Sådhanå (2023)48:187 Ó Indian Academy of Sciences

https://doi.org/10.1007/s12046-023-02242-9Sadhana(0123456789().,-volV)FT3](012345
6789().,-volV)

Early prediction framework for a rainfall-induced landslide:

validation through a real case study
PRASHANT SUDANI* and K A PATIL

Department of Civil Engineering, College of Engineering, Pune, Maharashtra 411005, India

e-mail: [email protected]

MS received 31 March 2023; revised 5 June 2023; accepted 22 July 2023

Abstract. Shallow landslide occurrence is most common in rainy season in the form of flow of debris, costs
heavy damages to the infrastructure and human lives. Early prediction framework of such disaster can help to
mitigate such damages. The present work deals with prediction framework for initiation of debris flow, which is
developed and validated with real case study. In order to test reliability of prediction framework, back analysis
of very recent landslide debris flow accrued in the study area, Taliye village of Konkan region of Maharashtra,
India on 22 July 2021 was carried out. Simulation results of landslide stability were compared with the leaky
barrel-based rainfall-water saturation algorithm. Relations of landslide stability with the water saturation were
established through physically based approach using Geo-Studio analysis module. Leaky barrel algorithm was
used for study location for monitoring effect of rainfall on water saturation. The result confirms the good
predictability of landslide occurrence through a developed early prediction framework. The methodological
framework was presented in this paper for prediction of shallow landslide occurrence and recommended for real-
time monitoring of landslide prone locations.

Keywords. Shallow landslide; debris flow; early warning; rainfall induced landslide.

1. Introduction debris flow. Availability of loosely placed soil, topography,

and antecedent moisture condition of soil determine the
The word ‘‘Landslide’’ swiftly got attention of stakeholders vulnerability of slope failure for rainfall-induced landslides
associated with, due to growing frequency of its adverse and are referred to as causative factor [9]. During rainfall,
incidences all over the world. Increased population and infiltration, distribution, and drainage of water lead to a rise
their preference of choosing hilly areas for residence, raises in soil water saturation/degree of saturation which could be
the disturbance to the slope. During 2004 to 2016, more considered as triggering factor for initiation of landslide.
than 55000 fatalities were reported globally due to land- This can cause a critical development of positive pore water
slide [1]. pressure and loss of soil suction, which eventually fail the
Landslide Early Warning System (LEWS) is an efficient slope [9].
tool to predict the hydro-geological state of the soil and For rainfall-induced landslides quantity of rain water
determine elevated level of landslide danger [2–5]. They which can most likely resulted into landslides is threshold
allow decision makers to timely warn the people at risk and [10]. Physical and empirical threshold are two broad cate-
move them to safer place. LEWS can apply on a single gories of rainfall threshold for landslides predictions
slope for site specific landslide and on region covered with [11, 12]. Failure of soil slope depends on morphological,
several kilometers of same geomorphology and hydrologi- hydrological, and geotechnical characteristics of location,
cal condition called as regional landslide [6]. In terms of and hence setting of physical threshold required all these
accuracy, site specific landslide warning system provides information [13]. Physical threshold on rainfall is deter-
higher degree of accuracy compared to regional landslide mined with integrating impact of rainfall on stability of
early warning system. Several working landslide early slope through physically based hydro geological models
warning system available globally, reviews for which are [14–17]. Presently many physical approaches have been
given for local [7] and regional warning system [8]. proposed by various researchers for early warning of
Failure of land slope is related to factors acting adversely landslide through linking rainfall distribution with soil
on stability of slope. Besides many reasons of landslides, profile, soil hydrology, geotechnical factors and antecedent
rainfall-induced landslide is too common, resulted in mud/ moisture condition [10, 17–19].
Empirically based models for landslide prediction are
*For correspondence widely used due to their simplicity. Most triggering rainfall
187 Page 2 of 13 Sådhanå (2023)48:187

threshold for landslides are determined through empirically presented in this paper for prediction of shallow landslide
based methods [12, 20, 21]. The thresholds derived based occurrence, recommend for real-time monitoring of land-
on empirical models that are popularly used are on relation slide prone locations.
between rainfall intensity duration which is log-log diagram
and past incidences of landslide. Threshold based on such
empirically based model lacks sufficient landslide infor- 2. Study area
mation which cannot provide accurate rainfall threshold.
Physically based models have been used the detailed Taliye village, nestled in the Mahad tehsil of the Raigad
information about landslide, providing best way to analyze District of Maharashtra, India and is situated approximately
problem, they are able to predict stability of slope. Models 90 km from the district headquarters in Raigad. This village
like this efficiently used the geotechnical properties, strat- is situated in the Deccan trap region of the majestic Wes-
ification, topography, and hydrology of the landmass. For tern Ghats of India. The village can be located precisely at
physically based models, strength parameters of soil, such N18.29 and E73.08 coordinates, and its geographical
as cohesion and angle of friction are involved. SINMAP, location can be seen in figure 1.
TRIGRS and SLOP/W SEEP/W are widely used physically On the ill-fated day of 22nd July 2021 at 17:00 IST, the
based stability models which provide the stability status of hill slope adjacent to the residential area suffered a catas-
land slope in terms of factor of safety. trophic failure. This incident has been recorded as one of
However, these models can determine the status of the the most devastating tragedies in the state in recent decades,
landslide stability based on soil type, strength parameters with over 80 people losing their lives. The calamity was
and hydrological properties i.e., degree of saturation, which triggered by a colossal hillock that came tumbling down on
yields the relation between water saturation and land mass the village, crushing 35–40 houses in its path. The current
stability, but it cannot provide rainfall threshold directly. condition of the Taliye land slope, both before and after the
Linking of rainfall events with the saturation of the soil incident, can be viewed in figure 2.
mass is possible, through leaky barrel model worked on A field work was conducted on November 15th, 2022, to
water balance principle, presented by Fiorillo and Keefer the location of the devastating landslide that occurred in
[22, 23]. Taliye village. The primary objective of this visit was to
The present study attempts to build cost effective and examine the failed land slope and obtain representative soil
efficient early prediction framework for landslide through samples for determination of soil properties in a geotech-
interlinking of rainfall to water storage in soil mass and nical engineering laboratory. During the site visit, the team
water storage in soil mass to stability of land slope. A gathered vital information from government officials,
complete experimental investigation was planned to map which included the number of homes in the village (135),
the effect of the saturation on the soil suction and unsatu- the village’s population (750), the death toll resulting from
rated shear strength of soil. The stability of the soil slope the landslide (87), the water table’s depth (in range of 10 to
was derived using limit equilibrium analysis using SEEP/ 12 m below the foothill), the bedrock’s location on the
W, SLOPE/W module of geo-studio for various water hillock (3 to 4 m from the surface), and the time it took
saturation levels considering saturated unsaturated shear for the debris to reach the village after the slope failure
strength of soil. Initiation of debris flow corresponding to (5–10 s).
the real-time rainfall is monitored using a leaky barrel
model. A leaky barrel-based algorithm is developed for
monitoring the saturation evolution in the soil slope cor-
responding to rainfall events, which provides the real-time
water saturation of the soil. Stability status from Geo-studio
limit equilibrium analysis corresponding to water saturation
and real-time monitoring of water saturation using newly
developed algorithm corresponding to rainfall event yield
the three types of warning status, i.e., green, yellow, and
red. The present study aims to outline and endorse the
configuration of the landslide early warning system and the
accompanying potential of the approach. The prediction
framework for debris flow initiation is developed and val-
idated with real case study. In order to test the reliability of
prediction framework, back analysis of very recent land-
slide debris flow accrued in the study area, Taliye village of
Konkan region of Maharashtra on 22 July 2021 was carried
out. The result confirms the good predictability of landslide
occurrence through the presented framework. Procedure Figure 1. The geographical location of the study area.
Sådhanå (2023)48:187 Page 3 of 13 187

