You are on page 1of 5

Geomorphology, 5 (1992) 213-217

213

Elsevier Science Publishers B.V., Amsterdam

Limitations of the system approach in geomorphology


A.E. Scheidegger
Technische Universitiit Wien, Institut J~r Theoretische Geod~ie und Geophysik, Abteilung Geophysik, Gusshausstrasse 27-29,
A-1040 Wien, Austria
(Received August 13, 1991; revised December 23, 1991; accepted January 6, 1992 )

ABSTRACT
Scheidegger, A.E., 1992. Limitations of the system approach in geomorphology. In: J.D. Phillips and W.H Renwick (Editors), Geomorphic Systems. Geomorphology, 5:213-217.
A system is defined as a set of interrelated elements which function together as a whole. For a statistical-mechanical
application of system theory, it is assumed that the number of elements is large and that the interrelations are complex so
that the individual interactions cannot be followed individually in detail. Under such circumstances, the individual processes can be considered as quasi-random, i.e. as if they were stochastic and, by the use of the limit theorems of probability
theory, statements can be made regarding the expected behavior of the system. This approach, however, has its limitations:
the assumption that the individual processes are quasi-stochastic does not necessarily hold. There are many instances when
the individual processes are uniform and correlated; in this case, they cannot under any circumstances be considered as
quasi-random. This applies in geomorphology mainly in connection with structural and (neo)tectonic predesign. The
application of system theory in such circumstances leads to faulty results.

Introduction
During the last 30 years, the system approach has become very popular in geomorphology: all kinds of geomorphic features have
been regarded as systems, and the rules of complex system dynamics and synergetics have
been applied to them. In this it was often forgotten that system dynamics are based on a set
of rather specific assumptions which may or
may not apply in connection with specific geomorphological problems. Thus, the aim of the
present paper is to review some of the basic
principles o f system dynamics and of landscape evolution, and then to compare the former with the latter. It will be seen that not all
geomorphic features can be treated by system
Correspondence to: A.E. Scheidegger, Technische Universit~it Wien, Institut ftir Theoretische Geod~isie und
Geophysik, Abteilung Geophysik, Gusshausstrasse 27-29,
A- 1040 Wien, Austria.

theory, but that some definitely have to be dealt


with by other methods.

Principles of system theory


Concept of a system
The concept of a system has been introduced by Bertalanffy ( 1932 ). Accordingly, in
order to speak of a system, one needs
(1) a set of elements identified with some
variable attributes,
(2) a set of relationships between attributes,
(3) a set of relationships between attributes
and the environment.
Thus, a system is a set of interrelated elements
which function together as an entity embedded
in an environment. The last condition above
assumes that the system is open to some external environment. In effect, the distinction of
where the system proper ends and the environ-

0169-555X/92/$05.00 1992 Elsevier Science Publishers B.V. All rights reserved.

214
ment begins is arbitrary. A natural delimitation occurs only ifa system is completely closed
within itself. However, most geomorphic systems are open in the sense that mass and energy enter at one instance, cascade through the
system and exit at another instance.

Quasi-stochastic approach
Formal system theory is, in fact, a mathematical discipline. Generally it is involved with
statistical methods. In common usage, the term
"system" is often taken more loosely: a "system" is simply represented by the interplay of
some parts, such as a hi-fi "system" which is
represented by the interplay of a CD-player, an
amplifier and loudspeakers. This loose concept, however, does not lend itself easily to formal analysis. The usefulness of formal system
theory is evident if the number of elements in
the set is large and the relationships between
their attributes complex. By "large" and
"complex" is meant that it is no longer feasible
to consider each element and every relationship separately and to calculate explicitly the
combined effect of these relationships on the
behavior of the whole.
In geomorphology, one deals essentially with
mechanical systems. The elements are mechanical entities (particles, river courses, slope
angles, etc. ) which interact in a complex way.
The state of the system is then defined by giving all the attribute-values of all the elements;
the set of all attribute-values can be represented by a point in a multi-dimensional phasespace. During the evolution of the system, this
point will describe a trajectory in phase space.
As noted, it is assumed that the number of
elements is large and that the attribute values
of each element can be ascertained in detail.
Therefore, there is a quasi-probability distribution of phase points which represents the
position likelihood of the system in phase space
within the limits of one's ignorance. This probability distribution can be taken as the basis for
statistical predictions of the behavior of the

