GIS-based Landslide Hazard Assessment - An Overview
GIS-based Landslide Hazard Assessment - An Overview
GIS-based Landslide Hazard Assessment - An Overview
548567
I Introduction
Landslides, defined as the mass movement of
rock, debris or earth down a slope (Cruden,
1991), can be triggered by various external
stimuli. These include intense rainfall, earthquakes, water-level changes, storm waves or
rapid stream erosion which cause a rapid
increase in shear stress or decrease in shear
*Author for correspondence. Tel., 81 774 38 4115; fax, 81 774 38 4300. E-mail: huabin-w@landslide.dpri.
kyoto-u.ac.jp
2005 Edward Arnold (Publishers) Ltd
10.1191/0309133305pp462ra
549
550
551
approach may not be suitable for studies concerning individual slopes or small areas while a
geotechnical engineering approach based on
the calculations of safety factor and/or associated failure probability would not be suitable
at the regional scale. Generally, landslide
susceptibility analysis methods used consist
of landslide distribution analysis, landslide
density analysis, landslide activity analysis,
geomorphologic analysis, qualitative map
combination and safety factor analysis. At a
regional scale, landslide distribution analysis,
landslide density analysis, geomorphologic
analysis and qualitative map combination
are used. At a medium scale, the relationship
between the landslide and contributing
factors is analysed statistically. Other methods used are landslide distribution analysis,
landslide activity analysis, geomorphologic
analysis and qualitative map combination.
At a large scale, safety factors analysis is
chosen for one main method to assess the
landslide hazard after the analysis of landslide
distribution, landslide activity and geomorphologic features. At a detailed scale, only
safety factor analysis is suitable to the landslide hazard evaluation.
For the geographic scale considering the
purpose of landslide assessment (Luzi and
Pergalani, 1996), the detailed scale is mainly
for the companies or municipal agencies
dealing with hazards on individual sites with a
maximum area of several hectares. The largescale maps are used for problems of local
slope instability, for planning of infrastructure, housing and industrial projects. The
size of the evaluation area is several tens
of square kilometres. In the detailed and
large-scale maps, the slope stability model
was applied or the physically processed model
used. Using this model, a safety factor is
calculated, and the index is calculated using
the score table. The medium-scale map is
principally for agencies dealing with intermunicipal planning and studies for local engineering works. At the medium scale, the
relationship between the landslide and contributing factors is analysed statistically. The
552
All the methods share a common conceptual model of landslide mapping, the mapping
of a set of environmental factors, which are
supposed to be directly or indirectly correlated with slope instability. Based on the
detected relationships between these factors
and the instability phenomena, the land
surface is partitioned into area units of
different landslide potential.
1 Landslide mapping units
Landslide hazards are site-specific, situationsensible and spatially heterogeneous. An
important step in any landslide hazard assessment is the preparation of landslide maps.
Landslide maps can be loosely grouped into
the three classes mentioned before: inventory, density and hazard maps. Inventory
maps simply show the location of known
landslides from direct mapping (Hansen,
1984). Density maps attempt to portray the
spatial abundance of landslides through indirect mapping. Hazard maps show the
inferred or computed degree of landslide
hazard obtained by modelling or by indirect
mapping (Carrara et al., 1995; Guzzetti
et al., 1999; Parise, 2001). In discussing the
advantages and limitations of the available
maps, and outlining possible applications
for decision-makers, land-developers and
environmental and civil defence agencies,
Guzzetti et al. (2000) have shown that GIS
technology makes it easy to prepare landslide
density maps from landslide inventories in a
research project carried out in the Upper
Tiber River basin in central Italy.
Landslide hazard mapping and assessment
require a preliminary selection of a suitable
mapping unit that refers to a portion of
the land surface. Each unit has a set of ground
conditions that are different in a definable
manner from those if its adjacent units
(Hansen, 1984). At the scale of the analysis,
a mapping unit represents a region that
maximizes intra-unit homogeneity and interunit heterogeneity for specific condition(s).
Various methods have been proposed to
partition the landscape for the purpose of
553
554
555
hazard. It is suggested that the use of finerresolution elevation data may reduce this
problem.
However, grid DEMs have several disadvantages: (1) they cannot easily handle
discontinuities in elevation; (2) the resolution
of the mesh effects the results and computational efficiency; (3) grid spacing needs to be
based on the roughest terrain in the catchment, resulting in redundancy in smoother
areas; (4) the computed flow paths tend to
zigzag, not following drainage lines, and are
systematically too long (Moore et al., 1991).
