3 Role of SAR in Surface Energy Flux Measurements Over Sea Ice
57
spread of negative ,:1(j0 as microwave energy is absorbed in the deep moist snow cover
over multiyear ice. Virtually all multiyear ice ,:1(j0 is more negative than -1 dB, whereas
stable ice signatures appear to be primarily associated with rubble ice. These changes
are also apparent in the bivariate histogram for the two images. Below a (j0 magnitude
of -IS dB the increase in (j0 is inversely proportional to the (j0 magnitude of the May 2
image. Negative change in (j0 occurs at all (j0 above -13 dB, while the range of (j0 magnitudes which change less than 1 dB is extremely narrow.
The May S and June 9 contrast set (T3) shows an area of conglomerated ice types
which display all three change classes. This ice, visible in the southern regions of the
image, appears to be a combination of multiyear ice, first-year ice and rubble. The multiyear ice floes are clearly seen as blue (,:1(jo > -ldB). Rubble appears green (,:1(j0 between
-1 dB and +1 dB) and first-year ice is visible as veins of red running through the conglomerate. At this point in the time series, it is apparent that the use of ,:1(j0 could be
used as an effective image segmentation tool. The separation of ice types is preserved
throughout the melt in difference images, with the boundaries between positive and
negative ,:1(j0 change representing first-year ice and multiyear ice respectively.
The T 4 and TS change images are very similar to T3. This indicates that the rate of
,:1(j0 or first-year ice and multiyear ice types is stable. Rubble ice ,:1(j0 has deviated less
than 1 dB from its original winter level. The patterns are consistent with the seasonal
evolution illustrated in the conceptual schematic described previously (Fig. 3). The T6
difference image represents the threshold between the melt onset and advanced melt
periods when snow melt water begins to pond on the ice surface. T6 indicates the first
evidence of a universal increase in (j0 for both first-year ice and multiyear ice due to
melt water ponding of both first year ice and multiyear ice surfaces. The positive ,:1(j0
component displays a magnitude-independent increase from an average of -14 dB to
an average of -12 dB between Ts and T6. This indicates the first effects of surface melt
water on raising first-year ice (j0.
It is apparent from the preceding discussion that change detection images may provide a useful mechanism for segmentation of SAR sea ice scenes. In the winter season,
separation of smooth first-year from multiyear and rubble is possible because of
responses of the first-year classes to changing brine volumes in the basal snow layer
and ice surface (Barber and Thomas 1997). In the seasonal transition period there is a
considerable amount of information available on the state of first-year, multiyear and
rubble forms of sea ice. Multiyear ice is identifiable by the decrease in its (j0 relative to
typical winter values and smooth first year ice by its increasing (j0. Finally, certain ice
types, notably rough ice and rubble, display (j0 fluctuations of less than 1 dB between
winter and the onset of surface ponding.
3.4.3
Predicting Daily Average Shortwave Albedo (a) from 0°.
In the preceding sections we have shown that the time series evolution of the microwave
scattering coefficient (j0 is driven by the electrical and phase proportions of water within snow covered sea ice. We have also shown that the climatological albedo is one of the
primary determinants in the initiation of the seasonal evolution of the marine cryosphere. It should also now be apparent that climatological shortwave (K*) and
microwave scattering are affected by the same physical changes in snow-covered sea
57
spread of negative ,:1(j0 as microwave energy is absorbed in the deep moist snow cover
over multiyear ice. Virtually all multiyear ice ,:1(j0 is more negative than -1 dB, whereas
stable ice signatures appear to be primarily associated with rubble ice. These changes
are also apparent in the bivariate histogram for the two images. Below a (j0 magnitude
of -IS dB the increase in (j0 is inversely proportional to the (j0 magnitude of the May 2
image. Negative change in (j0 occurs at all (j0 above -13 dB, while the range of (j0 magnitudes which change less than 1 dB is extremely narrow.
The May S and June 9 contrast set (T3) shows an area of conglomerated ice types
which display all three change classes. This ice, visible in the southern regions of the
image, appears to be a combination of multiyear ice, first-year ice and rubble. The multiyear ice floes are clearly seen as blue (,:1(jo > -ldB). Rubble appears green (,:1(j0 between
-1 dB and +1 dB) and first-year ice is visible as veins of red running through the conglomerate. At this point in the time series, it is apparent that the use of ,:1(j0 could be
used as an effective image segmentation tool. The separation of ice types is preserved
throughout the melt in difference images, with the boundaries between positive and
negative ,:1(j0 change representing first-year ice and multiyear ice respectively.
The T 4 and TS change images are very similar to T3. This indicates that the rate of
,:1(j0 or first-year ice and multiyear ice types is stable. Rubble ice ,:1(j0 has deviated less
than 1 dB from its original winter level. The patterns are consistent with the seasonal
evolution illustrated in the conceptual schematic described previously (Fig. 3). The T6
difference image represents the threshold between the melt onset and advanced melt
periods when snow melt water begins to pond on the ice surface. T6 indicates the first
evidence of a universal increase in (j0 for both first-year ice and multiyear ice due to
melt water ponding of both first year ice and multiyear ice surfaces. The positive ,:1(j0
component displays a magnitude-independent increase from an average of -14 dB to
an average of -12 dB between Ts and T6. This indicates the first effects of surface melt
water on raising first-year ice (j0.
It is apparent from the preceding discussion that change detection images may provide a useful mechanism for segmentation of SAR sea ice scenes. In the winter season,
separation of smooth first-year from multiyear and rubble is possible because of
responses of the first-year classes to changing brine volumes in the basal snow layer
and ice surface (Barber and Thomas 1997). In the seasonal transition period there is a
considerable amount of information available on the state of first-year, multiyear and
rubble forms of sea ice. Multiyear ice is identifiable by the decrease in its (j0 relative to
typical winter values and smooth first year ice by its increasing (j0. Finally, certain ice
types, notably rough ice and rubble, display (j0 fluctuations of less than 1 dB between
winter and the onset of surface ponding.
3.4.3
Predicting Daily Average Shortwave Albedo (a) from 0°.
In the preceding sections we have shown that the time series evolution of the microwave
scattering coefficient (j0 is driven by the electrical and phase proportions of water within snow covered sea ice. We have also shown that the climatological albedo is one of the
primary determinants in the initiation of the seasonal evolution of the marine cryosphere. It should also now be apparent that climatological shortwave (K*) and
microwave scattering are affected by the same physical changes in snow-covered sea