R.KwOK
niques under development for measurement of the ice thickness are reviewed by
Thorndike et al. (1992) and Wadhams and Comiso (1992). These instruments typically
provide a one dimensional transect of ice draft (or, equivalently, thickness). These data
are suitable for computing the volume transport of ice through a region or could be used
for building up a climatology of mean ice iliickness over a long period of time. However, iliey do not provide a spatial picture of ilie thickness distribution.
The use of ice type as inferred from active (Kwok et al. 1992) or passive microwave
(Cavalieri et al.1984) data sets as a proxy thickness indicator of ice thickness, in substitution for ilie direct resolution of the thicker ice classes, is at best a poor replacement of
direct measurement. Surface processes typically dominate the backscatter and emissivity observed by spaceborne sensors. At higher probing frequencies, atmospheric water
vapor and cloud are additional confounding factors in the identification of ice types.
Kwok et al. (1995) demonstrated that ilie age distribution of sea ice is a fundamental
quantity that can be measured using ice motion information by keeping track of the area
changes of deforming Lagrangian cells. The age distribution of sea ice specifies the fractional area covered by ice in different age classes as a function of time. In ilie Arctic Ocean
in winter, new ice forms from freezing sea water that is exposed by the opening of leads
in the ice cover. These horizontal openings manifest iliemselves as additions of areas to
the local ice cover and are directly observable in time sequences of SAR imagery. The
new ice in these leads ages and iliickens. Repeated temporal sampling of iliese elemental areas provides us wiili an indication of when new areas or leads were created, the
length of their existence, and thus a record of ilieir age. The resolution of age is dependent upon ilie sampling interval provided by SAR observations. With some knowledge
of the heat exchange between the atmosphere and ocean, the observed ice age distribution can be used to estimate the thickness distribution of the sea ice in iliese cells.
11.1.3
Melt Onset/Freeze-Up
The location and timing of melt onset and freeze-up on Arctic sea ice play significant
roles in a number of geophysical processes. A large decrease in surface albedo accompanies the onset of snow melt during the spring and thus increases the absorption and
heating of the ice by the shortwave radiation. The rapid temperature decrease during
freeze-up in the fall has the opposite effect on albedo and marks the beginning of ilie
ice growth season and the end of the summer. The length of the melt season is defined
by the two transitions described above. Changes in the timing and spatial patterns of
these transitions and the length of ilie melt season are important parameters in polar
climate and may also serve as sensitive indicators of climate change in the high latitudes.
Winebrenner et al. (1994 and 1996) have shown that the onset of melt and freeze-up
events are clearly detectaWe as changes in radar backscatter in ERS-1 imagery. During
melt onset, the appearance of liquid water in the snow cover on multiyear ice is marked
by a steep decrease (almost 9 dB) in the observed backscatter. At the end of the summer, the C-band backscatter from sea ice which survived the summer season increases rapidly over several weeks as temperatures fall below freezing and attains a stable
backscatter signature which is characteristic of multiyear ice in the winter (Kwok and
Cunningham 1994).
niques under development for measurement of the ice thickness are reviewed by
Thorndike et al. (1992) and Wadhams and Comiso (1992). These instruments typically
provide a one dimensional transect of ice draft (or, equivalently, thickness). These data
are suitable for computing the volume transport of ice through a region or could be used
for building up a climatology of mean ice iliickness over a long period of time. However, iliey do not provide a spatial picture of ilie thickness distribution.
The use of ice type as inferred from active (Kwok et al. 1992) or passive microwave
(Cavalieri et al.1984) data sets as a proxy thickness indicator of ice thickness, in substitution for ilie direct resolution of the thicker ice classes, is at best a poor replacement of
direct measurement. Surface processes typically dominate the backscatter and emissivity observed by spaceborne sensors. At higher probing frequencies, atmospheric water
vapor and cloud are additional confounding factors in the identification of ice types.
Kwok et al. (1995) demonstrated that ilie age distribution of sea ice is a fundamental
quantity that can be measured using ice motion information by keeping track of the area
changes of deforming Lagrangian cells. The age distribution of sea ice specifies the fractional area covered by ice in different age classes as a function of time. In ilie Arctic Ocean
in winter, new ice forms from freezing sea water that is exposed by the opening of leads
in the ice cover. These horizontal openings manifest iliemselves as additions of areas to
the local ice cover and are directly observable in time sequences of SAR imagery. The
new ice in these leads ages and iliickens. Repeated temporal sampling of iliese elemental areas provides us wiili an indication of when new areas or leads were created, the
length of their existence, and thus a record of ilieir age. The resolution of age is dependent upon ilie sampling interval provided by SAR observations. With some knowledge
of the heat exchange between the atmosphere and ocean, the observed ice age distribution can be used to estimate the thickness distribution of the sea ice in iliese cells.
11.1.3
Melt Onset/Freeze-Up
The location and timing of melt onset and freeze-up on Arctic sea ice play significant
roles in a number of geophysical processes. A large decrease in surface albedo accompanies the onset of snow melt during the spring and thus increases the absorption and
heating of the ice by the shortwave radiation. The rapid temperature decrease during
freeze-up in the fall has the opposite effect on albedo and marks the beginning of ilie
ice growth season and the end of the summer. The length of the melt season is defined
by the two transitions described above. Changes in the timing and spatial patterns of
these transitions and the length of ilie melt season are important parameters in polar
climate and may also serve as sensitive indicators of climate change in the high latitudes.
Winebrenner et al. (1994 and 1996) have shown that the onset of melt and freeze-up
events are clearly detectaWe as changes in radar backscatter in ERS-1 imagery. During
melt onset, the appearance of liquid water in the snow cover on multiyear ice is marked
by a steep decrease (almost 9 dB) in the observed backscatter. At the end of the summer, the C-band backscatter from sea ice which survived the summer season increases rapidly over several weeks as temperatures fall below freezing and attains a stable
backscatter signature which is characteristic of multiyear ice in the winter (Kwok and
Cunningham 1994).