The DLR SAR Calibration Center

Jens Reimann, Marco Schwerdt, Kersten Schmidt, Nuria Tous Ramon,

Gabriel Alfonzo Castellanos, Bjorn Doring, Daniel Rudolf, Sebastian Raab
John Mohan Walter Antony and Manfred Zink
Institut fur Hochfrequenztechnik und Radarsysteme
Deutsches Zentrum fur Luft- und Raumfahrt e.V.
82234 Wessling, Germany
Email: jens.reimann@dlr.de

AbstractA SAR system has to be calibrated for quantitative

measurement. During this process all essential parameters of a
SAR image are linked to their geophysical quantity. This includes
the location of the image pixel, its backscattering characteristics
(in amplitude and in phase) and polarimetric information. The
Microwave and Radar Institute of the DLR has gained a lot of
experience in these calibration procedures during the last decades
and has developed special methods and dedicated reference
targets for spaceborne SAR system calibration. In example of
calibration results obtained for different spaceborne SAR systems
the DLR SAR Calibration Center is presented.

I. I NTRODUCTION
In order to keep up with the growing demand on accurate
SAR data products on the one hand and the growing complex-
ity of innovative spaceborne SAR system on the other side
(with a multitude of different beams and novel operation modes
like TOPS, sliding spotlight, etc.), sophisticated concepts, pre-
cise algorithms, and adequate facilities are a must to calibrate
such complex spaceborne SAR systems efciently.
Different concepts and procedures together with the cor-
responding algorithms have been developed and implemented
in a number of tools for analyzing and evaluation the various
measurements. The resulting calibration parameters are then
being used to generate accurate SAR data products.
The state of the art of calibration accuracy for modern
SAR instruments will be illustrated with results from different
spaceborne SAR systems like TerraSAR-X, TanDEM-X and
Sentinel-1A. DLRs Microwave and Radar Institute has been
an active player in SAR system calibration from the very
beginnings and has built up more than 25 years of experience.
II. C HALLENGES AND S TRATEGY
An efcient calibration strategy has to be developed for
each and every SAR mission individually to satisfy its system
requirements and to ensure a product release as soon as
possible at the end of the commissioning phase. One generally
already proven concept [1] is based on an antenna model to
predict the performance of the active phased array antenna
and its hundredths of beams, which is often employed. The
calibration itself then consists of:
1) Internal Calibration, to guarantee a stable instrument,
2) Geometric Calibration, to assign the SAR data to the
geographic location on the Earths surface.
3) Antenna Pointing Determination, to obtain a well-
known beam pointing of the antenna.
Fig. 1. The DLR Calibration Field consists of 36 target positions with
20 corner reectors permanently installed. The newly installed six remote-
controlled targets are marked by blue crosses. (Map: OpenStreetMap
c
contributers, Overview: David Liuzzo, CC-BY-SA-2.0-de)
4) Antenna Model Verication, to ensure the provision
of precise reference patterns (including the gain offset
between different beams) required for all operation
modes.
5) Absolute Radiometric Calibration, for radiometric
bias correction of SAR data products.
III. DLR SAR C ALIBRATION FACILITY
The DLR SAR Calibration Facility is a well-equipped
center for the calibration of spaceborne SAR systems. An
important element is the infrastructure required for the prepara-
tion and execution of calibration campaigns even over extended
test sites. The Institute owns a multitude of calibration targets
and operates and maintains a large calibration site in Southern
Germany (Fig. 1).
In 2014 six new sites have been established for the ESA
Sentinel-1A mission. Three of them were equipped with new
precise remote-controlled trihedral corner reectors (Fig. 2),
the others with novel in-house developed C-band transponders
(Fig. 3) named Kalibri.
For deriving the different calibration parameters from
measurements executed during the commissioning phase and
during nominal operation of a spaceborne SAR system sev-
eral analysis and calibration tools have been developed and
adapted for the corresponding SAR missions (TerraSAR-X,
TanDEM-X, PAZ, ALOS, Sentinel-1).

978-1-4673-7297-8/15/$31.00 2015
c IEEE 169
 Leg Length 2.8m
RCS (X-,C-,L-Band) 54.26, 49.23, 36.56 dBm2
Mechanical Tolerance 1.0 mm
Abs. Radiometric Accuracy 0.2 dB (1)
Pointing Accuracy <0.1 deg

Fig. 2. New remote-controlled corner reector.

Fig. 4. Gain deviation over the life time of TarraSAR-X and TanDEM-X
since launch separated for receive (RX) and transmit (TX) path.

Center Frequency 5.405 GHz

Bandwidth 100 MHz
RCS 60 dBm2
Radiometric Stability 0.1 dB (1)
Abs. Radiometric Accuracy 0.2 dB (1)
Polarization exible backscatter matrix

Fig. 3. DLR C-band remote-controlled transponder Kalibri.