saturation and develop an algorithm for real-time moni-

toring of soil saturation levels corresponding to rainfall.
The first two phases of the study established a correlation
between the water saturation of the land slope and its sta-
bility status, which was measured in terms of the factor of
safety. The third phase resulted in the development of an
algorithm that provides real-time information on the satu-
ration levels of the land slope in response to rainfall.
Together, the results of the study offer a comprehensive
understanding of the stability status of land slopes under
varying saturation conditions. By simulating real-time sta-
bility status, the findings can inform effective decision-
making and management strategies to prevent landslides
and mitigate associated risks.

3.1 Phase 1: Investigating the geotechnical

properties of soil
3.1.1 Field density of soil To estimate the field density
of the study area, a core cutter test was performed at the
study location. The method involved a 1000 cc cylindrical
steel cutter with cutting edge at bottom, which has to drive
into soil to get core of soil of known volume of 1000 cc.
After removing the soil core from the steel cutter, the
weight of the soil core was measured, and the bulk density
of the soil was determined by calculating the ratio of the
volume of the cutter to its weight. The standard method of
performing filed density test using core cutter was used as
outlined in IS 2720 (part-29) code of procedure. To have
entire spectrum of soil layer over bed rock, field density test
Figure 2. Taliye Land slope before (top image [24]) and after was conducted at depth of 1 m, 2 m, and 3 m. The test
(bottom image) failure. provided the bulk density of the soil at various strata
ranging from 1.5565 to 1.5735 gm/cc, with a corresponding
water content range of 37.51 to 38.45%.
Upon closer inspection of the Taliye land slope, it was
discovered that the entire soil deposit had been washed 3.1.2 Soil Sample Collection The representative soil
away, revealing clear rock with a distinct failure cut mark samples were collected from different depths of soil layer at
in the remaining soil. The field visit also revealed that the study location. Hand operated augur boring was used to
soil layer’s thickness on the bedrock of the Taliye village collect the soil samples. Figure 3 offers a visual depiction
hill slope was 3.0 m. For a better understanding of the of both the field density test location and the location where
geotechnical stratification of soil on the land slope, field soil samples were collected.
density tests were performed on unfailed portion of soil The collected soil samples from the study area were
slope, provided the soil density at 1 meter, 2 meters, and 3 subjected to extensive geotechnical testing in the labora-
meters depths. tory. The collected representative soil sample was in dis-
turbed form, which then remolded at field density in
laboratory for performing geotechnical testing. The results
3. Methodology of these tests have been presented in table 2 and the result
and discussion section of this article, which provides a
The study followed a methodological approach consisting detailed outline of the various parameters that were tested.
of three significant phases to investigate the stability of land
slopes. The first phase focused on examining the geo- 3.1.3 Soil water characteristic curve (SWCC) The
technical properties of soil, while the second phase focused effect of water saturation on the soil suction is a significant
on investigating the impact of water saturation on the sta- factor to look upon when investigating stability analysis on
bility of land slopes. In the third phase, the study aimed to soil slope with varying water saturation levels. In this study
investigate the relationship between rainfall and soil the determination of SWCC curve, which represents the
187 Page 4 of 13 Sådhanå (2023)48:187

content. The soil matric suction was determined through a

calibration curve offered by ASTM D 5298 standard.

