You are on page 1of 21

Geomatics, Natural Hazards and Risk

ISSN: 1947-5705 (Print) 1947-5713 (Online) Journal homepage: https://www.tandfonline.com/loi/tgnh20

Hydrologically complemented deterministic slope


stability analysis in part of Indian Lesser Himalaya

John Mathew, S. Kundu, K. Vinod Kumar & Charu C. Pant

To cite this article: John Mathew, S. Kundu, K. Vinod Kumar & Charu C. Pant
(2016) Hydrologically complemented deterministic slope stability analysis in part of
Indian Lesser Himalaya, Geomatics, Natural Hazards and Risk, 7:5, 1557-1576, DOI:
10.1080/19475705.2015.1101026

To link to this article: https://doi.org/10.1080/19475705.2015.1101026

© 2015 Informa UK Limited, trading as Published online: 20 Oct 2015.


Taylor & Francis Group

Submit your article to this journal Article views: 931

View related articles View Crossmark data

Citing articles: 3 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=tgnh20
GEOMATICS, NATURAL HAZARDS AND RISK, 2016
VOL. 7, NO. 5, 15571576
http://dx.doi.org/10.1080/19475705.2015.1101026

Hydrologically complemented deterministic slope stability


analysis in part of Indian Lesser Himalaya
John Mathewa, S. Kundub, K. Vinod Kumara and Charu C. Pantb
a
National Remote Sensing Centre, ISRO, Hyderabad, India; bDepartment of Geology, Kumaun University,
Nainital, India

ABSTRACT ARTICLE HISTORY


This study uses a deterministic approach to evaluate the factor of safety Received 23 January 2015
(FS) of the terrain for different hydrological conditions, in part of Indian Accepted 17 September 2015
Lesser Himalaya. The results indicate sudden increase in the percentage KEYWORDS
unstable area from 7.5% to 13.8% for rainfall intensity variation from 50 to Deterministic model; factor
100 mm/day. For the rainfall intensity of 15 August 2007 which caused of safety; hydrological model;
many landslides in the study area, 18.5% of the total area was unstable ROC curve
and it increases to 21.7%, 23.5% and 24.7%, respectively, for rainfall
intensities corresponding to 10, 25 and 50 year return periods. This
increment stagnates at about 260 mm/day, making about 25% of the area
unstable. Higher rainfall intensities make progressively gentler slopes
unstable, but limited to 25 degrees of slope in this area. The area
underlain by granitic gneiss showed 23.1% of area as unstable for
135 mm/day of rainfall intensity, and was followed by those areas
underlain by amphibolite (16%), limestone (13.7%) and quartzite (10.4%).
Receiver operating characteristic (ROC) curve analysis has given 84.2%
accuracy for the model. Conversion of FS to failure probability through Z
scores enables identification unstable or marginally unstable areas, for
planning selective slope stabilization measures.

1. Introduction
Assessment of geological hazard of a region holds a crucial role in the pre-disaster phase of disaster
management cycle. In India, landslides are one of the most prevalent geological hazards causing
widespread damage in its mountainous landmass on an annually recurring cycle. The Himalayan
terrain and part of the Arakan Yoma ranges in the North Eastern region, account for nearly 80% of
the slope failures in the country. Since rainfall is the most common triggering factor, landsliding is
more frequent during the monsoon or immediate post-monsoon season (Dahal et al. 2009). The fre-
quency distribution of landslides during the monsoon season reveals the occurrence of a large num-
ber of shallow landslides in the Himalaya (Mathew et al. 2014). Such shallow landslides are also
destructive when contributed from large areal extent and can cause loss of lives, damage and loss to
agriculture as well as destruction of houses and other infrastructure. When the slope gets saturated
with water from excessive rainfall or cloudbursts, debris flows or earth flows are formed which are
even more destructive due to their potentially longer run-out distances and higher speed (Ellen &
Wieczorek 1982; Brabb & Harrod 1989; Prochaska et al. 2008).

CONTACT John Mathew john_isro@yahoo.com; john_mathew@nrsc.gov.in

© 2015 Informa UK Limited, trading as Taylor & Francis Group


1558 J. MATHEW ET AL.

The spatial distribution of slope failure susceptibility is assessed using heuristic or statistical
methods based on input from remote sensing field and ancillary data (Mantovani et al. 1996; Singh-
roy & Molch 2004; Mathew et al. 2007a, 2007b, 2009; Lee and Hyun-Joo 2012; Mann et al. 2012;
Pourghasemi et al. 2012; van Westen et al. 2012, etc.). However, assessing the temporal dimension
of slope failure involves establishing the relationship between the triggering factor and the slope sta-
bility (Terlien 1998; Qui et al. 2007). Cause-effect relationship is established for rainfall-induced
landslides, statistically, empirically or through process-based approach. In statistical and empirical
approaches (based on historical rainfalllandslide relationship), the landslide trigger probability or
the empirical threshold has to be integrated with the output from spatial landslide susceptibility
assessment (based on pre-conditioning factors) to translate the result on to spatial domain (Mathew
et al. 2014). Deterministic slope stability evaluation based on the factor of safety (FS, which is the
ratio of shear strength to the shear stress on the slope-forming material), is another approach to
incorporate the effect of rainfall as a trigger (van Westen & Terlien 1996). This approach uses one-,
two- or three-dimensional hydrological models to evaluate the variation of FS over an area in
response to variations in the rainfall (Terlien et al. 1995; Safaei et al. 2011). These models provide
the saturated soil depth; a factor that is essential in calculating the FS, using steady-state or transient
response models. Steady-state models assume steady water table heights and the flow surfaces are
parallel to the topography whereas the transient response models consider the slope normal pore
water pressure by transient rainfall infiltration (Iverson 2000). The infinite slope stability model
which is used in the study requires input on hydrological parameters and hence usually is integrated
with hydrological models for stability assessment.
Customized softwares like LISA (Hammond et al. 1992), SHALSTAB (Montgomery & Dietrich
1994; Dietrich et al. 1995), TOPMODEL (Beven et al. 1984; Beven et al. 1995; Franchini et al. 1996;
Lamb et al. 1998), CHILD (Tucker et al. 1999, Tucker et al. 2001), TRIGRS (Baum et al. 2002),
CHASM (Wilkinson et al. 2002), IDSSM (Dhakal & Sidle 2004), SINMAP (Pack et al. 1998, 2001),
GEOtop-FS (Simoni et al. 2008), etc., were developed and applied by researchers, by integrating
hydrological and slope stability models to asses slope response to rainfall events. Influence of topog-
raphy on the hydrological behaviour of slope has been studied by many scientists (Beven & Kirkby
1979; Iida 1984; O’Loughlin 1986; Okimura & Kawatani 1987; Hsu 1994; Montgomery & Dietrich
1994; Duan 1996; Iverson 2000; Casadei et al. 2003; Sharma & Nakagawa 2005 etc.). Hydrological
indicators as input to early warning for slope failure were also discussed by some researchers (Knive-
ton et al. 2000; Buchroithner 2002).
Deterministic models have been used by many researchers for slope stability analysis (G€okceoglu
& Aksoy 1996; Borga et al. 2002; Moon & Blackstock 2004; Zhang et al. 2005; Fall et al. 2006; Xie
et al. 2006; Muntohar & Liao 2009; Cervi et al. 2010; Formetta et al. 2014, etc.). Vanacker et al.
(2003) implemented a coupled hydrological-infinite slope stability model to evaluate the effect of
vegetation change on slope stability. Galang (2004) used a one-dimensional infinite slope stability
model whereas Xie et al. (2004) and Qui et al. (2007) developed three-dimensional deterministic
slope stability model. Godt et al. (2008) integrated a transient deterministic model with an infinite
slope stability calculation for shallow landslide susceptibility mapping. Van Beek and Van Asch
(2004) developed a coupled, distributed hydrological-probabilistic slope stability assessment model.
In comparison to data-driven or knowledge-driven slope stability assessments, number of studies
using geotechnical models in the Himalaya or other landslide prone areas in India is less. Some of
the studies involving deterministic slope stability studies in the Nepal Himalaya are those by
Acharya (2003), Sharma and Shakya (2008) and Acharya et al. (2014). Dahal et al. (2008) carried
out an integrated, DEM-based hydrological and infinite slope stability model for deterministic eval-
uation of slope failure susceptibility in the Lesser Himalaya of Nepal.
In the Kumaun Himalayan region, Kainthola et al. (2013) applied limit equilibrium method
for calculation of FS and probability of failure. Kainthola et al. (2012) evaluated the stability
of cut slope on state highway (SH-72), connecting Mahabaleshwar and Poladpur, Maharashtra,
India, using Hoek and Brown failure criterion. Kainthola et al. (2014) derived the FS and
GEOMATICS, NATURAL HAZARDS AND RISK 1559

