Ss, %s,
!h and L
Ri, Ra, Rb, $i, Is,
FrequencyPolarization
# of LooksIncidence Angle
CEOSCEOS
The Complex Dielectric Constant
•The complex dielectric constant consists
of a complex number
!% = !" + j!#
!" is the permittivity
!# is the loss
CEOSCEOS
The Complex Dielectric Constant
!
" # w = #w$ +#w0 %#w$
1+ (2&f'w )2
!
" " # w =2$f%w(#w0 &#w')
1+ (2$f%w )2
!
"w0(T ) = 88.045# 0.4147T + 6.295$10
#4T2+1.075$10
#5T3
!
2"#w(T ) = 1.1109$10
%10% 3.824 $10
%12T + 6.938$10
%14T2% 5.096$10
%6T3
where w0 is static dielectric constant of pure water, w! is high-frequency (or optical) limit of w ( w! = 4.9) ,
w is relaxation time of pure water (s); f is electromagnetic frequency (Hz).
The Debye Model - Water
CEOSCEOS
The Complex Dielectric Constant
!
" # b = #w$ +#b0 %#w$
1+ (2&f'b )2
!
" " # b =2$f%b(#b0 &#w')
1+ (2$f%b )2
+(b
2$#0f
!
"b0 =
(939.66#19.068T )
(10.737# T )
!
"# =(82.79+ 8.19T
2)
(15.68+T2)
!
2"# = 0.10990+ 0.13603$10%2T + 0.20894 $10
%3T2+ 0.28167$10
%5T3
The Debye Model - Brine
CEOSCEOS
Dielectric Mixture Models
Inclusion dielectric
in an air
background
!
"si = " i +3"si fa(1#" i )
2" si +1+" si fb("b #" i )
3
1
" si(1# Ni )+ Ni"b+
1
"si(1+ N1)+"b(1# N
1)
$
% &
'
( )
!
"ds
= 1+ 3"dsvi
"i#1
"i+ 2"
ds
$
% &
'
( )
!
"ws
= "ds
+mv"ws
3("
w#"
ds) ["
ws+ ("
w#"
ws)N
i]#1
i=1
3
$
Dry snow
Wet snow
Sea Ice
CEOSCEOS
Dielectric Mixture Models
. .
2
2.5
3
3.5
4
4.5
5
0 10 20 30 40 50
Ta = -5.0°C
Snow depth (Ds) in cm.
Si= 10pptDi = 100cmFi = 5GHz!s = 350kg
Ta = -10.0°C
Ta = -20.0°C
Ice
Surf
ace
Per
mit
tivit
y ("')
Change in !' and !'’ at the sea ice surface as a function of air temperature and snow thickness.
CEOSCEOS
Dielectric Mixture Models
. .
0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0 5 10 15 20
!s=300 kg·m-3" = 5.3 GHz
Salinity
#' m
ix
0 5 10 15 20 25
Salinity
#"
mix
!s=300 kg·m-3" = 5.3 GHz-4°C
-12°C
-20°C
0.00
0.05
0.10
0.15
0.20
0.25
0.30
-4°C
-12°C
-20°C
Modeled permittivity () and loss () as a function of snow salinity and snow temperatures.
CEOSCEOS
Dielectric Mixture Models
Water Volume (%/100)
!"
