The Envisat RA2 is the first spaceborne altimeter designed to collect data over all the earths surfa

1
Analysis of ENVISAT RA-2 Backscatter over Natural
Land Calibration Targets Bramer, S.M.S., Johnson,
C.P.D. Berry, P.A.M. Earth and Planetary
Sciences Laboratory, De Montfort University, The
Gateway, Leicester. LE1 9BH, UK smsb_at_dmu.ac.uk
http//www.cse.dmu.ac.uk/eaprs/
Natural land targets for ERS RA-1 backscatter,
including the Simpson desert, the Takla Makan,
parts of the Sahara and the Arabian desert, have
previously been identified and modelled using
ERS-1 Geodetic Mission and Tandem Mission data.
These have been used to perform cross-calibration
of ERS-1/2 ice and ocean mode altimeter signa0.
The advent of the Envisat RA-2 provides an
opportunity to study the behaviour of RA-2
backscatter over land, both at Ku and S
bands. This paper presents an analysis of RA2
backscatter performance over natural land
targets, and provides s detailed comparison with
ERS RA-1 data. Comparison is also made with
Topex Ku and C band data over the identified
calibration sites.

Takla Makan
The Pearson Correlation Coefficient was
calculated and plotted for a number of Envisat
passes over the model area. Fig. 1 shows the
values obtained. Fig. 2 shows the mean difference
from the model for each of these passes, and Fig.
3 shows the standard deviation of the mean
difference. A number of individual passes, for
all SGDR retrackers, show correlation of above
0.9 with ice 1 showing the highest degree of
overall correlation, averaging 0.66. Ice1 Mean
Offset -6.012 Std.Dev. 2.800 Ice2 Mean
Offset -7.333 Std.Dev. 2.763 SeaIce Mean
Offset 4.572 Std.Dev. 2.207 Fig. 4 shows s0
values for a single track together with modelled
values, and Fig. 5 shows the differences from
modelled values. For this track ice2 shows the
most stable correlation with the model. This is
an expected result, since along this track
ocean-like echo shapes predominate.
Figure 1
Figure 4
Using ERS-1 35 day and geodetic mission ice mode
datasets, s0 models of previously identified land
areas were created (manually removing
environmentally contaminated signals), creating a
s0 modelled surface using Delauney Triangulation
and Bilinear Interpolation. The model accuracy
was then tested by comparison with the actual
values on independent tracks. Extremely good
results were obtained. The model was then used to
cross calibrate between ERS-1 and ERS-2 ice
modes, giving a Least Squares offset of 7.55 dB
with a standard deviation of 0.02 dB. The high
offset is an expected result because ERS-1 ice
mode sigma0 contains a positive bias. A
comparison was then performed for Envisat Ku
band, ocean mode s0 using the four given values
in the SGDR, both using all data and selecting
only data with appropriate waveform shapes. This
latter approach provided improved correlation
with the model.
Figure 2
Figure 5
Topex
Fig. 6 Shows the mean offset between TOPEX Ku and
C band s0 and modelled values. Ku band shows a
mean offset oscillating around 1 and 1.5dB
values with a variation of about 3 dB. C band
values oscillate around -5dB with a variation of
about 3dB. There is a clear offset of 6 to 7.5dB
between Ku and C band s0
Figure 3
Arabian Desert
A number of passes, for all SGDR retrackers show
correlation of above 0.8, with ice 1 showing the
highest degree of correlation, averaging 0.77.
Ice 1 s0 show an average offset from modelled ERS
1 values of -8.82 dB, with Std. Dev. of 0.60.
Ice1 Mean Offset -8.222 Std.Dev. 0.596 Ice2 Mean
Offset -11.386 Std.Dev. 0.769 SeaIce Mean
Offset -2.608 Std.Dev. 1.920
TOPEX coverage over the Arabian Desert was found
to be intermittent, with large and variable
offsets from the model and low correlation
coefficients due to the poor coverage.
Fig ?
Simpson Desert
A number of individual passes, in all modes, show
correlation of above 0.7 with ice 1 showing the
highest degree of correlation, averaging 0.72.
Ice 1 s0 values show an average offset from
modelled ERS1 values of -7.79 dB, with Std. Dev.
of 1.25. Ice1 Mean Offset -7.790 Std.Dev. 1.252 I
ce2 Mean Offset -10.022 Std.Dev. 0.369 SeaIce Mea
n Offset 0.393 Std.Dev. 1.961
The Envisat RA-2 is the first space-borne
altimeter designed to collect data over all the
earths surfaces. Initial results indicate that
in Ku band, ocean mode, all the given SGDR
retracker s0 values show a significant degree of
correlation with the detailed calibration model
derived from ERS1 RA-1 data. Overall, the
greatest degree of correlation of the model with
the SGDR retracked s0 values is obtained with
ice1 , with mean offsets from the calibration
model of -6.0 to -8.2 dB (ice1), -7.3 to -11.4 dB
(ice2) and 4.6 to -2.6 dB (sea ice). These
results are broadly as expected, since the desert
calibration zones return many wide ocean-like
waveforms (but with a more steeply sloping
trailing edge then Brown model waveforms).
However, it is clear that for land applications,
sigma0 must in fact be recalculated for each
waveform according to the actual shape and the
given SGDR values are not generally appropriate.
Data point distribution for Envisat is broadly
similar to that of ERS-1/2, and far superior to
that provided by TOPEX not only because of the
higher sampling rate but also because TOPEX loses
lock more frequently. The relatively poor
correlations seen between the model and Topex
data is an expected consequence of the model
optimisation for ERS/Envisat . The calibration
models will now be tuned for Topex and
Jason-1. The next part of this work is to
recalculate the Envisat s0 values for land
applications and repeat the analysis over the
land calibration targets. Acknowledgments The
authors wish to thank the European Space Agency
and NASA for supply of data . This work was
partly undertaken within the Envisat RA-2
Calibration/Validation programme.