11

CHAPTER 2

LITERATURE REVIEW

2.1 GENERAL

In this chapter types of landslides, general landslide causes, and

previous landslide studies in India and around the world are briefly explained.
Landslide definition from literature can be organized into three stages:
Detection and classification of landslides, monitoring activity of existing
landslides and analysis and prediction of the slope failures in space (spatial
distribution) and time (temporal distribution) (Mantovani et al 1996). In
general way, the results of an international research projects dealing with the
application of Remote Sensing and GIS in Landslide analysis and prediction
of slope failures are briefly reviewed in this chapter.

Landslides are recognized as the third type of natural disaster in

terms of worldwide importance. Due to natural conditions or man- made
actions, landslides have produced multiple human and economic losses
(Fleming 1980, Guzzetti 1999). Individual slope failures are generally not so
spectacular or so costly as earthquakes, major floods, hurricanes or some
other natural catastrophes. Slope failures are more widespread, and over the
years they may cause more damage to properties than any other geological
hazards (Varnes 1984). Most of the damages and a considerable proportion of
the human losses associated with earthquakes and meteorological events are
caused by landslides, although these damages are attributed to the main event
 12

which leads to a substantial underestimation of the available statistical data on

landslide impact.

This is illustrated in Table 2.1, which shows the statistics of

landslides disasters per continent from April 1903 till January 2007 from the
Emergency Disaster Database, EM-DAT, (OFDA/CRED 2007). In this period
landslides have caused 57,028 deaths and affected more than 10 million
people around the world. The quantification of damage is more than US $5
billion. These losses have driven the scientific community to produce disaster
risk reduction plans for landslides, which imply first of all landslide risk
assessment.

Landslides are the damaging natural hazards in the mountainous

terrain such as Nilgiris. The study of landslides has drawn worldwide
attention mainly due to increasing awareness of the socio- economic impact
of landslides, as well as the increasing pressure of urbanization on the
mountain environment (Aleotti and Chowdhury 1999). Although it is yet
difficult to predict a landslide event in space and time, an area may be divided
into near-homogeneous domains and ranked according to degrees of potential
hazard due to mass movements (Varnes 1984). Such maps are called
Landslide Hazard Zonation (LHZ) or Landslide Susceptibility Zonation (LSZ)
maps.

Many examples can illustrate the catastrophic nature of landslides

in the world (Brabb and Harrod 1989, Brabb 1993). Schuster and Fleming
(1986) estimated annual losses in United States, Japan, India and Italy at one
billion or more each. Subsequently, Schuster and Highland (2001) analyzed
the socioeconomic impact of landslides in Western Hemisphere highlighting
extreme events such as a debris avalanche in 1970 in Huascaran, Peru with a
death toll of 20,000 people, a debris flow in 1985 in Nevado del Ruiz,
 13

Colombia killing 25,000 people and the 30,000 that were killed or are missing
as result of the 1999 landslides and floods in northern Venezuela. The
mismatch between these data and the ones from table 2.1 is due to the manner
in which events were recorded, and the minimum threshold for deaths and
economic impact, which is used to include an event in the official EM-DAT
database.

Statistical data about landslide impacts varies considerably when

comparing the reports of different scientific organizations. This reflects that
the comparisons are imprecise and that there is reason to assume higher losses
(both economic and human) by landslides than reported, due to the following
causes:

Landslides occur frequently, and per event they do not cause

such levels of damage as other types of events. Since many of
the disaster databases apply a minimum threshold of victims or
economic losses for disaster impact, most landslide disasters are
not recorded.

Landslide impacts in the past (historic events) are frequently not

recorded.

The records of other countries under similar natural conditions

show larger variations.

The registration of landslides in mountainous areas with low

risk but high hazard is cumbersome as not many people are
affected.
 Table 2.1 World statistics for landslides

Total Damage
Continents Events Killed Injured Homeless Affected
affected US (000'S)
Africa 23 745 56 7,936 13,748 21,740 No data
Average
32 2 345 598 945 No data
per event
Americas 145 20,684 4,809 186,752 4,485,037 4,676,598 1,226,927
Average
143 33 1,288 30,931 32,252 8,462
per event
Asia 255 18,299 3,776 3,825,311 1,647,683 5,476,770 1,534,893
Average
72 15 15,001 6,462 21,478 6,019
per event
Europe 72 16,758 523 8,625 39,376 48,524 2,487,389
Average
23 7 120 547 674 34,547
per event
Oceania 16 542 52 18,000 2,963 21,015 2,466
Average
34 3 1,125 185 1,313 154
Per event
Total 511 57,028 9,216 4,046,624 6,188,807 10,244,647 5,251,675

Source: EM-DAT database for the period 1903-2007 (OFDA/CRED 2007)

14
 15

Landslides occur as a consequence of various triggering factors.

Rainfall is one such factor. But the human intervention like deforestation may
cause the soil to lose its capacity and ultimately leads to landslides during
heavy rainfall. The Nilgiris in the Western Ghats entered an anxious era of
landslides since the calamitous landslides of 1978. The frequency of
landslides has been increased in recent years with major slides occurring in
1993, 1995, 2002, and 2007 and very recently in November 2009. The
Nilgiris landslides have been demonstrated to be the reflection of pore
pressure increase during the rainy seasons (Ramasamy et al 2006). The major
problem in Nilgiris district is deforestation. Between 1849 and 1992, the
shoals were decreased from 8,600 ha to 4,225 ha (Newspaper article
reference). Previous studies on deforestation and land use changes in Western
Ghats (Jha et al 2000) showed a loss of 25.6% in forest cover between 1973
and 1995 in the southern part. The present study aims to find the extent of
deforestation in Nilgiris district and the increase of landslides due to
deforestation.

