Beruflich Dokumente
Kultur Dokumente
Geological
Survey
DFID
Department for
International
Development
A report prepared for the Department for International Development (DFID) under the
Knowledge and Research Programme as part of the UK provision of technical assistance
to developing countries. The views expressed are not necessarily those of DFID.
DFID classification :
Subsector: Geoscience
Theme: G2 - Improve geological, geochemical and geotechnical hazard avoidance strategies in development planning
Project title: An implementation strategy for landslide hazard preparedness
Project reference: R6839
Bibliographic reference:
Marsh S H. 2000. Landslide hazard mapping: summary report. BGS Technical Report WC/OO/ll
Front cover illustration: Top left depicts a Radarsat image over the Javorniky Mountains, Slovakia. Top right shows
the digital geological map for the Slovakian study area. The final hazard map is at the bottom, draped on a terrain
model and shown in perspective view.
0NERC 2000
Keyworth, Nottingham, British Geological Survey, 2000
EXECUTIVE SUMMARY
This document is the final summary report resulting from successive research projects on landslide
hazard mapping undertaken by the British Geological Survey with support from the Department for
International Development (DFID). The aim of the studies has been to develop a generic approach to
landslide hazard modelling that can be applied and adapted in any developing country. The purpose
of this report is to summarise a strategy for landslide hazard preparedness based on this generic
approach. The ultimate goal of the work is to prevent or minimise loss of life and damage to property.
Conventional landslide hazard mapping involves detailed and costly ground geotechnical surveys. These
result in accurate maps that are essential in site-specific planning and development. However, such surveys
are time-consuming and expensive, and are therefore impractical and usually unjustified for large regions.
The approach developed in this research and described here provides an alternative, rapid and cost-effective
solution to the problem of providing regional hazard information. It is based on the concept that the past is
the key to the future; that is, landslides are most likely to happen in areas where the ground conditions that
caused them in the past still persist today. By identifylng the distribution of past landslide events using
remote sensing, understandingwhy they occurred using spatial analysis and mapping the relevant ground
condtions over wider areas, it is possible to predict where landslides are most likely to occur in the future.
The resulting regional hazard susceptibilitymaps are not as reliable as those produced by conventional
ground surveys and should not be used for site-specificprojects, but they do at least provide a preliminary
indication of hazard over an entire region. They can be used for general developmentplanning purposes, as
a guide to where detailed ground surveys are required, or in disaster mitigation and contingency planning.
This report is aimed at non-technical people in countries where landsliding is a problem. For the
technically minded, a brief outline of the methodology is included but reference should be made to the
full implementation strategy reported elsewhere. This summary is more concerned with strategic
issues that determine the skills and resources needed to plan and cost a landslide hazard preparedness
work programme. For planners, decision-makers and other professionals involved with landslide
hazard issues, it should provide an understandable guide to landslides, why they occur, and how
regional landslide hazard maps can possibly help in their mitigation. The report is written in general
terms that stress the need to assess the problem in the context of local and national constraints and
requirements. Implementation of the methods requires an investment in resources and probably staff
training but it should be achievable in many cases within a reasonable time scale and, given the
falling price of information technology, at reasonable cost.
~~
Summary report on
landslide hazard preparedness
CONTENTS
1. INTRODUCTION ..........................................................................................................................
2.1 Significance of the problem ......................................................................................................
2.2 Assessment and management of landslide hazards ....................................................................
2.3 What is a landslide?..................................................................................................................
2.4 Landslide causes .......................................................................................................................
2.5 Landslide hazard maps .............................................................................................................
2.5.1 Use of landslide hazard maps .............................................................................................
2.5.2 Limitations of landslide hazard maps .................................................................................
2.5.3 Accuracy of landslide hazard maps ....................................................................................
2.6 National landslide hazard preparedness strategies.....................................................................
3. HAZARD MAP PREPARATION.................................................................................................
3.1 Concepts of the remote sensing and spatial analysis approach .................................................
3.2 Developing the landslide inventory .........................................................................................
3.2.1 Remote sensing data types ................................................................................................
3.2.2 Image processing .............................................................................................................
3.2.3 Data interpretation and verification..................................................................................
3.3 GIS hazard modelling .............................................................................................................
3.3.1 Concepts ..........................................................................................................................
3.3.2 Raster GIS terminology....................................................................................................
3.3.3 Building the database .......................................................................................................
3.3.4 GIS Analysis....................................................................................................................
3.3.5 Hazard map outputs .........................................................................................................
3.4 Discussion ..............................................................................................................................
4 . IMPLEMENTATION ISSUES .....................................................................................................
4.1 The skills base and training needs ...........................................................................................
4.1.1 Remote Sensing ...............................................................................................................
4.1.2 Databasing and GIS .........................................................................................................
4.1.3 Other Considerations .......................................................................................................
4.2 Computer hardware and software requirements.......................................................................
4.2.1 Hardware Requirements ...................................................................................................
4.2.2 Software Requirements.....................................................................................................
4.3 Project planning......................................................................................................................
4. 4 Public awareness....................................................................................................................