Fig. 5. Phase deviation over the life time of TarraSAR-X and TanDEM-X
IV. I NTERNAL C ALIBRATION since launch separated for receive and transmit path.

In addition to nominal calibration pulses generated within

the SAR instrument and applied for drift compensation during
SAR data processing, an innovative pulse coded calibration
technique (PCC) has been developed [2]. The PCC technique
can be applied to monitor and characterize the individual
transmit-receive-modules (TRMs) of the antenna. This tech-
nique also known as PN-gating method was successfully veri-
ed in-ight by TerraSAR-X and then applied for TanDEM-X
and Sentinel-1A [1]. This technique allows to monitor the
gain and phase of each module individually for the transmit
(TX) and for the receive path (RX). The resulting values are
compared with module specic references; the deviations are
regularly monitored over time to detect possible drifts and
module degradations.
The deviation of each individual module has been measured
for TerraSAR-X and TanDEM-X. The results for both satellites

170 2015 IEEE 5th Asia-Pacific Conference on Synthetic Aperture Radar(APSAR)

TABLE I. S TATISTICS OF GAIN AND PHASE T/R- MODULE DEVIATIONS
FOR TRANSMIT AND RECEIVE PATH FOR T ERRA SAR-X AND TAN DEM-X
SATELLITES .
Mean over the -values TX RX
of all modules Gain[dB] Phase[deg] Gain[dB]
TerraSAR-X 0.08 2.04 0.17
TanDEM-X 0.03 1.72 0.15
are shown for gain (Fig. 4) and phase (Fig. 5); red lines and
black error bars indicate the mean value and standard deviation
for each module, respectively. While the gain in the transmit
path is very stable for both satellites the receive path gain
is more module depended but still well within the specica-
tion. From phase deviations no SAR module degradation is
detectable for both satellites.
The standard deviations of all modules are averaged to
get an overview of the instrument quality for each parameter.
The results are summarized in Table I and showing a very
stable instrument for both systems since launch (TerraSAR-X:
2007, TanDEM-X: 2010). Hence, the PCC technique allows
to measure the actual settings of individual TRMs down to an
accuracy of one tenth of a dB in amplitude and one degree in
phase.
V. G EOMETRIC C ALIBRATION
Geometric calibration links each pixel in the radar image to
its geographic location on the Earth surface. The localization is
done in the SAR instrument by time measurements for range
direction and using the Doppler-zero-plane for azimuth. For
geometric calibration both the satellite and the target position
has to be known precisely. Two major effects inuence the
range delay:
the internal electronic delay of the instrument and
propagation effects (ionosphere and troposphere).
While the effect of the ionosphere is negligible for high
frequencies, the impact of the troposphere has to be considered.
The additional (one-way) correction due to the troposphere is
calculated according to [3] and [4] by taken into account the
zenith path delay (ZPD), the incidence angle (), the altitude
of the point target (h), and a scaling height (hs ) which is set
to 8 km, a typical upper cloud border in the midlatitudes.
1 correction across the scene) but also the gain offsets between
rtropo = ZP D eh/hs
cos
In case of TerraSAR-X and TanDEM-X a pixel localization
accuracy in the order of 10 cm was achieved [5]. This accuracy
is about one order of magnitude better than the requirement
with 1 m. By this accurately calibrated TerraSAR-X system
the impact of propagation effects on the range delay could
be veried for the rst time by a SAR system, as shown in
Fig. 6. The theoretical path extension caused by tropospheric
refraction along the hydrostatic path and well known from the
global positioning system (GPS) is depicted by the pink line.
If we compare furthermore the range delay offset of
TerraSAR-X and TanDEM-X systems, i.e. concentrating on the
mean value of the measurements for the respective satellite, the
range delay offset of both systems differs only about 54 psec.
I.e. the geometric offset between both satellites in a distance
2015 IEEE 5th Asia-Pacific Conference on Synthetic Aperture Radar(APSAR)
5,00
s023
4,00
s016
Phase[deg] s009 s026
Delta Range [m]
1.15 3,00
s019
1.02 s003
s012
2,00
s005
1,00
0,00
15 20 25 30 35 40 45 50 55
Look Angle [deg]
Fig. 6. Range offset derived from the range delay and measured for
different beams of TerraSAR-X (triangles). The theoretical path extension by
hydrostatic effects is depicted by the pink line.
of about 600 km is only 8 mm. This is only one quarter of the
wavelength. Hence, the user of SAR data products is not able
to distinguish geometrically whether the scene was acquired
by the TerraSAR-X or by the TanDEM-X satellite.
VI. I N -F LIGHT A NTENNA C HARACTERIZATION
For correct illumination of the scene and the precise
antenna pattern correction during SAR data processing, the
antenna pointing has to be known precisely. The calibration
estimates the difference between the real antenna pointing and
the desired one. This can be efciently done by commanding
so called notch beams on the phase array antenna in elevation