-x.~.SCHEIDEGGER
system. Within the limits of one's ignorance, a
whole ensemble of states is possible. Because
of some fundamental natural laws, e.g. conservation of mass (or energy), not all thinkable
states in phase space are possible, but these may
be restricted to certain regions.
An observable quantity is generally a gross
characteristic built on a conglomerate function
of attributes. The expectation value for this
observable quantity is the average of the conglomerate over all states of the system that are
possible. On occasion, the conglomerate function has been calculated for those attributes
that correspond to the most probable state of
the system, but this is not in conformity with
the principles of statistical physics as they were
developed by Boltzmann and Gibbs in connection with gas dynamcs (see e.g. Sommerfeld,
1964 ). Thus, "most probable" and "expected"
characteristics are not the same; it is logically
evident that ensemble averages have to be
taken of the attributes and not those attributes
for the most probable state of the system.
The approach presented above has been
called quasi-stochastic. The word "quasi" indicates that the evolution of a large system (e.g.
landscape ) is not in reality a stochastic process
at all, but a well determined mechanical one.
This is also the case tbr thresholds in systems
where one or more parameters possess one or
several critical "bifurcation" values at which
the structure changes of the control parameter
lead to further hierarchically arranged bifurcations which may be run through until a quasistationary state of complete chaos is reached
(Schuster, 1984; Brun, 1986; Harrison and
Biswas, 1986 ). The latter is also, in principle,
completely defined; however, the knowledge of
the details of the processes is so incomplete that
it is convenient to treat them, within the limits
of one's ignorance, as if they were stochastic.

Equilibrium conditions
In equilibrium conditions the basic probabilities are stationary. In phase-space, this is

215

LIMITATIONS OF THE SYSTEM APPROACH IN GEOMORPHOLOGY

expressed by the confinement of the phase


points to certain fixed regions. If there is a constant of the motion H, this region is represented by the subspace that has the equation:
H = const.

( 1)

If the interactions are many and complex, the


phase point will move in time; because we suppose equilibrium conditions, all points of the
subspace will, in time, be equally closely approached by the phase-point: this is the ergodic theorem embodying the idea that the
time averages can be replaced by ensemble averages (cf. e.g. Sommerfeld, 1964).
If the system consists of a large number of
subsystems which fluctuate and contribute
part-values H to the constant of the motion H,
then the distribution P of these values in canonical (cf. e.g. Sommerfeld, 1964) form is:

Pi(H~) = ( 1/Z) e x p ( - H i / k T )

(2)

dynamic variables led to valid statements


regarding landscape development. The statistical justification of these analogies lends a
much better foundation to the latter.
An extension of the equilibrium case is to the
steady state, for which the entropy production
rate a must be a minimum:
a=-

Ih'

~TJ d x = m i n

(4)

where J is the mass flux per unit time. From


the above equations, it is possible to calculate
equilibrium river profiles, etc.
The equilibrium theory immediately leads to
the process-response concept: as soon as an
equilibrium is disturbed, the system responds
by the adjustment of the remaining variables
to a new equilibrium configuration.

Nonequilibrium conditions

where Z is the partition function required for


normalization and k T a parameter. This is a
consequence of the limit theorems of probability theory under the assumption that the interactions between subsystems are many and uncorrelated. On this basis, it is possible to set up
a complete analogy between thermodynamics
and landscapes; one can define analogs of all
the thermodynamic functions (notably the entropy) and corresponding landscape variables
(Scheidegger, 1967). It is clear that the constant of the motion is mass (this is conserved
in the landscape process), the temperature
equivalent is the topographic height h, and for
the entropy analog S one has:

The analogy can be extended to nonequilibrium cases. If the (large) system possesses a
large number of subsystems in which a nonnegative quantity is transferred by a statistically fluctuating transfer process whose exact
nature is unspecified, then the central limit
theorem of probability theory leads to a diffusivity equation for the non-negative quantity,
provided the fluctuations are linearly additive
(Tomkoria and Scheidegger, 1967). This immediately leads to an understanding of the
degradation of landscapes.

d S = dM/h

Quite apart from the discussion above, landscape evolution has been described by a set of
heuristic fundamental principles (Scheidegger, 1987 ); these will be briefly described.