Due to the lack of a landslide inventory, the knowledge about

geological, geomorphological, tectonic and hydrological conditions under
which these events happen is limited or even unknown in Nilgiris. The
interpretative criteria to identify and recognize these phenomena in aerial
photos or satellite images for the case of Nilgiris have been studied.

Likewise, limited work has been done so far where landslides

correlate with environmental variables like soils, slope, etc. to produce
susceptibility and hazard maps. Even fewer studies have been carried out in
areas where the landslide hazards are correlated to elements at risk to generate
risk maps.
 16

2.2 TYPES OF LANDSLIDES IN GENERAL

The term "landslide" describes a wide variety of processes that

result in the downward and outward movement of slope-forming materials
including rock, soil, artificial fill, or a combination of these. The materials
may move by falling, toppling, sliding, spreading, or flowing. Figure 2.1
shows a graphic illustration of a landslide, with the commonly accepted
terminology describing its features.

Crown Cracks
Crown

Minor scarp
Transverse Cracks

Transverse ridges

Radial
Cracks
Surface of
Toe rupture
Main Body
Toe of surface of
Foot
rupture
Surface of
separation

Figure 2.1 An idealized slump-earth flow showing commonly used

nomenclature for labeling the factors of a landslide

The various types of landslides can be differentiated by the kinds of

material involved and the mode of movement. A classification system based
on these parameters is shown in Table 2.2. Other classification systems
 17

incorporate additional variables, such as the rate of movement and the water,
air, or ice content of the landslide material.

Table 2.2 Types of landslides and the Abbreviated version of Varnes

classification of slope movements (Varnes 1978)

Type of materials
Engineering soils
Type of movement
Bedrock Predominantly Predominantly
coarse Fine
Falls Rock Fall Debris fall Earth fall
Topples Rock Topple Debris topple Earth topple
Rotational
Rock slide Debris slide Earth slide
Slides Translational
Lateral spreads Rock spread Debris spread Earth spread
Flows Rock flows Debris flows Earth flow
(deep creep) (soil creep)
Complex Combination of two or more principal types of movement

Although landslides are primarily associated with mountainous

regions, they can also occur in areas of generally low relief. In low-relief
areas, landslides occur as cut-and-fill failures (roadway and building
excavations), river bluff failures, lateral spreading landslides, collapse of
mine-waste piles (especially coal), and a wide variety of slope failures
associated with quarries and open-pit mines. The most common types of
landslides are described as follows and are illustrated in Figure 2.2.
 18

A B C Surface
rupture

Surface of
rupture

Rotational landslide Translational landslide Block Slide

D E F

Rockfall Topple Debris flow

G H I
Source area
Curved tree trunks
Main track Tilted Pole
Depositional area

Soil ripples

Fence out of alignment

Debris avalanche Earthflow Creep

J Firm clay

Soft clay with water-bearing

silt and sand layers
Bedrock
Lateral spread

Figure 2.2 Illustration of major types of landslide movement

2.2.1 Slides

Although many types of mass movements are included in the

general term "landslide," the more restrictive use of the term refers only to
mass movements, where there is a distinct zone of weakness that separates the
slide material from more stable underlying material.
 19

The two major types of slides are rotational slides and translational
slides.

Rotational slide: This is a slide in which the surface of rupture is

curved concavely upward and the slide movement is roughly rotational about
an axis that is parallel to the ground surface and transverse across the slide
(Figure 2.2A).

Translational slide: In this type of slide, the landslide mass

moves along a roughly planar surface with little rotation or backward tilting
(Figure 2.2B). A block slide is a translational slide in which the moving mass
consists of a single unit or a few closely related units that move down slope as
a relatively coherent mass (Figure 2.2C).

2.2.2 Falls

Falls are abrupt movements of masses of geologic materials, such

as rocks and boulders, which become detached from steep slopes or cliffs
(Figure 2.2D).

Separation occurs along discontinuities such as fractures, joints,

and bedding planes and movement occurs by free-fall, bouncing, and rolling.
Falls are strongly influenced by gravity, mechanical weathering, and the
presence of interstitial water.

2.2.3 Topples

Toppling failures are distinguished by the forward rotation of a unit

or units about some pivotal point, below or low in the unit, under the actions
of gravity and forces exerted by adjacent units or by fluids in cracks
(Figure 2.2E).
 20

2.2.4 Flows

There are five basic categories of flows that differ from one another
in fundamental ways.

(i) Debris flow: A debris flow is a form of rapid mass movement in

which a combination of loose soil, rock, organic matter, air, and water
mobilize as slurry that flows down slope (Figure 2.2F). Debris flows include
<50% fines. Debris flows are commonly caused by intense surface-water
flow, due to heavy precipitation or rapid snowmelt that erodes and mobilizes
loose soil or rock on steep slopes. Debris flows also commonly mobilize from
other types of landslides that occur on steep slopes, are nearly saturated, and
consist of a large proportion of silt- and sand-sized material. Debris-flow
source areas are often associated with steep gullies, and debris-flow deposits
are usually indicated by the presence of debris fans at the mouths of gullies.