or in azimuth. Based on these notch patterns, the gain drop
in boresight direction can precisely be recognized, i.e. in
elevation using rainforest scenes or in azimuth by transponder
measurements. For TerraSAR-X and TanDEM-X the pointing
knowledge was proven to be better than 0.002 deg in azimuth
and in elevation [5]. Hence, by readjusting the attitude of the
satellite, the detected mispointing of the antenna beam can be
precisely removed.
A key element of an efcient calibration strategy is a
precise antenna model because then most of the antenna char-
acterization can be shifted from the commissioning phase to
pre-launch activities. Furthermore, the antenna model provides
not only the shape of the patterns (required for the radiometric
(1) different beams. Thus, absolute radiometric calibration can
rely on one absolute calibration factor valid for all beams and
modes. However, a suitable set of beams has to be measured
in-ight verifying the antenna model. For this purpose the
Institute has derived several rules and recommendations [6].
One-way pattern of the antenna in azimuth can be recorded
by transponders. Due to the movement of the satellite, an
azimuthal cut through the pattern is seen by the transponder.
For example, the performance of the antenna model was
veried to be better than 0.1 dB within the 3 dB beamwidth
for Sentinel-1A (Fig. 7). A detailed description of the results
from Sentinel-1A would exceed the content of this paper, and
the interested reader is referred to [7]).
Long term statistic of the antenna patterns measured since
launch for TerraSAR-X and TanDEM-X (using Amazon rain-
forest scenes) are shown in Fig. 11. TerraSAR-X (TSX-1)
171
Fig. 7. Relative difference between the azimuth antenna pattern measured
by the DLR transponders and derived by the antenna model within the 3dB
beam width.
Fig. 8. Comparison of the very complicated azimuth pattern in TOPS
operation between antenna model and transponder measurements.
datatakes are displayed in blue, TanDEM-X (TDX-1) in green.
The mean values are shifted to zero to focus on relative devia-
tions (Note: absolute calibration is preformed against accurate
reference targets with well-known RCS, see section VII). The
minimum and maximum deviation between the measured and
the calculated set of patterns (4 beams in ScanSAR operation)
are represented by an error bar per acquisition. The standard
deviations are depicted by triangles. The timeline shows a very
stable behavior of the antenna patterns. While the standard
deviations are always inside a limit of +/- 0.2 dB, the extreme
values do not exceed +/- 0.3 dB. There is no remarkable
difference between the two satellites.
Analyzing the whole azimuth antenna pattern becomes
really a challenge for the Terrain Observation by Progressive
Scans (TOPS) modes as realized for the Sentinel-1 mission due
to the sophisticated steering (up to 5 sub-swathes each steered
by more than 800 azimuth beams, see Fig 8). Nevertheless,
the DLR was able to verify this complicated one-way antenna
pattern in TOPS operation [8].
VII. A BSOLUTE R ADIOMETRIC C ALIBRATION
In a nal step, i.e. for absolute radiometric calibration, the
whole SAR instrument is measured against reference targets
with well-known RCS [9], [10]. As furthermore, the shape
of the antenna pattern and the gain-offset between different
beams are already compensated for by applying the antenna
172 2015 IEEE 5th Asia-Pacific Conference on Synthetic Aperture Radar(APSAR)
Fig. 9. Timeline of radar cross section (RCS) derived from permanently
installed corner reectors for TSX-1.
Fig. 10. Timeline of radar cross section (RCS) derived from permanently
installed corner reectors for TDX-1.
TABLE II. DLR C ALIBRATION T ECHNIQUES / M ETHODS
Drift Compensation: Amplitude / Phase <0.1dB / < 1 deg
TRM Characterization (PCC): Amplitude / Phase setting <0.2dB / < 2 deg
Geometric Calibration: Azimuth / Range 10cm / 10cm
Pointing Determination: Azimuth / Elevation < 2mdeg / < 2mdeg
Antenna Model: Verication < 0.2dB
Radiometric Calibration
Accuracy <0.3dB
Stability <0.2dB
model during SAR data processing, the absolute calibration
factor is be independent of both, of the target position within
the swath and the beam being operated.
And indeed, displaying the RCS of accurate reference
targets derived from all measurements executed for TanDEM-
X as shown in Fig. 9 and Fig. 10, no signicant dependency
can be observed [11]. Consequently, the absolute radiometric
accuracy can be derived by the standard deviation of all
measurements. For TanDEM-X we have achieved an absolute
radiometric accuracy of 0.14 dB for StripMap operation during
the commissioning phase. Considering furthermore the same
radiometric stability (0.15 dB as derived for TerraSAR-X two
years after launch of TerraSAR-X by an comprehensive re-
calibration campaign executed in 2009 [1]) and propagation
uncertainties of 0.25 dB, an absolute radiometric accuracy of
less than 0.5 dB could be achieved for TanDEM-X [12].
Fig. 11. Timeline of antenna pattern statistics derived from distributed target scenes (Amazon rainforest) by operating the instrument in ScanSAR mode for
TSX-1 (blue) and TDX-1 (green).