(3)

The analogy between temperature T and


topographic height h and the corresponding
definition of"geomorphic" entropy S had been
postulated earlier by Leopold and Langbein
( 1962 ) on entirely heuristic grounds, from the
analogy with a heat engine. These authors observed that the thermodynamic principle applied to the landscape analogs of the thermo-

Landscape evolution

Principle of antagonism
The most fundamental of the landscape
principles is that of the antagonism of endogenic (originating inside the solid Earth, i.e.

216

tectonic) and exogenic (originating outside the


solid Earth ) processes. In this, the condition of
the surface of the Earth represents the instantaneous result of the action of these two types
of processes: uplift (tectonic) and degradation
(exogenic).
The intensity of the antagonism determines
the character of the landscape: if the upliftdenudation rate is millimetres per year ( m m /
a), one has a youth type, if it is 0.5 ram/a, a
mature-type, and if it is < 0.1 m m / a , an old
age type landscape. The primarily endogenetically caused features are morphologically systematic, the exogenetically caused ones are
morphologically random (Scheidegger, 1979 ).

A.E. SCHEIDEGGER

sists of flat-steep-flat elements. This is a consequence of the instability principle, inasmuch


as the erosion rate increases with topographic
gradient (positive feedback) leading to a characteristic deviation from uniformity.

Selection principle
A principle of a completely different nature
is the selection principle (Gerber, 1969). It
states that in exogenic processes, the statically
stable formations are selected for preservation. One has here a directivity in the erosive
action which is toward stable forms.

Principle of tectonic control


Principle of instability
The principle of antagonism describes important features of landscape evolution. However, it does not act alone. The antagonistic action of endogenic and exogenic processes
effects a fundamental instability of a landscape. Thus, we arrive at a principle of instability. This instability expresses itself in two
ways: first, many individual landscape elements are impermanent, although their general character appears as permanent. The bestknown example of this type is the meandering
of rivers: the meanders change constantly, but
the meander character stays essentially the
same. The second aspect of the instability is
that the individual landscape elements are not
only impermanent, but also remove themselves from the state of uniformity: straight
rivers become curved, brooks end up as a sequence of riffles and pools, valleys become
stepped. It is as if there were a positive feedback between a nonuniformity and its growth
(Scheidegger, 1983 ).

Finally, the principle of tectonic control


states that many landscape features are conditioned by deep-seated tectonic processes. Such
features are more c o m m o n than had been usually thought. Thus, the U-shape of glacial valleys (Hantke, 1978, p. 70), the orientation
patterns of river nets (Scheidegger, 1982, pp.
28-31) and the directions of landslides
(Scheidegger and Ai, 1987) are primarily
predesigned by endogenic processes and not by
exogenic ones as usually thought. It is therefore appropriate to elevate these observations
to a "principle of tectonic control" of landscape evolution (Scheidegger and Ai, 1986).

Applicability of system theory in landscape


evolution
Finally, one can investigate the applicability
of the system approach in landscape evolution,
the latter being described by the heuristic principles outlined in the last section.

Catena principle

The fundamental features of the system


approach

A further principle is related to the instability principle; this is the catena principle
(Scheidegger, 1986 ). It states that a slope con-

When we review the system approach we


note that the most important feature of the latter is that there are many elements related by

217

LIMITATIONS OF THE SYSTEM APPROACH IN GEOMORPHOLOGY

complex relations. All of system theory, then,


is essentially based on probabilistic considerations. For this, one needs indeed many elements and uncorrelated relations so that averages can be taken and the assumption of quasistochasticity holds. If these assumptions, many
elements and quasi-randomness, are not correct, the system approach leads to faulty results.

