ISPRS Commission III, Vol.34, Part 3A ,,Photogrammetric Computer Vision'*, Graz, 2002 
  
It can be seen that negative errors tend to be on the left hand 
side and positive errors on the right. 
  
  
  
  
Figure 11 — Height errors represented as arrows in x direction 
This error pattern suggests that coefficient B in equation (8), 
which was assumed in a first approach as a constant, should be 
considered as a function of the x image coordinate: 
B(x, y)» B, * Bx (10) 
There are two coefficients to determine, which require that 
heights of at least 2 tie-points are known. These 2 points should 
have some separation in x direction in order to determine Bl 
with some accuracy. If more heights are known all should be 
used in a least-squares adjustment. 
There is a theoretical explanation for equation (10). SPOT 
parameters extracted from the header are not exact, which 
explain the systematic errors in parallaxes. Errors in the orbit 
elements have a constant effect in the image location on the 
ground. However that is not the case with errors in attitude 
angles, in particular the roll angle. The SPOT sensor has a field 
of view of 4°. When projecting from image to ground, the effect 
of an error in roll is variable from the left to right side if the 
image. As the field of view is small there is an approximately 
linear relation between the image coordinate and the error on 
the ground coordinates. The points used in the image 
registration have variable errors, affecting parallaxes with a shift 
that varies linearly with x image coordinate. It is then possible 
to slightly improve the model for height determination. 
S. DISCUSSION AND CONCLUSIONS 
It was possible to verify that a SAR image can help in 
improving the orientation of a SPOT image, requiring as little 
as one height control point. Heights of SAR-SPOT tie-points 
could be determined from parallaxes. These heights were found 
to have a RMS error of 12 m when compared with map heights. 
The tie-points were transformed into GCPs, which allowed for 
the SPOT image resection with accuracy a bit larger than one 
pixel. 
An essential requirement is that an accurate SAR image 
orientation must be known, which was not the case of the 
Radarsat image used. For practical reasons the Radarsat image 
orientation was improved. This improvement required ground 
control, which is against the principle of trying to use as little 
ground control as possible. The actual application of this 
methodology would require an ERS SAR image. 
Serious difficulties can be found when looking for tie-points in 
the SAR image. However, as water surfaces are clearly 
identified on the SAR image, their boundaries could be used to 
look for their conjugates on the SPOT image. Relatively 
accurate tie-points can be obtained in this way provided that the 
shape of the features does not change from one image to the 
other. That might be the case with reservoirs due to changes in 
water levels. Provided that some stable features, common to 
both images, exist tie-points can be obtained. 
The ground control data required is only altimetric. It is much 
simpler to obtain the height of a lake, for example, than precise 
planimetric coordinates of a point. 
The methodology proposed, although not completely 
eliminating ground control, strongly reduces it. It may be useful 
in order to simplify the process of topographic mapping from 
images of remote regions. 
REFERENCES 
CNES, 2000. SPOT4 internet web pages: http://spot4. 
cnes.fr/spot4. gb/index.htm (visited on 20-09-2001). 
Curlander, J., 1982. “Location of Space borne SAR Imagery". 
IEEE Transactions on Geoscience and Remote Sensing. 20 (3): 
359-364. 
Gongalves, J., 2001, “Integration of SAR and SPOT data for 
topographic mapping”. Ph.D. Thesis, University of London. 246 
pages. 
Gugan, D. and I. Dowman, 1988. Topographic mapping from 
SPOT imagery. Photogrammetric Engineering and Remote 
Sensing, 54 (10): 1409-1414. 
Mohr J., and S. Madsen, 2001. "Geomtric calibration of ERS 
satellite SAR images ". IEEE Transactions on Geoscience and 
Remote Sensing, 39 (40) 842-850. 
Olander, N., 1998. “Modelling Spaceborne and Airborne 
Sensors in Software”. International Archives of 
Photogrammetry and Remote Sensing. 32(2): 223-228 
Renouard, L. and F. Perlant, 1993. "Geocoding SPOT products 
with ERS-1 geometry". Proceedings of the Second ERS-I, 
Space at the Service of the Environment, Hamburg: 653-658. 
Raggam, J., A. Almer, and D. Strobl, 1994. *A combination of 
SAR and optical line scanner imagery for stereoscopic 
extraction of 3-D data”. ISPRS Journal of Photogrammetry and 
Remote Sensing. 49 (4): 11-21. 
Toutin, T., 2000. “Stereo-mapping with SPOT-P and ERS-1 
SAR images”. International Journal of Remote Sensing, 21 (8): 
1657-1674. 
Rufenacht, H., R. Proulx, and P. Cefola, 1997. “Improvement of 
RADARSAT image localization”. In Geomatics in the Era of 
Radarsat; Directorate of Geographical Operations, Canadian 
National Defense; CD-ROM. 
Westin, T., 1990. “Precision rectification of SPOT imagery”. 
Photogrammetric Engineering and Remote Sensing, 56(2): 247- 
253. 
ACKNOWLEDGEMENTS 
Radarsat images were provided by the Canadian Space Agency, 
under the Application Development Research Opportunity 
(ADRO). 
Digital topographic map data was provided by the Instituto 
Geográfico do Exército, from the Portuguese Army. 
A - 130