Grid size, production styles and interpolation
algorithms of DEMs, therefore, vary from
different geomorphological regions to obtain
the reliable data for the calculation of slope
and aspect. Whatever the DEM/DTM is
used for, both geometric and semantic
aspects of terrain representation should be
emphasized with information concerning the
quality provided to reliable digital terrain
models. Reliable terrain modelling involves
the producer as well as the user of the
DEM/DTM by presuming that the former is
able to specify requirements for the DEM/
DTM, and that the latter makes available a
quality report of his/her products.
3 GIS-based modelling of landslide hazard
Of particular interest are discussions and
applications of GIS to landslide hazard and
general slope instability research (Carrara,
1983; Wadge, 1988; Gupta and Joshi, 1989;
Niemann and Howes, 1992; Kingsbury et al.,
1992; Wang and Unwin, 1992; van Westen,
1993; Carrara et al., 1995; Clouatre et al., 1996;
Dhakal et al., 1999; Cavallo and Norese, 2000;
Carrasco et al., 2000; Corominas, 2000;
Barredo et al., 2000; C.F. Lee et al., 2001).
One of the crucial issues in GIS-based hazard
assessment is the availability of suitable input
data, which remain fundamentally inadequate
in quantity and quality for the intended task.
Another issue is related to many sources of
errors and uncertainties associated with data
representation, acquisition and manipulation.
It has been demonstrated clearly that landslide
556
Qualititative map
combination
Bivariate statistical No
analysis
Multivariate
No
statistical
analysis
Safety factor
analysis
Probability
of failure
Heuristic analysis
Statistical analysis
Deterministic
analysis
No
No
No
Yes
Yes
Yes
Medium
No
Yes
Regional
Advantages
No
Large
Technique
Type of analysis
Table 1
Disadvantages
558
559
560
561
562
integration of data sets, (2) weak documentation of data sets and data sources, and
(3) missing links between the data sets
(Jochen et al., 1999). More significantly, these
problems lead to the risk of losing data or
misinterpreting data. If the hidden information behind data sets can be made explicitly,
it can be valuable for improving landslide
hazard assessment including mapping or
zonation. Data integration is therefore clearly
an important research task, particularly in
landslide hazard assessment and, generally,
in geosciences research. Object-orientated
data modelling techniques can be used to
model geospatial data in an integrative way.
This approach could lead to the development
of new types of information systems capable
Figure 1 A GIS-based conceptual system integrated with data mining for landslide
hazard assessment
563
experts judgements and social issues associated with landslide hazard evaluation. Within
this framework, better-quality historic landslide databases could be constructed as the
basis for all components of landslide hazard
assessment, and the potential to effectively
evaluate the landslide posterior to data mining
and expert system should be explored in
future research of regional landslide zonation.
Acknowledgements
Funding of this research was provided by a
Grant-in-aid for Scientific Research from
the Japan Society of the Promotion for
Science and the China Postdoctoral Science
Foundation. The authors wish to express
their sincere appreciation for the generous
support received from these two organizations. We would also like to thank
B.W. Atkinson for his helpful comments.
References
Ahlberg, P., Stigler, B. and Viberg, L. 1988:
Experiences of landslide risk considerations in land
use planning in Sweden. In Proceedings of 5th
International Symposium on Landslides, Rotterdam:
Balkema, 109196.
Aleotti, P. and Chowdhury, R. 1999: Landslide
hazard assessment: summary review and new perspectives. Bull Engineering Geology and Environment
58, 2144.
Baeza, C. and Corominas, J. 2001: Assessment of
shallow landslide susceptibility by means of multivariate statistical techniques. Earth Surface Processes
and Landforms 26, 125163.
Barredo, J., Benavides, A., Hervs, J. and van
Westen, C.J. 2000: Comparing heuristic landslide
hazard assessment techniques using GIS in the
Tirajana basin, Gran Canaria Island, Spain.
International Journal of Applied Earth Observation
and Geoinformation 2, 923.
Bonham-Carter, G.F. 1994: Geographic Information
Systems for geoscientists: modelling with GIS. Ottawa:
Pergamon.
Brabb, E.E. 1984: Innovative approaches to landslide
hazard mapping. In Proceedings of 4th International
Symposium on Landslides, Toronto, 30724.
Brand, E.W. 1988: Special lecture: landslide risk
assessment in Hong Kong. In Proceedings of 5th
International Symposium on Landslides, Rotterdam:
Balkema, 105974.
Carrara, A. 1983: A multivariate model for landslide
hazard evaluation. Mathematical Geology 15, 40326.
564
565
566
567
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.