3.1.4 Shear test The direct shear test has proven to be

an effective means of evaluating the shear characteristics of
landslide-prone soil due to its compatibility with thin
specimens and the similarity of its shearing mechanism to
that of landslides along bedding planes. To assess the shear
parameter of the landscape soil, the suitability of
performing a drained or undrained test is a matter of
consideration. Stark (2005) recommends selecting an
appropriate shear test based on the involved loading rate
and hydraulic conductivity [25]. Watabe et al (2009) [26]
suggest performing an undrained test for low permeability
values where draining water during the shear test is not
feasible. A drained test is necessary to assess the soil’s
shear parameter for high permeability values where water
may escape from the soil sample during the shear stage.
Based on the loading rate, Farhan et al (2016) [27] indicate
that an unconsolidated undrained test is preferable for a
Figure 3. Location of soil sample collection and field density quick response of rain on the slope. After considering the
test on Taliye hill slope. aforementioned findings, the undrained box shear test was
chosen for evaluating the soil’s shear parameter.
The impact of saturation on the apparent shear strength
relationship between soil matric suction and water content, parameters was extensively studied in the geotechnical
was estimated using filter paper method using ASTM engineering laboratory, utilizing the direct shear test. The
D-5298-03 standard. By utilizing derived SWCC curve, soil sample preparation methodology involved pulverizing
study was able to investigate the behavior of soil in and air-drying the field-obtained soil samples overnight
different water saturation. before testing. A soil sample size of 90 cm3 was used, and
In this study, Whatman No. 42 filter paper was utilized the soil was carefully weighed to achieve the same field
for measuring soil matric suction. Before the test, the filter density as observed in the field. The soil was compacted in
paper was oven-dried overnight and precisely weighed to the shear box with limited compaction to attain the desired
an accuracy of 0.0001 gm. A soil sample that passed density. A strain-controlled direct shear machine was
through a 2mm sieve was used to prepare the test specimen. employed as the shear test equipment, consisting of a shear
The soil was subsequently oven-dried and mixed with dif- box, soil container, loading unit, proving ring, and a dial
ferent quantities of water, which were then determined at gauge for measuring shear deformation and shear force.
the end of the test as 14.11%, 17.35%, 20.13%, 22.17%, The soil container utilized in this study was a two-piece
25.25%, 26.16%, 32.02, and 40.66%. Each soil sample was square shear box, while a proving ring was used to measure
placed in a sealed container for approximately 24 hours to the shear load experienced by the soil along the shearing
reach moisture equilibrium. The soil was then compacted at plane. A constant shear strain rate of 0.125 mm/min was
the field density through static compaction. Each soil preferred to ensure consistent test conditions. During the
sample was compacted into two parts to accommodate the shear tests, various normal stresses of 50 kPa, 100 kPa, and
placement of the filter paper for direct contact measure- 200 kPa were applied to the soil samples. Each soil sample
ment. The filter paper was covered with two additional filter was tested three times under every normal stress. The
papers to prevent soil particles from coming into contact. failure envelope of the soil was determined by analyzing
This assembly of soil and filter paper was then hermetically the relationship between normal stress and shear stress at
sealed in a container, allowing for moisture equilibrium for failure, and the shear strength parameters, namely cohesion
one week in a desiccator set at a constant temperature of and internal friction angle, were determined.
20°C. The container was sealed with adhesive tape to
ensure no air or moisture leakage. After reaching moisture
equilibrium within the container, the wet filter papers were 3.2 Phase 2: Investigating the effect of water
carefully weighed to the nearest 0.0001 g. The wet soil, saturation on the land slope stability
obtained from the mid-height of the specimen, was col-
lected, and precisely weighed to determine its water con- This phase of the study was aimed to investigate the sta-
tent. Subsequently, the soil was oven-dried at a constant bility status of land slopes under different levels of water
temperature of 105°C and re-weighed to calculate its water saturation. To achieve this, the physical Limit Equilibrium
Sådhanå (2023)48:187 Page 5 of 13 187

Method (LEM) modeling using SLOP/W module of Geo- 3.3 Phase 3: Investigating the effect of rainfall
studio software was employed. Physical Limit Equilibrium on the evolution of soil saturation
Method (LEM) Modeling is a numerical technique used in
geotechnical engineering to analyze the stability of slopes This part of the study focuses on assessing the evolution of
and structures. It is based on the principles of limit equi- soil saturation in real time based on rainfall precipitation.
librium, which assumes that the soil mass reaches a state of To achieve this, the leaky barrel concept was used to
equilibrium when the forces acting on it are balanced. In monitor infiltration and percolation through the soil. By
Physical LEM Modeling, the slope is divided into discrete balancing the water in the soil barrel through infiltration
slices, and equilibrium equations are applied to each slice. and percolation, researchers were able to map the water
The method considers the shear strength parameters of the stored in the soil sample, allowing for the determination of
soil. The modeling process involves determining the factor the degree of saturation in the soil slope.
of safety, which is a measure of the stability of the slope or To develop an algorithm for real-time monitoring of
structure. The factor of safety is calculated by comparing amount of water stored in the soil sample corresponding to
the resisting forces to the driving forces. If the factor of rainfall, Once the water stored in the soil is known, satu-
safety is less than one, it indicates potential instability and ration of soil can be determined through properties of the
the possibility of slope failure. These modules require the soil, such as density, specific gravity, and void ratio as
input of important parameters such as the geotechnical mentioned in equations (1) and (2).
characteristics of the soil, soil suction characteristics, and G  cw
piezometric line. The piezometric line refers to a line that cd ¼ ð1Þ
1þe
represents the pressure head of water within the soil. It
signifies the water pressure at different points within the W G
e¼ ð2Þ
soil mass. Sr
Through the use of these analytical tools, the relationship
between soil saturation and the stability status of the land where cd is dry density of soil, G is specific gravity of soil,
slope was established. The analysis allowed for a compre- e is void ratio of soil and Sr is water saturation of soil mass.
hensive understanding of the behavior of the soil slope Once these properties and the amount of water stored in the
under varying levels of water saturation, enabling soil were known, the algorithm was able to effectively
researchers to accurately predict the stability status of the monitor the saturation in the soil slope corresponding to the
land slope and identify potential risks of slope failure. rainfall. The geographical methodological approach of this
study is presented in figure 4, which visually represents the
methodology used in the study.

4. Result and discussion

4.1 Geotechnical properties of soil

To obtain a comprehensive understanding of the soil
properties associated with the slope of the Taliye village, a
detailed geotechnical engineering experimental investiga-
Figure 4. Graphical methodology of landslide early warning tion was conducted. The geotechnical investigation
approach. involved assessing various soil properties, including natural
moisture content, specific gravity, liquid limit and plastic

Table 1. Geotechnical investigation of soil and results.