5 . CONCLUDING REMARKS.........................................................................................................
ACKNOWLEDGEMENTS ..............................................................................................................
REFERENCES.................................................................................................................................
2
3
3
4
6
6
7
7
7
8
11
11
12
13
13
14
14
14
15
15
16
17
17
19
19
19
19
20
20
20
20
21
22
23
23
24
Summary report on
landslide hazards preparedness
1 INTRODUCTION
This report is the final output from successive research projects on landslide hazard modelling
undertaken by the British Geological Survey with support from the UK governments Department for
International Development (DFID). The original work (Project R5554:Greenbaum, 1995, Greenbaum
and others, 1995 a and b) was carried out in Papua New Guinea and Fiji and focused on geologically
young and active terrains. The second project (R6839) aimed to extend the methodology to other
regions in order to develop a generic approach that could be adapted and applied to developing
countries world-wide. Regions studied include Jamaica (Northmore and others, 2000) and Slovakia
(OConnor and others, 2000). The findings of the research have been used to define a generic strategy
for landslide hazard preparedness (Greenbaum and others, 2000) that draws on six years of
cumulative experience. This report summarises the approach developed, which extracts common
themes that are likely to apply whatever the region under study.
Landslides are a common natural hazard throughout much of the developing and developed world.
They occur when extreme events, such as heavy rainfall or seismic activity, trigger mass movements
of ground that is only marginally stable. They tend to be regarded as unusual occurrences because, on
a human scale, the interval between landsliding events may seem large and erratic. However, from a
geological viewpoint landslides are nothing exceptional. Indeed, over longer (i.e. geological) periods,
landsliding is probably the main erosional process operating in many regions. Thus, prevention may
only be possible to a limited degree. Improved understanding of the causes may, however, be used to
prevent or modify human actions that increase ground instability. For example, a land clearance in
inappropriate terrain could be prohibited or a road cutting redirected across less hazardous ground.
As landslides are a geologically common event, their distribution in space and time can be examined
and used to predict statistically the likelihood of such events occurring in the future in the same type
of terrain. This is the rationale behind landslide hazard modelling. The conventional approaches to
mapping the distribution of landslides operate at the local level. They involve detailed ground surveys,
including subsurface investigation, that determine the nature of the slip surface and the groundwater
pressures acting upon it, rock and soil strength and other key factors. Such methods result in detailed
maps that can be used for site-specific planning. However, such surveys are time-consumingand
therefore expensive. They are justifiable, and even essential, where major new infrastructure or other
significant development work is planned in an area with a history of previous landsliding. But because
of the expense and slowness of coverage they are impractical for larger regions.
The approach described here provides an alternative, cost-effective solution to this problem for large,
mainly rural areas. The technique is rapid and low-cost. It uses information that is either already
available or that can be obtained easily in order to classify broad regions in terms of their probable
susceptibility to landsliding. It is based on the concept that the past is the key to the future; that is,
landslides are most likely to happen in areas where the ground conditions that caused them in the past
persist today. By mapping the existing landslides using remote sensing techniques, understanding
why they occurred and then mapping the relevant ground conditions over wider areas, it is possible to
predict where landslides are most likely to occur in the future. The resulting maps are not as detailed
or reliable as those produced by conventional ground surveys for small areas, but they at least provide
a preliminary indication of hazard over a whole region. This information can be used for a variety of
general planning purposes and as a guide to where more detailed studies are required.
This summary report is intended primarily for non-technical people in developing countries where
there is a landslide problem. It also forms a key deliverable for DFID, who commissioned the work. It
should provide an understandable guide to landslides, why they occur, how landslide hazards can be
mapped and how to use the resulting regional landslide hazard maps to help develop a landslide
hazard preparedness strategy. There is also information on the skills and resources needed to plan,
cost and implement such a work programme. The report is written in general terms that emphasise
the need to assess the problem, mitigation requirements and implementation constraints in terms of
local and national needs. Full details on the methodology and examples of its implementationare
provided in the reports referred to above.
BGS Technical Report WC/OO/ll
Issue. 1.0
Summary report on
landslide hazards preparedness
2. LANDSLIDE HAZARDS
Landslide hazard is the potential for harm caused by existing and future landslides. It is dynamic,
because landslides vary in extent, magnitude and frequency over time due to environmental change.
Its interaction with human activities is increasing due to population growth and development. The
diverse causes, types, setting and size of landslides mean they range from innumerable, insignificant
failures to occasional, destructive high-magnitude events that have disastrous impacts on society.
Consequently, perceptions of landslide hazards vary considerably. To the general population, the
causes of landslides are obscure and often thought to be beyond human control. This attitude often
reduces the perceived need for investigations to assess the potential for landslide hazards. In other
instances, underestimation of the ways human activities can make slope instability worse, coupled
with over-confidence in engineering solutions, can result in excessive earthworks being constructed
with limited stabilisation measures. Inadequate or inappropriate engineering work during building
and road construction frequently reactivates old failures or generates new landslides.