(ii) Debris avalanche: This is a variety of very rapid to extremely

rapid debris flow (Figure 2.2G).

(iii) Earth flow: Earth flows have a characteristic "hourglass"

shape (Figure 2.2H). The slope material liquefies and runs out, forming a
bowl or depression at the head. The flow itself is elongate and usually occurs
in fine-grained materials or clay-bearing rocks on moderate slopes and under
saturated conditions. However, dry flows of granular material are also
possible.

(iv) Mudflow: A mudflow is an earth flow consisting of material

that is wet enough to flow rapidly and that contains at least 50 percent sand-,
silt-, and clay-sized particles. In some instances, for example in many
newspaper reports, the mudflows and debris flows are commonly referred to
as "mudslides."
 21

(v) Creep: Creep is the imperceptibly slow, steady, downward

movement of slope-forming soil or rock. Movement is caused by shear stress
sufficient to produce permanent deformation, but too small to produce shear
failure.

There are generally three types of creep: (1) Seasonal, where

movement is within the depth of soil affected by seasonal changes in soil
moisture and soil temperature. (2) Continuous, where shear stress
continuously exceeds the strength of the material, and (3) progressive, where
slopes are reaching the point of failure as other types of mass movements.
Creep is indicated by curved tree trunks, bent fences or retaining walls, tilted
poles or fences, and small soil ripples or ridges (Figure 2.2I).

2.2.5 Lateral Spreads

Lateral spreads are distinctive because they usually occur on very

gentle slopes or flat terrain (Figure 2.2J). The dominant mode of movement is
lateral extension accompanied by shear or tensile fractures. The failure is
caused by liquefaction, the process whereby saturated, loose, cohesion less
sediments (usually sands and silts) are transformed from a solid into a
liquefied state. Failure is usually triggered by rapid ground motion, such as
that of experienced during an earthquake, but can also be artificially induced.
When coherent material, either bedrock or soil, rests on materials that liquefy,
the upper units may undergo fracturing, extension and may then subside,
translate, rotate, disintegrate, or liquefy and flow off. Lateral spreading in
fine-grained materials on shallow slopes is usually progressive.

The failure starts suddenly in a small area and spreads rapidly.

Often the initial failure is a slump, but in some materials movement occurs for
no apparent reason. Combination of two or more of the above types is known
as a complex landslide.
 22

In our study area, numerous numbers of landslides have occurred in

last two decades. Nilgiris experienced debris avalanche, debris flow, Rock
fall, creep and translational types of landslides have occurred.

2.3 LANDSLIDE CAUSES IN GENERAL

There are three major causes for landslides in general, i.e.

geological causes, morphological causes and human causes. They are
describes below.

(i) Geological causes

a) Weak or sensitive materials.
b) Weathered materials.
c) Sheared, jointed, or fissured materials.
d) Adversely oriented discontinuity (bedding, schistosity,
fault, unconformity, contact, and so forth).
e) Contrast in permeability and/or stiffness of materials.

(ii) Morphological causes

a) Tectonic or volcanic uplift
b) Glacial rebound
c) Fluvial, wave, or glacial erosion of slope toe or lateral
margins
d) Subterranean erosion (solution, piping)
e) Deposition loading slope or its crest
f) Vegetation removal (by fire, drought)
g) Thawing
h) Freeze-and-thaw weathering
i) Shrink-and-swell weathering
 23

(iii) Human causes

a) Excavation of slope or its toe
b) Loading of slope or its crest
c) Drawdown (of reservoirs)
d) Deforestation
e) Irrigation
f) Mining
g) Artificial vibration
h) Water leakage from utilities

Although there are multiple types of causes of landslides,

specifically the three that cause most of the damaging landslides around the
world are these:

2.3.1 Landslides and Water

In Nilgiris, slope saturation by water is a primary cause of

landslides. This effect can occur in the form of intense rainfall, snowmelt,
changes in ground-water levels, and water- level changes along coastlines,
earth dams, and the banks of lakes, reservoirs, canals, and rivers.

Landsliding and flooding are closely allied because both are related
to precipitation, runoff, and the saturation of ground by water. In addition,
debris flows and mudflows usually occur in small, steep stream channels and
often are mistaken for floods; in fact, these two events often occur
simultaneously in the same area.

Landslides can cause flooding by forming landslide dams that block

valleys and stream channels, allowing large amounts of water to back up. This
causes backwater flooding and, if the dam fails, subsequent downstream
flooding. Also, solid landslide debris can "bulk" or add volume and density to
 24

otherwise normal stream flow or cause channel blockages and diversions

creating flood conditions or localized erosion. Landslides can also cause
overtopping of reservoirs and/or reduced capacity of reservoirs to store water.

2.3.2 Landslides and Seismic Activity

Many mountainous areas that are vulnerable to landslides have also

experienced at least moderate rates of earthquake occurrence in recorded
times. The occurrence of earthquakes in steep landslide-prone areas greatly
increases the likelihood that landslides will occur, due to ground shaking
alone or shaking-caused dilation of soil materials, which allows rapid
infiltration of water.

2.3.3 Landslides and Volcanic Activity

Landslides due to volcanic activity is melting of volcanic lava at a

rapid rate, causing a deluge of rock, soil, ash, and water that accelerates
rapidly on the steep slopes of volcanoes. These volcanic debris flows (also
known as lahars) reach great distances, once they leave the flanks of the
volcano, and can damage structures in flat areas surrounding the volcanoes.

