m Ce o
Area Feature
orge ca Part of
UHUT
m
Pa rea Entity
is-a t Part of
UST.
3 grid cells,
—\ dead areas
a
Nr
ucture for a DTM
' the presentation
Earth' surface is
Very often, a
iat can be derived
l For the points
nsists of n regions
y) € V(1) then pi
(pi, pj) is the half-
than to p;, then
pi, Pj)
pi, Dj)
d to derive the tri-
lines dual to this
. A defining prop-
at the circumcircle
her point in its in-
ph property of the
1gulation.
am
data structure of a
ded DTM delivers
Em
a similar pictogram (see fig.4 ). The advantage of an irreg-
ular triangulated network (TIN) compared with a gridded
DTM is taken from the fact that triangles of a TIN rep-
resent very close geomorphological features. This results
simply from the primary data acquisition process that dis-
cretizes the continous Earth’ surface mainly in those points
in which geomorphological characteristics change.
Zo Class Cures cess) died Class 3
P
c db ee
As Belong to fes Belong to ^ Belong to
S TT
N ‘
c int roe) e Feature À noi Frais
Repre re Represented We Part of Represented te Part of
AN BN be
Point Entity (im Entity v) C^ Area Entity
hii ba
a a^ ^e Part of °F Part of
= SRE CE Begin a left y
Triangle Node = Edge \T——— (rose)
ER End Cur right
AL
Xyz J
5 (N
AEN
Figure 4: Geometric-topological data structure of a TIN
Fig. 4 is derived from a fully geometric-topological data
structure. Further structures that define topological TIN
organizations according to triangles and edges, respect-
ively, are given by D. Fritsch (1991). :
2.1 Layer-oriented DTM in GIS
The layer approach is the oldest data structure that
emerged from the superimposition of various analog maps.
The link of the DTM with planimetric features is real-
ized by the coordinate reference system, which has to be
identical for the layers superimposed with each other. This
superimposition should not be called ’ integration ' because
no real integration step is carried out.
Contrary to the primary DTM interpretation problem there
is no doubt that the layer approach of DTM derivates in
vector and raster form, for instance contour lines, slope
and aspect information, shaded reliefs etc. delivers an ad-
equate data storage model. In some cases, the superimpos-
ition of a few derivates comes out with new information in
the sense, that the terrain can better be interpreted than
before, and spatial analysis is performed in an excellent
manner.
2.2 Fully integrated 2.5D data model
The integration of DTM data structures with their plani-
metric counterparts is also dealt with in P. van Osterom
et al. (1994), M. Pilouk/O. Kutoniyi (1994) and K. Kraus
(1995). Above all, the integration consists of a link between
planimetry and topography. That means, a geomorpholo-
gical feature should have a counterpart in planimetry, but
must not necessarily. The way of implementation is not to
be defined in a strict sense, as also shown in the section of
open system architectures.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
Point Class
Belong to
Point Feature
Reprasanted
by
Area Feature Line Feature
Represented Represanted Pert of
by by
Coordinates
£x, y, 2)
Figure 5: 2.5D data structure of integrated TIN (Copy-
right/Courtesy: M. Pilouk/O. Kutoniyi, 1994)
The proposal of M. Pilouk/O. Kutoniyi integrates a TIN
DTM and multitheme geoinformation as represented by
fig. 5 in a most rigorous approach. In their model, terrain
features are classified into three geometric types: point (0
simplex), line (1 simplex) and area (2 simplex) in each
mapping theme. In addition, features are grouped into
mutually exclusive thematic classes in each layer. These
classes are simply represented by class labels. The type to
which a feature belong will be decided during the imple-
mentation step. For the representation of terrain features,
point, line and area entities are used. The same data struc-
ture is worked out by K. Kraus (1995).
The implementation of the integrated 2.5D data structure
can be realized in a purely relational scheme (M. Pilouk/O.
Kutoniyi, 1994). Only eight normalised relations are ne-
cessary.
R1: AREA (a id, af id, layer, a name, a, class)
R2: LINE (1_id, lf id, layer, 1l. name, l1_class)
R3: POINT (pid, pf .id, layer, p.name, p.class)
R4: PNODET (p.id, p.node)
Rb: ARC (arcjnr, beg, end, l.tri, r tri)
R6: NODE (node-nr, x coord, y coord, z coord)
R7: ARCLINE (arc nr, al. id)
R8: TRIANGLE (tri nr, ta id)
These relations are derived from six dependency state-
ments. The data types and the link types serve as field
names in the final relational structure.
3 THREE-DIMENSIONAL DATA
MODELS
Three-dimensional data models often refer to spatial data
structures used for mapping of nodes, edges, faces (areas),
and volumes. Applications can be found in the geosciences
in which particularly solid bodies have to be modelled,
analyzed and visualized.
Volume modelling technology and its integration with vari-
able prediction provide a range of options for performing
more precise volumetrics analyses. This includes determin-
ation of the volume of any irregular shape, or the volume of
217
