Geoscience Reference
In-Depth Information
first case high thresholds for the peakiness are necessary to exclude off-nadir leads.
In the second case off-nadir leads cause decreased ice elevations which is shown in
Fig.
4
a for example between 200 and 400 km. In the FYI zone, we do not observe a
similar effect. Figure
2
shows biased waveforms of FYI and MYI. Both are a
composition of an off-nadir lead re
ections from sea ice. The biased
MYI waveform shows a high left-peakiness of 38.7 while the right-peakiness is 5.2
which is close to the value for mean MYI. Here the off-nadir lead seems to dom-
inate the peak power. Thus the waveform is dominated by the off-nadir lead
re
fl
ection and re
fl
ection and the range is tracked at the leading edge of the lead waveform con-
tribution, resulting in a range bias
fl
d (Fig.
1
b). Considering PPl
l
and PP
r
allows us
to characterize waveforms and to identify biased waveforms.
FYI waveforms can exhibit similar shapes and properties as biased MYI. As a
consequence, FYI waveforms might be discarded if they are classi
D
ed as MYI in
the OSI SAF ice type.
We also note that for FYI we
find fewer outliers than for MYI (Fig.
4
a). We can
speculate that the backscatter from FYI is usually higher than from MYI (Fig.
2
a).
An off-nadir lead re
ection is then in certain cases still distinguishable from the sea-
ice echo as shown in Fig.
2
(green line). We can identify two peaks where the
fl
rst
represents the sea-ice re
ection and the second the off-nadir lead that is well sep-
arated from the ice waveform. Therefore the retracker algorithm captures the
leading edge of the sea-ice echo correctly and hence a range bias does not occur.
Therefore we can use higher thresholds of PPl
l
and PP
r
for FYI than for MYI to
avoid discarding FYI waveforms erroneously. However, another reason for fewer
outliers in the FYI zone could be a different pattern and distribution of leads in the
considered FYI area.
The uncertainties of the range retrieval are discussed in more detail in Ricker
et al. (
2014
). Besides a bias due to the choice of the retracker, the uncertainty is
dominated by the speckle noise (Laxon et al.
2013
; Wingham et al.
2006
) that is
around 0.1 m for a single measurement. Using the OSI SAF ice-type product for the
waveform classi
fl
cation uncertainty (Eastwood
2012
)
and might lead to an underrepresentation of FYI within the MYI zone. Therefore a
combined ice-type classi
cation also induces a classi
cation using the CryoSat-2 waveforms as well as the OSI
SAF ice-type product might be a reasonable approach for the future. The thresholds
in Table
1
for FYI and MYI are empirical and where chosen considering the
distribution of PPl
l
and PP
r
in Fig.
4
b. Valid outliers in Fig.
4
a could not be
identi
ed as biased waveforms and still affect the freeboard retrieval.
5 Conclusion
In this study we present a method to classify CryoSat-2 waveforms using a com-
bination of parameters that characterize the radar echo. We use a left- and right-
peakiness to characterize surface types and to identify waveforms that are biased by
off-nadir-leads. Those waveforms can cause a decrease in surface elevation,
Search WWH ::
Custom Search