International Archives of Photograminetry and Remote Sensing. Vol. XXXII Part 7C2, UNISPACE III, Vienna. 1999 
88 
I5PR5 
UNISPACE III - ISPRS/EARSeL Workshop on 
“Remote Sensing for the Detection, Monitoring 
and Mitigation of Natural Disasters” 
2:30-5:30 pm, 22 July 1999, VIC Room B 
Vienna, Austria 
synthetic antenna of great length by integrating the complex 
signals in amplitude and phase returned from targets over a dis 
tance along the track. In the case of a satellite SAR with a 10 m 
long antenna at 5.3 GHz and at 800 km altitude (ERS) the an 
tenna ‘sees’ a stretch on the ground of about 6 km in the azimuth 
direction - determined by the real aperture. A satellite has a 
ground velocity of about 6 km/sec and since the satellite accord 
ing to theory must not move more than half the length of the 
antenna between the emitted pulses the pulse repetition fre 
quency becomes of the order of 1 kHz. The received signal rep 
resents therefore 1000 samples of observation at 5 meter. With a 
pulse length of 60 nanoseconds the range resolution is 9 metres. 
The signal processing comprises a technique called focusing, 
which is carried out within each band of 9 meter within the swath 
of the antenna. This is a major computational task. In the case of 
ERS SAR the final resolution is of the order of 12.5 in. Multi 
looking, i.e. combination of three sets of resolution cells to re 
duce noise, i.e. speckle that is a characteristic of any coherent 
systems results in a spatial resolution of 30 m in a 100-km 
swath. 
In principle, a veiy fine spatial resolution may be achieved by 
means of SAR by applying veiy short pulses and improving the 
integration teclmiques. However, the volume and therefore the 
bandwidth of these signals transmitted to ground lead to a trade 
off between the resolution that may be achieved and the swath of 
the system. Thus, fine-resolution systems will in general have a 
smaller swath than medium-resolution systems. Future satellite 
SAR systems with a resolution of, say 5 m, may provide a swath 
of only 10 km. 
However, fine ground resolution in even very wide swaths, ex 
emplified by the ScanSAR system of RADARSAT, may be de 
vised by beam swinging as already described. This type of sys 
tem needs an active (scanning) array antenna with a great number 
of active antenna elements (each a transmitter and a receiver)that 
increases cost and complexity of the system. However, beam 
swinging requires pre-programming of the SAR system that 
entails a time delay. 
In connection with the ERS mission a system of alert lias been 
implemented by ESA/Eurimage under the name of Earth Watch 
ing. SAR operation is accomplished so that the user will have 
processed data at disposal within 24 hours after request, orbit 
configuration permits. RADARSAT International has estab 
lished a similar system which also covers beam-swinging within 
the wide coverage of 500 km. This will also be the case with 
ENVISAT ASAR, although ‘only’ within a coverage of 400 km. 
I have referred only to satellite SAR but airborne SAR are 
equally useful. Polarimetric airborne SAR have been imple 
mented with resolutions down to 2 metre (single look) in a swath 
of 8 km, for instance. 
Radar features 
Two important features are the lay-over effect and shadowing. 
The former stems from the fact that in essence a radar system is a 
time measurement, i.e. the distance to a target is calculated from 
the time lapse between emission and reception of a pulse. Thus, 
there may be cases where an elevated target is ‘seen’ before a 
lower target. A hill will therefore be seen as leaning towards the 
radar. This is a general feature of systems with a small angle of 
incidence such as ERS SAR with its 23° angle of incidence. It 
may be corrected for by using a DEM, for instance. 
In contrast, shadowing that is observed in hilly terrain by systems 
with a large angle of incidence may not be corrected for except 
when observing the hills from two or more directions which is 
only feasible with airborne systems or by satellite systems with 
two identical radar’s looking in opposite directions (and in that 
case on a later orbit) or from a complementary satellite. Wishful 
thinking! 
The interested reader may obtain detailed information on SLAR 
and especially SAR in the book by C. Elaclii (1987). 
Interferometric SAR - InSAR 
Satellite SAR data may be used for interferometry as demon 
strated with SEASAT and ERS data. The techniques is based on 
raw SAR data that are acquired from the same area on different 
passes, made possible since the repeated tracks normally will be 
at a horizontal distance of for example hundred meters. The 
techniques is based on comparing the phase of the return signals 
acquired from repeated passes by so-called umvrapping. With a 
coherent system of great frequency stability this is meaningful 
even over very long time intervals - counted in hundreds of days, 
as has been demonstrated. The teclmiques is basically based on 
measurements in range so that the image that results constitutes a 
diagram of fringes separated by half a wavelength in the across- 
track direction DEM’s may be worked out with contours of 
about 50 m. for instance. Detailed information about this appli 
cation may be obtained from Zebker and Goldstein (1986), for 
instance. 
More importantly is perhaps the feature that materialises when 
distances to the same target are determined by so-called three- 
pass interferometiy. With one observation regarded the ‘master’ 
two interferograms are produced based on data from the other 
two passes that normally will be at different positions and have 
different look-angles. By comparing the two interferograms any 
shift that may have occurred in the period between the observa 
tions may be determined. This techniques does not require avail 
ability of a DEM, since in principle one interferogram may be 
regarded a DEM to which the other one is referred - after careful 
registration to a fraction of a pixel.