Summary report on
landslide hazards preparedness
Total Risk is the expected number of lives lost, persons injured, value of property damaged, and cost
of disruption to economic activity. Hazard is the probability of occurrence of a particular magnitude
of landslide in a specified area and within a specified time period. The Elements at Risk measures
the total value of the population, properties and economic activities threatened within the area under
consideration. Vulnerability is the actual degree of loss within the area under consideration inflicted
on a given set of elements at risk by the occurrence of the landsliding event.
Thus, a large hazard in a desert affects few vulnerable elements at risk like people or infrastructure
and so may pose a small risk. On the other hand, a modest hazard in an over-populated urban setting
can in fact threaten a large number of vulnerable elements and so pose a higher risk. This research
considered both ends of the spectrum, drawing on studies in Papua New Guinea that concern huge
mass movements in remote areas and others in Fiji and elsewhere that looked at smaller movements
affecting large numbers of people. It is only possible to develop estimates of potential loss, and hence
assess total risk, if hazard and vulnerability assessments are combined. To do this requires that the
location, magnitude and timing of landslide events be forecast. This can be done using probabilistic
analysis but it is a difficult task and may only be feasible in specific situations. It also requires the
vulnerability of the elements at risk to be quantified. These complexities make it hard to achieve an
accurate total risk assessment. Management of landslides is therefore currently focused on local or
regional landslide hazard assessments.
Local hazard assessment concerns site-specific work, often associated with a proposed construction
project. It follows a well-established methodology that involves surface and sub-surface investigation,
instrumentation, testing and analysis. A factor of safety approach is used in which the stresses acting
on a slope (destabilising forces) and the shear strength of the slope materials (resisting forces) are
calculated. It is critical that the location and nature of any pre-existing slip, or shear, surfaces are
known and groundwater pressures acting upon them measured (Hutchinson, 1993). Such techniques
are too expensive and time-consumingto be employed in regional hazard assessments.
Regional landslide hazard assessments enable preliminary landslide hazard maps to be prepared
quickly and at reasonable cost and are of particular relevance to developing countries. They provide
information on the statistical likelihood of landsliding at a coarse scale and should not be used at the
detailed, site-specific level. Rather, regional landslide hazard maps are a tool for the strategic, predevelopment planning of infrastructure, the development of disaster response plans that mitigate loss
to life and property, and the identification of sites for local assessment. The logic behind a regional
landslide hazard assessment is that the distribution of past landslides reflects the occurrence of the
combined factors that resulted in the initial slope failure. If those causal factors can be identified and
mapped, then comparison of their mapped distributions allows identification of areas with varying
potential for landsliding in the future. Various regional hazard assessment techniques exist whose
application depends on the extent of the area under consideration, type of landsliding and available
baseline data. The methodology for rapid landslide hazard assessment developed in this study suits a
variety of terrain conditions.
Regional landslide hazard assessments have rarely been applied systematically in the developing
world for the reasons discussed above. This report describes ways to redress this, at least in part, by:
Increasing general awareness of the nature, distribution, causes and significance of landsliding;
Improving the accuracy and applicability of landslide hazard assessments;
Improving awareness of the potential of regional landslide hazard assessment for formulating
hazard preparedness strategies and thereby reducing landslide losses; and
Transferring knowledge, expertise and technology to developing countries such that national
awareness and capabilities with respect to landslide hazards may be raised and sustained.
Summary report on
landslide hazards preparedness
acting on a slope exceed the resisting strength of its materials, the slope will fail and a landslide will
occur.
It is as natural for land to slide as it is for water to run downhill. Landslides are one of the most
common of the erosional processes that shape the earths landscape. One or more of three main
mechanisms cause the displacement of the material involved in landslide movements:
Falling: detachment of blocks from steep slopes that descend by free-fall, bouncing or rolling;
Sliding: movement of materials as a coherent body over a basal shear surface; and
Flowing: turbulent motion of wet or dry material that moves as a viscous fluid which may or may
not be bounded by a basal shear surface.
These processes produce a bewildering spectrum of slope movement landforms and behaviours. Size
varies and displacementsrange from a few metres to tens of kilometres. Some movements are rapid
while others occur slowly. In some cases the majority of displacement may be achieved in a single,
short-lived event, whilst in others movement may be gradual, cyclical or pulsed. Displaced material
may move in coherent masses to create a characteristic terrain of scars, ridges, humps and hollows. In
other situations it may lose coherence and run away from the initial failure site like dry sugar (e.g. in
dry sands) or freshly-mixed wet concrete (e.g. mudflows). One movement mechanism may operate
alone, but usually several occur together to produce a landslide complex. Alternatively, the movement
mechanism may change progressively down slope to give what is known as a complex landslide.
The variation of slope forming materials, failure triggers and movement types means that a single,
globally acceptable landslide classification scheme has still not been produced. Yet a valid, global
classification scheme is a pre-requisite for both the field identification and reasoned generalisation of
landslides. Most importantly, it facilitates communication between earth scientists and the engineers,
planners, legislators and general public concerned with, or affected by, landslides. The terminology
used in this report is based on one of the most widely used schemes (Varnes, 1978), which primarily
uses the type of movement and nature of the displaced material (Figure 1). Varnes scheme is also
recommended by the International Geotechnical Societies (IGS) and United Nations Educational,
Scientific and Cultural Organisation (UNESCO) Working Party on the World Landslide Inventory.