deformational mechanism for the same road section. Singh et al. (2015) carried out characteri-
zation of rock mass along the right bank of river Sutlej, Luhri, Himachal Pradesh, India.
Kuthari (2007) estimated the stability index of part of the Alakananda catchment (Garhwal
Himalaya) using SINMAP whereas Mondal and Maiti (2012) carried out a one-dimensional
slope stability model in part of Darjiling Himalaya. This study is one among these attempts of
deterministic slope stability studies in the Indian Himalaya, where rainfall-triggered landslides
(mainly during the monsoon period) create widespread damage every year, to understand
slope stability variation in response to varying hydrological conditions. It focuses on assessing
the terrain stability by coupling an infinite slope stability model with a steady-state hydrologi-
cal model for determination of FS using the spatial analysis capabilities of geographic informa-
tion system (GIS). Deterministic approach is adopted here because it helps direct derivation of
spatial distribution of terrain stability for modelling the cause-effect relation of slope failures
using geotechnical parameters and hydrological conditions. Furthermore, being process-based
approach, it is superior to other methods and can most accurately predict the terrain stability
under various climatic/environmental conditions and hydrological scenario (Sharma & Shakya
2008, Safaei et al. 2011).

1.1. Study area setting


The study has been carried out in the Garigaon watershed within the catchment of Birahi valley
which is a tributary of Alakananda River in Uttarakhand, India. It is bounded by latitudes
30 210 0700 N to 30 230 1100 N and longitudes 79 230 1800 E to 79 250 3000 E and lies in the
Chamoli district of Uttarakhand. The location of the study area is shown in figure 1. The
Garigaon watershed covers an area of 6.45 km2, with elevations ranging from 1140 to 3070 m
above mean sea level. The area has slope varying from near flat to steep; the steeper zones being
located towards the west, south and southwest portion of the watershed. The principal stream
of the Garigaon watershed is 4.43 km long and it descends with an average gradient of 4%. The
average rainfall in the Alakananda valley is about 1600 mm (Agrawal et al. 2010). The dominant
rainfall period is from June to September, although the highest intensity rains take place in the
month of July and August. The landslides in the catchment of the Birahi River are the result of
varying pre-conditioning factors dominated by geological and topographical aspects and trig-
gered by the monsoon rainfall.
The study area comprises of rocks belonging to the Baijnath crystallines of undifferentiated Pro-
terozoic age and Garhwal Group of Meso-Proterozoic age (Kumar 2005). The Baijnath crystallines
are represented in this area by granitic gneisses showing porphyroblastic and augen structure. Biotite
gneiss is also present in the area. The Berinag and Pithoragarh Formations of the Garhwal group are
present here. The Berinag Formation consists basically of quartzites which are thickly bedded, fine-
to-medium grained and occasionally gritty. The Pithoragarh Formation is calcareous and consists
mainly of limestone with minor intercalations of shale, slate and phyllite. The area also has amphib-
olitic intrusion. The lithological map of the area is given in figure 2. The study area has been divided
into four zones based on the lithology. These four zones have been named as zone-1 to zone-4 and
are the areas underlain by limestone, amphibolite, granitic gneiss and quartzite, respectively. It is
assumed in this study that the geotechnical property is uniform in a zone and is primarily influenced
by the lithology.

2. Methodology
Deterministic slope stability assessment using infinite slope stability model has been coupled with
steady-state hydrological model for assessing FS of the slope-forming material for varying hydrolog-
ical conditions. A hydrological model is necessary for estimating the depth of saturated flow for a
given event, which reduces the shear strength of the slope-forming material. The rainfall intensity
1560 J. MATHEW ET AL.

Figure 1. Location map of the study area.

determines the amount saturated through flow, its depth and the pore water pressure, depending on
the terrain parameters. Here, the steady-state hydrological model of Iida (1984) has been used to cal-
culate the saturated flow depth (h) based on rainfall intensity, curvature, effective porosity, saturated
hydraulic conductivity and the horizontal velocity component of bedrock parallel flow,

R h  e  2i
hD tC Vs t ; (1)
n 2
GEOMATICS, NATURAL HAZARDS AND RISK 1561

Figure 2. Lithological map showing landslide distribution.

where
ks
Vs D sin ucos u; (2)
n

and h is the depth to saturated flow (m), R is the rainfall intensity (m/day), t is the time (day), e is the
curvature of a particular terrain cell (DEM’s second derivative) (m¡1),VS is the bedrock parallel
flow’s horizontal velocity component (m/day), n is the effective porosity, ks is the saturated hydrau-
lic conductivity (m/day), u is the slope angle (degree).
The stability condition of an infinite slope can be represented by relating the slope parallel com-
ponent of the gravitational force (shear stress) and the resisting frictional force, adjusted by pore
water pressure (shear strength) (Iverson 2000). FS is computed (Brunsden & Prior 1979) as

C tan ’ g w h cos2 u tan ’


FS D C ¡ ; (3)
g s Hsin u cos u tan u g s H sinu cos u

where C is the soil cohesion (KN/m2), ’ is the angle of internal friction (degree), g s is the unit weight
of soil (KN/m3), H is the depth-to-bedrock (m), h is the depth to saturated flow (m), u is the slope
angle (degree), g w is the unit weight of water (KN/m3).
Equations (1)(3) are used to compute the value of FS for the terrain. The analysis has been car-
ried out using ArcGIS 9.3 with the spatial analyst module. The advantage of the deterministic model
implemented in a GIS is the versatility with which rapid reassessment of FS of the terrain can done
in response to variations in the dynamic components like rainfall intensity. Since rainfall is highly
varying in time and space, the FS of the terrain also changes. In this study, specific hydrologic situa-
tions varying from dry condition to extreme wet conditions have been analyzed to understand the
response of the slope-forming material.
1562 J. MATHEW ET AL.