1
1.5
2
2.5
3
3.5
0.0
01
0.0
11
0.0
21
0.0
31
0.0
41
0.0
51
0.0
61
0.0
71
0.0
81
0.0
91
0.1
01
0.1 gm·cm-3
0.2 gm·cm-3
0.3 gm·cm-3
0.4 gm·cm-3
0.5 gm·cm-3
Frequency = 5.3 GHz
Bulk dielectric permittivity (!') of snow as a function of water volume and snow density
CEOSCEOS
Link between dielectrics and
scattering/emission
Snow
Columnar layer
S1 , TPhy1
S2 , TPhy2
S3, TPhy3
!1", !1""
!2", !2""
!3", !3""
Frazil layer
T(z)
Layer properties defined by the Fresnel reflection (&) coefficient
'。total = '。ss
+ (as
()) * '。sv
()') + (s()')
* '。is + (
si ()") * '。iv ()")
CEOSCEOS
Classes of Models
DMRT; Dense Medium Radiative Transfer; DMT: Dense Medium Theory, PS: physical optics under the scalar
approximation, GO: geometric optics approximation, SP: small perturbation method
Volume scatteringRough surface
scatteringMulti-layer reflection
Integrated surface-
volume signature
model
Reflection-volume
signature
Multilayer Fresnel formula
•No scattering effects
•Applicable only to
Young saline ice
Many layer SFT
•Scattering effects
to some extent
•Applicable to
Young saline ice,
frozen melt ponds
(up to 40 GHz)
•bubbly ice (up to
20 GHz)
Rayleigh scattering
DMRT
DMT
•No reflection effects
•Poor agreements (at
H-pol) over young ice,
frozen melt ponds
•Bubbly ice up to 19
GHz
DMT-integration model
PS
GO
SP
Integration
•No significant
improvement from
DMT
•This indicated no
significant rough
surface scattering
effects
Numerical method
Empirical method
•Good potential
•Backscattering
problem
•Snow application
(Winebrenner et al., 1992; Nassar et al., 2000; Wiesmann and Matzler, 1999)
CEOSCEOS
Emission/Scattering Theory
Snow
Columnar layer
S1 , TPhy1
S2 , TPhy2
S3, TPhy3
!1", !1""
!2", !2""
!3", !3""
Frazil layer
T(z)
oooo
sosvossokTkTk !!!! )()()( 22
++=
Backscattering
siaaTkkT )()0,( 00 !=
! +""= #$$%%&
' ddkkkkRkvaaha
sin)],(),([4
11)( 00
2
0
Emission
snow surface
snow volume
Snow/sea ice
interface
Interface
reflection
Volume
Scattering loss
! += "##$$%
ddkkkkRoavoahs
sin)],(),([4
1
CEOSCEOS
'。total = '。ss
+ (as
()) * '。sv
()') + (s()')
* '。is + (
si ()") * '。iv ()")
Snow Surface
Snow Volume
Ice Surface
Ice Volume
'h and L
Ri, Rw, *s, Wv, Ss, +s,
'h and L
Ri, Ra, Rb, *i, Is,
FrequencyPolarization
# of LooksIncidence Angle
Forward scattering model defined for
Microwave Scattering
CEOSCEOS
physical optics/geometric optics
• Surface Scattering
!
" <#
32cos$RMS height
Correlation length
CEOSCEOS
Volume scattering
0.0
6.0
9.0
12.0
15.0
18.0
10.0 20.0 30.0 40.0
-8 -7 -6 -5 -4
Ts
Vb
!s
Grain Size3.0
Snow-Ice
Air-Snow
Number density
Volume fraction
Scattering physics
CEOSCEOS
Forward scattering model defined for
Microwave Scattering
!" (#) = 2 | $%%
|2 cos2 # exp (-4 K
o
2 !
2 cos
2 #)
• &n=1
!
n!
(4Ko
2 !
2cos
2 #)
n
•
(4Ko
2 sin
2 # + n
2/ l
2)3/2
(Ko
2 n / l)
s
!HH
= "
2 x cos # + "
1 x cos # '
"2 x cos # - "
1 x cos # '
Where:
!1 =
"' + ""2
1, for material #1 (air)
Surface Scattering
CEOSCEOS
Forward scattering model defined for
Microwave Scattering
!v
o(") =
2Ke
!v cos "
'
(1 -
(exp (Ked sec("
')))
2
1 )s
!v = N
i !
bi
+ Nw !
bw
Where:
!b =
"o
4
64 #5 r6
|K|2
Volume Scattering
CEOSCEOS
Features in the mm range affect scattering
0.0
6.0
9.0
12.0
15.0
18.0
Snow
Dep
th (!s;
cm
)
10.0 20.0 30.0 40.0
Snow Grain Size (mm-2)
-8 -7 -6 -5 -4
Snow Temperature (Ts; °C)
0.00 0.02 0.04 0.06
Snow Brine Volume (Vb; %/100)
0.0 100 200 300 400 500
Snow Density ("s; Kg·m-3)
Grain Size3.0
Snow-Ice
Air-Snow
"s
#°
Vb
Ehn et al.
