hey
ons
ate
noi
by
nes
the
and
the
ode
cial
ode
fied
ype
oth
is a
ach
our
lo-
pri-
ated
dif-
ent.
rom
g is
ring
t by
t in-
, the
eak.
dial
the
vail-
le to
the
half equidistance) in case of a pit or (H1 + the half
equidistance) in case of a peak. Other points on the
medial axis get adjusted values between the middle
point and the concerned contour point.
For type (2): The selection of the saddle point follows
the symmetry criterion, too. The saddle point gets
then the mean value of H1 and H2. Other points get
adjusted values as described above.
For type (3): If H1 < H2 holds the geomorphological
element represents a ridge line else a drainage line.
The elevation assignment by a linear interpolation be-
tween H1 and H2 is a simple and also widely accepted
way.
7
(NC
— A —
\
NN
zZ
Figure 8. Geomorphological elements from contours.
4. DIGITAL TERRAIN MODELLING USING
RASTER ALGORITHMS
Contours are often used as reference data for DTM
generation. A HQ-DTM from contours means that
the given contours ought to be restored from it on the
one hand and the intermediate contours derived from
it should reasonably be shaped on the other hand. The
former can be guaranteed by the constrained triangu-
lation, and the latter requires, however, geomorpho-
logical elements for assistance.
At first, geomorphological elements are derived by
means of the medial axis approach described above. A
TIN-DTM is then generated by the raster-based trian-
gulation of the given contours and the derived geo-
morphological elements. Finally, various follow-up
products can be derived from the TIN-DTM.
To evaluate the presented approaches two practical
examples are given in the following:
571
Example "Thalham": It covers an area of 400x400
square meters on the ground. The contours with a
constant equidistance of 2 meters were manually di-
gitized from a topographic map of a scale of 1:10000.
The total number of contour points amounts to 1168.
The given contours and the derived geomorphological
elements are shown in Figure 8. Figure 9 is the TIN-
DTM.
Example "Koralpe": It covers an area of 1480x1620
square meters on the ground. The contours with a
constant equidistance of 10 meters were manually di-
gitized from a topographic map of a scale of 1:10000.
The total number of contour points amounts to 4252.
Figure 10 (a) shows the given contours and the
ME N A Us EEE
SC
> a
SSID
cl
[2
v
J)
I
7
4
A
—— SZ
NJ
Dax SN Xe - | A S6 /
N tés 7 E
> AY, > SM M NN CLP \/ \/ NZ /\
(SAS [^ 7 > - N SC XN [Se 4
S X
N
N
N
uM
7
L
e]
U
av», A
AZ
i]
LA
2
%
A
CN
4
7
À
S
[/ y
a
MN 7
b
NN
ce
&
S
x
CP
~
NS
VAN
V
VAX
E
N
=
0)
e
NI
N
V
7
AT
N
+ "wv
b) NN
«
N
h.l
A
ho +,
RA
X
Ss}
1%
Ss
A
NOS
N
NJ IA
eS
LS
v4
[/
zs
aS
RY
[29 71
d
"lol 12
SS
LT
a
SR
CHE
LCS
A
A
7
LS
SA
EE
C
Ci
/A
> Y A7
0 N SKA ?-- dx ON
FS IS
SSSR A SSI (ESA
BESSER DO
NASA P PL A
Ded D
STN DA CEE AUS
P ce eta >< SAT
1
DS
NS:
NS
NON
S
5 A ES
D L^ VA
x
N
D
\
N
EN
pas
CV
Figure 9. Example "Thalham": the TIN constructed
from the given contours and the derived geomorpho-
logical elements.
derived geomorphological elements. Following the
procedure described above, a TIN-DTM was gener-
ated from these data at first. It was then converted
into a raster DTM by a planar interpolation of the
triangular facets. The 10-meter contours and the 2.5-
meter intermediate contours were derived from the
raster DTM (cf. Figure 10 (b)). The improvement of
the DTM quality is evident.
With regards to the computational complexion the
raster-based triangulation was compared with an
existing vectorial algorithm for the constrained Delau-
nay triangulation. The comparsion was carried out on
a Hewlett Parckard graphics workstation HP 9000/350
CHX, which is equipped with a MC68020/25MHz
CPU, a MC68881/20MHz floating point processor
and 8 MBytes main memory. In addition, a test on the
computational time for a triangulation without con-
