The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
461
Clearly, errors in knowledge of the quantities in this equation
impact the total INSAR performance. The propagation of a
baseline uncertainty to a topographic height error has already
been derived by Zebker and Goldstein (1986). The high
sensitivity of the InSAR height on B, a and O puts
extremely high requirements on the knowledge of these
parameters. The trade-off between these errors was a large part
of the mission design. Supposing effective across-track
baselines B for high resolution DEM’s is in the order of 1 km,
Table 2 represents the error in elevation caused by errors of the
INSAR system.
3. GPS APPLICATION
Dual frequency GPS are capable of eliminating the ionospheric
signal errors and thus to make best use of the high accuracy
GPS carrier phase measurements. Aside from space
applications like real-time positioning, precise orbit
determination, attitude determination of spacebome sensors,
dual-frequency GPS receivers are considered as practical and
cost-effective system for precise baseline determination and
time & frequency synchronization in DSS missions.
3.1 Position Determination
INSAR accuracy is obtained by careful measurement of the
baseline length and orientation and the location of the platform
relative to the reference coordinate system.
3.1.1 Baseline Estimation
In INSAR processing the knowledge of precise geometrical
parameters of a DSS is essential. The precise baseline vector
determination is based on direct evaluation of dual frequency
GPS carrier phase measurements. Recent studies have
demonstrated the usefulness of GPS receivers for relative
positioning of formation flying satellites using dual-frequency
carrier-phase observations(Meyer,2004). The analysis
performed with GFZ's Earth Parameter and Orbit System
utilizing an adapted GRACE configuration shows that a
relative position knowledge of 1 mm can be achieved in most
cases( Kroes,2005).
The measured position is that of the GPS antenna. Considering
the impact of satellite attitude errors and uncertainties in both
the GPS and the SAR antenna phase centre positions, Baseline
estimation could be realized with 8mm accuracy(three axis).
3.1.2 Precise Orbit Determination
Precise knowledge of the orbit is required in order to obtain the
required 1-m position determination. Furthermore, formation
flying DSS also require an accurate real-time knowledge of the
position. In accordance with INSAR applications, dual
frequency GPS receivers are also preferred for precise orbit
determination and navigation of LEO spacecraft. Tracking
Accuracy can reach: < 10m (real time); < 10 cm RMS (post-
processed); velocity accuracy can reach:< 10cm/sec (real-time).
3.2 Time & Frequency Synchronization
Time & frequency synchronization error occurs because of the
different formation flying platforms and different frequency
sources. The phase coherence of different echoes, which is the
key to SAR imaging and INSAR processing, must be
guaranteed. Hence there must be higher requirement of the
accuracy for time and frequency synchronization. For a DSS
operation, some means must be provided to ensure that signals
are received at the proper time and frequency.
3.2.1 Frequecny synchronization
Frequency synchronization is required to position the receive-
only signal within the azimuth prefilter bandwidth, also the
range shift of the impulse response will be dominated by
deviations between the PRFs of the transmitter and receiver.
Since the PRF is usually derived from the STALOs , the shift
in slant range may be derived as
2.M0 = _l^_ (6)
ct 10 c-/
where r is the range resolution, c is the light velocity, t is the
allowable payload working duration, y is the relative frequency
deviation between the two STALOs . By solving formula (11),
we get y<5E-l 1.
All required transmitter and receiver frequencies are generated
from a stable local oscillator (STALO) by means of
multiplication, division and phase-locked loops. The internal
STALO with 5 x 10-Ell short term stability determines the
short time stability which is required by master satellite SAR
prcessing, while the long time stability is disciplined to the
stability of the GPS satellites atomic clocks. The atomic clock
onboard the GPS satellite has an accuracy of Af/f <10-14. As
long as the receiver is locked to the GPS, the long time stability
of STALO can reach 10E-11. In this way the highest possible
coherency between master and slave satellite is ensured.
Synchronising master and slave satellite internal time base to
GPS timing is practical and cost effective.
In the case of GPS receiver unlock, the stability of the STALO
unit will determine the drift due to aging and temperature
changes. The idea way is to use atomic clock which is also
disciplined to the GPS.
3.2.2 pulse-timing synchronization
Time synchronization includes pulse-timing synchronization
and absolute time synchronization. Pulse-timing
synchronization is used to synchronize the SAR data window
start for the transmit/receive radar and the receive-only radar so
that both radars can be triggered at a fixed time delay to meet
the INSAR swath. Pulse-timing synchronization accuracy cr r
should be less than 0.01 times of the PRT (pulse repetition time)
to provide sufficient overlap of swath, so <J r < 2.5us is
required.
GPS receiver outputs pulse-per-second (PPS) signal once a
second. The rising edge of PPS which define each second start
has an accuracy of 50ns. PPS signal is suitable for appropriate
positioning of the data window and can be used for timing pulse
transmission and range estimation for Distributed SAR Satellite.
The receiver of slave SAR is always triggered at a fixed time
delay after the signal is transmitted to compensate for the range
difference between the distributed satellites.
3.2.3 bsolute Time Synchronization
Absolute time synchronization accuracy cr T is also essential
for DSS. Absolute time synchronisation needs to be used to
trigger the data acquisition events and precisely timestamp the
SAR acquisition data and baseline estimation data. cr T should