CHAPTER 4: METHODOLOGY FOR LANDSLIDE
ASSESSMENT
“Research means that you don’t know, but are willing to find out”
- Charles F. Kettering
This chapter provides the various methodologies adopted in the research work at various
levels of study (regional, sub-regional and plot-scale). The research work tries to follow
the hierarchical approach of landslide assessment from top to bottom (small to large scale).
The regional scale and sub-regional scale/level of study are carried out to assess the
landslide susceptibility using AHP and FR approaches. Considering the specific condition
of data availability, the plot scale study was conducted using a geological based slope-mass
rating approach to understand the slope failure mechanism at identified past landslide
locations. An overview of the methodologies for each level of study is presented in this
chapter although; the details of the outcomes and procedures will be explained in the later
chapters.
4.1 Regional Scale Methodology
Different methodological approaches have been adopted for the landslide susceptibility
mapping (LSM) at different scale of analysis. The selection of methodology is based on
the type and nature of study, the scale of study, and data availability. Although the advance
methodologies such as Machine Learning and Deep Learning are not adopted as, they
require detailed and large number of good quality data for producing accurate and reliable
results. Discussions of the accuracies of the various model are still continued (El Jazouli et
al. 2019) and the best LSM approach cannot be declared (Goetz et al. 2015). The capability
of AHP model in producing good LSM using a few available data and its relative
effectiveness in regional-scale LSM assessment, which allows fast and practical study, has
encouraged the utilization of AHP in this research work. While FR approach is statistical
method, which assumes that occurrence of landslides are directly related to the factors
related to landslides. The simplicity of FR approach, which allows practical analysis using
55
the relationship of landslide factors and landslide occurrence, has urged to adopt this
method.
The AHP and FR approaches are adopted for the regional-scale LSM assessment. The
regional scale LSM assessment involves the utilization of eight thematic data layers-
LULC, NDVI, slope, curvature, elevation, aspect, soil type, and rainfall. The LSM
assessment is conducted at spatial resolution of 30m. The regional scale analysis is carried
out at two different stages- the first stage analysis is done to develop the LSM of the whole
region for understanding the potential landslide area using two different approaches AHP
and Frequency Ratio. In addition, the second stage of analysis is conducted to analyse the
effect of landslide contributing factors in the LSM of the region. The methodological
framework for the Regional-scale assessment is shown in Fig 4-1.
Different sets of data are obtained from various data sources. The LANDSAT imageries
are used to developed LULC and NDVI maps. Various topographic factors, elevation,
slope, aspect, and curvature are derived from the DEM date while rainfall layer is derived
from Tropical Rainfall Measuring Mission (TRMM) rainfall data and the soil information
layer is derived from NBSS&LUP (ICAR). The derived thematic data are assigned weights
using AHP approach and finally the weighted factors are integrated to develop landslide
susceptibility maps (LSMs) of the region. Validation of the LSMs are done using the
landslide inventory and comparisons of the accuracies are also performed to find out the
more accurate and reliable LSM.
4.1.1 Landslide susceptibility mapping using AHP
The Analytic Hierarchy Process (AHP) is a semi-qualitative approach. It is one of the most
predominant multi-criteria decision making procedure in which the pairwise comparison
of causative factors is used to derive priority scales (Saaty 2008). Many decision-making
methods has utilized and successfully adopted AHP from the past decades. AHP translates
complex decision-making problems as hierarchy consisting of the factors and their
alternatives.
56
DATA PREPARATION
LANDSAT 8 TRMM SOIL MAP DEM DATA
MAGES DATA
SA
NDVI RAINFALL SOIL TYPE SLOPE
PARAMETER MAPS
LULC ELEVATION
CURVATURE
ASPECT
M
ANALYTIC HIERARCHY PROCESS FREQUENCY RATIO ANALYSIS
LANDSLIDE SUSCEPTIBILITY MAP LANDSLIDE SUSCEPTIBILITY MAP
VALIDATION USING VALIDATION USING
LANDSLIDE LOCATIONS LANDSLIDE LOCATIONS
COMPARISION
FINAL LANDSLIDE SUSCEPTIBILITY
MAP OF MANIPUR
Fig 4-1 Methodological framework for Regional landslide susceptibility assessment
(LSM)
57
Decision-making involves both subjective and objective approach (Yalcin 2008). The
pairwise comparison of the causative factors is used to assign weights to the factors and
alternatives based on a nine-point continuous scale Table 4-1.
Table 4-1 The fundamental scale of judgment (Saaty 1977)
Ratings Degree of Description
preference
Equally important
1 Equal
3 Moderate One factor moderately favored over another
5 Strong Judgment strongly favors one factor over other
Very strongly favored to one over other
7 Very strong
Extreme Judgment extremely favors one factor over other
9
importance with highest degree
2, 4, 6, 8 In between Compromise needed for weights 1, 3, 5, 7 and 9
Reciprocals Converse A reasonable assumption
The weights of factors and their alternatives are computed as Eigen vector/Eigen value. A
significant characteristic of AHP is the calculation of the consistency ratio (CR) or the
inconsistency rate that allows reviewing judgments. The CR is the ratio of Consistency
Index (CI) and Random Consistency Index (RI). Saaty (1980) has developed consistency
index using the following:
λmax – n
CI = (2)
(n – 1)
where λmax is the maximum eigenvalue and n is the size of the matrix.
