Geology Reference
In-Depth Information
images. The anisotropy parameter is not useful for ice
classification. It displays uniform gray tone distribution
across the image, which indicates that at least one second-
ary scattering mechanism is effective (Figure 7.35). The
correlation parameter seems to be useful only in identify-
ing thin ice as can be seen in the bright zone, which likely
represents an open or refrozen lead.
As more studies of SAR polarimetric data for sea ice
are expected to emerge in the future, more data sets will
be generated, hopefully accompanied with detailed meas-
urements of snow and sea ice parameters. Identification
of polarimetric data with surface conditions is important
because ambiguity of the scattering mechanism is mostly
caused by the surface conditions (particularly snow).
The multidimensionality of the radar polarimetric data
should allow linking the data to gross surface features
such as roughness, melting or snow metamorphism. With
a comprehensive database linking backscatter with ice
and snow conditions, it would be possible to retrieve
information on the snow metamorphosis, surface melt
onset, degree of surface roughness (i.e., pancake versus
rubble ice), thickness of thin ice, and better surface‐based
ice type classification.
observations from two different frequencies
f
1
and
f
2
at
the same polarization
p
is written as
TT
TT
bf p
,,
b fp
,
,
GR
fpfp
(8.11)
1
2
12
bf p
,,
b fp
,
,
1
2
These ratios account for the diurnal effects on sur-
face temperature as well as seasonal variations of the
temperature. Therefore, they are commonly used (instead
of the simple difference in polarization or gradient) to
compare observations obtained at different times of day
or different seasons.
It should be noted that ice and snow compositions may
change modestly or considerably in response to changes
in atmospheric temperature and that leads to change in
the emissivity of the medium. That is an indirect effect of
temperature change on the measured brightness tempera-
ture. This should be taken into consideration when
interpreting microwave brightness temperature observa-
tions. Data on microwave emissivity are presented in
section 8.4.
Many studies were conducted in the 1970s and 1980s
to establish databases of
T
b
for ice types and OW in the
microwave region. These efforts started after the launch
of the first passive microwave sensor (ESMR) on Nimbus
5 (section 7.2), which carried a single‐channel radiometer
operating in the 19.53 GHz (1.55 cm) with horizontal
polarization. One of the earliest records of
T
b
was com-
piled by
Parkinson et al
. [1987] from Arctic sea ice meas-
urements of EMSR over 4 years (1973-1976). The
authors presented a temporal plot of 3 day averaged
T
b
from FY and MY ice surfaces (Figure 8.16). It demon-
strates the steady values of
T
b
for both ice types during
winter months. MY ice has less
T
b
because air bubbles
scatter the energy before reaching the point of emission
at the surface. This makes identification of these two
types relatively easy in passive microwave data. The diffi-
culty starts in the late spring (May and June) and extends
throughout the summer. The figure shows a sharp drop in
brightness temperature from FY ice starting in June and
a milder drop from MY ice. The
T
b
from FY ice becomes
significantly lower than MY ice (note the exception in the
1975 data). The decrease must have been caused by mois-
ture increase in the snowpack in June then accelerated
due to melt pond formation then ice breakup. It reaches a
minimum in September with values close to
T
b
from sea-
water. Summer is the only period where
T
b
from FY and
MY overlap. During the fall, brightness temperatures are
restored to their normal winter values. The scattering
of the data points (i.e., the short‐term variation of
T
b
)
observed in Figure 8.16 might be caused by changes in ice
surface temperature, which impacts
T
b
directly according
8.2. MicRowave BRightness
teMpeRatuRe data
The definition of brightness temperature (
T
b
) is given
in Section 7.3.3. In the microwave region,
T
b
is the multi-
plication of the physical temperature of the medium by
its emissivity [equation (7.28)]. It is mentioned in the
same section also that the emissivity of sea ice, snow, and
OW in the TIR region are almost equal and their values
are between 0.95 and 0.96. Because of these stable values,
there is no need to establish databases of
T
b
in the TIR
region. In fact, the brightness temperature in this spectral
range is very close to the physical temperature.
Emissivity is an intrinsic radiative property of the
material, but brightness temperature is not because it
depends on the physical temperature of the observed
medium. For that reason the polarization and gradi-
ent ratios (both derived from brightness temperature)
are usually used instead of brightness temperature
because they are independent of the physical tempera-
ture. The polarization ratio PRf
f
at a given frequency
f
is defined as
TT
TT
bfv
,,
b fh
,,
PR
f
(8.10)
bfv
,,
b fh
,,
where
v
and
h
refer to vertical and horizontal polari-
zation, respectively. The gradient ratio between two