2.4 LITERATURE REVIEW ON LANDSLIDE RISK ANALYSIS

Risk is the result of the product of probability (of occurrence of a

landslide with a given magnitude), costs (of the elements at risk) and
vulnerability (the degree of damage of the elements at risk due to the
occurrence of a landslide with a given magnitude). A complete risk
assessment involves the quantification of a number of different types of losses
(FEMA, 2004), such as:

Losses associated with general building stock: structural and

nonstructural cost of repair or replacement, loss of contents.
 25

Social losses: number of displaced households, number of people

requiring temporary shelter, casualties in four categories of severity (based on
different times of day).

Transportation and utility lifelines: for components of the lifeline

systems: damage probabilities, cost of repair or replacement and expected
functionality for various times following the disaster;

Essential facilities: damage probabilities, probability of

functionality, insufficiency of beds in hospitals;

Indirect economic impact: business inventory loss, relocation

costs, business income loss, employee wage loss, loss of rental income, long-
term economic effects on the region

In many areas hazard and risk assessment procedures have been

implemented, for example in California (Blake et al 2002), Hong Kong
(Hardingham et al 1998), New Zealand (Glassey et al 2003), Australia
(AGSO 2001, Michael-Leiba et al 2003), France (Flageollet 1989) and
Switzerland (Lateltin 1997). In Australia, the National Geohazards
Vulnerability of Urban Communities Project (or Cities project) was a program
of applied research and technique development designed to analyze and assess
the risks posed by a range of geo-hazards to urban communities (AGSO
2001). The Cities Project initiated a series of case studies in Australian cities,
e.g. Southeast Queensland, Cairns, and Mackay.

The quantification of landslide risk is often a difficult task, as both

the landslide intensity and frequency will be difficult to estimate for an entire
area, even with sophisticated methods in GIS. In practice, often simplified
qualitative procedures are used, such as the one developed in Switzerland
(Lateltin 1997).
 26

2.5 PREVIOUS STUDIES ON LANDSLIDE AROUND THE

WORLD

Landslides have been occurring on the mountainous areas since

time immemorial. But they have been studied, with some scientific curiosity,
only since the 19th century. The need to overcome this geohazard has been felt
since long. In fact the earliest records of regional landslide maps date back to
1783 when numerous huge landslides in parts of Calabria in Italy had affected
many settlements and blocked rivers and streams creating 215 lakes as a co-
seismic effect of a major earthquake (Cotecchia and Melidaro 1974).
Surprisingly such important geohazard, whose menacing power had been
recognized since long, received the attention of specialist, towards correlation
of various parameters in varied geoenvironments to establish their slidability
only in the early sixties of the twentieth century, though a singular attempt of
field data based landslide map on 1:5,00,000 scale was prepared by Almagia
in 1910.

A few of such studies dealing with application of remote sensing

and GIS technology in landslides are briefly discussed in the following
paragraphs.

Carrara et al (1999), in an interesting overview paper on the use of

GIS technology for the prediction and monitoring of landslide hazards,
indicated some of the negative aspects of the extensive use of GIS in the
process, such as:

Computer-generated results are considered to be more objective

and accurate than products derived by experts in the
conventional way through extensive field mapping;
 27

The use of GIS and the production of less accurate hazard maps
by users that are not experts in earth sciences;

The increased focus on the use of new computational techniques

for landslide hazard assessment, and less interest on the
collection of reliable data;

For the average earth scientist it is difficult to keep up with the

rapid developments in the field of Geo-information Science and Earth
Observation. The number of new sensors and platforms, and the amount of
acronyms is overwhelming. Also the change of GIS software from one
version to the next, in which the methods that had been developed earlier on
do no longer function, because of changes in file structure or interface, can be
frustrating to many earth scientists. Nevertheless, GIS has become an almost
compulsory tool in landslide hazard and risk assessment, and it is the
challenge to keep on using it as a tool, and not as an objective in itself. When
using GIS, the following components of a landslide risk project can be
differentiated: data collection, data entry, data management, and data
modeling.

Powers et al (1996) developed a digital method for visual

comparison between two sets of multi temporal aerial photographs, of the
active portion of the Slumgullion earth flow in Colorado, to determine
horizontal displacement vectors from the movements of visually identifiable
objects, such as trees and large rocks. Baum et al (1998) report on the result
of displacement gradients obtained through photogrammetrical work of multi-
temporal aerial photos in Honolulu, Hawaii. Maas and Kersten (1997) present
two practical studies on the helicopter-based use of a high-resolution digital
still-video camera for digital aero-triangulation and the automatic generation
of digital elevation models and ortho-photos. Test regions were an alpine
village and a landslide area in Switzerland.
 28

Moon and Blackstock (2003) used an entirely different approach in

their study on deterministic landslide hazard assessment for the city of
Hamilton in New Zealand. They selected representative slope profiles from a
DEM within the various geomorphological units. For the slope stability
analysis both circular (using the Bishop Simplified method) and noncircular
(using the Spencer- Wright method of analysis) failure surfaces were used,
taking into account variations in water table and seismic accelerations.

Miller and Sias (1998) worked with a two dimensional finite-

element model (MODFE) to simulate unconfined groundwater flux and to
calculate water table elevations and factors of safety for large landslides using
Bishop's simplified method of slices along individual slope transects.

