2.12 Data Handling Cookbook
2.12.1 Hints And Algorithms For Product Use
The evaluation of the quality of the EnviSat
Altimetry geophysical data products
and the cross calibration on ERS-2
and other flying altimeters (e.g. TOPEX, Jason,
GFO,...) has been performed during
the Commissioning Phase within a team
of scientists drawn up from the pre-launch
announcement of opportunity: the
RA-2/ MWR Cross-Calibration and Validation
The product validation approach has been to
verify, with real data, the
consistency of the product package (document,
media, format and actual data set), to quantify
the inherent validity and accuracy
of the range, altitude, wave height,
wind speed measurements. etc. and geophysical corrections.
Here follow the main results of the work
performed by the CCVT in terms of
altimetric parameters performance, errors,
proposals for future algorithmic improvements, etc.
RA-2 measures low wave heights much
better than ERS, due
to the presence of two additional
located on the waveform leading edge to
improve the slope estimation. The
been reduced (although not completely
EnviSat, wrt ERS-2, due to a better SWH
EnviSat SWH compares well with ERS-2,
buoys and models:
Envisat SWH are 21 cm (or 3%) higher
than ERS-2, there is
a 10 cm bias with buoys, a 16 cm
bias with WAM SWH model and 21 cm rms/WAM.
RA2 wind speed rms is only 1.4 m/s,
compared to ECMWF fields.
Ku-band sigma0 was artificially aligned
with ERS-2 to
satisfy modified Witter and Chelton wind model.
The estimated value of the sigma0 bias,
coming from the
absolute calibration activities, is
around 0.95 dB (RA-2
higher), pretty much in line with the
results of the RA2 passive calibration
(0.91 dB). The
passive calibration activities have
provided also with a
(preliminary) value of the
interfrequency bias, set currently to
This value will be further tuned with
larger sets of data.
The actual values of Ku sigma0 coming
from the absolute
calibration will be applied to the data
only when a new wind model will be developed.
The range noise, estimated on short
segments over low
waveheight areas of the ocean, is 1.8 cm
and 5.2 cm (at 1 Hz)
for Ku-band and S-band, respectively.
From the cross-calibration with ERS-2, a
bias of around -40
cm was found (but this value could be
still sensitive to preliminary sea state
An improved noise spectrum, with a higher
(than in ERS) is present: 14 Km, and
less when 20 Hz data
Using the output of the ocean retracker
and the geophysical
corrections in the product, the sea
surface height rms at cross-overs was
estimated by CLS,
Newcastle Univ. and MSSL, at
7.0 and 8.5 cm, with a further reduction
to 6.25 cm when an
orbit error correction was applied
The pseudo time-tag was estimated
cross-overs over ocean. No significant
slope was detected
in the sea surface height as a function
of the altitude
rate, therefore confirming that there
is no significant time tag error.
The square estimated mispointing is less
than 0.01 deg2. The
mispointing value in the L2 products
(around 0.026 deg2) is erroneous.
The algorithm in the operational
processors is being investigated.
The dual frequency ionospheric correction
against JPL GIM model (depending on the
SSB which is not yet final on the
limited data set
available) with good agreement between
both. The DORIS
model, instead, showed an
limited dynamic range compared to the
GIM and dual-freq
The DORIS iono has been in the meantime improved.
All other geophysical corrections have
MWR calibration and validation:
The Level 2 wet tropo correction is in
very good agreement
with the ERS-2 MWR one (relative bias
< 5 mm) and with
MWR 36 GHz channel may require anyway
due to aging and thermal cycling of
the components. The current drift from
the beginning of the
mission is of 0.25 counts/K.
A new MWR side lobes correction, to be
implemented in a
near future in the operational L1b
improve the MWR corrections in the
A new neural network algorithm has been
for the computation of the MWR corrections.
Continuous monitoring of the 36 GHz
channel gain drift
and local oscillator behaviour is
assess possible stabilisation.
Orbit heights from CNES, NCL, DUT and
ESOC agree to within
2-3 cm rms. Radial orbit error on the
GDRs is estimated
< 4 cm rms.
The orbit propagated into the NRT
does not yet use the DORIS Navigator, so
it is very poor: 20
to 30 m error in the radial
component with jumps of several meters
between orbit state vectors.
The orbit upgrade is currently under
will be implemented in a near future.