NDVI 
| to un- 
nd use/ 
inly in 
in the 
ed over 
nsuper- 
n. The 
of the 
'sponse 
factors 
method 
e fields 
olution 
vith the 
the re- 
ion did 
lid sig- 
. Digi- 
ith the 
FSPOT 
ult was 
[here’s 
re ana- 
sic The 
a high 
e ratio. 
fusion 
tant to 
antee a 
| 2002) 
c maps 
pdated 
ements 
uch as 
IAPRS & SIS, Vol.34, Part 7, "Resource and Environmental Monitoring", Hyderabad, India,2002 
settlements, water, paddy fields, woodlands and so on were 
extracted. For extracting these elements different methods and 
means were used. 
The land use map was edited in ArcView. In the first step the 
data of settlements, water, vegetation, roads and rivers were 
overlaid on the basis of merged SPOT images, fused images 
from different sensors, referring to topographic data, admini- 
strations and DEM data from topographic maps. Then, spatial 
data of every element was put in and the attributes were named 
according to the requirements for land use data in the project. 
At last, topological relation of data was established in the Arc/ 
INFO coverages. 
  
  
ee BÓ eei pure Ren 
8 60000 665000 670000 675000 680000 685000 690000 
| 
Ó 
e 
3535000 3540000 
O00SESE 0000FSE O00SFSE 
3530000 
O000€£S€E 
3525000 
000SZSE 
3515000 3520000 
000SLSE 0000ZSE 
3510000 
00001SE 
  
  
  
000S0SE 
  
660000 665000 670000 675000 680000 685000 690000 
0 5 10 15 20 Kilometers 
  
uncultivated, waste land EE vegetation, bamboo 
arable, irrigated M water plane 
{| arable, non-irrigated Elli settlement, town 
..... uncultivated, other | settlement, village 
L__ | vegetation, orchard |__| transportation 
i54 vegetation, mulberry feel mining area 
___) vegetation, tea garden 
__| vegetation, mixed conif + decid. 
vegetation, coniferous 
| industry, large development area 
| industry, independent factory 
special landuse 
   
Figure 7. Land use map for the SILUP study area 
3.7 Two methods for DEM reconstruction 
A reliable DEM is the base for many applications of the SILUP 
project. All these applications would like to have a perfect 
DEM. Because of the known limitations this isn't possible. The 
different applications can tolerate different limitations of the 
DEM. The DEM used for orthophoto generation should be 
absolutely as exact as possible. For the hydrological applica- 
tions this is not so important, but those DEMS have to be hydro- 
logical correct. The flowing direction of the water has to be 
correct for each pixel, the absolute high of each pixel is not that 
important. These two different expectations to a good DEM 
leads to two different methods creating a DEM. 
For the first type of DEM, SPOT PAN Stereo Images have been 
used. Because of the variations in neighbouring heights, this 
DEM couldn't be used for hydrological applications, but for the 
generation of an orthoimage these small variations are not 
important. This DEM was generated using the Virtuozo? soft- 
ware package which was developed at the Wuhan University 
and is now being distributed by Supresoft Inc. 
Ground control pixels situated near water bodies made prob- 
lems in the DEM generation. This may be caused because of 
different water levels between the image acquisition time and 
the map production time. The other difficulty is that the water 
body areas haven't much texture in the remotely sensed data. 
Specular reflections do also occur in the water areas, making 
even more problems. Because of these problems the resulting 
DEM wasn't very exact and couldn't be used especially not for 
the hydrological applications. For further aspects on the accu- 
racy of DEM generation from SPOT data see (Al-Rousan et al., 
1998). 
   
  
s 
  
  
  
Figure 8. DEM generated from Stereo-SPOT images (left) and 
DEM generated with the TOPOGRID module 
(right) 
For the hydrological applications the flowing direction of the 
water is important and has to be correct. As shown in Figure 8 
the DEM created with the Stereo-SPOT images are not 
hydrological correct. In the middle of the left image of Figure 8 
the river would have to climb over a small ridge. 
To build an DEM which was hydrologically correct the TOPO- 
GRID module from Arc/INFO (Hutchinson, 1988) was used. 
For generating the DEM, TOPOGRID needs input data. The 
input data were in detail: polder areas, water bodies and river 
middle axes. The river middle axes were digitised manually 
from a topographic map. All rivers inside polder areas were 
removed since polder areas were decided to be treated as closed 
units. All river segments were oriented downstream, which is a 
prerequisite for the TOPOGRID module. 
The TOPOGRID module computes gradient directions perpen- 
dicular to the contour lines and then iteratively respects the 
oriented river network (Hutchinson, 1996). To generate the fi- 
nally used DEM, 2000 iterations were calculated. After the 
DEM calculation the digitised lake and polder polygons were 
used to level all DEM cells inside the same polygon to the same 
height. For the hydrological applications this again was an im- 
portant step, because this levelling ensures a flatness of lake 
and polder areas. The DEM generation was performed in close 
551