ESF |
ugh
rec-
but
e in
n all
)ec-
uch
ner-
sts:
3go-
[hat
cal-
jints
yoint
ob-
ated
nore
tion
M of
from
the hill-slope. This is contrary to the results Steinmetz
has found out for the program system SCOP, where
there is a very strong dependence between the corre-
lation coefficient and the flatness of the terrain, yield-
ing correlations of nearly 1 in flat areas (Steinmetz,
1992).
o Foran experienced operator there does not seem to
be a dependence between the correlation coefficient
from the original data (obtained from an analytical
plotter) and the hill-slope up to a hill-slope of 5096.
5 CONCLUSION AND FINAL REMARKS
The influence on the correlation between the elevation
errors seems mainly to be an effect of the interpolation
technique, when static profiling is used on an analytical
plotter. On the other hand, precisely a high correlation of
the elevation errors in flat areas is necessary in order to
obtain a smooth surface. Falling short of this, the results
are crossing slope vectors or noisy hill-shading represen-
tations in flat areas. On the other hand well-designed low
pass filters just pretend a higher accuracy through the
higher correlation; that is one reason for the small correla-
tion of the height errors when compared to a data set of
superior quality as done here, even when using low pass
filters.
Quality information for the DEM frequently is given through
the RMS error of point elevations. This RMS error is still a
very good means to describe a DEM's quality, since it
takes into consideration the deviations from the smoothed,
"artificial" surface which is often wanted from DEM- and
GIS-users for more appealing products. It should, however,
kept in mind that this error is not really an "error", but ra-
ther a wanted deviation from the real surface. So the RMS
error is partly a real error, partly the result of modelling an
approximate but smooth surface. The small correlation of
the height errors especially in flat terrain as shown in this
paper partly reflects the roughness of the terrain against
the artificial (i. e. modelled) surface.
Nevertheless there is a need for as exact surface repre-
sentations as possible, especially in mountainous areas, in
order to model environmental hazards like river floods or
avalanches. For that purpose further investigation work will
be done in order to check the behaviour of manual mea-
surement from aerial photographs (on an analytical plotter)
as well as image matching techniques in dependence from
hill-slope, especially in steep areas where there is a strong
need for quality control of the slope. This leads very close
to the question how to define the surface, which certainly
has to be done specific for the respective application:
There is still a lot of effort to be done in order to define the
real demands of the user on the DEM.
It should be a task of photogrammetry to provide the user
not only with realistic estimations for the quality of its pro-
ducts but also to consult him in learning about his real
needs.
REFERENCES
Ackermann, F., 1979. The Accuracy of Digital Height Mo-
dels. Proceedings of the 37th Photogrammetric Week,
Stuttgart 1979.
Goodchild, M., Gopal, S. (Eds), 1989. The Accuracy of
Spatial Databases. Taylor & Francis.
695
Hochstôger, F., 1995 Verwaltung landesweiter Gelände-
hôhendaten. Proceedings of the AGIT, Salzburg, Austria.
Hutchinson, M. F., 1989. A new procedure for gridding
elevation and stream line data with automatic removal of
spurious pits. Journal of Hydrology, Vol. 106, pp. 211-232.
Kóstli, A., Sigle, M., 1986. The random access data struc-
ture of the DTM program SCOP. In: International Archives
of Photogrammetry and Remote Sensing, Edinburgh, Vol.
26, Commission IV, pp. 45-52.
Kraus, K., 1994. Visualization of the Quality of Surfaces
and their Derivatives. PE&RS, Vol. 60, pp. 457-462.
Kraus, K., 1995. From Digital Elevation Model to Topogra-
phic Information System. Proceedings of the 45th Photo-
grammetric Week, Stuttgart, pp. 277-285.
Kraus, K., Haussteiner, K., 1993. Visualisierung der Ge-
nauigkeit geometrischer Daten. GIS, Vol. 6, no. 3, pp 7-12.
Krzystek, P., 1991. Fully Automated Measurement of digital
Elevation Models with MATCH-T, Proceedings of the 43rd
Photogrammetric Week, Stuttgart, pp. 203-214.
Makaroviè, B., 1972. Information transfer in reconstruction
of data from sampled points. Photogrammetria, no. 4.
Mark, D. M., 1979. Phenomenon-based data-structuring
and digital terrain modelling. Geoprocessing, Vol. 1, no. 1,
pp 27-36.
Rieger, W., 1992. Hydrolgische Anwendungen des digita-
len Gelándemodelles. Dissertation. Geowissenschaftliche
Mitteilungen, Heft 39, Technical University Vienna.
Stechauner, A., Ehartner, M., 1988. Praktische Móglich-
keiten für die Bewertung der Bodenerosion in Osterreich.
Osterreichische Zeitschrift für Vermessungswesen,
Steinmetz, D., 1992. Etude de précision du modéle numéri-
que de terrain SCOP. Memoires de soutenance de diplóme
d'ingenieur. Ecole Nationale Superieure des Arts et In-
dustries de Strasbourg.
Tempfli, K., 1982. Genauigkeitsabschátzung digitaler Hó-
henmodelle mittels Spektralanalyse. Geowissenschaftliche
Mitteilungen, Heft 22, Technical University Vienna.
Theobald, D. M., 1989. Accuracy and bias issues in sur-
face representation. The Accuracy of Spatial Databases.
Goodchild, Gopal (Eds), Taylor & Francis.
Toomey, M. A. G., 1988. The Alberta Digital Elevation
Model. In: International Archives of Photogrammetry and
Remote Sensing, Kyoto, Japan, Vol. 27, Comm. Ill, pp.
775-783.
Zhang, Y., 1988. Accuracy Estimation of DTM Using High
Sampling Density Profiles. In: International Archives of
Photogrammetry and Remote Sensing, Kyoto, Japan, Vol.
27, Part B3, Comm. Ill, pp. 857-864.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996