When the FS value of a terrain pixel falls below 1, it is assigned an ‘unstable’ condition in physical
terms. Conversely, any pixel with an FS value greater than 1 is assigned a ‘stable’ condition. In order
to represent the FS on a continuous measure of stability, it can be converted to slope failure proba-
bility as proposed by Terlien et al. (1995). This conversion is also useful for selective slope manage-
ment measures and also for assessing the accuracy of the model using receiver operating
characteristic (ROC) curve analysis. The deviation of a given FS value from FS D 1 is indicative of
the probability of slope failure, assuming normal distribution of FS values. The area under the curve
of the normalized deviation distribution gives the probability that FS is less than 1. When the FS
value of a give pixel is less than 1, the area under the standard normal curve (from 1 to the nor-
malized deviation value corresponding to that particular FS value) will be greater than 0.5, indicating
that the probability with which the FS falls below 1 (failure probability) is more than 0.5. This indi-
cates tendency towards slope failure. The case becomes opposite when the FS is greater than 1, and
the estimated failure probability of the cell will be less than 0.5. The conversion of the FS value to
the normalized deviation is done (van Westen 1993) as

ð1 ¡ FSÞ
ZD ; (4)
s

where Z is the Z score value, FS is the FS value of each terrain cell, s is the standard deviation of the
FS distribution.
The FS values have been converted to the corresponding failure probability in order to have an
output which can have multiple classes of landslide susceptibility, rather than a binary output with
stable and unstable areas. Such zonation is helpful for prioritizing the areas for slope stability man-
agement measures. The methodology flow chart is given in figure 3.

Figure 3. Methodology flow chart.


GEOMATICS, NATURAL HAZARDS AND RISK 1563

3. Analyses and results


3.1. Determination of input parameters
Various analyses to derive the input factors and their assimilation into the deterministic model as
described in the methodology section are summarized below.

3.1.1. Direct shear test


The cohesion and friction angles were determined using the direct shear test (ASTM D5321). Undis-
turbed soil samples were collected from the four zones in the field from pits dug to depths of
6070 cm from the four zones. Cylindrical, steel core cutters of 1 foot diameter and 1 foot length
were hammered into the soil without disturbing the natural setting and once the core has completely
penetrated the soil, it was removed with the intact soil sample inside, by scrapping off the contact
soil. The shear force ‘F’, at failure corresponding to the normal load ‘N’ has been measured and a
number of identical specimens had been tested under increasing normal loads for recording ‘F’ at
failure. The normal stress and shear stress values are given in table 1. A graph was plotted for the
calculated normal stress (s, along the X-axis) and the shear stress (t, along the Y-axis) values. The Y
intercept of the plot gives the cohesion of the soil sample and the angle of internal friction is the
inverse of the tangent of the ratio of shear stress (t) and normal stress (s) (Punmia 1987). The cohe-
sion and friction angles calculated for the four zones in the study area are given in table 2.
The soils developed over zone-4 underlain by quartzite show relatively high cohesion and fric-
tional angle values. On the other hand, the soils from zone-3 (granitic gneiss) show lower values of
cohesion and friction. The soil texture can be influential in determining the cohesion and friction
and the lower cohesion and friction of zone-3 relative to zone-4 could be because of textural
variations.

3.1.2. Soil texture


The particle size distribution analysis was carried out using sieve analysis and sedimentation analysis
(ASTM D422). Oven dried samples were used for fine sieve analysis and the percentage passing
through various sieves were calculated. The fraction finer than 75 micron size was subjected to sedi-
mentation analysis using hydrometer. The particle size distribution curve has been prepared using
the results of analysis and the proportion of various size fractions of the soils in the study area is
given in table 3. Zone-2 has maximum percentage of particles finer than 0.1 mm (56%), followed by
zone-4 (50%), zone-1 (40%) and zone-3 (30%). Based on the USDA textural classification (Buol
et al. 2003), these soils are sandy loam except in zone-2 which has silt loam. Soils of zone-4 are

Table 1. Normal and shear stress values of soil samples from the four zones.
Zone-1 Zone-2 Zone-3 Zone-4

Normal stress (kg/cm2) Shear stress (kg/cm2) Shear stress (kg/cm2) Shear stress (kg/cm2) Shear stress (kg/cm2)

0.25 0.292 0.278 0.257 0.319


0.5 0.5 0.472 0.424 0.531
0.75 0.703 0.656 0.578 0.734
1 0.906 0.844 0.75 0.938

Table 2. Cohesion and angle of internal friction of soils of various zones in the study area.
Geotechnical property Zone-1 Zone-2 Zone-3 Zone-4

Cohesion (KN/m2) 7.848 8.339 7.358 8.829


Angle of internal Friction (degrees) 39 38 36 39
1564 J. MATHEW ET AL.

Table 3. Particle size distribution of soils in the study area.

Size fraction Zone-1 Zone-2 Zone-3 Zone-4

Gravel (%) 24.50 16.00 23.00 10.00


Sand (%) 39.00 30.00 51.50 44.00
Silt (%) 33.50 48.00 23.50 42.50
Clay (%) 3.00 6.00 2.00 3.50

characterized by lesser amount of gravels compared to those from other zones. On the other hand,
the soils from zone-3 have high content of gravels and relatively higher proportion of sand (as com-
pared to the other zones). The least proportion of finer (than sand) particles makes this zone the
least cohesive. The texture information has been used in estimating saturated hydraulic conductivity
and effective porosity as mentioned in the following sections.

3.1.3. Saturated hydraulic conductivity


Saturated hydraulic conductivity (ks) is the rate of water movement through the soil under saturated
conditions and specific hydraulic gradient and it depends on the porosity, texture and structure of
the soil. ks of soil samples in study area were estimated based on an empirical approach. The soil
water characteristics (SWC) module of soil-plant-air-water (SPAW), version 6.02.75 software has
been used to estimate the saturated hydraulic conductivity of soils. The estimated saturated hydrau-
lic conductivity values of soils in the study area are given in table 4. The soil of zone-2 shows the
lowest values of saturated hydraulic conductivity and porosity. This can be explained in terms of
high proportion of finer particles (this zone has the highest silt and clay percentage).

3.1.4. Porosity
The porosity (n) values have been determined for the soil samples from different zones of the study
area using standard laboratory procedure; ASTM D7263. The total porosity values have been con-
verted to effective porosity by subtracting the field capacity values estimated using the SWC module
of SPAW. The total and effective porosity values are also given in table 4. The effective porosity
shows inverse relation to the proportion of percentage of finer particles. Thus, zone-2 has the least
effective porosity and zone-3 has the highest.

3.1.5. Unit weight of soil


Unit weight (g s ) of soil samples of different zones have been determined based on ASTM D7263 and
these values of soils in the four zones are given in table 4. Zone-2 has the highest unit weight for soil.
This zone is underlain by amphibolites and the presence of weathered residues of ferro-magnesium
minerals is the reason for higher unit weight. The quartzitic area (zone-4) shows the lowest unit
weight of soils.

3.1.6. Terrain parameters


Terrain elevation derived from analogue or digital photogrammetric techniques (Toutin & Gray
2000; Hirano et al. 2003; Kaab 2005; Roer et al. 2005) provides spatially continuous source for

Table 4. Saturated hydraulic conductivity, porosity and unit weight of soils in the study area.

