Mapping the Thickness of Pancake Ice Using Océan Wave Dispersion...
27
The quality of the inversion results was evaluated considering both the squared
error and the corrélation index pattern between the observed SAR spectrum and
the simulated SAR spectrum, as well as the cut-off wavelength measured on the
simulated spectra. The corrélation index and the square error are defined as
„
jp(ü)p(k)dk
J ,
J IP(P -P(k)}'-dk
C =
— and ez =
—,
V f P2(k)dk- iP\k)dk
\ \ P2(k)dk- ÏP^kïdk
respectively. Both parameters hâve been estimated in the wave number domain
defined by
2ti
_
2n
— <\k\<---- ,
700
100
where SAR data are expected to be more reliable.
Table 1 reports the inversion results from the 15 Windows and includes the
square error, the corrélation index, the wavelength and propagation angle of the
dominant wave computed with respect to the range direction and the ratio
between the real and simulated cut-off wavelength. Results show that the dominant wavelength is affected by the presence of sea ice so that its value is lower than
the one propagating in open sea, as expected. Figure 2 shows: (1) the retrieved
Table 1. Inversion results of the 15 Windows
Window
X
e2
C(%)
P
1
167.49
42.88
2.08
0.093
95.38
293/265
2
173.54
40.60
2.38
0.10
94.97
292/286
3
141.42
45.00
2.52
0.097
95.41
280/283
4
145.30
50.52
2.28
0.16
95.10
271/271
5
****
****
****
****
****
****
6
128.83
49.90
1.93
0.22
92.77
257/252
7
129.24
43.36
1.78
0.27
93.47
249/203
8
140.32
52.12
1.73
0.15
94.93
253/232
9
132.87
48.36
2.15
0.095
95.80
261/268
10
132.87
48.36
2.42
0.14
95.60
267/267
11
196.57
47.49
1.71
0.17
93.54
265/262
12
153.78
54.58
2.47
0.26
90.59
283/289
13
150.51
41.18
2.62
0.12
93.72
285/285
14
187.91
40.23
2.32
0.11
94.38
316/298
15
196.57
43.12
2.37
0.17
92.77
290/275
X, dominant wave wavelength (in métrés).
, propagation direction of the dominant wave relative to range direction.
Hs, estimated height (in métrés) of the dominant wave.
e2, square error (see text).
C(%), corrélation index (see text).
p, ratio of cut-off wavelength of SAR-simulated spectrum to cut-off wavelength of SAR-observed
spectrum.
****, quality of image did not permit parameters to be extracted.
27
The quality of the inversion results was evaluated considering both the squared
error and the corrélation index pattern between the observed SAR spectrum and
the simulated SAR spectrum, as well as the cut-off wavelength measured on the
simulated spectra. The corrélation index and the square error are defined as
„
jp(ü)p(k)dk
J ,
J IP(P -P(k)}'-dk
C =
— and ez =
—,
V f P2(k)dk- iP\k)dk
\ \ P2(k)dk- ÏP^kïdk
respectively. Both parameters hâve been estimated in the wave number domain
defined by
2ti
_
2n
— <\k\<---- ,
700
100
where SAR data are expected to be more reliable.
Table 1 reports the inversion results from the 15 Windows and includes the
square error, the corrélation index, the wavelength and propagation angle of the
dominant wave computed with respect to the range direction and the ratio
between the real and simulated cut-off wavelength. Results show that the dominant wavelength is affected by the presence of sea ice so that its value is lower than
the one propagating in open sea, as expected. Figure 2 shows: (1) the retrieved
Table 1. Inversion results of the 15 Windows
Window
X
e2
C(%)
P
1
167.49
42.88
2.08
0.093
95.38
293/265
2
173.54
40.60
2.38
0.10
94.97
292/286
3
141.42
45.00
2.52
0.097
95.41
280/283
4
145.30
50.52
2.28
0.16
95.10
271/271
5
****
****
****
****
****
****
6
128.83
49.90
1.93
0.22
92.77
257/252
7
129.24
43.36
1.78
0.27
93.47
249/203
8
140.32
52.12
1.73
0.15
94.93
253/232
9
132.87
48.36
2.15
0.095
95.80
261/268
10
132.87
48.36
2.42
0.14
95.60
267/267
11
196.57
47.49
1.71
0.17
93.54
265/262
12
153.78
54.58
2.47
0.26
90.59
283/289
13
150.51
41.18
2.62
0.12
93.72
285/285
14
187.91
40.23
2.32
0.11
94.38
316/298
15
196.57
43.12
2.37
0.17
92.77
290/275
X, dominant wave wavelength (in métrés).
Hs, estimated height (in métrés) of the dominant wave.
e2, square error (see text).
C(%), corrélation index (see text).
p, ratio of cut-off wavelength of SAR-simulated spectrum to cut-off wavelength of SAR-observed
spectrum.
****, quality of image did not permit parameters to be extracted.