8 Satellite Microwave Radar Observations of Antarctic Sea Ice
157
spray from nearby leads is thought to be a contributor to the mean background level
of surface snow salinity of 2 pSu.
Rough First-Year Ice Floes
Well-deformed, ridged, or rubbled first-year ice floes are difficult to measure using a
shipborne scatterometer, because a complete incidence angle scan is not possible from
a fixed viewing position when the local topography is punctuated by piles of ice blocks.
Shipborne scatterometer measurements were limited to five sites where surface roughness was relatively evenly distributed, and where the signature was not significantly
biased by one or more features within the scan. The result is the mean signature in Fig. 5C.
Values of croShip are lower at near normal incidence, and fall between -11.7 and -15. dB for
the SAR incidence angle band. The gradient in crOShip is approximately -0.52 dB/degree
in the 20-60· range, but the small sample size and averaging of individual ridges or
local roughness elements cause the variability in measurement points around the logarithmic fit. Cross-polarized backscatter is greater and more variable for these rough
ice floes, especially in the 20-40· range, and is explained by second-order scattering
effects such as multiple scattering from blocks.
Multiyear Ice Floes
According to the field scatterometer data in Fig. 5d, multiyear ice floes (largely secondyear ice) are practically indistinguishable from rough first-year ice forms in the ERS incidence angle range. Undeformed, old floes were distinguished on the basis of snow depth,
salinity, and thickness as second-year ice. Mean net annual thermodynamic ice growth is
estimated to be around 1.5 m in the absence of dynamic thickening, based on the lifecycle,mean snow depth, and annual net freezing rate in the Weddell Sea (Fischer 1995). However, perennial ice floes observed during WWGS'92 comprised massive undeformed ice
with considerable snow accumulation. Snow depths exceeding 0.75 m were commonly
measured (Massom et al.1997) while ice thickness ranged from around 2 m upwards. Shipborne scatterometer crOShip values from these floes fall between -12 and -15.5 dB, and the
low gradient of -0.36 dB/degree observed in the 30-50· incidence angle range is caused
by snow volume scattering from layers and distinctive ice lenses and ice pipes (Massom
et al.1997). To account for the volume scattering characteristics exhibited in Fig. 5C, a modified third order polynomial fit is applied instead of a single parameter exponential, to
accommodate the flattening at 25· and rolloffbeyond 55· . The distinctive plateau and higher backscatter value in the range 40-50· may be used in 40· -incidence EScat images as a
method for discriminating high concentrations of thick, snow-covered perennial ice.
8.4.1.2
Regional Winter (-band Backscatter Variability
The map of the Weddell Sea region in Fig. 2 indicates (in bold) the regional locations
of several winter SAR image frames listed in Table 1. Corresponding image pdf's of calibrated backscatter coefficients are shown in Fig. 4. Each pdf comprises individual pixel values from 8 x 8 averaged and calibrated full-resolution SAR.PRI images (originally 8000 x 8000 pixels), resulting in 10 6 pixel samples per 100 x 100 km frame.
A global pdf histogram, indicated as the thick black stepped line in Fig. 4 (shown as
a bar-histogram with 0.1 dB bin width), shows the combined pdf from all images indi-
157
spray from nearby leads is thought to be a contributor to the mean background level
of surface snow salinity of 2 pSu.
Rough First-Year Ice Floes
Well-deformed, ridged, or rubbled first-year ice floes are difficult to measure using a
shipborne scatterometer, because a complete incidence angle scan is not possible from
a fixed viewing position when the local topography is punctuated by piles of ice blocks.
Shipborne scatterometer measurements were limited to five sites where surface roughness was relatively evenly distributed, and where the signature was not significantly
biased by one or more features within the scan. The result is the mean signature in Fig. 5C.
Values of croShip are lower at near normal incidence, and fall between -11.7 and -15. dB for
the SAR incidence angle band. The gradient in crOShip is approximately -0.52 dB/degree
in the 20-60· range, but the small sample size and averaging of individual ridges or
local roughness elements cause the variability in measurement points around the logarithmic fit. Cross-polarized backscatter is greater and more variable for these rough
ice floes, especially in the 20-40· range, and is explained by second-order scattering
effects such as multiple scattering from blocks.
Multiyear Ice Floes
According to the field scatterometer data in Fig. 5d, multiyear ice floes (largely secondyear ice) are practically indistinguishable from rough first-year ice forms in the ERS incidence angle range. Undeformed, old floes were distinguished on the basis of snow depth,
salinity, and thickness as second-year ice. Mean net annual thermodynamic ice growth is
estimated to be around 1.5 m in the absence of dynamic thickening, based on the lifecycle,mean snow depth, and annual net freezing rate in the Weddell Sea (Fischer 1995). However, perennial ice floes observed during WWGS'92 comprised massive undeformed ice
with considerable snow accumulation. Snow depths exceeding 0.75 m were commonly
measured (Massom et al.1997) while ice thickness ranged from around 2 m upwards. Shipborne scatterometer crOShip values from these floes fall between -12 and -15.5 dB, and the
low gradient of -0.36 dB/degree observed in the 30-50· incidence angle range is caused
by snow volume scattering from layers and distinctive ice lenses and ice pipes (Massom
et al.1997). To account for the volume scattering characteristics exhibited in Fig. 5C, a modified third order polynomial fit is applied instead of a single parameter exponential, to
accommodate the flattening at 25· and rolloffbeyond 55· . The distinctive plateau and higher backscatter value in the range 40-50· may be used in 40· -incidence EScat images as a
method for discriminating high concentrations of thick, snow-covered perennial ice.
8.4.1.2
Regional Winter (-band Backscatter Variability
The map of the Weddell Sea region in Fig. 2 indicates (in bold) the regional locations
of several winter SAR image frames listed in Table 1. Corresponding image pdf's of calibrated backscatter coefficients are shown in Fig. 4. Each pdf comprises individual pixel values from 8 x 8 averaged and calibrated full-resolution SAR.PRI images (originally 8000 x 8000 pixels), resulting in 10 6 pixel samples per 100 x 100 km frame.
A global pdf histogram, indicated as the thick black stepped line in Fig. 4 (shown as
a bar-histogram with 0.1 dB bin width), shows the combined pdf from all images indi-