Geotechnical property Zone-1 Zone-2 Zone-3 Zone-4

Saturated hydraulic conductivity (m/day) 1.611 1.194 2.256 1.986


Porosity (total) (%) 44.5 35.1 43.4 45.7
Porosity (effective) (%) 28.6 15.0 33.5 30.4
Unit weight of soil (KN/m3) 14.42 16.87 14.80 14.13
GEOMATICS, NATURAL HAZARDS AND RISK 1565

important derivatives like slope, its aspect, curvature, etc. A DEM of the study area has been gener-
ated using IRS Cartosat-1 stereo pair (Fore and Aft) images (Giribabu et al. 2013). Ten GCPs col-
lected using DGPS survey were used for the stereo processing, out of which seven had been used as
control points and the remaining had been used as check points. The average RMS error for the con-
trol points was 7.2 m in Z and less than 1 m in X and Y. Automatic tie points were generated with
correlation coefficient > D 0.82 and were augmented with manually added tie points. The DEM
was then generated with all tie points and control points as seed points. Considering the undulation
of the terrain, the DEM grid cell size was kept as 10 m. Terrain editing on the relevant portion of
DEM pertaining to the Garigaon watershed was executed using Leica Photogrammetric Suite. A
small segment in the eastern portion of the study area has interpolated elevation values due to the
complete masking of that area by shadow and the elevation values were generated by interpolation
of the surrounding elevation values.
The slope map of the area generated from the DEM shows that this range between 0 and 71
degrees. The study area has predominant slope in the 30 to 60 degree slope class. The gentler slopes
are found in the NW, western and eastern portions of the watershed whereas the southern, SE and
northern portions have predominantly steeper slopes. The slope curvature, which is the second
derivative of DEM, has also been derived and it has negative values indicating concave and positive
values indicating convex curvature. The study area has dominantly concave to planar curvature
reflecting dominant overland flow, absence of deep weathering and also the influence of bedding
and gneissocity planes on the morphology. Areas of predominantly higher weathering with deeper
soils (e.g. the western to NW portion) exhibit convex curvature.

3.1.7. Land cover, land use and landslides


The land use/land cover characteristics of the area influence the terrain stability through root cohe-
sion (positive influence of vegetation), by increasing the weight on the slope material (negative influ-
ence of vegetation) and through removal of slope support (negative influence of anthropogenic
activities like road cutting, construction activity, etc.) (Sidle et al. 2006). Orthorectified LISS IV data
of IRS Resourcesat-1 satellite has been used in this study for land use/land cover mapping through a
supervised classification method (maximum likelihood classification). The classes derived are dense
forest, moderate dense forest, open forest, agricultural land, scrub land, barren land and settlement
area. The land use/land cover map is shown below in figure 4.
The LISS IV (5.8 m resolution) and Cartosat-1 PAN data (2.5 m resolution) data were fused using
Brovey transformation and the resultant high-resolution data have been used for mapping the land-
slides. A total of 17 landslides have been mapped in the study area. All these are shallow landslides
involving the slope material above the bedrock. The locations of the landslides are shown on the
lithological map of the area (figure 2). Comparison with pre-monsoon satellite data has shown that
12 of these slides occurred during the unusually high rainfall event recorded on 15 August 2007.

3.1.8. Depth-to-bedrock
The depth-to-bedrock is the maximum depth of the slope material extending up to the hard rock
and defines the factor ‘H’ in equation (3). This is the maximum depth of slope failure in case of shal-
low landslides. The depth-to-bedrock measurements have been carried out at selected locations dur-
ing the field work through creation of trenches and by observations at existing slope cuttings. The
range of depth values and the average values considered for various land use/land cover classes have
been given in table 5. In order to compute the depth values across the study area (using weighted lin-
ear combinations), two more factors have been considered. The first factor is the terrain slope. The
soil depth is found to decrease with increasing slope in the area. This has been accounted by using
empirical depth reduction factors as given in table 6. Similarly, the underlying rock type also influen-
ces the soil formation. The limestone dominated area shows the maximum depth of the soil develop-
ment and the quartzitic areas show the least of the soil depth values. Similar to the depth reduction
factor defined for the slope categories, empirical depth reduction factors have been defined for
1566 J. MATHEW ET AL.

Figure 4. Land use/land cover map of the study area.

various lithological units (table 6). Within a given slope class, the variation of depth to bedrock
amongst the different lithological classes is only within 10%. However, for a given lithology, it has
been seen in the field that the depth to bedrock variation is up to 70% from flat areas to steep slope
areas. The depth reduction factors account for this relation also.

Table 5. Depth-to-bedrock values of various land use/land cover classes from field observations.
Land use/land cover class Depth range (m) Average depth (m)

Dense forest 2.02.4 2.2


Moderately dense forest 1.52.0 1.8
Open forest 1.01.5 1.4
Agriculture 1.82.2 2.0
Scrub land 0.91.3 1.2
Barren land 0.10.4 0.3
Settlement 1.82.2 2.0

Table 6. Depth reduction factors of various slope and lithological categories.

Depth reduction factors for slope classes Depth reduction factors for lithological classes

Slope class (degrees) Depth reduction factor Lithology Depth reduction factor

010 1.00 Limestone 1.00


1020 0.95 Ambhibolite 0.96
2030 0.85
3045 0.75 Gneiss 0.94
4560 0.60
> 60 0.30 Quartzite 0.92
GEOMATICS, NATURAL HAZARDS AND RISK 1567

Figure 5. Depth-to-bedrock map of the study area.

Using the depth reduction factors and observed depth-to-bedrock values, the depth-to-bedrock
map of the study area has been prepared as given in figure 5. The estimated depth-to-bedrock values
range from 0.12 to 2.11 m. High depth-to-bedrock values are in the forested areas which are gently
sloping. The NW, W, NE and eastern portions of the watershed show depth to bedrock values reach-
ing to the maximum of 2.11 m. The steep areas occupied by quartzite have relatively lesser depth to
bedrock values. This is clearly visible in the northern and central portions of the watershed where
the slope is more than 40 degrees and are either barren or covered with scrub or open forest (< 10%
canopy cover).

3.1.9. Rainfall return period


Return period analysis (Subramanya 2001) of long-term (1976 to 2007) daily rainfall data from
nearby rain gauge stations at Chamoli (for the duration from 1976 to 2003) and Pipalkoti (figure 1)
(for the duration from 2004 to 2007) have been carried out to determine rainfall events of specific
recurrence interval. The annual rainfall values for this period ranged from 900 to 1533 mm with
average value of 1140 mm. Gumbel’s (2004) extreme value distribution method was used for the
return period analysis of rainfall intensity values corresponding to the return periods of 10, 25 and
50 years. These rainfall intensities are 166.23, 208.28 and 239.5 mm/day, respectively. The value for
the 5-year return period (132.9 mm/day) is very close to the actual rainfall intensity of 15 August
2007 (135 mm/day). Since the FS varies with the dynamic factor of depth of saturated flow, it has
been computed for various hydrological conditions varying from dry condition, actual rainfall event
of 15 August 2007 and rainfall intensities corresponding to return periods of 10, 25 and 50 years. In
addition, two more rainfall intensity values (50 and 100 mm/day) have also been used to understand
the trend of stability condition for rainfall intensity values below 135 mm/day. The hydrological
conditions used in the study are given in table 7.
1568 J. MATHEW ET AL.

