DTM Representation
Once DTM information has been collected, it may be repre-
sented in a variety of compressed or non-compressed formats.
The terms format, representation, and structure are used
interchangeably here. Keep in mind that most DTM produc-
tion systems today collect only elevation information,
either as discrete spot heights or line profiles. DTM's,
however, may be represented by elevations, contours,
tessellated polygons, surface polynomials, fractal numbers,
slope maps, power spectra, convolved displays, etc. Each
representation is some approximation to the data collected,
and is done to satisfy the storage and input requirements
of the DTM user. |
ACCURACY CONSIDERATIONS
The term accuracy related to DTM's means many things to
different people. It would be appropriate to define
accuracy in terms of geomorphological quality, precision
of information, positional accuracy, data commonality, data
compatibility, and compression (representation).
Geomorphological Quality
The geomorphological quality of a DTM is the degree to
which the DTM represents the actual landform. This concept
is extremely difficult to define quantitatively. It con-
siders all of the statistically measurable quantities of
the DTM as well as the visually apparent anomalies, texture,
and fit to the actual landform. Forstner (19835) presents
a good discussion on sampling interval and form of data
collected on the quality of a DTM. He presents evidence
that demonstrates that slope and curvature information are
extremely important in the quality of a DTM. Note however
that most DTM production systems do not collect this in-
formation.
Faintich et. al. (1982) demonstrate that variations in
production such as different types of equipment, source
scale and type, etc., produce a variety of textures and
anomalies in DTM's when the pieces are assembled over large
cells of data. Although such DTM's may meet numerical
accuracy requirements, the overall quality of the DTM is
affected by the non-uniform appearance.
Precision of Information
The precision of the data described by a DTM is a statistical
representation of the primarily random error or noise in
the model. Precision is generally determined by statistical