Random consistency index (RI) was generated by Saaty (1980) based on the number of
random samples (Table 4-2).
Table 4-2 Random consistency index (RI) (Saaty 1980)
n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
RI 0 0 .58 .9 1.12 1.24 1.32 1.41 1.45 1.49 1.51 1.53 1.56 1.57 1.59
A CR value of 0 specifies that the comparison is perfectly consistent while CR>0.1
specifies inconsistent comparison and needs a revision.
58
AHP method is advantageous in rating the factors and sub-classes, but some uncertainties
are associated with its rating method. However, the pairwise comparison method used in
AHP offer a simple and satisfactory decision rule. Although, there is certain degree of
dependency of the landslide causative factors in influencing landslides but AHP considers
each factor in hierarchy as an independent entity. AHP delivers a simple and flexible
decision-making.
The regional scale and the sub-regional scale adopted this AHP approach, where the
corresponding LCFs are integrated using eq. 3 for developing the. Landslide susceptibility
index (LSI) maps.
N
LSI = ∑j=1 Weighted Factors(Fj)Weights of factor classes(Wj) (2)
4.1.2 Landslide susceptibility mapping using FR
The frequency ratio, FR is a bi-variate statistical approach, which is based on the
relationships landslides distribution and the landslides causative factors. This method
compares every data layer to the landslide locations (Kanungo et al. 2009). Generally, it
assumes that the landslides occurrences are directly related to landslide occurrences.
The FR model is a probabilistic model, which is simple and easy to understand. The FR
values helps in defining weights of factors, and it is defined as the ratio of the landslides
area to the total study area. In other words, FR is also expressed as the ratio of the
probability of a landslide occurrence to a non-occurrence (Pradhan and Lee 2010). A FR
value >1 means the relationship between the landslide occurrence and the particular factor
is strong whereas, FR lower than unity indicates that there is a slight relationship between
the landslide occurrence and the factor. The FR value is computed as-
% landslide occurence in each classes ( points in factor class⁄total points)
FR = = (3)
% area of an independent factor (factor class area⁄total area)
Landslide susceptibility index (LSI) map is developed using eq. 4.
LSI=∑(Fij×Wij) (4)
where Fij =Weighted factors and Wij=weights obtained from Prediction rate.
59
4.2 Sub-regional scale methodology
The sub-regional scale LSM assessment utilizes nine landslide causative factors- LULC,
NDVI, slope, aspect, curvature, elevation, distance to roads, distance to rivers, and distance
to fault layers. The generation of the LSM is done using the AHP approach, and carried
out at three different spatial resolutions 10m, 30m, and 90m. The methodological
framework for the LSM assessment at sub-regional scale is displayed in Fig 4-2.
DATA PARAMETER
SOURCES MAPS
LANDSAT 8 / NDVI
SENTINEL 2 IMAGES LULC
SA
FAULT DISTANCE
TO FAULTS ANALYTIC HIERARCHY PROCESS
DATA
STREAM DISTANCE
LAYER TO STREAMS LANDSLIDE SUSCEPTIBILITY MAPS
DEM DATA ELEVATION
VALIDATION
(SRTM 90m, CURVATURE
ASTER 30m, ASPECT
CARTOSAT 10m) SLOPE
Fig 4-2 Methodological framework for sub-regional landslide susceptibility assessment
(LSM)
For the sub-regional scale LSM assessment, thematic data layers are generated/developed
for the three resolutions 10m, 30m, and 90m. After acquiring the thematic layers for the
different resolutions, AHP approach is adopted to assign weights to each of the factors
using the frequency of landslide distribution within the sub-classes.
The weighted factors are then integrated to produce three scenarios of LSM having
different spatial resolutions. The accuracies of the LSMs are calculated using the validation
60
procedures. Comparisons of the accuracies are performed and the role of the resolution in
the LSM assessment in the particular study area is analysed.
4.3 Plot scale methodology
A different methodological approach is adopted in the plot-scale analysis. The plot scale
study is done to analyse and understand the slope failure mechanism of identified historical
landslide using the slope-mass rating techniques in a test site (Langdeibung area, Noney)
along NH-37. The plot scale methodological approach uses data collected using expert
knowledge during field investigations. Information, such as, number of joints, dip, strike,
amount of dip, discontinuity condition, ground water condition, and strength of rock are
utilized in the assessment. The methodological framework for the Plot scale is shown in
Fig 4-3.