Sl. no. Description Unit Taliye soil Reference

1 Natural water content % 38.40 IS 2720; Part-2
2 Specific gravity NA 2.66 IS 2720; Part-3
3 Liquid limit % 48 IS 2720; Part-5
4 Plastic limit % 39.40 IS 2720; Part-5
5 Plasticity index % 8.6 IS 2720; Part-5
6 Classification NA MI IS 2720; Part-4
7 Cohesion kN/m2 18.9 IS 2720; Part-13
8 Angle of internal friction ˚ 27.26 IS 2720; Part-13
9 Permeability (Falling head) cm/sec 1.87 910-6 IS 2720; Part-17
187 Page 6 of 13 Sådhanå (2023)48:187

limit, plasticity index, sieve analysis, free swell index,

Degree
Shear parameters

18
27.26
shear parameters, and permeability. These assessments

27

28
were carried out in the geotechnical laboratory of the
College of Engineering Pune (COEP). The findings of the
Cohesion
kN/m2.
geotechnical engineering experimental investigation have

17
18.63

17.65

18.63
been tabulated in table 1 for reference.
A thorough investigation was conducted to determine the
Specific
Gravity

geotechnical properties of soil at various depth of soil

Gs
15
2.66

2.66

2.66
present on the bedrock of the Taliye hill slope. The soil
properties were meticulously examined in the geotechnical
laboratory at the College of Engineering Pune. The findings
Moisture
Natural

Wn %

of this investigation, including test results and analysis,

14
38.45

37.51

38.40
have been presented in table 2 and section 4.2.
Atterberg’s limit

density
gm/cc
Bulk

13
1.56

1.55

1.56
4.2 Effect of saturation on apparent shear strength
parameters
Plasticity
Index

The measurement of field density and specific gravity was

12
8.93

7.8

8.5

utilized to determine the necessary water content for

achieving 100% saturation in the soil of the Taliye hill
slope with using equations (1) and (2). It was found that a
Plastic
Limit
11

water content of 50% is necessary to achieve complete

39.6

40.2

40

saturation. An investigation was conducted to examine the

impact of water saturation on the apparent shear strength
Liquid
Limit
10

parameters, namely cohesion and angle of internal friction.

48.5

48.5
48

The results obtained from this investigation are depicted in

Silt % Clay %

figure 5.
- 0.002 - Less

0.002
than

mm

17.28

16.58

16.98
9

Figure 5 presents the influence of water saturation on the

apparent cohesion and angle of internal friction, which are
critical factors for soil strength. Initially, the cohesion
to 4.75 0.075
mm
to

54.18

54.42

55.02
8

exhibited a 10.32% increase, reaching its highest point at

Particle size analysis

20% water saturation, potentially due to its impact on soil

Summary of exploration soil data of Taliye village hill slope soil.

softness. As saturation increased further to 100%, the

0.075
Sand

mm
%-

7
28.5

cohesion value declined, experiencing an 87.87% reduction

29

28

from its peak value. The angle of internal friction demon-

Gravel

mm &
above

strated a gradual and decreasing trend with increasing sat-

4.75
%-

uration. Over the range of 0% to 100% saturation, the angle

0

of internal friction decreased from 33.03° to 27.35°. At

classification
IS

5
MI

MI

MI

40 80
Angle of Internal Friction (Φ)

35 70
soil & stratum
description of

Clayey Silt with

Clayey Silt with

Clayey Silt with

30 60
Cohesion (KN/m2)
Intermediate

Intermediate

Intermediate
Relatively

Relatively

Relatively
Plasticity

Plasticity

Plasticity
Visual

25 50
no.
4

20 40
15 30
10 20
5 10
Sample

0 0
no.
3

0.0 20.0 40.0 60.0 80.0 100.0

1

Degree of Saturation (Sr in %)

in mtr. To

Sr Vs Φ Sr Vs C
2
1.00

2.00

3.00
Table 2.

Figure 5. Effect of saturation variation on apparent shear

Depth

0.00

1.00

2.00

strength parameters.
1
Sådhanå (2023)48:187 Page 7 of 13 187

100% saturation, both parameters significantly decreased in functions in Geo-studio, was thus adopted to solve the
value. stability problem in the present case. The slope stability
analysis was performed on the actual geometry of the
Taliye hill slope, as depicted in figure 6. The slope was
4.3 Effect of saturation on stability found to be 190 m high and 158 m wide at the base. The
geotechnical characteristics of a hill slope are critical in
To determine the stability status of the landscape, the Limit understanding its stability and susceptibility to failure.
Equilibrium Method (LEM) analysis was conducted using Table 3 showcases the essential parameters governing the
Geo-studio software. The SEEP/W and SLOP/W modules behavior of the geomaterials forming the slope, including
of Geo-studio were utilized to analyze the slope for both
their saturated state, cohesion, and angle of internal friction.
hydrological and stability status in sequences. For the sta-
A saturated-unsaturated material model was employed in
bility analysis, Morgenstern and Price’s [28] method was
the stability analysis of a slope using GeoStudio software.
employed, which utilizes a half-sine function to effectively
This model incorporated functions for volumetric water
consider interslice force, satisfying both force equilibrium content and hydraulic conductivity, which were crucial in
and moment equilibrium conditions. The inclusion of accurately representing the behavior of the soil. The volu-
interslice forces in the stability analysis is crucial for metric water content function depended on the amount of
obtaining accurate results; simpler methods that do not water in the soil and the matric suction. Data points were
account for interslice forces and equilibrium equations can
added to the stability model using test results of soil-water
lead to erroneous conclusions. The Morgenstern Price
characteristic curve (SWCC) to determine the volumetric
method, which considers both shear and normal interslice
water content function. On the other hand, the hydraulic
forces and allows for various user-selected interslice force
conductivity function was developed using the software’s
built-in function, utilizing a saturated hydraulic conduc-
tivity (permeability) value of 1.87 910-6 cm/s, which had
been determined for the soil. These functions played a
significant role in accurately modeling the behavior of the
soil, thereby enhancing the reliability of the slope stability
analysis.
To accurately model the behavior of the slope and make
reliable predictions, appropriate assumptions in terms of
boundary conditions were applied in the modeling process.
A rigid boundary condition was used to simulate the bed-
rock and other rigid constraints present in the slope. On the
other hand, a slope face boundary condition was applied to
model the open surface of the slope. To simulate pore water
pressure and seepage water level in the soil, boundary
conditions were applied corresponding to the water satu-
ration level of the slope. This involved applying a constant
head of pressure in the soil layer to represent the saturation
condition required for accurate analysis. In addition to these
boundary conditions, slip surface analysis was conducted
using the entry and exit method. This allowed for identi-
fying potential slip surfaces within the slope, which is vital
in accurately assessing the slope stability.
To analyze the stability of Taliye hill, an investigation
was conducted to study the effect of water saturation on
unsaturated shear behavior of soil. This was done by ana-
Figure 6. The geometry of Taliye hill slope (value in meters).
lyzing the soil’s soil-water characteristic curve (SWCC)
and piezometric line with respect to saturation. The soil

Table 3. Geotechnical properties of the geomaterials forming the slope.