LL
f
A
L
L
8
T
0
C.
c
?
L
E
8
8
L
I
0
c
8
R
A
0
f
L
0
W
E
0
Y
c
L
e
X
Summary report on
landslide hazards preparedness
Summary report on
landslide hazards preparedness
can face a variety of consequences depending on land use. So, the simplest approach to risk map
preparation is to overlay a hazard map with a land-use map. However, the resulting risk maps may be
politically sensitive and may also oversimplify the information. For this reason, decision-makers and
planners often work directly from the landslide hazard map and omit the quantitative risk analysis
stage.
Summary report on
landslide hazards preparedness
other relevant parameters are often excluded from regional hazard assessmentsbecause the data do
not exist or are difficult to measure over a whole region. For example, variable joint and fracture
spacing influences a slopes stability but is extremely time-consuming and expensive to map. Finally,
it is not always possible to assign relative levels of significance to landslide causes with a high degree
of confidence. All these considerations will affect the accuracy of the hazard zones.
Deficiencies in the landslide distribution inventory will be transmitted to the hazard map. The main
reason for such deficiencies is the difficulty of recognising old, degraded landslides in steep, vegetated
terrain. These recognition problems are exacerbated in tropical regions where high relief and severe
rainfall lead to the widespread development of small landslides. Re-growth of the tropical vegetation
cover may obscure evidence of slips after only five to ten years. Recognition problems are particularly
important if the landslide distribution inventory is compiled by interpretation of the remotely sensed
imagery with only limited field checking. Competent interpretation skills and targeted field checking
are essential. The accuracy of the final hazard zonation is also sensitive to whether the landslides are
analysed in terms of their complete area, source area only or as initiation points. Each technique has
merits and limitations, depending on the type of landslides prevalent in the region under
consideration and the data available to compile the landslide inventory. The use of Geographic
Information System (GIS) capabilities in hazard map preparation does at least allow map accuracy to
be readily updated as additional data are acquired. Section 3 describes the procedures adopted for
compiling landslide inventories using remote sensing and for undertaking the statistical hazard
analysis using a GIS.
The assumption that responsive strategies are more cost effective than those aimed at prediction and
prevention is not borne out by the limited cost-benefit analyses that are available. Studies such as that
BGS Technical Report WC/OO/l 1
Issue. 1.0
Summary report on
landslide hazards preparedness
conducted by Hansen (1984) have demonstrated the importance of landslide hazard assessment and
planning initiatives in the developed world. Landslide hazards occur in developed and developing
countries; so should sensible hazard preparedness planning. A hazard preparedness strategy should be
developed that is designed to provide essential information prior to the catastrophic event occurring.
In this way, regional landslide hazard maps can be used as part of a strategy that aims to:
Recognise geographical areas where landsliding has occurred and future landsliding is likely;
Understand the scale of the landslide problem;
Deal with the problems that may arise because of landslides on marginally stable slopes;
Identify constraints on land use by relating land-use zonation to hazard zonation;
Regulate new development in hazardous areas through planning controls;
Prepare for, modify, andor mitigate the often disastrous effects of landslides on communities and
infrastructure by means of appropriate engineering practice and building codes; and
Enhance public education and general knowledge regarding these issues.
Whilst landslide hazard reduction can be undertaken as an individual exercise, it must also be
recognised that landslides often occur as a result of interrelated multiple natural-hazard processes in
which an initial event triggers one or more secondary hazard events. Volcanic eruptions, earthquakes,
hurricanes and extreme rainfall events all cause landslides. The resulting multiple-hazard problems
require a shift in perspective from the mitigation of individual hazards to a broader framework that
takes into account the characteristics and effects of all the processes involved. Therefore, in planning
landslide hazard reduction programmes, attention should also be paid to the possible relationships
between landslides and other hazards. For example, moving a building from a hillside to low lying,
flat ground to avoid landslides might increase the risk from floods.
Although earth scientists and engineers undertake the assessment of landslide hazards, responsible
government authorities or departments must apply the results through appropriate planning policies
and other instruments. It is beyond the scope of this report to recommend in detail how this may best
be achieved for specific countries or regions. However, some general principles of landslide hazard
reduction strategies are universally applicable. The development and implementation of a national
hazard reduction strategy by the responsible government authorities or departments may hinge on a
number of important issues, including:
The level of occurrence of landslides and other natural hazards in a region;
The need to sustain a robust economy that is not repeatedly setback by successive disasters;
The need for proper land-use planning and practices for sustainable development; and
The existing level of co-ordination between agencies in applying mitigation procedures.