Field data
No. of joints Rock Quality Designation (RQD)
Strength of intact material
Discontinuities spacing
Rock Mass Rating (RMR)
Discontinuities condition
Ground water condition
Dip direction (αj)
Line of discontinuities direction (αi)
Slope Mass Rating (SMR)
Line of discontinuities direction (αi)
Dip of joint (βj)
Slope Stability
Plunge of discontinuities (βi)
Condition
Inclination of slope (βs)
Fig 4-3 Methodological framework for slope stability assessment (plot scale)
61
4.3.1 Rock Quality Designation (RQD)
Deere (1963) introduces the rock quality designation (RQD) as an index to quantitatively
access rock quality. This method is particularly advantageous in providing a quantitative
estimate of rock mass quality using drill core logs. However, for this study, RQD value is
obtained using the indirect method developed by Palmstrom (1982).
RQD was calculated as-
RQD= 115-3.3Jv (5)
where Jv- sum of no. of joints/unit area (N) + (N/3).
4.3.2 Rock Mass Rating (RMR) Classification
RMR classification is adopted in this study to assess the slope stability. Bieniawski (1973)
has developed the RMR classified and later on modified by (Bieniawski 1979, 1989). RMR
is adopted in engineering practices because of its versatility and easy to use. The calculation
of RMR requires certain input parameters- RQD, strength of rock, joints spacing,
conditions of discontinuities, and groundwater conditions, derived during field
investigations.
Table 4-3 Field estimates of strength of rock (after Hoek et al. 1995)
Grade Term σc (MPa) I (MPa) Field estimates of rock Strength
R6 Extremely >250 >10 Repeated hammer blows can chip the rock
Strong
R5 Very 250-100 10-4 Many blows of a geological hammer for
Strong breaking the intact rock specimens
R4 Strong 100-50 4-2 A single blow of geological hammer can
break hand held rock specimens
R3 Mod. 50-25 2-1 Knife just scrapes surface
strong
R2 Weak 25-5 - Knife can cut but too hard to shape
R1 Very 5-1 - Can be cut with knife
Weak
R0 Extremely 0.25-1 - Gouged by thumbnail
Weak
σc= Uniaxial Compressive Strength of intact rock; I-Point Load Index
62
RMR is calculated as (Bieniawiksi 1979, 1989) –
RMR= Ratings of (RQD + Strength of material + Discontinuities spacing +
Condition of discontinuities + Ground Water condition) (6)
The RMR classification in this particular study has adopted the RMR classification of
Bieniawiksi (1979, 1989) and Hoek et al. 1995 it is presented in Table 4-4.
4.3.3 Slope Mass Rating (SMR)
Slope Mass Rating (SMR) system is a quantifying method used to evaluate the stability of
rock on slopes Romana (1985). RMR has a significant role in the assessment of rock slopes
stability. SMR is resultant from RMR by the addition of factorial adjustment factors for
the discontinuity orientation. The SMR is calculated as-
SMR= RMRb + (F1×F2×F3) + F4 (7)
F1- Degree of parallelism between the slope and the discontinuity
F2- Depends on the relationship of dip of discontinuity and inclination of slope.
F3- Depends on the dip of discontinuity
F4- Adjustment Rating for Methods of Excavation of Slopes
The adjustment factors F1, F2, F3 are derived from discontinuity orientation and
orientation data of slope. The F4 value is derived with respect to the excavation type. The
rating for method of excavation (F4) is taken as 0 as blasting or mechanical excavation has
been done in all slopes.
4.4 Remarks
The landslide assessment studies conducted at different scale of studies in the region of
Manipur, India. The main goal of carrying out landslide assessment in the study area is to
better understand the scenario of landslide occurrence in the area. The different scales of
study, applying different methodologies, give ideas on the landslide potential
areas/susceptible zones and in understanding the slope failure mechanisms.
63
Table 4-4 Ratings used for RMR classification
PARAMETER RATING
Strength Point load > 10 MPa 10-4 MPa 4-2 MPa 2-1 MPa
of intact Uniaxial > 250 MPa 250-100 MPa 100-50 MPa 50-25 MPa 25-5 MPa 5-1 MPa < 1MPa
material compressive
Rating 15 12 7 4 2 1 0
RQD 90-100% 75-90% 50-75% 25-50% <25%
Rating 20 17 13 8 3
Discontinuities spacing >2m 2-0.6m 60-20 cm 20-6 cm <6 cm
Rating 20 15 10 8 5
Condition of Very rough surface, Slightly rough Slightly rough Slickensides Soft gouge >5mm thick OR
discontinuities Not continuous, No surface, surface, surface OR separation >5mm, Continuous
separation, Separation Separation Gouge <5mm
Unweathered wall <1mm, <1mm, thick OR
rock Slightly Highly Wea- separation
weathered thered walls 1.5mm,
walls Continuous
Rating 30 25 20 10 0
Ground water dry Damp Wet Dripping Flowing
condition
Rating 15 10 7 4 0
64