Soil/rock Unit weight (kN/m3) Cohesion (kN/m2) Friction angle (degree)

Soil (Top layer) 15.30 18.9 27.26
Basalt rock (Bed layer) 27.8 750.5 38.2
187 Page 8 of 13 Sådhanå (2023)48:187

100.0 3 the soil. Void ratio of Taliye hill soil was found to be 1.36.
90.0 Thickness = 3 m
Eff Porosity = 11 % 2.5
As the volumetric water content (VWC) increased up to
Degree of Saturation (%)

80.0 Total Porosity = 57 %

Piezometric level (m)

46% equivalent to 80.70% of water saturation, water was
70.0 2
60.0
gradually accumulated and retained in the capillary pores of
Retention Zone θmax = n - neff = 46% the soil. The effective porosity of the soil, which is the
50.0 1.5
40.0 θ < F.C. θ = 46% θsat = 57% difference between total porosity and the field capacity
1
30.0 [22], was found to be 0.11. i.e. 0.57–0.46. Figure 7 illus-
20.0 0.5 trates that when the volumetric water content (VWC)
10.0
exceeds 46%, an increase of 11% in VWC within the soil
0.0 0
0 10 20 30 40 50 60 can result in the rise of the piezometric line to a depth of 3
Vollumetric water content (%) meters, ultimately bringing it to the surface of the soil.
VWC vs Saturation Piezometric level
The volumetric water content (h) varies between the
Figure 7. VMC, Saturation, and Piezo metric relationship for maximum value of field capacity (representing zero suc-
Taliye soil. tion) and the residual value (hresidual) that represents the soil
status during the summer season. Any excess water beyond
the field capacity (h [ hF.C.) may have been stored tem-
0.8 porarily, generating positive pore water pressure in the soil.
0.7 However, this excess water is eventually removed from the
Degree of Saturation

0.6 system by percolation.

0.5 In order to investigate the effect of water saturation on
0.4 the shear behavior of the unsaturated soil, the Soil Water
0.3 Characteristic Curve (SWCC) was derived in accordance
0.2 with ASTM D-5298-03, a standard test method for mea-
0.1 suring soil suction using filter paper. Figure 8 presents the
0 SWCC, which provides important information about the
1 10 100 1000 10000 100000 1000000
Metric Suction (kPa)
soil’s water retention characteristics and helps to better
Best fit SWCC Test result - 1 Test result - 2
understand the soil’s shear behavior under unsaturated
conditions.
Figure 8. Soil Water Characteristic Curve (SWCC) of Taliye The study examined the relationship between slope sta-
hill slope soil. bility and varying levels of water saturation, as depicted in
figure 9. The results showed that the slope’s stability ini-
exhibits maximum suction (negative pore water pressure) tially increased slightly up to 20% water saturation levels
during initial saturation conditions at relatively drier state and then continuously decreased with a positive change in
compared to other saturation levels and reaches maximum water saturation. The initial rise in stability could be
positive pore pressure when fully saturated (Sr = 100%). attributed to the effect of water on dry soil samples, which
The point at which the soil attains zero water pressure lies mobilized the soil cohesion. However, it was observed that
between these two extremes and is known as the field if water saturation continued to build up, the slope would
capacity [22]. To determine the field capacity of the soil, lose its stability upon reaching 90.5 % water saturation. The
samples were taken from the Taliye village slope and stability factor was found to be 1 at this point, and any
prepared in cylindrical barrels at the field’s density and further increase in saturation would result in slope failure.
moisture content. Water was then introduced drop-by-drop
onto the soil sample, and the test was performed for a week
to allow for even diffusion of the retained water in the soil’s 4
voids. As water began to accumulate in the bottom, capil- 3.5
Factor of Safety

lary porosity became completely saturated, and matric 3

suction became zero. Further accumulation of water in the 2.5
system generated positive pore water pressure in the soil. 2
The slope stability analysis was carried out for different 1.5
levels of water saturation to consider the effect of saturation 1
on the unsaturated shear behavior of the soil. 0.5
The water accumulation in the top 3 m of the weathered
0
soil of the Taliye hill slope is illustrated in figure 7. Ini- 0 10 20 30 40 50 60 70 80 90 100
tially, the porosity of soil if found to be 0.57. The porosity Degree of Saturation (%)
of the soil was determined by calculating the void ratio of
the soil, which was obtained using equation (1). This Figure 9. Effect of water saturation on the stability of the Taliye
equation considers the field density and specific gravity of hill slope.
Sådhanå (2023)48:187 Page 9 of 13 187

The findings suggest that monitoring the water saturation

in the soil mass could provide early warnings of landslide
hazards. The monitoring of water saturation in the field was
carried out using an algorithm, which is comprehensively
described in the subsequent section titled ‘‘The Impact of
Rainfall on the Soil Saturation Process.’’ To mitigate the
risks associated with slope failure, it is recommended to
implement a warning system with three warning levels,
providing clear messages on appropriate actions that need
to be taken by affected individuals. The study underscores
the importance of monitoring the water saturation levels in
the soil mass for prediction of landslide hazards.
Figure 10. Graphical setup of Leaky barrel.