On the basis of collective experience from various national programmes, and as reviewed by Swanston
and Schuster (1989), a successful unified national programme of landslide hazard reduction would be
likely to include the following key elements:
Identification of a central organisation for management of a landslide loss-reduction programme;
Establishment of the limits of responsibility of national, regional, local, and municipal authorities
and private sector organisations in dealing with landslide hazards;
National efforts to identify and map hazards, define process characteristics, and determine degree
of vulnerability and risk;
Development of guidelines for the application of reduction techniques to the identified hazards;
Development of minimum standards of application and professional practice by professional
societies in collaboration with regional and national governments;
Regulation of minimum standards of application and professional practice through their periodic
review and the upgrading of building codes and land-use practices;
Strong support of university research and regional and national government dealing with process
mechanics, reduction techniques and warning systems;
Provision of a central clearing house for collection and distribution of publications and guidelines
to professionals, agencies and local authorities; and
Relief and compensation programmes through regional, national and private insurance funds.
Summary report on
landslide hazards preparedness
A comprehensive, national hazard reduction programme must clearly be tailored both to the severity
of the hazards and to what can be achieved within the funding constraints. A planned, strategic
approach is recommended that sets out realistic goals and tasks for making landslide studies, mapping
and evaluating the hazard, disseminating information to potential users, and subsequently evaluating
the use of the information.
10
Summary report on
landslide hazards preparedness
11
Summary report on
landslide hazards preparedness
interpretations by reference to published information and through field checking. Use ancillary
information to revise and update the preliminary remote sensing interpretations.
5. Transfer interpretation based on any uncorrected aerial photography to a base map, orthorectified
aerial photograph or geocorrected satellite image
6. Unless already captured digitally, digitise and attribute the landslide inventory and convert to a
raster format suitable for GIS analysis.
Issue. 1.0
13
Summary report on
landslide hazards preparedness
in each wavelength band across the full dynamic range of the display system. Images that occupy only
a small portion of the possible brightness range appear dark and lack contrast.
This process exaggerates small differences inherent in the data to achieve maximum discrimination
between surface materials. For a landslide interpretation, the contrast stretch should provide
maximum discrimination of bare rock and soil, possibly at the expense of other surface materials.
These processes are most effective on optical data. Radar data require more specialised processing
beyond the scope of this summary report. All these processes, whilst quantitative, rely on the
qualitative judgement and experience of the operator.
14
Summary report on
landslide hazards preparedness
developed, tested and updated once new information becomes available. The analyst uses the GIS to
quantify the apparent significance of the potential controlling factors to landsliding in the study area.
Theme A particular spatial variable that can be characterised by a map, such as geology.
Class A sub-division of a theme; for example, limestone is a class within the geology theme.
The analysis examines the correlation between the landslide inventory map and the various themes in
the database using cross tabulation. The themes are best represented in a raster format by nominal
data. Maps in raster format consist of a regular grid of cells, each containing a number representing
the value of the theme in that cell. Neighbourhood queries and other spatial operations around cells
and overlay operations between themes can all be performed rapidly and efficiently. Results are
obtained for every ground cell, unlike vector GIS analysis where the answers relate to large, variably
shaped areas. Nominal data simply use numbers as codes to represent classes. For example, in the
geology theme the code value 1 might represent the class limestone and the value 2, granite. Themes
such as elevation are numerical measurements on a continuous scale and must be converted to
nominal data before the analysis. Dividing the overall range of theme values into a smaller set of
classes achieves this. For elevation information, this level slicing might result in classes that represent
a range of 100 m, so that 100 m - 200 m becomes code value 1,200 m - 300 m becomes code value 2,
and so on.
15
Summary report on
landslide hazards preparedness
be digitised in a similar way to elevation contours, and data restricted to specific lithologies digitised
as raster polygons.
Land use data are often important, depicting for example the proportion of vegetated to cleared
ground. In some cases land use, forestry or agriculture maps may be available from the same sources
as soil maps. Paper maps need to be digitised in a similar way to the geology. Alternatively, a land use
classification of the remotely sensed data could form part of the project using standard IA software
and ground truth information. The digital data so produced can be included in the database easily.
Cuttings associated with road construction may themselves cause landsliding. The road network may
be digitised from existing maps, augmented by reference to a recent satellite image. A narrow buffer
could be generated around the road network for use in the GIS analysis. However, it is probably not
appropriate to do so as road construction is more like a triggering effect rather than a controlling
factor. Nevertheless, the digitised road network is a valuable way to link hazard to risk using the final
hazard map. It may also explain why certain landslides occur where they do.
1. Cross tabulate the individual themes with the landslide inventory map
The first step is to examine the amount of landsliding in every class of each theme using cross
tabulation. An area cross tabulation is a two-dimensional table summarising the extent of overlap
between all possible class combinations of two input theme maps. The landslide inventory map has
two classes, landslide and not landslide. It is systematically cross tabulated with each of the
controlling factor theme maps. The result is a table showing the percentage area affected by and not
affected by landslides for each class, which compares the incidence of landsliding to the average.
2. Calculate the significance of each theme
The significance of each theme to landsliding in general must be estimated in a consistent way (e.g.
how important is the geology theme for predicting landslides?). This can be used to determine if a
theme is a controlling factor and, if so, how much weight it should be given in the model. The method
used to establish the significance of a theme is known as Cramers co-efficient (V) and varies from 0,
representing no influence on landsliding, to a maximum of 1, for strong influence on landsliding.
