Classification of CryoSat-2 Radar Echoes - Towards an Interdisciplinary Approach in Earth System Science - page 141

Table 1 Waveform parameter and ice concentration thresholds used in the CryoSat-2 processing
to identify the surface types Lead and multi-year (MYI) and first-year ice (FYI)
Surface type PP K SSD PP l PP r IC (%)

Lead ≥ 40 ≥ 40 ≤ 4 ≥ 40 ≥ 30 ≥ 70

Sea ice (FYI) ≤ 60 ≤ 25 ≥ 70

70
Pulse peakiness PP, stack kurtosis K, stack standard deviation SSD, peakiness PP l left of the power
maximum, peakiness PP r right of the power maximum and sea-ice concentration IC

Sea ice (MYI) ≤ 18 ≤ 15 ≥

max(WF)

PP l ¼

WF i max þ 3 Þ

3 ð 2 Þ

mean ð ½ WF i max þ 1 ;

where WF i is the echo power at range bin i and max(WF) the peak power of the
waveform. PPl l and PP r are a measure for the peakiness left and right of the power
maximum as we consider the ratio of the maximum power to the mean power of
only three range bins left and right of the maximum. In the case of a nadir lead the
waveform power distribution is narrow and shows a high maximum echo power as
well as high PPl r and PP l values (Fig. 2 a, b).
For the lead identi cation we further use the parameter stack kurtosis (K), also a

measure of peakiness (Wingham et al. 2006 ), and the stack standard deviation
(SSD), which is a measure of the variation in surface backscatter depending on the
incident angle (Wingham et al. 2006 ). The term stack refers to the multi-look SAR
processing (Wingham et al. 2006 ). Leads are associated with a high K and a low
SSD because of their specular re ection. Table 1 shows a set of waveform

parameters used for the discrimination between sea ice and leads. The threshold
values were determined by test-processing of CryoSat-2 ground tracks. All wave-
forms that do not comply with these constraints are discarded.
After the identi

fl

cation of leads, the actual sea level can be interpolated and
subtracted from the CryoSat-2 elevations that were identi ed as sea ice. As a result

we receive the radar freeboard according to (Ricker et al. 2014 ).

3 Results

Here we show exemplary results from a CryoSat-2 ground track from December 15.
The track is directed south-east and first passes the MYI region north of Greenland
before it passes over FYI in the marginal ice zone in the Fram Strait (Fig. 3 ). For the
ice-type discrimination we use the OSI SAF ice-type product (Eastwood 2012 ).
Figure 4 a reveals the range retrieval after subtracting the mean sea-surface
height. Applying the waveform discrimination according to Table 1 we nd 50

leads over a distance of 2,300 km. Radar echoes with waveform parameters that do
not comply with the thresholds in Table 1 were discarded. Overall 22 % of the FYI
and 21 % of the MYI waveforms are discarded. The fraction of detected leads is
0.7 % for FYI and 0.5 % for MYI. The difference of waveform characteristics