Geology Reference
In-Depth Information
value for rough pond water at steep incidence angles
reflect the high values of
and lowest for DFYI (0.79-0.84). It decreases with the
incidence angle and can be used to discriminate between
DFYI and OW. On the other hand, the entropy H increases
with an increase in the incidence angle and can be used to
separate between DFYI and OW. The anisotropy A can
discriminate between OW and DFYI on one hand and
SFYI on the other hand. The high values of the alpha
angle α of DFYI are triggered by a mixture of surface
and volume scattering. The OW signature is affected by
the wind speed and direction relative to the radar line of
sight, which were not identified in this data set.
( Gill and Yackle [2012] present also the probability
density function of a few polarimetric parameters for
the same surface types (DFYI, RFYI, SFYI, and OW).
Figure  8.14 shows results of selected parameters. The
SFYI is separated from the other surface types using
0 . The co‐polariza-
tion ratio R vv / hh does not vary with incidence angle for FY
and MY ice but increases almost linearly with the angle
for the rough water surface. The standard deviation of
the three parameters at each angle is much less than the
corresponding values for the co‐ and cross‐polarization
parameters. This may be considered an advantage in an
ice discrimination scheme, but it should be noted that
R vv / hh and R depol have almost equal values for FY and MY
ice at all angles. Rough melt pond water produces lower
values of R depol than R vv / hh or the corresponding value for
shown ice types. Among the conclusions that can be
drawn from these graphs is that the co‐polarization ratio
can be used to discriminate between the rough water and
sea ice surfaces at high incidence angles.
Gill and Yackle [2012] published one of the first sets of
polarimetric signatures of ice types obtained from
Radarsat‐2. They explored a few parameters from three
surface types: smooth FY ice (SFYI), rough FY ice
(RFYI), and deformed FY ice (DFYI) in addition to
wind‐roughened open water at different ranges of radar
incidence angles. The data were obtained from nine
Radarsat‐2 fine mode images in quad‐polarization at spa-
tial resolution between 5.2 and 7.6 m in April 2008. The
study area comprised land‐fast FY and marginal ice
located on the east and west of the Parry Peninsula in
Franklin Bay (70°N and 125°W), western Arctic. The
images vary in incidence angle, with scene centers ranging
between 22.26° and 37.04°. Table 8.5 includes results of
11 polarimetric parameters obtained at 4 different radar
incidence angles. The parameters are defined in sec-
tion 7.6.2.3. The authors related some of those parame-
ters to physical characteristics of the surface using field
measurements and observations. They also assessed the
use of the parameters, individually and in combination,
in ice classification schemes. The OW data in the table are
presented in terms of the incidence angle only without
considering the wind speed.
Data in Table 8.5 show that none of the examined sur-
faces strongly depolarizes the signal, although DFYI
causes limited depolarization as indicated by its relatively
high values of the cross polarization parameters. Some
parameters demonstrate a capability to discriminate between
ice and water or between ice types. The significantly lower
values of R hh / vv from OW compared to values from all ice
types make this parameter suitable for ice/OW discrimina-
tion. The SPAN parameter increases with surface roughness,
but it overlaps heavily between ice types and OW. Therefore,
it is not suitable as an ice discriminator parameter. The co‐
polarization phase difference ( φ hh vv ) is incapable of dis-
criminating between any ice types. The mean value of the
co‐polarized correlation coefficient ( r hh vv ) is highest for
OW (0.96-0.98), lower for  SFYI or RFYI (0.89-0.93),
0
and
hh
vv
0 ,
hh
0
are within the noise floor of the sensor. As expected,
deformed ice (DFYI) has remarkably higher
0 ,
0
and the SPAN parameter, but the values for
vv
hv
hv
0 since it
triggers a multiscattering mechanism and consequently
depolarization of the scattered signal. Open water can-
not be easily identified using any of the above parame-
ters. The apparent high variability in the co‐polarization
ratio R hh / vv for ice types limits its use as an ice classifier
parameter. Entropy is highest from DFYI and low-
est  from OW. Recall that high values correspond to
equal  contribution from all scattering mechanisms
and  low values signify a single scattering mechanism,
(section 7.6.2.3). Anisotropy is not useful for any surface
discrimination, while the alpha angle can be used to sepa-
rate RFYI. The striking observation from Figure 8.14 is
the heavy overlap of polarimetric parameters from the
given ice types. More data from same ice types are needed
to confirm the given distributions. Data from other ice
types in different seasons are also needed to compile a
more comprehensive database. The incidence angle may
also affect the ability of different parameters to separate
between ice classes.
Another set of multipolarization results from the co‐
polarized C‐band ASAR on ENVISAT are presented in
Geldsetzer and Yackle [2009]. Observations were obtained
from an area surrounding Cornwallis Island, Nunavut,
Canada, for the period 3 April-30 May 2004. The pur-
pose was to find out if the dual co‐polarized channels can
discriminate OW from sea ice (regardless of the type) and
if it can effectively discriminate thin ice from FY and MY
ice. A limited set of statistics of polarimetric parameters
from MY ice, FY ice (smooth and rough), thin ice, and
OW is presented. Visual analysis was also conducted
during the study using enhanced techniques to generate
color composite images from the dual polarization data.
In addition to the backscatter coefficients
hv
0
0 ,
and
vv
hh
0
0
the study uses the co‐polarization ratio
/
.
vv
hh
Search WWH ::




Custom Search