Classification of CryoSat-2 Radar Echoes - Towards an Interdisciplinary Approach in Earth System Science - page 139

satellite altimeter mission of the European Space Agency (ESA) and was launched
in April 2010, with special emphasis on Arctic sea ice. It is equipped with a Ku-
Band SAR radar altimeter synthetic aperture interferometric radar altimeter (SI-
RAL) that uses along-track beam sharpening (Wingham et al. 2006 ) to reduce
footprint size compared to previous radar altimeter missions (ERS1/2, Envisat). By
using the effect of the Doppler shift the radar footprint can be divided into stripes
called Doppler cells (for CryoSat-2 approximately 250 m). Each cell is illuminated
from different incident angles as the satellite passes by (Fig. 1 a). The echoes of each
illumination are stacked to reduce noise. This method results in a higher resolution
than pulse-limited radar altimeters like onboard ERS1/2 and Envisat.
Since the uncertainties of freeboard can easily reach the magnitude of freeboard
itself, optimized algorithms that reduce errors and uncertainties in CryoSat-2
freeboard retrieval are necessary. The

first step in obtaining sea-ice freeboard is to
determine the main scattering horizon to receive geolocated surface elevations
(Kurtz et al. 2014 ; Ricker et al. 2014 ). In this study a threshold first-maximum
retracker with a 40 % threshold (TFMRA40) (Helm et al. 2014 ; Ricker et al. 2014 )
is applied to the geolocated radar echoes (waveforms) that are provided by the
European Space Agency. Within this retracker algorithm the waveform is over-
sampled and smoothed. We compute the derivative to first maximum of the

waveform and assign the main scattering horizon at 40 % of this

nd the

rst peak. The
effects of different thresholds and retrackers on the freeboard retrieval can be
substantial and have been investigated in Ricker et al. ( 2014 ) and Kurtz et al.
( 2014 ). In the second step the geolocated CryoSat-2 elevations have to be refer-
enced to the sea level to obtain the freeboard. We apply a waveform classi cation

algorithm (Ricker et al. 2014 ) in order to detect leads which are narrow open water
areas in the ice surface. At leads the sea level can directly be obtained by the
CryoSat-2 range measurement. The lead elevations are interpolated along the
CryoSat-2 ground tracks to receive the actual sea-surface height which is then
subtracted from the sea-ice elevations to get the sea-ice freeboard.
Armitage and Davidson ( 2014 ) have shown that off-nadir re ections from leads

can bias the range retrieval since elevation retrievals are based on the assumption
that the main re

fl

fl ector is in the nadir of the satellite. They typically occur when

specular re ection on the edge of the main radar lobe still dominate the return signal

(Figs. 1 b and 2 ). These biased waveforms are mostly a composition of re

fl

fl ections of

leads and sea ice. They can potentially affect elevations of leads if classi ed as leads

as well as ice elevations if classi d

(Fig. 1 b). In this study we present our method to discriminate waveforms that are
biased by off-nadir re

ed as sea ice and cause a range bias of

D

ections from leads and valid sea-surface height information.
In addition the waveform classi

fl

cation scheme is extended to also discriminate

different ice types.