4.4 Effect of rainfall on the evolution of soil

relationship with the water stored within the system, Z
saturation (Darcy’s law is valid). The coefficient of proportionality,
The primary objective of this study is to monitor the known as the drainage coefficient, is represented by the
dynamic changes in the degree of soil saturation in response symbol Kd and is measured in units of inverse time as
to varying rainfall events. The leaky barrel concept is presented in equation (3) [31].
employed to model the experimental setup for obtaining the
DrainageRate ¼ K d  Z ð3Þ
evolution of saturation in the soil mass. Specifically, the
focus is on examining how rainfall triggers infiltration from where Kd is drainage coefficient and Z is water stored in
the slope’s surface, leading to water diffusion in all direc- soil system.
tions. Some of the water infiltrates into the soil voids, while The quantity of water stored within soil masses is subject
the rest percolates and drain out from slope. By leveraging to change, with fluctuations depending on the intensity of
these infiltration and percolation rates, this study endeavors rainfall (I(t)) and the corresponding drainage rate (KdZ).
to construct a comprehensive map of the saturation varia- Specifically, the net variation in water storage can be
tion in the soil mass in response to rainfall events. The modeled using a linear first-order differential equation (4)
research findings offer valuable insights into the dynamics [31].
of water saturation within soil slopes during rainfall.
Overall, the results demonstrate the effectiveness of the dZ
¼ I ðt Þ  K d  Z ð4Þ
employed approach in accurately mapping the temporal dt
changes in water saturation in the soil mass. Where, ‘I(t)’ is rainfall intensity in any time interval ‘t’
To simulate the storage of infiltrated rainwater within the To effectively monitor the behavior of water storage
soil mass of a slope, a numerical model based on the leaky within soil systems, it is necessary to consider two distinct
barrel concept is employed, as previously proposed by scenarios as (i) an increase in water retention during rainfall
Keefer [23]. This model bears similarity to the widely used events, and (ii) a subsequent decrease in retained water
‘‘tank model’’ in Japan for rainfall-runoff simulation after rainfall has ceased. During periods of rainfall, the
[29, 30]. The leaky barrel model is capable of accurately level of retained water within the soil system, denoted by
simulating the evolution of rainfall into water saturation, the symbol Z(t), can be characterized using the relationship
utilizing a cylinder open at the top to receive rainfall and a presented in equation (5) [31]:
water collection system at the bottom for percolation
through the soil sample as illustrated in figure 10. This I  
Z ðt Þ ¼ 1  eK d t ð5Þ
approach is not only effective in reproducing the observed Kd
behavior of water infiltration and saturation in soil slopes
but also provides a practical and robust tool for predicting Following the cessation of rainfall, the quantity of water
the behavior of soil slopes under varying rainfall condi- retained within the soil system gradually decreases back, as
tions. The leaky barrel model represents a crucial contri- described by a logarithmic decay function as presented in
bution to the field of hydrology, as it enables the equation (6) [31].
quantification and prediction of water movement and stor-  
Z ðtÞ ¼ Z 0 eK d t ð6Þ
age within soil masses.
The water temporarily held within the system, which Where, Zo = Retained water in the soil sample when
corresponds to the free pore water in the zone of saturation, rainfall ceases, T* = Time elapsed since rainfall is ceased
is denoted by the symbol ‘‘Z.’’ The drainage rate of the An effective numerical simulation capable of accurately
leaky barrel system is assumed to exhibit a linear predicting the behavior of soil water storage in response to
187 Page 10 of 13 Sådhanå (2023)48:187

Figure 11. Leaky barrel model test setup.

rainfall can be obtained by integrating both scenarios allow for water to percolate and drain from the soil system.
described in equations (3) and (4), through the development The complete configuration of the leaky barrel test setup is
of a comprehensive algorithm presented in equation (7) illustrated in figure 11.
[31]. This algorithm offers a powerful tool for continuous, The soil was carefully placed into a cylindrical barrel at
real-time monitoring of soil water retention, providing the same density and moisture content as found in the field.
valuable insights into the dynamics of soil slopes. In order to achieve the same density in the barrel as in the
field, the required weight of dry soil was added to the mold.
  IðtÞ   The soil was compacted in the leaky barrel at field density
Z n ¼ ðZ n1 Þ eK d Dt þ 1  eK d Dt ð7Þ
Kd with controlled static loading. The compaction process
ensures that the soil particles are closely bound together at
By integrating both scenarios into a comprehensive
the desired density. Next, water was introduced to the top of
numerical simulation, the behavior of water storage within
the soil sample, one by one drop, until water began to move
soil systems can be monitored in a step-wise fashion. Here,
out of the bottom outlet. Each test was conducted over the
the subscript ‘‘n’’ represents the time step, while Dt denotes
course of a week to allow for the diffusion of any retained
the time increment and I(t) represent the intensity of rainfall
water within the soil sample. The water was also allowed to
during the considered time interval. Through iterative cal-
infiltrate the soil mass, and the percolation of the water was
culations, the amount of water stored within the soil system,
carefully recorded. Infiltration focuses on the entry of water
denoted by Zn, can be computed for each time step based on
into the soil from sources such as precipitation, while
the recorded rainfall data.
percolation emphasizes the downward movement of water
Equation (7) has been further modified in order to
beyond the point of infiltration.
enhance its capability to detect water saturation at each
By closely monitoring the rate of drainage and the
time step. This enhanced version is presented as equation
amount of water stored in the soil system, we were able
(8).
to determine that the drainage coefficient (Kd) of the soil
n    o
100  ðZ n1 Þ eK d Dt þ IðtÞ
Kd 1  e K d Dt
Sr ¼ ð8Þ 350 6000
Z max
300 5000
Cumulative Rainfall (mm)

Where Zmax represent the maximum amount of water 250

that can be retained in the soil mass. The resulting algo- 4000
Rainfall (mm)

200
rithm is both robust and easily programmable, making it a 3000
150
valuable tool for real-time monitoring and prediction of soil
2000
water saturation, with important implications to landslide 100

research. 50 1000

The leaky barrel model test setup was constructed using 0 0

polyvinyl chloride (PVC) material, with a cylindrical shape
measuring 110 mm in diameter and 250 mm in height. In
Date
order to accurately estimate infiltration rates, the top 50 mm Rainfall Cumulative rainfall
of the model was specifically cast with thread marks to
enable the storage of 20 ml of water per thread. Meanwhile, Figure 12. Rainfall data of Taliye village landslide location of
a perforated disk was placed at the bottom of the setup to the 2021 year (Landslide year).
Sådhanå (2023)48:187 Page 11 of 13 187
96 350
In order to accurately approximate the behavior of soil
94
300 saturation levels over time, an hourly record of rainfall in
92
250 the region was utilized. This data was then used to run a
90
Saturation (%)

