A New Method to Filter Out Radio-Frequency Interference (RFI) from SMOS Level 1C Data for Sea Ice Applications - Towards an Interdisciplinary Approach in Earth System Science - page 86

Fig. 1 Statistical overview on SMOS L1C data higher than 50 N on 25 Dec 2010. The upper plot

(a) illustrates the number of all observations per incidence angle bin per grid cell per over

°

ight.
The color shows the number of occurrences. The red line marks the threshold of at least 3 data
point for a valid average brightness temperature. The lower plot (b) illustrates the standard
deviation criteria on all cases which ful lled the 3 data point criteria. The chosen threshold of 10 K
for the standard deviation is marked by the red line. Red hatched areas mark discarded data

fl

change of brightness temperature with the incidence angle leads to a smooth curve
of average brightness temperatures with low standard deviations. The threshold of
10 K is exceeded at no incidence angle bin. In contrast, on the right (b) the result is
much less homogeneous even though it is the same grid cell on the same day, but at
a later over

°
in TBv cause high standard deviations in many incidence angle bins exceeding the
10 K threshold. An averaging of brightness temperatures will not give represen-
tative values as the strong variations are not caused by natural surface emissions but
most probably by RFI.
On 25 Dec 2010 the Kara and Barents Seas are highly affected by RFI, which are

fl ight (12:32 UTC). The strong variations especially between 40 and 50

in uencing the brightness temperatures and therefore the retrieved thin SIT values

(Huntemann et al. 2014 ). Figure 3 demonstrates the RFI

fl

filter qualities of this