Arctic Multiyear Ice Concentration Retrieval from SSM/I Data Using the NASA Team Algorithm with Dynamic Tie Points - Towards an Interdisciplinary Approach in Earth System Science - page 96

In the NASA Team algorithm, polarization ratio (PR) and gradient ratio (GR) are
calculated as below:

T B ; 19V T B ; 19H
T B ; 19V þ T B ; 19H ;

PR ¼ ð 1 Þ

T B ; 37V T B ; 19V
T B ; 37V þ T B ; 19V ;

GR ¼ ð 2 Þ

where T B ; 19V , T B ; 19H and T B ; 37V are the brightness temperature at 19 GHz V,
19 GHz H and 37 GHz V channel. Multiyear ice concentration retrieval is mainly
determined by GR in the GR/PR space because of its high sensitivity for multiyear
ice concentration retrieval. Based on the average brightness temperature differences
between dynamic and

fixed tie points in Table 2 , GR of averaged dynamic tie points

for first year ice and multiyear ice are:

T BFY ; 37V T BFY ; 19V þ 0 : 74 þ 0 : 86

GR FY ¼

86 ;

ð 3 Þ

T BFY ; 37V þ T BFY ; 19V þ 0 : 74 0 :

T BMY ; 37V T BMY ; 19V 0 : 36 þ 2 : 5

GR MY ¼

5 ;

ð 4 Þ

T BMY ; 37V þ T BMY ; 19V 0 : 36 2 :

where T BMY ; 37V and T BMY ; 19V are the brightness temperature of fixed tie points for

first year ice at 37 and 19 GHz V, T BFY are those for multiyear ice. They are larger
than GR of the fixed tie points. In addition, the GR differences between that of xed

and dynamic tie points for multiyear ice are bigger than those for first year ice,

which makes GR of second year ice closer to that of multiyear ice. As a conse-
quence, the method using dynamic tie points yields higher estimates than that using

fixed tie points in most winters, which is shown in Fig. 3 .
There is a slightly declining trend of minimum ice extent as shown in Fig. 1 ,
which is consistent with the multiyear ice extent retrieved from both methods. In
most winters, multiyear ice extent decreases as expected. One reason for the
multiyear ice reduction is the export to more southern regions. Figure 4 shows the
reduction of multiyear ice area and the ice out ow through Fram Strait (Kwok

2009 ) from 1988 to 2007. It is noted that the multiyear ice area takes the pixels with
multiyear ice concentration below 30 % into account. In all, the multiyear ice
reduction is larger than however not highly correlated to the ice export through
Fram, with a correlation coef

fl

cient of 0.15 and 0.22 for the retrieval from original
and dynamic NASA Team algorithm respectively. The difference between Fram
strait export and multiyear ice area could be caused discrepancy in considered
periods and the export through regions like Canadian archipelago.
In Fig. 3 , multiyear ice extent retrieved from both methods increases during
some winters, such as the winter of 1999/2000. It could be caused by the ice drift
from central Arctic, which is not covered by SSM/I data. Besides, the multiyear ice
extent changes rapidly from day to day, which could be caused by the sensor noise.