Table 7. Rainfall intensity values considered in the analysis.

Sl no. Event Rainfall intensity (mm/day)

(1) Dry condition 0


(2) Rainfall intensity value 50
(3) Rainfall intensity value 100
(4) Rainfall recorded on 15 August 2007 135.00
(5) Gumbel’s 10 year return period 166.23
(6) Gumbel’s 25 year return period 208.28
(7) Gumbel’s 50 year return period 239.50

3.2. Estimation of factor of safety and failure probability


The parameters computed as described in the previous sections have been input into spatial analysis
to compute the FS of the terrain. The raster grid size for the analysis has been kept as 10 m. The
effective porosity and saturated hydraulic conductivity values have been converted to spatial data by
assigning their values to the respective zone of the study area. Using the slope, effective porosity and
saturated hydraulic conductivity, the subsurface flow velocity (equation 2) raster has been com-
puted. Using this as input, along with the curvature of the terrain, the saturated flow depth (equation
1) raster layers have been computed for the hydrological conditions mentioned in table 7. The next
step has been the estimation of FS values using the saturated flow depth raster layers corresponding
to different hydrological conditions, through equation 3. The cohesion, angle of internal friction
and unit weight of soil parameters have been rasterized for compatibility in spatial analysis.
The dry condition has not yielded any unstable pixel. Corresponding to the rainfall intensity
value of 135 mm/day on 15 August 2007, the FS values show that 18.5% of the pixels in study area
are unstable (figure 6). The FS estimated using the rainfall values corresponding to return periods of
10, 25 and 50 years show that nearly 21.7%, 23.5% and 24.7%, respectively, become unstable at these
rainfall intensity conditions. The percentage of unstable areas at rainfall intensities of 50 and
100 mm/day are 7.5% and 13.8%, respectively.
This study has revealed that the percentage of unstable areas with FS less than unity (table 8)
shows initial rapid increase with increasing rainfall intensity and stagnates after reaching a critical
value. This trend continues to about 150 mm/day (figure 7). The second-order polynomial trend-
line (dark line in figure 7) shows that the intensity corresponding to the maximum unstable area of
25% is about 260 mm/day.
This study also evaluated the FS values in terms of failure probability. The FS values of the pixels
have been converted to Z scores and then the corresponding failure probability values have been
estimated using tabled values of standard normal distribution. For example, for the rainfall intensity
value of 135 mm/day, the failure probability values which are classified based on breaks in their fre-
quency distribution are given in figure 8. It shows three classes with failure probabilities ranging
from 0 to 0.36, 0.36 to 0.5 and more than 0.5, as stable, marginally unstable and unstable areas,
respectively.

3.3. Accuracy assessment


Accuracy assessment of the slope stability model used in this study has been carried out by using the
ROC curve analysis. ROC curve represents a plot of probability of true positive (a slided terrain cell
is correctly classified) cases of landslide cells (sensitivity) along the Y-axis against the probability of
false positive (a non-slided terrain cell is incorrectly classified) cases of landslide cells ([1-specificity],
where specificity is the true negative rate) along the X-axis, with varying cut-off probability (Mathew
et al. 2007b). The area under the curve represents the probability that the model calculated failure
probability value for a randomly chosen slided cell will exceed that of a randomly chosen non-slided
GEOMATICS, NATURAL HAZARDS AND RISK 1569

Figure 6. Factor of safety map of the study area for 135 mm/day rainfall intensity.

cell. Thus, the area under the ROC curve can be used as a measure of the accuracy or the success rate
of the model (Hanley & McNeil 1982).
For the ROC curve analysis, the failure probability values (corresponding to the rainfall intensity
of 15 August 2007, i.e. 135 mm/day) of landslide pixels (431 cells) have been compared with that of
an equal number of randomly selected non-landslide pixels in the study area. The input for the
ROC curve analysis is a table that contains the model derived failure probabilities of the samples
(both the slided cells and the randomly selected non-slided cells) and their actual state of stability
condition as 1 (slided) and 0 (non-slided) (Mathew et al. 2007b). The ROC curve of the model is
shown in figure 9. The area under the ROC curve is 0.842, giving an accuracy of 84.2% for the model
used in this study for the rainfall event of 15 August 2007.

Table 8. Unstable terrain areas under different hydrological conditions.


Unstable area

Rainfall intensity (mm/day) In sq. km In percentage

0.00 0 0.00
50 0.485 7.52
100 0.891 13.82
135.00 1.190 18.45
166.23 1.399 21.69
208.28 1.517 23.52
239.50 1.592 24.68
1570 J. MATHEW ET AL.

Figure 7. Percentage of unstable area vs. rainfall intensity.

Figure 8. Failure probability map of the area for 135 mm/day rainfall intensity.
GEOMATICS, NATURAL HAZARDS AND RISK 1571

Figure 9. Receiver operating characteristic curve.

4. Discussion
Process-based slope stability analyses require extensive geotechnical data and large-scale topograph-
ical information for calculating FS of the terrain and hence it is usually applied to small areas. The
infinite slope stability model assumes the slope profile length to be longer than the depth of the slope
material and also considers the failure plane to be planar. The terrain condition and the observed
slope failures in the Garigaon watershed are in conformity with these assumptions. One of the chal-
lenges in implementing the deterministic model is the estimation of various geotechnical parame-
ters. The in situ determination of geotechnical parameters, collection of samples and their
laboratory analyses are time consuming, labour intensive and physically challenging in inaccessible
terrains like that in the Himalaya. Due to the constraints related to ruggedness and inaccessibility of
the terrain and also due to the constraints in analytical infrastructure, only representative samples
have been analyzed for different zones, defined and limited by the underlying rock type. Hence, the
geotechnical parameters have been determined and represented zone-wise in the study area where
each zone is assumed to be covered by slope-forming material of homogeneous geotechnical proper-
ties. Though it is not the optimum or ideal sampling strategy for deterministic slope stability studies,
it provides input for rapid assessment of the FS of the terrain under varying hydrological conditions.
The results indicate a sudden increase in the percentage unstable area from 7.5% to 13.8% for
rainfall intensity variation from 50 to 100 mm/day. This range is significant as it conforms to the
intensity-duration thresholding of rainfall-induced landslides (Mathew et al. 2014) along the road
corridor from Rishikesh to Mana, which passes through close to Garigaon watershed. The authors
described in that study that a rainfall intensity of 4 mm/hr for 12 hours (i.e. 48 mm in 12 hours or
96 mm/day) is the threshold for slope failure initiation along the road corridor.
Distribution of unstable areas in case of the hydrological condition of 135 mm/day (figure 6) in
various lithological units shows that the granitic gneiss areas (figure 2) had the maximum percentage
of unstable area (23.1%) and the quartzitic areas had the least percentage of unstable area (10.4%).
Out of the remaining areas, those underlain by amphibolite had 16% of its area in the unstable cate-
gory, whereas the areas underlain by limestone had about 13.7% of its area having FS less than 1 at
this rainfall intensity. The granitic gneiss areas which yielded the maximum unstable area for
1572 J. MATHEW ET AL.