In the field of landslide run out modeling also GIS has been used
extensively (Hungr 1995). Dymond et al (1999) developed a GIS-based
computer simulation model of shallow landslides and associated sediment
delivery to the stream network, for different rainstorm events and landuse
scenarios. A high resolution DEM is one of the major components in the
model. Cellular automata have also been used extensively in modeling the
flow velocity and extend of landslides (Aviolo et al 2000).

The use of physical distributed models for landslide hazard

zonation with GIS also has a number of drawbacks. As the input data
normally have a high degree of uncertainty, the values that result from the
calculations should not be taken as absolute values of landslide occurrence,
and therefore cannot directly serve for quantitative landslide risk assessment.
Furthermore, a considerable parameterization is needed, and from sensitivity
analysis the estimated soil depth appears to be a crucial factor, which is also
most difficult to measure. The models are also not suitable in predicting the
development of complex landslides with a complex hydrological system (Van
Asch et al 1999).
 29

Van Westen et al (2003) have analyzed the evolution of the Tessina

landslide using sequential aerial photographs and direct field mapping. The
interpretations were converted to large scale multi temporal topographical
maps, resulting in detailed geomorphological maps of the Tessina landslide
for different periods. Further, he emphasizes the need of series of digital
elevation models for different time steps to calculate the total volumes of
material removed and accumulated for the entire Tessina landslide using
quantitative volumetric analysis.

Clerici et al (2002) have discussed landslide susceptibility zonation

by the conditional analysis method which is applied to a sub division of
landscape in Unique Condition Units (UCU), their discussions have evolved
as the conceptual simplicity of this method, however does not necessarily
imply that it is simple to implement, especially at it requires rather complex
operations and a high number of GIS commands. More over, there is the
possibility that, in order to achieve satisfactory results, the procedure has to be
repeated a few times changing the factors or modifying the class sub division.
To solve this problem, a shell program which, by combining the shell
commands, the GIS Geographical Research Analysis Support System
(GRASS) commands and the gawk language commands, carries out the whole
procedure automatically, this makes the construction of a landslide
susceptibility map easy and fast for large areas too, and even when a high
spatial resolution is adopted.

Hervas et al (2003) have proposed a method for mapping new

landslide occurrence and monitoring ground surface changes related to land
activity using optical remote sensing imagery. The method is based on
automatic digital image change reduction and thresholding techniques. They
suggested that the image processing techniques should be used with caution
on digital aerial photographs in the absence of suitable satellite imagery
 30

covering a major reactivation episode. Further, they demonstrated the

usefulness of this method on panchromatic images separated by long time
intervals, typically more frequent observations would be needed to more
precisely monitor ground surface dynamics.

Chau et al (2004) have presented landslide inventory based and GIS

based frame work for systematic landslide hazard analysis by employing
historical landslide data in Hong Kong, coupling with slope angles, elevation,
lithology, soil deposit distribution potential run out area of landslide, rainfall
and population. Based on the above parameters, they proposed the landslide
hazard zonation and risk maps using raster calculations.

Wu et al (2002) have focused on the zonation of the landslide

hazards using an integrated information model based on the investigation and
statistics of landslides. It is divided into destructive, disastrous, slightly
disastrous, likely disastrous and essentially non-disastrous areas. They
suggested that the zonation of landslide hazards may serve the purpose of
providing some geological environment data for the constriction of the
project.

Flageollet (1996) has studied the temporal dimensions of mass

movements. He observed that the time terminology was often inaccurate or
incomplete. So he has tried to clarify such as the state and mode of activity,
dormancy, the return time and the age of a movement.

Dikau et al (1996) have discussed the use of database and GIS for
temporal occurrences and forecasting of landslides. They state the temporal
landslide database information is correlated with recent ad historical
triggering factors to calculate temporal probabilities for landslide forecasting
using landslide frequency analysis. Their discussions have evolved that at the
medium and broad scales different combinations of landslide data with factor
 31

maps lead to static susceptibility and hazard assessments, which allow

probability evaluations for future landslide occurrences and local scales
process and deterministic slope stability models are in use.

Cheng et al (2004) have discussed the landuse / landcover change

detection for locating landslides using remotely sensed images. Their
discussions have evolved, that a grey level threshold of the band ratio
difference images is determined as the value whose exceeding probability
equals the aerial percentage of landuse change. DTM data were also used to
further restrict landslides areas to steep slope areas.

Wilcke et al (2003) have assessed the impact of landslides on soil

fertility and compared the properties of shallow translational debris slides
with those of adjacent undistributed soils. Their discussions have evolved as
the most obvious change in soil properties caused by the landsliding was
partial of complete removal of the organic layer, which was not restored
during the 20 years covered by the chronosequence. Their decreased the
topsoil fertility of the landslide area.

Dai et al (2002a) have demonstrated that slope instability modeling

by using GIS technology and logistic multiple regression analysis. They state
that the GIS tools have made possible the production of innovative slope
instability maps. In particular, they have facilitated the application of the
logistic multiple regression analysis technique which is applied to training
samples collected from existing data layers considered to be relevant to
landslide occurrences was able to predict slope instability at a rate of about
85% concordance. The predicted susceptibility was used to produce a map of
relative landslide susceptibility.

Mantovani et al (1996) have presented the inventory of researches

concerning the use of remote sensing for landslide studies and hazard
 32

zonation as mainly carried out in the countries belonging to the European

Community. They emphasized the applicability of remote sensing in the
following three oases of landslide studies are detection, monitoring and
hazard analysis.

