ymbol 
MUN 
  
SNfIOA 
w 
= 
L——— — 
CIO 
°k to 
ı rough 
sed 
‘ay 
ar Cross 
response 
ote 
ze has 
L et al. 
yhted the 
a study 
ne 
ions, but 
vave 
arized 
ater; and 
ay et al. 
  
  
During the cold conditions which characterize the winter months in the 
Beaufort Sea, a well defined set of microwave properties may be indicated by the 
active/passive signature at a single look angle and frequency for old and FY 
ice. The 20 m resolution profiling sensor set depicted in Figure 1 supports 
this statement and demonstrates the improvement a combined measurement may make 
over a single passive or active measurement alone. Although there is a 
separation in these diagrams of FY and young ice from SY and older ice, some 
overlap exists between FY and young ice rough and smooth (deformed and ridged as 
opposed to flat smooth ice) which complicates a classification exercise. 
Nevertheless, classifier tests based solely on radiometric levels show that a 
two-feature classifier can yield major class accuracies of 93% and subclass 
accuracies of 72% (Gluch, 1981). Guindon et al. (1982) have shown that the 
addition of textural features can enhance classification accuracy and that level 
contrast was particularly useful with texture feature X-band SAR imagery. 
The general success of microwave techniques for ice reconnaissance 
which the above results and those of other workers demonstrate in cold winter 
conditions is not shared in results from the same region taken at the height of 
melt season in June and July, 1980. Figure 2 shows a section of the photo 
mosaic from this period, crossing (from left to right) a MY floe with large 
aggregate melt ponds and followed by FY ice with elongated lacey drainage 
patterns. Neither the scatterometer, nor the radiometer trace below the 
photograph, register a significant change as the transition between these major 
ice classes is made. This is a general comment for the ice we studied in this 
season*. There are however, some aspects of the summer season data set which 
may be regarded as positive. The microwave radiometer, for instance, is a 
sensitive indicator of free water whether ponded on the ice surface or between 
floes and this feature might be used to decide on ice classes from a high 
resolution (aircraft) passive microwave image because of the distinctive melt 
pond patterns present in FY and MY ice. A direct relationship appears to exists 
between measured brightness temperature and the percentage of ponded water in 
the radiometer footprint, although the accuracy with which concentration can be 
measured under heavy melt conditions is unknown. The Convair-580 has returned 
to the Beaufort Sea to gather data on two missions during freeze-up conditions: 
Fall, 1980 and 1981. Although new ice had not grown to FY thickness on either 
mission; grey-white ice and younger forms were present and appeared to yield 
microwave signatures close to the midwinter values. The total seasonal 
variation of the Ku-band scatterometer data is shown in Figure 3 with a 
reference curve for a low wind speed, open water situation**. 
  
* Onstott, stulying sea-ice backscatter later in the season found that FY was | 
brighter than MY by as much as 8 dB in his X-band, VV polarization results and M 
that the MY ice had retained characteristics found in other seasons. We can IM 
find small regions of our SAR X-VV imagery which qualitatively exhibit the | 
behaviour found by Onstott, but these are the exception rather than the rule in 
all our results. It is also worth noting that examples may be found where FY 
which has become smooth from melting can appear much darker than MY in the same 
imagery, and we therefore attribute these effects to specific surface anomalies. 
The general conclusions given here for wet surface conditions have later been 
verified by Onstott (Onstott et al., 1982a) and it appears that ice signatures 
may change through the summer melt period as brine drainage and 
recrystallisation take place. 
** The wind alters the radar cross section of the sea considerably and has been 
well characterized at Ku-band by Jones et al., 1978. The radar cross section of 
ice is unchanged unless the wind results in ice breakup or flooding. 
  
791 
A 
E i 
EA RR a E ORDRE Dio D Mu MEM EM ERE REESE wey Ms