Geology Reference
In-Depth Information
freeboard portion of MY ice [
Steffen et al
., 1992].
Maslanik and Key
[1993] developed a technique that
combines AVHRR and SSM/I data to derive more accu-
rate estimates of IST by incorporating the estimated ice
concentration from passive microwave data into the esti-
mated temperature from a heterogeneous footprint. It
also allows estimating IST under cloudy sky.
There are two main differences between the TIR and
microwave observations that impact the way the surface
temperature is retrieved from the observations. The first
is that the radiances in the TIR spectral region are more
affected by the atmosphere. That is the reason for using
the split‐window technique to account for atmospheric
contribution to the emitted radiation as explained in
section 7.5. The second is the large difference in emissiv-
ity between open water and sea ice types in the micro-
wave bands compared to the insignificant difference in
the TIR band. For that reason the concept of the “com-
posite” emissivity to account for the heterogeneity of
emissivity within the footprint from a microwave sensor
has been developed (the footprint typically measures a
few kilometers or tens of kilometers).
The first algorithm to calculate ice surface tempera-
ture from passive microwave observation was applied to
SMMR data [
Gloersen et al
., 1992]. A simple relation-
ship between temperature of the radiating layer
T
RL
and
the brightness temperature at 6.6 GHz vertical polariza-
tion is used:
where
total
is the total surface emissivity based on the
emissivity tie points of the retrieved ice types,
T
air
(
T
e
) is a
weighted atmospheric temperature as a function of
T
e
,
and
τ
(
T
e
) is the atmospheric attenuation coefficient. All
variables in the above equation are either measured or
estimated; and
T
e
can be determined using an iterative
approach to solve the equation. The ice temperature
T
ice
can then be determined based on the assumption that the
footprint includes two ice types, FY and MY ice with
concentrations
C
fy
and
C
my
, respectively:
TT C
272 .
OW
C
C
ice
ice
e
OW
FY
MY
ice
ice
(10.100)
where
ice
is the composite emissivity given by the
following equation:
ice WOWFYFY YMY
C
C
C
(10.101)
Other forms can be constructed similarly according to
the number and types of the ice composition. Similar to
the algorithm by
Gloersen et al
. [1992], ice temperature
can only be determined from microwave observations if
the footprint contains more than 80% ice.
Steffen et al
. [1992] presented a map of 9 year average
sea ice temperatures derived from SMMR observations
for the month of February and compared it with the
corresponding climatological map of surface air temper-
ature for validation purposes. Climatological data show
colder temperature in the central Arctic and warmer tem-
perature at the peripherals. This is compatible with the
expectation of thicker ice in the center. SMMR‐derived
ice temperature is a weighted mean temperature of the
radiating layer in the snow and ice and air temperatures.
Comiso et al
. [2003] used a simple algorithm to estimate
the temperature of the radiating layer from snow‐covered
ice using AMSR‐E data from the 6.9 GHz channel. The
algorithm was used by NASA to generate regular maps
of ice surface temperature in the polar regions (this was
one of three sea ice AMSR‐E standard products at 25 km
resolution; the other two maps were ice concentration
and snow depth). The method involves calculation of
the composite emissivity
ice
from an equation similar to
equation (10.101) where ice types are combined into one
term with concentration
C
ice
. This is based on the assump-
tion that at 6.9 GHz frequency the emissivity of sea ice is
constant regardless of the ice type. The physical tempera-
ture of the radiating layer
T
RL
can then be determined
from equation
(
10.98) assuming a constant value of 0.98
for ice emissivity. Since this represents the composite tem-
perature from a likely heterogeneous footprint (contain-
ing ice and OW), the ice surface temperature
T
TRL − ice
(this
T
TV
V
66
66
.
.
b
(10.98)
RL
ice
A value of the ice emissivity
ice
= 0.96 is selected. The
6.6 GHz channel is chosen because of the deep pene-
tration of the signal. This ensures that the derived
temperature can be taken to represent the snow‐ice inter-
face temperature or the top layer of the ice surface.
However, the very coarse footprint of this channel
(150 km) means that its composition is almost always
heterogeneous. The calculated temperature is sensitive to
the heterogeneity and is more reliable when ice concentra-
tion exceeds 80%. For ice concentration <80% the algo-
rithm sets the surface temperature to 271 K.
This simple algorithm was modified as described in
Steffen et al
. [1992] to model the effect of the Arctic
atmosphere. The model is known as the University of
Massachusetts algorithm. The effective temperature
of the radiating layer
T
e
of a heterogeneous footprint
is related to the brightness temperature
T
b
according
to the following radiative transfer equation:
T
T
(10.99)
T
total
T e TTe
(
1
)
e
e
b
e
air
e