Carmen Solana and Christropher Kilburn (2003) have demonstrated

the public awareness programme of landslide hazards around populations
existing perception of landslides. They state that it is crucial that vulnerable
communities are aware of the hazards they face and know how to respond in
an emergency, as a case study to gauge the awareness of the landslide
hazards, a survey has been conducted among vulnerable communities in the
Barranco de Tiranjana (Bolt) basin on Canaria, one of the most active zones
of slope movement in the canary Islands. Results from a formal questionnaire,
together with anecdotal evidence, suggest that the communities are generally
aware that landslides occur in the basin and can be dangerous.

Julian et al (1996) have discussed the different aspects of landslide

activity. They stated that the context for landslide development is a
particularly favorable one, both in terms of the geomorphic and structural
setting and of the climatic, hydrologic and seismic factors that triggers such
failures.

Dattilo and Spezzano (2003) have demonstrated parallel simulator

developed by a problem solving environment, called Cellular Automata
Environment for systems modeling Open Technology (CAMELOT) that
handles debris / mud-flows. They stated that CAMELOT is a simulation
environment that uses the cellular automata formalism to model and simulate
dynamic complex phenomena on parallel machines and it combines
simulation, visualization, control and parallel processing into one tool which
allows to interactively explore a simulation, visualize the state of the
 33

computation as it progresses and change parameters, resolution or

representation on the fly.

Dai et al (2003) have discussed the characterization of rainfall

induced landslides. The main objective of this study is to characterize the
initiation process and the subsequent travel distance of the landslides resulting
from recent rainstorms by integrating aerial photogrammetry with GIS.
Further, this emphasized this approach could serve as an effective means of
landslide characterization.

Jibson et al (1998) have presented a method for producing digital

probabilistic seismic hazard maps. They have used the data sets are inventory
of triggered landslides, about 200 strong motion records of the main shock,
1:24,000 scale geological mapping of the region, engineering properties of
geologic units and high resolution digital elevation models of the topography.
Combining these data sets in a dynamic model based a Newmarksa
permanent-deformation (sliding-blocks) analysis yields estimates of co-
seismic landslide displacement in grid cell from the Northrige earthquake.
Then, they compared this modeled displacements with the digital inventory of
landslides triggered by the earthquake. They anticipate that this mapping
procedures will be used construct seismic landslide hazard maps that will
assist in emergency preparedness planning and in making rational decisions
regarding development and construction in areas susceptible to seismic slope
failure.

Jibson (1993) has discussed the Newmarkss sliding block analysis

method for modeling a landslide as a rigid plastic block sliding on an inclined
plane provides a workable means of predicting earthquake-induced landslide
displacements. He stated that this method yields much more useful
information than pseudostatic analysis and if for more practical then finite
element modeling. Further, he suggested that a simplified Newmark method
 34

can be used for approximate results; it can be used, which estimates Newmark
displacements as a function of landslide critical acceleration and earthquake
shaking intensity.

Dai et al (2002b) have reviewed the recent advances in landslide

risk assessment and management, and discussed the availability a variety of
approaches to assessing landslide risk. Firstly, they proposed a framework for
landslide risk assessment and management by which landslide risk can be
reduced. This is followed by a critical review of the current state of research
an assessing the probability of landsliding runout behavior, and vulnerability.
Further described the effective management strategies for reducing economic
and social losses due to landslides followed by problems in landslide risk
assessment and management is also examined. Finally they concluded that the
modern technologies, such as GIS and remote communications, should have
wider application in landslide risk assessment and management.

Parise and Jibson (2000) have analyzed the frequency, distribution

and geometries of triggered landslides near the earthquake epicenter. Their
analyses have evolved the landslide morphologies by computing simple
morphometric parameters (area, length, width, aspect ratio, slope angle). They
calculated two indices: the susceptibility index and the frequency index to
quantify and rank the relative susceptibility of each geologic unit to seismic
landsliding. They divided the susceptibility categories into very high, high,
moderate and low.

Refice et al (2001) have presented the applications of Differential

SAR interferometric techniques to the assessment of the stability of landslide
prone areas. They have shown the technique, known as the Permanent scatters
approach, to give excellent results over areas with high densities of man mode
target. This technique deals with some of the applications over different areas
such as those affected by slope instability phenomena.
 35

2.6 PREVIOUS LANDSLIDE HAZARD ZONATION STUDIES

IN INDIA

Landslide Hazard Zonation studies in India were initiated by the

Geological Survey of India (GSI), the GSI has been involved in the site-
specific investigations of a number of landslides particularly those related to
communications routes, urban settlements and River Valley projects. The
pioneering investigation related to the stability of slopes for urban settlement
dates back to 1896 when survey was called upon to study the stability of
slopes around Nainital, an important hill resort. Prior to this, a classical
documentation was carried out by the Sir T.H.Holland of the survey in 1893,
of the catastrophic rock slide in Brihaiganaga valley that lead to creation of
huge reservoir. This was instrumental in obviating loss of life by flooding due
its partial breach that was predicted with uncanny accuracy.