0.0
6.0
9.0
12.0
15.0
18.0
10.0 20.0 30.0 40.0
-8 -7 -6 -5 -4
Ts
Vb
!s
Grain Size3.0
Snow-Ice
Air-Snow
CEOSCEOS
Many layer SFT model (scattering and emission)
(Winebrenner et al., 1992)
Description
-assumes the snow/sea ice is a piecewise-continuous random
medium and accounts for the interference between waves reflected
and transmitted coherently by the various planar layers
- accounts for the mean propagation and first-order multiple scattering
effects by using bilocal and distorted born approximations
CEOSCEOS
(Winebrenner et al., 1992)
CEOS
Input parameters
General: frequency (GHz), angles
For ice: temperature, salinity, density, ice grain size (mm), air
bubble size, brine aspect ratio and tilted angle.
For snow: temperature, density, snow wetness (fractional volume)*,
snow grain size (mm).
*liquid water distributed between grains as well as around grains
(Stogryn, 1985).
Output parameters
Microwave brightness temperature (emissivity) and backscattering
(sigma) for V and H polarizations.
Many layer SFT model (scattering and emission)
CEOSCEOS
Extending signatures temporally - Ice emissivity simulated by the
many layer strong fluctuation theory model. The brine skim/wet slush
was set to be 5 mm, and ice salinity based on field data
Hwang and Barber, JGR, in press
CEOSCEOS
Snow Thickness
• Scattering Response. .
-26 -24 -22 -20 -18 -16 -14 -12 -10 -8 -6 -4
-24
-22
-20
-18
-16
-14
-12
-10
-8
-6
-4
-26
!° (dB) - May 6 (Clear)
!°
(dB
) -
M
ay 9
(C
loudy)
N= 900 x 2
May 6 (Clear) May 9 (Cloudy)
Barber and Thomas, 1998
0.0
6.0
9.0
12.0
15.0
18.0
10.0 20.0 30.0 40.0
-8 -7 -6 -5 -4
Ts
Vb
!s
Grain Size3.0
Snow-Ice
Air-Snow
Snow water equivalent (SWE)
Sea ice
CEOSCEOS
5 10 15 20
Site Number
0
20
40
60
Dep
th (
cm
)
0
60
115
170
225S
WE
(mm
)
Snow
thicknessSWE
SWE and Radiometry (20 sites)
Barber et al. 2000.
CEOSCEOS
Explained variance versusƒ,P, and )
0
10
20
30
40
50
60
70
80
90
100
19V
30
19V
50
19H
35
19H
55
37V
40
37V
60
37H
45
85V
30
85V
50
Frequency (19, 37, 85)Polarization (V, H)Incidence (30-60 in 5º increments)
PercentExplainedVariance
Barber et al. 2000.
CEOSCEOS
Explained variance versusƒ,P, and )
0
10
20
30
40
50
60
70
80
90
100
19V
30
19V
50
19H
35
19H
55
37V
40
37V
60
37H
45
85V
30
85V
50
Frequency (19, 37, 85)Polarization (V, H)Incidence (30-60 in 5º increments)
PercentExplainedVariance
Barber et al. 2000.