Using Cramers V, themes having a high correlation with landslides can be given more weight in the
hazard model. For example, in the Jamaican case study geology has V = 0.038 whereas aspect has V
= 0.024. This indicates that geology is the more important controlling factor and can be given more
weight in constructing the Jamaican hazard model.
3. Reclassify each theme map on the basis of the significance of each class within the theme
The significance of each class within each theme must also be addressed (e.g. does alluvium host
more or less landslides than the other classes in the geology theme?). This is used firstly to generate a
hazard map for individual themes, and secondly to combine all the themes into a final landslide
hazard model. The method recommended for determining the significance of each class within a
theme is called association and was proposed by Yule (Fleiss, 1991). Association values for a class
range from -1 (maximum negative correlation with landslides) though 0 (no influence on landslides)
to + 1 (maximum positive correlation). Cross tabulation and assessment of significance are performed
for each theme. The themes are then reclassified using their association values for each class to
produce a set of maps showing the varying degree of landslide susceptibility within each theme.
4.
Sum and normalise the reclassified themes on a cell-by-cell basis to form the final model
If two themes individually influence landsliding then those themes taken together should provide an
even better predictor. So, after assessing each theme it is necessary to look at their combined effect.
BGS Technical Report WC/OO/l 1
Issue. 1.0
16
Summary report on
landslide hazards preparedness
This is done by adding the individual susceptibility maps, after first weighting each reclassified theme
or landslide association map by its Cramers V value. The result is then normalised by the sum of the
Cramers coefficients to give an overall landslide susceptibility model with values between -1 to +1
that can be interpreted in the same way as association. Providing an objective measure of the success
of the model is difficult but there are two options; compare the model with the landslide inventory, or
with landslides identified by fieldwork. Both operations can be done by visual comparison, or by using
a further cross-tabulation analysis.
Hazard zones: A raster image showing three or four levels of hazard, each with an assigned colour.
Topography: Either as a vector contour overlay or a shaded relief image generated from the DTM.
Vector data:
Roads, rivers, towns, hospitals, police stations, water tanks etc., and map grid.
Map Legend: An explanation of how the map was made, what the colour zones mean, how the map
should be used, and its scale and location that can be understood by the nonspecialist.
A major consideration is how to represent the topographic information. Conventionally,topography is
displayed on maps as contours but they can be difficult to understand for those not familiar with maps
and may also crowd the presentation. An alternative is to represent topography as a grey-scale shaded
relief image, which provides an easily understood visualisation of the landscape. This image can be
generated from the DTM using either IA or raster GIS software. The next problem is how to display
the two raster layers together; this can be approached in at least two ways depending on the software
options available and the experience of the operator. One approach is to adjust the transparency of
the hazard colours so that a grey scale shaded relief is visible through them. Alternatively, an
intensity-hue-saturation transformation can be used to combine the two, raster images into one where
colours are controlled by the hazard zonation but shading is modulated by the intensity of the shaded
relief. Both approaches give similar results, but each requires experimentation to achieve the best
output. Another useful visualisation technique involves the use of perspective views in which the
hazard map can be viewed obliquely from different vantage points and at different angles to the
horizon, creating a series of birds eye views of the terrain. This is done by draping the hazard zones
over the DTM. Such presentations can be especially helpful in scenario testing and public education.
Figure 3 shows a 1 :75 000 scale hazard map produced for Jamaica, with the hazard zones draped on
top of a shaded relief image. Reproduction within an A4 report naturally reduces the detail compared
to the full-scale version of the map, but it illustrates the type of product that can be produced.
3.4 Discussion
In most instances, regional landslide hazard maps produced using the methods outlined here will only
approximate the truth. A principal limitation of the method is likely to be the incompleteness of the
input controlling factor database. Taking the Slovakian case study as the example, although most of
the landslides occurred in the weathered surface layers, no physical property data for this weathered
zone was available to be included in the analysis. This deficiency must affect the accuracy of the
Slovakian landslide hazard model. However, an advantage of the GIS approach is that the analysis
can easily be repeated when additional information becomes available.
Another uncertainty in the analysis is the assumed independence of the themes. Most statistics are
designed for use with variables that are independent of one another, but this is unlikely to be true in
BGS Technical Report WC/OO/11
Issue. 1.0
17
Summary report on
landslide hazards preparedness
4. IMPLEMENTATION ISSUES
Previous sections of this report have outlined a generic strategy for, and the technical approach to,
landslide hazard mapping. Any geological survey or other organisation considering implementing a
project will need to develop a detailed implementationplan based on their own needs and resources.
Having decided that hazard maps are needed, priority areas must be selected and desk investigations
undertaken to establish whether sufficient data on controlling factors exist for the project to succeed.
Consideration should be given at the outset to the needs of the potential users and the availability of
other resources including suitably qualified staff, computing facilities and funding.
