530
M. Hallikainen
the amount of ice experiencing summer melt is limited. The same polarization ratio/spectral
gradient ratio approach is used, but the coefficients in the two equations are slightly different
from those in Equations (19.10) and (19.11). In the absence of reliable in situ data, the two ice
types in the Southern hemisphere are called Type A and Type B ice (Parkinson and Gloersen,
1992). Figure 19.2 depicts the ice concentration nomograms for the Northern and Southern
hemisphere.
If hemispherically chosen brightness temperatures are used as tie points, the variations of TBw
(wind speed, foam, atmosphere), TBFY (surface roughness, temperature, snow cover, atmosphere), and TBMy (surface roughness, temperature, snow cover, atmosphere) over time and
space are ignored. To account for these factors, ice concentrations based on local tie points determined in the area of interest can be derived. Local-ocean tie point values are usually higher
than the hemispheric ones, due to e.g. wind speed. Differences as high as 18 K between the
hemispheric and local tie points for water have been observed (Steffen and Schweiger, 1991).
Other ice concentration algorithms include those by Swift et al. (1985), Svendsen et al. (1992),
Comiso (1986). The algorithm discussed above has been widely applied. Accuracy estimates
for the calculated sea ice concentrations are 5 to 9% for the Nimbus-7 SMMR data (Cavalieri
et al., 1984; Steffen and Maslanik, 1988).
The algorithm used for calculating sea ice concentrations from SSM/I data has the same formulation as the SMMR algorithm presented above; only the coefficients are different due to
slightly different channel frequencies.
A basic problem in developing ice type discrimination algorithms for the polar regions is poor
availability of in situ data. Validation of microwave observations by Landsat optical sensors
can be done only in clear weather conditions, even this may give a bias to the results.
In the Baltic Sea winter shipping traffic is handicapped by deformed ice (ice ridges, hummocks).
In the winter of 1992 classification of ice types and water with a microwave radiometer was
investigated. Especially, measurements were made on compact pack ice and close pack ice that
are difficult for ship navigation. Compact pack ice is formed from crushed fast or new ice,
which has been packed and refrozen. This ice type is common in the open sea where waves
and wind crush existing fast or new ice. The ice surface is rough and, hence, existing snow is
spatially unevenly distributed. Close pack ice is formed from crushed ice and slush and it is
not solid. This ice type exists between a solid ice field and open sea. It absorbs waves which
keep it fractured and wet, without any snow cover.
The microwave radiometer system operates at 24, 35, 48, and 94 GHz, vertical polarization,
incidence angle 45 degrees. Measurements were made under three different ice/snow conditions:
dry, wet, and refrozen. Figure 19.3 shows the observed signatures for three ice categories and
open water.
In the presence of dry snow cover on top of the ice layer, the brightness temperature decreases
with increasing frequency. This can be used as a rough indicator of the ice type. Open water
can be discriminated easily in all measurements.
The results suggest that a high-frequency receiver (90 GHz) should be combined with a lowerfrequency (35 GHz) receiver for ice type classification. The high frequency is sensitive to dry
and refrozen snow; hence, it is useful for discrimination of old first-year ice (compact pack ice
and fast ice) which is often covered by snow. The brightness temperature of water may, under
suitable wind conditions, be nearly the same as that of old snow-covered ice. Using the lower
frequency, water can be discriminated under all weather conditions. A limiting factor in the
Baltic Sea is the spatial resolution; information with resolution worse than 10 to 20 km is not
useful.
M. Hallikainen
the amount of ice experiencing summer melt is limited. The same polarization ratio/spectral
gradient ratio approach is used, but the coefficients in the two equations are slightly different
from those in Equations (19.10) and (19.11). In the absence of reliable in situ data, the two ice
types in the Southern hemisphere are called Type A and Type B ice (Parkinson and Gloersen,
1992). Figure 19.2 depicts the ice concentration nomograms for the Northern and Southern
hemisphere.
If hemispherically chosen brightness temperatures are used as tie points, the variations of TBw
(wind speed, foam, atmosphere), TBFY (surface roughness, temperature, snow cover, atmosphere), and TBMy (surface roughness, temperature, snow cover, atmosphere) over time and
space are ignored. To account for these factors, ice concentrations based on local tie points determined in the area of interest can be derived. Local-ocean tie point values are usually higher
than the hemispheric ones, due to e.g. wind speed. Differences as high as 18 K between the
hemispheric and local tie points for water have been observed (Steffen and Schweiger, 1991).
Other ice concentration algorithms include those by Swift et al. (1985), Svendsen et al. (1992),
Comiso (1986). The algorithm discussed above has been widely applied. Accuracy estimates
for the calculated sea ice concentrations are 5 to 9% for the Nimbus-7 SMMR data (Cavalieri
et al., 1984; Steffen and Maslanik, 1988).
The algorithm used for calculating sea ice concentrations from SSM/I data has the same formulation as the SMMR algorithm presented above; only the coefficients are different due to
slightly different channel frequencies.
A basic problem in developing ice type discrimination algorithms for the polar regions is poor
availability of in situ data. Validation of microwave observations by Landsat optical sensors
can be done only in clear weather conditions, even this may give a bias to the results.
In the Baltic Sea winter shipping traffic is handicapped by deformed ice (ice ridges, hummocks).
In the winter of 1992 classification of ice types and water with a microwave radiometer was
investigated. Especially, measurements were made on compact pack ice and close pack ice that
are difficult for ship navigation. Compact pack ice is formed from crushed fast or new ice,
which has been packed and refrozen. This ice type is common in the open sea where waves
and wind crush existing fast or new ice. The ice surface is rough and, hence, existing snow is
spatially unevenly distributed. Close pack ice is formed from crushed ice and slush and it is
not solid. This ice type exists between a solid ice field and open sea. It absorbs waves which
keep it fractured and wet, without any snow cover.
The microwave radiometer system operates at 24, 35, 48, and 94 GHz, vertical polarization,
incidence angle 45 degrees. Measurements were made under three different ice/snow conditions:
dry, wet, and refrozen. Figure 19.3 shows the observed signatures for three ice categories and
open water.
In the presence of dry snow cover on top of the ice layer, the brightness temperature decreases
with increasing frequency. This can be used as a rough indicator of the ice type. Open water
can be discriminated easily in all measurements.
The results suggest that a high-frequency receiver (90 GHz) should be combined with a lowerfrequency (35 GHz) receiver for ice type classification. The high frequency is sensitive to dry
and refrozen snow; hence, it is useful for discrimination of old first-year ice (compact pack ice
and fast ice) which is often covered by snow. The brightness temperature of water may, under
suitable wind conditions, be nearly the same as that of old snow-covered ice. Using the lower
frequency, water can be discriminated under all weather conditions. A limiting factor in the
Baltic Sea is the spatial resolution; information with resolution worse than 10 to 20 km is not
useful.