Geography Reference
In-Depth Information
they can be classified by their approach into three
groups: the parameter method, the stochastic
historical method and the triggering threshold
method.
activity. This is both an advantage, because it does
not rely on sample values, and a disadvantage,
because to be of use in prediction, it assumes
temporal stability in causative factors.
Above all, the method demands a good
database. Unfortunately, few countries have
standard protocols and procedures for establishing
reliable records of landslide activity. Consequently,
in employing this approach, much original
research of information sources is required,
including information from media sources, public
organisations, private consultants, etc. The record
can be extended beyond the historical period by
investigating landslide deposits in the geological
record. For example, lake sediments in an unstable
area of New Zealand have revealed the occurrence
of 395 landslide events in the last 6000 years (Eden
and Page 1998). By dividing the number of events
by the period of observation, a historical
frequency (probability of occurrence) can be
determined for different regions.
Parameter method
The parameter method requires a knowledge of
the type, distribution and effectiveness of causative
factors for different components of the terrain.
The choice of which factors to investigate is
determined from prior knowledge or by
discriminating between the factors associated with
stable and unstable terrain (Gee 1992). Commonly,
the initial mapping or investigation units are areas
homogeneous for important stability factors such
as geology or slope angle. These may be analysed
subsequently and perhaps subdivided by other
stability factors. Experience can be used to provide
a semi-quantitative weighting to stability factors
(Sinclair 1992), and summed values can be
obtained for ranking each class or areal unit.
Usually, factors indicating the presence and
activity of any existing landslides are weighted
heavily as indicators of the degree of hazard. GIS
are of particular value as a tool for analysis,
synthesis and computation within the parameter
approach to hazard assessment.
Whereas the parameter method is the most
common form of regional hazard assessment, it
provides only a ranking of susceptibility—not true
hazard. By itself, the method does not provide any
indication of the probability of occurrence or the
magnitude of landslide to be expected. Two
examples of the parameter method are presented
in Box 6.2.
Triggering threshold method
The triggering threshold approach is more
complex than other methods but has greater
potential for forecasting landslide activity and
determining the mass movement response to
climatic change and other triggering factors. This
approach couples the forcing process with a
process response. For rainfall-triggered landslides,
this involves establishing an initiating threshold
between rainfall parameters and landslide
occurrence (Julian and Anthony 1994; Crozier
1989) (Figure 6.3). For earthquake-triggered
landslides, the threshold between non-occurrence
and occurrence is usually either a function of
shaking intensity or earthquake magnitude (Keefer
1984).
Thresholds established in this way simply
measure the susceptibility of the terrain under
study to the landslide-triggering process. Clearly,
inherent stability conditions and consequently
thresholds will vary from place to place. A reliable
regional threshold, however, may be used to
determine the probability of occurrence (statistical
frequency) of landslide activity by reference to the
Stochastic historical method
One of the main difficulties with both stability
analysis and the parameter method is the difficulty
in obtaining accurate and representative values for
the parameters involved. This is because many
stability factors exhibit a high degree of spatial
variability. The historical method, on the other
hand, works on the principle of 'precedence',
indicated by the record of previous landslide