19
Summary report on
landslide hazards preparedness
terms of the broader benefits that the extra computing facilities bring to the organisation. Although
costs have reduced significantly in recent years, the total hardware/software requirement of the image
analysis system may still be substantial. The investment may not be justified if the proposed landslide
hazard mapping were the only application and any such decision should be taken in the context of the
wider needs of the organisation as a whole. Other applications of remote sensing include geological
mapping, mineral exploration, groundwater and waste management and environmental monitoring.
Organisations not able to make this investment should consider whether it is feasible to use the
facilities of another institution (e.g. a university) on a collaborative basis.
20
Summary report on
landslide hazards preparedness
Goal
The higher level objective to which the project hopes to contribute. It is useful to state
this as it may help funding departments to justify the award of funds in terms of their
own policy objectives (which the project proposers would be well advised to consult).
It may be little more than a motherhood statement, like: Protect population and
infrastructure from natural and man-made disasters.
Purpose
outputs
Unlike the Goal and Purpose, the project is held accountable for delivering Outputs.
These are not the same as deliverables and may be better thought of as an immediate
desired effect (e.g, better informed decision-makers, planners, emergency services).
Tangible deliverables (e.g. the maps themselves) are better considered as verifiable
indicators (see below) of the Outputs.
21
Summary report on
landslide hazards preparedness
Activities
Specific actions to be undertaken in order to achieve each stated Output. There will
usually be one main Activity for each Output, subdivided into an ordered series of
individual tasks. They should be described in a narrative and shown graphically on a
bar (Gantt) chart. Each task is given a time frame, milestone(s) and team member
responsible. For large projects, each component (e.g. a hazard map for a specific
region) may be described as a separate phase.
Indicators
Assumptions Internal or external factors that can influence the projects success should be
considered and listed. Success to at least the Output level requires appropriate
contingencyplanning to deal with any damaging, uncontrolled event.
Team
All staff necessary to carry out the project should be described in terms of skills and
specialisms. The plan should identify whether they exist in-house, need to be trained,
or need to be recruited. It should describe the projects management and responsibility
structure, external links and the steering committees role.
Resources
Finances
The project should be fully costed to take account of all elements. This will include:
staff costs (including overheads); training; field costs (including transport and living
expenses); computing hardware and software (including ongoing maintenance);
purchase of imagery/photography; consumables etc.
The form that the proposal takes will depend on the requirements of the department or funding
agency. A good project plan, besides being required to secure funding, is also a tool for project
management. However, it should not be so inflexible that it cannot be adapted to deal with specific
events or delays that occur. Experience gained at each stage will help improve subsequent plans.
4 . 4 Public awareness
For the project to be successful the results must not only be applied but the concept of landslide hazard
mapping must also gain acceptance in the community. An active and enthusiastic Steering Committee
may also serve to raise public awareness and understanding. Such publicity can act for the good of the
project provided it is planned and not merely responsive. However, one of the pitfalls of informing the
general public about hazards is that the information is open to misinterpretation, which can result in
bad press, public alarm and planning blight of certain areas. So, a campaign of public awareness,
dissemination and education might therefore be required as part of the project plan to pre-empt this
reaction. It is important that everyone involved in and affected by the project, from the emergency
services, to the insurance industry, to the population at large should understand in general terms what
hazard maps are, what they are used for, and what they do - and dont - reveal.
22
Summary report on
landslide hazards preparedness
This report is the product of successive three-year research projects on landslide hazard modelling
undertaken by the British Geological Survey under support from the UK Department for International
Development. The original work focused on geologically very young and active terrains. The second
project (R6839) was aimed at extending the methodology to other regions in order to develop a
generic approach that could be adapted and applied to any developing country. This report
summarises such a strategy, but it must be stressed that it provides no more than a skeleton around
which to develop a national programme. The reasons for this include:
Differences in the scale and causes of the landslide problem;
Climatic and geographical differences;
Consequent variations in the usefulness and availability of different types of remote sensing data;
The availability of information layers relevant to landslide modelling;
The local technical and especially Information Technology skills base;
Local computer hardware and software systems and support;
Political and/or local government constraints; and
Funding.
There is a significant volume of published literature on the modelling of landslide hazards, including
a number of variants on the approach presented here. These differences reflect both the stance taken
by different workers and the uncertain nature of the science with which we are dealing. There will
always be a compromise between the effort that goes into the model and the reliability of the results.
The technique suggested here charts a middle course; one that attempts to provide a useful product
with a moderate effort, as this probably represents the lowest common denominator on which most
implementationscan build. It recognises that local circumstanceswill dictate what is possible and
useful. It allows users to develop their own approach based on the method suggested. It recognises
that technology is fast changing. Ikonos-2 satellite imagery did not become available in time to be
used in these case studies, but it is appreciated that this and other similar very high spatial resolution
imagery may well prove to be the way forward. This is important, since a good landslide inventory is
fundamental to developing a reliable model of the hazard.
For those considering developing a landslide hazard work programme, we suggest adopting a staged
approach. The initial area chosen should be regarded as a test bed to develop and evaluate techniques
and outputs. Following this, a review of the results and further consultation with the potential users
should be undertaken in order to define a more formal approach to be adopted in subsequent phases.
