Geology Reference
In-Depth Information
140
120
100
80
60
40
20
0
-20
-40
-60
-80
-100
-120
Difference bet. NT2 and CIS estimate of thin ice concentration
Difference bet. AES-York and CIS estimate of thin ice concentration
-140
-110 -100 -90
-80
-70
-60 -50 -40 -30
Thick minus thin ice concentration from CIS estimates (%)
-20
-100 10
20
30
40
50
60
70
80
90
100 10
140
120
100
80
60
40
20
0
-20
-40
-60
-80
-100
Difference bet. NT2 and CIS estimate of thick ice concentration
Difference bet. AES-York and CIS estimate of thick ice concentration
-120
-140
-110 -100 -90
-80
-70
-60 -50 -40 -30
Thick minus thin ice concentration from CIS estimates (%)
-20
-100 10
20
30
40
50
60
70
80
90
100 10
Figure 10.23
Deviation of thin (top) and thick (bottom) FY ice from CIS estimates plotted against the difference
between CIS estimates of thick minus thin ice concentrations. Data points are from the two algorithms: NT2 and
AES‐York. The points from each algorithm are shifted by ±1% relative to each other to improve the visibility of
the data.
from different ice concentration algorithms provide
information on the relative suitability and accuracy of
the algorithms.
Anderson
[2000] compared results from
eight ice concentrations algorithms using a radiative
transfer model. They found that the NT algorithm is
least sensitive to geophysical noise over consolidated ice,
whereas the frequency mode of the BS algorithm was
the most stable over open water.
Andersen et al
. [2007]
compared ice concentration from seven algorithms: BRI,
BS (from frequency and polarization modes), N90 GHz,
NT, NT2, and TUD. They used data from the Arctic per-
ennial ice cover during the cold season from 31 October
to 31 December 2003. Their validation data included
ship observations, the Radarsat Geophysical Processor
System (RGPS), and classified SAR imagery statistics.
They showed that ice concentration from the seven