135 mm/day rainfall intensity had lowest values of cohesion and angle of internal friction (7.358
KN/m2 and 36 degrees, respectively). This area is also characterised by highest value of saturated
hydraulic conductivity (2.256) and effective porosity (33.5%). The quartzitic areas which exhibited
the least proportion of unstable area have higher angle of internal friction (39 degrees) and cohesion
(8.829 KN/m2) compared to the granitic gneiss areas. The proportion of unstable areas showed the
same relative trend for higher and lower rainfall intensities. For example, for the rainfall intensity
corresponding to 50 year return period, the unstable area has increased to 27.8% (highest) and
18.2% (lowest) of the gneissic and quartzitic areas, respectively. These values are 9% and 4.5%,
respectively, for gneissic and quartzitic areas at 50 mm/day of rainfall intensity.
Distribution of FS values with respect to slope and rainfall intensity reveals that incremental rain-
fall intensity causes progressively gentler slopes to become unstable. This is due to the fact that for a
given duration, as the quantity of rainfall increases, the saturated flow depth also increases, leading
to lowering of the FS. Higher value of saturated flow depth also means that gentler slopes get
included in the saturated zone. This is revealed from the slope distribution in the unstable zone (FS
< 1) for different rainfall intensities. The slope range in the unstable zone for rainfall intensity of
50 mm/day is 37 to 54 degrees, and for 100 mm/day, it is 34 to 55 degrees. For 135 mm/day of rain-
fall intensity, the unstable slopes range between 31 and 56 degrees whereas at 239.5 mm/day, the
unstable areas’ slope distribution is between 25 and 56 degrees. Additional unstable areas for higher
rainfall intensities are extensions (involving mainly gentler slopes) of areas which are stable for
lower intensities. The mean slope of the unstable areas for different rainfall intensities shows only
marginal variation. For example, at 50 mm/day, the mean slope of unstable area is 43 degrees, at
135 mm/day, the mean slope is 41.8 degree, whereas at 239.5 mm/day, the mean slope of areas with
FS < 1 is about 40.2 degree. It has also been seen that the upper limit of slope of the unstable areas
has not increased beyond 56 degree for higher rainfall intensities. Though steep, these slopes remain
stable, unless disintegrated and disturbed by discontinuities.
Figure 8 gives idea about the failure probability of the terrain for the rainfall intensity of
135 mm/day. At this scenario, the areas marked as stable do not require slope stabilization measures.
For example, the NW corner of the watershed, some of the areas of in the eastern portion and NE
corner of the watershed tend to be stable at this rainfall intensity. It also shows that the marginally
unstable areas are extensions of the unstable areas, towards gentler slopes. This is clear from the
eastern, southern and western portions of the watershed. The unstable and marginally unstable areas
are those which require slope management measures to avoid extreme events of mass wasting. For
example, aggregation of unstable and marginally unstable in the western and central portions of the
watershed could lead to potentially damaging landslides of dimensions to the order of about 0.9
km2 area at intense rainfall events.
The behaviour of FS values in response to higher rainfall intensity values are direct reflection of
the influence of the saturated flow depth (h, which higher for higher rainfall intensities, but limited
to H, depth-to-bedrock). As the saturated flow depth increases, the FS values drop, which makes
more areas unstable and this continues until the critical rainfall intensity has been reached, which is
about 260 mm/day for the study area.

5. Conclusion
This study demonstrates an integrated slope stability hydrological model for estimating the FS and
its conversion to failure probability of the slope-forming material for various hydrological condi-
tions. The geotechnical parameters which have been determined in laboratory and in the field,
together with the topographical and hydrological parameters form the input for the model. The
results show that for various rainfall intensity values, the maximum unstable area is presented over
the slopes underlain by granitic gneisses where the slope-forming material have the lowest cohesion
and angle of internal friction values compared to the areas underlain by the other three types of
lithological units. The slopes underlain by quartzite are the most stable in Garigaon watershed.
GEOMATICS, NATURAL HAZARDS AND RISK 1573

These materials on such slopes display the highest cohesion and have low degree of weathering.
Rainfall intensity has definite influence in the slope stability and it has been seen in this study that
higher rainfall intensities make progressively gentler slopes unstable, but limited to about 25 degree
slope in the study area. It has also been found that the percentage of unstable area (FS < 1), which is
nil for dry condition, increases with rainfall intensity, but not beyond for an intensity value of about
260 mm/day. The deterministic model in this study has shown an accuracy of 84.2% and can be
used to evaluate the slope stability condition of the area for different hydrological conditions includ-
ing for those with long return periods or extreme events. Using the conversion from FS to failure
probability used in this study, this model can be used to delineate areas which would be marginally
unstable or unstable, for rainfall intensity values corresponding to different return periods, so that
appropriate slope management measures can be planned in advance.

Acknowledgments
The authors are thankful to Dr V. K. Dadhwal, Director, NRSC, and Dr P. G. Diwakar, Deputy Director, NRSC, for
their motivation to carry out this study. Dr A. Ghosh, Head, Geo-technical Engineering Division, CBRI, Roorkee, has
permitted the soil analyses at CBRI, Roorkee. John Mathew and S. Kundu acknowledge the use of rainfall data col-
lected from Border Roads Organization in this study. The district administration of Chamoli has also shared the avail-
able rainfall data. The critical comments from the anonymous reviewers have helped to consolidate the study in a
focused way.

Disclosure statement
No potential conflict of interest was reported by the authors.

References
Acharya G. 2003. GIS approach for slope stability risk analysis: a case study from Nepal [M.S. dissertation]. Brussels
(Belgium): Free University of Brussels.
Acharya KP, Yatabe R, Bhandary NP, Dahal RK. 2014. Deterministic slope failure hazard assessment in a model catch-
ment and its replication in neighbourhood terrain. Geomatics, Nat Hazards Risk. [Internet]. [cited 2014 Feb 14];
30. Available from: http://dx.doi.org/10.1080/19475705.2014.880856
Agrawal DK, Lodhi MS, Panwar S. 2010. Are EIA studies sufficient for projected hydropower development in Indian
Himalayan region? Curr Sci. 98:154161.
Baum RL, Savage WZ, Godt JW. 2002. TRIGRS - A FORTRAN programme for transient rainfall infiltration and grid-
based regional slope stability analysis. USGS Open File report 02-424.
Beven KJ, Kirkby MJ, Schofield N, Tagg AF. 1984. Testing a physically-based flood forecasting model (TOPMODEL)
for three UK catchments. J Hydrol. 69:119143.
Beven KJ, Kirkby MJ. 1979. A physical based variable contributing area model of basin hydrology. Hydrol Sci Bull.
24:4369.
Beven KJ, Lamb R, Quinn P, Romanowicz R, Freer J. 1995. TOPMODEL. In: Singh VP, editor. Computer models of
watershed hydrology. Colorado (CO): Water Resources Publications; p. 627668.
Borga M, Fontana, GD, Gregoretti C, Marchi L. 2002. Assessment of shallow landsliding by using a physically based
model of hillslope stability. Hydrol Process. 16:28332851.
Brabb EE, Harrod BL, editors. 1989. Landslides: extent and economic significance. Rotterdam (the Netherlands): AA
Balkema.
Brunsden D, Prior DB, editors. 1979. Slope instability. Chichester (UK): John Willey and sons.
Buchroithner MF. 2002. Meteorological and earth observation remote sensing data for mass movement preparedness.
Adv Space Res. 29:516.
Buol SW, Southard RJ, Graham RC, McDaniel PA. 2003. Soil genesis and classification. Ames (IA): Willey Blackwell
Publishing.
Casadei M, Dietrich WE, Miller NL. 2003. Testing a model for predicting the timing and location of shallow landslide
initiation in soil-mantled landscapes. Earth Surf Process Landf. 28:925950.
Cervi F, Berti M, Borgatti L, Ronchetti F, Manenti F, Corsini A. 2010. Comparing predictive capability of statistical
and deterministic methods for landslide susceptibility mapping: a case study in the northern Apennines (Reggio
Emilia Province, Italy). Landslides. 7:433444.
1574 J. MATHEW ET AL.