VIII. C ONCLUSION [7] M. Schwerdt, K. Schmidt, N. T. Ramon, G. C. Alfonzo, B. Doring,

M. Zink, and P. Prats, Independent Verication of the Sentinel-1A
Calibration of SAR systems is a traditional R&D eld in System Calibration, in IEEE Journal of Selected Topics in Applied
the DLR Microwave and Radar Institute. During the last 20 Earth Observations and Remote Sensing (JSTARS), Special Issue on
years, not only the demand on high accuracy of spaceborne Synthetic Aperture Radar (SAR) New Techniques, Missions and Appli-
cations, accepted, 2015.
SAR data products has increased but also the level of com-
[8] G. C. Alfonzo, M. Schwerdt, B. Doring, N. T. Ramon, and K. Schmidt,
plexity in the SAR instrument architecture and calibration. To First Results of the Sentinel-1A In-Orbit Antenna Characterization Per-
keep pace with this development, the Microwave and Radar formed by DLR, in 10th European Conference on Synthetic Aperture
Institute has built up, maintained and extended the DLR SAR Radar, 2014.
Calibration Center, including innovative targets like remote- [9] B. J. Doring, M. Jirousek, D. Rudolf, S. Raab, J. Reimann, and
controlled trihedral corner reectors or the in-house developed M. Schwerdt, The three-transponder method: A novel method for
C-band transponders Kalibri, as well as novel tools for accurate transponder rcs calibration, Progress in Electromagnetics
Research (B), 2015.
product quality control and performance analysis.
[10] B. J. Doring, K. Schmidt, M. Jirousek, R. Daniel, J. Reimann, S. Raab,
In example of different spaceborne SAR systems, important A. John, and M. Schwerdt, Hierarchical Bayesian data analysis in
radiometric sar system calibration: A case study on transponder cal-
aspects of SAR system calibration were described (including ibration with RADARSAT-2 data, Remote Sensing, vol. 5, no. 12, pp.
internal calibration, geometric calibration, antenna pointing 66676690, 2013.
determination, antenna model verication, and absolute radio- [11] K. Schmidt, G. C. Alfonzo, N. T. Ramon, M. Bachmann, and M. Schw-
metric calibration) and the accuracy of the different calibration erdt, Calibration Performance of the TerraSAR-X and TanDEM-X
techniques/methods were shown, as summarized in Tabel II. Satellites since Launch, in 10th European Conference on Synthetic
Aperture Radar, 2014.
[12] M. Schwerdt, J. H. Gonzalez, M. Bachmann, D. Schrank, B. Doring,
R EFERENCES N. T. Ramon, and J. M. W. Antony, In-Orbit Calibration of the
TanDEM-X System, in 30th International Geoscience And Remote
[1] M. Schwerdt, B. Brautigam, M. Bachmann, B. Doring, D. Schrank, and Sensing Symposium, Vancouver, Canada, 2011.
J. H. Gonzalez, Final TerraSAR-X Calibration Results Based on Novel
Efcient Methods, in IEEE Transaction on Geoscience and Remote
Sensing, 2010, Vol. 48, no.2.
[2] D. Hounam, M. Schwerdt, and M. Zink, Active Antenna Module Char-
acterisation by Pseudo-Noise Gating, in 25th ESA Antenna Workshop
on Satellite Antenna Technology, Noordwijk, Netherlands, 2002.
[3] M. Jehle, D. Perler, D. Small, A. Schubert, and E. Meier, Estimation
of atmospheric path delays in TerraSAR-X data using models vs.
measurements, Sensors, vol. 8, no. 12, pp. 84798491, 2008.
[4] O. Frey, E. Meier, D. Nuesch, and A. Roth, Geometric error budget
analysis for TerraSAR-X, in 5th European Conference on Synthetic
Aperture Radar EUSAR, 2004, pp. 513516.
[5] M. Schwerdt, D. Schrank, M. Bachmann, J. H. Gonzalez, B. Doring,
N. Tous-Ramon, and J. W. Antony, Calibration of the TerraSAR-X and
the TanDEM-X Satellite for the TerraSAR-X Mission, in 9th European
Conference on Synthetic Aperture Radar, 2012.
[6] M. Schwerdt, B. Doring, M. Zink, and D. Schrank, In-Orbit Calibration
Plan for Sentinel-1, in 8th European Conference on Synthetic Aperture
Radar, 2010.

2015 IEEE 5th Asia-Pacific Conference on Synthetic Aperture Radar(APSAR) 173