d part 
r field 
erived 
digital 
essing 
Ways. 
ftware 
'essing 
(1) the 
2) the 
ig, and 
In this 
/sed in 
ta sets, 
on the 
ccurate 
e shift 
asuring 
coarse 
| check 
coarse 
fts. 
  
irtesy of 
| scenes 
rt of the 
ntarctica 
surface 
rrelation 
between 
direction 
  
Figure 2: Interferogram with a shift -2/-5 (Image courtesy of 
University Freiburg) 
Assuming a sufficient number of fringes, the fine registration 
can be performed. The accuracy of the fine registration depends 
mainly on a precise knowledge of orbital parameters. À further 
improvement can be achieved by deploying corner reflectors. 
These can be used as ground control points as well as other 
points which can be clearly identified. The use of ground 
control points becomes essential if data are used from systems 
with a low precision tracking (e.g. data from the shuttle mission 
SIR-C/X-SAR). 
To reduce the amount of noise at each stage of the processing, 
filtering techniques can be applied. The use of any filter 
technique improves the images visually; however, it reduces the 
information content at the same time. Until now, only filtering 
techniques have been applied in SAR interferometry which 
were originally developed for other applications. This opens a 
new field for the development of new filtering techniques due 
to an optimal conservation of the phase information given by 
the complex imagery. 
Based on the finely registered images the interferogram as well 
as the coherence image can be calculated. The quality of the 
coherence depends mainly on the way the resampling is 
performed. There are several implementations given in the 
literature (Lin et. al., 1992; Small et. al., 1993; Geudtner, 1995) 
which are based on different considerations due to the time 
required for the data processing. 
3. IMPROVEMENTS 
There are, in general, two possibilities to improve the quality of 
SAR interferometric products. The first aspect is the accuracy 
and amount of the data introduced into the data processing. 
For the use of ERS-1/ERS-2 data, for example, precise orbit 
parameters are provided by ESA. Based on a better knowledge 
of the orbit, the geometry is more accurate, which leads to 
better results for the registration of the images. 
Data from the SIR-C/X-SAR shuttle mission provide the user 
with data sets from different wavelengths. Because of the 
different backscattering behaviour of the surface topography 
109 
due to the wavelength, data are obtained which could be used, 
¢.g., as additional informaton for solving the ambiguity of the 
phase to improve the phase unwrapping. 
The influence of atmosphere effects is assumed to decrease by 
averaging as many as possible data sets over the same area, 
which is the same approach used for reducing speckle. 
The performance of the data processing itself is the other aspect 
to improve the quality of the results. For example, by using 
cubic splines instead of a bilinear interpolation during the 
resampling process, the signal-to-noise ratio can be increased. 
This leads to an improvement of 10 % of the coherence in the 
coherence image (Geudtner, 1995). The data processing is a 
complex task which is still on the way to reach an operational 
status. At present, there are no established commercial software 
packages for SAR interferometry on the market. 
4. CONCLUSIONS 
One of the most sufficient ways for a complete description of 
the complex structure of the influencing factors is an error 
propagation model. This needs to be further developed in order 
to estimate the influence of single parameters due to an 
interferometric product. 
The influence of the atmosphere is assumed to be one of the 
most limiting factors due to the accuracy which can be reached 
by SAR interferometric techniques. The distortions caused by 
atmospheric effects appear locally and vary in time and are 
therefore difficult to correct. This aspect still needs to be further 
investigated. 
At present, one of the main issues in the development in SAR 
interferometry is to reach an operational level for the different 
applications. The basic techniques are well studied and 
understood. However, there are several aspects in the data 
processing scheme to be optimised in terms of performance, 
accuracy and time. This includes especially the development of 
new filtering techniques. 
ACKNOWLEDGEMENTS 
This study was carried out in cooperation with the German 
project ‘Dynamic Processes in Antarctic  Geosystems 
(DYPAG) coordinated by the Institute for Physical 
Geography, University of Freiburg. 
The ITC research on SAR interferometry forms part of the 
CEC’s Human Capital and Mobility Programme Research 
Network “Synergy of Remotely Sensed Data”. Contract No. 
CHRX-CT93-0310. 
REFERENCES 
Askne, J. and Hagberg, J.O., 1993. Potential of interferometric 
SAR for classification of land surfaces. Proceedings of 
IGARSS ‘93, Tokyo, Japan, pp. 985-987. 
De Fazio, M. and Vinelli, F., 1993. DEM Reconstruction in 
SAR Interferometry: Practical Experiences with ERS-1 SAR 
Data. Proceedings of IGARSS ‘93, Tokyo, Japan, 
pp. 1207-1209. 
Gens, R. and Genderen, J.L. van, 1996. SAR Interferometry - 
Issues, Techniques, Applications. International Journal of 
Remote Sensing (in press). 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996