Dahal RK, Hasegawa S, Nonomura A, Yamanaka M, Dhakal S. 2008. DEM-based deterministic landslide hazard anal-
ysis in the Lesser Himalaya of Nepal. Georisk. 2:161178.
Dahal RK, Hasegawa S, Nonomura A, Yamanaka M, Dhakal S, Bhandary NP, Yatabe R. 2009. Comparative analysis of
contributing parameters for rainfall-triggered landslides in the lesser Himalaya of Nepal. Environ Geol.
58:567586.
Dhakal AS, Siddle RC. 2004. Distributed simulations of landslides for different rainfall conditions. Hydrol Process.
18:757776.
Dietrich WE, Reiss R, Hsu M, Montgomery DR. 1995. A process-based model for colluvial soil depth and shallow
landsliding using digital elevation data. Hydrol Process. 9:383400.
Duan J. 1996. A coupled hydrologic-geomorphic model evaluating the effects of vegetation change in watersheds
[Ph.D. dissertation]. Corvallis (OR): Oregon State University.
Ellen SD, Wieczorek GF. 1982. Landslides, flood and marine effects of the storm of January 3-5, 1982 in the San Fran-
sisco Bay region, California. USGS professional paper, 1434, p. 310.
Fall M, Azzam R, Noubactep C. 2006. A multi-method approach to study the stability of natural slopes and landslide
susceptibility mapping. Eng Geol. 82:241263.
Formetta G, Ragob V, Capparellia G, Rigonc R, Mutob F, Versacea P. 2014. Integrated physically based system for
modeling landslide susceptibility. Procedia Earth Planet Sci. 9:7482.
Franchini M, Wendling J, Obled C, Todini E. 1996. Physical interpretation and sensitivity analysis of the TOPMO-
DEL. J Hydrol. 175:293338.
Galang J. 2004. A Comparison of GIS Approaches to slope instability zonation in the central Blue Ridge mountains of
Virginia [ Ph.D. dissertation]. Blacksburg (VA): Virginia Polytechnic Institute and State University.
Giribabu D, Kumar P, Mathew J, Sharma KP, Murthy YVNK. 2013. DEM generation using Cartosat-1 stereo data:
issues and complexities in Himalayan terrain. Eur J Remote Sens. 46:431443.
G€okceoglu C, Aksoy H. 1996. Landslide susceptibility mapping of the slopes in the residual soils of the Mengen region
(Turkey) by deterministic stability analyses and image processing techniques. Eng Geol. 44:147161.
Godt JW, Baum RL, Savage WZ, Salciarini D, Schulz WH, Harp EL. 2008. Transient deterministic shallow
landslide modeling: requirements for susceptibility and hazard assessments in a GIS framework. Eng Geol.
102:214226.
Gumbel EJ. 2004. Statistics of extremes. New York: Dover Publication.
Hammond C, Hall D, Miller S, Swetik P. 1992. Level I stability analysis (LISA): documentation for version 2.0, gen.
Technical Report INT-285, USDA, Forest Service, Intermountain Research Station; p. 190.
Hanley JA, McNeil BJ. 1982. The meaning and use of the area under a receiver operator characteristic (ROC) curve.
Radiology. 143:2936.
Hirano A, Welch R, Lang H. 2003. Mapping from ASTER stereo image data: DEM validation and accuracy assess-
ment. ISPRS J Photogramm Remote Sens. 57:356370.
Hsu M. 1994. A grid-based model for predicting soil depth and shallow landslides [Ph.D. dissertation]. Berkeley (CA):
U.C.
Iida T. 1984. A hydrological method of estimation of the topographic effect on the saturated throughflow. Trans Jpn
Geomorphol Union. 5:112.
Iverson RM. 2000. Landslide triggering by rain infiltration. Water Resour Res. 36:18971910.
Kaab A. 2005. Combination of SRTM3 and repeat ASTER data for deriving alpine glacier flow velocities in the Bhutan
Himalaya. Remote Sens Environ. 94:463474.
Kainthola A, Singh PK, Singh TN. 2014. Stability investigation of road cut slope in basaltic rockmass, Mahabaleshwar,
India. Geosci Front. [Internet]. [cited 2014 Mar 17]; 9. Available from: http://dx.doi.org/10.1016/j.gsf.2014.03.002
Kainthola A, Singh PK, Wasnik AB, Sazid M, Singh TN. 2012. Finite element analysis of road cut slopes using Hoek &
Brown failure criterion. Int J Earth Sci Eng. 5:11001109.
Kainthola A, Verma D, Singh TN. 2013. Probabilistic and sensitivity investigation for the hill slopes in Uttarakhand,
lesser Himalaya, India. Am J Numer Anal. 1:814.
Kniveton DR, De Graff, PJ, Granica K, Hardy RJ. 2000. The development of a remote sensing based technique to pre-
dict de-bris flow triggering conditions in the French Alps. Int J Remote Sens. 21:419434.
Kumar G. 2005. Geology of Uttar Pradesh and Uttaranchal. Bangalore: Geological Society of India.
Kuthari S. 2007. Establishing precipitation thresholds for landslide initiation along with slope characterisation using
GIS-based modeling [M.Sc. dissertation]. Enschede: ITC.
Lamb R, Beven KJ, Myrabo S. 1998. A generalized topographic-soil hydrological index. In: Lane S, Richard K,
Chandler J, editors. Landform monitoring, modelling and analysis. Chichester: Wiley; p. 263278.
Lee S, Hyun-Joo Oh. 2012. Ensemble-based landslide susceptibility maps in Jinbu area, Korea. In: Pradhan B,
Buchroithner MF, editors. Terrigenous mass movements. Berlin: Springer; p. 193220.
Mann U, Pradhan B, Prechtel N, Buchroithner MF. 2012. An automated approach for detection of shallow landslides
from LiDAR derived DEM using geomorphological indicators in a tropical forest. In: Pradhan B, Buchroithner
MF, editors. Terrigenous mass movements. Berlin: Springer; p. 122.
GEOMATICS, NATURAL HAZARDS AND RISK 1575

