Geology Reference
In-Depth Information
conclusion applies in the case of the dual frequencies
(the figure shows the S‐band combinations with other
frequencies). It appears that the S‐band deteriorates the
classification accuracy perhaps because of its larger
footprint (8.2 × 5.8 m
2
) compared to (3.7 × 2.6 m
2
), (2.0 ×
1.4 m
2
), and (1.2 × 0.9 m
2
) from the C‐, X‐, and Ku‐bands,
respectively. Classification of old ice improves as more
combinations of frequency are used and when higher fre-
quencies are included. Once again, the value of this study
resides in its use of a multifrequency scatterometer sys-
tem, though it has not yet been envisioned for a future
space‐borne SAR.
The most commonly used sensor for estimating ice con-
centration has been the passive microwave (PM). That is
mainly because of the sharp contrast between emissivity
of water and ice and partly because of the wide coverage
that allows mosaicking the data for the entire polar region
on a daily basis. The disadvantage, however, is the large
footprint that generates the information at coarse resolu-
tion. The footprint is almost always heterogeneous, con-
taining a few ice types plus open water. Decomposition
of the single observation (brightness temperature) from a
footprint to its components from the different ice types
and open water is necessary in this case.
The fine resolution of SAR, on the other hand, ensures
the homogeneous contents of the pixel. Therefore, the
task of estimating the ice concentration is reduced to a
simple classification of the pixel. Ice concentration can
then be determined as the summation of ice pixels (or ice
type pixels) in a given area divided by the total number
of pixels. The challenge in using SAR for this purpose is
the wide range of backscatter from the open water due to
wind‐roughened water surface. It overlaps with the back-
scatter from most ice types (Figure 8.4). The premise for
estimating ice concentration from any remote sensing
instrument is the contrast between the radiometric meas-
urement from ice and open water.
In a heterogeneous footprint that comprises ice and
open water the following linear model decomposes the
observation
R
obs
into its components from the two entities
is commonly used:
10.2. Ice concenTraTIon
Sea ice concentration is defined as the fraction of ice
cover within a given area. This can be a footprint of a
satellite sensor or a delineated polygon in a satellite image.
The latter is commonly generated when producing the
operational ice charts. The charts include estimates of the
total ice concentration as well as the partial concentra-
tion of presumed ice types. In addition to total ice con-
centration, a few algorithms are capable of producing
concentration of a limited number of ice/surface types
(e.g., thin ice, MY ice, or ice with metamorphosed snow
at the surface). Lack of identifying ice types (at least the
major type) by the ice concentration algorithms is one of
the reasons that most operational ice centers still depend
on visual analysis of remote sensing data.
In addition to its importance for marine vessel and
structure operations, ice concentration is also an impor-
tant climatic and oceanic parameter. It plays an impor-
tant role in controlling the heat and moisture fluxes
between ocean and atmosphere. In winter, ice concen-
tration determines the latent and sensible heat as well as
the longwave radiative exchange between ocean and
atmosphere. In summer it affects the amount of
absorbed solar radiation by the ocean in polar regions.
Climatologists have become increasingly interested in
ice concentration in order to improve climate models by
assimilating the ice concentration data [
McLaren et al.
,
2006;
Caya et al.
, 2010;
Tietsche et al.
, 2013]. Ice con-
centration can be also used in determining a number of
climate‐related parameters such as surface albedo and
emissivity as well as heat and moisture fluxes between
the warm ocean and the cold atmosphere in winter.
Combined with an estimate of ice thickness, ice volume
can be determined. Additionally, combined with growth
rate of ice, salt migration from the sea ice to the under-
lying ocean water can also be determined. Ice concen-
tration is used also to define ice edge [
Comiso el al.
,
2001] and ice extent [
Zwally et al.
, 2002a].
Stroeve et al.
[2004] used ice concentration to make projections of
decadal trends of Arctic sea ice.
R CRC
,
where
C
1
C
(10.12)
obs
i
i ww
w
i
where
C
is the concentration,
R
i
and
R
w
are the typical
values (tie point) from ice and open water, respectively.
The larger the contrast between
R
i
and
R
w
the easier and
more accurate the calculation of
C
i
would be. The con-
trast is large enough in the cases of visible albedo, physi-
cal temperature, and microwave brightness temperature
but not the backscatter. Within the microwave region,
lower frequency channels provide higher contrast
between brightness temperatures from sea ice and
OW (much lower
T
b
is observed from OW as shown in
Figure 8.17). The two deviations from this favorable
contrast are found in the cases of very thin ice (as it
represents the transition between OW and developed
ice) and flooded ice surface during the spring and sum-
mer. In both cases the microwave radiation from the ice
and OW overlap.
Ice concentration is produced regularly on a daily or
weekly basis in the form of maps/charts at a few centers.
A partial list of those centers is included in Table 10.2
along with the links to access the products (more than ice
concentration).
