2 Identifying Ice Floes and Computing Ice Floe Distributions in SAR Images
27
ly first-year ice and is placed by segmentation into the water/ice mixture category.
While this may seem confusing or incorrect, it should be borne in mind that this ice is
most likely to be thinner than the floes and melt away during the summer months, leaving the thicker, older, more rounded floes intact at the end of the summer. Thus, for
comparison purposes with end of summer data, it seems satisfactory that this brighter,
nonrounded ice falls into the water/ice mixture rather than floe category. Note that both
dates are shortly after melt has begun in the respective latitudes, which was measured
by Winebrenner et al. (1994) to have occurred between June 13 and June 20 in 1992.
Figure 13 shows two four-image segments in the high Arctic which cover nearly the
same area, one (Fig. 13 a,b) from June 30 (which includes the same data as Fig. 12) and
Fig. 13. a Sequence
JUNE30
SEGMENTATION
of four ERS-1 SAR
a
b
images from June
30,1992, centered
84
at 82°N, 140 0 W
(copyright ESA).
- 128
- 128
b Output display of
floes, discarded
floes, open water
- 132
• R.OES
-132
and substrate for a.
•
DISCARDED
C Sequence of four
ERS-1 SAR images
83
-'38
•
SUBSTRATE
83
-'36
from July 4, 1992,
centered at 82°N,
]
OPEN WATER
138°N (copyright
ESA). d Output
-101()
- 101()
display of floes,
discarded floes,
open water, and
82
substrate for c
- 1'"
- 1'"
8 1
8'
JULY4
SEGMENTATION
b
- '28
- 128
B3
83
• FLOES
-132
- 132
•
DISCARDED
•
SU8STIlATE
-136
]
OPEN WATER
-136
82
- 101()
-101()
81
81
eo
eo