Mantovani F, Soeters R, Van Westen CJ. 1996. Remote sensing techniques for landslide studies and hazard zonation in
Europe. Geomorphology. 15:213225.
Mathew J, Jha VK, Rawat GS. 2007a. Weights of evidence modelling for landslide hazard zonation mapping in part of
Bhagirathi valley, Uttarakhand. Curr Sci. 92:628638.
Mathew J, Jha VK, Rawat GS. 2007b. Application of binary logistic regression analysis and its validation for landslide
susceptibility mapping in part of Garhwal Himalaya, India. Int J Remote Sens. 28:22572275.
Mathew J, Jha VK, Rawat GS. 2009. Landslide susceptibility zonation mapping and its validation in part of Garhwal
lesser Himalaya, India, using binary logistic regression analysis and receiver operating characteristic curve method.
Landslides. 6:1726.
Mathew, J, Babu DG, Kundu S, Kumar KV, Pant CC. 2014. Integrating intensityduration-based rainfall threshold
and antecedent rainfall-based probability estimate towards generating early warning for rainfall-induced landslides
in parts of the Garhwal Himalaya, India. Landslides. 11:575588.
Mondal S, Maiti R. 2012. Application of 1-D slope stability model in landslide susceptibility mapping of Shivkhola
watershed, Darjiling Himalaya. Int J Geol, Earth Environ Sci. 2:3450.
Montgomery DR, Dietrich MT. 1994. A physically based model for the topographic control of shallow landsliding.
Water Resour Res. 30:11531171.
Moon V, Blackstock H. 2004. A methodology for assessing landslide hazard using deterministic stability models. Nat
Hazards. 32:111134.
Muntohar AS, Liao HJ. 2009. Analysis of rainfall-induced infinite slope failure during typhoon using a hydrologi-
calgeotechnical model. Environ Geol. 56:11451159.
Okimura T, Kawatani T. 1987. Mapping of the potential surface failure sites on granite mountain slopes. In: Gardiner
V, editor. International geomorphology part I. New York: Wiley; p. 121138.
O’Loughlin EM. 1986. Prediction of surfaces saturation zones in natural catchments by topographic analysis. Water
Resour Res. 22:794804.
Pack RT, Tarboton DG, Goodwin, CN. 1998. SINMAP- A stability index approach to terrain stability hazard mapping
(users manual). Logan (UT): C.N. Goodwin Fluvial System Consulting.
Pack RT, Tarboton DG, Goodwin CN. 2001. SINMAP approach to terrain stability mapping. In: Moore DP, Hungr O,
editors. Proceedings of International Conference of International Association for Engineering Geology and
Environment; 1998 Sep 2125; Vancouver (BC), Canada. Rotterdam: AA Balkema.
Pourghasemi HR, Pradhan B, Gokceoglu C, Moezzi KD. 2012. Landslide susceptibility mapping using a spatial multi
criteria evaluation model at Haraz watershed, Iran. In: Pradhan B, Buchroithner MF, editors. Terrigenous mass
movements. Berlin: Springer; p. 2349.
Prochaska AB, Santi PM, Higgins JB, Cannon SH. 2008. A study of methods to estimate debris flow velocity. Land-
slides. 5:431444.
Punmia BC. 1987. Soil mechanics and foundations. Delhi: Standard book house.
Qui C, Esaki T, Xie M, Mitani Y, Wang C. 2007. Spatio-temporal estimation of shallow landslide hazard triggered by
rainfall using a three-dimensional model. Environ Geol. 52:15691579.
Roer I, Kaab A, Dikau R. 2005. Rockglacier kinematics derived from small-scale aerial photography and digital air-
borne pushbroom imagery. Zeitschrift fur Geomorphologie. 49:7387
Safaei M, Omar H, Huat BK, Yousof ZBM, Ghiasi V. 2011. Deterministic rainfall induced landslide approaches,
advantages and limitation. Electron J Geotech Eng. 16:16191650.
Sharma RH, Nakagawa H. 2005. Shallow landslide modeling for heavy rainfall events. Kyoto (Japan): Annuals of
Disaster Prevention Research Institute, Kyoto University.
Sharma RH, Shakya NM. 2008. Rain induced shallow landslide hazard assessment for un-gauged catchments. Hydro-
geol J. 16:871877.
Sidle RC, Ziegler AD, Negishi JN, Nik AR, Siew R, Turkelboom F. 2006. Erosion processes in steep terrain—
truths, myths, and uncertainties related to forest management in Southeast Asia. For Ecol Manag.
224:199225.
Simoni S, Zanotti F, Bertoldi G, Rigon R. 2008. Modelling the probability of occurrence of shallow landslides and
channelized debris flows using GEOtop-FS. Hydrol Process. 22:532545.
Singh PK, Kainthola A, Singh TN. 2015. Rock mass assessment along the right bank of river Sutlej, Luhri, Himachal
Pradesh, India. Geomatics, Nat Hazards Risk. 6:212223.
Singhroy V, Molch K. 2004. Characterizing and monitoring rockslides from SAR techniques. Adv Space Res.
33:290295.
Subramanya K. 2001. Engineering hydrology. Calcutta: Tata.
Terlien MTJ. 1998. The determination of statistical and deterministic hydrological landslide-triggering thresholds.
Environ Geol. 35:124130.
Terlien MTJ, van Westen CJ, Asch ThWJ. 1995. Deterministic modelling in GIS-based landslide hazard assessment.
In: Carrara A, Guzzetti F, editors. Geographical information systems in assessing natural hazards. Dordrecht (the
Netherlands): Kluwer Academic Publishers; p. 5777.
1576 J. MATHEW ET AL.

Toutin T, Gray L. 2000. State-of-the-art of elevation extraction from satellite SAR data. ISPRS J Photogramm Remote
Sens. 55:1333.
Tucker GE, Gasparini NM, Bras RL, Lancaster SL. 1999. A 3D computer simulation model of drainage basin and
floodplain evolution: theory and applications. Technical report prepared for U.S. Army Corps of Engineers Con-
struction Engineering Research Laboratory.
Tucker G, Lancaster S, Gasparini N, Bras R. 2001. The channel-hillslope integrated landscape development
model (CHILD). In: Harmon RS, Doe WW, editors. Landscape erosion and evolution modeling. Dordrecht
(the Netherlands): Kluwer Academic Publishers; p. 349388.
Vanacker V, Vanderschaeghe M, Govers G, Willems E, Poesen J, Deckers J, De Bievre B. 2003. Linking hydrological,
infinite slope stability and land-use change models through GIS for assessing the impact of deforestation on slope
stability in high Andean watersheds. Geomorphology. 52:299315.
Van Beek LP, Van Asch TJ. 2004. Regional assessment of the effects of land-use change on landslide hazard by means
of physically based modelling. Nat Hazards. 31:289304.
van Westen CJ. 1993. Application of geographical information systems to landslide hazard zonation. Enschede: ITC
Publication.
van Westen CJ, Terlien MJT. 1996. An approach towards deterministic landslide hazard analysis in GIS. A case study
from Manizales (Colombia). Earth Surf Process Landf. 21:853868.
van Westen CJ, Jaiswal P, Gosh S, Martha TR, Kuriakose SL. 2012. Landslide inventory, hazard and risk assessment in
India. In: Pradhan B, Buchroithner MF, editors. Terrigenous mass movements Berlin: Springer; p. 239282.
Wilkinson PL, Anderson MG, Lloyd DM. 2002. An integrated hydrological model for rain-induced landslide predic-
tion. Earth Surf Process Landf. 27:12851297
Xie M, Esaki T, Qiu Cheng, Wang C. 2006. Geographical information system-based computational implementation
and application of spatial three-dimensional slope stability analysis. Comput Geotech. 33:260274.
Xie M, Esaki T, Zhou G. 2004. GIS-based probabilistic mapping of landslide hazard using a three-dimensional deter-
ministic model. Nat Hazards. 33:265282.
Zhang LL, Zhang LM, Tang WH. 2005. Rainfall-induced slope failure considering variability of soil properties. Geo-
technique. 55:183188.

You might also like