TERRAIN MORPHOLOGY MODELLING
Massoud Sharif
Alfred Zinck
International Institute for Aerospace Survey and Earth Sciences (ITC)
P.O. Box 6, 7500 AA Enschede, The Netherlands
Tel. +31-53-4874345, Fax. +31-53-4874335
E-mail:SHARIF 9 ITC.NL
Commission Ill, Working group IV
KEYWORDS: morphologic feature, geometric primitives, Selective modelling, semi automatic modelling, skeleton
information, filling information, transfer function, rule base.
ABSTRACT
Terrain morphology can be described in terms of morphographic and morphometric attributes. Morphographic attributes refer
essentially to the geometry of the geoforms, including shape and profile of the topography, aspect, configuration and contour
design of the forms, and drainage pattern. Morphometric attributes refer to the dimensions of the geoforms, including relative
elevation, valley density and slope steepness. Both morphographic and morphometric attributes of geoforms can be derived
from a topographic map by visual interpretation or from a DTM by either visual or automated procedures. The first issue
concerns the acquisition of terrain morphologic features either by photogrammetric selective modelling or by means of an
automated process. Selective procedure is a highly subjective one which calls for an objective automated method using
digital terrain modelling (DTM). The second issue concerns the quality assessment models. Different approaches allow to
assess the accuracy of modelling. They deal mostly with the terrain morphologic representation in terms of statistical
consideration, namely, the standard deviation of the discrepancies. However in some applications, it may be more relevant
to assess the fidelity of the terrain representation in relation to the real terrain morphologic features. In this context, the
mathematical background of an analytical approach is discussed, and a model to assess the quality of the representation
is presented. The third issue refers to the determination of an optimum model for grid densification. Idealized simulated
terrain primitives were used for that purpose as well as for extracting selective modelling rules. Morphologic modelling was
carried out using the primitives and applying different grid densities to determine an optimum model for grid densification.
The performance of the optimum digital morphologic model was verified using not only geometric primitives but also real
terrain morphology. The effects of the skeleton information on terrain morphologic features were analyzed, and the rules
drawn from idealized geometric primitives were tested on a real terrain using a moderately rough terrain model. The analysis
of the test results provides a feedback for optimizing the procedure to generate an optimum terrain morphologic model.
INTRODUCTION Optimum representation — combines selective with
semiautomated representation. The aim is to portray terrain
Terrain morphology modelling is the process of extracting morphology faithfully, without excessive redundancy of the
and representing the spatial location of morphologic points, presented information. For that purpose, an expert system
lines or features on the terrain surface. This requiresthe for optimum representation of morphologic terrain features,
fulfilement of three basic objectives: (a) the acquisition of integratedin a GIS was developed. It is practically
terrain morphologic features either by photogrammetric impossible to represent either the global terrain surface in
selective modelling or by an automated process; (b) the an exact mathematical expression ( approximateld by a
assessement of the modelling quality in terms of statistical reference ellipsoid) or the macro-reliefs (via different
accuracy but also as tothe fidelity of the terrain polynomials or via superimposition of sinusoids with
representation in relation to the real terrain morphologic variable amplitudes and frequencies, etc ...). Therefore
features; and (c) the determination of an optimal model for ^ we have decided to simulate the terrain morphology by
grid densification using terrain geometric primitives and computer generation of ideal geometric primitives, and the
expert rules. These three issues are discussed in the combination of those primitives. These are the geometric
present paper. primitives for which, on one hand, a mathematical definition
is possible and, on the other hand, the assimilation to
1. TERRAIN MORPHOLOGY REPRESENTATION terrain morphologic features is probable. These are: semi-
spheroidal surface, semi- ellipsoidal surface, conical
The representatin of morphologic terrain features can be surface, gaussian surface , parabolohyperboloidal surface,
performed either by ^ manually controlled selective fault , ridge, and any combination of those.
representation of those features, referred to as the skeleton
information (2), or by means of semi-automated The simulation of terrain morphology via geometric
representation of more homogeneous morphologic features, primitives allows us to know the input signal to the
called filling information (I1). processing and evaluate the error of terrain morphology
792
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
terr
ver
pro
rea
Bas
in t|
Thi
Ba:
ske
The
figu
Be
sys
ser
sor
The
sel
are
exr
prir
