Geology Reference
In-Depth Information
determine the fast ice thickness. The premise behind
using passive microwave observations is that the salin-
ity of thin ice near its upper surface varies with ice
thickness due to the rapid desalination process that
takes place during the initial ice formation phase
(Section 2.3.3.1). This, in turn, affects the complex
permittivity of the ice and consequently the emitted
radiation. Many studies have shown strong negative
correlation between the polarization ratio and the thick-
ness of thin ice in the absence of snow cover [
Martin
et al.
, 2004;
Tamura et al.
, 2007;
Hwang et al.
, 2007;
Tamura and Ohshima
, 2011;
Singh et al.
, 2011;
Iwamoto
et al.
, [2013].
Shokr et al.
[2009] verified this relation
even in the presence of snow (to some extent). The
polarization ratio at frequency
f
is given by equation
(8.10), but it can also take the following form, which is
commonly used in ice thickness retrieval algorithms:
175 km
1
(10.77)
PR
f
(
RR
1
)/(
)
f
f
where
0
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12 0.13 0.14
Ice thickness (m)
RTT
f
/
(10.78)
b fv bfh
,
,
Figure 10.29
AVHRR‐derived thin ice thickness (color) over-
laid on a Radarsat SAR image (B&W) of a coastal polynya in
the Bering Sea west of Alaska. Images were acquired on 9
January 1999. Black areas within the ice thickness map are
cloud masked. The circles and the squares show locations of
ULS (see text) and salinity sensors, respectively [
Drucker et al.
,
2003, Figure 4, with permission from AGU]. (For color detail,
please see color plate section).
While both PRf
f
and
R
f
are identical in terms of their use
to retrieve the ice thickness, the polarization ratio has an
advantage of being independent of the surface tempera-
ture. Most studies use PRf,
f
, but
Martin et al.
[2004] used
R
f
instead. The retrieval algorithm is a simple regression
equation that relates thin ice thickness to
R
f
or PR
f
(only
PR
f
is considered in the rest of this discussion). The
equation is generated from a scatterplot of the thickness
estimated using TIR data against PRf.
f
. An example of a
scatterplot between AVHRR‐derived thin ice thickness
and SSM/I PR
37
or PR
85
, is presented in
Tamura and
Ohshima
[2011] and reproduced in Figure 10.30. The data
are obtained from three different polynyas in the Arctic:
the North Water Polynya, the Chukchi Polynya, and the
Laptev Polynya. The ice thickness is correlated to the
polarization ratio up to about 15 cm. Beyond that limit
the thickness becomes insensitive to the polarization
ratio. Two least square curve fittings are shown in the fig-
ure; a linear and an exponential. Both are used in the ice
thickness retrieval as suggested by
Martin et al.
[2004].
The linear curve fitting is shown for data from each one
of the three aforementioned polynyas plus the overall
data set from the three polynyas. Following
Tamura and
Ohshima
[2011], the linear equations that give the ice
thickness
H
are
the locations marked by the circles in the figure to verify
the ice thickness results. The thickness from these sensors
at those locations was around 8 mm. The figure shows
that ice thickness increases away from the coast. This
contradicts the findings from the ULS measurements,
which show a decrease of thickness away from the coast.
Drucker et al.
[2003] provide a plausible explanation in
terms of Langmuir circulation, which causes the distribu-
tion of frazil at depth near the coast to become strong
while ocean wave damps the circulation away from the
coast causing less frazil distribution at depth.
10.4.2. Passive Microwave Observations
Passive microwave radiometry has been used to esti-
mate ice thickness of snow‐free thin ice. The algorithms
use observations from the medium‐ and high‐frequency
channels (37 and 85 GHz on SSM/I or 36 and 89 GHZ
on AMSR‐E). Most of the studies aim at mapping
sea ice thickness in polynyas and marginal ice zones,
although
Tamura et al.
[2007] made an attempt to
H
2 055
.
PR
0 1765
.
for
0 081
.
PR
0 0494
.
85
85
(10.79)
