'anbul 2004
verlap)
ping areas
'ovider) :
Area 6
E s
-6
27
10,0
7,1
4
32
19,5
6,5
-9
27
8,5
7,8
bias with
)m for area
iny bias in
5. The bias
16 but also
by the fact
ace DEM
gs and the
viation for
n in area4,
EM some
en off the
-obably for
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004
Red = reference
Green = H. Raggam DEM
Maximal shift = 250m
Fig 15c: Yellow profile
Fig 15: Example: Valley filled up in H. Raggam DEM
2.8 Qualitative analysis
The statistics results show small differences between the 3
providers, each one used his own processing. We don’t know
exactly the software used by W. Kornus and H. Raggam, P.
Reinartz being the only one to specify the software in his
report used the DLR software for modelisation and matching.
To easily compare all the DEM, we converted all altitudes in
meters, and if necessary took off the bias
PA A 5 POE "
Di
In W. Kornus DEM (Fig.17) there are see some default in this
DEM but the altitude is quite well restituted, the biggest relief
shapes are visible (see next profiles). The P. Reinartz DEM
(Fig.18) is a little smoother but the resolution is also different
(15 m instead of 10 m for W. Kornus DEM). We can see
some regular small patterns (nearly 3m every 50m), which
look like a sampling quantification problem. H. Raggam
global DEM (Fig 19) is smoother, we can't see all the valley
in relief and there have also regular defaults but different
from P. Reinartz DEM ones, they look more like waffles or
scratches. H. Raggam local HRS DEM (Fig 20) is more or
less the same for this local DEM. It is a bit better according
to the resolution (5m). H. Raggam local HRG/HRS DEM is
quite noisy but we can clearly see all the details in the relief.
This is confirmed by the following profiles. (Fig 21)
This is confirmed by the following profiles (Fig 22) which
has been drawn in purple in the previous DEM. The altitudes
along this profile are bounded by 435 and 703m.
