International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
be covering the whole of Germany with a Universal Trans- 
verse Mercator projection (Gauß-Krüger) in the third strip 
(9° East) and will have a consistent grid width of 50 m. 
Table 2 lists the quality and the grid width of the input 
datasets of the states. Because of data acquisitions being 
still in progress there are datasets of lower quality in the 
list. This data will be replaced with precise elevation data 
during the next year. 
  
  
    
  
  
  
  
  
  
  
  
  
  
  
Quality | Accuracy | Grid width | Percentage 
DEM5 +0 5m = 5-10 m 24 
DEM10| +1,0m = 10-20 m 6 
DEM25| +2,0m = 20-50 m 54 
DEM50| +4,0 m = 50m 16 
  
  
  
  
  
Table 2: DEM Quality delivered from the states to build up 
the DEM25 
42 QUALITY CONTROL 
Quality control of the incoming datasets is another task of 
BKG. For this inspection a method was developed (Hoven- 
bitzer et al., 2004), which was based on the DGPS techno- 
logy SAPOS. The height information is tested on site by 
walking or driving through the test areas. When a point is 
measured the survey controller of the GPS receiver calcu- 
lates the difference to the DEM or DTM data in real time, 
so that the field staff can evaluate the correctness of the 
tested dataset. 
5 PRACTRICAL EXPERIENCE 
For varying reasons there are differences in the overlapping 
areas of the DEM/DTM datasets. The main causes are : 
e Different dates of data acquisition 
e Different methods of data acquisition 
e Different grid widths of the data 
e No use of morphological information for building up 
a DEM 
Figure 2 shows a garbage dump, which was measured at 
two different times. The contour lines in the figure depict 
the ongoing dumping of waste in the north of the garbage 
dump. To solve this difference problem we have to cut off 
the older measuring points, so that the current data will 
appear in the final dataset. 
In figure 3 great differences in a quarry are determined. 
The reason is again a different date of data acquisition. Be- 
cause of the digging in the quarry the surface has changed 
between the two dates of data acquisition. 
A narrow valley is shown in figure 4. The reason for the 
differences in this case is a different grid width of the two 
overlapping datasets. The raster points of the grid with the 
lower spacing cannot model this narrow valley correctly. 
   
    
3 
  
Figure 2: Contour lines of differences in an overlapping 
zone of two states (garbage dump) 
  
  
Figure 3: Contour lines of differences in an overlapping 
zone of two states (quarry) 
  
  
   
7 
“A 
Figure 4: Contour lines of differences in an overlapping 
zone of two states (different grid widths) 
1242 
  
Inter 
  
The 
fore 
qual 
renc 
in pi 
aceti 
DE 
sed. 
For 
plen 
accu 
ons. 
area: 
REF 
AdV 
http: 
Hove 
DGN 
des E 
Fran!