Since those early days, the officers of the survey have carried out
detailed evaluation of mechanism of failure of specific slides in different
geoenvironments and have evolved treatment measures for communications
routes, natural and curt slopes, engineering projects as well as urban
settlements. These studies, though essential for designing safe slope cuts and
for evolving treatment measures for failing slopes. Even though they do not
answer the queries of environment conscious communities and for planners of
developmental activities to arrive where, when and how much is the hazard in
a particular domain. Though in regional hazard evaluation, all these questions
may not be possible to be replied to. But the most central one is how much
hazardous a particular domain for failure. This question has to be addressed
objectively in any zonation exercise.

In most cases, landslide hazard zonation exercise assesses the

relative hazard by comparing the slopes with one another, using the
influencing parameters without calculation of safety factors. The quality as
 36

well as, the utility of such maps is dependent on the scale at which these maps
are prepared, because of this choice and treatment of the stability influencing
parameters would be scale dependent.

For example, the first generation, small scale maps (1:1 million)
could take into consideration the parameters like the general physical
characteristics of the slope forming materials (lithology), the general relief
and annual rainfall precipitation. The regional small scale zonation, maps are
the simple thematic representation of terrain evaluation and they serve the
purpose of synoptic representation of areas where this natural phenomenon is
prevalent. Contrary to this, the medium scale second generation maps on
1:50,000 scale would take into consideration the shear characteristics of the
slope forming materials, the slope morphometry, the landuse, geomechanical
behavior of the discontinuity surfaces etc., as the inputs. Local networks of
rainfall measurements as well as, the groundwater conditions are now
available and these could be used for hazard evaluation as well as mapping.

During the last few decades, attempts at landslide hazard zonation

studies have been made in different parts of the country. Since then a large
number of landslides were investigated, but Landslide Hazard Zonation, as it
is commonly understood today, is relatively a new concept. Different
approaches to zonation have been followed by different investigators.

Krishnaswamy (1980) was perhaps the first to attempt landslide

zonation at the national level. He made the three fold geomorphic division of
India into the penisular, the Indo-Gangetic plain and the Extra-Peninsular as
the basis for evaluating the relative incidence of landslides.

The first attempt on regional level landslide hazard zonation studies

in the North Eastern region (Majundar 1980) and in the North West Himalaya
 37

(Narula et al 1996) was made by GSI. The next major attempt on regional
zonation was made in 1982 for the Nilgiris district of Tamil Nadu.

These maps were prepared taking into consideration of the

lithology, general physiography, rainfall patterns, seismicity and domains of
crustal adjustments. The basic approach in both these maps has been similar
to the one suggested by Krohn and Slossen (1976) in which the landslide
prone or resistant bed-rock and steepness of the relief were used and the area
categorized as high, moderate and low. The perusal of these maps would
indicate that these maps could serve only the thematic representation of the
lithoilogical, tectonic and physiographic conditions and would be of very
limited use for planning and execution of developmental activities or for
mitigation purposes. As because within the very high domains demarcated in
these maps, these area could again be subdivided into various vulnerability
classes on larger scales when more rigorous analysis of the parameters is
carried out on 1:50,000 scale. To explain this, larger scale map of a window
of high hazard zone of the smaller scale if reproduced.

The first attempts of the second generation landslide hazard

zonation maps on 1: 50000 scale was attempted by the GSI in the Nilgiris
hills, southern part of India, in which more than the overlay, the numerical
method with ratings were given for slope angles, thickness of soils, drainage
and landuse. Five landslide susceptibility zones were identified. (GSI 1982) it
addresses soil and debris slides. Landslide Hazard Zonation maps on 1: 50000
scale have been prepared of an area aggregating about 12,000 sq. km. in the
Chenab, Sutlej, Beas and Ganga basins utilizing the overlay methods. In all
these studies the remotely sensed database was also utilized for making the
landslide incidence maps with representative field checking. The inputs for
these included the detailed morphometry, characterization of slope forming
materials, and the geomechanical behavior of the discontinuity surfaces which
 38

contain low strength sheared material, the critical angles of failure of different
materials, identification of type of failure in a particular material and given
natural conditions as derived from landslide incidences (Gupta 1988, Sharan
1992). These inputs give normalized conditions for identifying areas of
different landslide potentials.

One of the early projects on zonation was carried out by Central

Road Research Institute in 1984, in which hazard zonation techniques were
used to choose a most suitable alignment from the possible alternative
alignments on landslide affected stretches in Sikkim area. Subsequent
monitoring has shown that the choices made have been proved to be
successful. During 1989, a hazard zonation map was prepared for a part of
kathgodam- Nainital highway. This map was prepared with the objective of
enabling the department to evolve a suitable maintenance strategy to keep the
hill slopes along the road free of landslide problem (Sharma 1999).

Landslide hazard zonation studies in oarts of Beasvalley, Himachal

Pradesh (Prakash Chandra 1996) and in parts of Bhagirathi valley, Garhwal in
North Western Himalayas (Gupta 1996, Sharma 1996) are mostly confined to
small area and limited number of slides. Studies along NH31A of Sikkim in
Eastern Himalayas (Sengupta and Gohosh 1996) are mostly based on the
Landslide Hazard Evaluation Factors (LHEF) rating scheme, which is mainly
a quantitative way to ascertain relative importance to factors for slope
instability.

Landslide Hazard Zonation along the ilgrim road routes in the

Himalayan regions of Uttranchal and Hiamchal Pradesh was done using
remote sensing and GIS techniques based on the Analytical Hierarchical
Process and Saatys principle of pair wise Comparison model by NRSA,
Hydrabad (2001). They modeled landslide hazard zonation based on true
topographic conditions without the effect of triggering factors.
 39

Ramakrishnan et al (2002) have been made the attempt to identify

landslide prone areas using photogrammetry with 3D GIS techniques. The
advantage of the high resolution data helps in deriving 2m contour, which is
ideal to get the elevation and slope values of the terrain.

