3.4 Resolution and Speckle 
  
Operationally, the nominal radar resolution or pixel size does not 
guarantee identification of an ice floe of similar dimension. Normally it is 
necessary to have spatial information (shape of floes, feature identification 
etc.) in order to classify and interpret sea-ice imagery. However, in 
situations in which there is significant radar contrast (e.g. iceberg in open 
water at incidence angles 2409. small multi-year floes in first-year ice) it is 
possible to predict detectability even when the feature occupies a small number 
of pixels in the final image. Gray et al. (1982) have presented some 
preliminary results for a moderate resolution ( ^20 m) radar by fitting to a 
gaussian distribution scatterometer data of the spatial variability of o © for 
particular ice classes using a Rayleigh fading model to describe the imaging 
radar speckle contribution (Ulaby et al. 1981). On this basis, it is possible 
to estimate classification errors for target detection as a function of average 
contrast and number of looks. Figure 7 illustrates the percentage of target 
(class B) misclassifications if A, representing a first-year ice background, has 
a gaussian backscatter variability with standard deviation of 2.5 dB and B, the 
multi-year ice, has a standard deviation of 1.0 dB and a variable average 
brightness or contrast above the background. 
  
It is clear from Figure 7 that single-look performance is poor for 
contrasts up to 14 dB but it is good for 6 looks or more and contrasts of 9 dB 
or more. Normally Ku-band, multi-year and first-year ice contrasts are in the 
7-10 dB range at Ku-band but appear to be approximately 4 dB lower at C band 
(Onstott et al. 1982). Thermal noise has not been included in this estimate so 
that it appears that detection of small multi-year floes (a few pixels) in a 
background of first-year ice will be difficult at C band. 
  
Multi-looking can be achieved in SAR imagery by degrading the 
resolution or, if a particular resolution is required, by increasing pulse 
bandwidth and decreasing antenna length. To .exercise these options would 
require an increased transmitter power and a reduced swath width so that there 
are inevitable trade-offs in spaceborne SAR design. 
Conclusion 
Both active and passive microwave remote sensing of sea ice and oceans 
have a demonstrated promise for future satellite ani airborne systems. In the 
paper, we have shown that under cold conditions good discrimination between ice 
classes can be achieved when suitable choices of fundamental system parameters 
are made. Optimizing these parameters for ice work to meet operational needs is 
the object of current research. 
REFERENCES 
Dunbar, M. (1969) "A glossary of ice terms (WHO Terminology)." Ice Seminar, 
Special Volume 10, The Canadian Institute of Mining and Metallurgy, pp. 
105-110. 
Gloerson, P.W. Nordberg, T.J. Schmugge, T.T. Wilheit (1973) 
"Microwave signatures of first-year and multi-year sea ice." 
J. Geophysical Research 78, pp. 3564-3572. 
  
, D. Cavalievi, W.J. Campbell (1982) "Derivation of sea-ice 
concentration, age and surface temperature from multispectral microwave 
radiances obtained with Nimbus VII scanning multichannel microwave 
radiometer.” Oceanography From Space, J.F.R. Gower el., Plenum Press, 
pp 823-830, N.Y. 
  
794 
E: 
N E po 
mu d CC SASL 7/07 
E ns oem " A 
i m 
—Ó— te 
EN. 
m 
<À