Tb and Ts
CEOSCEOS
Late season sea ice
Melt Pond DielectricsMelt Pond Dielectrics Snow Patch DielectricsSnow Patch Dielectrics
!% = !" + j!#
!% = 65.8065.80+ j36.5136.51
for pure water at
0°C, 5.3 GHz
for wet snow at
-1°C, 0.3 gm.m-3, 0.1 Wv
!% = 1.911.91+ j0.110.11
CEOSCEOS
, = 0.40
))
))
Small Small -'-'
Ice Patches 0 .4 5
Light Melt Pond 0 .1 7
Dark Melt Pond 0 .3 8
Albedo 0 .4 0
, = 0.40
Ice Patches 0 .1 2
Light Melt Pond 0 .5 0
Dark Melt Pond 0 .3 8
Albedo 0 .2 8
..=0.28=0.28
))
))
LargeLarge -'-'
CEOSCEOS
The Temporal Evolution of
Sigma Naught
• Several variables must be taken into
consideration:
– The effect of incidence angle
– The effect of wind
– The contribution to backscatter (!º) by
volume and surface scattering as dictated
by the dielectrics of the system
CEOSCEOS
The Effect of Incidence Angle
• During periods of cold temperatures the dielectrics ofthe system are considered static and changes tobackscatter are a function of incidence angle andsurface roughness
• The Incident Angle Calibration Model (IACM)standardized !º to the near range of the RADARSAT-1 swath (!º)– The IACM explained in excess of 99% of the variability in
backscatter which resulted from changes in incidence angle
CEOSCEOS
The Effect of Wind
• Under calm conditions, volume scattering withinbare ice (!i) results in backscatter which isgreater than backscatter caused by surfacescattering from melt ponds (!m) or !i > !m.– This allows for an estimation of melt ponds from SAR
• Over melt ponds, there is an amplification of !º asa function of wind speed provided wind directionis orthogonal to the SAR pulse– Between 1.5ms-1 - 2.5ms-1 !i = !m.
– Above 2.5ms-1 !i < !m.
CEOSCEOS
Pond fraction (PF),
wind speed (W),
wind direction in
degrees (D) and
weather are all
indicated for each
image. The images
have been calibrated
to ASF gamma
values and areas of
low backscatter
appear dark. All
images are courtesy
of the Canadian
Space Agency (©
CSA, 2002)
CEOSCEOS
Surface AlbedoSurface Albedo
R2 = 0.913
! = 0.269 - 0.015 * "o
0.5
0.52
0.54
0.56
0.58
0.6
-21 -20 -19 -18 -17
Integrated Shortwave Al
bedo (
!)
0.5
0.52
0.54
0.56
0.58
0.6
! = 0.379 - 0.012 * "o
R 2 = 0.786
-18 -17 -16 -15 -14 -13 -12
Transect 8
Wind Speed = 3.2 m/s
July 3 Desc; 13: 27 UTC
Standard Beam 2
-28 -26 -24 -22 -20 -180.52
0.54
0.56
0.58! = 0.617 + 0.003 * "o
R2 = 0.1877
Transect 10
Wind Speed = 1.5 m/s
July 6 Asc; 23 :36 UTC
Standard Beam 6
RADARSAT-1 Scattering Coefficient ("o)
Transect 8
Wind Speed = 5.3 m/s
July 3 Asc; 23 :24 UTC
Standard Beam 5
Light Wind
Moderate Wind
Strong Wind
Extracting Summer Ice Information from SARExtracting Summer Ice Information from SAR
CEOSCEOS
The Temporal Evolution of !º
..
-20
-15
-10
-5
Multiyear
First-Year
Winter EarlyMelt
MeltOnset
AdvancedMelt
Pen
dula
r
Funic
ula
r
Pondin
g
Dra
inag
e
!°
(ER
S-1
)
Freeze-up
CEOSCEOS
Conclusions
• Electro-thermophysical model (heuristic to physical)
• Emission/scattering models
• Geophysical vs Thermodynamic state (processes)
• Initialization and steering of models, data
assimilation
• Scale related science (micro to hemispheric)
• Merger of environmental science and technologies
CEOSCEOS
Acknowledgements:
John Yackel
John Hanesiak
Tim Papkyriakou
Ryan Galley
Alex Langlois
Phil Hwang
John Iacozza
CJ Mundy
Rob Kirk
Sheldon Drobot
Theresa Fisico
Theresa Nichols
Chistina Blouw
Isabelle Harouche
Andrew Thomas
Klaus Hochheim
Jens Ehn
Mats Granskog
Wanli Woo
Randy Scharien
Katherine Wilson
Wayne Chan
Jennifer Lukovich
The real forces
behind this work