Geology Reference
In-Depth Information
Ice crystals emit higher energy than water molecules in
the microwave region; i.e., the emissivity of ice is sig-
nificantly higher than that of water. As mentioned before,
this is the premise for using passive microwave sensing for
discriminating sea ice from its surrounding seawater.
Moreover, the gradual evolution of emissivity from the
low values for OW to the higher values of ice can also be
utilized to identify thin ice and estimate its thickness.
Stringer et al.
[1984] compared microwave brightness
temperature from OW and sea ice using the following
numerical example. For a wavelength of 1.5 cm (nearly
19 GHz) the brightness temperature from OW at 273 K
with emissivity 0.5 is 136 K and for FY ice at 250 K with
emissivity 0.92 it is 230 K [from using equation 7.28].
Therefore, although it is physically colder than the sur-
rounding OW, ice emits more energy in the microwave
band and therefore becomes radiometrically warmer.
In general, observational frequencies of the operational
PM sensors are selected within the range 4.9-94 GHz such
that spectral regions of high atmospheric opacity (20-
24 and 40-80 GHz) are avoided. Operational frequen-
cies of PM sensors shown in Table 7.3 are suitable for
sea ice applications in the polar regions because clouds
emit low radiation in this frequency range, especially at
the lower frequencies. Due to the weak radiation emit-
ted from Earth's surface in the microwave region, the
spatial resolution from PM sensors is usually coarse (a
few kilometers or tens of kilometers). This does not
allow resolving the retrieved information at small scale
to identify, for example, leads of a few hundreds of
meters in dimensions. However, the wide swath of the
data is useful for regional and global ice monitoring.
Passive microwave remote sensing has provided the
most comprehensive long‐term observations of sea ice
in all regions.
Recent operational microwave sensors that have been
used widely in sea ice applications include SSM/I, AMSR‐E
and AMSR‐2. Details about their radiometric and geo-
metric characteristics can be found in
Hollinger et al.
[1987],
Kawanishi et al.
[2003] and
Kachi et al.
[2008];
respectively. The radiometers on these sensors are conical
that look backward and scan the swath at a constant angle
(53° in case of SSM/I). The scan lines draw nonconcentric
arcs that tend to converge more toward the edge of the
swath as the distance from the ground track of the sub‐
satellite point increases (Figure 7.24). This viewing geom-
etry produces footprints in the form of ellipses with minor
and major axes in the direction of scan lines and across
scan lines, respectively. More overlap of the footprints is
visible near the end of each scan line. The dimensions of
the footprint of observation from commonly used chan-
nels in sea ice application from SSM/I, AMSR‐E, and
AMSR‐2 are included in Table 7.4. These are the inte-
grated field of view (IFOV) dimensions.
Satellite position
3
2
1
Scan angle
3
2
1
Grid point
IFOV
Figure 7.24
Scanning geometry of SSM/I showing footprints of
a given radiometric channel (not to scale). Data are sampled
over 104.2° from end to end per scan line.
Table 7.4
Ground dimensions of the effective field of view for
each channel of commonly used passive microwave sensors
in sea ice.
Resolution
(IFOV)
Grid Pixel
Dimens
i
ons
Frequency
(GHz)
Along
Scan
Cross
Scan
Along
Scan
Cross
Scan
SSM/I
19
43
69
25
25
37
28
37
25
25
85
13
15
12.5
12.5
AMSR‐E
18.7
16
27
10
10
36.5
8
14
10
10
89
6
4
5
5
AMSR2
18.7
14
22
10
10
36.5
7
12
10
10
89
3
5
5
5
The sampling of the measured radiation can be
achieved at a rate finer than that of the footprint of the
radiometer. This is common when generating gridded
data. For example, the National Snow and Ice Data
Centre (NSIDC) resamples the SSM/I 19 and 37 GHz
data at 25 km × 25 km grid spacing. This is less than the
IFOV dimensions of the same channels, as shown in
Table 7.4. Therefore, the observation at each grid point is
produced from about 7-9 overlapped footprints in the
case of the 19 GHz channel and 4-5 footprints in the case
of the 37 GHz channel (more pixels are averaged near the
end of the swath than at nadir, as shown in Figure 7.24).
On the other hand, data from the 85 GHz channels were
sampled at 12.5 km pixel spacing. This is very close to the