As GIS becomes more widely available and used, we see there being much greater interaction with
other government, university and commercial users since many of the data sets will be of common
interest. This is particularly true of digital terrain models but we see it extending to land use data and
geocorrected aerial photography and Earth observation imagery. It will have many benefits including
the availability of new data from external sources to use within the project. Since database building is
one of the major inputs, such shared involvement could result in significant savings in time and
money. Finally, we wish to stress again the importance of collaborating with any other parties who
have an interest in the work or an involvement at the take-up level. Not only will this ensure that the
products are useful to and understood by those they are intended for, but it will also create a feeling of
shared intellectual ownership and thus encourage their adoption. The success of such ventures can
only be measured in terms of the practical use to which the results are put.
This work was funded under the Knowledge & Research Programme of the Infrastructure and Urban
Development Department, DFID. Many people contributed to the individual case studies, including
all the authors of the case studies listed in the references. Collaborators from participating institutes
23
Summary report on
landslide hazards preparedness
provided invaluable local knowledge in Papua New Guinea, Fiji, Jamaica and Slovakia. The report is
published with permission of the Director, British Geological Survey (NERC).
REFERENCES
Aleotti, P. and Chowdhury, R., 1999. Landslide hazard assessment: summary, review and new
perspectives. Bulletin of Engineering Geology and Environment, Vol58, 21-44.
Bonham-Carter, G. F., 1994, Geographic Information Systems for Geoscientists: Modelling with GIS.
Pergamon Press (Elsevier Science Ltd.), Oxford, 398p.
Fleiss, J. L., 1991, Statistical Methods for Rates and Proportions. Wiley & Sons, New York, 321p.
Greenbaum, D., 1995, Project Summary Report: Rapid methods of landslide hazard mapping. British
Geological Survey Technical Report WC/95/30, 12p.
Greenbaum, D. and others., 1995a, Rapid methods of landslide hazard mapping: Papua New Guinea
case study. British Geological Survey Technical Report WC/95/27, 112p.
Greenbaum, D. and others., 1995b, Rapid methods of landslide hazard mapping: Fiji case study.
British Geological Survey Technical Report WC/95/28, 103p.
Greenbaum, D. G, OConnor, E. A., Northmore, K. J., Jordan, C. J., Marchant, A. P., Marsh, S. H.
and McDonald, A. J. W. 2000. An implementation strategy for landslide hazard preparedness. British
Geological Survey Technical Report WC/OO/8,54pp.
Hansen, A., 1984. Landslide hazard analysis. In Slope Instability, Brunsden, D. and Prior, D. B.
(eds). John Wiley & Sons. 523-602
Hutchinson, J. N., 1993. The assessment of sub-aerial landslide hazard. In Landslides Hazard
Mitigation with Particular Reference to Developing Countries: Proceedings of a Conference held on
12 November, 1993 at the Royal Society, London, 57-66.
Hutchinson, J. N., 1998. General report: Morphological and geotechnical parameters of landslides in
relation to geology and hydrogeology. In Proceedings of the Fifth International Symposium on
Landslides, C. Bonnard (ed), A.A. Balkema, Rotterdam, Netherlands, Vol 1, 3-35.
Jones, D. K. C., 1992. Landslide hazard assessment in the context of development. In Geohazards
Natural and Man-made, McCall, G. C. H., Laming, D. J. C. and Scott, S. C. (eds). Chapman & Hall,
London. 117-141.
OConnor, E. A., Kovacik, M., Northmore, K. J., Greenbaum, D., Marchant, A. P., Jordan, C. J.,
Macdonald, A. J. W., Kacer, S., Kovak, P. and Marsh, S. H. 2000. Landslide hazard mapping:
Slovakia case study. British Geological Survey Technical Report WC/00/9,29pp.
Northmore, K. J., Ahmed, R., OConnor, E. A., Greenbaum, D., Macdonald, A. J. W., Jordan, C. J.,
Marchant, A. P., and Marsh, S. H. 2000. Landslide hazard mapping: Jamaica case study. British
Geological Survey Technical Report WC/OO/lO, pp. 41.
Schuster, R. L. 1978., Introduction (Economics of slope movements). In Landslides: analysis and
control. Schuster, R.L. and Krizek, R.J. (eds). Special Report 176, Transportation Research Board,
1978, National Academy of Sciences: Washington, D.C. 1-10
Swanston, D. N. and Schuster, R. L., 1989. Long-term landslide hazard mitigation programmes:
structure and experience from other countries. Association of Engineering Geologists Bulletin, Vol.
26, NO. 1, 109-133.
Varnes, D. J., 1978. Slope movement types and processes. In Landslides: analysis and control.
Schuster, R. L., and Krizek, R. J. (eds). Special Report 176, Transportation Research Board, 1978,
National Acadamy of Sciences, Washington, D.C. 11-33.
24
Summary report on
landslide hazards preparedness
Varnes, D. J., 1984. Landslide hazard zonation: a review of principles and practice. IAEG
Commission On Landslides And Other Mass Movements, UNESCO, Paris, 63 pp.
25
Summary report on
landslide hazards preparedness