3.0 Choice of Fundamental Radar Parameters for Detection and
Identification of Sea Ice
In this section, we discuss the influence of the choice of four
fundamental radar parameters (frequency, incidence angle, polarization and
resolution) on the effectiveness of the radar for sea-ice reconnaissance.
3.1 Frequency
The most extensive study of the frequency behaviour of radar cross
section for sea ice has been done by workers at the University of Kansas
(Onstott et al., 1979, 1981, 1982) who have facilities to do a calibrated scan
at discrete frequencies on the same ice. Other workers are often restricted to
a single frequency or uncalibrated imagery. For cold winter conditions, Onstott
has reported considerable variation in radar contrast with frequency between FY
and MY ice, with neglible contrast at L-band and contrasts of approximately
10 dB for frequencies of X-band or greater.
Until recently the band of frequencies between L and X-band has been
largely unexplored. Preliminary new results reported by Onstott et al. (1982b),
show that under the conditions which prevailed near Mould Bay, N.W.T., in
October, 1981, radar contrasts between MY and FY are ^5 dB at 4 GHz and
increase to ~10 dB at 16 GHz with the absolute value increasing steadily with
frequency having -15 dB variation over the frequency range. Figure 4 shows a
comparison of simultaneous SAR imagery taken later in the same season in the
Beaufort Sea. In each case, X-band data are compared to C-band and L-band in
steep geometry. These examples demonstrate that indeed the radiometric contrast
between ice classes is greatest at X-band and lowest at L-band with C-band being
in a intermediate position consistent with Onstott's results later in this
season. Although the conclusion is based on imagery which is seriously degraded
due to limitations in the C-band channel, it is believed to be a general result,
at least in the ice conditions studied.
3.2 Incidence angle
The incidence angle dependence of o © over the range from nadir to 60°
has been well studied using scatterometer techniques (Gray et al., Parashar,
1973, Onstott, 1979a,b). Figure 3 shows this behaviour over three seasons
for Beaufort Sea ice. From these curves there seems to be a slight advantage in
radar contrast between FY and MY at the larger incidence angles, but greater
benefit is derived when the ice-water contrast is considerel. A choice of
incidence angle greater than 35° would allow sufficient contrast at Ku band to
separate MY from open ocean even at windspeeds of 20 m/s* and most ice types
from open water at more moderate winlspeeds.
Much of the variability of radar cross section for the same ice class
can be attributed to associated changes of local scene roughness (for example
frost flowers on young ice). Scene roughness, however, couples together both
792
a = »
= ee er pa i ee - A
ne ee LU ER
|
st
cc
st
<A