Rainfall (mm)
200
numerical approximation algorithm, which generated a
88
series of iterations, each providing the degree of soil satu-
86 150
84
ration at the end of each hour. After completing a full
100 iteration process for the Taliye village location, were able
82
80
50 to generate a comprehensive set of results, which are pre-
78 0 sented in figure 13. These findings are highly significant, as
26-May 05-Jun 15-Jun 25-Jun 05-Jul 15-Jul 25-Jul 04-Aug they provide a detailed understanding of the dynamics of
Date (Year 2021)
soil saturation in the region over time.
rainfall Saturation Variation The output of the leaky barrel algorithm, illustrated in
figure 11, depicts the iterative approximation steps used to
Figure 13. Monitoring of water saturation in the slope soil
predict the water stored in the soil system. Remarkably, the
corresponding to the hourly rainfall.
algorithm was able to determine that the percentage of
water saturation in the land slope was more than 90.5% on
on the study location is 0.03 per hour. As part of our
the day of the landslide. Further, stability analysis revealed
research efforts, we also collected rainfall data from the
that the land slope would become unstable if the water
year 2021, which was a particularly significant year due
saturation increased over 90.5%. This finding was the result
to a major landslide that occurred at the study location.
of a rigorous stability analysis that included both limit
This data was gathered from a rain gauge situated in
equilibrium analysis and saturation evolution monitoring.
close proximity to the landslide site, as shown in fig-
Consequently, this study was able to accurately predict the
ure 12. Logarithmic equation (8), which takes into
occurrence of a landslide in Taliye village on 22nd July
account crucial inputs such as the drainage coefficient
2021.
(Kd) obtained from the leaky barrel test setup and the
The results illustrated in figure 14 interpret the remark-
rainfall data, was utilized for each time step. This
able predictive capability of the newly developed leaky
equation was employed to determine the saturation
barrel algorithm in anticipating landslides in response to
variation within the soil mass.
water saturation and stability conditions of land slopes. The
analysis revealed that a critical level of 90.5% water satu-
ration exists, beyond which the slope is susceptible to
failure. Interestingly, the real-time monitoring of water
saturation on an hourly basis recorded a value of 91.45% on
the day the landslide occurred, which effectively validates
the findings of this study.
The implementation of an early warning system (EWS)
for landslides based on this concept could prove to be a
significant step toward mitigating the potential impact of
such natural disasters in vulnerable regions. By integrating
this system into vulnerable areas of the country, it could be
better prepared to anticipate and respond to the risks posed
by landslides. Overall, this study highlights the potential for
innovative approaches to predict and prevent landslides,
which could ultimately save lives and reduce the damage
caused by these destructive events.

5. Conclusions

The study aims to provide early prediction framework for

rainfall-induced shallow landslide events. The present
study proposes the utilization of a leaky barrel-based
saturation monitoring algorithm for the real-time predic-
tion of rainfall-induced landslides. To evaluate the reli-
ability of this prediction approach, a back analysis of a
Figure 14. Landslide early warning prediction and validation. recent landslide event that occurred in Taliye, Raigad
187 Page 12 of 13 Sådhanå (2023)48:187

area of Maharashtra on July 22, 2021, was conducted. References

The results demonstrate that the leaky barrel model
simulations accurately depict the real-time saturation [1] Froude M J and Petley D N 2018 Global fatal landslide
variation in the slope corresponding to the rainfall, which occurrence from 2004 to 2016. Natural Hazards and Earth
is further linked to the slope’s stability and effectively System Sciences 18(8): 2161–2181. https://doi.org/10.5194/
identifies prominent landslide events. The approach con- nhess-18-2161-2018
sists of a specific multi-methodological approach, that [2] Wicki A, Lehmann P, Hauck C, Seneviratne S I, Waldner P and
Stähli M 2020 Assessing the potential of soil moisture
considers variable soil moisture and rainfall conditions,
measurements for regional landslide early warning. Landslides
thereby enabling the definition of constantly updated
17(8): 1881–1896. https://doi.org/10.1007/s10346-020-01400-y
triggering scenarios. Overall, this research suggests the [3] Sudani P and Patil K A 2023 Investigating effect of water
efficacy of the leaky barrel-based saturation monitoring saturation on stability of hill slope: Malin case study.
algorithm for the real-time prediction of rainfall-induced Advances in Civil and Architectural Engineering. https://doi.
landslides, as demonstrated through the back analysis of org/10.13167/2023.26.9
a recent landslide event and simulation. The results [4] Sudani P and Patil K A 2023 Evaluating the significance of
highlight the model’s ability to accurately represent slope saturation in shallow landslide failure: saturated–unsaturated
saturation and stability, with minimal false alarms. This soil perspective. Multiscale and Multidisciplinary Modeling,
is attributed to its reliance on factual data rather than a Experiments and Design.. https://doi.org/10.1007/s41939-
probabilistic approach. As a result, the model becomes a 023-00175-8
[5] Sudani P and Patil K A 2023 Assessment of variation in
valuable tool for mitigating and managing landslide
degree of saturation due to rainfall for landslide study.
hazards.
Geotechnical Engineering Journal 54(2): 36–40
[6] Sudani P, Patil K A and Kolekar Y A 2023 Historical
development of landslide early warning system (LEWS): a
List of symbols review. Lecture Notes in Civil Engineering. https://doi.org/
EWS Early warning system 10.1007/978-981-19-7245-4_24
LSEWS Landslide early warning system [7] Michoud C, Bazin S, Blikra L H, Derron M-H and
SWCC Soil water characteristics curve Jaboyedoff M 2013 Experiences from site-specific landslide
LEM Limit equilibrium model early warning systems. Natural Hazards and Earth System
VWC Volumetric water content Sciences 13(10): 2659–2673. https://doi.org/10.5194/nhess-
PVC Poly vinyl chloride 13-2659-2013
[8] Guzzetti F, Gariano S L, Peruccacci S, Brunetti M T,
°C Degree Celsius
Marchesini I, Rossi M and Melillo M 2020 Geographical
cd Dry density landslide early warning systems. Earth-Science Reviews 200:
G Specific gravity 102973. https://doi.org/10.1016/j.earscirev.2019.102973
cw Density of water [9] Bogaard T A and andGreco R, 2016 Landslide hydrology:
e Void ratio from hydrology to pore pressure. WIREs Water 3(3):
W Percentage water content 439–459. https://doi.org/10.1002/wat2.1126
Sr Degree of saturation [10] Harilal G T, Madhu D, Ramesh M V and Pullarkatt D 2019
C Cohesion Towards establishing rainfall thresholds for a real-time land-
U Angel of internal friction slide early warning system in Sikkim, India. Landslides 16(12):
h Volumetric water content 2395–2408. https://doi.org/10.1007/s10346-019-01244-1
I Rainfall intensity [11] Guzzetti F, Reichenbach P, Ardizzone F, Cardinali M and
Galli M 2006 Estimating the quality of landslide suscepti-
Z Rainfall stored in barrel
bility models. Geomorphology 81(1–2): 166–184. https://doi.
Zc Capacity of system org/10.1016/j.geomorph.2006.04.007
D Drainage rate [12] Aleotti P 2004 A warning system for rainfall-induced
Kd Drainage Coefficient shallow failures. Engineering Geology 73(3–4): 247–265.
T Time https://doi.org/10.1016/j.enggeo.2004.01.007
[13] Lagomarsino D, Segoni S, Fanti R and Catani F 2013
Updating and tuning a regional-scale landslide early warning
system. Landslides 10(1): 91–97. https://doi.org/10.1007/
Acknowledgements s10346-012-0376-y
[14] Badoux A, Graf C, Rhyner J, Kuntner R and McArdell B W
We express our gratitude to the AICTE-NDF 2009 A debris-flow alarm system for the Alpine Illgraben
Scheme (Application number 60400) for awarding us a catchment: design and performance. Natural Hazards 49(3):
research fellowship, which has enabled us to pursue our 517–539. https://doi.org/10.1007/s11069-008-9303-x
research endeavors. Additionally, we extend our appreci- [15] Baum R L and Godt J W 2010 Early warning of rainfall-induced
ation to the College of Engineering Pune, which has shallow landslides and debris flows in the USA. Landslides 7(3):
259–272. https://doi.org/10.1007/s10346-009-0177-0
facilitated our academic pursuits.
Sådhanå (2023)48:187 Page 13 of 13 187