Thus, the usefulness of the system approach


is principally limited to purely exogenic processes; regarding endogenic processes the underlying tectonic processes must be described
by explicit deterministic mechanics. It is the
main argument of this paper that the systematic endogenic predesign in geomorphology is
far more widespread than commonly assumed.

Principles of landscape evolution and system


theory

References

We shall now review the fundamental landscape evolution principles in the light of a possible correctness of the basic assumptions of
system theory.
Evidently, the principle of antagonism applies to a landscape consisting of many elements that are complexly related. So does the
principle of instability and the catena principle.
However, the selection principle embodies a
directionality based on well-defined static considerations: certain forms are preferred whose
configurations have nothing to do with random interactions. This is even more the case
in connection with the principle of tectonic
predesign. Here, tectonic conditions, such as
the intraplate stress field, act on large numbers
of landscape points and this action is anything
but uncorrelated and quasi-random. In principle, the system approach cannot be expected to
work in connection with "structural landscapes", or "tectonic landscapes": Tectonism
is well-ordered and acts over dimensions that
are of the order of those of the tectonic plates.
Thus, in earlier statements of the principle of
antagonism (Scheidegger, 1979) it was noted
that the endogenically conditioned features are
not random. It is only the exogenic processes
that act in a quasi-random fashion. In effect,
this may only be a question of scale: water and
air also move in a completely mechanistic way,
but the uniformity scale in the motion is of the
order of cm; in tectonic processes, their order
may be several hundreds or thousands of km.

Bertalanffy, L.V., 1932. Theoretische Biologie. Springer,


Berlin, 170 pp.
Brun, E., 1986. Ordnungs-Hierarchien. Neujahrsblatt
Naturforsch. Gesellsch. Ziirich, 188. Stiick Orell-Fiissli,
Ziirich, 40 pp.
Gerber, E., 1969. Bildung von Gratgipfeln und Felsw~inden in den Alpen. Z. Geomorph. Supp., 8:94-118.
Hantke, R., 1978. Eiszeitalter, 1. Ott, Thun, 468 pp.
Harrison, R.G. and Biswas, D.J., 1986. Chaos in light.
Nature, 321: 394-401.
Leopold, L.B. and Langbein, W.B., 1962. The concept of
entropy in landscape evolution. U.S. Geol. Surv. Prof.
Pap., 500A: A1-A20.
Scheidegger, A.E., 1967. A complete thermodynamic
analogy for landscape evolution. Bull. Int. Assoc. Sci.
Hydrol., 12(4): 57-62.
Scheidegger, A.E., 1979. The principle of antagonism in
the Earth's evolution. Tectonophysics, 55: 7-10.
Scheidegger, A.E., 1982. Principles of Geodynamics, 3d
ed. Springer, Berlin, 395 pp.
Scheidegger, A.E., 1983. The instability principle in geomorphic equilibrium. Z. Geomorph., 27( l ): 1-19.
Scheidegger, A.E., 1986. The catena principle in geomorphology. Z. Geomorph., 30 (3): 257-273.
Scheidegger, A.E., 1987. The fundamental principles of
landscape evolution. Catena Suppl., 10: 149-210.
Scheidegger, A.E. and Ai, N.S., 1986. Tectonic process and
geomorphological design. Tectonophysics, 126: 285300.
Scheidegger, A.E. and Ai, N.S., 1987. Clay slides and debris- flows in the Wudu-Region and their tectonic implication. Sci. Explor. (Bejing), 7( 1 ): 253-264.
Schuster, H.G., 1984. Deterministic Chaos. Physik, Berlin, 220 pp.
Sommerfeld, A., 1964. Thermodynamics and Statistical
Mechanics. Lectures of Theoretical Physics, 5. Academic Press, New York, 401 pp.
Tomkoria, B.N. and Scheidegger, A.E., 1967. A complete
thermodynamic analogy for transport processes. Can.
J. Phys., 45: 3569-3587.

You might also like