509 
In: Wagner W., Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
It can be seen in Table 6 TSX SL DSM has a better relative 
accuracy than IKONOS DSM. That means the interior integrity 
of TSX model is powerful, limitations may be caused by phase 
unwrapping. 
4.3 Differential DEMs 
Here differences between height models (DIFFDEMs) have 
been created for visualization. During the generation of three 
dimensional DIFFDEMs with program LISA, a view direction 
of 160° (0= North, 90= East) and an inclination angle of 35° 
was chosen for all DIFFDEMs. An exaggeration factor of 5 was 
used to provide an optimal visibility of the differences. Figure 
13 shows the DIFFDEMs between IKONOS and TSX SL 
DSMs. 
Figure 13. DIFFDEM between IKONOS and TSX SL 
Height differences are obvious in patches especially in woody 
areas (black parts). Except woody regions TSX SL and 
IKONOS DSMs are coherent and the height differences are in 
between 0-5m 
4.4 Shading 
The shading is one of the best possibilities of visualizing a 
DSM. All buildings, vegetation and forest coverage in a DSM 
can be seen clearly. By this method, the shadings of IKONOS 
and TSX SL DSMs which have the same grid spacing (3m) 
have been generated and compared. Figure 14 and 15 illustrate 
the results of shadings. 
Figure 15. Shading (TSX SL) 
Looking at the results, it can be mentioned that the 
visualizations of both DSMs are pretty good. They illustrate the 
ground objects like roads (in red circles), buildings, vegetation 
and forest. 
5. CONCLUSION AND FUTURE TARGETS 
In this study, DSMs which have 3m grid spacing have been 
generated from TSX SL and IKONOS Pan image-pairs. And 
these models have been compared in Istanbul test field, Turkey 
using a reference digital elevation model, derived from 1:1000 
scale aerial photos and has 10cm up to lm accuracy. 
In the densely build up area of Istanbul DSMs are strongly 
influenced by buildings, partly also by trees and forest. The 
influence of the buildings and the vegetation can not be filtered 
totally because of missing satisfying number of points on the 
ground. 
InSAR height models have a clear dependency upon the 
aspects. TSX SL DSM have a better relative accuracy as 
absolute accuracy, which may be explained by the influence of 
phase unwrapping, but also the case that DSMs are compared 
with reference elevation model containing the height of the bare 
ground. 
It can be mentioned that TSX SL image-pair which has suitable 
baseline length (135m) presents an absolute accuracy 
competitive to the height model based on the very high 
resolution optical images. A dominating effect of the accuracy 
is still the vegetation and buildings. 
Future investigations and development will be DSM fusion with 
optical DEMs based on space and/or aerial images and support 
high resolution optical images for generation of maps 1/5000 
scale and below. 
ACKNOWLEDGEMENTS 
Thanks are going to TUBITAK, Turkey and DLR, Germany for 
their supports to this research. And to Dipl. -Ing. Abdalla 
Alobeid for supporting the generation of IKONOS DSM. 
REFERENCES 
Figure 14. Shading (IKONOS) Alobeid, A., Jacobsen, K., Heipke, C., 2009. Building Height 
Estimation in Urban Areas from Very High Resolution Satellite 
Stereo Images, ISPRS Hannover Workshop, 2-5 June, 
Hannover, Germany