[16] Bai S, Wang J, Thiebes B, Cheng C and Yang Y 2014 [23] Keefer D K, Wilson R C, Mari R K, Brabb E E, Iii W M B,
Analysis of the relationship of landslide occurrence with Ellen S D, Harp E L, Wieczorek G F, Alger C S and Zatkint
rainfall: a case study of Wudu County. China. Arabian R S 1987 Real-time landslide warning during heavy rainfall.
Journal of Geosciences 7(4): 1277–1285. https://doi.org/10. Science 238(13): 921–925
1007/s12517-013-0939-9 [24] Petley D 2021 The disastrous Taliye village landslide in
[17] Sudani P and Patil K A 2022 Evolution of early warning Maharashtra. Landslide blog, India
system for landslides. Disaster Advances 15(8): 46–59. [25] Stark T D, Choi H and McCone S 2005 Drained shear
https://doi.org/10.25303/1508da046059 strength parameters for analysis of landslides. J. Geotech.
[18] Salciarini D, Brocca L, Camici S, Ciabatta L, Volpe E, Geoenvironmental Eng. 131(5): 575–588. https://doi.org/10.
Massini R and Tamagnini C 2019 Physically based approach 1061/(asce)1090-0241(2005)131:5(575)
for rainfall-induced landslide projections in a changing [26] Watabe Y, Tanaka M and Kikuchi Y 2009 Practical
climate. Proceedings of the Institution of Civil Engineers - determination method for soil parameters adopted in the
Geotechnical Engineering 172(6): 481–495. https://doi.org/ new performance based design code for port facilities in
10.1680/jgeen.18.00216 Japan. Soils Found. 49(6): 827–839
[19] Wei Y, Wu X, Xia J, Miller G A, Cai C, Guo Z and Arash H 2019 [27] Ali F, Farooq K, Mujtaba H, Riaz A and Ulhaq E 2016
The effect of water content on the shear strength characteristics of Influence of saturation on rainfall generated landslides in shale
granitic soils in South China. Soil and Tillage Research 187: along Murree-Kohala road. Pakistan J. Geol. Soc. India 88(6):
50–59. https://doi.org/10.1016/j.still.2018.11.013 718–724. https://doi.org/10.1007/s12594-016-0539-x
[20] Mondini A, Carlà R, Reichenbach P, Cardinali M and [28] Morgenstern N U and Price V E 1965 The analysis of the
Guzzetti F 2009. Use of a remote sensing approach to detect stability of general slip surfaces. Geotechnique 15(1): 79–93
landslide thermal behaviour. In: EGU General Assembly [29] Hiura H and Sassa K 1985. Estimation of the landslide
Conference. 11, 7790 drainage effect in use of tank model. In: A. Takei (Ed.),
[21] Abraham M T, Satyam N, Pradhan B, Segoni S and Alamri Procedings of international symposium on errosion, debris
A 2022 Developing a prototype landslide early warning flow and disaster prevention pp. 355–360
system for Darjeeling Himalayas using SIGMA model and [30] Tan K S, Lim S B, Tan T L and Yang S L 1987. Landslide
real-time field monitoring. Geosciences Journal 26(2): problems and their control in Singapore. In: Southeast Asian
289–301. https://doi.org/10.1007/s12303-021-0026-2 Geotechnical Conference., 25–36
[22] Fiorillo F and Wilson R C 2004 Rainfall induced debris [31] Wilson R C and Wieczorek G F 1995 Rainfall thresholds for
flows in pyroclastic deposits, Campania (southern Italy). the initiation of debris flows at La Honda, California.
Engineering Geology 75(3–4): 263–289. https://doi.org/10. Environmental & Engineering Geoscience 1(1): 11–27.
1016/j.enggeo.2004.06.014 https://doi.org/10.2113/gseegeosci.i.1.11