International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
generally very reliable even in wooded areas but are frequently
missing or erroneous in the river valley and adjoining quarries,
both of which are open areas.
Both DEDSs are recalculated to a regular grid with 20m
posting. and the differences are analyzed.
^
| -9 |
-8 |
E44 |
-5 1
o 4 |
o 3 |
c 2 |
o A |
5 © |
1 |
= 2 |
© 3 |
- 4 |
+ 5 |
ES 5 |
o 7 |
8 |
L 9 |
10 |
»10 |
0 500 1000 1500|
# grid points |
Figure 3. Test area 1. Z-value of DSM minus z-value of DTM based on
contour lines.
For area 1, figure 3 shows a big deviation. The interesting
values here are the differences with absolute values larger than
10m. In both cases they indicate situations where one of the two
DEDSs is clearly not reflecting the terrain surface. Negative
values indicate areas where the DTM based on contour lines is
higher than the DSM. This is the case in quarries where contour
lines where falsely interpolated and to a certain extent it reflects
the lack of contour lines in the river valley which leads to a
DTM somewhat floating above the valley bottom. Positive
/alues indicate areas where the DSM yields values higher than
the DTM, generally indicating wooded areas.
<
-9
— -8
= =f
2 E
e. 3
9 7
o 0
1
E
3
ad 4
L 5
Ds
o 7
8
+ 9
10
>10
0 2000 4000 6000 8000
# grid points
Figure 4. Test area 2. Z-value of DSM minus z-value of DTM based on
contour lines.
For area 2 the average difference and variation are smaller than
in area 1. There are no pronounced peaks above the absolute
536
value of 10m. Both DEDSs here result in roughly the same
DTM .
The differences can also be stored in a rastermap for analysis of
the relation between differences and the topographical situation,
as depicted in the following figure.
t
Figure 5. area 1: Differences in height between DTM based on digitized
contours and DSM for area |. Red to yellow = DTM higher than DSM.
Green = DTM lower than DSM.
Figure 6. Orthophoto showing area |.
The two datasets are combined into a third grid by taking for
each gridpoint the lowest available value. This results in à
genetically hybrid DTM which reflects the terrain better than
either parent grid separately. In quarries and the river valley the
DSM-derived grid points reflect much better the actual terrain
Inte
hei;
the
alw
hyb
We
NC
dat
act
nec
fle;
dat