Prabu et al (A1 2009) have developed a new model for landslide

hazard mapping through the integration of GIS, Remote Sensing and Neural
networks. He compared conventional method of landslide mapping with the
use of neural networks in landslide mapping.

Anbalagan (1992) has been evolved new quantitative approach

based on major causative factors of slope instability. He adopted a landslide
hazard evaluation factor rating scheme for Landslide Hazard Zonation.

Sanjeevi Kumar et al (2004) have developed the web based GIS for
landslide inventory for the Nilgiris district. It includes spatio-temporal
landslide database, different landslide inducing factors and landslide hazard
zonation. This application was developed in ArcIMS to view the landslide
information together with other data layers.

Bureau of Indian Standards has published a code (IS 14496 (part

2):1998) on preparation of Landslide Hazard macro Zonation Maps in
mountainous terrains Guidelines based on LHEF rating scheme for
different causative factors.

The BMPTC (2003) has taken the effort to produce the Landslide
Hazard Zonation Atlas of India on 1: 6 million scale. This small landslide
hazard maps only provide a mega view of landslide hazard distribution across
our country.
 40

Since then, a number of landslide mapping programmes have been

carried out in different parts of the country, mostly confined to small scales
and with limited terms of references. These examples have been cited to
explain how the utility of the landslide hazard maps goes on increasing with
the scale of the maps.

The national and regional level Landslide Hazard Zonation maps

which depict the thematic representation of slide prone areas based on general
lithological, tectonic, climatic and physiographic conditions. Thus would be
of limited use and for realistic mitigation efforts for larger scale maps at least
on 1: 10,000 scale will have to be prepared.

There are vast tracks of the Nilgiris, Which are landslide prone and
needs quick survey for zonation would take lot of resources, for the large
scale. From the small scale maps, high susceptibility areas should be
identified as a first priority, and then highest hazard areas should be selected
for large scale analysis. This will help to choose favorable locations for sitting
development schemes such as townships, dams, roads and other development.

2.7 SUMMARY

This literature on landslide risk assessment indicates that a lot of

developments have taken place in the last decade in Nilgiris, and that
quantitative risk assessment for individual locations is feasible (Wu et al
1996, Morgenstern 1997, Einstein 1997, Fell and Hartford 1997, Wong et al
1997). However, the generation of quantitative risk zonation maps, expressing
the expected monetary losses as the product of probability (of occurrence of a
landslide with a given magnitude), costs (of the elements at risk) and
vulnerability (the degree of damage of the elements at risk due to the
occurrence of a landslide with a given magnitude) seems still a step to far.
 41

In the meantime, risk maps are produced for many municipalities,

following a pragmatic and qualitative approach (Michael-Leiba et al 2003).
Such risk maps form the basis for development and regulatory planning.
Geo- Information tools have become essential for landslide hazard,
vulnerability and risk assessment. For obtaining landslide probability
information the following approaches are possible:

At large scales deterministic models are used for determining

factors of safety. Dynamic models are used to model trajectories
of landslides. With probabilistic methods, failure probability can
also be obtained.

At medium scales landslide data is combined with factor maps

(e.g. slope angle, lithology etc) using heuristic or statistical
methods in landslide susceptibility maps. It is also possible to
obtain landslide probabilities, when combining the landslide
frequency analysis with landslide information from temporal
databases.

Finally, the various components of landslide risk assessment should

be integrated in risk information management systems which should be
developed as spatial decision support systems for local authorities of Nilgiris
district dealing with risk management.

The use of statistical methods has a number of drawbacks. One of

these is the tendency to simplify the factors that cause landslides, by taking
only those that can be relatively easily mapped in an area, or derived from a
DEM. Another problem is related to generalization, assuming that landslides
happen under the same combination of factors throughout the study area. The
third problem is related to the fact that each landslide type will have its own
set of causal factors, and should be analyzed individually.
 42

The statistical models generally ignore the temporal aspects of

landslides, and are not able to predict the impact of changes in landslide
controlling conditions (e.g. water table fluctuations, or landuse changes).

Hence by keeping all these drawbacks in our mind, we have chosen

a soft computing technique called neural networks for mapping the landslide
susceptibility. An artificial neural network is trained by the use of a set of
associated input and output values. The method is not available within
existing GIS systems, and can been programmed in systems like MATLAB
(Lee et al 2003).

Based on the above literature survey, the main conclusions are

arrived as follows.

For planning and execution of any environment friendly

developmental activity and mitigation of these hazards, its zonation is a
prerequisite. The zonation of landslide hazard must be the basis for any
landslide mitigation project.

Hence it is necessary to develop landslide hazard zonation model

by the combining remote sensing data, socio economic analysis and GIS
technology to generate the Landslide Hazard Zonation. This will help the
planners and decision makers so as to enable them in making quick decisions
and better planning to execute the landslide mitigation measures.

This integrated study on landslide hazard zonation at micro level

will indicate the methodology can be adopted for assessing the landslide on
large scale and also provide an action plan for landslide management as an
optimal solution to the planners and decision makers for better planning. This
study could be applied to similar problematic areas and terrain conditions.